Datasets:

blesspearl

/

arxiv-summarization

like 0

1111.3272

we present a systematic yet very basic review of the construction of homological evolutionary vector fields on the infinite jet spaces and on several natural classes of partial differential equations such as the gauge models . for a long time , this geometric structure developed in parallel in mathematics and physics . since mid-70 s , it has been being used intensively in theoretical physics : specifically , within the brst- or bv - technique for the quantization of gauge - invariant systems @xcite . in mathematics ,the concept first stemmed over smooth manifolds in the problem of the homological vector field realizations of lie algebroids @xcite , which encompass lie algebras and vector bundles and are best known in the framework of symplectic and poisson geometries .although the notion of lie algebroids in the geometry of manifolds had been known for a long time @xcite , the proper generalization of this omnipresent structure to the geometry of jet bundles ( c.f .@xcite ) appeared in @xcite . in principle , it _ could _ be contemporary to the discovery of the brst - technique ( @xcite vs @xcite ) . in the meantime , a steady progress in the geometry of jet spaces @xcite created the platform for important applications of the future concept .let us name only a few of them : * the variational poisson dynamics for the kdv - type systems @xcite , also @xcite ; * modern revisions of the theory of liouville - type hyperbolic equations @xcite , in particular , the 2d toda chains associated with the dynkin diagrams @xcite ; * the cohomological approach to gauge fields @xcite and ( quantum ) poisson theory @xcite . in this paper , we recall the jet - bundle geometry that stands behind the structures @xmath0 .it allows us to address , from a unified standpoint @xcite , many relevant physical models ranging from the korteweg de vries equation and its poisson structures @xcite to the yang mills theories , gravity , and poisson sigma - models @xcite . in section [ secinvolutive ]we first consider the evolutionary vector fields @xmath1 whose generating sections @xmath2\bigr)$ ] belong to the images of matrix linear operators @xmath3 , @xmath4 , @xmath5 in total derivatives ( and those derivatives are equal to @xmath6 ) .the three above - mentioned classes of geometries are regular sources of collections of such operators .we impose the berends burgers van dam hypothesis @xcite : namely , we let the images of the operators be subject to the collective commutation closure . this involutive setup gives rise to the bi - differential structure constants @xcite and bi - differential analogs of christoffel symbols @xcite . for any number @xmath7 of the operators @xmath3 , @xmath4 , @xmath5 , we reduce the setup at hand to the variational lie algebroids @xcite with one variational anchor . having endowed the new bundle geometry with the cartan connection , we reverse the parity of the fibres in it by brute force and then construct the odd , parity - reversing evolutionary vector field @xmath8 that encodes the entire initial geometry via the homological equality @xmath0 . in section [ secexamples ]we describe three natural examples of such geometries : the variational poisson algebroids @xcite , 2d toda chains viewed as variational lie algebroids @xcite , and gauge algebroids @xcite .first , in section [ secham ] we recall the construction of necessary superbundles using the notion of the variational cotangent bundle @xcite and derive the representation via @xmath0 for the variational poisson structures @xcite . as usual, the renowned kdv equation @xcite offers us the minimally possible nontrivial illustration .the odd vector fields @xmath8 are themselves hamiltonian and we write the corresponding @xmath9-charges explicitly . by exploiting the profound relation between the variational poisson geometry and 2d toda chains @xcite, we obtain the evolutionary fields @xmath8 for these models of liouville type @xcite .we emphasize that the differentials @xmath8 and their cohomology carry much more information than the ordinary chevalley eilenberg differentials for the semi - simple lie algebras at hand . being obtained from the gauge - invariant yang mills equations under a symmetry reduction @xcite , the 2d toda chains represent the vast class of nonlinear euler lagrange systems of liouville type @xcite .this class is very interesting by itself ; moreover , the construction of the homological vector fields @xmath8 in section [ sectoda ] remains valid uniformly for all such equations .finally , in section [ secgauge ] we expose the true geometric nature of the brst - differentials @xcite or the ` longitudinal ' components of the bv - differentials @xcite for gauge systems . in agreement with the secondnoether theorem , the differential operators @xmath3 , @xmath4 , @xmath5 which we study here emerge from the noether relations between the equations of motion ; we explain why the same approach of @xmath0 can successfully grasp more subtle geometries without any further modifications .the gauge parameters ( that is , the arguments of the operators @xmath10 ) , being the parity - reversed _neighbours _ of the _ ghosts _ , together with the _ antifields _ ( which will have appeared before in section [ secham ] ) and the newly introduced _antighosts _ constitute the classical bv - zoo .similarly to the variational poisson case , the evolutionary bv - differentials @xmath11 , which incorporate the odd vector fields @xmath8 , are then determined by the schouten bracket of the bv - charges with the inhabitants of the zoo . in the appendix , we report the results of a direct search for the linear differential operators with involutive images ; their coordinate - free understanding as the variational anchors in variational lie algebroids is yet to be achieved .we conclude that the presence of the evolutionary differentials @xmath8 is not specific to either the variational poisson or gauge systems alone : indeed , it is an immanent property of a much wider class of models .the study of the arising cohomology theories not only sheds new light on each system or structure in particular but also reveals far - reaching links between them .the geometry of jet fibre bundles is an established and well developed domain @xcite .any deviations between the notations from different sources are easily tractable ; we follow the notation of @xcite and adopt dirac s convention on bra- covectors @xmath12 and -ket vectors @xmath13 , fixing the orientation @xmath14 on the cotangent bundles . by default , we pronounce the standard incantation about the ground field @xmath15 , the infinite smoothness of all mappings at hand , the locality along the oriented base @xmath16 of the bundles ( whence emerge the jets of sections of _ vector _bundles @xmath17 instead of the jet spaces @xmath18 for maps of manifolds ) , and the locality with respect to the jet order ( whence the filtration in @xmath19 ) .we assume that all total differential operators are local , i.e. , polynomial in the total derivatives . by convention , the symmetries are infinitesimal and we identify the evolutionary vector fields @xmath20 with their generating sections @xmath21 . for the sake of transparency, we consider the purely even , non - graded initial geometry of commutative vector spaces , manifolds , and bundles ; the odd objects which we deal with appear later , through the explicit use of the parity reversion @xmath22 .our reasonings are independent of the choice of local coordinates .we denote the independent variables by @xmath23 , whereas the notation @xmath24 runs through all the dependent variables such as the gauge fields ( if any in the model under study ) at the points of space - time or say , the realizations of the `` string '' @xmath16 in the space - time @xmath25 .the notation is multi - faceted : in each class of examples ( see section [ secexamples ] ) , we shall recall the common specific notation so that , on one hand , an object in this review may acquire several interpretations in different contexts , while on the other hand , we shall of course meet with the habitual ghosts , antighosts , and antifields .we work with the infinite jet bundles from the very beginning .let us remember that many elementary constructions on usual manifolds first of all , the leibniz rule in the commutator @xmath26 $ ] and the schouten bracket @xmath27\!]$ ] or in the poisson brackets under the multiplication of the arguments by smooth functions either entirely disappear over the jet spaces or ( seldom ) survive after a careful consideration and proper amendments .furthermore , we deal with the variational poisson structures @xcite over the spaces @xmath28 of jets of sections for vector bundles @xmath17 rather than suppose that the target space @xmath29 for @xmath18 alone is a poisson manifold ( c.f .the variational poisson theory we exploit does not amount to the canonical formalism which one applies to a lagrangian system after taking its legendre transformation ( except for section [ sectoda ] where we do use that approach as well ) , c.f. @xcite . for completeness, we explicitly endow the horizontal infinite jet bundles @xcite over the spaces @xmath28 with the cartan connection .when introducing the modules of sections of the horizontal bundles such as the spaces of the gauge parameters , never we attempt to treat these sections as the `` collections of functions '' or use any other , equally weak interpretation of their structure because it does not grasp properly the geometry of the building blocks instead , this shadows it with irrelevant fragmentations .indeed , let us remember that , under a differential change of the coordinates @xmath30 and @xmath24 , the genuine reparametrization rules for those sections are imperatively prescribed by their geometric nature ( see @xcite for detail and examples ) .we thus ought to be careful with the @xmath19-module structure for the spaces of such sections ; this is specific in particular to the variational anchors in variational lie algebroids ( see section [ secinvolutive ] and @xcite ) which are neither the anchors nor lie algebroids for usual manifolds , respectively , unless the base manifold @xmath16 shows itself in space - time as no more than a point , or unless all gauge fields are uniformly constant .this also means that we deal _ ab initio _ with the field - dependent gauge parameters @xmath31\bigr)$ ] ; moreover , their rigorous introduction ( see section [ secgauge ] ) demonstrates that the gauge models must be particularly restrictive in order for these sections @xmath32 to not depend on a part of the variables that encode the points of @xmath28 . in sections [ secinvolutive ] and [ secham ] we work on the empty jet spaces ; later , in sections [ sectoda ] and [ secgauge ] , we do on - shell . we assume the off - shell validity of the berends burgers van dam hypothesis stating the collective commutation closure of the evolutionary fields@xmath33 whose generating sections @xmath34 belong to the images of linear differential operators @xmath10 .this may require us to quotient out the trivial symmetries that vanish on - shell .we recall that the euler lagrange systems @xmath35 contain as many equations as there are unknowns in it .( in contrast with @xcite , we do not assume that the density of the action functional @xmath36 is a differential polynomial : indeed , see section [ sectoda ] . )we point out also that _ usually _ the equations , either in evolutionary systems or in the euler lagrange systems , are enumerated ( more precisely , labelled ) by the respective unknowns @xmath37 which explicitly occur in the left - hand sides .however , we emphasize that the admissible reparametrizations of the fields @xmath24 and of the equations @xmath38 are entirely unrelated .this produces its due effect on the transcription of the infinitesimal symmetries @xmath39\bigr)$ ] in the former case and , in the latter , on the objects that lie in or are dual to the horizontal module @xmath40 of the equations , e.g. , on the antifields @xmath41 ( thus , more appropriately denoted by @xmath42 ) , the generating sections @xmath43 of conservation laws ( for the euler lagrange systems , emerging from the noether symmetries @xmath44 by the first noether theorem ) , or the objects which obey the second noether theorem such as the ghosts @xmath45 and the antighosts @xmath46 .the 2d toda chains in section [ sectoda ] offer a perfect example of such discorrelation between @xmath44 and @xmath43 , which is brought in by force due to a purely sthetic tradition of writing hyperbolic systems with their symbols cast into diagonal shape .this requires a bit of attention ( see @xcite ) : the noether map @xmath47 is not always the identity . in this paper , we understand as the _ noether maps _ not only the linear differential operators @xmath48 but , for @xmath49 , the operators @xmath50 at each @xmath51th generation of the _ syzygies _ @xmath52\in p_i$ ] , i.e. , the equations @xmath53 at @xmath54 , noether s relations @xmath55\equiv0 $ ] at @xmath56 , etc . ; the notation @xmath57 corresponds to the modules of sections dual to @xmath58 . in this sense , the hamiltonian operators @xmath59 and the generators @xmath60 of gauge symmetries are noether operators . on the same footing, we note that the arguments @xmath61 of the linear differential operators @xmath10 which determine the evolutionary vector fields @xmath62 do possess an ambiguity .this is due to several independent reasons and occurs not only because of the on - shell equivalence : for a given equation @xmath63 , we let @xmath64 whenever @xmath65 on @xmath63 . butindeed , the kernels @xmath66 in reducible gauge theories carry the freedom as large as the presence of the free functional parameters in their description if the operator equations @xmath67 admit nontrivial solutions @xmath68 . independently , the differential equations @xmath69\bigr)\bigr)=0 $ ] may have non - empty non - linear spaces of solutions @xmath70 .this shows that the structure constants ( c.f .@xcite and @xcite ) , which are bi - differential operators with respect to the arguments @xmath71 , @xmath4 , @xmath72 of the operators @xmath3 , @xmath4 , @xmath5 , are in fact equivalence classes of mappings .because of this , considerable efforts are required to achieve the canonical normalization of the homological evolutionary vector fields on the jet - bundle extensions of such domains , see @xcite and @xcite .the construction of gauge symmetries , which appear in a class of geometries very particular by themselves , is reviewed in section [ secgauge ] .but let us recall in advance that these symmetries have the form @xmath62 with arbitrary @xmath73\bigr)$ ] . in this context , the class of hyperbolic partial differential equations @xmath74 that admit arbitrary functions @xmath75 and @xmath76 in their symmetries is _ close to but different _ from the renowned class of liouville - type hyperbolic quasilinear systems ( see section [ sectoda ] or @xcite and @xcite ) . summarizing, we always highlight the geometric nature of the structures at hand and describe the links between them , as carefully as possible .the literature which is available on the rapidly developing topic of this review is so vast that , instead of the botany of new remarkable advancements , we now do the sociology of repeatedly practised oversights .let @xmath77 and @xmath78 , @xmath79 , be vector bundles over the base manifold @xmath80 and denote by @xmath81 and @xmath82 the respective projections ( the forgetful maps ) to the base of the spaces of infinite jets of sections for these bundles .consider the _ horizontal modules _@xmath83 of sections of the induced bundles @xmath84 and denote their elements by @xmath85\bigr)$ ] .the main example of such modules is @xmath86 , which is the @xmath19-module of the sections of evolutionary derivations @xmath87 on @xmath28 ; we shall also use the modules @xmath88 of the highest horizontal forms - module structure for the spaces of hamiltonians @xmath89 or lagrangians @xmath90 is practically useless because it does not pass via any reasonable leibniz rule through the variational poisson or schouten brackets . ] @xmath91 and the modules @xmath92 of the _ variational covectors _@xmath43 which are dual .let us add that in this paper we never focus on the concrete choice of that metric and also we do not pay any _ particular _ attention to the upper or lower location of the indices .] to the sections @xmath93 with respect to the coupling @xmath94 that takes values in @xmath88 .( generally speaking , the construction of the bundles @xmath78 can be very involved even in the regular and natural classes of geometries , see section [ sectoda ] and @xcite . ) not accenting it at all times , we shall project to the @xmath95th horizontal cohomology spaces @xmath96 that are determined by the lifting @xmath97 onto @xmath28 of the de rham differential on @xmath16 . by thiswe gain the right to integrate by parts ; we denote by @xmath98 or by @xmath99 $ ] the equivalence classes of differential forms @xmath100 .suppose that @xmath101 are linear total differential operators that take values in the space of evolutionary vector fields for each @xmath79 .that is , the domains of the @xmath102 operators may be different but the target space is common to them all .the hamiltonian operators @xmath103 on empty jet spaces @xmath28 are an example of such structures @xcite .assume further that the images of the @xmath102 operators @xmath10 are subject to the collective commutation closure , @xmath104\subseteq\sum\nolimits_{k=1}^n\text{im}\,a_k.\ ] ] the sum of the images in the right - hand side is not direct because in effect they can overlap @xcite .assumption gives rise to the bi - differential structure constants @xcite via @xmath105 = \sum\limits_{k=1}^n a_k\bigl(\mathbf{c}_{ij}^k(\boldsymbol{p}_i,\boldsymbol{p}_j)\bigr);\ ] ] each @xmath106 is the equivalence class of bi - differential operators ( possibly , with _ _non__constant , field - dependent coefficients ) with both its arguments and its values taken modulo the kernels of the operators @xmath10 , @xmath107 and @xmath108 , respectively .( the kernels can be sufficiently large unless the extra nondegeneracy assumptions are made ; for the sake of brevity , we shall not emphasize the presence of such kernels in _ all _ formulas but rather in few ones , c.f .@xcite . ) using the permutability of the evolutionary vector fields with the total derivatives and taking into account the leibniz rule that always holds for the evolutionary derivations @xmath109 , we obtain the canonical decomposition of the bi - differential structure constants , @xmath110 the symbols @xmath111 absorb the bi - linear action of @xmath106 by total differential operators on the arguments ; the notation is justified @xcite because the terms @xmath111 do indeed transform as the christoffel symbols under the reparametrizations @xmath112 invoked by a differential change of the jet coordinates ( whence the @xmath113 s are linear operators in total derivatives ) .we emphasize that generally speaking , the symbols @xmath111 alone that is , _ without _ the first two standard terms in do not determine any lie algebra structure : the true jacobi identity is given by where the standard terms are taken into account properly .it is likely that the restoration of the standard component in for known bi - differential symbols @xmath111 is the cause of a fierce struggle in the modern double field theory and in the theories which are based on courant - like brackets .we notice further that the bi - differential christoffel symbols introduced in must not be confused with the connections @xmath114 in principal fibre bundles in gauge theories .it is interesting to pursue further this emerging parallel between the geometry over the jet bundles and the known connection - based structures over usual manifolds .we remark that , in general , @xmath115 if @xmath116 , that is , the image of an operator alone may not be involutive even if the operator ( in particular , its domain of definition ) is well defined regardless of the entire collection @xmath3 , @xmath4 , @xmath5 . also , we note that for only one operator ( @xmath117 ) the lie algebra of evolutionary vector fields in its image is , generally , non - abelian , which must not be confused with the abelian gauge theories ( e.g. , maxwell s electrodynamics ) corresponding to one - dimensional lie groups , which are always commutative .we finally notice that the nature of the arguments @xmath61 is firmly fixed by the reparametrization rules @xmath112 .therefore , the isolated components of the sections @xmath61 , which one may be tempted to treat as `` functions '' and by this fully neglect their geometry , may not be well - defined as true _ functions_. we repeat that this primitivization discards much of the information about the setup and shadows the geometry which we have addressed so far .we now reduce , in a regular way , the setup of _ many _ operators and their individual domains to one operator on the large domain @xcite .namely , we take the whitney sum over @xmath28 of the induced bundles @xmath84 , which in down - to - earth terms means that we compose the new `` tall '' sections [ bothtallwide ] @xmath118 obviously , property implies that @xmath119\subseteq\text{im}\,a.\ ] ] by definition , operator is the variational anchor in the variational lie algebroid @xcite ; its total space is the quotient of the sum over @xmath28 of the initial domains for @xmath10 by the kernel @xmath120 .we emphasize that variational lie algebroids are not lie algebroids and the variational anchors are not anchors as described by their traditional definition @xcite for usual manifolds ( here , for the target @xmath29 alone but not @xmath18 ) . the operator @xmath121 transfers the commutator @xmath26 $ ] in the lie algebra of evolutionary vector fields to the lie algebra structure @xmath26_a$ ] on the quotient of its domain by the kernel @xcite .it is now entirely obvious that for a given collection @xmath3 , @xmath4 , @xmath5 of the operators with a _common _ domain ( and not necessarily involutive images , but this does not matter ) the induced bracket @xmath26_a$ ] may not determine a well defined lie structure on a single copy of that common domain ( c.f .@xcite ) : indeed , it is the concatenations @xmath32 of the @xmath102 elements @xmath61 from the same space for which the new bracket actually appears .in fact , it is not the horizontal modules @xmath83 of sections@xmath122\bigr)$ ] of the induced fibre bundles @xmath84 which we really need for the representation of the geometry at hand in terms of the homological vector fields , but it is the product @xmath123 over @xmath16 of the infinite jet spaces for @xmath77 and @xmath124 .this means that we operate with the jet coordinates @xmath125 corresponding to the multi - indices @xmath126 instead of the derivatives @xmath127\bigr)$ ] , whence the linear differential operators @xmath10 become linear vector - functions of the jet variables @xmath125 .( in practice , this often means also that the number of the `` fields '' _ doubles _ at exactly this moment . )the following diagram endows the total space @xmath128 with the cartan connection @xmath129 ( c.f .@xcite ) , @xmath130 this justifies the application of the total derivatives to the variables @xmath32 ( including each @xmath61 alone ) and , from now on , allows us to consider on @xmath128 the evolutionary vector fields with their sections depending ( non)linearly on the fields @xmath24 and the variables @xmath61 .the conversion of the sections @xmath61 to jet variables creates the miracle : the kernel @xmath131 of the operator @xmath121 becomes a linear subspace in @xmath128 for each point of @xmath28 .this is a significant achievement because it allows us to operate with the quotient spaces , which we anticipated when the bi - differential structure constants @xmath106 were introduced in .next , we take the fibres of the vector bundles @xmath78 ( hence , of @xmath132 ) and reverse their parity , @xmath133 , the entire underlying jet space @xmath28 remaining intact .this produces the horizontal infinite jet superbundle @xmath134 with odd fibres , the coordinates there being @xmath135 for @xmath79 and @xmath136 .( in a special class of geometries , see section [ secgauge ] , we shall recognize the variables @xmath137 as the _ ghosts _ , which are denoted usually by @xmath138 . )the operators @xmath10 tautologically extend to @xmath139 and become fibrewise - linear functions in @xmath135 ; the equivalence classes of the bi - differential christoffel symbols @xmath111 are represented by bi - linear functions on that superspace .whatever representatives in the equivalence classes of the bi - differential symbols @xmath111 be taken , the odd - parity evolutionary vector field @xmath140 is homological @xmath141 the proof is obtained immediately by passing from many arguments of many operators to the tall sections of wide operators , see , and then literally repeating the first half of the proof of the main theorem in @xcite .however , we still prefer to prove the theorem straightforwardly ; this will attest that the nature of the variables @xmath137 at different @xmath51 s may be entirely uncorrelated .the halved anticommutator @xmath142=\boldsymbol{q}^2 $ ] of the odd vector field @xmath8 with itself is an evolutionary vector field . in this anticommutator , the coefficient of @xmath143 , which determines the coefficients of @xmath144 at all @xmath145 , is equal to @xmath146 by the definition of the symbols @xmath111 .second , let us consider the jacobi identity for the lie algebra of evolutionary vector fields with the generating sections belonging to the images of the operators @xmath3 , @xmath4 , @xmath5 viewed as the fibrewise - linear functions on the total space of the bundle @xmath147 : @xmath148\notag \\ { } & \quad{}=\sum\limits_{\substack{\circlearrowright \\ ( imn ) } } \sum\limits_k a_k\bigl\ { \partial^{(\boldsymbol{q})}_{a_i(\boldsymbol{p}_i ) } \bigl(\gamma_{mn}^k(\boldsymbol{p}_m,\boldsymbol{p}_n)\bigr ) + \sum\limits_\ell \gamma^k_{i\ell}\bigl(\boldsymbol{p}_i , \gamma^\ell_{mn}(\boldsymbol{p}_m,\boldsymbol{p}_n)\bigr ) \bigr\},\notag\\\intertext{where we have relabelled the indices $ \ell\rightleftarrows k$ in the second sum , } { } & \quad{}=-\sum\limits_{\substack{\circlearrowright \\ ( imn ) } } \sum\limits_k a_k\bigl\{ -\partial^{(\boldsymbol{q})}_{a_i(\boldsymbol{p}_i ) } \bigl(\gamma_{mn}^k(\boldsymbol{p}_m,\boldsymbol{p}_n)\bigr ) + \sum\limits_\ell \gamma^k_{\ell i}\bigl ( \gamma^\ell_{mn}(\boldsymbol{p}_m,\boldsymbol{p}_n ) , \boldsymbol{p}_i\bigr)\bigr\}.\label{jacobigamma}\end{aligned}\ ] ] thirdly , the velocity of each odd variable @xmath149 induced by the ( for convenience , _ not _ halved ) anticommutator @xmath150=2\boldsymbol{q}^2 $ ] is obtained as follows , @xmath151 this completes the proof .we remark that the above construction of the homological vector fields @xmath8 in absence of the equations of motion but with an _ a priori _ given collection @xmath3 , @xmath4 , @xmath5 of generators for the infinitesimal transformations of the model patterns upon the construction of nontrivial gauge theories with the zero action functional and the prescribed nontrivial gauge group , see @xcite .the definition of hamiltonian operators on jet spaces @xcite , which are a class of linear differential operators in total derivatives , reads as follows @xcite .consider a total differential operator @xmath103 that maps variational covectors to evolutionary fields .let @xmath152 , @xmath153 be two hamiltonian functionals , that is , the equivalence classes of two elements in the coupling module @xmath88 of the highest horizontal forms .the operator @xmath121 is _ hamiltonian _ if the bracket @xmath154 defined by the formula @xmath155 is a poisson structure , i.e. , if it is bi - linear ( which comes automatically ) , is skew - symmetric ( whence @xmath156 ; the @xmath157-graded case is more subtle , c.f .@xcite ) , and satisfies the jacobi identity .there are many formulations of convenient criteria for the verification of the latter ( see @xcite and also @xcite ) .it is important that they reveal or are based effectively on the commutation closure for the images of hamiltonian operators . for a hamiltonian operator thus defined on the empty jet space @xmath28, a determined autonomous evolution equation is called _ hamiltonian _ with respect to that given structure if the equation can be cast into the form @xmath158 the left - hand side(s ) of the equation @xmath63 belong to a suitable horizontal module @xmath159 of form @xmath160 for some @xmath161 ; _ usually _ , this module is identified with the module @xmath162 of the velocities .this identification conveniently fills in several small gaps in the count of the base dimensions , the location of the hamiltonian @xmath89 for @xmath63 in the highest horizontal cohomology group but not in the preceding one , and the ( in / ex)clusion of the time - derivatives of @xmath24 in the module @xmath159 of equations .however , we emphasize that the reparametrizations of the unknowns @xmath24 and equations @xmath38 are _ not _ correlated in principle .let us remember that the arguments @xmath32 of the hamiltonian operator for @xmath63 belong to the module @xmath163 which is @xmath94-dual to @xmath159 . while the system @xmath63 remains in evolutionary transcription , the variational derivatives @xmath164 of conserved densities @xmath165 for @xmath63 constitute a linear subspace of @xmath163 , but this correspondence vanishes under more general reparametrizations of the equations . reversing the parity of the sections @xmath32 , we arrive at the following diagram : @xmath166we remark that , under the labelling of the evolution equations @xmath167 by the fields @xmath24 themselves , the _ antifields _ are then , on these grounds , usually denoted by @xmath41 or @xmath168 .but in view of the non - identical correspondence between the fields @xmath24 and the left - hand sides @xmath169 of the equations @xmath170 , the proper notation for the antifields would be @xmath42 ( resp . , @xmath171 ) .not only the images of hamiltonian operators are involutive so that @xmath172= a\bigl([\boldsymbol{p}_1,\boldsymbol{p}_2]_a\bigr),\ ] ] but the brackets @xmath26_a$ ] are calculated effectively for each given @xmath121 , see , e.g. , @xcite .therefore , we have all the data of section [ secinvolutive ] ready at hand : taking the odd jet variables @xmath173 , we obtain the odd evolutionary vector fields @xmath174 which encode the variational poisson structures ( primarily , the jacobi identity , see above ) in terms of the homological condition @xmath0 .the homological vector fields @xmath8 for hamiltonian operators were constructed in the recent paper @xcite by taking the advantage of the known geometric interpretation for @xmath175 as the parity - reversed variational covectors but without references to variational poisson algebroids .namely , this was done by explicitly calculating the induced velocities of the sections @xmath31\bigr)$ ] for _ a priori _ specified velocities @xmath176 of the unknowns @xmath24 ; we shall apply the same technique in section [ secgauge ] .there are many examples when the hamiltonian operators @xmath121 are addressed in the context of the brackets @xmath26_a$ ] and bi - differential symbols @xmath177 .for instance , the notorious kdv equation ( say , upon @xmath178 ) yields ( @xcite , see also the appendix ) @xmath179 which is the minimally possible nontrivial illustration ; we refer to @xcite for the boussinesq system , the kaup boussinesq equation was considered in @xcite ; the hamiltonian operators and the bi - differential christoffel symbols for the drinfeld sokolov equations associated with the root systems of rank two are contained in @xcite .the calculation of variational poisson cohomology groups determined by the evolutionary poisson differentials @xmath8 is a rapidly developing topic , see @xcite .we regret to notice the extent to which the progress in the cohomology theory for variational poisson algebroids is retarded by three decades at the least in comparison with the brst- and bv - cohomology technique for the gauge algebroids ( see section [ secgauge ] ) .the evolutionary vector field @xmath8 in is itself hamiltonian with respect to the even @xmath9-charge @xmath180 and the canonical symplectic structure @xmath181 that stems from the volume form @xmath14 in the fibres over points of @xmath16 in the variational cotangent bundle @xcite to the jet space @xmath28 : @xmath182 we emphasize that this property of the field @xmath8 associated with a hamiltonian operator @xmath121 does not require the target @xmath183 in the initial geometric setup @xmath18 to be a poisson manifold , c.f .@xcite .the condition @xmath184=0 $ ] upon the evolutionary vector field @xmath8 is equivalent to the variational classical master equation @xmath185\!]=0 $ ] upon the variational poisson bi - vector @xmath186 and the variational schouten bracket @xmath187\!]= \left<\overrightarrow{\delta}\boldsymbol{\omega}_1\wedge \overleftarrow{\delta}\boldsymbol{\omega}_2\right>$ ] ( the odd - parity poisson bracket or the _ antibracket _ , see @xcite or @xcite and references therein ) , which in the adopted notation and with dirac s convention reads @xmath188\!]&=\bigl\langle \bigl(\frac{\overrightarrow{\delta}\boldsymbol{\omega}_1}{\delta\boldsymbol{q } } \delta\boldsymbol{q } + \frac{\overrightarrow{\delta}\boldsymbol{\omega}_1}{\delta\boldsymbol{b } } \delta\boldsymbol{b}\bigr ) \wedge \bigl(\delta\boldsymbol{q } \frac{\overleftarrow{\delta}\boldsymbol{\omega}_2}{\delta\boldsymbol{q } } + \delta\boldsymbol{b } \frac{\overleftarrow{\delta}\boldsymbol{\omega}_2}{\delta\boldsymbol{b } } \bigr ) \bigr\rangle \\ { } & = \frac{\overrightarrow{\delta}\boldsymbol{\omega}_1}{\delta\boldsymbol{b } } \langle\delta\boldsymbol{b } , \delta\boldsymbol{q}\rangle \frac{\overleftarrow{\delta}\boldsymbol{\omega}_2}{\delta\boldsymbol{q } } - \frac{\overrightarrow{\delta}\boldsymbol{\omega}_1}{\delta\boldsymbol{q } } \langle\delta\boldsymbol{b } , \delta\boldsymbol{q}\rangle \frac{\overleftarrow{\delta}\boldsymbol{\omega}_2}{\delta\boldsymbol{b } } = \left [ \frac{\overrightarrow{\delta}\boldsymbol{\omega}_1}{\delta\boldsymbol{b}}\cdot \frac{\overleftarrow{\delta}\boldsymbol{\omega}_2}{\delta\boldsymbol{q } } - \frac{\overrightarrow{\delta}\boldsymbol{\omega}_1}{\delta\boldsymbol{q}}\cdot \frac{\overleftarrow{\delta}\boldsymbol{\omega}_2}{\delta\boldsymbol{b}}\right],\end{aligned}\ ] ] where the brackets @xmath189 $ ] mark the cohomology class _ after _ the volume form @xmath190 is fixed in the coupling @xmath94 .the composition of the antibracket calculates the commutator @xmath191=\partial^{(\boldsymbol{q})}_{\varphi_1}(\varphi_2)-\partial^{(\boldsymbol{q})}_{\varphi_2}(\varphi_1)$ ] of the two evolutionary vector fields @xmath192 : namely , we have that @xmath193\!]= \langle\boldsymbol{b},[\varphi_1,\varphi_2]\rangle$ ] .this formula illustrates our earlier warning about the absence of any reasonable leibniz rule for either @xmath26 $ ] or @xmath27\!]$ ] . ]relies on the arrangement of the jet ( super)fibre variables in the coupling - dual pairs of opposite parity : here , we have the fields @xmath24 and the antifields @xmath194 so that ( up to the minus sign in @xmath195 which is due to dirac s convention ) @xmath196\ ! ] } + \partial^{(\boldsymbol{q}^\dagger)}_{\textstyle[\! [ \bar{\boldsymbol{\omega}},\boldsymbol{q}^\dagger]\ ! ] } \qquad \text{so that}\quad \boldsymbol{q}(\xi)=[\![\bar{\boldsymbol{\omega}},\xi]\!]\quad \forall\ \xi.\ ] ] gauge models , which we address in section [ secgauge ] , provide a different geometric interpretation for the antifields and require the introduction of at least one generation of the ghost - antighost pairs. the definition of variational poisson structures admits its further generalization to _ non_-evolutionary systems @xmath63 .while the bi - linearity of the bracket remains intact , its skew - symmetry imposes a nontrivial constraint upon the operator @xmath59 in terms of the linearization of the system @xmath63 at hand , whereas the jacobi identity amounts to the same condition @xmath0 upon the odd vector fields .a significant progress in this direction has been achieved only very recently , see @xcite .do there exist hamiltonian operators @xmath3 , @xmath4 , @xmath5 which are compatible in the sense of , and what then could be the geometry standing behind the bi - differential christoffel symbols ( c.f .@xcite ) ? finally , we recall that the root systems of semi - simple complex lie algebras are a regular source of variational poisson structures for kdv - type models @xcite ; this is achieved by using the 2d toda chains .the hyperbolic two - dimensional non - periodic toda chains @xcite associated with the root systems of semi - simple complex lie algebras @xmath197 of rank @xmath198 are the exponential - nonlinear euler lagrange systems @xmath199 where @xmath200 is the cartan matrix of @xmath197 .the action functionals for the 2d toda chains are expressed in terms of the roots and are non - polynomial , thus exceeding the assumptions of @xcite . also ,it is very instructive to inspect the non - identical correlation between the noether symmetries and the generating functions of conservation laws for these lagrangian equations , see @xcite for detail .the 2d toda chains are the best representatives of the vast class of euler lagrange systems of liouville type ( see @xcite and references therein , also @xcite ) . by definition , the systems of this type possess as many liouville s _ integrals _ @xmath201\bigr)$ ] such that @xmath202 on - shell and similarly , @xmath203\bigr)$ ] satisfying @xmath204 ,as there are unknowns @xmath24 . for example , consider the liouville equation @xmath205 and let @xmath206 so that @xmath207 .we have shown in @xcite that the integrals for this class of euler lagrange equations in fact depend differentially on the momenta @xmath208 which emerge from the kinetic part of the action ; this allows us to cast @xmath63 into the evolutionary form @xmath209 , @xmath210 .the differential orders of the integrals @xmath211 $ ] the concide with the exponents of the lie algebra @xmath197 , see @xcite ; there are constructive procedures to obtain the integrals @xmath212 and @xmath213 ( see @xcite and @xcite illustrated in @xcite ) .in @xcite we proved that the adjoint linearizations of the integrals with respect to the momenta , @xmath214 are linear differential operators with involutive images , see .these operators produce symmetries of the liouville - type system at hand ( in particular , its noether symmetries @xmath215\bigr]/\delta\mathfrak{m}=\square\bigl(\delta\mathcal{h}[w]/\delta w\bigr)$ ] ) . by construction ,each column of the operator @xmath216 stems from the respective integral for @xmath63 .however , the image of a particular column may not itself be closed under commutation : we encounter an example for the root system @xmath217 ( here , for the second , higher - order integral and , respectively , the second column in , see @xcite ) .the geometric nature of the domains of the operators @xmath216 is the most nontrivial in comparison with the other two sample geometries which we discuss in sections [ secham ] and [ secgauge ] of this review . in brief , the arguments @xmath218\bigr]\bigr)$ ] in each such domain belong to the variational cotangent bundle to an equation _ other than _@xmath63 , while the miura substitutions @xmath211 $ ] relate the systems ; we refer to @xcite and @xcite for detail .this reasoning implies that the components of the sections @xmath32 do not exist individually as well - defined `` functions '' and therefore the operators @xmath216 may not be split in separate columns .( let us remember that the operators whose domains appear from the variational __ co__tangent bundles @xcite , like the hamiltonian operators , were referred to in @xcite as the operators of _ second kind_. ) the term `` miura '' is indeed justified in this context @xcite : the brackets @xmath26_\square$ ] on the domains of @xmath216 are calculated by using the correlation @xcite of the entire geometry of the 2d toda chains with the previous case of variational poisson algebroids ( section [ secham ] ) .the correspondence yields the symbols @xmath219 explicitly in terms of the integrals @xmath212 and the roots of @xmath197 ; this calculates the commutation relations in the symmetry algebras for the 2d toda chains . as usual , we reverse the parity of the arguments @xmath32 for the operators @xmath216 by taking @xmath220 and obtain the homological evolutionary vector field @xmath221 the operator @xmath222 for the liouville equation @xmath223 specifies the differential @xmath224 ; the equality of the even velocity @xmath225 of the odd variable @xmath226 to the respective velocity which we obtained in for the korteweg de vries equation is no coincidence @xcite .is it true that for each operator @xmath216 of _ second kind _ andfor the associated differential @xmath8 there is the _operator @xmath121 with the same domain such that @xmath227 ?finally , we recall that the 2d toda chains related to the root systems are obtained by imposing the cylindric symmetry reduction in the yang mills equations , the distinguished gauge models @xcite . in turn , the latter do carry their own homological vector fields ; we discuss them in the next section .however , the yang mills equations are a class of models for which ( unlike , e.g. , gravity ) the bi - differential christoffel symbols standing in the fields @xmath8 are structurally simple .this shows the `` preservation of complexity '' in its genuine sense of logic under the reduction ( large , simple ) @xmath228 ( small , complex ) of the yang mills systems to the 2d toda chains .it is interesting to track the details of the correspondence between the homological vector fields @xmath8 , the compositions of the schouten bracket , and the respective syzygies and charges in the two models and by this , to approach the origin of liouville s integrals @xmath212 along a new direction .this will give us a clue to their analogs in the affine case of @xmath229 .the euler lagrange systems @xmath230 contain as many equations as there are unknowns @xmath24 . by construction ,such equations @xmath231 are _ conveniently _ labelled by the respective fields @xmath37 ; let us remember that the horizontal module @xmath159 of the sections @xmath232 is then @xmath233 .because of this , the generating sections @xmath234 of conservation laws for @xmath63 acquire the nature of noether symmetries @xmath235 , which is indeed the case by virtue of the first noether theorem ( up to , possibly , the non - identical noether maps @xmath236 , as we recalled in section [ sectoda ] ) .although the systems @xmath63 are determined , there may appear the differential constraints ( also called _ syzygies _ , _ noether identities _ , or _ bianchi identities _ ) between the equations of motion , @xmath237\equiv0\quad\forall\ \boldsymbol{q}= \boldsymbol{q}(x),\qquad \phi\in p_1.\ ] ] the relations @xmath238\equiv0 $ ] between the relations , valid identically for all @xmath239 ( here @xmath240 ) give rise to possibly several but finitely many generations of the horizontal modules @xmath241 for @xmath242 . in the sequel , we shall assume that the given action functional @xmath36 determines the system @xmath243 of equations of motion with one generation of the constraints @xmath244=0 $ ] and that there are no further relations between the already known ones .in the following diagram we summarize the notation and therefore not grasp the subtle difference between the transcripts @xmath245 and @xmath246 for the antighosts ( see , e.g. , @xcite ) .we repeat that we do not pay any _ particular _ attention to the upper or lower location of indices . ] and interpret the objects at hand in terms of the bv - theory @xcite : @xmath247\in p_1 & \ltexto & \lefteqn{\text{noether identities}}\phantom{\text{usual identification . } } \\ { } & & & & \boldsymbol{q}^\dagger\in\pi p_0 & \ltexto & \lefteqn{\text{antifields}}\phantom{\text{usual identification . } } \\ { } & & & & \utexto^{\pi } \dtexto & & { } \\ \text{noether symmetries } & \rtexto & \varphi_{\mathcal{l } } \doteq\psi\in\varkappa(\pi)\simeq\widehat{p_0 } & \rbothways^{*}_{\phantom{m}\langle\,,\,\rangle\phantom{m } } & \boldsymbol{f}\in p_0\simeq\widehat{\varkappa(\pi ) } & \ltexto & \text{usual identification . }\end{diagram}\ ] ] each relation @xmath248=0 $ ] between the equations @xmath249 , @xmath250 , yields the linear total differential operator @xmath251 that generates symmetries of the system @xmath63 .these symmetries are parametrized by arbitrary sections @xmath85\bigr)\in\widehat{p_1}$ ] and are called _gauge symmetries_. namely , suppose that the identity @xmath252\bigr]\equiv0 $ ] holds irrespective of a section @xmath253 . consequently , this identity is indifferent to arbitrary infinitesimal shifts , which are given by the @xmath254-vertical evolutionary vector fields @xmath255 on @xmath28 .the chain rule implies that @xmath256\bigr)=\partial^{(\boldsymbol{a})}_\varphi(\psi)$ whenever this right\/-\/hand side is well\/-\/defined for the given~$\boldsymbol{a}$ and~$\psi$ ) , we conclude that } \ell^{(\boldsymbol{f})}_\phi \circ \ell^{(\boldsymbol{q})}_{\boldsymbol{f } } ( \varphi ) & = 0,\\ \intertext{where the composition is a linear mapping from $ \varkappa(\pi)$ to~$p_1$. let us now couple this zero velocity along~$p_1 $ with \emph{any } element~$\boldsymbol{p}\in\widehat{p_1}$ from the dual module and then integrate by parts , staying in the equivalence class of the zero in the cohomology .we obtain } { \bigl(\ell^{(\boldsymbol{q})}_{\boldsymbol{f}}\bigr)}^\dagger \circ { \bigl(\ell^{(\boldsymbol{f})}_{\phi}\bigr)}^\dagger ( \boldsymbol{p } ) & = 0.\end{aligned}\ ] ] but let us recall that the linearization @xmath257 _ coincide _ for the euler lagrange systems @xmath258 due to the helmholz criterion ( see @xcite ) @xmath259 this implies that @xmath260 is , on - shell , a symmetry of the system @xmath63 for any section @xmath261 .moreover , for each noether identity @xmath248\in p_1 $ ] we have constructed explicitly the operator @xmath262 that yields the symmetries @xmath263 of the model .we note that the constraints @xmath264 need not be linear .furthermore , in the cases of a very specific geometry of @xmath63 there may appear the symmetry - producing operators which do not issue from any differential relations between the equations of motion .the approach which we have formulated by so far treats all such structures in a uniform way .however , under the additional assumption that the symmetries obtained in the images of the operators are noether , @xmath265\in\overline{h}^n(\pi)$ ] , the existence of the respective sections @xmath264 is justified easily and besides , these constraints appear to be _actually , suppose that @xmath266 $ ] for all sections @xmath261 .this implies that the noether identity @xmath244=0 $ ] amounts to the linear differential relation @xmath267 between the equations of motion so that @xmath264 coincides with its own linearization .the maxwell equations @xmath268 upon the skew - symmetric field strength tensor @xmath269 satisfy the obvious relation @xmath270 ( that is , @xmath271 ) , which manifests the invariance of the system ( including its action functional ) under ( finite or infinitesimal ) gauge transformations @xmath272)\bigr)$ ] of the electromagnetic field @xmath273 , here @xmath274 is the de rham differential on the minkowski space - time .the conventional plus sign in the transcript of the bianchi identities yields the minus sign in front of the adjoint operator @xmath275 in the gauge symmetries .because the coefficients of the de rham differential , which spreads along the diagonal of the @xmath276-matrix operator @xmath121 , are constant , the bi - differential christoffel symbols @xmath177 vanish in this model .it is remarkable that the yang mills theories with non - abelian gauge groups still do constitute a regular class of examples with the field - independent ( and at most constant , whenever nonzero ) coefficients in the bi - differential christoffel symbols @xmath277 . on the other hand , _ gravity _produces the drastically more involved structure of these symbols with the explicit dependence on the unknown fields in their coefficients @xcite .this is immanent also to the liouville - type systems ( see section [ sectoda ] ) .let the operators @xmath3 , @xmath4 , @xmath278 be the entire collection of the gauge symmetry generators for the system @xmath63 .for the sake of clarity , we always assume the off - shell validity of the berends burgers van dam hypothesis @xcite about the collective commutation closure for the images of these operators , see .then the odd evolutionary vector field is the becchi rouet stora tyutin differential @xcite ; in the model at hand , _ all _ parity - reversed arguments @xmath137 belong to one module @xmath279 and are known as the _ ghosts _ , usually denoted by @xmath138 .the variational lie algebroid encoded by the evolutionary differential @xmath8 was named the _ gauge algebroid _ in @xcite . as in the variational poisson case( see section [ secham ] ) , we arrange the variables in pairs of opposite parity : even fields @xmath24 @xmath280 odd _ antifields _@xmath41 , and odd ghosts @xmath138 @xmath280 even _ antighosts _ @xmath281 .the evolutionary brst - field @xmath8 lifts to the odd derivation @xmath282 on the horizontal infinite jet superspace that now contains the antifield coordinates @xmath283 and similarly , the antighosts @xmath284 , here @xmath136 .the induced velocities of the newly incorporated fibre variables are calculated in agreement with their geometric nature .first , we recall that the off - shell determining condition @xmath285 for @xmath286 to be an infinitesimal symmetry of the equation @xmath38 states the existence of the linear total differential operator @xmath287 such that the right - hand side of the condition vanishes on - shell ( i.e. , by virtue of the equation and its differential consequences ) .let us remember also that the operator @xmath288 depends linearly on @xmath286 because the entire equality does so .specifically for the gauge geometry at hand , this means that @xmath289 or , after the parity reversion @xmath290 of the linear entry @xmath169 in both sides of the equality , we obtain the odd - parity component @xmath291 of the lifting @xmath282 for @xmath8 .second , we see that due to the chain rule the odd velocity of a noether relation @xmath244\in p_1 $ ] , or of the even antighost @xmath46 associated with it , is equal to @xmath292 the proper antifield - antighost lifting @xmath282 of the brst - differential @xmath8 reads @xmath293 the full batalin vilkovisky differential @xmath294 merges the parity - reversing ( and besides , vanishing on - shell ) koszul tate differential , @xmath295 with the lifting @xmath282 of @xmath8 ; however , @xmath11 may contain further terms of the correction that ensures the off - shell equality @xmath296 .let us now look at the problem of construction of the bv - differential from another perspective .the pairwise arrangement @xmath297 , @xmath298 for the @xmath94-dual variables of opposite parity prescribes the composition of the variational schouten bracket @xmath187\!]= \bigl\langle\overrightarrow{\delta}\boldsymbol{\omega}_1\wedge \overleftarrow{\delta}\boldsymbol{\omega}_2\bigr\rangle $ ] ( @xcite and , e.g. , @xcite ) , now extending to all generations of the ghost - antighost pairs : @xmath187\!]$ ] @xmath299 + \left[\frac{\overrightarrow{\delta}\boldsymbol{\omega}_1}{\delta\boldsymbol{\gamma}}\cdot \frac{\overleftarrow{\delta}\boldsymbol{\omega}_2}{\delta\boldsymbol{\gamma}^\dagger } - \frac{\overrightarrow{\delta}\boldsymbol{\omega}_1}{\delta\boldsymbol{\gamma}^\dagger}\cdot \frac{\overleftarrow{\delta}\boldsymbol{\omega}_2}{\delta\boldsymbol{\gamma } } \right].\end{gathered}\ ] ] in this notation , ] ] the bv - differential @xmath11 is determined ( @xcite , see also section [ secham ] ) by the variational schouten bracket of the bv - action @xmath300 and the respective super - jet variables , @xmath301\!]}+ \partial^{(\boldsymbol{q}^\dagger)}_{\textstyle[\ ! [ \bar{\boldsymbol{s}},\boldsymbol{q}^\dagger]\!]}+ \partial^{(\boldsymbol{\gamma})}_{\textstyle[\ ! [ \bar{\boldsymbol{s}},\boldsymbol{\gamma}]\!]}+ \partial^{(\boldsymbol{\gamma}^\dagger)}_{\textstyle[\![\bar{\boldsymbol{s}},\boldsymbol{\gamma}^\dagger]\ ! ] } , \qquad \text{i.e.,}\quad \boldsymbol{d}(\xi)=[\![\bar{\boldsymbol{s}},\xi]\!],\ ] ] where the extended action functional @xmath302 has even parity and @xmath303 equals @xcite , with the conventional einstein s summation over repeated indices and dirac s ordering of ( co)vectors , @xmath304 the possible necessity to introduce the correction terms of higher polynomial orders in @xmath45 , @xmath46 , or @xmath41 is legitimate , e.g. , when the coefficients of the bi - differential symbols @xmath111 depend explicitly on the fields @xmath24 and hence produce the redundant terms in the velocity of @xmath41 , which thus must be cancelled out properly .the renowned paper @xcite ( c.f .@xcite ) contains a perfect example of a gauge model with one generator @xmath121 of its gauge symmetries and an explicit calculation of the ( zero - order ) bi - differential christoffel symbols @xmath177 , brst- and bv - differentials , and the master - action @xmath303 , see @xcite where the parity reversion @xmath22 is the ` promotion ' and the brst - field @xmath305 is denoted by @xmath306 .the cohomological condition @xmath307=0 $ ] is equivalent to the classical master equation @xmath308\!]=0 $ ] . the bv - cohomology with respect to the differential @xmath11 ( or @xmath309\!]$ ] ) provides the resolvent for the algebra @xmath310 of _ observables _ in the model and opens a way for the quantization of gauge systems .describe the variational poisson structures that are pertinent to gauge - invariant systems .the most recent techniques of @xcite allow or soon will allow the regular search for variational poisson structures on non - evolutionary gauge - invariant models by using the respresentation of the structures via @xmath0 .it must be noted however , that the gauge algebroids are endowed _ a priori _ with the evolutionary differentials which suit well for the quantization of the systems .therefore , the variational poisson structures whenever found for a given gauge model will either provide only a few of its symmetries or , if there is a functional freedom in the domains of the hamiltonian operators , will definitely _ not _ provide the entire lie algebra of infinitesimal gauge transformations because of the apparent difference between the domains of definition for the hamiltonian operators and ordinary gauge generators ( compare the diagrams in this section and in section [ secham ] ) .this confirms that the bi - hamiltonian _ complete integrability _ paradigm ( @xcite , see @xcite ) , which relies on the construction of the poisson pencils for hierarchies of evolutionary systems , becomes insufficient in the geometry of gauge fields .the realization of nonlinear systems , or parts of the information which they carry , in terms of homological evolutionary vector fields @xmath8 is not specific to gauge - invariant models only . indeed ,other natural classes of partial differential equations can be addressed with convenience from the same viewpoint . moreover , even for the already known gauge systems the available approach is ready to capture more of their geometry without any extra modifications . for a long time , the physical and mathematical incarnations of the notion of lie algebroids over infinite jet spaces developed in parallel , the progress on the math side being seriously retarded with respect to the demands from physics .seemingly , the exchange between the two theories , foresighted in @xcite , channeled through the construction of the _ variational _ schouten bracket .nowadays , the systematic and well - motivated calculation of the variational poisson cohomology only begins while the brst- and bv - cohomology theories are an established domain of research and the subject of a vastest literature . in this review we traced the links between the variational poisson algebroids , gauge algebroids , and playing the rle of mediators 2d toda - like systems also viewed as variational lie algebroids .the author is grateful to the organizing committee of 7th international workshop ` quantum theory and symmetries7 ' ( august 713 , 2011 ; cvut prague , czech republic ) for a welcome and warm atmosphere during the meeting .the author thanks g. barnich , b. a. dubrovin , g. felder , e. a. ivanov , s. o. krivonos , j. w. van de leur , m. schlichenmaier , m. a. semenov tian - shansky , and m. a. vasiliev for helpful discussions and stimulating remarks . also , the author thanks j. krasilshchik and a. verbovetsky for constructive criticisms and interesting correspondence .this research was supported in part by nwo veni grant 639.031.623 ( utrecht ) and jbi rug project 103511 ( groningen ) .a part of this research was done while the author was visiting at the @xmath311 ( bures - sur - yvette ) ; the financial support and hospitality of this institution are gratefully acknowledged .we now list several operators which we obtained in a fixed system of local coordinates and whose images are then closed under commutation .we denote by @xmath312 the field and , for brevity , we use the notation @xmath313 instead of @xmath314 .let us fix the weights @xmath315 , @xmath316 that originate from the scaling invariance of the kdv equation @xmath317 ; we have that @xmath318 . using the method of undetermined coefficients, we performed the search for scalar operators that satisfy and which are homogeneous with respect to the weights not greater than @xmath319 .we obtained two compatible hamiltonian operators @xmath320 and @xmath321 , the odd powers @xmath322 of @xmath323 , and the hamiltonian operator @xmath324also , there are four non - skew - adjoint operators with involutive images , @xmath325@xmath326 finally , we have found the operators that contain arbitrary functions : @xmath327 and @xmath328 with vanishing christoffel symbols @xmath329 , and also @xmath330 the transformation rules for the sections that constitute the domains of the operators @xmath331 , @xmath332 , @xmath333 , @xmath4 , @xmath334 are yet to be found .likewise , the interpretation as variational anchors and the understanding of the physics of the models remain an open problem for the infinite class of linear total differential operators with involutive images which were found in @xcite .77 _ alexandrov m. , schwarz a. , zaboronsky o. , kontsevich m. _ ( 1997 ) the geometry of the master equation and topological quantum field theory , _ int .j. modern phys . _ * a12*:7 , 14051429 ._ barnich g. _ ( 2010 ) a note on gauge systems from the point of view of lie algebroids , aip conf .proc . * 1307 * xxix workshop on geometric methods in physics ( june 27 july 3 , 2010 ; biaowiea , poland ) , 718 . `arxiv : math - ph/1010.0899 ` _ barnich g. , henneaux m. _ ( 1996 ) isomorphisms between the batalin vilkovisky antibracket and the poisson bracket , _ j. math .phys . _ * 37*:11 , 52735296 ._ batalin i. a. , fradkin e. s. , fradkina t. e. _ ( 1990 ) generalized canonical quantization of dynamical systems with constraints and curved phase space , _ nucl .phys . _ * b332*:3 , 723736 ._ batalin i. , vilkovisky g. _ ( 1981 ) gauge algebra and quantization , _ phys .lett . _ * b102*:1 , 2731 ; _ batalin i. a. , vilkovisky g. a. _ ( 1983 ) quantization of gauge theories with linearly dependent generators , _ phys . rev . _ * d29*:10 , 25672582 . _becchi c. , rouet a. , stora r. _ ( 1976 ) renormalization of gauge theories , _ ann .* 98*:2 , 287321 ; _ tyutin i. v. _ ( 1975 ) gauge invariance in field theory and statistical mechanics , _ preprint _ lebedev fian no. 39 ._ belavin a. a. _ ( 1989 ) kdv - type equations and @xmath9-algebras , in : _ integrable systems in quantum field theory and statistical mechanics_. _ adv .stud . pure math ._ * 19 * , acad . press , boston , ma , 117125 ._ crainic m. , fernandes r. l. _ ( 2003 ) integrability of lie brackets , _ ann .( 2)*157*:2 , 575620 ._ de sole a. , kac v. g. _ ( 2011 ) the variational poisson cohomology , 130 p. _ preprint _ ` arxiv : math - ph/1106.0082 ` _ de sole a. , kac v. g. _ ( 2011 ) essential variational poisson cohomology , 30 p. _ preprint _ ` arxiv : math - ph/1106.5882 ` _ drinfeld v. g. , sokolov v. v. _ ( 1985 ) lie algebras and equations of korteweg de vries type , _ j. sov . math . _ * 30 * , 19752035 . _ dubrovin b. a. _ ( 1996 )geometry of 2d topological field theories , _ lect .notes in math ._ * 1620 * integrable systems and quantum groups ( montecatini terme , 1993 ) , springer , berlin , 120348 ._ dubrovin b. a. , novikov s. p. _ ( 1983 ) hamiltonian formalism of one - dimensional systems of the hydrodynamic type and the bogolyubov whitham averaging method , _ sov .* 27*:3 , 665669 ; _ dubrovin b. a. , novikov s. p. _ ( 1984 ) poisson brackets of hydrodynamic type , _ sov .dokl . _ * 30*:3 , 651654 . _fradkin e. s. , vasiliev m. a. _ ( 1976 ) model of supergravity with minimal electromagnetic interaction ._ preprint _ lebedev fian-76 - 197 , 23 p. _ fradkin e. s. , vilkovisky g. a. _ ( 1975 ) quantization of relativistic systems with constraints , _ phys .lett . _ * b55*:2 , 224226 ._ fulp r. , lada t. , stasheff j. _ ( 2003 ) noether s variational theorem ii and the bv formalism .proc . 22nd winter school `` geometry and physics '' ( srn , 2002 ) , _ rend .circ . mat .palermo _ ( 2 ) suppl .71 , 115126 . _gelfand i. m. , dorfman i. ja . _ ( 1981 ) schouten bracket and hamiltonian operators , _ functional anal .appl . _ * 14*:3 , 223226 ._ getzler e. _ ( 2002 ) a darboux theorem for hamiltonian operators in the formal calculus of variations , _ duke math .j. _ * 111*:3 , 535560 ._ gitman d. m. , tyutin i. v. _ ( 1990 ) _ quantization of fields with constraints_. springer ser .nucl . part ., springer - verlag , berlin ._ golovko v. a. _ ( 2011 ) _ variational poisson nijenhuis structures and integrable hamiltonian systems _ ( in russian ) .phd thesis , lomonosov msu , moscow , 189 pp ._ golovko v. a. , krasilshchik i. s. , verbovetsky a. m. _ ( 2008 ) variational poisson nijenhuis structures for partial differential equations , _ theor . math .phys . _ * 154*:2 , 227239 ._ gomis j. , pars j. , samuel s. _( 1995 ) antibracket , antifields and gauge - theory quantization , _ phys . rep . _ * 259*:1 - 2 , 1145 ._ henneaux m. , teitelboim c. _ ( 1992 ) _ quantization of gauge systems_. princeton university press , princeton , nj . _ henneaux m. _ ( 1985 ) hamiltonian form of the path integral for theories with a gauge freedom , _ phys ._ * 126*:1 , 166 ._ hussin v. , kiselev a. v. _ ( 2011 ) a convenient criterion under which @xmath157-graded operators are hamiltonian , _ j. phys .ser . _ * 284 * mathematical and physical aspects of symmetry .28th int .colloquium on group - theoretical methods in physics ( july 2630 , 2010 ; northumbria university , newcastle , uk ) , 012035 , 6 p. ` arxiv : nlin.si/1103.2542 ` _ igonin s. , verbovetsky a. , vitolo r. _ ( 2004 ) variational multivectors and brackets in the geometry of jet spaces , _ symmetry in nonlinear mathematical physics ( kiev , 2003)_. _ pr .inst . mat .nauk ukr . mat ._ * 50 * , 13351342 ._ kiselev a. v. , van de leur j. w. _ ( 2011 ) involutive distributions of operator - valued evolutionary vector fields and their affine geometry , proc .5th int . workshop ` group analysis of differential equations and integrable systems ' ( june 610 , 2010 ; protaras , cyprus ) , 99109 . `arxiv : math - ph/0904.1555 ` _ kiselev a. v. , van de leur j. w. _ ( 2009 ) a geometric derivation of kdv - type hierarchies from root systems , in : proc .4th int .workshop ` group analysis of differential equations and integrable systems ' ( october 2630 , 2008 ; protaras , cyprus ) , 87106 . `arxiv : nlin.si/0901.4866 ` _ kiselev a. v. , van de leur j. w. _ ( 2009 ) a family of second lie algebra structures for symmetries of dispersionless boussinesq system ,_ j. phys . a math .* 42*:40 , 404011 ( 8 p. ) ` arxiv : nlin.si/0903.1214 ` _ krasilshchik i. , verbovetsky a. _ ( 2011 ) geometry of jet spaces and integrable systems , _ j. geom .phys . _ * 61 * , 16331674 . `arxiv : math.dg/1002.0077 ` , _ krasilshchik i. s. , verbovetsky a. m. , vitolo r. _ ( 2011 ) a unified approach to computation of integrable structures , _ preprint _` arxiv : nlin.si/1110.4560 ` , 18 pp . _ krasilshchik i. s. , vinogradov a. m. _ , eds .( 1999 ) _ symmetries and conservation laws for differential equations of mathematical physics_. ( bocharov a. v. , chetverikov v. n. , duzhin s. v. _ et al_. ) ams , providence , ri ._ kumpera a. , spencer d. _ ( 1972 ) _ lie equations .i : general theory ._ annals of math . stud . * 73*. princeton university press , princeton , nj ; university of tokyo press , tokyo ._ kupershmidt b. a. _ ( 1980 ) geometry of jet bundles and the structure of lagrangian and hamiltonian formalisms ._ geometric methods in mathematical physics _nsf cbms conf .lowell , mass . , 1979 ) , lecture notes in math .* 775 * , springer , berlin , 162218 ._ leznov a. n. , saveliev m. v. _ ( 1979 ) representation of zero curvature for the system of nonlinear partial differential equations @xmath336 and its integrability , _ lett .phys . _ * 3 * , 489494 ; _ leznov a. n. , saveliev m. v. _ ( 1992 ) group - theoretical methods for integration of nonlinear dynamical systems .progr . in phys .* 15*. birkhuser verlag , basel ._ mackenzie k. _ ( 1987 ) _ lie groupoids and lie algebroids in differential geometry ._ london math .* 124*. cup , cambridge ._ magri f. _ ( 1978 ) a simple model of the integrable equation , _ j. math .phys . _ * 19*:5 , 11561162 ._ nutku y. _ ( 2001 ) lagrangian approach to integrable systems yields new symplectic structures for kdv , in : _ integrable hierarchies and modern physical theories _ ( h. aratyn and a. s. sorin , eds ) proc .advanced research workshop ( arw - uic 2000 ) ( chicago , il , july 22 - 26 , 2000 ) , nato sci . ser .ii math .phys . chem .* 18 * , kluwer acad .publ . , dordrecht , 203213 ._ olver p. j. _ ( 1993 ) _ applications of lie groups to differential equations _ , grad .texts in math . *107 * ( 2nd ed . ) , springer verlag , ny . _ pradines j. _ ( 1967 ) thorie de lie pour les groupodes diffrentiables .calcul diffrenetiel dans la catgorie des groupodes infinitsimaux , _ c. r. acad .paris _ sr .a - b * 264 * , a245a248 . _saunders d. j. _ ( 1989 ) _ the geometry of jet bundles . _ london math .note ser .* 142*. cup , cambridge ._ saveliev m. v. _ ( 1993 ) on the integrability problem of a continuous toda system , _ theor .* 92*:3 , 10241031 ._ schwarz a. _ ( 1993 ) geometry of batalin vilkovisky quantization , _ commun ._ * 155*:2 , 249260 ._ schwarz a. s. _ ( 1979 ) the partition function of a degenerate functional , _ commun . math .phys . _ * 67*:1 , 116 ._ shabat a. b. _ ( 1995 ) higher symmetries of two - dimensional lattices , _ phys .lett . _ * a200*:2 , 121133 ; _ shabat a. b. _ ( 1996 ) first integrals of the infinite toda lattice . _ symmetries and integrability of difference equations _( d. levi , l. vinet and p. winternitz , eds . ; estrel , pq , 1994 ) , crm proc .lecture notes * 9 * , ams , providence , ri , 345351 ._ vaintrob a. yu . _ ( 1997 ) lie algebroids and homological vector fields , _ russ .* 52*:2 , 428429 ._ vinogradov a. m. _ ( 1984 ) the @xmath337-spectral sequence , lagrangian formalism , and conservation laws . i. the linear theory , ii . the nonlinear theory , _ j. math .appl . _ * 100*:1 , 140 , 41129 ._ zhiber a. v. , sokolov v. v. _ ( 2001 ) exactly integrable hyperbolic equations of liouvillean type , _ russ .math . surveys _ * 56*:1 , 61101 ._ zinn - justin j. _ ( 1975 ) renormalization of gauge theories ._ trends in elementary particle theory _ ( lect . notes in phys . * 37 * h. rollnick and k. dietz eds ) , springer , berlin , 239 ._ zinn - justin j. _ ( 1976 ) _ mthodes en thorie des champs / methods in field theory . _( cole dt de physique thorique , session xxviii , tenue les houches , 28 juillet6 septembre , 1975 ; r. balian and j. zinn - justin , eds ) north - holland publ .co. , amsterdam etc .

<S> we review the construction of homological evolutionary vector fields on infinite jet spaces and partial differential equations . </S><S> we describe the applications of this concept in three tightly inter - related domains : the variational poisson formalism ( e.g. , for equations of korteweg de vries type ) , geometry of liouville - type hyperbolic systems ( including the 2d toda chains ) , and euler lagrange gauge theories ( such as the yang mills theories , gravity , or the poisson sigma - models ) . </S><S> also , we formulate several open problems . </S>

0705.1611

the hubbard model ( hm)@xcite is one of the simplest nontrivial models of interacting spin-1/2 electrons on a lattice .its hamiltonian reads @xmath0 where @xmath1 and @xmath2 are creation and annihilation operators for electrons with spin @xmath3 at site @xmath4 respectively , @xmath5 , @xmath6 is the hoping integral , and @xmath7 denotes the strength of on - site interaction . in one dimension ( 1d )the hm can be solved exactly by the bethe - ansatz method.@xcite the wave function and the energy of @xmath8 electrons on a chain with @xmath9 sites can be written in terms of @xmath10 pseudo - momentum variables and @xmath11 spin rapidities . although the energy spectra have been known for many years , the calculation of the correlation functions proved to be a delicate problem .@xcite the numerical evaluations of the correlation functions @xcite and the analytic results @xcite indicated clearly that the 1d hm is a tomonaga - luttinger liquid ( tll).@xcite assuming that the 1d hm is tll , it then becomes possible to calculate the correlation functions from the knowledge of the energy spectra @xcite . using this procedure schulz@xcite studied the correlation - function exponents for different @xmath7 and band filling @xmath12 .it is also shown@xcite that the large scale behavior of the spin and charge degrees of freedom can be described by two decoupled boson field theories with dynamics governed by the tll hamiltonian in the small and large @xmath7 regimes .another nontrivial model is the falicov - kimball model @xcite which consists of localized ions and itinerant spinless fermions .the hamiltonian of the falicov - kimball model reads @xmath13 where @xmath14 are creation operators for spinless fermions , and the configuration @xmath15 denote spatial distribution of ions .clearly , the fkm can be viewed as a modification of the hm in the sense that the one kind of fermions , such as down - spin fermions , has infinite mass , and hence does not move . nevertheless , the physics of the fkm is completely different . in the neutral case where each particle concentration equals @xmath16 , it was @xcite proved that the system always orders in an alternating `` chessboard '' phase at a finite transition temperature in all dimensions greater than 1. this ordered phase can be interpreted as the transition from a high - temperature homogeneous ( liquid / gas ) phase to a low temperature ordered ( solid ) phase .freericks@xcite showed that the model ( on a hypercubic lattice ) also displayed incommensurate order , segregation or phase separation .the 1d case of the fkm has also been extensively studied .since there is no finite - temperature phase transition , the system can have phase transition in the ground state .the numerical solutions @xcite produced a conjecture for the case @xmath17 with @xmath18 and the screened coulomb interaction @xmath7 is large enough , the system will segregate into an empty lattice ( with no ions and all the electrons ) , and a full lattice ( with all the ions and no electrons ) . this conjecture was later proven to be true by lemberger @xcite . for any dimensional hfm , freericks _ etal _@xcite gave a theorem that the strong correlation can leads to ps .the relation between the hm and the fkm is straightforward . in order to have a unified framework ,the asymmetric hubbard model ( ahm ) has been introduced naturally .@xcite its hamiltonian reads @xmath19 where @xmath20 is @xmath3-dependent hoping integral . clearly , if @xmath21 , the ahm becomes the hm , and if @xmath22 , it becomes the fkm .the hamiltonian ( [ eq : hamiltonian_ahm ] ) has u(1)@xmath23u(1 ) symmetry for general @xmath20 , and the electron number @xmath24 are conserved respectively . in the condensed matter physics ,the ahm is believed to describe many physical phenomena , such as superconductor , valence fluctuating , and heavy fermions.@xcite in the recent development of the optical lattice , it has been pointed out that the ahm can be used to describe a mixture of two species of fermionic atoms in optical lattices .@xcite according to the fact that the ground states of the hamiltonian ( [ eq : hamiltonian_ahm ] ) in its two limiting cases : the hm ( @xmath25 ) and the fkm ( @xmath26 ) belong to two different universality classes , a quantum phase transition was suggested to happen in the phase diagram defined in the @xmath27 plane.@xcite nevertheless , the quantitative phase diagram for the case away from half - filling had never been obtained until a recent work by gu , _ et al _ @xcite . in their work , the quantum entanglement @xcite between a local part and the rest of the system , and the structure factor of charge - density - wave ( cdw ) for down - spin electronsare used to identify the critical point . here, we are going to study the ground - state phase diagram of the ahm away from the half filling with the strategy of bosonization method . @xcitediffer from the numerical approach @xcite which captures the physics from the finite - size analysis for small systems , our work aim to give definitely analytical results from the point of view of field theory .the paper is organized as follows . in section [ sec : bos ] , we derive the bosonized form of the 1d ahm , and clarify the role of the some terms in the hamiltonian . in section [ sec : res ] , we first diagonalize the effective hamiltonian in which some irrelevant terms are ignored , then obtain the instability conditions for the phase separation and compare them with the numerical results of a finite sample. we also obtain the analytical expressions for the correlation functions of charge - density - wave , spin - density - wave ( sdw ) , singlet - superconductivity ( ss ) , triplet - superconductivity ( ts ) fluctuations , as well as the corresponding exponents .finally , a brief summary and acknowledgement are given in section [ sec : sum ] .the convenient way to analyze the 1d ahm is to bosonize the fermi operators and convert them to a quantum theory of two bose fields @xcite . in the framework of the standard bosonization method the hmare expressed in terms of canonical bose fields and their dual counterparts as , @xmath28 \\ & & + \frac{v_{s}}{2}\int dx\left [ \frac{1}{k_{s}}\left ( \partial _ { x}\phi _ { s}\right ) ^{2}+k_{s}\pi _ { s}^{2}\right]\nonumber \\ & & + \delta v\int dx\left [ \pi _ { c}\pi _ { s}+\partial _ { x}\phi _ { c}\partial _ { x}\phi _ { s}\right ] \nonumber \\ & & + \frac{u}{2\pi ^{2}a}\int dx\cos \left ( \sqrt{8\pi } \phi _ { s}\right ) \nonumber \\ & & + \frac{u}{2\pi ^{2}a}\int dx\cos \left ( \sqrt{8\pi } \phi _ { c}+2(k_{f\uparrow}+k_{f\uparrow})x\right ) , \nonumber\end{aligned}\ ] ] with @xmath29 } } , \\ v_{s } & = & a\sqrt{{t_{\uparrow } \sin ( k_{f\uparrow } a)+t_{\downarrow } \sin ( k_{f\downarrow } a)}\left ( { t_{\uparrow } \sin ( k_{f\uparrow } a)+t_{\downarrow } \sin ( k_{f\downarrow } a)}-\frac{u}{2\pi } \right ) } , \\ \frac{1}{k_{s } } & = & \sqrt{1-\frac{u}{2\pi \left [ t_{\uparrow } \sin ( k_{f\uparrow}a)+t_{\downarrow } \sin ( k_{f\downarrow } a)\right ] } } , \\\delta v & = & { a}\left [ t_{\uparrow } \sin ( k_{f\uparrow } a)-t_{\downarrow } \sin ( k_{f\downarrow } a)\right].\end{aligned}\ ] ] here , the bose field @xmath30 present the charge and spin degrees of freedom , respectively .@xmath31 and @xmath32 are the fermi momentum for up- and down - spin electrons , @xmath33 with @xmath34 and @xmath35 are the filling densities for up- and down - spin electrons , respectively ; and @xmath36 lattice constant .@xmath37 are the propagation velocities of the charge and spin collective modes of the decoupled model ( @xmath38 = 0 ) , and @xmath39 are the stiffness constants .it can be seen that this model is the standard hm at unpolarized case with both bose fields decoupling . and if model is at half - filling band , @xmath40 the umklapp term ( the last term in @xmath41 ) is important and the hm is in the sdw phase with @xmath42 or in the cdw phase with @xmath43 . as the model shifts away from the half - filling band , the umklapp interaction can be cancelled .the difference of hopping integral and the filling densities of up- and down - spin electrons in the system appear as an effect that breaks the spin - charge separation as reveals the presence of the third term in the last equation . here, we discuss the quantum transition in the case of away from the half filling which means that the umklapp interaction can be ignored . in general @xmath44 . in the parameter region of @xmath42 , since @xmath45, so the term of @xmath46 is irrelevant whenever the system is at unpolarized case , in the one - loop approximation we can cancel it directly and get @xmath47 \nonumber \\ & & + \frac{v_{s}}{2}\int dx\left [ \frac{1}{k_{s}}\left ( \partial _ { x}\phi _ { s}\right ) ^{2}+k_{s}\pi _ { s}^{2}\right ] \nonumber \\ & & + \delta v\int dx\left [ \pi _ { c}\pi _ { s}+\partial _ { x}\phi _ { c}\partial _ { x}\phi _ { s}\right ] .\label{equ9}\end{aligned}\ ] ]the hamiltonian ( [ equ9 ] ) can be diagonalized in terms of two new phase fields which contains a mixture of spin and charge degrees of freedom .the propagation velocities of these collective modes are @xmath48 . } \ ] ] as @xmath49 and @xmath50 . as @xmath38 increases, @xmath51 decreases until vanishes at the points @xmath52at these points , the freezing of the lower bosonic mode is accompanied by a divergence in the charge and spin response functions .the static charge compressibility @xmath53 diverges at @xmath38 = @xmath54 , or @xmath55= @xmath56 .it behaves as @xmath57 ^{-1},% \text { \ \ \ \ } \kappa _ { 0}=\frac{2k_{c}}{\pi v_{c}},\ ] ] beyond these points the susceptibilities becomes negatives .this behavior of the static response functions together with the vanishing of the collective modes velocity indicates that the system becomes unstable@xcite and undergoes a first order phase transition@xcite .this instability is known as phase separation and has been shown to occur in the extended hm@xcite and in the @xmath58 model@xcite . in our case , we obtain @xmath59 it is obvious that @xmath60 so the system is in ps phase state as @xmath61 i.e. , @xmath62 then we obtain the ps phase state condition @xmath63 let us now focus our attention on the correlation functions .our interest in this work is observing the algebraic decay of the instantaneous correlation functions at zero temperature and studying how the exponents get modified from the standard hm .the operators for cdw , sdw , ss , and ts fluctuations in their bosonized form are@xcite @xmath64 , \nonumber \\ & & o_{\rm sdw}^{+}(x)=\psi _ { 1\uparrow } ^{+}\psi _ { 2\downarrow } \nonumber \\ & & \;\ ; = \frac{1}{2\pi \epsilon}\exp\left[\sqrt{2\pi } i\left(\phi _ { \rho } ( x)-\int_{-\infty } ^{x}dy\pi _ { \sigma } ( y)\right)-2ik_{f}x\right ] , \nonumber \\ & & o_{\rm ss}^{+}(x)=\psi _ { 1\uparrow } ^{+}\psi _ { 2\downarrow } ^{+ } \nonumber \\ & & \;\ ; = \frac{1}{2\pi \epsilon } \exp\left[\sqrt{2\pi } i\left(-\int_{-\infty } ^{x}dy\pi _ { \rho } ( y)+\phi _ { \sigma } ( x)\right)\right ] , \nonumber \\ & & o_{\rm ts}^{+}(x)=\psi _ { 1\uparrow } ^{+}\psi _ { 2\uparrow } ^{+ } \\ \nonumber & & \;\ ; = \frac{1}{2\pi \epsilon } \exp\left[\sqrt{2\pi } i\left(-\int_{-\infty } ^{x}dy\pi _ { \rho } ( y)-\int_{-\infty } ^{x}dy\pi _ { \sigma } ( y)\right)\right],\end{aligned}\ ] ] which represent fluctuations of cdw , sdw , ss and ts phases , respectively .the correlation functions are defined as @xmath65 after some cacluation , we find that the correlation functions behavior as @xmath66 the exponents @xmath67 s determine the divergence of the corresponding phase .the expressions obtained for the @xmath67 are @xmath68 with @xmath69,\\ \mu _ { s}&=&\frac{v_s}{v_+ + v_-}\left [ 1+\frac{v_c^2}{v_+ v_-}\left(1-\frac{\delta v^2}{\delta v_1 ^ 2}\right)\right],\\ \nu _ { c}&=&\frac{v_c}{v_+ + v_-}\left [ 1+\frac{v_s^2}{v_+ v_-}\left(1-\frac{\delta v^2}{\delta v_1 ^ 2}\right)\right],\\ \nu _ { s}&=&\frac{v_s}{v_+ + v_-}\left [ 1 + -\frac{v_c^2}{v_+ v_-}\left(1-\frac{\delta v^2}{\delta v_2 ^ 2}\right)\right],\\ \gamma_{c}&=&\frac{\delta v}{v_+ + v_-}\left [ 1+\frac{\delta v_2 ^ 2-\delta v^2}{v_+ v_-}\right],\\ \gamma_{s}&=&\frac{\delta v}{v_+ + v_-}\left [ 1+\frac{\delta v_1 ^ 2-\delta v^2}{v_+ v_-}\right].\end{aligned}\]]the ground state is controlled and named by the most divergent correlation function , i.e. , with the largest @xmath70 .( color online ) the structure factor of the cdw for two different modes @xmath71 and @xmath72 at given @xmath73. here @xmath74 , and @xmath75 denotes @xmath76.,width=302 ] ( color online ) the boundary line between the density wave and phase separation predicted by the bosonization method ( solid line ) and exact diagonalization ( dotted line for @xmath77 ) at give concentration @xmath78 . here@xmath79.,width=302 ] ( color online ) the boundary line between the density wave and phase separation predicted by the bosonization method ( solid line ) and exact diagonalization ( dotted line , @xmath80 ) at give concentration @xmath81 . here@xmath82.,width=302 ] to have a deep understanding on the ps , it is very useful to study the structure factor of the cdw .since the dominating configuration of electrons with spin - down is quite different in two phase , so we introduce the following structure of down - spin electrons , @xmath83 where @xmath84 .we show the structure fact as a function of @xmath85 at a given @xmath73 for two different modes @xmath71 and @xmath72 in fig .[ figure_cdw05 ] . from the figure we find that in the small @xmath6 limit , @xmath86 dominate , this fact manifest the phase separation ; while in a relatively larger @xmath6 region, @xmath87 dominates , this suggests the density - wave state .so we can use the intersection point of the structure factors of two different modes to determine the transition point . in fig .[ figure_tphase10_4 ] and [ figure_tphase12_4 ] , we shown the phase diagram for different concentration @xmath88 in the small @xmath7 region .the results are very impressive . in both figures, we can see that if @xmath89 , the numerical results from the exact diagonalization method agree with eq .( [ equ12 ] ) excellently .that is the phase boundary in the small @xmath7 regime is proportional to @xmath90 .however , when @xmath7 becomes large , say @xmath91 , the bosonization results deviate from the numerical results apparently .we interpret it due to the fact the bosonization method becomes invalid in the large @xmath7 region . on the other hand, the excellent agreement between the results obtained from two approaches suggests that finite - size correction to the numerical data for a finite sample is very small .therefore , the bosonization results are wonderful in the small @xmath7 region and low concentration conditions . since the 1d ahm is equivalent to the fkm if @xmath92 @xmath93or @xmath94 , which has been proven that there exists ps phase at infinite-@xmath7 limit when the system shifts away form half - filling . from eq .( [ equ12 ] ) , we find that the ps phase always appears in the 1d fkm whenever the onsite interaction is small or large as the system shifts away form half - filling . on the other hand ,the numerical studies @xcite suggest there might exist a critical @xmath7 if the density of electrons is close to half - filling .this inconsistence may due to the effect of umklapp process around half - filling .in summary , we have studied the quantum phase transitions in the 1d ahm with the bosonization approach . in the framework of standard bosonization method, we first obtained an effective hamiltonian of 1d ahm .then we diagonalized the hamiltonian and obtained the propagation velocities of the collective modes for both spin and charge degree of freedom .based on the instability condition , we got the final critical conditions of the phase transition from dw to ps .we also obtained the analytical expressions for the correlation functions of cdw , sdw , ss , ts fluctuations , as well as the corresponding exponents .our results show that the difference between the hopping integrals for up- and down - spin electrons is crucial for the happening of the ps .when the difference is large enough , the phase separation will appear even if the on - site interaction is small . in the small-@xmath7 and low concentration region , critical conditions which scales like @xmath95 agree with the numerical results excellently .we thank hq lin for suggesting this problem to us .sj gu is grateful for the hospitality of the physics department at shanghai jiaotong university .this work is supported by rgc grant cuhk 401504 , nsfc grant no . 10304014 , and and foundations from sci . & tech .committee of shanghai municipality ( grant nos . 03qa14054 ) .l. m. falicov and j. c. kimball , phys .* 22 * , 997 ( 1969 ) ; r. ramirez , l. m. falicov , and j. c. kimball , phys .b * 2 * , 3383 ( 1970 ) ; j. k. freericks , ch .gruber and n. macris , phys .b * 60 * 1617 ( 1999 ) ; j. k. freericks and v. zlatic , rev .75 * , 1333 ( 2003 ) .h. q. lin , e. gagliano , d. k. campbell , e. h. fradkin , and j. e. gubernatis , in the hubbard model : _ its physics and mathematical physics _ , edited by d. baeriswyl , pp .315 ( plenum press , 1995 ) ; see also h. q. lin , d. k. campbell , and r. t. clay , chin .* 38 * , 1 ( 2000 ) .

<S> the quantum phase transitions in the one - dimensional asymmetric hubbard model are investigated with the bosonization approach . </S><S> the critical conditions for the transition from density wave to phase separation , the correlation functions and their exponents are obtained analytically . </S><S> our results show that the difference between the hopping integrals for up- and down - spin electrons is crucial for the happening of the phase separation . </S><S> when the difference is large enough , the phase separation will appear even if the on - site interaction is small . </S>

0808.1045

we have discussed the problems related to the ultra high energy cosmic rays ( uhecr ) quite many times in recent years after the results of the high resolution fly s eye ( hires ) started trickling out several years ago .the cosmic ray energy spectrum extracted from the the hires observations indicated that the spectrum suffers a strong decrease at energies above 5@xmath010@xmath1 ev @xcite , consistent with the gzk cutoff @xcite that is due to uhecr interactions with the microwave background .there was thus a strong disagreement with the energy spectrum derived from the previous largest exposure cosmic ray air shower array , agasa @xcite that seemed to continue with the same slope above 10@xmath2 ev .hires and agasa measure the cosmic ray events in different ways .hires is a fluorescent detector that follows the shower longitudinal development by detecting the fluorescent light emitted by the nitrogen atoms in the air ionized by the shower charged particles .the primary cosmic ray energy is obtained from integration of the shower longitudinal profile normalized by the average electron energy loss and the fluorescent efficiency of the air .agasa , on the other hand , is a classical air shower array that records the particle density at the observation level and uses monte carlo calculations to relate it to the primary cosmic ray energy .the dependence of this method on the hadronic interaction model used in the simulations is considered stronger in this approach . in 2007 the first results of the pierre auger observatory ( pao ) in argentina were published .this experiment became the hope for better understanding of uhecr several years ago when its construction started .it is a hybrid air shower experiment that applies both detection methods .pao consists of an air shower array of area 3,000 km@xmath3 where the individual detectors are 10 m@xmath3 water tanks placed at distances of 1.5 km in a hexagonal pattern ( sd ) .the shower array is surrounded by 24 fluorescent telescopes ( fd ) located in four groups of six .the idea is that during cloudless moonless nights when the fluorescent detectors can work the uhecr showers will be observed simultaneously by both types of detectors and this _ hybrid _ observation will unveil better the shower properties . here is a brief summary of the pierre auger observatory results .the energy spectrum derived from pao data @xcite shows a feature consistent with the gzk cutoff similar , but not identical to that of hires - see fig .[ augerhr ] .the spectrum is based on the much higher surface detector statistics normalized to the energy estimates of the fluorescent detector in hybrid events .the problem is that if the energy was estimated only from the sd data ( particle density at 1,000 meters from the shower core - @xmath4 ) the energy scale would be 25% higer @xcite .formally this is not a big problem because the uncertainty of the energy estimate is 22% .another problem appears when pao attempts to estimate the type of the primary cosmic ray nuclei .when it is estimated from the depth of the shower maximum measured by the fluorescent detector @xmath5 the average primary mass appears to be close to he @xcite . if the surface detector density @xmath4 is used to estimate the number of shower muons @xcite the average primary nuclei appear to be much heavier .the preliminary conclusion from these two disagreements is that showers are absorbed in the atmosphere slower than models predict .the auger collaboration succeeded , however , to eliminate a class of uhecr models that predicted that the primary particles are gamma rays . from studies of @xmath5 and other shower propertiesthe fraction of gamma rays in the cosmic rays above 10@xmath1 ev is limited to not more than 2% @xcite .the much smaller statistics at higher energy does not allow for such strong limits , but at least the bulk of uhecr , which we consider extra galactic , consists of nuclei .the auger result that impressed the most not only the astrophysical community , but also the general public , was the published correlation of their highest energy events with active galactic nuclei @xcite .twenty of the 27 events of energy above 5.7@xmath010@xmath1 ev ( 57 eev ) come within 3.1@xmath6 of the position of agn at redshift smaller than @xmath70.017 ( distance of 71 mpc ) while one expects only 7 in case of isotropic source distribution .the 3.1@xmath6 circle around the event direction ( that was obtained from a scan in energy , angle and distance ) is understood as scattering in the galactic magnetic field plus about 1@xmath6 auger angular resolution .accounting for the number of scans the chance probability is well below 1% .if the events that come from galactic latitude less than 10@xmath6 ( where catalogs are not full and cosmic ray may scatter more ) are excluded then 19 out of 21 events are closer than 3.1@xmath6 from nearby agn . since the auger collaboration published the arrival directions and energies of these 27 events in ref .@xcite this announcement was quickly followed by a number of other analyses and questions , like : + @xmath8 why only 71 mpc when @xmath960 eev cosmic rays should come from distances up to 200 mpc ? + @xmath8 why there are no events from the direction of the virgo cluster , that contains powerful agn ?+ @xmath8 what does the concentration of events from directions close to cen a means ?+ many of these questions were also asked by the auger collaboration @xcite .the energy spectrum at energies above 10@xmath10 ev depends mostly on two parameters : the acceleration spectrum at the sources and the source distance distribution , as the spectrum detected at earth evolves in propagation in the universal photon fields .most of the propagation calculations used for the determination of the acceleration spectrum have been done in two simplifying assumptions : + @xmath8 all sources have identical acceleration spectra , and + @xmath8 sources are isotropically distributed in the universe .since the energy of the background photons is much lower than 1 ev the center of mass energy for @xmath11 interactions is low , at the threshold for photoproduction interactions .the cross section of the @xmath11 photoproduction interactions is very well known from accelerator measurements - see fig . 3 in ref .@xcite as well as the fractional proton energy loss per interaction .another energy loss process for protons is the @xmath12 bethe - heitler pair production @xcite where the proton energy loss peaks at about 10@xmath1 ev and the adiabatic energy loss due to the expansion of the universe .the energy loss length for protons is shown in fig .[ xloss ] .nuclei heavier than h in addition to photoproduction and pair production ( scaled by @xmath13 and shifted up in total energy by @xmath14 ) suffer from a different energy loss - photo disintegration @xcite .there are two main processes - one nucleon loss at the giant nuclear resonance and two nucleon loss at higher photon energy in the nucleus frame .one has also to account for the decay of unstable nuclei generated this way .if uhecr are protons there are two general solutions for the cosmic ray acceleration spectrum - relatively steep acceleration spectrum @xmath15 with no cosmological evolution of the sources ( a l berezinsky @xcite ) and a much flatter ( @xmath16 = 1.2 - 1.3 ) with a strong cosmological evolution such as that of the star forming regions ( sfr ) .the cosmological evolution is not important for the highest energy cosmic rays that can not come to us from redshifts higher than 0.05 but does affect lower energy ones .note that the first ( ` dip ' ) model does not need any galactic cosmic rays above about 10@xmath10 ev .mixed composition models @xcite generally do not require any cosmological evolution of the uhecr sources and are consistent with acceleration spectra with @xmath16 = 1.2 - 1.3 .figure [ augerhr ] compares the cosmic ray energy spectra derived by the auger and the hires collaborations .the lines drawn through the data points are rough fits of the spectra in limited energy ranges .note these are not identical to the fits made by the auger collaboration @xcite .apart from the overall normalization of the two spectra they show a small difference in shape that is still statistically not very significant but currently favor different astrophysical models .the hires spectrum @xcite is fully consistent with the model of berezinsky et al that involves a steep acceleration spectrum ( @xmath16 = 1.7 ) .the ` dip ' at about 3@xmath010@xmath10 ev is formed at the edge of pair production and adiabatic energy loss .there are several good fits to the auger energy spectrum performed by the collaboration itself @xcite : a steep proton acceleration spectrum with @xmath16 = 1.55 without cosmological evolution , a flatter one with @xmath16 = 1.3 and strong @xmath17 cosmological evolution , and a mixed composition spectrum with @xmath16 = 1.2 some of these fits require the galactic cosmic ray spectrum to extend above 10@xmath18 ev .it is worth noting that if the auger energy scale were 25% higher , the spectrum would be much closer to the hires one in both overall normalization and in the position of the gzk feature and that they predict different chemical compositions of uhecr .the energy that uhecr lose in propagation end up in fluxes of gamma rays and neutrinos from @xmath19 and @xmath20 decays .gamma rays have even shorter energy loss length than nuclei and develop pair production / inverse compton cascades where synchrotron radiation may play important role . in the presence of noticeable ( 1 ng ) extragalactic field 10@xmath10 ev electrons quickly lose energy to gev synchrotron photons .neutrinos , however , have very low interaction cross section and arrive at earth with just redshift energy loss from almost any distance . for this reason the cosmological evolution of the cosmic ray sources are very importantinput in the calculation of these _ cosmogenic _@xcite neutrinos .two among the auger results presented above may have an important connection to the flux of cosmogenic neutrinos - the correlation of the highest auger events with nearby agn and the @xmath21 cosmological evolution evolution required by one of the spectrum fits with relatively flat ( @xmath16 = 1.3 ) proton models .the cosmological evolution of agn has been studied by their x - ray emission as a function of the redshift .most of the x - ray satellite measurements were sensitive to relatively low energies and the worlds richest data set is in the 0.5 to 2 kev range .anyway it is very difficult to know what the best proxy is for the evolution of the cosmic ray luminosity , but the available astronomical data are either in infrared ( for the sfr ) or in low energy x - rays .the agn observations of the rosat satellite were analyzed in ref .@xcite and the derived cosmological evolution is @xmath22 . a later analysis @xcite used much larger statistics and was able to divide the observed agn in four luminosity groups , each encompassing one order of magnitude above 10@xmath23 erg / mpc@xmath24 .it turns out , according to this analysis , that the highest x - ray luminosity agn also have the highest luminosity evolution : @xmath25 up to @xmath26 = 1.7 .such cosmological evolution would indeed generated very high fluxes of cosmogenic neutrinos .there are not , however , many such agn and it is not very likely that they are the sources of uhecr that have to come ( according to the analysis in ref .@xcite ) from many nearby sources .it maybe better to use for this purpose the cosmological evolution of the agn x - ray luminosity that could be obtained by summation from the data published in ref .@xcite and is shown in fig .[ table3z ] . , width=4 ] up to redshifts exceeding 1 the total agn luminosity has @xmath22 cosmological evolution , significantly stronger than the evolution of star forming regions .in addition to the cosmological evolution of the cosmic ray sources the flux of cosmogenic neutrinos depends on a number of parameters : * the average acceleration slope of the highest energy cosmic rays ( uhecr ) * the maximum acceleration energy @xmath27 . * the distribution of the cosmic ray sources in the universe .this is important to estimate the average uhecr emissivity since we only observe uhecr generated in our cosmological neighborhood , i.e. @xmath28 0.40 .auger analysis requires source density higher than 3.5@xmath010@xmath29 mpc@xmath30 .( number sources within 75 mpc > 61 ) we will use the following parameters in calculation of the cosmogenic neutrino flux : + uhecr emissivity of 2.25@xmath010@xmath31 erg / mpc@xmath24 , i.e. one half of the value derived by waxman and used in an earlier calculation @xcite + acceleration spectrum with @xmath16 = 1.3 with an exponential cutoff at 10@xmath32 ev + cosmological evolution @xmath22 up to @xmath26 = 1.7 , then flat to @xmath33 = 2.7 with exponential decline after that + cosmological model with @xmath34 = 0.73 + and new neutrino yields calculated with the same cosmological model for interactions on the microwave background figure [ all4 ] compares the spectrum of all four neutrino flavors calculated in this way to that of the previous calculation based on waxman s parameters @xcite .note that the higher energy peak consists mostly of @xmath35 , @xmath36 and @xmath37 while the lower energy one represents @xmath38 from neutron decay .the contribution of electron anti neutrinos to the higher energy peak comes from neutron interactions and is small as most neutrons decay .although the fluxes are not very distant from each other there are several significant differences : + @xmath8 the auger inspired flux cuts off at lower energy because of its steeper acceleration spectrum in spite of the identical maximum energy .if the maximum energy were lower the total normalization of the flux would be lower and the cutoff would be scaled down .+ @xmath8 the peaks in the neutrino energy distribution are slightly higher but at somewhat lower energy .both effects can be related to the higher contribution of high redshifts .it is difficult to estimate correctly the detectable event rates from these two distributions , but the auger inspired fit is certainly much less efficient in production of very high energy neutrinos that could be detected by experiments such as anita and generally all experiments based on radio detection of neutrino induced showers .a full calculation of the cosmogenic neutrino fluxes should also include neutrinos from interactions in the infrared / optical background .in such a calculation the relatively flat acceleration spectrum would not make a huge difference although the infrared background contribution would be stronger than the one for @xmath16 = 1 spectrum .the strong cosmological evolution of the sources would also not be very important since the infrared background evolves not as fast as the microwave one .models of uhe cosmic rays accelerated in a mixed cosmic ray composition generate cosmogenic neutrino fluxes with a different shape .since the energy measured by air showers is the total energy per nucleus such models generate smaller high energy peaks and much higher lower energy one from neutron decays after disintegration .the photoproduction threshold is proportional to the nuclear mass , @xmath14 times higher then 3@xmath010@xmath1 which is the proton threshold for interactions in mbr .he nuclei would then start interacting at energy higher than 10@xmath2 ev and iron nuclei - above 10@xmath39 ev .although the maximum energy achieved in acceleration depends on the nuclear charge @xmath40 such nuclei may not exist and the mixed composition models tested by auger only extend to 10@xmath2 ( 10@xmath39 ) ev , i.e. the energy per nucleon does not reach the he ( fe ) photoproduction threshold .the first approximation for the flux of cosmogenic neutrinos generated by mixed composition models is that only their proton component undergoes photoproduction interactions .this , however , is not well defined either because the cosmic ray composition continuously changes in propagation - it becomes lighter as nuclei disintegrate .a good estimate can only be obtained by a direct monte carlo simulation assuming a particular cosmic ray composition at the sources . assuming a composition introduces some more free parameters in the calculation .it is important to note that the lower energy @xmath38 peak is below the glashow resonance energy and would not give a high neutrino interaction rate .the first results of the auger collaboration did improve our knowledge of the ultra high energy cosmic rays .they are fully consistent with a gzk feature in the cosmic ray spectrum as suggested previously by the hires collaboration .being a hybrid experiment auger demonstrated that the energy estimate by fluorescent detectors and surface air shower array is different , as previously suggested by the agasa and hires spectra .now the uhe community is searching for the reasons for this disagreement that may hide in the currently used hadronic interaction models .auger started the direct search for the sources of uhecr by studies of the correlation of their highest energy events with active galactic nuclei .the published correlation inspired many other different analyses of correlation with powerful astrophysical objects .this work is in progress now and the uhecr sources will be identified eventually when the auger statistics grows - it will double at the end of august 2008 .since the current uhecr energy spectrum can be fit with several different models it is not obvious that we know better the expected flux of cosmogenic neutrinos .this may only happen when an unique fit of the spectrum is achieved .+ * acknowledgments * this talk is based on collaboration with many colleagues including peter biermann , daniel demarco , thomas gaisser , jrg rachen , & david seckel .the author was supported in part by nasa apt grant nng04gk86 g .99 the hires collaboration ( r. abbasi et al . ) , phys.rev.lett .100:101101 ( 2008 ) k. greizen , phys .lett . , 16:748 ( 1966 ) ; g.t .zatsepin & v.a .kuzmin , jetp lett 4:78 ( 1966 ) the agasa collaboration ( m. takeda et al . ) , phys .lett . , 81:1163 ( 1998 ) the auger collaboration ( j. abaraham et al .) , arxiv:0806.4302 the auger collaboration ( r. engel et al . ) , arxiv:0706.1921 the auger collaboration ( m. unger et al . ) , arxiv:0706.1495 the auger collaboration ( j. abraham et al . ) , astropart.phys .29:243 ( 2008 ) the auger collaboration ( j. abraham et al . ) , science 318:938 ( 2007 ) the auger collaboration ( j. abraham et al . ) , astropart .phys . 29:188 ( 2008 )a. mcke et al . , comput.phys.commun.124:290-314 ( 2000 ) a.m. hillas , proc 14th icrc ( mnchen ) , 2:717 ( 1975 ) v.s .berezinsky & s.a .grigorieva , a&a , 109:1 ( 1988 ) f.w .stecker , phys .180:1264 ( 1969 ) v.s .berezinaky et al . , phys .d74:043005 ( 2006 ) d. allard et al . ,a&a , 443:l29 ( 2005 ) the auger collaboration ( t. yamamoto et al . ) , arxiv : 0707.2638 the hires collaboration ( r.u .abbasi et al . )lett 92:151101 ( 2004 ) v.s .berezinsky & g.t .zatsepin , phys .lett . 28b:423 ( 1969 )stecker , astrophys .space sci ., 20:47 ( 1973 ) a. franceschini et al ., apj , 506:600 ( 1998 ) 1998 g. hasinger , t. miyaji & m. schmidt , a&a , 441 , 417 ( 2005 ) r. engel , d. seckel & t. stanev , phys .d 64:093010 ( 2001 ) e. waxman , ap.j ., 452:l1 ( 1995 ) * discussion * + * daniele fargion * in your last paper few weeks ago the main conclusive sentence was : following our 1995 paper one could think the auger events may reach us from cen a. i could not find this prediction in that paper .+ * todor stanev* + i should not have used a similar statement in the conclusions since this option was not really discussed in the paper .what i had in mind and tried to lead to is that with reasonable magnetic fields ( order of 1 @xmath41 g ) in the cen a lobe one can imagine uhecr accelerated at that object to bend at significant angles and appear to the observer as coming from nearby sources .+ * sergio colafrancesco * it seems to me that you are assuming a single population of agn in your calculations and not accounting of their astrophysical differences .what are the implications of such an assumption on your results .+ * todor stanev * you are absolutely correct , sergio .all current estimates are done in the assumption that all sources have identical parameters .there are only a few attempts to account for differences choosing a certain type of source and a luminosity and maximum energy for each source ( see p.l .biermann and students work ) .unfortunately in such analyses one has to rely on simple theoretical calculations of these important parameters .one can easily fit the uhecr spectrum in such models as they introduce more ` free ' parameters .

<S> we discuss the main results that were recently published by the auger collaboration and their impact on our knowledge of the ultra high energy cosmic rays and neutrinos . </S>

nucl-th0403056

in 1924 , rosseland@xcite studied the possibility of a self gravitating star on eddington s theory to contain a net charge where the star is modeled by a ball of hot ionized gas ( see also eddington @xcite ) .in such a system the electrons ( lighter particles ) tend to rise to the top because of the difference in the partial pressure of electrons compared to that of ions ( heavier particles ) .the motion of electrons to the top and further escape from the star is stopped by the electric field created by the charge separation .the equilibrium is attained after some amount of electrons escaped leaving behind an electrified star whose net positive charge is of about 100 coulomb per solar mass , and building an interstellar gas with a net negative charge . as shown by bally and harrison @xcite, this result applies to any bound system whose size is smaller than the debye length of the surrounding media .the conclusion is that a star formed by an initially neutral gas can not acquire a net electric charge larger than about @xmath4c per solar mass .it is expected that the sun holds some amount of net charge due to the much more frequent escape of electrons than that of protons .moreover , it is also expected that the escape would stop when the electrostatic energy of an electron @xmath5 is of the order of its thermal energy @xmath6 .this gives for a ball of hot matter with the sun radius , a net charge @xmath7 ( in coulomb ) .hence , the escape effect can not lead to a net electric charge much larger than a few hundred coulomb for most of the gaseous stars . for newtonian stars , the net charge of 100 c per solar massis obtained by the balance between the electrostatic energy @xmath8 and the gravitational energy @xmath9 ( glendenning@xcite ) .however , for very compact stars , the high density and the relativistic effects must be taken into account @xcite . in a strong gravitational field ,the general relativistic effects are felt and the star needs more charge to be in equilibrium .moreover , for very compact stars , the induced electric field can be substantially higher than in the case of the sun .for instance , the same amount of charge yields an electric field approximately @xmath10 times larger at the surface of a neutron star than at the surface of the sun .so , even a relatively small amount of net charge on compact stars can induce intense electric fields whose effects may become important to the structure of the star .this fact deserves further investigation .the general relativistic analog for charged dust stars was discovered by majumdar @xcite and by papapetrou @xcite , and further discussed by bonnor @xcite and several other authors@xcite .study for the stability of charged fluid spheres have been done by bekenstein@xcite , zhang et al.@xcite , de felice & yu@xcite , yu & liu@xcite , de felice et al.@xcite , anninos & rothman@xcite and others .this was indirectly verified by zhang et al.@xcite who found that the structure of a neutron star , for a degenerate relativistic fermi gas , is significantly affected by the electric charge just when the charge density is close to the mass density ( in geometric units ) . in the investigations by de felice et al . , and by anninos & rothman, they assumed that the charge distribution followed particular functions of the radial coordinate , and they were mostly interested in the extreme @xmath11 case .our basic consideration to incorporate charge into the system is in the form of trapped charged particles where the charge goes with the positive value .the effect of charge does not depend on its sign by our formulation .the energy density which appears from the electrostatic field will _ add up _ to the total energy density of the system , which in turn will help in the _ gaining _ of the total mass of the system .the modified tolman - oppenheimer - volkoff ( tov ) equation now has extra terms due to the presence of the maxwell - einstein stress tensor .we solve the modified tov equation for polytropic equation of state ( eos ) assuming that the charge density goes with the matter density and discuss the results .the formation of this extra charge inside the star is however left open .a mechanism to generate charge asymmetry for charged black holes has been suggested recently by mosquera cuesta et al . @xcite andthe same may be applied for compact stars too .this article is arranged in the following way . in section [ sec : rel - form ] , we show the basic formalism for the modified tov . in section [ sec : poly ] , we used this modified tov on a polytropic eos , discuss the results and the stability of the charged stars .finally we make our conclusions in section [ sec : conc ] .we take the metric for our static spherical star as @xmath12 the stress tensor @xmath13 will include the terms from the maxwell s equation and the complete form of the einstein - maxwell stress tensor will be : @xmath14 where p is the pressure , @xmath15 is the energy density ( = @xmath16 ) and @xmath17-s are the 4-velocity vectors . for the time component ,one easily sees that @xmath18 and hence @xmath19 consequently , the other components ( radial and spherical ) of the four vector are absent .now , the electromagnetic field is taken from the maxwell s field equations and hence they will follow the relation @xmath20_{,\nu}~=~4 \pi j^\mu \sqrt{-g } \label{eq : fmunu}\ ] ] where @xmath21 is the four - current density . since the present choice of the electromagnetic field is only due to charge , we have only @xmath22 , and the other terms are absent . in general , we can derive the electromagnetic field tensor @xmath23 from the four - potential @xmath24 .so , for non vanishing field tensor , the surviving potential is @xmath25 .we also considered that the potential has a spherical symmetry , i.e. , @xmath26 .the nonvanishing term in eq.([eq : fmunu ] ) is when @xmath27=r .this gives the electric field for both the @xmath28 and @xmath29 components as : @xmath30 where , @xmath31 is the electric field .so , the total charge of the system is @xmath32 where r is the radius of the star .the mass of the star is now due to the total contribution of the energy density of the matter and the electric energy ( @xmath33 ) density .the mass takes the new form as @xmath34 and the metric coefficient is given by @xmath35 the stress tensor is conserved ( @xmath36 ) .hence , one gets the form of the hydrostatic equation from it as : @xmath37(\epsilon+p)}{c^2r^2\left(1-\frac{2gm_{tot}}{c^2r}\right)}\\ & & + \rho_{ch}{\cal u}e^{\frac{\lambda}{2}}. \label{eq : dpdr}\end{aligned}\ ] ] we solve the eqns .( [ eq : mnew ] , [ eq : lamnew ] & [ eq : dpdr ] ) simultaneously to get our results for the charged compact stars .we study here the effect of charge on a model independent polytropic eos .we assume the charge is proportional to the mass density ( @xmath15 ) like @xmath38 where @xmath39 is in [ mev/@xmath40 . in geometrical units ,this can be written as @xmath41 where charge is expressed in units of mass and charge density in units of mass density .this @xmath42 is related to our charge fraction @xmath43 as @xmath44 our choice of charge distribution is a reasonable assumption in the sense that large mass can hold large amount of charge .the polytropic eos is given by @xmath45 where @xmath46 is the polytropic index and is related to the exponent @xmath47 as @xmath48 . in the relativistic regime , the allowed value of @xmath47 is @xmath49 to @xmath50 .we have considered the adiabatic case of @xmath51 and the corresponding value of @xmath46 is 1.5 .primarily , our units of matter density and pressure are in mev/@xmath52 .we chose a value of @xmath53 as 0.05 [ fm]@xmath54 for our polytrope that reproduces quite well realistic eos for neutron stars @xcite .it should be noted that the amount of charge we find implies a very small ratio @xmath55 which justifies to use an eos which is calculated for neutral matter . in fig.([fig : poly - e - m ] ) , we plot the mass as function of the central density , for different values of the charge fraction @xmath43 . for the charge fraction @xmath56 , we do not see any departure on the stellar structure from that of the chargeless case .this value of @xmath43 is @xmath57 because any increase in the value beyond this , shows enormous effect on the structure .the increase of the maximum mass of the star is very much non - linear , as can be seen from the fig .( [ fig : poly - e - m ] ) . in fig.([fig : poly - m - r ] ) we plotted the mass - radius relation . due to the effect of the repulsive force ,the charged stars have large radius and larger mass as we should expect . even if the radius is increasing with the mass , the m / r ratio is also increasing , but much slower . for the lower charge fractions ,this increase in the radius is very small , but a look at the structure for the fraction @xmath58 reveals that for a mass of 4.3 m@xmath3 , the radius goes as high as 35 km .though the compactness of the stars are retained , they are now better to be called as @xmath59 rather than @xmath60 .the charge fraction in the limiting case of maximally allowed value goes up to @xmath61 , for which the maximum mass stable star forms at a lower central density even smaller than the nuclear matter density .this extreme case is not shown in fig.([fig : poly - m - r ] ) because the radius of the star and its mass is very high ( 68 km & 9.7 m@xmath3 respectively ) which suppresses the curves of the lower charge fractions due to scaling .for this star , the mass contribution from the electric energy density is 10% than that from the mass density .it can be checked by using relation ( [ eq : no - dim - conv ] ) that this charge fraction @xmath61 corresponds to @xmath62 in geometrical units .the q@xmath63 m diagram of fig.([fig : poly - q - m ] ) shows the mass of the stars against their surface charge .we have made the charge density proportional to the energy density and so it was expected that the charge , which is a volume integral of the charge density , will go in the same way as the mass , which is also a volume integral over the mass density .the slope of the curves comes from the different charge fractions.the nature of the curves in fact reflects that charge varies with mass ( with the turning back of the curves all falling in the ` unstable zone ' and is not taken into consideration ) .if we consider that the maximum allowed charge estimated by the condition ( @xmath64 ) for @xmath65 to be negative ( eq . ( [ eq : dpdr ] ) ) , we see that the curve for the maximum charge in fig.([fig : poly - q - m ] ) has a slope of 1:1 ( in a charge scale of 10@xmath66 coulombs we get a relation for the charge and mev/@xmath52 which comes out as @xmath67^{1/2}.$ ] ] ) . this scale can easily be understood if we write the charge as @xmath68this charge q is the charge at the surface of the star where the pressure and also @xmath65 are zero .so , at the surface , the coulomb force is essentially balanced by the gravitational force and the relation of the charge and mass distribution we found is exactly the same for the case of charged dust sphere discussed earlier by papapetrou@xcite and bonnor@xcite .the total mass of the system @xmath69 increases with increasing charge because the electric energy density @xmath70 to the mass energy density .this change in the mass is low for smaller charge fraction and going up to 7 times the value of chargeless case for maximum allowed charge fraction @xmath61 . the most effective term in eq.([eq : dpdr ] )is the factor ( @xmath71 ) .@xmath72 is the effective pressure of the system because the effect of charge decreases the outward fluid pressure , negative in sign to the inward gravitational pressure . with the increase of charge , the value of p@xmath73 decreases , and hence the gravitational negative part of eq.([eq : dpdr ] ) decreasesso , with the softening of the pressure gradient , the system allows more radius for the star until it reaches the surface where the pressure ( and @xmath65 ) goes to zero . we should stress that because @xmath74 can not be too much larger than the pressure in order to maintain @xmath65 negative as discussed before , so we have a limit on the charge , which comes from the relativistic effects of the gravitational force and not just only from the repulsive coulombian partthis effect is shown in fig.([fig : poly - dpdr ] ) where we have plotted both the positive coulomb part and the negative matter part of the pressure gradient .the plots are for two values of the charge fraction @xmath75 and @xmath76 .the positive part of @xmath65 maintains its almost constant value because the charge fraction @xmath43 is the controller of the same , and in our case , they differ by a very small percentage . in the negative part ,the changes are drastic and are mainly brought by the effective pressure as we already discussed . in our high density system , the gravitational and the coulomb forcesare highly coupled .although it is difficult to disentangle the forces , but to a common belief , it can be considered that the charged particles , due to their self created huge field , will leave the star very soon .this process will however lead to an imbalance of the global forces acting on the star , which were previously balanced by the gravitational and the coulomb forces .this process will help in the star to further collapse to a charged black hole .we say it charged because , by the time the system collapses , all the charge has not left the star , and they get trapped inside the black hole @xcite .in our study , we have shown that a high density system like a neutron star can hold huge charge of the order of 10@xmath66 coulomb considering the global balance of forces . with the increase of charge ,the maximum mass of the star recedes back to a lower density regime .the stellar mass also increases rapidly in the critical limit of the maximum charge content , the systems can hold .the radius also increases accordingly , however keeping the m / r ratio increasing with charge .the increase in mass is primarily brought in by the softening of the pressure gradient due to the presence of a coulombian term coupled with the gravitational matter part .another intrinsic increase in the mass term comes through the addition of the electric energy density to the mass density of the system .the inside electric field of the charged stars are very high and crosses the critical field limit for pair creation ( bekenstein @xcite ) .however , this issue is debatable because the critical field has been calculated for vacuum and one does not really know what the value will be in a high density system .the stability of the charged stars are however ruled out from the consideration of forces acting on individual charged particles .they face enormous radial repulsive force and leave the star in a very short time .this creates an imbalance of forces and the gravitational force overwhelms the repulsive coulomb and fluid pressure forces and the star collapses to a charged black hole .

<S> in this work , we analyze the effect of charge in compact stars considering the limit of the maximum amount of charge they can hold . we find that the global balance of the forces allows a huge charge ( @xmath0 coulomb ) to be present in a neutron star producing a very high electric field ( @xmath1 v / m ) . </S><S> we have studied the particular case of a polytropic equation of state and assumed that the charge distribution is proportional to the mass density . </S><S> the charged stars have large mass and radius as we should expect due to the effect of the repulsive coulomb force with the m / r ratio increasing with charge . in the limit of the maximum charge </S><S> the mass goes up to @xmath2 m@xmath3 which is much higher than the maximum mass allowed for a neutral compact star . </S><S> however , the local effect of the forces experienced by a single charged particle , makes it to discharge quickly . </S><S> this creates a global force imbalance and the system collapses to a charged black hole . </S>

1612.01809

and @xmath0 . the vibrational mode is represented by the harmonic oscillator in - between the two electronic sites and we define @xmath1 the oscillator - controlled hopping between @xmath2 and @xmath3.[fig : schematic-1],scaledwidth=45.0% ] as illustrated in fig . [fig : schematic-1 ] , the system being considered consists of two coupled parts : * ( i ) * an electronic wire and * ( ii ) * a vibrational mode . here , we first focus on the hamiltonian part of the dynamics of each subsystem and their coupling .our model aims at capturing the salient features of the energy of a molecular wire .such system consists of a three - modular molecular junction encompassing a left ( @xmath4 ) , right ( @xmath5 ) and central ( @xmath6 ) region .the @xmath4 and @xmath5 part of the junction are connected , through leads , to an external potential , thus putting the junction out of equilibrium and allowing for the circulation of an electronic current .the conductance of the junction depends on the conformation of the molecule itself : by putting part @xmath6 out of the plane containing both @xmath4 and @xmath5 ( which we assume to be coplanar ) , the resistance offered by the junction to the current can be varied in light of the modifications induced to the overlap of the electronic wave - functions that is responsible for the conduction . in what follows, we will focus on the case of small fluctuations around an otherwise stable molecular conformation ( i.e. a stable relative angle between the central and peripheral parts of the molecular junction ) .the relative angle between the in - plane and out - of - plane parts of the junction would thus oscillate around the macroscopically stable configuration , such oscillations being treated quantum mechanically . a minimal model that is able to capture the essential conformational influences over electric conductance is as follows .we consider two single - occupation sites whose energy is given by the on - site hamiltonian ( where we have neglected the coulomb interaction and have assumed units such that @xmath7 ) @xmath8 here , @xmath9 is the annihilation ( creation ) operator of an electron occupying the left ( @xmath4 ) and right ( @xmath5 ) site . for simplicitywe restrict ourself to @xmath10 .we have assumed that the electrons do not hop directly between sites , an assumption invoked only to simplify our approach but that does not affect the phenomenology that will be illustrated here .our model also includes a vibrational mode , which is described under harmonic approximation as @xmath11 here , @xmath12 is the vibrational frequency , @xmath13 is the mode annihilation ( creation ) operator , and @xmath14 is the associated position - like operator .we consider the case of a harmonic oscillator driven by a constant force of strength @xmath15 .electronic and vibrational subsystems are coupled according to the model @xmath16 which describes phonon - assisted inter - site hopping at a rate @xmath17 , which in turn depends on the position of the oscillator , i.e. the molecular conformation .the total hamiltonian of the system is thus @xmath18 the assumption of small oscillations around a stable configuration justifies a series expansion of the hopping rate as @xmath19 .this gives rise to a standard hopping mechanism connecting the two sites being considered , and a phonon - assisted one , occurring at rate @xmath20 , that depends explicitly on @xmath21 .thermal excitations are a possible physical mechanism responsible for the oscillations around the stable molecular conformation .we thus consider a thermal reservoir coupled to the vibrational mode .this leads to the dynamical model @xmath22+\mathcal{l}_{\text{v}}[\hat{\rho}_{\text{m}}],\ ] ] where @xmath23 is the density matrix of the system and we have introduced the lindblad dissipator @xmath24=\frac{\gamma\bar{n}}{2 } d[\hat a^\dag,\bullet]+\frac{\gamma(\bar{n}+1)}{2 } d[\hat a,\bullet ] , \label{eq : ldiss-1}\end{aligned}\ ] ] where @xmath25 is the coupling strength to the bosonic bath , @xmath26 is the mean number of excitations in the bath ( related to the bath s temperature by @xmath27 - 1)$ ] ) , and @xmath28=(2\hat o\bullet\hat{o}^\dag-\ { \hat{o}^\dag\hat{o},\bullet\})$ ] .the equation of motion of the annihilation operator of the vibrational mode reads @xmath29 we assume that the oscillatory mode reaches its steady - state in a time much shorter than the characteristic time of the evolution of the system , so that we can advocate for the validity of the adiabatic approximation , according to which the state of the oscillator can be assumed to be stationary and unaffected by the coupling to the electronic wire .its degrees of freedom can thus be traced out to seek for an effective reduced dynamics of the electronic system . from eq ., the steady - state position of the oscillator is @xmath30 the heisenberg evolution of the wire operator is @xmath31 where @xmath32 $ ] and with @xmath33 and @xmath34 are either @xmath5 or @xmath4 . replacing @xmath35 with its steady - state solution @xmath36 in eq . , we have @xmath37 where @xmath38 and @xmath39 .( [ eq : l-1 ] ) can be interpreted as the heisenberg equation for the operator @xmath40 evolving according to the effective hamiltonian @xmath41 which does not contain the oscillator s degrees of freedom .the adiabatic elimination induces a significant change of the coherent part of the dynamics of the electronic subsystem , but also of the incoherent part .indeed , the reduced electronic density matrix @xmath42 $ ] evolves according to the master equation @xmath43+\mathcal{l}_{\text{eff}}[\hat{\rho}]\ ] ] with @xmath44 that describes the effective dissipation induced by the vibrational mode on the wire subsystem @xmath45 = \gamma_{\text{v}}[\hat{s}_{l}^{\dagger}\hat{s}_{r}\bullet\hat{s}_{l}^{\dagger}\hat{s}_{r}+ d[\hat{s}_{l}^{\dagger}\hat{s}_{r},\rho]+l\leftrightarrow r ] \label{eq : ldiffeff-1}\;,\ ] ] with @xmath46 .this incoherent part of the wire subsystem dynamics encompasses two processes .the first one , corresponding to the first term in eq .( and the analogous one where label @xmath4 is swapped with @xmath5 ) , randomly swaps the coherence between single occupancy states , while the second term ( and analogous with @xmath47 ) allows for the incoherent hopping between both sites .it is worth noticing that this last process leads to an enhancement of the conduction through the wire .moreover the strength of such an incoherent process depends directly on the temperature of the bath attached to the vibrational mode through the mean occupation number @xmath26 .such decoherence induced by the molecular vibration conserves the number of excitations , stating that , under the assumptions considered here , there is no energy exchange between the wire and the vibronic system .we find it convenient , for the continuation of our analysis , to move to the eigenbasis of the effective hamiltonian in eq . , which involves non - local states @xcite , so as to get the diagonal operator @xmath48 where @xmath49 are the system energies ( ordered so that @xmath50 ) , corresponding to the non - local states @xmath51 .in what follows , we use a notation such that the states @xmath52 ( @xmath53 ) represents configurations with @xmath54 and @xmath55 electrons in the left and right site , respectively .we call these _ local - basis _ states .a close analysis reveals that the first and last eigenstates in the non - local basis are identical to those of the local one , that is @xmath56 where no electrons are in the system and @xmath57 where one electron is on each sites .notice that the hamiltonian in eq .does not couple subspaces with a different total number of electrons in the sites .the intermediate states @xmath58 and @xmath59 are instead coupled to give rise to the non - local states @xmath60 as @xmath61 with @xmath62 the change - of - basis matrix @xmath63 where we have introduced the parameters @xmath64 and @xmath65 .let us now connect each electronic site to a lead .the picture we have in mind can be sketched as in fig .[ fig : schematic-1 ] , where each electronic site is individually coupled to its own fermionic bath .the hamiltonian of each lead is @xmath66 where @xmath67 refers to the left or right lead and @xmath68 ( @xmath69 ) the fermonic destruction ( creation ) operator for the @xmath70 mode in the @xmath71 bath .the coupling between the electronic sites and each lead is of the hopping form @xmath72 we will now eliminate the leads degrees of freedom . writing eq . in the eigenbasis of the electronic subsystem, we have @xmath73\label{eq : g}\end{aligned}\ ] ] with @xmath74 and @xmath75 .the first ( second ) line of eq . describes the transition between the wire ground state @xmath76 ( highest energy state @xmath77 ) and an intermediate state ( either @xmath78 or @xmath79 ) . as in refs .@xcite , assuming the born - markov approximation , we eliminate the degrees of freedoms of both leads , giving rise to two effective dissipators @xmath80 , i.e. one per lead , as defined in appendix [ sub : dissipation - of - the ] . for simplicitywe assume the leads to act as two similar ohmic baths with fermi distribution at chemical potential @xmath81 given by @xmath82 the number of excitations in the leads is taken to be the same ( that is @xmath83 ) and we choose the coupling strength @xmath84 . in this way , the only difference between the two leads consists in the respective chemical potentials @xmath85 .we decompose the dissipation due to the leads , taken in the local picture as @xmath86 , where the first refers to usual amplitude damping channel and the second one to decoherence .both are explicitly defined in appendix [ sub : dissipation - of - the ] [ eq . andrespectively ] . notice that last term @xmath87 is not of a dephasing type because it changes the system energy as well as coherences . despite a rather simple system the dynamics taking place andespecially the dissipation acting on the system is rich .we will now focus on this dynamics in order to determine the exchange taking place between the system and the leads .in this section we discuss the evolution of the electronic subsystem driven by its effective dynamics and connect it to the exchange statistics with the leads , using the thermodynamics of trajectories ( sec .[ sub : exchange - statistics ] ) .we then develop a systematic approach to identify the regime maximizing the enhancement effect provided to the current crossing the devices ( sec .[ sub : systematic ] ) arising from the presence of the vibrational mode . in order to find the stationary state of the systemwe restrict ourselves to the evolution of the diagonal elements of the density matrix @xmath88 in the local basis .however , due to the interaction and possible hybridization of intermediate levels , terms such as @xmath89 and @xmath90 are also crucial .the dynamics of the electronic system can be tracked by writing its density matrix in vector form as @xmath91 @xmath92 , and the master equation as @xmath93 the form of the superoperator @xmath94 , which is not essential for the discussions to follow , is provided in appendix [ sub : superop ] .the resulting dynamics is similar to the one found in ref .@xcite , albeit with specific features that should be stressed . among them ,the most significant is that the incoherent hopping induced by the vibrational mode acts directly on both the coherences ( @xmath95 and @xmath96 ) of the electronic density matrix , and the occupations of the intermediate states ( @xmath97 and @xmath98 ) . to determine the exchange statistics taking place between the system and the leads we now use the formalism of thermodynamics of trajectories @xcite: we define a counting process of the net exchange of excitations between the system and the right leads such as @xmath99 , where @xmath100 ( @xmath101 ) refers to an increment related to excitations leaving ( entering ) the system to ( from ) the lead @xmath71 inducing a transition between level @xmath102 and @xmath103 .the introduction of such process modifies the dynamics of the electronic density matrix , which now obeys the biased master equation @xmath104+\mathcal{l}_{s}[\hat\rho_{s}]$ ] where the biasing contribution to the evolution @xmath105 is defined in eq . and where @xmath106 is the biased density matrix with @xmath107 a projector over the subspace where the selected counting process results in @xmath108 excitations being exchanged .as done previously , we consider the evolution of the vector of relevant density matrix elements @xmath109@xmath110 @xmath111 , which occurs as @xmath112 , where @xmath113 is given in appedix [ sub : superop ] .the interesting feature of this system is that even the evolution of the coherences may induce changes on the counting statistics . at this stage, we can formally access the large deviation function @xmath114 ( scaled cumulant generating function ) encoding the full counting statistics defined as @xmath115\boldsymbol{\rho}_{s}(0)\right)\;,\ ] ] where @xmath116 , @xmath117 and @xmath118 the time ordering operator . through the diagonalization of the propagator @xmath119one can access directly @xmath114 .the latter is aptly determined as the eigenvalue of @xmath119 with the longest decay time .more directly we can access also the first cumulants of the chosen exchange statistics using a bypassing approach not requiring numerical coarse - graining , similar to the one introduced in ref .we thus have , as the first cumulant ( moment ) , the mean current flowing through the system @xmath120\right\ } = -\mathbf{x}^{t}\boldsymbol{\mathcal{l}}_{0}'\tilde{\boldsymbol{\rho}}\ ] ] with @xmath121 the steady - state solution of the density matrix @xmath122 and @xmath123 the steady - state solution of its vectorized version and with @xmath124 .the second cumulant is related to the variance of the current crossing the device , and can be determined as @xmath125 + 2\mathcal{l}_{0}'[\tilde{\rho}']\right\ } = \mathbf{x}^{t}\left(\boldsymbol{\mathcal{l}}_{0}''\tilde{\boldsymbol{\rho}}+2\boldsymbol{\mathcal{l}}_{0}'\tilde{\boldsymbol{\rho}}'\right),\ ] ] where @xmath126 is the steady - state solution of the first order biased density matrix evolving according to @xmath127 and where @xmath128 is the identity matrix . notice that , by definition , we have @xmath129 . according to this framework @xcite , after finding the steady state solution of the density matrix @xmath123 we can access the mean current , while by solving @xmath123 and @xmath130 we access the variance of the current flux . in order to understand the dynamics taking placewe now assume that the incoherent hopping strength @xmath131 is independent of the other parameters .the idea here is to find out which range of parameters gives rise to the largest enhancement due to the presence of the vibrational mode .in order to do so we distinguish between six different cases depending on the leads configuration , as summarized in the table reported in fig .[ fig : cases ] * a. * , where we have set @xmath132 . with the value taken by @xmath133 reported in the third column of the table . *b. * illustration of the energy diagram corresponding to case 5.[fig : cases],scaledwidth=50.0% ] for simplicity we consider here the case where the electronic leads are at zero temperature , leading to the following simplification @xmath134 where @xmath135 is the heaviside function of argument @xmath136 . in this conditionwe find that cases 1 and 3 , with no bias voltage applied between the two leads and with their chemical potential respectively below @xmath137 and above @xmath138 , give a zero mean current .conversely case 2 , where @xmath139 lies between @xmath137 and @xmath138 , corresponds to a non - zero net flux from @xmath4 to @xmath5 .this case is of special interest as it allows for conduction without need of bias voltage .this derives from the conformation of the system .in fact , such a net flux is also present without taking into account the molecular vibration .however , the presence of the vibrational mode gives rise to a significant enhancement on the conduction process , as shown in fig .[ fig : cases2 ] , where we present the difference between the net flux crossing the system with and without incoherent hopping as a function of @xmath140 and @xmath141 .( where @xmath142 is the net exchange neglecting incoherent hopping , @xmath143 ) , describing the impact of the presence of the vibrational mode . in the vertical axiswe have as a parameter the splitting between the energy level of each site @xmath144 , normalized by the coupling strength to the leads @xmath145 .the horizontal axis corresponds to coupling strength @xmath141 between the two physical electronic sites , normalized by @xmath145 .the black dashed line represents the maximum enhancement for given @xmath141 and @xmath140 .the parameter are such as @xmath146[fig : cases2],scaledwidth=45.0% ] as we can see , in case 2 the maximum enhancement occurs for @xmath147. moreover , the enhancement gets more significant as @xmath141 and @xmath140 tend to zero .this comes from the fact that for @xmath148 we have clearly @xmath149 if @xmath150 , while for @xmath151 and @xmath132 , @xmath152 tends to be constant .consequently , in this case , for small @xmath140 and @xmath141 , the conduction process is dominated by incoherent hopping induced by the vibrational mode . in this casewe find that the relation @xmath132 gives rise to the smallest fano factor ( @xmath153 ) depending on @xmath141 and @xmath140 , which is smaller than 1 and converging monotonously to 1 while increasing @xmath141 and @xmath140 , which indicates anti - bunching of the electron emission .the remaining three cases all include a bias voltage ( cf .[ fig : cases ] ) . in all of themwe have observed a similar conduction enhancement provided by the presence of the vibrational mode .the only difference with respect to fig .[ fig : cases2 ] is the ratio between @xmath141 and @xmath140 associated to the maximal enhancement , which are as listed in fig .[ fig : cases ] .* a.*. however , while the quantitative enhancement is very similar in all such cases , the qualitative behaviours are not necessary similar . for example , case 4 may present reversed conduction flux in some region of the parameter space ( for @xmath154 ) .cases 5 and 6 , conversely , present always a positive current , whereas in case 2 an almost constant current is observed for @xmath132 . considering the reduction of the fano factor ,its behaviour does not generally correspond to the maximal enhancement . except for case 2, the smallest fano factor does not coincide to @xmath132 with @xmath133 as listed in fig .[ fig : cases ] .* a.*. the smallest fano factor was observed in both cases 5 and 6 for @xmath155 . as the main objective of our investigation was the demonstration of conduction enhancement arising from the presence of the vibrational mode, we will not go deeper into the analysis of the six cases presented herein .instead , we now focus on a more physical configuration where the interdependence of the parameter is fully taken into account .in this section we will draw a less systematic but more practical picture of the situation under scrutiny to highlight the relevance of the effect induced by the vibrational mode on the electric conduction .firstly , from a device perspective we will illustrate how the enhancement takes place under realistic conditions and how the control of the vibrational mode can lead to a control of the electronic flux .next we will link the considered system to ( i ) high efficiency light - energy harvesting in molecular photosynthetic complexes and ( ii ) high - efficiency molecular identification in olfaction processes .both these classes of processes are currently believed to benefit from quantum features @xcite .[ fig : net - exchange - as ] shows the mean current @xmath156 passing through the wire as a function of the applied biased @xmath157 , which is the most natural and accessible control parameter to adjust , and for different values of the incoherent hopping strength @xmath131 . in the insetwe represent the corresponding fano factor .we notice the appearance of a plateau in the conduction in correspondence of the passage from one of the cases described in the previous section to the next .the smoothing of the edges of those plateaux directly results from the finite temperature used for the electronic leads .panel * a. * corresponds to a case where the coherent hopping strength @xmath141 between the left and right sites is one order of magnitude less than for the lower panel * b. * , while keeping the ratio @xmath158 identical . from these two graphswe can clearly see that the enhancing effect is stronger as we decrease @xmath141 ( and @xmath140 ) .the picture regarding the fano factor , on the other hand , is less straightforward . for high @xmath141 ,the changes are relatively small and monotonic with respect to @xmath131 . for smaller @xmath141the changes induced by varying @xmath131 are more complex .the spike observed close to @xmath159 for some line , corresponds to cases where the mean emission @xmath160 changes sign .notice that the dependence on the bias voltage is normalized with respect to @xmath141 to focus on the enhancing effect provided by the vibrational mode .this renormalisation is at the origin of a more important smoothing effect of the edge due to the temperature .notice also that at zero bias we have conduction taking place , and this conduction process is not intrinsic to the presence of the vibrational mode even if it is strongly enhanced by it . * a. * + for @xmath161 for various values of the incoherent hopping strength @xmath131 ( respectively @xmath162 , @xmath163 , @xmath164 and @xmath165 for the blue , yellow , green and red curves ) .inset represents corresponding fano factor .panel * a. * is for @xmath166 , while panel * b. * is for @xmath167 .other parameters are such as @xmath168 , @xmath169 , and @xmath170[fig : net - exchange - as],title="fig:",scaledwidth=40.0% ] * b. * + for @xmath161 for various values of the incoherent hopping strength @xmath131 ( respectively @xmath162 , @xmath163 , @xmath164 and @xmath165 for the blue , yellow , green and red curves ) .inset represents corresponding fano factor .panel * a. * is for @xmath166 , while panel * b. * is for @xmath167 .other parameters are such as @xmath168 , @xmath169 , and @xmath170[fig : net - exchange - as],title="fig:",scaledwidth=40.0% ] however @xmath141 is , in principle , not easily accessible , in particular in light of its dependence on other relevant parameters of the system . indeed, we have @xmath171 where @xmath12 is the oscillator frequency , @xmath25 its damping rate and @xmath15 is the driving force ( if any ) , @xmath172 is the bare hopping rate , independent of the molecular vibrational properties [ @xmath173 order expansion of @xmath1 ] , while @xmath175 is the coupling strength between the oscillator position and the electronic hopping .it is worth reminding that , as shown previously , a key requirement to maximize the enhancement effect is to get the ratio @xmath158 as in fig .[ fig : cases ] * a. * with @xmath140 and @xmath141 as small as possible . given eq ., fine tuning of @xmath141 can be done through modulating the amplitude @xmath15 of the driving force applied to the oscillator .the other parameter directly connected to the oscillator properties is the incoherent hopping strength @xmath176 where @xmath177 - 1\right)^{-1}$ ] corresponds to the density of excitation in the bosonic bath connected to the vibrational mode . from eq .we see that raising the temperature of the bath @xmath178 leads to a direct increase of @xmath131 .consequently , by playing with a driving force @xmath15 applied to the oscillator and the temperature of the corresponding bath @xmath178 , one can independently manipulate both @xmath141 and @xmath131 .notice also that the energy level of the system depends on @xmath179 .this interdependence of the key parameters makes the previous systematic approach difficult to sustain in this context .nevertheless , eq . suggests that , for the amplitude @xmath180 of the driving force , the effective hamiltonian of the wire eq . leads to a short - circuit scenario where the two electronic sites become uncoupled , thus inhibiting the conduction through the wire . in fig .[ fig : real ] we plot the mean current through the device as a function of the voltage bias @xmath157 and the driving force @xmath15 . as @xmath157 is varied , we clearly identify the two situations illustrated in the previous section . more specifically: for @xmath181 we have a strongly conducting regime for high @xmath157 . if , starting from this point in the parameter space , we increase slightly the applied driving force toward @xmath182 , we see a strong modification of the conduction , which leads to the short - circuit regime . differently from the mechanism highlighted in ref .@xcite , the switching on and off of the conduction does not require a complex driving of the electronic sites , although both our scheme and ref .@xcite allow for a non - zero conduction even if no voltage bias is applied between the electronic leads [ cf .[ fig : real ] ] . ) as a function of the voltage bias @xmath157 and the driving force @xmath15 .the elected parameters are such as @xmath183 , @xmath184 , @xmath185 , @xmath186 and @xmath187 .[ fig : real],scaledwidth=45.0% ] consequently , beside allowing for an independent control of the key parameters @xmath131 and @xmath141 ( through the control of @xmath15 and @xmath178 , respectively ) , in our model the conduction can be switched on and off through a simple mechanical pressure on the molecule . herewe discuss possible connections that the enhancement effect has with existing quantum descriptions of biological processes , and more specifically energy harvesting in photo - systemic molecular complex and olfaction @xcite .to better mimic biological conditions we will consider only a situation where no voltage bias and driving force are applied . even in such conditions ,the interesting features observed previously are strong making this model a good candidate to explain these highly efficient biological processes .molecular complexes are highly efficient in the conduction of charge produced by photosynthesis @xcite .the general idea is that the transport across the molecular complex is in some way enhanced through interaction with vibrational modes of the molecule . among the different phenomena which were considered to explain this high efficiencyone strong candidate is known as noise - assisted transport @xcite .this phenomenon , related to stochastic resonance @xcite , assumes that the fluctuations introduced in the electronic channel by the vibrational mode are responsible for enhanced conductivity .this corresponds exactly to what we reported in the previous section .the higher @xmath26 , the bigger will be @xmath131 the incoherent hopping strength and then lead to an enhanced conductivity . to illustrate such effect , in fig .[ fig : real-1 ] we report the current through the device as a function of @xmath175 .the various lines refer to different vibrational and lead temperatures .as discussed previously , an increase of the vibrational mode s temperature , e.g. fluctuations of larger amplitudes , leads to an enhanced conduction . in order to maximize this effect , we need to fulfill the criteria detailed in sec . [ sub : systematic ] ( @xmath188 and @xmath140 small ) .the enhancement effect appears to be present in a wide range of values of @xmath175 , even weak ones .notice that for large @xmath175 the adiabatic approach used above ceases to be justified .finally , the difference between the full and dashed lines indicates that a further increase of the leads temperature results in a drop of the conductivity of the wire .however , the enhancement effect appears to be robust against the leads temperature , making this model a very good candidate to explain light - energy harvesting in photosynthesis complex . unlike other existing microscopic model presenting thermal - assisted transport ,the effect here is strong and robust .this is because the fluctuation of the vibration mode ( @xmath26 ) only affects the incoherent hopping @xmath189 and not the other key parameter .neither @xmath141 nor the energy eigenstates of the system depend on it , which make highly feasible the observation of such effect in actual molecular electronic systems . ) as a function of the coupling strength to the oscillator @xmath175 ( logarithmic scale ) for different temperature .full lines correspond to leads temperature @xmath190 while the dashed lines to warmer situation @xmath191 .the three colored lines , blue , green and red respectively correspond to a vibrational mode at equilibrium with a thermal bath at temperature @xmath192 , @xmath164 and @xmath165 where @xmath187 . on the lower panel mean emission ( @xmath152 in logarithm scale ) as a function of the vibrational frequency @xmath12 ( logarithm scale ) .as previously the full line corresponds to leads at temperature @xmath193 and @xmath184 , while the dashed line to @xmath194 and @xmath195 .the other parameters are such as @xmath196 , @xmath197 , @xmath198 , @xmath199 and @xmath200 .[ fig : real-1],scaledwidth=45.0% ] ) as a function of the coupling strength to the oscillator @xmath175 ( logarithmic scale ) for different temperature .full lines correspond to leads temperature @xmath190 while the dashed lines to warmer situation @xmath191 .the three colored lines , blue , green and red respectively correspond to a vibrational mode at equilibrium with a thermal bath at temperature @xmath192 , @xmath164 and @xmath165 where @xmath187 . on the lower panel mean emission ( @xmath152 in logarithm scale ) as a function of the vibrational frequency @xmath12 ( logarithm scale ) .as previously the full line corresponds to leads at temperature @xmath193 and @xmath184 , while the dashed line to @xmath194 and @xmath195 .the other parameters are such as @xmath196 , @xmath197 , @xmath198 , @xmath199 and @xmath200 .[ fig : real-1],scaledwidth=45.0% ] the high efficiency of smell receptors has attracted attention , recently @xcite .an accredited model of the way such receptors work consists of molecules taking physically place between two branches of the receptor .each one of these branches acts as an electronic lead .an electronic current would then have to go through or near the molecule .this proximity makes the current exchange strongly sensitive to the molecular properties , thus allowing for its characterization @xcite .such description @xcite is very close to the model explored in this paper .however , the microscopic coupling between electronic part and vibrational part in our system is rather different to the one considered in ref . @xcite . in the latter ,an electron was assumed to pass physically through a molecule eigenstate carrying out necessary information to analyze the molecule , whereas in our case the molecule only affects the way the hopping processes take place .moreover we do not need to assume any electronic bias between the two electronics leads . in the lower panel of fig .[ fig : real-1 ] we plot the mean current in logarithmic scale as a function of the mechanical mode frequency @xmath12 . in the condition considered we see that the emission depends strongly on @xmath12 varying of many order of magnitude . as every key parameter of the effective model depends on this frequency , tuning it in suitable ways leads to a strong dependence on the emission as pictured in the lower panel of fig .[ fig : real-1 ] .if for the given frequency the system is at the verge of one of cases previously detailed , changing @xmath12 may change the system configuration therefore affecting dramatically the conductance . in the caseconsidered the molecule acts as a band - stop filter which can be used to spectrally distinguish different molecules .we have presented a detailed study of the conduction properties of a molecular wire coupled to a vibrational system via electronic hopping exchange .this description is inspired by the idea that physically the vibrational mode does not need to change the energetic structure of the electronic part but can just perturb the exchange taking place on this subsystem .we showed that the presence of the vibrational system can give rise to strong enhancement of the wire conductivity .moreover through the control of the vibrational properties ( temperature and position ) one can accurately control the electronic flux crossing the device .an increase of the temperature enhances the conduction , while the control of the equilibrium position of the oscillator can switch on and off the conduction .we have suggested that the identified mechanism can be related to two biological quantum processes in light of the strong similarities between our model and mechanisms for noise - assisted transport .such connections deserve more analysis , which is the objective of forthcoming studies .we thank the eu fp7-funded collaborative project thermiq , the john templeton foundation ( grant number 43467 ) , the julian schwinger foundation ( grant number jsf-14 - 7 - 0000 ) , the uk epsrc ( grants ep / m003019/1 and ep / p00282x/1 ) and the french anr ( anr-16-achn-0027 ) .l. venkataraman , klare , nuckolls , hybertsen , steigerward , dependence of single molecule junction conductance on molecular conformation , http://dx.doi.org/10.1038/nature05037[_nature_ * 442 * , 904 ( 2006 ) ] .a. gambetta , c. manzoni , e. menna , m. meneghetti , g. cerullo , g. lanzani , s. tretiak , a. piryatinski , a. saxena , r. l. martin and a. r. bishop . real - time observation of nonlinear coherent phonon dynamics in single - walled carbon nanotubes .* 2 * , 515 ( 2006 ) ] .s. tretiak , a. saxena , r. l. martin and a. r. bishop , conformational dynamics of photoexcited conjugated molecules .http://dx.doi.org/10.1103/physrevlett.89.097402 [ _ phys .lett . _ * 89 * , 097402 ( 2002 ) ] .f. santandrea , l. y. gorelik , r. i. shekhter and m. jonson .cooling of nanomechanical resonators by thermally activated single - electron transport .lett . _ * 106 * , 186803 ( 2011 ) ] .s. zippilli , g. morigi , and a. bachtold .cooling carbon nanotubes to the phononic ground state with a constant electron current .http://dx.doi.org/10.1103/physrevlett.102.096804 [ _ phys ._ * 102 * , 096804 ( 2009 ) ] .r. hrtle and m. thoss , resonant electron transport in single - molecule junctions : vibrational excitation , rectification , negative differential resistance , and local cooling .http://dx.doi.org/10.1103/physrevb.83.115414 [ _ phys .b _ * 83 * , 115414 ( 2011 ) ] . s. walter , b. trauzettel and thomas l. schmidt , transport properties of double quantum dots with electron - phonon coupling .http://dx.doi.org/10.1103/physrevb.88.195425 [ _ phys .b _ * 88 * , 195425 ( 2013 ) ] . d. boese and h. schoeller , influence of nanomechanical properties on single - electron tunneling : a vibrating single - electron transistor .http://dx.doi.org/10.1209/epl/i2001-00367-8 [ _ epl _ * 54 * , 668 ( 2001 ) ] .chen , r. l and b .- f .zhu , effects of electron - phonon interaction on nonequilibrium transport through a single - molecule transistor , http://dx.doi.org/10.1103/physrevb.71.165324 [ _ phys .b _ * 71 * , 165324 ( 2005 ) ] . j. koch , m. semmelhack , f. von oppen , and a. nitzan , current - induced nonequilibrium vibrations in single - molecule devices .http://dx.doi.org/10.1103/physrevb.73.155306 [ _ phys .b _ * 73 * , 155306 ( 2006 ) ] .a. yar , a. donarini , s. koller and m. grifoni , dynamical symmetry breaking in vibration - assisted transport through nanostructures .b _ * 84 * , 115432 ( 2011 ) ] . y. kim , h. song , f. strigl , h .- f .pernau , t. lee , and e. scheer , conductance and vibrational states of single - molecule junctions controlled by mechanical stretching and material variation .http://dx.doi.org/10.1103/physrevlett.106.196804 [ _ phys ._ * 106 * , 196804 ( 2011 ) ] .m. l. perrin , c. j. o. verzijl , c. a. martin , a. j. shaikh , r. eelkema , j. h. van esch , j. m. van ruitenbeek , j. m. thijssen , h. s. j. van der zant , d. duli , large tunable image - charge effects in single - molecule junctions .nanotechnol . _* 8 * , 282 ( 2013 ) ] .u. harbola , m. esposito and s. mukamel , quantum master equation for electron transport through quantum dots and single molecules .http://dx.doi.org/10.1103/physrevb.74.235309 [ _ phys .b _ * 74 * , 235309 ( 2006 ) ] . s. pigeon , l. fusco , a. xuereb , g. de chiara and m. paternostro , thermodynamics of trajectories of a quantum harmonic oscillator coupled to @xmath201 baths .http://dx.doi.org/10.1103/physreva.92.013844 [ _ phys .rev . a _ * 92 * , 013844 ( 2015 ) ] . f. levi , s. mostarda , f. rao and f. mintert .quantum mechanics of excitation transport in photosynthetic complexes : a key issues review . ttp://dx.doi.org/10.1088/0034 - 4885/78/8/082001 [ _ rep .phys . _ * 78 * , 082001 ( 2015 ) ] r. de j. len - montiel and j. p. torres , highly efficient noise - assisted energy transport in classical oscillator systems .http://dx.doi.org/10.1103/physrevlett.110.218101 [ _ phys .lett . _ * 110 * , 218101 ( 2013 ) ] .s. viciani , m. lima , m. bellini and f. caruso , observation of noise - assisted transport in an all - optical cavity - based network .http://dx.doi.org/10.1103/physrevlett.115.083601 [ _ phys .* 115 * , 083601 ( 2015 ) ] .j. c. brookes , f. hartoutsiou , a. p. horsfield and a. m. stoneham .could humans recognize odor by phonon assisted tunneling ?_ , * 98 * 038101 ( 2007 ) ] .[ [ appendix-1-dissipation - of - the - wire - to - the - leads - and - biased - contribution - to - the - evolution - subdissipation - of - the ] ] appendix 1 : dissipation of the wire to the leads and biased contribution to the evolution [ sub : dissipation - of - the ] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ in this appendix we detail the calculation leading to the lindblad form dissipator induced by coupling the wire subsystem to fermionic leads as explained in section [ sub : coupling - to - electronic ] . based on refs .@xcite we determine the the dissipator in the manybody picture such as @xmath202 ( with @xmath203 or @xmath4 ) @xmath204 & = & \frac{\pi}{4}\gamma^{\nu}n_{\nu}(\epsilon_{a})\vert t_{2,a}^{\nu}\vert^{2}\bigg[f_{\nu}(\epsilon_{a})\left(2\hat{c}_{a}^{\dagger}\bullet\hat{c}_{a}-\left\ { \hat{c}_{a}\hat{c}_{a}^{\dagger},\bullet\right\ } \right)+(1-f_{\nu}(\epsilon_{a}))\left(2\hat{c}_{a}\bullet\hat{c}_{a}^{\dagger}-\left\ { \hat{c}_{a}^{\dagger}\hat{c}_{a},\bullet\right\ } \right)\bigg]\\ \tilde{\mathcal{l}}_{a\leftrightarrow3}^{\nu}[\bullet ] & = & \frac{\pi}{4}\gamma^{\nu}n_{\nu}(\epsilon_{a})\vert t_{1,a}^{\nu}\vert^{2}\bigg[f_{\nu}(\epsilon_{a})\left(2\hat{c}_{a}^{\dagger}\hat{c}_{3}\bullet\hat{c}_{3}^{\dagger}\hat{c}_{a}-\left\ { \hat{c}_{3}^{\dagger}\hat{c}_{a}\hat{c}_{a}^{\dagger}\hat{c}_{3},\bullet\right\ } \right)+(1-f_{\nu}(\epsilon_{a}))\left(2\hat{c}_{3}^{\dagger}\hat{c}_{a}\bullet\hat{c}_{a}^{\dagger}\hat{c}_{3}-\left\ { \hat{c}_{a}^{\dagger}\hat{c}_{3}\hat{c}_{3}^{\dagger}\hat{c}_{a},\bullet\right\ } \right)\bigg]\;,\end{aligned}\ ] ] where @xmath205 is the density of state in the lead @xmath71 ( @xmath5 or @xmath4 ) at a given energy and @xmath206 $ ] the fermi distribution of a given lead @xmath71 having @xmath81 as chemical potential . notice that the total dissipation is related to the leads dissipation @xmath207 plus@xmath208 the one induced by the vibrational mode ( eq . ) .each dissipation induced by the leads can be decomposed in terms of amplitude damping and coherence damping channels such as @xmath209 , with for the amplitude part @xmath210=\sum_{{\scriptscriptstyle xy=\{lo , lf , ro , rf\}}}\bigg[\alpha_{xy}\left(\hat{l}_{xy}\bullet\hat{l}_{xy}^{\dagger}-\frac{1}{2}\left\ { \hat{l}_{xy}^{\dagger}\hat{l}_{xy},\bullet\right\ } \right)+\beta_{xy}\left(\hat{l}_{xy}^{\dagger}\bullet\hat{l}_{xy}-\frac{1}{2}\left\ { \hat{l}_{xy}\hat{l}_{xy}^{\dagger},\bullet\right\ } \right)\bigg]\label{eq : la}\;,\ ] ] with @xmath211 , @xmath212 , @xmath213 and @xmath214 and for the coupling strength we have @xmath215\\ \alpha_{lf } & = & \frac{\gamma}{2}\bigg[\left(1-a\right)^{2}\left(f_{l}(\epsilon_{1})\left(1-a\right)^{2}+f_{r}(\epsilon_{1})b^{2}\right)+\left(1+a\right)^{2}\left(f_{l}(\epsilon_{2})\left(1+a\right)^{2}+f_{r}(\epsilon_{2})b^{2}\right)\bigg]\\ \alpha_{ro } & = & \frac{\gamma}{2}\bigg[b^{2}\left(f_{l}(\epsilon_{1})+f_{l}(\epsilon_{2})\right)+\left(f_{r}(\epsilon_{1})\left(1-a\right)^{2}+f_{r}(\epsilon_{2})\left(1+a\right)^{2}\right)\bigg]\\ \alpha_{rf } & = & \frac{\gamma}{2}\bigg[\left(f_{l}(\epsilon_{1})\left(1-a\right)^{2}+f_{l}(\epsilon_{2})\left(1+a\right)^{2}\right)+b^{2}\left(f_{r}(\epsilon_{1})+f_{r}(\epsilon_{2})\right)\bigg]\label{eq : alpha}\;.\end{aligned}\ ] ] for simplicity we assume the leads density of states to be homogenous and identical @xmath83 as for the coupling strength @xmath84 . the coefficients @xmath216 are obtained replacing in the definition above @xmath217 .the effective decoherence is written as @xmath218=\sum_{\begin{array}{c } { \scriptscriptstyle xy=\{lo , lf , ro , rf\}}\\ { \scriptscriptstyle ab=\{ro , rf , lo , lf\ } } \end{array}}\bigg[\alpha_{xy}^{c}\left(\hat{l}_{xy}\bullet\hat{l}_{ab}^{\dagger}-\frac{1}{2}\left\ { \hat{l}_{ab}^{\dagger}\hat{l}_{xy},\bullet\right\ } \right)+\beta_{xy}^{c}\left(\hat{l}_{ab}^{\dagger}\bullet\hat{l}_{xy}-\frac{1}{2}\left\ { \hat{l}_{xy}\hat{l}_{ab}^{\dagger},\bullet\right\ } \right)\bigg]\label{eq : lc}\;.\ ] ] it is worth to notice that those dissipator acting on the coherence are not of a dephasing form because they change the energy of the system .they give rise to or suppress coherences between the 2 intermediate states through absorption or emission of quanta of energy with the leads .the coupling strength attached to those channels are such as @xmath219\\ \alpha_{f}^{c}=\alpha_{lf}^{c}=\alpha_{rf}^{c } & = & \frac{\gamma}{2}\bigg[\left(1-a\right)\left(f_{l}(\epsilon_{1})\left(1-a\right)^{2}+f_{r}(\epsilon_{1})b^{2}\right)-\left(1+a\right)\left(f_{l}(\epsilon_{2})\left(1+a\right)^{2}+f_{r}(\epsilon_{2})b^{2}\right)\bigg]\;.\label{eq : alphac}\end{aligned}\ ] ] in order to retrieve the exchange statistics we define as done in the text ( section [ sub : exchange - statistics ] ) a counting process @xmath108 which is related to a biased contribution of the evolution @xmath220 where @xmath221 & = & \sum_{b=\{o , f\}}\sum_{a=\{l , r\}}\bigg[\left(e^{\left(-1\right)^{b}s}-1\right)\bigg(\alpha_{b}^{rc}\vert a\rangle\langle b\vert\bullet\vert b\rangle\langle\bar{a}\vert+\alpha_{ab}^{r}\vert10\rangle\langle b\vert\bullet\vert b\rangle\langle10\vert\bigg)\nonumber \\ & & + \left(e^{\left(-1\right)^{b+1}s}-1\right)\bigg(\beta_{b}^{rc}\vert b\rangle\langle a\vert\bullet\vert\bar{a}\rangle\langle b\vert+\beta_{ab}^{r}\vert b\rangle\langle a\vert\bullet\vert a\rangle\langle b\vert\bigg)\bigg]\label{eq : ls}\end{aligned}\ ] ] with @xmath222 is the complementary of @xmath102 ( @xmath5 for @xmath4 and reciprocally ) .[ [ appendix-2-superoperator - of - the - effective - electronic - dynamics - subsuperop ] ] appendix 2 : superoperator of the effective electronic dynamics [ sub : superop ] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ here we provide the explicit form of the superoperator responsible for the effective open - system dynamics of the electronic system discussed in sec .[ sec : wire - dynamic ] . with the definitions of the previous appendix, we have @xmath223 with @xmath224 and @xmath144 .the explicit definition of each coefficient is given in appendix [ sub : dissipation - of - the ] .notice that each coefficient @xmath225 and @xmath226 can be easily rewritten in terms of leads contribution such as @xmath227 .

<S> we present a detailed study of the conduction properties of a molecular wire where hopping process between electronic sites is coupled to a molecular vibrational mode . </S><S> the molecule is sandwiched between two electronic leads at finite temperatures . </S><S> we show that the electro - mechanical coupling leads , in some cases , to a strong enhancement of the conduction from one lead to another . moreover , the device can act as a transistor with proper driving of the molecular vibrational mode , passing sharply from enhanced conduction to short - circuit configuration . </S><S> we also relate the system behavior to quantum biological phenomena such as light - energy harvesting and olfaction to show how the original microscopic model considered here can be a good candidate to describe those phenomena . </S><S> molecular electronics was initially dedicated to the study of transport properties of molecule sandwiched between electronic leads . </S><S> it was originally thought as an alternative platform to silicon electronics but it has matured lately as a rich and promising field of research beyond its initial scope @xcite . </S><S> numerous theoretical works have been dedicated to enlighten the behavior of such metal - molecule - metal junction depending on the microscopic description of the molecule . </S><S> most of them have focused on a description including solely connected electronic sites for building devices such as current rectificators @xcite or thermal transistors @xcite for example . </S><S> rich features were predicted such as stochastic pumps @xcite or laser - induced phase - controllable transport @xcite to mention only some recent results . </S><S> based on the rich nature of the molecule under study , more recently an additional ingredient was added to the microscopic description of these devices : the coupling of the conduction properties to the vibrational degrees of freedom of the molecule which is expected to play a crucial role on the transport properties of these systems @xcite . </S><S> this configuration was first considered in order to cool ( through side - band cooling ) the mechanical / vibrational degree of freedom of the molecule @xcite . </S><S> the study of the impact of this coupling on transport was also considered showing negative differential conductance @xcite or other vibration - assisted transport phenomena @xcite . however </S><S> , this extended literature on the subject is only , to our knowledge , based on anderson - holstein - like models . </S><S> the anderson - holstein model consists of electronic sites sandwiched between two electronic leads while the vibrational mode is coupled to an electronic eigenstate , raising or lowering its energy depending on the vibrational state @xcite . in this paper </S><S> we present a similar setup , in which the molecular vibration is not coupled to an electronic eigenstate but to the hopping process taking place between two microscopic electronic sites . consequently </S><S> , as it will be detailed , controlling the vibration properties of the molecule we can tuned the flux of electrons get through the system . </S><S> this allows not only for an enhancement of the conduction , even when no voltage bias is applied between the electronic leads , but also for a switching regime in which the current flowing is deterministically turned on and off . </S><S> moreover , we illustrate how the original microscopic model considered can be a very good candidate for recently proposed quantum biological processes . </S><S> the remainder of this paper is organized as follows : sec . </S><S> [ sec : description ] introduces the system and the working conditions considered throughout our work . in sec . </S><S> [ sub : vibrational - mode - damping ] , we adiabatically eliminate the vibrational degree of freedom to end up with an effective dynamics for the conducting part of the system . </S><S> this analysis is then complemented by tracing out the leads degrees of freedom ( sec . </S><S> [ sub : coupling - to - electronic ] ) . </S><S> sec . </S><S> [ sec : wire - dynamic ] is dedicated to the dynamics and related exchange statistics with the electronic leads . </S><S> we detail the unraveling approach used to access the statistics of exchange taking place between the system and one of the electronic leads . </S><S> we then pass to a systematic study of the mean current flowing through the device as a function of the electronic - lead configuration . </S><S> for such a study , we focus on the low - temperature regime . </S><S> we then switch to the study of more realistic conditions . </S><S> we first illustrate the conduction enhancement together with demonstrating the transistor regime ( sec . [ </S><S> sub : conduction - enhancement ] ) before to connect the observed phenomena to potential quantum biological processes such as light - energy harvesting and olfaction ( sec . </S><S> [ sub : light - energy - harvesting ] ) . in sec . </S><S> [ sub : conclusion ] we summarise our results and comment on other interesting features left unaddressed in our study . </S>

1003.0499

the magnetism of conventional metal is composed of two different contributions , the spin component known as the pauli paramagnetism and the orbital component as the landau diamagnetism . in condensed matter systems ,the orbital magnetism sensitively depends on the detail of the electronic band structure , and sometimes largely deviates from the landau diamagnetism .particularly , narrow gap materials such as graphite@xcite or bismuth @xcite exhibit a singular behavior in the orbital susceptibility near the energy gap . in this paper , we show that the anomalous orbital magnetism in narrow gap systems can be understood in terms of pseudo - spin paramagnetism , which arises from the extra degree of freedom in the orbital motion of electrons .graphene monolayer recently fabricated@xcite is a zero - gap system in which the conduction and valance bands stick together at @xmath0 and @xmath1 points located at inequivalent corners of the brillouin zone , called valleys.@xcite the system is characterized by chiral quasiparticles with opposite chirality in each valley and a linear dispersion reminiscent of massless dirac fermions . at the dirac point where two bands cross each other ,the magnetic susceptibility has a singularity expressed as a delta function in fermi energy @xmath2 and disappears otherwise.@xcite bilayer graphene composed of a pair of graphene layers@xcite has a zero - gap structure with quadratic dispersion,@xcite leading to a less singular , logarithmic peak of the susceptibility.@xcite the orbital magnetism was also studied for related materials , such as graphite intercalation compounds,@xcite carbon nanotube,@xcite few - layer graphenes,@xcite and organic compounds having dirac - like spectrum.@xcite in both monolayer@xcite and bilayer graphenes,@xcite certain asymmetric potential opens an energy gap at the band touching point . in this paper, we calculate the orbital magnetism of several dirac - like systems with gap and show that the pseudo - spin degree of freedom such as valleys in graphene produces paramagnetism in the same manner as the real spin and gives an essential contribution to the singular diamagnetic behavior . in sec .[ sec_mono ] , we calculate the susceptibility of the monolayer graphene with varying gap . the singular susceptibility - change in varying @xmath2is understood in terms of valley - induced paramagnetism .we extend the analysis to the bilayer graphene in sec .[ sec_bi ] and to a three - dimensional dirac system corresponding to bismuth@xcite in sec .[ sec_3d ] .a brief conclusion is presented in sec .[ sec_conc ] .graphene is composed of a honeycomb network of carbon atoms , where a unit cell contains a pair of sublattices , denoted by @xmath3 and @xmath4 .electronic states in the vicinity of @xmath0 and @xmath1 points in the brillouin zone are well described by the effective mass approximation.@xcite let @xmath5 and @xmath6 be the bloch functions at the @xmath0 point , corresponding to the @xmath3 and @xmath4 sublattices , respectively . in a basis @xmath7 , the hamiltonian for the monolayer graphene around the @xmath0 point becomes@xcite @xmath8 where @xmath9 is the velocity , @xmath10 , and @xmath11 with vector potential @xmath12 giving external magnetic field @xmath13 . in the following, we shall completely neglect the spin zeeman energy because the spin splitting is much smaller than landau - level separations .the hamiltonian at the @xmath1 point is obtained by exchanging @xmath14 in eq .( [ eq : h_mono ] ) .the diagonal terms @xmath15 represent the potential asymmetry between @xmath3 and @xmath4 sites , which opens an energy gap at the dirac point.@xcite this can arise when graphene is placed on a certain substrate material .in fact , band - gap opening is observed in graphene epitaxially grown on a sic substrate.@xcite from a theoretical point of view , the singular behavior in ideal graphene with vanishing gap is better understood by taking the limit @xmath16 , as will be shown below .we can safely assume @xmath17 without loss of generality .the energy band at @xmath18 is given by @xmath19 with electron momentum @xmath20 and @xmath21 .the density of states is @xcite @xmath22 where @xmath23 and @xmath24 represent the degrees of freedom associated with spin and valley , respectively , and @xmath25 is a step function , defined by @xmath26 the landau - level spectrum can be found using the relation @xmath27 and @xmath28 , where @xmath29 is magnetic length and @xmath30 and @xmath31 are raising and lowering operators for usual landau - level wave functions , respectively .the eigenenergy at @xmath0 point becomes @xmath32 where @xmath33 and @xmath34 the corresponding wavefunction is @xmath35 where @xmath36 satisfies @xmath37 and @xmath38 is the usual landau - level wave function , where @xmath39 with @xmath40 should be regarded as 0 .the landau level @xmath41 lies just at the top of the valence band , i.e. , @xmath42 because @xmath43 , and its amplitude is only at the @xmath4 site .similarly , for landau levels lying in the vicinity of the valence - band top , i.e. , @xmath44 satisfying @xmath45 , we have @xmath46 , showing that the amplitude of the wave function is significant only at the b site . for those in the conduction band @xmath47 , on the other hand , we have @xmath48 , showing that the amplitude is significant only at the @xmath3 site . for the @xmath1 point , on the other hand, the eigenenergy is given by eq .( [ eq : mono_landauk ] ) with @xmath49 being replaced by @xmath50 and the eigenfunction is given by @xmath51 where @xmath52 is obtained from @xmath36 by replacing @xmath49 by @xmath50 . the landau level@xmath41 lies just at the bottom of the conduction band , i.e. , @xmath53 because @xmath54 , and its amplitude is only at the @xmath3 site .similarly , for low - lying landau levels in the conduction band @xmath55 , we have @xmath56 , showing that the amplitude of the wave function is significant only at the @xmath3 site . for those in the valence band @xmath40 , on the other hand , we have @xmath57 , showing that the amplitude is significant only at the @xmath4 site .the landau levels of @xmath58 are doubly degenerate between the @xmath0 and @xmath1 valleys , while those of @xmath41 are not .therefore , by defining @xmath59 the thermodynamical potential at temperature @xmath60 then becomes @xmath61 \big(1-\frac{\delta_{n0}}{2}\big ) , \label{eq : omega}\end{aligned}\ ] ] where @xmath62 and @xmath63 , \ ] ] with @xmath64 being the chemical potential . in weak magnetic field , using the euler - maclaurin formula , the summation in @xmath65 in eq .( [ eq : omega ] ) can be written as an integral in continuous variable @xmath66 and a residual term as @xmath67 dx \nonumber\\ & & \qquad - \frac{(\hbar\omega_b)^4}{12 } \frac{\partial \varphi[\varepsilon_s(x)]}{\partial x } \big|_{x=0 } \bigg ) + o(\delta^3 ) .\label{eq : omega_2}\end{aligned}\ ] ] the magnetization is given by @xmath68 and the magnetic susceptibility by @xmath69 in eq .( [ eq : omega_2 ] ) , the first term represents the thermodynamic potential in the absence of a magnetic field and only the second term depends on the magnetic field .we have @xmath70 with @xmath71 where @xmath72 is the fermi distribution function .the susceptibility at zero temperature is given by @xmath73 with @xmath2 being the fermi energy .in the limit of @xmath74 , the susceptibility approaches a delta function @xmath75 in agreement with the previous result.@xcite the susceptibility and the density of states , given by eqs .( [ eq : chi_mono ] ) and ( [ eq : densityofstatesinmonolayergraphene ] ) , respectively , are shown in fig .[ fig_chi_mono ] .note that the upward direction represents negative ( i.e. , diamagnetic ) susceptibility .the susceptibility is not zero at zero electron density , @xmath76 , because the completely filled valence band gives a constant diamagnetic susceptibility .when the fermi energy enters the conduction band , the susceptibility jumps downs to zero , resulting in zero total magnetism . in the limit of @xmath74 , the step height at the band edge increases in proportion to @xmath77 and the susceptibility approaches a delta function . .note that the upward direction represents negative ( i.e. , diamagnetic ) susceptibility . ]because the hamiltonian is equivalent to that of a dirac electron with a nonzero mass , the magnetic susceptibility around the band edge should correspond to that of a conventional electron .this is clearly illustrated by the effective hamiltonian expanded in the vicinity of @xmath78 .for the conduction band , @xmath79 , the effective hamiltonian for the a site near the band bottom @xmath80 is written apart from the constant energy as @xmath81 where @xmath82 is the bohr magneton , given by @xmath83 with @xmath84 being the free electron mass , and we used the relation @xmath85 = i\hbar e b / c$ ] and defined @xmath86 the last term in each hamiltonian can be regarded as the pseudo - spin zeeman term , where the different valleys @xmath0 and @xmath1 serve as pseudo - spin up ( @xmath87 ) and down ( @xmath88 ) , respectively .this agrees with the zeeman energy expected for an intrinsic magnetic moment , that originates from the self - rotation of the wave packet in bloch electron .@xcite the combined hamiltonian is written as @xmath89 obviously , the pseudo - spin zeeman term gives the pauli paramagnetism and the first term containing @xmath90 gives the landau diamagnetism in the usual form as @xmath91 with density of states @xmath92 the total susceptibility @xmath93 actually agrees with the amount of the jump at the conduction band bottom in @xmath94 of eq .( [ eq : chi_mono ] ) .because @xmath95 in the present case , we have @xmath96 as in the free electron , giving the paramagnetic susceptibility in total .therefore the susceptibility exhibits a discrete jump toward the paramagnetic direction when the fermi energy moves off the dirac point .the jump height goes to infinity as the gap closes , because the susceptibility is inversely proportional to the effective mass . .dashed lines connecting the levels of @xmath0 and @xmath1 represent corresponding levels with opposite pseudo - spins . ] in the original hamiltonian , the landau - level energies in eq .( [ eq : mono_landauk ] ) can be rewritten as @xmath97 figure [ fig_mono_ll ] shows energy levels for @xmath98 and the relationship between the different labeling schemes of eqs .( [ eq : mono_landauk ] ) and ( [ eq : mono_landau_uni ] ) . forthe conduction band , the levels of the same @xmath99 with opposite pseudo - spins @xmath100 share the same landau level function @xmath101 on the @xmath3 site , on which the states near the conduction - band bottom ( @xmath102 ) have most of the amplitude as has been discussed above .for the valence band , similarly , @xmath99 describes the index of the landau - level function at the @xmath4 site .bilayer graphene is a pair of graphene layers arranged in ab ( bernal ) stacking and includes @xmath103 and @xmath104 atoms on layer 1 and @xmath105 and @xmath106 on layer 2.@xcite the low energy states are again given by the states around @xmath0 and @xmath1 points in the brillouin zone .the hamiltonian at the @xmath0 point for the basis @xmath107 , @xmath108 is given by @xmath109 where @xmath110 describes potential difference between layer 1 and 2 ( not @xmath3 and @xmath4 sites ) and @xmath111 represents interlayer coupling between @xmath104 and @xmath105.@xcite the hamiltonian at the @xmath1 point is obtained by exchanging @xmath14 in eq .( [ eq : h_bi ] ) .the energy band at @xmath18 is given by @xmath112^{1/2 } \big)^{1/2 } , \label{eq : bilayer_band}\end{aligned}\ ] ] with @xmath113.@xcite the index @xmath114 and @xmath115 give a pair of bands further and closer to zero energy , respectively , and @xmath79 and @xmath115 in each pair represent the electron and hole branches , respectively .the band - edge energies corresponding to @xmath116 are given by @xmath117 for @xmath118 , where @xmath119 for @xmath120 , the band minimum becomes @xmath121 which corresponds to an off - center momentum.@xcite the density of states diverges here as @xmath122 .the energy bands and the density of states with several @xmath110 s are plotted in figs .[ fig_chi_bilayer ] ( a ) and ( b ) , respectively .vertical lines in ( a ) indicate the energies of @xmath123 , @xmath124 , and @xmath125 for @xmath126 . , 0.2 , and 0.5 . vertical lines in ( a ) indicate the energies of @xmath123 , @xmath124 and @xmath125 for @xmath127 .the upward direction represents negative ( i.e. , diamagnetic ) susceptibility in ( c ) .] in a magnetic field , the eigenfunction of the hamiltonian at the @xmath0 point is written as @xmath128 with integer @xmath129 . for @xmath130 , the hamiltonian matrix for @xmath131 becomes@xcite @xmath132 for @xmath41 ,the first component does not actually exist because @xmath133 .the matrix for @xmath134 becomes @xmath135 for @xmath136 , only the component @xmath137 survives and the hamiltonian is @xmath138 for the @xmath1 point , the eigenfunction is written as @xmath139 . for @xmath130 ,the hamiltonian matrix for @xmath131 is @xmath140 for @xmath41 , the matrix for @xmath141 becomes @xmath142 and for @xmath136 , that for @xmath143 is @xmath144 ) with @xmath145 and @xmath146 .the dashed horizontal lines represent the landau level which originally belongs to @xmath136 at opposite valley .the quantum number @xmath147 is indicated below @xmath148 . ]if we extend the definition of the matrix of eq .( [ eq : bilayer_matrix_k ] ) to @xmath41 , its three eigenvalues agree with those of @xmath149 and the rest with that of @xmath150 .similarly , the matrix of eq .( [ eq : bilayer_matrix_k2 ] ) with @xmath41 gives eigenvalues of @xmath151 and @xmath152 .thus we can use eqs .( [ eq : bilayer_matrix_k ] ) and ( [ eq : bilayer_matrix_k2 ] ) with @xmath153 to produce the full spectrum . by introducing the pseudo - spin variable @xmath100 ,the hamiltonian is combined into a single expression , @xmath154 with @xmath155 and @xmath156 we write the eigenvalues of @xmath157 as @xmath158 in the ascending order in energy ( @xmath159 and 2 for valence bands and @xmath160 and 4 for the conduction bands ) .the second argument in @xmath161 represents the dependence on @xmath4 which are not included in @xmath162 .figure [ fig_bi_ll ] shows the example of the landau - level spectrum at @xmath163 and @xmath164 , where the thick dashed lines represent the landau level which originally belongs to @xmath136 at opposite valleys .the correspondence between quantum numbers @xmath148 and @xmath147 are indicated in the figure .the thermodynamic potential becomes @xmath165 \nonumber\\ & = & -\frac{1}{\beta } \frac{g_s}{4\pi \hbar^2 v^2}\ ! \sum_{\xi , j } \bigg [ \ ! \int_{0}^\infty \!\ !\varphi\big[\varepsilon_{j}(x,\xi\delta ) \big ] dx \nonumber\\ & & + \frac{\delta^2}{24 } \frac{\partial \varphi[\varepsilon_{j}(x,0)]}{\partial x } \big|_{x=0 } \bigg ] + o(\delta^3 ) , \label{eq : omega_bilayer}\end{aligned}\ ] ] where we used the euler - maclaurin formula in the second equation .the first term in the bracket can be transformed by changing the integral variable from @xmath66 to @xmath166 as @xmath167 dx = \int_{-\infty}^\infty \!\ !f(\varepsilon ) \ , n_{j}(\varepsilon,\xi\delta ) d\varepsilon,\ ] ] where we used @xmath168 and defined @xmath169 where @xmath170 is a real and positive solution of @xmath171 and @xmath172 if there are more than one solution of @xmath173 , we regard @xmath174 as their sum . the quantity @xmath175 represents the electron density below @xmath166 for the conduction band and the hole density above @xmath166 for the valence band .by expanding @xmath176 we can further expand @xmath177 of eq .( [ eq : omega_bilayer ] ) in terms of @xmath178 .we have @xmath179 \frac{\partial \varepsilon_{j}(x,0)}{\partial x } \big|_{x=0 } \biggr ] . \quad\label{eq : chi_form}\end{aligned}\ ] ] for the hamiltonian of eq .( [ eq : bilayer_matrix ] ) , the eigenequation @xmath180 can be solved for @xmath66 @xmath181 as @xmath182 which gives @xmath183 when being real and positive .let us first consider the case @xmath184 , where two conduction bands are occupied by electrons .in this case @xmath185 are both real and positive and we have @xmath186 , @xmath187 , and @xmath188 .then , we have @xmath189 independent of @xmath190 .therefore , @xmath191 identically vanishes , resulting in susceptibility independent of energy in the region @xmath192 .the same is true for @xmath193 . because @xmath194 for @xmath195 , i.e. , in the case of empty or filled band , we can conclude that the susceptibility vanishes for @xmath184 and @xmath196 independent of interlayer interaction @xmath111 and asymmetry @xmath110 . similarly , the density of states for @xmath197 is independent of @xmath111 and @xmath110 and becomes twice as large as that of monolayer .in fact , we have @xmath198 this feature of the density of states is apparent in fig . [ fig_chi_bilayer ] ( b ) . in the vicinity of the bottom of the excited conduction band , @xmath199, we have @xmath200_{\delta=0 } , \label{eq : n4}\end{aligned}\ ] ] where we used @xmath201 and @xmath202 . using eq .( [ eq : chi_form ] ) , we find that the susceptibility makes a discrete jump at @xmath125 as @xmath203 where the first term in the bracket comes from the integral of the delta function in eq .( [ eq : n4 ] ) and the second term from the step function in eq .( [ eq : chi_form ] ) . near @xmath125, the eigenstates are given primarily by the dimer states composed of @xmath204 and @xmath205 .the effective hamiltonian is described by the second order in interband interation with the conduction - band bottom @xmath206 and the valence - band top @xmath207 , where each process gives a term @xmath208 or @xmath209 . in symmetric bilayer with @xmath210 , the terms @xmath211 and@xmath212 have the same coefficient and the pseudo - spin zeeman term identically vanishes .when @xmath110 becomes nonzero , the two coefficients shift from each other linearly in @xmath110 because of the band - gap opening , leading to a nonzero zeeman term .the resulting effective hamiltonian is given by eq .( [ eq : h_conv ] ) with @xmath213 the susceptibility is written as pauli and landau magnetism in eqs .( [ eq : chi_conv - p ] ) and ( [ eq : chi_conv - l ] ) , respectively , which together give a susceptibility jump of eq .( [ eq : chi_bi_jump ] ) .the paramagnetic component @xmath214 is zero at @xmath210 and monotonically increases as @xmath110 becomes larger . at @xmath215 or @xmath216 ,@xmath214 exceeds @xmath217 and the susceptibility step changes from diamagnetic to paramagnetic . in the limit @xmath218, we have @xmath219 as in the monolayer .this is to be expected , because the bilayer graphene in this limit can be regarded as a pair of independent monolayer graphenes , where interlayer coupling @xmath111 opens an energy gap at each dirac point .similar argument also applies to the behavior around @xmath124 . in the energy region @xmath220 near the top of the valence band , both @xmath221 and @xmath222 are real and positive , giving the states at outer and inner equi - energy circle of the band @xmath223 , respectively. then we have @xmath224^{3/2}}.\end{aligned}\ ] ] when the energy approaches to @xmath225 from the negative side , the integral of @xmath226 , thus the susceptibility , diverges in positive direction as @xmath227 in the same manner as the density of states .the same divergence occurs at the bottom of the conduction band , @xmath228 , because of the electron - hole symmetry .full analytic expression of the susceptibility @xmath229 is complicated and presented in appendix [ sec_app ] .figure [ fig_chi_bilayer ] ( c ) plots the susceptibility for @xmath230 , 0.2 , and 0.5 . in accordance with the above analytic consideration , we actually observe that the susceptibility vanishes in the region @xmath231 and @xmath232 and that the susceptibility step at @xmath199 changes from diamagnetic to paramagnetic with increasing @xmath110 .we also see that the susceptibility for @xmath233 diverges in the paramagnetic direction at @xmath234 .the results in monolayer graphene in sec . [ sec_mono ] can be directly extended to three - dimensional dirac hamiltonian , which is also known to describe the approximate electronic structure of bismuth with strong spin - orbit interaction.@xcite in bisumuth , electronic states near the fermi level is approximately described by a @xmath235 matrix hamiltonian , given by @xmath236 where four components consist of two orbital and two spin degerees of freedom . the density of states at zero magnetic field is @xmath237 where @xmath238 is the valley degeneracy allowing the presence of different @xmath239 points described by the above hamiltonian in the first brillouin zone .the landau levels in a uniform magnetic field in @xmath240 direction are given by @xmath241 with @xmath242 , @xmath243 , and @xmath244 .this is equivalent to the two - dimensional dirac system , eq .( [ eq : mono_landau_uni ] ) , when the term @xmath245 is replaced with @xmath246 .the susceptibility @xmath229 is thus calculated by integrating eq . ( [ eq : chi_mono ] ) in @xmath247 as @xmath248 where @xmath249 is a cut - off energy . in the limit of @xmath74 ,the susceptibility at zero energy logarithmically diverges . at an energy @xmath166 just above the band bottom @xmath250, we obtain the paramagnetic contribution @xmath251 where @xmath252 with @xmath253 .this is nothing but the magnetic susceptibility , dominated by the pauli paramagnetism , of a three - dimensional metal with mass @xmath254 and @xmath255 factor @xmath219 .figure [ fg_chi_3d_dirac ] shows the susceptibility and the density of states in the present system .the singular decrease of the susceptibility at the band edges is fully understood in terms of the appearance of the dominant spin paramagnetism inside the band .we have shown that the orbital magnetism singularly dependent on the fermi level , appearing in narrow gap electronic systems described by the dirac hamiltonian , can be understood in terms of pseudo - spin pauli paramagnetism induced by extra degree of freedom in the orbital motion .this has been demonstrated by explicit calculations of orbital susceptibility in monolayer and bilayer graphenes with band gap and also in three dimensional dirac systems such as bismuth .this work was supported in part by grant - in - aid for scientific research on priority area `` carbon nanotube nanoelectronics , '' by grant - in - aid for scientific research , and by global center of excellence program at tokyo tech `` nanoscience and quantum physics '' from ministry of education , culture , sports , science and technology japan .using eqs . ( [ eq : chi_form ] ) and ( [ eq : x_sol ] ) , the susceptibility of @xmath229 of bilayer graphene with energy gap is calculated as @xmath256 with @xmath257 where @xmath258 , \quad\end{aligned}\ ] ] and @xmath259 } { 4(\gamma_1 ^ 2 + 4\delta^2)^{3/2 } } , \\ \tilde\chi_- \!\ ! & = & \!\ !-f(\varepsilon_- ) + \frac{|\delta|}{3\gamma_1 } , \\ \tilde\chi_+ \!\ !& = & \!\ !-f(\varepsilon_+ ) - \frac{\delta^2\sqrt{\delta^2+\gamma_1 ^ 2}}{\gamma_1(2\delta^2 + \gamma_1 ^ 2 ) } + \frac{2\delta^2 + \gamma_1 ^ 2}{12\gamma_1\sqrt{\delta^2+\gamma_1 ^ 2}}. \qquad \end{aligned}\ ] ] in the symmetric bilayer with vanishing gap , @xmath210 , in particular , we simply get @xmath260 which agrees with the previous results.@xcite

<S> the orbital diamagnetic susceptibility is calculated in monolayer and bilayer graphenes with band gap as well as in three - dimensional dirac systems . </S><S> it is demonstrated that the pseudo - spin degree of freedom such as valleys produces paramagnetic susceptibility in an equal manner as the real spin dominating over the landau diamagnetism . </S><S> the pseudo - spin paramagnetism explains the origin of a singular diamagnetism which is present only in the band - gap region and disappears rapidly inside the conduction and valence bands . </S>

q-bio0412020

[ [ a .- gr1-as - a - satisfiability - problem ] ] a. gr1 as a satisfiability problem + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in this appendix we show how a realization of gr1 can be formulated as a satisfiability problem ( sat ) @xcite , an optimization problem where @xmath0 boolean variables are constrained by @xmath77 conjunctive normal form ( cnf ) constraints ( i.e. by a boolean polynomial constructed as a product ( @xmath78 ) of @xmath77 disjunctive monomials ( @xmath79 ) ) .equation ( [ eq : fp ] ) is equivalent to the xor boolean constraint @xmath80 .this can be recast in cnf , as @xmath81 clauses involving @xmath82 variables .each clause corresponds with a simple map to each line @xmath83 ( including the output ) in the truth table of @xmath84 .namely , if the value of variable @xmath85 is @xmath86 in the truth table line , it will be negated in the cnf clause .viceversa , it will be affirmed if its value is @xmath87 .the formula @xmath88 , defines a satisfiability problem on the variables @xmath89 , with @xmath90 .a realization of gr1 differs from a sat problem for the intrinsically asymmetric form of the constraints , which `` force '' the value of @xmath91 .moreover , considering random instances of the problem , the space of allowed functions of gr1 is much smaller than the corresponding sat problem .for example , for a clause with fixed connectivity @xmath53 , while there are @xmath92 possible boolean functions , all of which can appear in sat , only @xmath93 of these can appear in gr1 .the two above features make the leaf removal technique useful for the latter model . in this appendix, we discuss the average of @xmath29 on the realizations of the constraints @xmath94 , for @xmath35-gr1 .one can write @xmath95 here , the randomness is contained : ( i ) in the specification of the network structure , @xmath96 , i.e. in the coordinates @xmath13 ; ( ii ) in the specification of the functions @xmath97 within the class @xmath98 of all boolean functions .an overbar ( @xmath99 ) indicates an average on both distributions , @xmath100 and @xmath101 . taking for @xmath98 the class of all boolean functions , it is straightforward to obtain @xmath102 , independently from the specification of the network structure .this result remains true considering a sub - family @xmath103 of functions that satisfy @xmath104 .the reason for this is that one finds @xmath105 .we should note that the conventional average of @xmath29 might be biased by the weight of exceptions @xcite . to access the typical behavior of the system ,the correct quantity to compute is the `` quenched average '' , @xmath106 , usually accessed with the replica , or similar methods @xcite . in the case under exam we have computed the _ annealed _ average @xmath107 ( in general @xmath108 ) . for k - gr1 withpoisson distributed out - degree , we have shown the presence of a self - averaging property for @xmath29 . that is , the quantity @xmath109^{2 } } - \left[\overline{\mathcal{n}}\right]^2 ) / ( \left[\overline{\mathcal{n}}\right]^2 ) $ ] vanishes in the thermodynamic limit @xmath110 so that , in the simple and complex control regimes , an equality holds between quenched and annealed average @xcite .

<S> a great part of the effort in the study of coarse grained models of transcription networks concentrates on their dynamical features . in this letter </S><S> , we consider their _ equilibrium _ properties , showing that the backbone underlying the dynamic descriptions is an optimization problem . </S><S> it involves @xmath0 variables , the gene expression levels , and @xmath1 constraints , the effects of transcriptional regulation . in the case of boolean variables and constraints </S><S> , we investigate the structure of the solutions , and derive phase diagrams . </S><S> notably , the model exhibits a connectivity transition between a regime of simple gene control , where the input genes control o(1 ) other genes , to a regime of complex control , where some `` core '' input genes control o@xmath2 others . </S><S> [ [ introduction . ] ] introduction . </S><S> + + + + + + + + + + + + + identity , response and architecture of a living system are central topics of molecular biology . </S><S> presently , they are largely seen as a result of the interplay between a gene repertoire and the regulatory machinery @xcite . </S><S> gene transcription in mrna form is an important step in this process . at this level , </S><S> the regulatory machinery is embodied by the transcription factors , proteins that bind to special sites along dna and control the activity of rna polymerase @xcite ( fig . </S><S> [ fig : fagraph ] ) . </S><S> this process is referred to as signal integration . </S><S> together , the _ cis_-regulatory regions establish a set of interdependencies between transcription factors and genes , including other transcription factors : a `` transcription network '' @xcite . </S><S> some of such networks of living organisms are now being explored experimentally @xcite , and show a modularity that has important biological implications @xcite . </S><S> understanding the gene expression patterns determined by these networks is an enormous challenge . </S><S> the problem is that transcription networks are fairly large , so that a coarse graining is needed . </S><S> this fact has many consequences , mainly related to the dynamics . for example , the pioneering approach of kauffman @xcite suggesting a synchronous deterministic update for a boolean ( i.e. on / off ) representation of the network is still being debated @xcite . </S><S> microscopically , it is well accepted that the gillespie algorithm @xcite correctly describes the events of chemical kinetics involved . on the other hand , with a mesoscopic average in time , </S><S> it is not clear what the emergent time scales might be . . </S><S> each diamond node represents a constraint , while each black circle is a variable . ] </S><S> we approach this problem with a model based on two features . </S><S> firstly , it focuses , rather than on dynamics , on the compatibility between gene expression patterns and signal integration functions . </S><S> simply put , a cell with @xmath0 genes can express them in exponentially many ways , @xmath3 in the boolean representation . </S><S> however , the cell never explores all the patterns of expression . </S><S> it only knows clusters of correlated configurations . </S><S> an elementary example is the ci - cro switch of @xmath4-phage @xcite . </S><S> looking at the system , one could observe the states 10 , 01 or perhaps 00 , but not 11 . </S><S> one can think that in larger systems many configurations are ruled out for the same compatibility reasons . </S><S> secondly , the model takes explicitly into account that some genes are essentially `` free '' from the point of view of transcription ( fig . </S><S> [ fig : foglia ] ) . </S><S> this fact is evident looking at the available data @xcite . while the biological situation is more complex , we regard these genes as input receptors , connected to external stimuli . </S><S> the simplest formulation ( gr1 , from gene - regulation ) assumes boolean variables and functions . </S><S> we use it to investigate theoretically the control exerted by the free genes on the expression patterns for large @xmath0 , and for random realizations of the constraints @xcite . </S><S> analysis of the satisfying configurations leads to the introduction of a `` core '' of network variables . </S><S> depending on the number of free genes in the core and the connectivity of the constraints , there are three distinct regimes of gene control . in the first regime , </S><S> the core is empty . </S><S> each free gene controls the state of a small number of genes ( `` simple control '' phase ) . in the second regime </S><S> ( `` complex control '' ) , the core contains free genes that control , both directly and indirectly , order @xmath0 others . </S><S> thus , in the complex control phase , the free core genes can be interpreted as the subset of genes that determine a choice of an expression program . in the third regime , there are no free genes in the core , and the system can not control the simultaneous expression of all its genes . the transition can be tuned by varying the connectivity and the number of constraints . </S><S> [ [ model . ] ] model . </S><S> + + + + + + the two main ingredients of our representation of a transcription network are : ( i ) a set of @xmath0 discrete variables @xmath5 associated to genes or operons ( identified with their transcripts and protein products ) . </S><S> these variables represent the expression levels . </S><S> ( ii ) a set of @xmath1 interactions , representing the signal integration , @xmath6 , with @xmath7 . </S><S> it is useful to represent variables and interactions in a so - called factor graph , as illustrated in fig . </S><S> [ fig : fagraph ] . </S><S> note that the constraints @xmath8 contain the topology of the graph . </S><S> the @xmath9 represent real or coarse - grained expression levels and can take values in @xmath10 or even continuous ones . in general , using the shea and ackers model of gene activation by recruitment @xcite , one can construct a local free energy associated to each signal integration node , that generates the constraints @xcite . here </S><S> we consider the simplest possible scenario , gr1 , treating the expression levels as boolean variables ( i.e. setting @xmath11 ) , and the signal integration functions as boolean functions @xmath12 . </S><S> the coordinates @xmath13 point at the variable occupying place @xmath14 in the @xmath15th constraint . </S><S> the expression @xmath16 imposes that the variable @xmath17 is the output of the function @xmath18 . </S><S> for example , let us consider a graph with three variables and one constraint , labeled by @xmath19 . supposing the first two variables regulate the third , @xmath20 , @xmath21 . if , for instance , the transcription can occur only when both regulators are present , then @xmath22 only if @xmath23 ( boolean and function ) . </S><S> the local connectivity of a function node is @xmath24 . </S><S> @xmath25 is called the `` in - degree '' . </S><S> the factor graph is also associated to an `` out - degree '' @xmath26 , where @xmath27 is the number of functions connected to @xmath28 . </S><S> the fact that variables and constraints are boolean make gr1 an optimization problem of the satisfiability type ( sat ) @xcite . </S><S> a very special one , given the structure of the signal integration functions . </S><S> the properties of the model depend on the class of graphs and boolean functions used . </S><S> the results presented in this work hold for a rather large class of boolean functions ( see appendix a ) . [ [ structure - of - the - solution - space . ] ] structure of the solution space . </S><S> + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the phase - space structure has never been explored for this particular case . </S><S> we set out to analyze the number of compatible configurations @xmath29 , for large @xmath0 and @xmath1 and random instances of the problem . to this end , consider the following argument , which focuses on the control exerted by the @xmath30 `` free '' input genes on the compatible solutions . </S><S> together with the free genes , the network contains some genes which are regulated but do not regulate any other </S><S> . we can refer to them as `` leaves '' ( fig [ fig : foglia ] ) . </S><S> given a realization , a leaf can always be adjusted to the output value of its function , which is then satisfied . </S><S> let us now remove from the graph each leaf and iterate this procedure ( a variation of the so - called `` leaf removal '' algorithm @xcite ) . </S><S> there are two possible outcomes : either erasing all the graph , leaving the free genes as isolated points , or stopping at a core of constraints that contains loops . in other words , </S><S> the algorithm identifies the tree - like components of the graph connected to outputs . </S><S> the core will be composed of @xmath31 genes and @xmath32 constraints . </S><S> let us now imagine to have a compatible configuration , flip a free gene , and try to construct another compatible configuration . </S><S> in the case where the core is empty , since the graph is tree - like , it will always be possible to perform this operation by local rearrangements , which propagate the output of the functions . </S><S> thus , flipping all the free genes , we find @xmath33 satisfying configurations . in the presence of a core , because of the loops , flipping a free gene of the core will in general rearrange all the core genes , and is not guaranteed to lead to a new satisfying state . </S><S> in fact , it is not granted there will be a compatible state to start with . </S><S> provided there is , the output propagation procedure can be applied to the non - core free genes to construct another . </S><S> thus , in general the @xmath30 degrees of freedom given by the free genes can not guarantee a solution . </S><S> the relevant parameter is the number of core free genes @xmath34 . </S><S> let us for the moment restrict to the case of fixed in - degree ( `` @xmath35-gr1 '' ) . in appendix b </S><S> , we show that the average of @xmath29 on the class of all boolean functions is @xmath33 . thus : ( a ) if the core is empty , the number of compatible configurations constructed by flipping the free genes are on average all the possible ones . </S><S> ( b ) if the core is not empty , in the average case it will still be possible to construct @xmath33 solutions by flipping the free genes . if @xmath36 , there will be @xmath37 clusters of solutions , and ( c ) in the case where the core contains no free genes there will be generally contradictions . </S><S> we can thus distiguish the three regimes : ( a ) simple control , ( b ) complex control , ( c ) no control . considering ensembles of random graphs </S><S> , the regimes above depend on the value of @xmath38 , so that a proper order parameter to adopt is @xmath39 @xcite . </S><S> the phase diagram can be explored studying the rank and the kernel of the connectivity matrix in the ensemble . </S><S> [ [ example - the - case - of - poisson - distributed - c_i . ] </S><S> ] example : the case of poisson - distributed @xmath27 . </S><S> + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + this is the simplest ensemble to consider , where @xmath40 @xcite . </S><S> this probability distribution does nt exclude that genes may appear in the functions that regulate them , leaving some freedom of choice . in the simplest case </S><S> , one finds @xmath41 and @xmath42 , where @xmath43 is defined by the relation @xmath44 . </S><S> this gives the phase diagram as a function of @xmath45 . </S><S> for @xmath46 there is simple control . </S><S> for @xmath47 complex control , and for @xmath48 no control . </S><S> for example , for @xmath49-gr1 , @xmath50 and @xmath51 . </S><S> the regimes of gene control correspond to thermodynamic phases , commonly referred to as sat , hard - sat , and unsat phase respectively @xcite . as a function of @xmath45 for different values of @xmath52 in the multi - poisson case . </S><S> the discrete jumps are due to the onset of complex control phases for the different values of @xmath53 . </S><S> @xmath54 can become negative many times , giving rise to reentrant no control phases ( inset ) . </S><S> the figure refers to a connectivity distribution with a cutoff at @xmath55 . ] </S><S> furthermore , it is possible to show rigorously the clustering of solutions argued above . in the simple control phase </S><S> , one cluster contains all the solutions , and a free gene controls @xmath56 other genes . </S><S> the reason for this is that , for poisson distributed out - degree , the average number of controlled genes is finite @xmath57 , while the number of free genes is extensive . </S><S> conversely , in the complex control phase , the free genes within the core control @xmath58 other genes ( while there is still @xmath56 control outside of the core ) . from a physical point of view </S><S> , the clusters are separated by an extensive distance , i.e. by free energy barriers . </S><S> the number of clusters is related to the ( computational ) complexity @xmath59 of the system , defined by the relation @xmath60 . </S><S> here @xmath61 , the entropy , measures the width of each cluster , while @xmath59 `` counts '' the number of clusters . </S><S> therefore , by definition , @xmath59 is directly related to @xmath62 , i.e. to the partitioning of the free genes in and out of the core . how the system explores ( or not ) these clusters depends on details of its dynamics . </S><S> generically , one can say that the dynamics in a cluster will be residual : many variables are fixed , the others can change . this matches a qualitative feature of many cells , where some genes are constantly expressed , and the rest vary @xcite . [ [ multi - poisson - phase - diagram . ] ] multi - poisson phase diagram . </S><S> + + + + + + + + + + + + + + + + + + + + + + + + + + + + while the fixed @xmath53 case is useful to get some theoretical insight , the known transcription networks are far from having fixed in - degree . </S><S> for example , in e. coli , the in - degree has poisson distribution , while the out - degree resembles a power law . </S><S> for this reason , a biologically more interesting case is when both the in- and the out - degree vary along the network . </S><S> considering @xmath63 , the conditioned probability that a variable is in @xmath64 clauses of the @xmath53 kind , we have @xmath65 . </S><S> the leaf removal equations can be applied separately to sets of clauses with a given connectivity , defining @xmath66 and @xmath67 , where @xmath68 . </S><S> choosing @xmath69 does not affect the exponential asymptotic decay of @xmath70 for large @xmath64 . </S><S> we can call this case multi - poisson , as the graph is a superposition , following a poisson distribution , of graphs with fixed in - degree and poisson - distrubuted out - degree . for the multi - poisson case . </S><S> the dashed line , a power law with exponent @xmath71 , represents the mean value of the numerically evaluated critical parameter @xmath72 for the simple - complex control transition of network realizations with @xmath73 . ] the behavior of gr1 on such a topology is nontrivially different from the fixed connectivity case . </S><S> the main reason for this is that , while @xmath74 is still locally decreasing , many new discontinuities emerge , due to the influence of clauses with different connectivities . </S><S> this gives rise to different phenomena . </S><S> firstly , @xmath62 can increase globally with increasing @xmath45 . </S><S> indeed , it does increase step - wise with @xmath45 after @xmath75 , to decrease again before @xmath76 . </S><S> its discrete jumps are due to the onset of complex control phases for the different values of @xmath53 ( fig . </S><S> [ fig : mp - delta ] ) . </S><S> this fact has an influence on the number of compatible states as a function of @xmath45 . </S><S> secondly , @xmath62 can become negative and then jump back to a positive value , creating a reentrant unsat phase ( fig . </S><S> [ fig : mp - pd ] ) . </S><S> this means that , on average , by increasing the number of constraints one can pass from unsolvable problems to solvable ones . a heuristic explanation for this counterintuitive fact is that , at fixed @xmath0 , the addition of a constraint might connect a closed loop in the core to external free genes , thereby solving a contradiction . </S><S> interestingly , the simple to complex control boundary is a power - law in @xmath45 ( fig . </S><S> [ fig : mp - pd ] ) . </S><S> [ [ discussion . ] ] discussion . </S><S> + + + + + + + + + + + to conclude , we established a simple framework for the modeling of large scale transcription networks . </S><S> it is a compatibility analysis on the constraints established by transcription . </S><S> its advantage is that , while avoiding to deal directly with the dynamics , it gives non - trivial results . in the absence of an explicit knowledge of the emergent time scales </S><S> , we feel this is an appropriate approach , particularly in the boolean approximation treated here . from a technical standpoint </S><S> , gr1 is different from other problems of the satisfiability kind because of the particular structure of its constraints . </S><S> this makes it possible to apply the leaf removal technique , which is ineffective for other models , such as random - k - sat @xcite . from a general standpoint </S><S> , our model shows that the `` biological '' complexity is not simply measured by the number of genes . </S><S> a more proper indicator is @xmath62 which depends on the order parameter @xmath52 , or - roughly - on the number of transcription factors per gene . at fixed number of genes </S><S> , it is known that this quantity increases in bacteria that need to react to more environments @xcite . imagining that prokaryotes are naturally found in a simple control phase </S><S> , our phase diagram predicts an intrinsic limit to this process , represented by the phase boundary with the complex control phase . </S><S> the multi - poisson case gives an interesting prediction for the behaviour of this boundary at fixed @xmath0 . </S><S> namely , at criticality , the average number of constraints scales as a power - law with @xmath45 . </S><S> this feature , together with the existence of a core and the predictions on the control exerted by free genes , can possibly be tested experimentally . </S><S> the approach presented here is new , and largely unexplored . </S><S> it is naturally fit to study networks with non - boolean variables and probabilistic constraints . </S><S> it can be of use for models that describe other regulation mechanisms than just transcription . </S><S> more far - reached extensions include evolutionary models . </S><S> it is not clear yet exactly how useful it can be for the study of concrete biological networks . together with the general trends , a biologically significant model has to be able to deal with the details of an individual realization the system </S><S> this is , we think , the main challenge to our approach , and the direction we are currently exploring . we would like to acknowledge interesting discussions with j. berg , m. caselle , l. finzi , m. leone , a. sportiello , p.r . </S><S> tenwolde , r. zecchina . </S><S> we thank an anonymous referee for help improving our manuscript . </S>

1510.02484

we study _ contact representations _ of planar graphs in 3d , where vertices are represented by interior - disjoint axis - aligned boxes and edges are represented by shared boundaries between the corresponding boxes . a contact representation of a planar graph @xmath0 is _ proper _ if for each edge @xmath1 of @xmath0 , the boxes for @xmath2 and @xmath3 have a shared boundary with non - zero area .such a contact between two boxes is also called a _ proper contact_. _cubes _ are axis - aligned boxes where all sides have the same length .a contact representation of a planar graph with boxes is called a _ cube - contact _ representation when all the boxes are cubes . in a weighted variant of the problema _ proportional box - contact _ representation is one where each vertex @xmath3 is represented with a box of volume @xmath4 , for any function @xmath5 , assigning weights to the vertices @xmath6 . note that this `` value - by - volume '' representation is a natural generalization of the `` value - by - area '' cartograms in 2d .* related work : * the history of representing planar graphs as contact graphs dates back at least to koebe s 1930 theorem @xcite for representing planar graphs by touching disks in 2d .proper contact representation with rectangles in 2d is the well - known _ rectangular dual _ problem , for which several characterizations exist @xcite .representations with other axis - aligned and non - axis - aligned polygons @xcite have been studied .related graph - theoretic , combinatorial and geometric problems continue to be of interest @xcite . the weighted variant of the problem has been considered in the context of rectangular , rectilinear , and unrestricted cartograms @xcite .contact representations have been also considered in 3d .thomassen @xcite shows that any planar graph has a proper contact representation with touching boxes , while felsner and francis @xcite find a ( not necessarily proper ) contact representation of any planar graph with touching cubes .recently , bremner _et al . _@xcite asked whether any planar graph can be represented by proper contacts of cubes .they answered the question positively for the case of partial planar 3-trees and some planar grids , but the problem remains open for general planar graphs .the weighted variant of the problem in 3d is much less studied , although recently alam _ et al . _@xcite have presented algorithms for proportional representation of several classes of graphs ( e.g. , outerplanar , planar bipartite , planar , complete ) , using 3d l - shapes . *our contribution : * here we expand the class of planar graph representable by proper contact of cubes .we also show that several classes of planar graphs admit proportional box - contact representations .specifically , we show how to compute a proportional box - contact representation for plane 3-trees , while a cube - contact representation for the same graph class follows from @xcite .we also show how to compute a proportional box - contact representation and a cube - contact representation for _ nested maximal outerplanar graphs _ , which are defined as follows .nested outerplanar graph _ is either an outerplanar graph or a planar graph @xmath0 where each component induced by the internal vertices is another nested outerplanar graph with exactly three neighbors in the outerface of @xmath0 .a _ nested maximal outerplanar graph _ is a subclass of nested outerplanar graphs that is either a maximal outerplanar graph or a maximal planar graph in which the vertices on the outerface induce a maximal outerplanar graph and each component induced by internal vertices is another nested maximal outerplanar graph .a _ 3-tree _ is either a 3-cycle or a graph @xmath0 with a vertex @xmath3 of degree three in @xmath0 such that @xmath7 is a 3-tree and the neighbors of @xmath3 form a triangle .if @xmath0 is planar , then it is called a _planar 3-tree_. a _ plane 3-tree _ is a planar 3-tree along with a fixed planar embedding .starting with a 3-cycle , any planar 3-tree can be formed by recursively inserting a vertex inside a face and adding an edge between the newly added vertex and each of the three vertices on the face @xcite . using this simple construction, we can create in linear time a _ representative tree _ for @xmath0 @xcite , which is an ordered rooted ternary tree @xmath8 spanning all the internal vertices of @xmath0 .the root of @xmath8 is the first vertex we have to insert into the face of the three outer vertices . adding a new vertex @xmath3 in @xmath0 will introduce three new faces belonging to @xmath3 .the first vertex @xmath9 we add in each of these faces will be a child of @xmath3 in @xmath8 .the correct order of @xmath8 can be obtained by adding new vertices according to the counterclockwise order of the introduced faces .an _ outerplanar graph _ is one that has a planar embedding with all vertices on the same face ( outerface ) .an outerplanar graph is _ maximal _ if no edge can be added without violating its outerplanarity .thus in a maximal outerplanar graph all the faces except for the outerface are triangles . for @xmath10 , a @xmath11-outerplanar graph@xmath0 is an embedded graph such that deleting the outer - vertices from @xmath0 yields a graph where each component is at most a @xmath12-outerplanar graph ; a @xmath13-outerplanar graph is just an outerplanar graph .note that any planar graph is a @xmath11-outerplanar graph for some integer @xmath14 .let @xmath0 be a planar graph .we define the _ pieces _ of @xmath0 as follows .if @xmath0 is outerplanar , it has only one piece , the graph itself . otherwise , let @xmath15 , @xmath16 , @xmath17 , @xmath18 be the components of the graph obtained by deleting the outer vertices ( and their incident edges ) from @xmath0 .then the pieces of @xmath0 are all the pieces of @xmath19 for each @xmath20 , as well as the subgraph of @xmath0 induced by the outer - vertices of @xmath0 .note that each piece of @xmath0 is an outerplanar graph .since @xmath0 is an embedded graph , for each piece @xmath21 of @xmath0 , we can define the _ interior _ of @xmath21 as the region bounded by the outer cycle of @xmath21 .then we can define a rooted tree @xmath22 where the pieces of @xmath0 are the vertices of @xmath22 and the parent - child relationship in @xmath22 is determined as follows : for each piece @xmath21 of @xmath0 , its children are all the pieces of @xmath0 that are in the interior of @xmath21 but not in the interior of any other pieces of @xmath0 .a piece of @xmath0 has _ level _ @xmath23 if it is on the @xmath23-th level of @xmath22 .all the vertices of a piece at level @xmath23 are also _ @xmath23-level _ vertices .a planar graph is a _ nested outerplanar graph _ if each of its pieces at level @xmath24 has exactly three vertices of level @xmath25 as a neighbor of some of its vertices . on the other handa _ nested maximal outerplanar graph _ is a maximal planar graph where all the pieces are maximal outerplanar graphs .here we prove that planar 3-trees have proportional box - representations in two different ways .the first one is a more intuitive proof ; the second one includes a direct computation of the coordinates for the representation .[ th : p3t - box ] let @xmath26 be a plane 3-tree with a weight function @xmath9 . then a proportional box - contact representation of @xmath0 can be computed in linear time ..,scaledwidth=80.0% ] * first proof : * let @xmath27 , @xmath28 , @xmath29 be the outer vertices of @xmath0 .we construct a representation @xmath30 for @xmath0 where @xmath28 occupies the bottom side of @xmath30 , @xmath27 occupies the back of @xmath31 and @xmath29 occupies the right side of @xmath32 ; see fig .[ fig : prop - p3t-2 ] . here for a set of vertices @xmath33 , @xmath34 denotes the representation obtained from @xmath30 by deleting the boxes representing the vertices in @xmath33 .the claim is trivial when @xmath0 is a triangle , so assume that @xmath0 has at least one internal vertex .let @xmath35 be the root of the representation tree @xmath8 of @xmath0. then @xmath35 is adjacent to @xmath27 , @xmath28 and @xmath29 and thus defines three regions @xmath15 , @xmath16 and @xmath36 inside the triangles @xmath37 , @xmath38 and @xmath39 , respectively ( including the vertices of these triangles ) . by induction hypothesis @xmath19 , @xmath40 has a proportional box - contact representation @xmath41 where the boxes for the three vertices in @xmath42 occupy the bottom , back and right sides of @xmath41 .define @xmath43 .we now construct the desired representation for @xmath0 . first take a box for @xmath35 with volume @xmath44 and place it in a corner created by the intersection of three pairwise - touching boxes ; see fig .[ fig : prop - p3t-2 ] . for each @xmath42 , @xmath40, there is a corner @xmath45 formed by the intersection of the three boxes for @xmath42 .we now place @xmath46 ( after possible scaling ) in the corner @xmath45 so that it touches the boxes for the vertices in @xmath42 by three planes . note that this is always possiblesince we can choose the surface areas for @xmath27 , @xmath28 and @xmath29 to be arbitrarily large and still realize their corresponding weights by appropriately changing the thickness in the third dimension .this construction requires only linear time , by keeping the scaling factor for each region in the representative tree @xmath8 at the vertex representing that region .then the exact coordinates can be computed with a top - down traversal of @xmath8 . *second proof : * assume ( after possible factoring ) that for each vertex @xmath3 of @xmath0 , the weight @xmath4 is at least 1 .let @xmath8 be the representative tree of @xmath0 .for any vertex @xmath3 of @xmath8 , we denote by @xmath47 , the set of the descendants of @xmath3 in @xmath8 including @xmath3 .the _ predecessors _ of @xmath3 are the neighbors of @xmath3 in @xmath0 that are not in @xmath47 . clearlyeach vertex of @xmath8 has exactly three predecessors .we now define a parameter @xmath48 for each vertex @xmath3 of @xmath8 .let @xmath49 , @xmath50 and @xmath51 be the three children of @xmath3 in @xmath8 ( where zero or more of these three children may be empty ). then @xmath48 is defined as @xmath52{w(v)}]$ ] , where @xmath53 is taken as zero when @xmath54 is empty .we can compute the value of @xmath48 for each vertex @xmath3 of @xmath8 by a linear - time bottom - up traversal of @xmath8 .once we have computed these values , we proceed on constructing the box - contact representation as follows . .] let @xmath27 , @xmath28 , @xmath29 be the three outer vertices of @xmath0 in the clockwise order and let @xmath35 be the root of @xmath8 .we start by computing three boxes for @xmath27 , @xmath28 and @xmath29 with the correct volume as illustrated in fig .[ fig : prop - p3t](a ) , so that the volume of the dotted box @xmath55 is @xmath56 .we will now construct a box representation of @xmath57 inside @xmath55 so that all the vertices in @xmath57 adjacent to an outer vertex is represented by a box with a face co - planar on the face of @xmath55 adjacent to box representing that outer vertex .we do this recursively by a top - down computation on @xmath8 .let @xmath3 be a vertex of @xmath8 with the three predecessors @xmath58 , @xmath59 and @xmath60 .let @xmath61 be a box with volume @xmath48 and let @xmath62 , @xmath63 , @xmath64 be three faces of it with a common point . while traversing @xmath3 , we compute a proportional box - contact representation of @xmath47 inside @xmath61 where the vertices in @xmath47 adjacent to @xmath65 for some @xmath66 is represented by a box with a face co - planar with @xmath67 .let @xmath49 , @xmath50 and @xmath51 are the three children of @xmath3 in @xmath8 ( where zero or more of these children may be empty ) . also assume that @xmath68 , @xmath69 , @xmath70 are the length , width and height of @xmath61 , respectively and @xmath71 is the common point of @xmath62 , @xmath63 and @xmath64 . then first compute a box @xmath72 of volume @xmath4 for @xmath3 with a corner at @xmath71 where @xmath73 , @xmath74 and @xmath75 are the length , width and height of @xmath72 , such that @xmath76{w(v)}}{w(v_1)+\sqrt[3]{w(v)}}$ ] .these choices of @xmath77 s also creates three boxes @xmath78 with volume at least @xmath79 , @xmath66 , as illustrated in fig .[ fig : prop - p3t](b ) .finally we recursively compute the box representations for @xmath80 inside @xmath78 for @xmath66 to complete the construction .@xcite let @xmath0 be a plane 3-tree .then a cube - contact representation of @xmath0 can be computed in linear time .the proof of this claim also relies on the recursive decomposition of planar 3-trees .we prove the following main theorem in this section : any nested maximal outerplanar graph has a proper contact representation with cubes . [ th : nested ] we prove theorem [ th : nested ] by construction , starting with a representation for each piece of @xmath0 , and combining the pieces to complete the representation for @xmath0 .let @xmath0 be a nested maximal outerplanar graph .we first augment the graph @xmath0 by adding three mutually adjacent dummy vertices @xmath81 on the outerface and then triangulating the graph by adding dummy edges from these three vertices to the outer vertices of @xmath0 such that the graph remains planar ; see fig .[ fig : cube - merge1](a ) . call this graph the _ extended graph _ of @xmath0 . for consistency ,let the three dummy vertices have level @xmath82 .the observation below follows from the definition of nested maximal outerplanar graphs .[ obs : extended ] let @xmath0 be a nested - maximal planar graph and let @xmath83 be the extended graph of @xmath0 . then for each piece @xmath21 of @xmath0 at level @xmath23 , there is a triangle of @xmath25-level vertices adjacent to the vertices of @xmath21 and no other @xmath11-level vertices with @xmath84 are adjacent to any vertex of @xmath21 . given this observation , we use the following strategy to obtain a contact representation of @xmath0 with cubes . for each piece @xmath21 of @xmath0 at level @xmath23 , let @xmath85 , @xmath86 and @xmath87 be the three @xmath25-level vertices adjacent to @xmath21 s vertices . let @xmath88 be the subgraph of @xmath0 induced vertices of @xmath21 as well as @xmath85 , @xmath86 and @xmath87 ; call @xmath88 the _ extended piece _ of @xmath0 for @xmath21 .we obtain a contact representation of @xmath88 with cubes and delete the three cubes for @xmath85 , @xmath86 and @xmath87 to obtain the contact representation of @xmath21 with cubes .finally , we combine the representations for the pieces to complete the desired representation of @xmath0 . before we give more details on this algorithm , we have the following lemma , that we use in this section .furthermore this result is also interesting by itself , since for any outerplanar graph @xmath89 , where each face has at least one outer edge , lemma [ lem : square ] gives a contact representation of @xmath89 on the plane with squares such that the outer boundary of the representation is a rectangle .[ lem : square ] let @xmath0 be planar graph with outerface @xmath90 and at least one internal vertex , such that @xmath91 is a maximal outerplanar graph . if there is no chord between any two neighbors of @xmath85 and no chord between any two neighbors of @xmath86 , then @xmath0 has a contact representation @xmath30 in 2d where each inner vertex is represented by a square , the union of these squares forms a rectangle , and the four sides of these rectangles represent @xmath85 , @xmath86 , @xmath28 and @xmath27 , respectively .we prove this lemma by induction on the number of vertices in @xmath0 .denote the maximal outerplanar graph @xmath92 ; see fig .[ fig : square](a ) .if @xmath0 contains only one internal vertex @xmath3 , then we compute @xmath30 by representing @xmath3 by a square @xmath72 of arbitrary size and representing @xmath85 , @xmath86 , @xmath28 and @xmath27 by the left , bottom , right and top sides of @xmath72 ..,scaledwidth=80.0% ] we thus assume that @xmath0 has at least two internal vertices .let @xmath2 be the unique common neighbor of @xmath93 in @xmath94 .if @xmath2 is a neighbor of @xmath85 , then @xmath95 is a maximal outerplanar graph . by induction hypothesis, @xmath96 has a contact representation @xmath97 where each internal vertex of @xmath96 is represented by a square and the left , bottom , right and top sides of @xmath97 represent @xmath85 , @xmath86 , @xmath28 and @xmath2. then we compute @xmath30 from @xmath97 by adding a square @xmath98 to represent @xmath2 such that @xmath98 spans the entire width of @xmath97 and is placed on top of @xmath97 ; see fig .[ fig : square](b ) .a similar construction can be used if @xmath2 is a neighbor of @xmath86 ; see fig .[ fig : square](c ) . we thus compute a contact representation for @xmath0 ; see fig .[ fig : square](d ) .[ lem : piece ] let @xmath21 be a piece of @xmath0 at level @xmath23 and @xmath88 be the extended piece for @xmath21 with @xmath25-level vertices @xmath85 , @xmath86 , @xmath87 .then @xmath88 has a cube - contact representation .let @xmath35 be a common neighbor of @xmath86 and @xmath87 ; @xmath99 a common neighbor of @xmath85 and @xmath87 ; @xmath100 a common neighbor of @xmath85 and @xmath86 .it is easy to find a contact representation of @xmath88 if @xmath35 , @xmath99 and @xmath100 are the only vertices of @xmath21 , so let @xmath21 have at least four vertices .the outer cycle of @xmath21 can be partitioned into three paths : @xmath101 is the path from @xmath99 to @xmath100 , @xmath102 is the path from @xmath35 to @xmath100 and @xmath103 is the path from @xmath35 to @xmath99 .note that all vertices on the path @xmath101 ( @xmath102 , @xmath103 ) are adjacent to @xmath85 ( @xmath86 , @xmath87 ) .a chord @xmath1 is a _ short chord _ if it is between two vertices on the same path from the set @xmath104 .( note that a chord between two vertices from the set @xmath105 is also a short chord . )we have the following two cases .* case a : there is no short chord in @xmath21 . * in this case all the chords of @xmath21 are between two different paths .we consider the following two subcases ..,title="fig:",scaledwidth=70.0% ] + ( a)(b ) * case a1 : there is no chord with one end - point in @xmath106 . * in this case , due to maximal - planarity there exist three vertices @xmath27 , @xmath28 and @xmath29 , adjacent to @xmath85 , @xmath86 , and @xmath87 , respectively such that ( i ) @xmath107 is the chord between vertices of @xmath101 and @xmath102 farthest away from @xmath100 , ( ii ) @xmath108 is the chord between vertices of @xmath102 and @xmath103 farthest away from @xmath35 , and ( iii ) @xmath109 is the chord between vertices of @xmath101 and @xmath103 farthest away from @xmath99 ; see fig .[ fig : cube - merge1](a ) .we can then find three interior - disjoint subgraphs of @xmath88 defined by three cycles of @xmath88 : @xmath15 is the one induced by all vertices on or inside @xmath90 ; @xmath16 is induced by all vertices on or inside @xmath110 ; and @xmath36 is induced by all vertices on or inside @xmath111 .each of these subgraphs has the common property that if we delete two vertices from the outerface ( two vertices from the set @xmath81 in each subgraph ) , we get an outerplanar graph . from the representation with squares from the proof of lemma [ lem : square ], we find a contact representation of @xmath19 , @xmath40 where each internal vertex of @xmath19 is represented by a cube and the union of all these cubes forms a rectangular box whose four sides realize the outer vertices .we use such a representation to obtain a contact representation of @xmath88 with cubes as follows .we draw pairwise adjacent cubes ( of arbitrary size ) for @xmath85 , @xmath86 , @xmath87 . we need to place the cubes for all the vertices of @xmath21 in the a corner defined by three faces of the cubes for @xmath85 , @xmath86 , @xmath87. then we place three mutually touching cubes for @xmath27 , @xmath28 and @xmath29 , which touch the walls for @xmath85 , @xmath86 and @xmath87 , respectively ; see fig .[ fig : cube - merge1](b ) .we also compute a contact representation of the internal vertices for each of the three graphs @xmath15 , @xmath16 and @xmath36 with cubes using lemma [ lem : square ] , so that the outer boundary for each of these representation forms a rectangular pipe .we adjust the sizes of the three cubes for @xmath27 , @xmath28 and @xmath29 in such a way that the three highlighted rectangular pipes precisely fit these three representations ( after some possible scaling ) .note that this construction works even if one or more of the subgraphs @xmath15 , @xmath16 and @xmath36 are empty .this completes the analysis of * case a1*. .,title="fig:",scaledwidth=70.0% ] + ( a)(b ) * case a2 : there is at least one chord with one end - point in @xmath106 . *due to planarity all such chords will have the same end - point in @xmath112 .suppose @xmath99 is this common end point for these chords ; see fig .[ fig : cube - merge2](a ) .let @xmath113 and @xmath114 be the first and last endpoints in the clockwise order of these chords around @xmath99 .then we can find two subgraphs @xmath15 and @xmath16 induced by the vertices on or inside two separating cycles @xmath115 and @xmath116 .we find contact representations for the internal vertices of these two graphs @xmath15 and @xmath16 using lemma [ lem : square ] so that the outer - boundaries of these representation form rectangular pipes .we then obtain the desired contact representation for @xmath88 , starting with the three mutually touching walls for @xmath85 , @xmath86 and @xmath87 at right angles from each other , placing the cubes for @xmath99 and @xmath113 , @xmath17 , @xmath114 as illustrated in fig .[ fig : cube - merge2](b ) , and fitting the representations for @xmath15 and @xmath16 ( after some possible scaling ) in the highlighted regions .* case b : there are some shord chords in @xmath21 . * in this case , we find at most four subgraphs from @xmath88 as follows . at each path in @xmath117, we find the _outermost chord _, i.e. , one that is not contained inside any other chords on the same path . suppose these chords are @xmath118 , @xmath119 and @xmath120 , on the three paths @xmath101 , @xmath102 , @xmath103 , respectively .then three of these subgraphs @xmath121 , @xmath122 and @xmath123 are induced by the vertices on or inside the three triangles @xmath124 , @xmath125 and @xmath126 . the fourth subgraph @xmath127 is obtained from @xmath88 by deleting all the inner vertices of the three graphs @xmath121 , @xmath122 and @xmath123 ; see fig .[ fig : short - chord ] .a cube representation of @xmath127 can be found by the algorithm in * case a * , as @xmath127 fits the condition that there is no chord between any two neighbors of the same vertex in @xmath81 .note that by moving the cubes in the representation by an arbitrarily small amount , we can make sure that for each triangle @xmath128 in @xmath127 , the three cubes for @xmath129 , @xmath130 and @xmath131 form a corner surrounded by three mutually touching walls at right angles to each other . now observe that each of the three graphs @xmath121 , @xmath122 and @xmath123 is a planar 3-tree ; thus using the algorithm of either @xcite or @xcite , we can place the internal vertices of these three graphs in their corresponding corners , thereby completing the representation .* proof of theorem [ th : nested ] : * let @xmath0 be a nested maximal outerplanar graph .we build the contact representation of @xmath0 by a top - down traversal of the rooted tree @xmath22 of the pieces of @xmath0 .we start by creating a corner surrounded by three mutually touching walls at right angle to each other .then whenever we traverse any vertex of @xmath22 , we realize the corresponding piece @xmath21 at level @xmath23 by obtaining a representation using lemma [ lem : piece ] and placing this in the corner created by the three already - placed cubes for the three @xmath25-level vertices adjacent to @xmath21 ( after possible scaling ) .in this section we prove the following main theorem . [ th : nested - prop ] let @xmath132 be a nested outerplanar graph and let @xmath5 be a weight function defining weights for the vertices of @xmath0 .then @xmath0 has a proportional contact representation with axis - aligned boxes with respect to @xmath9 .we construct a proportional representation for @xmath0 using a similar strategy as in the previous section : we traverse the construction tree @xmath22 of @xmath0 and deal with each piece of @xmath0 separately .each piece @xmath21 of @xmath0 is an outerplanar graph and hence one can easily construct a proportional box - contact representation for @xmath21 as follows . any outerplanar graph@xmath21 has a contact representation with rectangles in the plane .in fact in @xcite , it was shown that @xmath21 has a contact representation with rectangles on the plane where the rectangles realize prespecified weights by their areas .thus by giving unit heights to all rectangles we can obtain a proportional box - contact representation of @xmath21 for any given weight function .however if we construct proportional box - contact representation for each piece of @xmath0 in this way , it is not clear that we can combine them all to find a proportional contact representation of the whole graph @xmath0 .instead , we use this construction idea in lemmas [ lem : stair ] and [ lem : double - stair ] to build two different proportional rectangle - contact representations for outerplanar graphs and we use them in the proof of theorem [ th : nested - prop ] .suppose @xmath89 is an outerplanar graph and @xmath30 is a contact representation of @xmath89 with rectangles in the plane .we say that a corner of a rectangle in @xmath30 is _ exposed _ if it is on the outer - boundary of @xmath30 and is not shared with any other rectangles .[ lem : stair ] let @xmath89 be a maximal outerplanar graph with a weight function @xmath9 .let @xmath13 , @xmath17 , @xmath133 be the clockwise order of the vertices around the outer - cycle . then a proportional rectangle - contact representation @xmath30 of @xmath89 for @xmath9 can be computed so that rectangle @xmath134 for @xmath13 is leftmost in @xmath30 , rectangle @xmath135 for @xmath133 is bottommost in @xmath136 , and the top - right corner for each rectangle is exposed in @xmath30 .we give an algorithm that recursively computes @xmath30 .constructing @xmath30 is easy when @xmath0 is a single edge @xmath137 .we thus assume that @xmath0 has at least 3 vertices .let @xmath129 be the ( unique ) third vertex on the inner face that is adjacent to @xmath137 .then graph @xmath0 can be split into two graphs at vertex @xmath129 and edge @xmath137 : @xmath138 $ ] consists of the graph induced by all vertices between @xmath13 and @xmath129 in clockwise order around the outer - cycle ; while @xmath139 $ ] consists of the graph induced by the vertices between @xmath129 and @xmath133 .recursively draw @xmath138 $ ] and remove the rectangles for @xmath13 and @xmath129 from it ; call the result @xmath140 .again recursively draw @xmath139 $ ] and remove @xmath129 and @xmath133 from it ; call the result @xmath141 .now draw a rectangle @xmath142 for @xmath129 with area @xmath143 .let @xmath144 and @xmath145 be the width and height of @xmath142 , respectively .then draw the rectangles @xmath134 and @xmath135 for @xmath13 and @xmath133 touching the left and the bottom sides of @xmath142 , respectively with necessary areas .select the widths and heights of these two rectangles such that the area @xmath146 can contain @xmath140 while the area @xmath147 can contain @xmath141 , where @xmath148 and @xmath149 denote the width and height of @xmath150 , respectively for @xmath151 .finally place @xmath140 ( after possible scaling ) touching the right side of @xmath134 and the top side of @xmath142 and place @xmath141 ( after possible scaling ) touching the right side of @xmath142 and the top side of @xmath135 to complete the drawing ; see fig .[ fig : stair ] . .] . ]note that in the layout obtained above the top right corners of the rectangles for vertices @xmath152 have increasing @xmath129-coordinates and decreasing @xmath130-coordinates .thus we refer to them as * staircase * layouts and to the algorithm as the * staircase algorithm*. [ lem : double - stair ] let @xmath89 be a maximal outerplanar graph with a weight function @xmath9 .let @xmath13 , @xmath17 , @xmath133 be the clockwise order of the vertices around the outer - cycle .then a proportional rectangle - contact representation @xmath30 of @xmath89 for @xmath9 can be computed so that rectangle @xmath134 for @xmath13 is leftmost in @xmath30 , rectangle @xmath135 for @xmath133 is bottommost in @xmath153 , and the top - right corners of all rectangles for vertices @xmath154 and the bottom - right corners of all rectangles for vertices @xmath155 are exposed in @xmath30 .we again compute @xmath30 recursively .constructing @xmath30 is easy when @xmath0 is a single edge @xmath137 .we thus assume that @xmath0 has at least 3 vertices .let @xmath129 be the ( unique ) third vertex on the inner face that is adjacent to @xmath137 .define the two graphs @xmath138 $ ] and @xmath139 $ ] as in the proof of lemma [ lem : stair ] ; see also fig .[ fig : double - stair](a ) . if @xmath156 , then recursively draw @xmath138 $ ] and remove the rectangles for @xmath13 and @xmath129 from it ; call the result @xmath140 . draw @xmath139 $ ] using the * staircase algorithm*. now draw three mutually touching rectangles @xmath134 , @xmath142 and @xmath135 for @xmath13 , @xmath129 and @xmath133 , respectively with necessary areas such that the right side of @xmath134 touches both @xmath142 and @xmath135 and the right side of @xmath142 has greater @xmath129-coordinate than the right side of @xmath135 ; see fig .[ fig : double - stair](b ) .finally place @xmath140 ( after possible scaling ) touching the right side of @xmath134 and the top side of @xmath142 such that the right side of the rectangle for @xmath157 extends past @xmath142 .also place @xmath141 ( after @xmath158 clockwise rotation and possible scaling ) touching the bottom side of @xmath142 and the right side of @xmath135 to complete the drawing ( the width of @xmath142 can be chosen long enough so that @xmath141 can be contained between the bottom side of @xmath142 and the right side of @xmath135 ) .on the other hand if @xmath159 we follow almost the same procedure as in the previous paragraph .however , instead of the drawing of @xmath138 $ ] recursively , we compute it by the * staircase algorithm * and then delete from it @xmath13 and @xmath129 to obtain @xmath140 .we compute @xmath141 as in the previous section .we also draw @xmath134 , @xmath142 and @xmath135 in the same way. then we place @xmath140 ( after possible scaling ) touching the right side of @xmath134 and the top side of @xmath142 ( again the width of @xmath142 is chosen long enough so that this can be done ) .we also place @xmath141 as the same manner as before to complete the drawing ; see fig .[ fig : double - stair](c ) . finally if @xmath160 , then we draw @xmath138 $ ] by the * staircase algorithm * and delete from it @xmath13 and @xmath129 to obtain @xmath140 . however ,to compute @xmath141 , we recursively draw @xmath139 $ ] and delete @xmath129 and @xmath133 from it .we now draw @xmath134 , @xmath142 and @xmath135 as before but this time the right side of @xmath135 should extend past @xmath142 .we now place @xmath140 ( after possible scaling ) touching the right side of @xmath134 and the top side of @xmath142 ( this is again possible for suitable choice of the height of @xmath134 ) .finally we complete the drawing by placing @xmath141 ( after possible scaling ) touching the right side of @xmath142 and the top side of @xmath135 so that the right side of the rectangle for @xmath161 extends past @xmath135 ; see fig .[ fig : double - stair](d ) . note that in the layout obtained above the top - right corners for vertices @xmath154 and the bottom - right corners for vertices @xmath162 form two staircases .thus we refer to this as a * double - staircase * layout , to the algorithm as the * double - staircase algorithm * , and to vertex @xmath163 as the _ pivot vertex_. let @xmath89 be a maximal outerplanar graph and let @xmath30 be either a * staircase * or a * double - staircase * layout .then any triangle @xmath164 in @xmath89 is represented by three rectangles and the shared boundaries of these rectangles define a _t - shape_. the vertex whose two shared boundaries are collinear in the t - shape is called the _ pole _ of the triangle @xmath164 . *proof of theorem [ th : nested - prop ] .* let @xmath22 be the construction tree for @xmath0 .we compute a representation for @xmath0 by a top - down traversal of @xmath22 , constructing the representation for each piece as we traverse it .let @xmath21 be a piece of @xmath0 at the @xmath23-th level .if @xmath21 is the root of @xmath22 , then we use the * staircase algorithm * to find a contact representation of @xmath21 with rectangles in the plane and then we give necessary heights to these rectangles to obtain a proportional contact representation of @xmath21 with boxes . otherwise , the vertices of @xmath21 are adjacent to exactly three @xmath25-level vertices @xmath85 , @xmath86 , @xmath87 that form a triangle in the parent piece of @xmath21 . since @xmath85 , @xmath86 ,@xmath87 belong to the parent piece of @xmath21 , their boxes have already been drawn when we start to draw @xmath21 . to find a correct representation of @xmath0 ,we need that the boxes for the vertices in @xmath21 have correct adjacencies with the boxes for @xmath85 , @xmath86 , and @xmath87 ; hence we assume a fixed structure for such a triangle .we maintain the following invariant : _ let @xmath164 be three vertices in a piece @xmath21 of @xmath0 forming a triangle .then in the proportional contact representation of @xmath21 , the boxes for @xmath71 , @xmath165 , @xmath35 are drawn in such a way that ( i ) the projection of the mutually shared boundaries for these boxes in the @xmath166-plane forms a t - shape , ( ii ) the highest faces ( faces with largest @xmath131-coordinate ) of the three rectangles have different @xmath131 coordinates and the highest face of the pole - vertex of the triangle has the smallest @xmath131-coordinate. _ note that by choosing the areas of the rectangles in the * staircase * layout , we can maintain this invariant for the parent piece by appropriately adjusting the heights of the boxes(e.g ., incrementally increasing heights for the vertices in the recursive * staircase algorithm * ) .we now describe the construction of a proportional box - contact representation of @xmath21 with the correct adjacencies for @xmath85 , @xmath86 and @xmath87 . by the invariantthe projection of the shared boundaries for @xmath167 forms a t - shape in the @xmath166-plane . without loss of generality assume that @xmath85 is the pole of the triangle and the highest faces of @xmath86 , @xmath87 and @xmath85 are in this order according to decreasing @xmath131-coordinates .also assume that @xmath21 is a maximal outerplanar graph ; we later argue that this assumption is not necessary .let @xmath107 be a common neighbor of @xmath85 and @xmath86 ; @xmath108 a common neighbor of @xmath86 and @xmath87 ; @xmath168 a common neighbor of @xmath87 and @xmath85 .then the outer cycle of @xmath21 can be partitioned into three paths : @xmath101 is the path from @xmath168 to @xmath107 , @xmath102 is the path from @xmath107 to @xmath108 and @xmath103 is the path from @xmath108 to @xmath168 .all the vertices on the path @xmath101 ( @xmath102 , @xmath103 , respectively ) are adjacent to @xmath85 ( @xmath86 , @xmath87 , respectively ) .we first assume that there is no chord in @xmath21 between @xmath168 and a vertex on path @xmath101 .we consider the following two cases .* case 1 : no vertex of @xmath21 is adjacent to all of @xmath167 . *we label the vertices of @xmath21 in the clockwise order starting from @xmath169 and ending at @xmath133 , where @xmath133 is the number of vertices in @xmath21 .let @xmath163 and @xmath170 be the indices of vertices @xmath108 and @xmath107 , respectively .let @xmath129 be the index of the vertex that is the ( unique ) third vertex of the inner face of @xmath21 containing the edge @xmath137 .define the two graphs @xmath138 $ ] and @xmath139 $ ] as in the proof of lemma [ lem : stair ] .we first find a proportional contact representation of @xmath21 for @xmath9 restricted to the vertices of @xmath21 using rectangles in the plane , then we give necessary heights to this rectangles .draw @xmath139 $ ] using the * staircase algorithm * and delete the rectangles for @xmath129 and @xmath133 to obtain @xmath141 . draw rectangles @xmath142 and @xmath135 for @xmath129 and @xmath133 , respectively , so that the bottom side of @xmath142 touches the top side of @xmath135 , the left sides for both the rectangles have the same @xmath129-coordinate and and the right side of @xmath135 extends past @xmath142 .now place @xmath141 ( after possible scaling ) touching the right side of @xmath142 and the top side of @xmath135 ( this is possible since we can make the width of @xmath135 sufficiently long ) ; see fig .[ fig : prop ] .place the rectangle @xmath134 for @xmath13 touching the left sides of @xmath142 and @xmath135 such that its bottom side is aligned with @xmath135 and its top side is aligned with the top side of the rectangle for @xmath170 . to complete the rest of the drawing, we have the following two subcases : * case 1a : @xmath171 .* we draw @xmath138 $ ] using the * staircase algorithm * and delete from it the rectangles for @xmath13 and @xmath129 to obtain @xmath140 .we finally place @xmath140 ( after @xmath158 counterclockwise rotation and possible scaling ) touching the top side of @xmath134 and left side of @xmath142 ( this is possible by choosing a sufficiently large height for @xmath142 ) ; see fig .[ fig : prop](a ) . * case 1b : @xmath156 .* we draw @xmath138 $ ] using the * double - staircase algorithm * where @xmath163 is the pivot vertex . from this drawing, we delete the rectangles for @xmath13 and @xmath129 to obtain @xmath140 . finally place @xmath140 ( after @xmath158 counterclockwise rotation and possible scaling ) touching the top side of @xmath134 and left side of @xmath142 such that the topside of the rectangle for @xmath157 goes past the top side of @xmath142 ; see fig .[ fig : prop](b ) . .construction of the representation for a piece @xmath21 of @xmath0 , when ( a)(b ) no vertex of @xmath21 is adjacent to all of @xmath167 , and ( c)(e ) a vertex of @xmath21 is adjacent to all of @xmath167.,title="fig:",scaledwidth=88.0% ] .construction of the representation for a piece @xmath21 of @xmath0 , when ( a)(b ) no vertex of @xmath21 is adjacent to all of @xmath167 , and ( c)(e ) a vertex of @xmath21 is adjacent to all of @xmath167.,title="fig:",scaledwidth=88.0% ] so far we used the function @xmath9 to assign areas for the rectangles and obtained proportional box - contact representation of @xmath21 from the rectangles by assigning unit heights .however , by changing the areas for the rectangles , we can obtain different heights for the boxes .we will use this property to maintain adjacencies with @xmath167 , as well as to maintain the invariant . specifically , once we get the box representation of @xmath21 , we scale it by increasing the heights for the boxes , so that when we place it at the corner created by the t - shape for @xmath167 it will not intersect the representation for any of its sibling pieces in @xmath22 .consider the point @xmath71 which is the intersection of the lines containing the right side of the rectangle for @xmath163 and the top side of the rectangle for @xmath170 .we place @xmath30 such that the point @xmath71 superimposes on the corner for the t - shape in the projection on the @xmath166-plane . since the highest faces of @xmath86 , @xmath87 and @xmath85 are in this order according to @xmath131-coordinate , the adjacencies of the vertices in @xmath21 with @xmath167 are correct . by appropriately choosing the areas for the rectangles , we ensure that all the boxes for the vertices of @xmath21 have their highest faces above that of @xmath86 and that the invariant is maintained .* case 2 : a vertex of @xmath21 is adjacent to all of @xmath167 . * in this case at least one of @xmath167 has only one neighbor in @xmath21 .assume first that a vertex @xmath28 ( @xmath172 ) of @xmath21 is adjacent to all of @xmath167 and this is the only neighbor of @xmath86 ; see fig . [fig : prop](c ) .then we follow the steps for _ case 1a _ with @xmath173 ( and some vertex between @xmath129 and @xmath28 as @xmath163 ) .but when we finally place this representation of @xmath21 on the corner for the t - shape of @xmath167 we find the point @xmath71 to superimpose on this corner as follows .the point @xmath71 is on the line containing the top side of the rectangle for @xmath28 and has @xmath129-coordinate between the right sides of the rectangles for @xmath28 and @xmath174 .if a vertex @xmath29 ( @xmath175 ) is adjacent to all of @xmath167 and is the only neighbor of @xmath87 in @xmath21 , then we follow the steps of _ case 1b _ with @xmath176 and @xmath177 ; see fig .[ fig : prop](d ) .we find the point @xmath71 to superimpose on the corner for the t - shape of @xmath167 as follows .the point @xmath71 is on the line containing the top side of the rectangle for @xmath177 and has @xmath129-coordinate between the left sides of the rectangles for @xmath13 and @xmath178 .if a vertex @xmath27 ( @xmath179 ) is adjacent to all of @xmath167 and is the only neighbor of @xmath85 in @xmath21 , then we number the vertices of @xmath21 in the clockwise order starting from the clockwise neighbor of @xmath27 and ending at @xmath180 ; see fig .[ fig : prop](e ) .we use the * staircase algorithm * to find a representation of @xmath21 with rectangles and give necessary heights to obtain a representation with boxes . on the corner for the t - shape of @xmath167 , we superimpose the intersection point for the lines containing the top side of the rectangle of @xmath133 and the right side of the rectangle for @xmath108 .finally , we consider the case when there is a chord between @xmath168 and another vertex on the path @xmath101 .take the innermost such chord and let its other end - vertex be @xmath100 .then consider the two subgraphs @xmath181 and @xmath182 induced by all the vertices outside the chord and inside the chord ( along with the two vertices @xmath168 and @xmath100 ) .@xmath181 does not contain any chord from @xmath168 ; thus we use the algorithm above to obtain a representation of @xmath181 ; denote this by @xmath97 . in this representation @xmath168 and @xmath100 will play the roles of @xmath13 and @xmath133 , respectively .each vertex of @xmath182 is adjacent to @xmath85 and we find a proportional contact representation of @xmath182 and attach it with @xmath97 as follows .we use the * staircase algorithm * to find a proportional contact representation of @xmath182 with rectangles in the plane and delete the rectangles for @xmath168 and @xmath100 from it to obtain @xmath183 . in @xmath97 , we change the height of the rectangle @xmath134 for @xmath169 to increase its area so that its bottom side extends past the bottom side of the rectangle @xmath135 for @xmath184. then we place @xmath183 ( after reflecting with respect to the @xmath129-axis and possible scaling ) touching the right side of @xmath134 and the bottom side of @xmath135 .since the * staircase algorithm * can accommodate any given area for the layout , we can change the heights of the boxes for the vertices in @xmath182 to maintain the invariant .thus with the top - down traversal of @xmath22 , we obtain a proportional contact representation for @xmath89 .we assumed that each piece of @xmath89 is maximal outerplanar .however in the contact representation , for each edge @xmath1 , either a face of the box @xmath185 for @xmath2 is adjacent to the box @xmath186 for @xmath3 and no other box ; or a face of @xmath186 is adjacent to @xmath185 and no other box . in both cases the adjacency between these two coxes can be removed without affecting any other adjacency .thus this algorithm holds for any nested outerplanar graph @xmath89 .we proved that nested maximal outerplanar graphs have cube - contact representations and nested outerplanar graphs have proportional box - contact representations .these classes of graphs are special cases of @xmath11-outerplanar graphs , and the set of @xmath11-outerplanar graphs for all @xmath14 is equivalent to the class of all planar graphs . even though our approach might generalize to large classes , cube - contact representations and proportional box - contact representations are still open for general planar graphs .

<S> we study two variants of the problem of contact representation of planar graphs with axis - aligned boxes . in a _ cube - contact representation </S><S> _ we realize each vertex with a cube , while in a _ proportional box - contact representation _ each vertex is an axis - aligned box with a prespecified volume . </S><S> we present algorithms for constructing cube - contact representation and proportional box - contact representation for several classes of planar graphs . </S>

1409.3499

cellular responses to environmental fluctuations often involve biochemical reactions that are intrinsically stochastic .for example , stochasticity ( noise ) plays an important role in processes leading to gene expression @xcite and in biochemical switching between distinct states @xcite .regulation of noise in these processes is critical for the maintenance of cellular functions as well as for the generation of phenotypic variability among clonal cells .quantitative modeling of mechanisms of noise regulation is thus a key step towards a fundamental understanding of cellular functions and variability .noise regulation in cells is typically implemented by regulatory proteins such as transcription factors ( tf ) .recent research has demonstrated that , at the single - cell level , regulatory proteins are often produced in bursts @xcite .such proteins can further be involved in autoregulation ( e.g. the tat regulatory protein which controls the latency switch of hiv-1 viral infections ) @xcite or in downstream regulation of biochemical switches ( e.g. switching of flagellar rotation states in bacterial chemotaxis ) @xcite .some interesting questions arise from these observations : how does feedback from proteins produced in bursts regulate noise in gene expression and biochemical switching ?how can gene expression parameters be tuned to optimize noise in the presence of bursting and feedback ?the aim of this letter is to address these questions by developing a gene expression model that combines bursting and feedback for which we obtain the exact stationary distribution .previous work on noise in gene expression has focused on exact analytical solutions for models with : a ) bursting but no feedback effects @xcite or b ) feedback effects but no protein production in bursts @xcite .similarly , previous work on noise - induced biochemical switching @xcite does not consider the case of input noise source produced in bursts . in this letter , we introduce a _model that addresses these gaps in the field .our model reduces to multiple previously studied models in limiting cases .we obtain exact analytical distributions that significantly extend previously obtained results and lead to new insights .* model : * a schematic representation of the model is shown in fig.[fig : model ] . here 0 and 1 represent the inactive and active state of the promoter , respectively .note that the terms inactive / active are simply used to label the two states since protein production can occur from either state .specifically , protein production from the inactive ( active ) state occurs with rate @xmath0 ( @xmath1 ) .each production event results in a random burst of proteins and we assume that these bursts are distributed geometrically with mean size @xmath2 .the degradation rate of proteins is denoted by @xmath3 .the rate of switching from active to inactive state is denoted by @xmath4 .the rate at which the inactive state switches to the active state has two contributions : the spontaneous contribution with rate @xmath5 , and the feedback contribution , with rate @xmath6 ( where @xmath7 is the number of proteins , and @xmath8 measures the strength of the feedback ) .the linear dependence on @xmath7 for the feedback term is consistent with experimental characterization of the genetic circuit for expression of hiv-1 tat protein @xcite . .rate of transition from inactive to active state is @xmath9 , and that from active to inactive is @xmath4 .( b ) for @xmath10 the model maps onto a two - state switch driven by a bursty input source.,width=321 ] since we allow protein production from both active and inactive states , the model can be used to analyze the effects of positive feedback as well as negative feedback .when @xmath11 , the feedback term enhances protein production leading to positive feedback . in contrast , @xmath12 leads to negative feedback . for @xmath13 ,protein production is independent of the promoter state . as indicated in fig .1b , the model then corresponds to a bursty input noise source controlling switching between the two states .thus the same model can be used to analyze the impact of input protein noise on the statistics of a simple two - state switch @xcite .let @xmath14 denote the probability to find , at time @xmath15 , the promoter in state @xmath16 ( @xmath17 ) with @xmath7 proteins in the cell .the temporal evolution of @xmath14 is given by the following master equations @xcite : @xmath18p_{0,n},\nonumber\\ \partial_t p_{1,n}&=&k_1\sum_{p=0}^ng(p)p_{1,n - p}+\mu ( n+1)p_{1,n+1}\nonumber\\&+&(\alpha+\tilde{\alpha}n ) p_{0,n}-[k_1+\beta+\mu n]p_{1,n},\end{aligned}\ ] ] where @xmath19 is the protein burst distribution .to proceed further , let us define the generating functions @xmath20 with @xmath17 . correspondingly , eq . ( [ mastereq1 ] )can be recast as @xmath21 where @xmath22 is the generating function of the protein burst distribution given by @xmath23 . in the long time limit ,( [ g0g1 ] ) is used to derive an equation for the generating function of the protein steady - state distribution , @xmath24 . after a sequence of transformations( see supplementary information ) eq .( [ g0g1 ] ) reduces to a hypergeometric differential equation , leading to the solution : @xmath25^{k_1/\mu } \\ & & \times \frac{_2f_1[u , v|u+v+1-w|1-\phi\{1+b(1-z)\}]}{_2f_1[u , v|u+v+1-w|1-\phi]},\nonumber\end{aligned}\ ] ] where the quantities , @xmath26 , @xmath27 , @xmath28 and @xmath29 are related to model parameters by @xmath30 @xmath31 with @xmath32 , and @xmath33 represents the gaussian hypergeometric function .this solution for the generating function is the central result of this paper .it can be shown that our result reduces to previously obtained results in different limiting cases ( supplementary information ) .it can be used to derive exact analytical results for several quantities of interest .for example , the steady - state probability that the promoter is in state @xmath34 ( @xmath35 ) is given by ( see supplementary information ) @xmath36}{{_2f_1}[u , v , u+v+1-w,1-\phi]}.\end{aligned}\ ] ] furthermore , eq.([solution ] ) can be used to obtain an analytical expression for the protein steady - state distribution @xmath37 and to analyze the corresponding moments .these expressions lead to quantitative insights into multiple topics of current research interest as discussed below .* regulation of protein noise : * there has been considerable focus in previous work on analyzing the effects of feedback on the noise @xmath38 characterizing the protein steady - state distribution @xcite . to analyze the impact of feedback , we first compare the noise for the case with feedback ( @xmath39 ) to the case without feedback ( @xmath40 ) in fig.[fig : fig2](a , b ) .it is interesting to observe that negative feedback increases the noise when compared to the case without feedback : @xmath41 . on the other hand, positive feedback leads to a decrease of noise @xmath42 . while this may appear surprising given previous results @xcite , this observation is consistent with recent results from simulations @xcite .it should be further noted that negative feedback leads to a reduction in mean levels whereas positive feedback increases mean levels ; thus the changes in @xmath43 can be driven largely by changes in the mean levels .it follows that to determine the effects of feedback on noise , it is desirable to compare models which give rise to the same mean levels . as a function of @xmath2 and @xmath8 for a ) negative feedback ( @xmath44 ) andb ) positive feedback ( @xmath45 ) . in the lower panel ,comparison between the original and effective models : density plots for @xmath46 are plotted as a function of @xmath2 and @xmath8 for c ) negative feedback ( @xmath44 ) and d ) positive feedback ( @xmath45 ) .other parameters are : @xmath47.,width=340,height=226 ] to address this issue , we introduce an effective model with no feedback that is characterized by a _ constant _rate @xmath48 for promoter switching from the inactive to active state .the parameter @xmath48 is determined analytically ( supplementary information ) by the condition that the mean protein levels @xmath49 and @xmath50 are identical in the original and effective models .the remaining parameters are the same as in the original model . in the following ,we compare noise in protein distributions for the original and effective models .fig.[fig : fig2](c , d ) illustrates the ratio @xmath46 for negative as well as positive feedback . for negative feedback ,protein noise in the original model is lower than the noise in the effective model , i.e. the effect of negative feedback is to reduce noise . for positive feedback , as shown in fig.[fig : fig2]d , we observe that feedback increases the noise when compared to the noise for the effective model .thus , in the context of regulation of protein noise , the results obtained indicate that the choice of reference model plays a critical role . in relation to the model without feedback , we observe that negative ( positive ) feedback increases ( decreases ) the noise .however , in relation to the effective model , which preserves the average number of proteins , the opposite behaviour is observed .fig.[fig : fig2](c , d ) also indicates that the effective model provides a useful approximation to the original model for a wide range of parameters , in particular for positive feedback . in this case , for the range of parameters considered in fig .[ fig : fig2](d ) , the effective model provides a good approximation in regions of parameter space for which a ) @xmath51 or b ) @xmath52 . in the former case ,fluctuations in protein levels make a negligible contribution to the promoter switching rate , whereas the latter condition represents ( almost ) constitutive production of protein bursts , making the effective model indistinguishable from the original model .however , there is a class of problems for which the effective model is inadequate and it is necessary to analyze the complete model .an important example includes noise optimization in the presence of feedback by varying system parameters , as discussed in the following .* noise minimization : * recent work @xcite has analyzed noise minimization due to negative feedback for a model similar to the one outlined in fig .1 . in this model ,the binding of a tf switches its promoter to a repressed state , ( i.e. set @xmath53 ) and the switching rate @xmath4 corresponds to the dissociation rate of the tf from its promoter . in the limit @xmath54 , there is no feedback , since proteins bursts are effectively produced constitutively with rate @xmath0 . to examine noise minimization , the system parameters @xmath0 and @xmath4are varied subject to the constraint that the mean protein number @xmath49 is held fixed .in particular , it is of interest to determine : a ) the minimum dissociation rate , @xmath55 , required for negative feedback to result in a reduction of noise relative to the model with no feedback .b ) the optimal rate @xmath56 at which noise suppression is maximal . in the following ,we explore the insights gained for this problem ( for the model in fig .1 ) using exact analytical results for moments derived using eq .( [ solution ] ) .3 illustrates the variation of protein noise @xmath43 as a function of tf dissociation rate @xmath4 , keeping the mean protein levels fixed by changing the transcriptional rate @xmath0 . in the limit @xmath54 ( i.e no feedback ), we have @xmath57 .as @xmath4 is reduced , the noise initially decreases , reaches a minimum value at @xmath58 and subsequently increases .in contrast , for the corresponding effective model with a constant rate of promoter transitions ( as defined in the preceding paragraphs ) , we have @xmath59 for all finite @xmath4 , i.e. there is no noise reduction .this indicates that it is essential to consider the role of fluctuations in the rate of promoter transitions ( from active to repressed state ) in understanding noise reduction due to negative feedback .the parameter @xmath55 can be determined by the condition that for @xmath60 , we have @xmath61 . the exact expression for @xmath43 in combination with some approximations , specifically @xmath62 ,can be used to derive the result obtained in @xcite for @xmath63 ( supplementary information ) .our analysis indicates that this rate corresponds to @xmath64 which implies that for noise reduction , the rate of tf dissociation must be greater than the rate of arrival of nonzero bursts of proteins .our results can also be used to analyze the optimal value @xmath65 at which noise suppression is maximal .the results derived in @xcite using moment - closure approaches serve as a good approximation in the limit of large @xmath49 .since our exact results apply for arbitrary parameter values , they can be used to connect large @xmath49 results with those for low @xmath49 .as discussed below , this analysis leads to some interesting observations . for @xmath66 .corresponding variations for the optimal dissociation rate @xmath56 and the probability @xmath50 are plotted for different values of @xmath49 in the inset , dotted lines representing the prediction of ref .other parameters are : @xmath67.,width=207,height=151 ] in ref .@xcite , it was shown that the optimal value of dissociation rate @xmath56 is linearly dependent on @xmath68 , i.e. @xmath69 remains constant as @xmath49 is varied by changing @xmath0 . as expected , we recover this feature when @xmath49 is large ( see fig .however , for small @xmath49 we see a strong deviation from the large @xmath49 limit , characterized by non - monotonic variation .furthermore , this non - trivial variation in the optimal dissociation rate is also reflected in the probability of the promoter state being transcriptionally active , @xmath50 .as shown in the figure , @xmath50 decreases monotonically with @xmath49 and in the limit of large @xmath70 it approaches the result derived in ref .thus our results predict that , for optimal noise suppression in low abundance tfs , the fraction of time that the promoter is active decreases as we increase tf concentration . * switching statistics : * as noted in fig .1b , when @xmath71 , the model analyzed can be mapped to a two - state system driven by a _bursty _ input signal .several cellular systems can be modeled ( at a coarse - grained level ) as two - state switches ; thus it is of interest to explore how such switches respond to fluctuating inputs @xcite .the results obtained in this work lead to exact analytical expressions for the corresponding switch statistics .the quantity of interest is the variance of the switch , @xmath72 , with @xmath50 given by eq.([ep0 ] ) .note that eq.([ep0 ] ) is valid for proteins produced in geometrically distributed bursts with mean burst size @xmath2 .on the other hand , previous work @xcite has considered the case such that each burst leads to creation of exactly one protein ( i.e. protein dynamics is a simple birth - death process ) .remarkably , there exists a mapping between the analytical solutions in these two cases ( supplementary information ) . using this mapping , we obtain the following exact result for the problem considered in previous work @xcite @xmath73.\nonumber\end{aligned}\ ] ] as expected , the above expression , eq.([ep02 ] ) , reduces to analytical results derived in @xcite ( for @xmath74 ) in limiting cases .for example , in the slow switching limit , i.e. @xmath75 , eq . ( [ ep02 ] ) leads to @xmath76 , which is identical to the result obtained in @xcite .similarly , in the fast switching limit @xmath77 , if we further set @xmath78 and @xmath79 , we obtain @xmath80 , consistent with the result obtained in @xcite . the exact result derived above , eq.([ep02 ] ) , allows for analysis of switching statistics beyond these limits , i.e. throughout parameter space .furthermore , the results derived can be used to explore how bursty protein production affects switching statistics .fig.[fig : switch]a shows how the switch variance @xmath81 depends on the burst size , @xmath2 , and the average number of proteins , @xmath68 .some interesting observations can be made which highlight the nontrivial variation of @xmath81 with bursting . for large @xmath49 values, @xmath81 shows a non - monotonic variation with @xmath2 , with a maximum at a critical burst size , @xmath82 ( see fig.[fig : switch]b ) . on the other hand , for low @xmath68 , we observe that @xmath81 decreases monotonically with @xmath2 with the maximum corresponding to @xmath83 ( fig.[fig : switch]b ) .these different behaviors can be understood based on the following observations : 1 ) for fixed @xmath49 , @xmath50 increases with increasing burst size @xmath2 , and 2 ) , for fixed @xmath2 , @xmath50 decreases with increasing @xmath49 .thus , in the limit @xmath84 , for @xmath49 such that @xmath85 we obtain a monotonic decrease in the variance @xmath72 as @xmath2 is increased . on the other hand , for @xmath49 such that @xmath86 ( in the limit @xmath87 ) we obtain non - monotonic variation with @xmath2 . is shown as a function of @xmath2 and @xmath70 .b ) variations of @xmath81 with @xmath2 for two different values of @xmath88 , 1(solid ) and 10(dashed ) . in both a and b , @xmath89 .c)variation of @xmath90 with @xmath2 , different lines correspond to different values of feedback strength @xmath91 ( solid ) , @xmath92 ( dashed ) , @xmath93 ( dotted ) . in all plots ,other parameters are : @xmath94 , @xmath95.,width=302,height=151 ] next , we focus on the variation of @xmath81 with mean protein @xmath49 for a fixed @xmath2 . as can be seen in fig.[fig : switch]a , it shows a non - monotonic variation , with @xmath81 being maximum at a critical mean protein level , @xmath96 . often , it is of interest to estimate the value @xmath96 which maximizes the noise in switching statistics . in fig.[fig : switch]c, we compare the mean - field estimate @xmath97 ( obtained by replacing the fluctuating @xmath7 by its mean value , @xmath98 ) with the corresponding exact value . as indicated in the figure , deviation from the mean - field estimateis significant and increases with increasing burst - size and feedback strength .the analytical results derived are thus useful in obtaining accurate estimates of parameters that maximize noise in switching statistics . to conclude, we have studied an exactly solvable model that integrates key features of regulation of gene expression , specifically : bursting , promoter switching and feedback , in a single model .the derived results provide new insights into the roles of bursting and feedback ( both positive and negative ) in fine - tuning noise in protein distributions .furthermore , the results obtained can serve as building blocks for future studies focusing on noise optimization strategies by varying the underlying parameters .the model developed can also be applied to study the statistics of a simple two - state switch driven by a bursty protein noise source .our results show that such bursty input noise can induce strong deviations in the optimal parameters for switch variance from the corresponding mean - field predictions .the development of analytical approaches , as outlined in this work , is thus an important ingredient for accurate quantitative modeling of stochastic cellular processes .the authors acknowledge funding support from the nsf through awards no .phy-1307067 and dms-1413111 .a. raj and a. van oudenaarden , cell * 135 * , 216 ( 2008 ) .a. eldar and m. b. elowitz , nature * 467 * , 167 ( 2010 ) .a. sanchez , s. choubey , and j. kondev , annual review of biophysics * 42 * , 469 ( 2013 ) .j. l. spudich and d. e. koshland , nature * 262 * , 467(1976 ) .h. c. berg and e. m. purcell , biophys .j. * 20 * , 193 ( 1977 ) .h. park , p. oikonomou , c.c .guet and p. cluzel , biophys .j. * 101 * , 2336(2011 ) .b. hu , d. a. kessler , w. j. rappel , and h. levine , phys .lett . * 107 * , 148101 ( 2011 ) .l. cai , n. friedman , and x. s. xie , nature ( london ) * 440 * , 358 ( 2006 ) .j. yu , j. xiao , x. ren , k. lao , and x. s. xie , science * 311 * , 1600 ( 2006 ) .i. golding , j. paulsson , s. m. zawilski , and e. c. cox , cell * 123 * , 1025 ( 2005 ) .a. raj , c. s. peskin , d. tranchina , d.y .vargas , and s. tyagi , plos biol .* 4 * , e309 ( 2006 ). j. chubb , t. trcek , s. shenoy , and r. singer , curr .* 16 * , 1018 ( 2006 ) .l. s. weinberger , j. c. burnett , j. e. toettcher , a. p. arkin , and d. v. schaffer , cell * 122 * , 169 ( 2005 ) .tsz - leung to and n. maheshri , science * 327 * , 1142 ( 2010 ) .a. singh and j. p. hespanha , biophys .j. * 96 * 4013 ( 2009 ) .m. voliotis and c. g. bowsher , nucleic acids res .* 40 * , 7084 ( 2012 ) .h. feng , b. han , and j. wang , j. phys .b * 115 * , 1254 ( 2011 ) .j. e. m. hornos et .al , phys .e * 72 * , 051907 ( 2005 ) .p. visco , r. j. allen , and m. r. evans , phys .* 101 * , 118104 ( 2008 ) .v. shahrezaei , and p. s. swain , proceedings of the national academy of sciences * 105.45 * , 17256 ( 2008 ) .r. grima , d. r. schmidt , and t. j. newman , j. chem . phys . * 137 * , 035104 ( 2012 ) . y. vandecan and r. blossey , phys .e * 87 * , 042705 ( 2013 ) . n. g. van kampen .stochastic processes in physics and chemistry .elsevier , 3rd edition , 2007 .m. l. simpson , c. d. cox , and g. s. sayler , proceedings of the national academy of sciences * 100 * , 4551 ( 2003 ) .a. becskei , and l. serrano .nature * 405 * , 590 ( 2000 ) .a. gronlund , p. lotstedt , and j. elf , nat . commun . * 4 * , 1864 ( 2013 ) . * supplementary material :in the steady state , eq . ( 2 ) in the main text can be written as @xmath99g_0-\left[\mu(1-z)-\tilde{\alpha}z\right]\partial_zg_0,\nonumber \\ \alpha g_0+\tilde{\alpha}z\partial_zg_0&=&\left[k_1(1-\tilde{g})+\beta\right]g_1-\mu(1-z)\partial_zg_1.\end{aligned}\ ] ] by eliminating @xmath100 in eq .( [ ed1 ] ) , we can write a single equation in terms of @xmath101 , @xmath102 with @xmath103+\mu ( 1-z)k_0\partial_z \tilde{g } , \nonumber\\ b(z)&=&\mu(1-z)\left [ ( k_0 + k_1)(1-\tilde{g})+\alpha+\tilde{\alpha}+\beta+\mu \right]-z\tilde{\alpha}k_1(1-\tilde{g}),\nonumber \\ c(z)&=&\mu(1-z)\left[\mu(1-z)-z\tilde{\alpha}\right].\end{aligned}\ ] ] writing @xmath104 in the second equation of eq .( [ ed1 ] ) , we can express @xmath101 in terms of @xmath105 which reads as @xmath106 using eq .( [ ed4 ] ) in the first equation of eq.([ed1 ] ) , and taking @xmath107 followed by a transformation @xmath108 leads to the following equation for @xmath105 .@xmath109 where @xmath110 ^ 2}\left [ bx\left\{k_0k_1+\alpha k_0+\beta k_1 \right\}+ \alpha k_0+\beta k_1\right ] , \nonumber \\ b(x)&= & \frac{\mu}{1+bx}\left [ bx\left\ { k_0+k_1+\alpha+\beta+\mu\right\ } + \alpha+\beta \right ] - \frac{\tilde{\alpha}(1-x)}{1+bx}b(k_1+\mu),\nonumber \\ c(x)&=&\mu\left[x(\mu+\tilde{\alpha } ) - \tilde{\alpha } \right].\end{aligned}\ ] ] applying the transformation , @xmath111f(x)$ ] , eq .( [ ed6 ] ) can be rewritten as @xmath112 with @xmath113,\nonumber\\ \bb(x)&=&b(x)+2c(x)\partial_xf,\nonumber\\ \cc(x)&=&c(x).\end{aligned}\ ] ] setting @xmath114 and changing @xmath115 to @xmath116 using the transformation , @xmath117 with @xmath118 , the resulting equation for @xmath119 reduces to the hypergeometric differential equation : @xmath120\partial_\xi f(\xi)+\xi(\xi-1)\partial_\xi^2 f(\xi)=0,\ ] ] where @xmath26 , @xmath27 and @xmath28 are given by eq.(4 ) , and corresponding solution for @xmath121 is given in eq .( 3 ) in the main text .here we discuss how the exact steady - state generating function , @xmath121 , reduces to known results in different limiting cases : + ( 1 ) in the absence of feedback , i.e. @xmath122 , the generating function reduces to @xmath123/[1+b(1-z)]^{k_1/\mu},\ ] ] which is identical to the result first derived in @xcite .+ ( 2 ) when the protein production rate in the inactive state vanishes ( @xmath124 ) and for no spontaneous switching to the active state ( @xmath74 ) , the system reaches , in the long - time limit , an absorbing state with no proteins .correspondingly , the expression for the generating function reduces to @xmath125 .+ ( 3 ) in the limit @xmath126 , we have a model with constitutive production of geometric bursts with rate @xmath0 .it follows that @xmath127^{k_0/\mu},\ ] ] which is the generating function for a negative binomial distribution as expected @xcite .+ ( 4 ) for @xmath128 or @xmath129 , we get @xmath127^{k_1/\mu}.\ ] ] thus the result obtained encompasses previously derived results and extends them to include the effects of bursting and feedback .starting from ( 2 ) , and adding the equations for @xmath101 and @xmath100 , we get @xmath130 taking the limit @xmath131 , this leads to @xmath132 replacing @xmath70 by its value we get @xmath133}{{_2f_1}[u , v , u+v+1-w|1-\phi]},\ ] ] which can be expressed as equation ( 5 ) .here we derive expressions for the mean protein levels and the constant rate of switching from the state 0 to state 1 in the effective model that maintains the same mean protein numbers as that of the original model . from the expression for the steady - state generating function ,we obtain the following expression for the mean number of proteins @xmath135}{_2f_1[u , v , u+v+1-w|1-\phi]}.\end{aligned}\ ] ] for the effective model with a constant rate of promoter transitions we have @xmath136 equating the two means , we get the general equation determining @xmath134 . the specific cases of positive and negative feedback in the main text are discussed below . for positive feedback ( with @xmath0=0 ) , the mean number of proteins for the original model can be expressed as @xmath137 whereas for the effective model with the same mean we have @xmath138 using eqs.([esn1 ] ) and ( [ esn2 ] ) we get @xmath139 similarly , for negative feedback ( with @xmath140 ) we get @xmath141here we illustrate how we can map a model with arrival of proteins in geometric bursts to a model for which each arrival leads to the production of a single protein .first , consider the geometric burst distribution conditional on the production of at least one protein ( i.e. ignore bursts which do not result in the production of any proteins ) .if the original burst distribution has mean @xmath2 , then the generating function of the corresponding conditional geometric distribution is given by @xmath142 .it is straightforward to see that the mean of the conditional geometric distribution is @xmath143 .if we now take the limit @xmath144 for the conditional geometric distribution , this corresponds to the case in which exactly one protein is produced in every burst .thus the simple poisson arrival process can be recovered as a limit of the model with arrival of _ conditional _ geometric bursts .now we note that the model described in fig .1b can be equivalently viewed from two perspectives : ( i ) geometric bursts ( with mean @xmath2 ) arriving at rate @xmath145 or ( ii ) conditional geometric bursts ( with mean @xmath146 ) arriving with rate @xmath147 , where @xmath148 is the probability that the burst produces at least one protein .thus the steady - state solution obtained in the main text is also the solution for a model with arrival of conditional geometric bursts .based on this observation , the results obtained also lead to exact results for models with conditional geometric bursts , using the mapping @xmath149 carrying out this mapping and taking the limit @xmath84 in eq .( 5 ) leads to the exact result for the problem considered in previous work @xcite .using eq . ( 9 ) , expression for the mean number of proteins is @xmath151 where @xmath152}{_2f_1[u , v , c,1-\phi]},\ ] ] @xmath153 and expressions for @xmath26 , @xmath27 , @xmath28 and @xmath29 are given by eq .( 4 ) . similarly , using eq . ( 3 ) , the exact expression for the protein noise can be written as a sum of two terms @xmath154 where @xmath155 is the noise contribution when proteins are produced constitutively , i.e. in the limit of @xmath156 , and @xmath157\ ] ] is the noise contribution due to promoter switching , and @xmath158 is given as @xmath159 _ 2f_1[u+2,v+2,c+2,1-\phi ] } { _ 2f_1[u+1,v+1,c+1,1-\phi]^2}-1.\ ] ] it is interesting to note that when the transition rate from the state 0 to 1 is a constant , the second term @xmath160 is a positive definite quantity and leads to monotonically increasing noise with decreasing @xmath4 . however , in the presence of feedback , the second term can be positive or negative or zero . the value of finite @xmath4 where @xmath161 corresponds to @xmath150 , which we wish to find .this is the minimum value of @xmath4 which marks the beginning of noise suppression due to negative feedback . to proceed further, we need to determine how @xmath0 varies with @xmath4 , given the constraint that @xmath49 is held fixed .this can be simplified by making the mean - field approximation : @xmath164 in combination with the exact equation @xmath165 .defining @xmath166 , the preceding approximation gives us @xmath167 . substituting this in eq .( [ eb6 ] ) , we obtain the solution @xmath168 this is identical to the result derived in @xcite .the preceding analysis highlights the assumptions needed to obtain this result from the exact results derived in the main text .the derivation also shows that we have @xmath169 the right - hand side of the above equation represents the rate of arrival of non - zero bursts of proteins at steady - state .thus the minimum dissociation rate has to be greater than the rate of arrival of non - zero bursts of protein production. 1 v. shahrezaei , and p. s. swain , proceedings of the national academy of sciences * 105.45 * , 17256 ( 2008 ) .b. hu , d. a. kessler , w. j. rappel , and h. levine , phys .lett . * 107 * , 148101 ( 2011 ) .a. gronlund , p. lotstedt , and j. elf , nat . commun . * 4 * , 1864 ( 2013 ) .

<S> stochasticity in gene expression can give rise to fluctuations in protein levels and lead to phenotypic variation across a population of genetically identical cells . </S><S> recent experiments indicate that bursting and feedback mechanisms play important roles in controlling noise in gene expression and phenotypic variation . </S><S> a quantitative understanding of the impact of these factors requires analysis of the corresponding stochastic models . </S><S> however , for stochastic models of gene expression with feedback and bursting , exact analytical results for protein distributions have not been obtained so far . here </S><S> , we analyze a model of gene expression with bursting and feedback regulation and obtain exact results for the corresponding protein steady - state distribution . </S><S> the results obtained provide new insights into the role of bursting and feedback in noise regulation and optimization . </S><S> furthermore , for a specific choice of parameters , the system studied maps on to a two - state biochemical switch driven by a bursty input noise source . </S><S> the analytical results derived thus provide quantitative insights into diverse cellular processes involving noise in gene expression and biochemical switching . </S>

astro-ph9801310

the development of the unified model for seyfert galaxies has opened interesting questions regarding the nature of the optical uv continuum in seyfert 2s .this continuum seems to be well understood only in the prototypical seyfert 2 galaxy ngc1068 ( antonucci & miller 1985 , miller , goodrich & mathews 1991 ) , where it is attributed to a combination of a red stellar population contribution plus a blue featureless continuum scattered towards us by electrons , and in mrk477 ( heckman et al .1997 ) , where the observed blue continuum is dominated by a dusty starburst .polarimetric studies by miller & goodrich ( 1990 ) , and later by tran , miller & kay ( 1992 ) , kay ( 1994 ) and tran ( 1995a , b , c ) have revealed polarized lines and continua in a number of additional seyfert 2 s , indicating the presence of a scattered seyfert 1 component . however , the low levels of polarization ( miller & goodrich 1990 ) , lack of detectable broad lines in the direct spectra ( cid fernandes & terlevich 1995 , heckman et al .1995 ) , and larger polarizations in the broad lines than in the continuum ( tran 1995a , b , c ) indicate the existence of another source of continuum , not accounted for by the starlight template nor by the reflected nuclear component .the origin of `` fc2 '' , as this extra component became known ( miller 1994 , tran 1995c ) is one of the most important current issues in seyfert 2s studies . the two currently contending scenarios are free - free emission from the scattering region ( tran 1995c ) , and young - stars , as advocated by cid fernandes & terlevich ( 1995 ) and heckman et al .( 1995 ) .when discussing the nature of the optical continuum in active galaxies , one must be aware of the importance of evaluating and removing the starlight `` contamination '' , particularly in seyfert 2s , where the stellar component often overwhelms the scattered flux .indeed , as emphasized by miller & goodrich ( 1990 ) , antonucci ( 1993 ) , kay ( 1994 ) and tran ( 1995a ) , the derived polarizations are very sensitive to the adopted stellar population template . the simple fact that different authors obtain very different starlight fractions for the same object ( e.g. , kay 1994 , tran 1995a ) is a clear demonstration of how critical and uncertain is the starlight removal procedure . in most polarimetric studies , as in fact in most optical spectroscopy studies of agn since koski ( 1978 ) , the stellar population template used to subtract the stellar contribution is the nuclear spectrum of an elliptical or early type galaxy . given a library of normal galaxy spectra spanning a sufficiently wide range of stellar population mixtures , this approach will certainly produce a satisfactorily agn - looking , smooth residual spectrum .however , given the non - uniqueness of the solution , one must always bear in mind that the adopted template may not correspond to the true stellar population of the host galaxy . andeven if the adopted template accounts for , say , 80% of the true stellar component , the error induced by not accounting for the remaining 20% may be critical when it comes to interpreting the nature of the residual continuum . in order to investigate the stellar populations in active galaxies using a different approach ,we ( cid fernandes , storchi - bergmann & schmitt 1997 , hereafter paper i ) have recently concluded a study where we have used long - slit optical spectroscopy of 42 galaxies to measure the variation of equivalent widths ( hereafter w ) of several stellar absorption lines , as well as continuum ratios as a function of distance from the nucleus .our goals were to map the stellar populations in the sample galaxies , and to obtain the stellar population characteristics right at the active nucleus via extrapolation of the properties measured in neighboring extranuclear regions .such an approach has been previously used by fosbury et al .( 1978 ) and by ourselves in previous works ( e.g. schmitt , storchi - bergmann & baldwin 1994 ; storchi - bergmann et al . 1995 , 1997a ) .in paper i we found that the presence of a featureless continuum in the nucleus of seyfert 1 galaxies was revealed by a bluening of the continuum and by a decrease of the w s of the absorption lines towards the nucleus . in galaxies with nuclear starbursts or rings of star formation, the same behavior was observed : the star - forming regions present bluer colors and smaller w s .one unexpected result was the finding that , in the majority of seyfert 2 galaxies , there is neither a dilution of the w s nor a bluening of the continuum towards the nucleus , indicating that the featureless continuum , if present at the nucleus , contributes with less than @xmath210% of the flux in the optical waveband .the only clear cases of detection of dilutions and blue continua were the seyfert 2 galaxies already known to harbor `` composite nuclei '' , with both seyfert and starburst activity , as evidenced by the emission - line ratios and/or the presence of absorption features of young stars .these results are illustrated in figure 1 , where we show the variation of w(ca ii k ) , w(g band ) , w(mg i@xmath3mgh ) and of the continuum ratio @xmath45870/@xmath44020 as a function of distance from the nuclei for the seyfert 1 galaxy ngc6814 , for the galaxy with composite nucleus ( seyfert 2 and starburst ) ngc7130 and the seyfert 2 galaxy mrk348 . while the decrease of the w s and bluening of the continuum is obvious in the two former galaxies ,they are not observed in the latter . in this paper , we investigate the above results more closely for four seyfert 2 galaxies : mrk348 , mrk573 , ngc1358 and mrk1210 , as previous works have reported dilution factors up to @xmath550% for these galaxies . in paperi we have shown that our method allows the detection of dilutions larger than 510% ( which correspond to the uncertainties in the w s ) , but for these galaxies we have not detected significant dilutions in the absorption features of the stellar population up to at least @xmath5 10 from the nucleus . this behavior can be observed in fig . 1 for mrk348 and in similar plots for mrk573 , ngc1358 and mrk1210 ( figs. 17 , 21 and 25 from paper i ) .we investigate the reason for this lack of dilution and the implications for the nature of the nuclear continuum and stellar population .the data discussed here have been described in paper i. in summary , they consist of long - slit spectra obtained with the cassegrain spectrograph and reticon detector at the 4 m telescope of the cerro tololo inter - american observatory .the pixel size was @xmath2 1 and the slit width 2 .for the four galaxies discussed here , the observations were obtained under photometric conditions with a seeing of @xmath2 1 .the slit was oriented along the radio axis and/or along the extended emission , but the observations were scheduled to make this angle coincide with the paralactic angle , in order to avoid the effects of differential refraction .the spectral resolution ( fwhm of the calibration lamp lines ) is @xmath2 7 .a summary of the observations is presented in table 1 .the spectra were reduced and flux calibrated using standard procedures in iraf , and one - dimensional spectra were extracted along the spatial direction in windows of 2@xmath62 . prior to the analysis , the spectra were corrected for the foreground milky way reddening , using the empirical selective extinction function of cardelli , clayton & mathis ( 1989 ) .the corresponding e(b - v ) values are listed in the last column of table 1 and were extracted from the nasa extragalactic database .mrk348 , mrk573 , ngc1358 and mrk1210 are nearby galaxies , whose bulges have effective radii ranging from 6 to 36 , as determined by near - infrared surface photometry ( alonso - herrero , ward & kotilainen , 1996 ; heisler , de robertis & nadeau 1996 ; xanthopoulos 1996 ) . thus , the bulge is completely resolved in these galaxies , and spectra extracted at less than 6 from the nucleus will still be sampling the bulge stellar population .we believe that using a stellar population template from the bulge of the same galaxy to represent the stellar population of the nucleus is in principle a more robust approach than adopting a population template from other galaxies , as paper i shows that the stellar population varies substantially from galaxy to galaxy .the smooth radial variation of the w s of absorption lines found in paper i , also supports this approach .the w s only show abrupt changes when there is evidence of the presence of a blue stellar population in the starburst galaxies or in the rings of star formation or a featureless continuum in the seyfert 1 galaxies . moreover ,as pointed out in the introduction , previous works usually adopt an elliptical galaxy template to represent the stellar population in seyfert nuclei , whereas recent studies ( wyse , gilmore & franx 1997 ) show that stellar populations in bulges frequently differ from that in ellipticals .jablonka , martin & arimoto ( 1996 ) show that bulges occupy regions similar to that of ellipticals in the fundamental plane , but are displaced towards smaller line strengths .balcells & peletier ( 1994 ) showed that , although bulges follow a color - magnitude relationship similar to that of ellipticals , they have a larger scatter and are bluer than ellipticals of similar luminosity .the use of an elliptical galaxy template to subtract the stellar population contribution of the bulge of an spiral or lenticular galaxy is thus likely to produce a residual blue continuum due to a `` template mismatch '' ( cid fernandes & terlevich 1992 ) . these facts , together with the discrepant results obtained by different authors using the traditional elliptical galaxy template method , motivated us to seek for an alternative starlight evaluation technique . in this sectionwe compare the bulge extranuclear spectra with the nuclear spectra of mrk348 , mrk573 , ngc1358 and mrk1210 , in order to investigate if the former can be used as stellar population templates for the latter .two criteria were used to select the windows for the extranuclear spectra : ( 1 ) to keep the smallest possible angular distance from the nucleus always smaller than the effective radius of the bulge in order to minimize changes in the stellar population characteristics ; ( 2 ) to ensure that the distance was large enough to avoid contamination from nuclear light due to seeing effects .the nuclear and extranuclear windows all have sizes of 2@xmath62 .in order to construct the extranuclear spectra , we have skipped the windows adjacent to the nucleus and averaged the two extractions centered at 4 at both sides of the nucleus , which include light from 3 to 5 from the nucleus .as the seeing was @xmath21 in the observations , we do not expect significant contamination by nuclear light at these windows .the corresponding average distances from the nuclei at the galaxies are 1.2 , 1.4 , 1.0 and 1.0 kpc , respectively for mrk348 , mrk573 , ngc1358 and mrk1210 .we compare the nuclear with the extranuclear optical ( 3500 to 6000 ) spectrum of each galaxy in figs .2 to 5 . as all galaxies present extended emission, we have chopped the emission lines from the extranuclear spectra in order to allow a more clear comparison of the absorption features .an extranuclear spectrum is considered to be an acceptable fit to the nuclear one if the nuclear w s are reproduced within 10% and the continuum within 5% . at the blue end of the spectra ,somewhat larger differences in the continua were allowed ( @xmath210% ) due to the larger noise in this spectral region .these values correspond to the maximum uncertainties in the measurements of w s and continua in paper i. the residuals in the continuum were calculated as averages of the difference spectra ( nuclear@xmath7extranuclear ) , in windows avoiding emission lines .the difference spectra are also shown in figs .25 . in all cases ,the nuclear continuum is slightly redder than the extranuclear one ( see paper i ) .assuming that this is due to an excess reddening of the nuclear relative to the extranuclear spectra ( in agreement with the accepted scenario that the nuclear regions in seyfert 2 galaxies are dusty ) , we have reddened the extranuclear spectra by the necessary amount to make the continuum slopes to match that of the nuclear spectra .we now discuss the results for each galaxy .figure 2a shows the comparison of the nuclear with the reddened [ e(b - v)=0.03 ] extranuclear spectrum of mrk348 , after normalization of the latter to the flux of the nuclear spectrum near 5500 , as well as the residual between the two .the nuclear continuum is well reproduced ( within 5% ) by the extranuclear one after a small reddening of the latter , and , in paper i , we have also concluded that the equivalent widths of absorption lines are the same within the uncertainties ( @xmath810% ) .these results indicate that the nuclear stellar population can be reproduced by that of the extranuclear spectrum ; no extra spectral component seems to be needed within the margins of our matching criteria .we now compare our results with those obtained by tran ( 1995a , b , c ) .tran has shown that mrk348 presents a seyfert 1 spectrum in polarized light , with a continuum polarization of 5.7% ( after starlight correction ) , increasing to more than twice this value in broad h@xmath9 .the stellar component , represented by the elliptical galaxy ngc 821 , was found to contribute 73% of the total flux at @xmath45500 , while our results suggest a much larger fraction .tran s analysis of the spectropolarimetry data also showed that fc1 ( the scattered seyfert 1 component in tran s nomenclature ) , amounts to 5% of the flux at @xmath45500 , and that it has a @xmath10 shape .unlike the starlight fraction , the fc1 strength is not strongly dependent on the choice of template , since tran s determination takes into account the strength necessary to match the polarization of the continuum to that of the broad lines .similarly , the @xmath11 shape is not critically affected by the template , as it is implicit in the polarized spectrum ( @xmath12 ) .since we know that a scattered component is present , the spectral decomposition in fig .2a is not physically acceptable , as it corresponds to 100% starlight . yet , as shown below , there is room for a small contribution from the scattered component . adopting a f@xmath13 shape for the reflected continuum , a contribution of 5% at @xmath45500 corresponds to a decrease of @xmath25% of w(mgi+mgh ) , somewhat larger for w(g band ) , and up to @xmath210% for w(caii k ) , thus within our measurement uncertainties for the w s . in order to illustrate that both w s and continuum shape are in agreement with the presence of such a scattered nuclear continuum, we have combined the extranuclear spectrum with a f@xmath13 power - law in proportions of 95% and 5% at @xmath45500 , respectively , and compared this spectrum with the nuclear one .the best reproduction of the nuclear spectrum was obtained when , prior to the combination with the power - law component , we reddened the extranuclear spectrum by an additional e(b - v)=0.06. the result is shown in figure 2b .the essential difference between tran s results and ours is that , while he attributes the remaining 22% of the photons at 5500 to an extra continuum source ( fc2 ) , we account for the whole spectrum with either just the extranuclear stellar population template ( fig .2a ) or , more realistically , with a combination of this template plus the reflected nuclear spectrum ( fig .we remark that the slit width used in our observations ( 2 ) , is similar to that used by tran ( 2.4 ) , and the same used by kay ( 1994 ) .the 2 length of our nuclear extraction window was smaller that the typical 57 used by kay , while tran does not give this information , but it is unlikely that it is smaller than ours .we thus rule out the possibility that we have found a larger galaxy contribution due to a larger nuclear window . instead , the different starlight fractions found by tran and us stem from the different stellar population templates used .note that tran s galaxy fractions are already larger than the ones obtained by miller & goodrich ( 1990 ) for the objects in common , including mrk348 and mrk1210 .our results show that an even larger stellar contribution is possible .figure 3a shows the comparison of the nuclear spectrum of mrk573 with the reddened [ e(b - v)=0.1 ] extranuclear spectrum . in this case, we find some mismatch of the continuum between 4000 and 5000 , the reddened extranuclear spectrum having @xmath1410% less flux than the nuclear spectrum in this spectral region .mrk573 has a polarized continuum observed by kay ( 1994 ) . allowing for a few percent contribution of such component ( represented by a f@xmath15 power - law , in analogy with mrk348 ) , we get an even worse fit in the @xmath164000 - 5000 region . we have then tried a different stellar population template , obtained from the nuclear spectrum of the elliptical galaxy ic4889 , from our sample of paper i ( thus , observed with the same setup as mrk573 ) , combined with the same power - law component above .figure 3b shows that this template , combined with a 10% contribution of the power - law at @xmath45500 gives a better fit for the continuum , and at the same time reproduces the nuclear w values .this result is in approximate agreement with previous studies , like that of kay ( 1994 ) , who derived a stellar population contribution of 80% at @xmath44400 , while our results imply a stellar population contribution of 85% at this wavelength . but how to understand the lack of dilution of the w s in the nucleus of this galaxy ?one possibility is that the stellar population shows a small variation with distance from the nucleus , more evident in the shape of the continuum than in the absorption lines , and that the scattered ( power - law ) component is _ extended _ , keeping the w s approximately at the same values .an extended blue continuum , probably due to scattered light , has indeed been found in near - uv images of mrk573 by pogge & de robertis ( 1993 ) along the ionization cones observed in this galaxy .if this interpretation is correct , spatially resolved spectropolarimetry should reveal a seyfert 1 spectrum up to a few arcseconds from the nucleus along the cone axis .figure 4a shows the comparison of the nuclear spectrum of ngc1358 with the reddened [ e(b - v)=0.1 ] extranuclear spectrum .the two spectra present @xmath175% differences in the continuum and @xmath18% differences in the w s , suggesting no contribution by a featureless continuum .nevertheless , kay ( 1994 ) finds a continuum polarization of 1.65% ( after starlight subtraction ) , indicating the presence of a scattered component , albeit weak .she derives a stellar population contribution of 83% at @xmath19 , while our results indicate a larger contribution , of at least 90% at the ca ii k line .allowing for a contribution of up to 10% of a f@xmath20 component in the blue , at @xmath45500 this contribution amounts only to @xmath23% .the fit we obtain for the nuclear spectrum , combining 3% of this component with 97% of extranuclear spectrum at @xmath45500 is slightly improved , as illustrated in figure 4b . at @xmath44400the stellar population contribution amounts to @xmath295% .we note that the 83% starlight fraction found by kay yields an uncomfortably low polarization for ngc 1358 as compared to the typical values expected in the unified model ( miller & goodrich 1990 ) . only for fractions above 95% , and thus within the range of the acceptable stellar population contributions according to our method , would the polarization raise above 10% .5a shows a comparison of the nuclear spectrum of mrk1210 with the extranuclear one scaled and reddened by e(b - v)=0.07 .the latter is a good reproduction of the nuclear spectrum down to the ca ii k line , with differences in the continuum @xmath175% and @xmath1710% in the w s . but bellow 3900 , there is an excess continuum in the nuclear spectrum .this continuum could be due to the balmer recombination continuum of the emitting gas , which is significant in this galaxy because of its strong emission lines .the gaseous temperature , as derived from the [ oiii]@xmath44363/5007 emission - line ratio is about 20,000k . we have then used the corresponding h@xmath21 recombination coefficient and hi continuum emission coefficient from osterbrock ( 1989 ) , and the observed h@xmath21 flux ( 5.38@xmath610@xmath22 ergs @xmath23 s@xmath24 ) to derive the contribution of the balmer continuum to the nuclear spectrum .this contribution amounts to @xmath23% of the total flux at @xmath45500 , @xmath27% at @xmath43700 and 30% for @xmath253646 .we note that a balmer continuum has also been detected in mrk477 ( heckman et .1997 , hereafter h97 ) , a galaxy in many respects similar to mrk1210 ( see below ) .mrk1210 was also observed by tran ( 1995a , b , c ) , who found a polarized continuum contribution of 6% at @xmath45500 . as for mrk348, this value is derived by tran as being the one necessary to make the polarization in the continuum to agree with that of the broad lines .tran finds a 75% stellar contribution , while the remaining 19% `` non - stellar '' continuum is attributed to fc2 .the spectral distribution of the scattered continuum is derived to be f@xmath26 .we have then combined this power - law contribution normalized to 6% at @xmath45500 , the balmer continuum described above ( 3% at @xmath45500 ) and the extranuclear spectrum normalized to 91% of the nuclear flux at @xmath45500 after applying an additional reddening of e(b - v)=0.08 .the result is the spectral distribution illustrated in fig .5b , which is an improved fit to the nuclear spectrum .again , the different stellar fractions obtained by us and previous works result from the different stellar population templates . as for mrk348, what tran ( 1995c ) calls fc2 is somehow included in our stellar population template. a close look at the nuclear emission - line spectrum shows that this galaxy seems to present a `` wolf - rayet feature '' , which can be observed in fig .we point out the similarity of the whole complex to that found in mrk477 by h97 .we identify 5 narrow emission - lines in this complex : [ feiii]@xmath44658 , heii@xmath44686 , [ ariv]@xmath164711,4740 and [ neiv]@xmath44725 , to which we have fitted narrow gaussians , constraining their fwhm to the average value of the other strong emission lines in the spectrum ( 720 km s@xmath24 ) .the subtraction of these lines leaves the residual shown in figure 6b , centered around @xmath44670 , with flux 1.22@xmath27 erg @xmath23 s@xmath24 ( and luminosity @xmath28 erg s@xmath24 ) . following the methodology of h97we have compared our spectrum with those from wr stars ( smith , shara & moffat 1996 ; conti & massey 1989 ) , in order to estimate the type of wr stars producing the emission lines .we reach a similar conclusion : that the residual of fig. 6b can be identified as broad emission from heii@xmath44686 and niii@xmath44640 , and the wr stars in mrk1210 can be classified as of subtype wn6 - 7 or later . in a starburst , the ratio of the flux of the wolf - rayet complex to the h@xmath21 flux gives a measure of the relative importance of wr and o stars ( h97 and references therein ) .in mrk1210 , this value is @xmath220% , which is a lower limit , as the h@xmath21 flux includes contribution from ionization by the agn .this value is twice the value found for mrk477 ( h97 ) before correcting for the contribution from the agn , and indicates the presence at the nucleus of mrk1210 of a stellar burst with an age between 3 and 8 myr ( h97 , meynet 1995 ) .since we have concluded that the difference between the starlight fractions obtained here with respect to those in previous works ( particularly for mrk 348 and mrk 1210 ) is due to the different stellar population templates , it is natural to ask what do our templates have that ordinary elliptical galaxies do nt ? in order to answer this question , we have performed a population synthesis analysis using the base of stellar clusters of bica ( 1988 ) and the method described in schmitt , bica & pastoriza ( 1996 ) . as input data we have used the w s of ca ii k , cn , g - band and mg i + mgh absorption features , as well as the continuum fluxes at 3660 , 4020 , 4570 and 6630 , all normalized to the flux at 5870 .the results of the synthesis are summarized in table 2 .the stellar population of mrk573 is represented by the elliptical template from ic4889 , in the last line of the table , while for the other galaxies the results correspond to the extranuclear templates discussed in 3 .the columns of the table show the percentage contribution of the age bins corresponding to 10 gyr ( globular clusters ) , 1 gyr , 100 myr , 10 myr and hii regions , respectively , to the spectrum at @xmath45870 . we also list , between parentheses , the corresponding fractions at @xmath44510 . inspection of the table shows that in ngc1358 , as in the case of the elliptical galaxy template , the population is composed almost only by the 10 gyr and 1 gyr age bins . in mrk348 and mrk1210 , however , there are additional contributions from the younger age bins ( @xmath8100 myr ) , amounting to 10% and 23% , respectively . since these populations are blue , the corresponding fractions increase for shorter wavelengths , reaching 14% and 33% at 4510 respectively ( table 2 ) .we thus conclude that the difference between our extranuclear templates and the more commonly used elliptical galaxy spectrum is the presence of a substantial contribution from a 0100 myr old population .a combination of a red bulge , power - law continuum , young and intermediate age components has been also found in previous works on the analysis of the optical continuum of active galaxies .tadhunter , dickinson & shaw ( 1996 ) , for example , have concluded that an intermediate age component was required in order to reproduce the optical spectrum of the nearby radio galaxy 3c321 , in addition to a power - law and an old bulge component . using a similar method as used here , storchi - bergmann et al .( 1997b ) have also found a mixture of old bulge , power - law and `` aging starburst '' components in the continuum spectrum of the inner region of cen a. cid fernandes & terlevich ( 1992 ) have also suggested that the contribution from young and intermediate age stars in seyfert 2s may have been sistematically underestimated due to a `` template mismatch '' effect .such a mismatch results from the known difficulty in determining the contribution of early type stars to the optical spectrum of agn ( see discussions in miller & goodrich 1990 and kay 1994 ) .further studies suggesting the presence of young to intermediate age stars in active galaxies are the ones about the strength of the ca ii triplet around 8500 ( terlevich , diaz & terlevich 1990 ) , and the infrared work of oliva et al .had we used an elliptical galaxy as a template for mrk348 and mrk1210 , we would inevitably obtain larger featureless continuum ( fc ) contributions .experiments using the nuclear spectrum of the elliptical galaxy ic 4889 confirm this ( see figures 7 and 8) . in order to match the nuclear spectrum of mrk 348 , a combination of 85% contribution of the elliptical template ic4889 and a 15% @xmath11 power - law contribution at @xmath45500 would be required . for mrk 1210, the proportions would be 69% of starlight and 28% of a @xmath29 power - law ( plus 3% contribution of the balmer continuum ) , in better agreement with tran s ( 1995a ) results .it is important to emphasize that the fits obtained with an elliptical galaxy are as good as those obtained with the extranuclear templates ! however , these are by no means equivalent choices , since the differences in the starlight fractions imply completely different scenarios for the nature of the residual non - stellar continuum . which leaves us with the question : how is one to decide which template to use ?whilst it is not possible to decide _ a priori _ which template suits best the true stellar population in the nucleus , we believe that a spectrum extracted from the bulge of the same galaxy , should be , in most cases , a better representation of the nuclear stellar population than a spectrum from an external galaxy , whose star formation history may differ substantially from that of the target galaxy .furthermore , the much higher fc fractions obtained above with the elliptical galaxy template would imply w and color radial gradients , if the fc is concentrated in or close to the nucleus , which should be detectable with our method , whereas we find none ( paper i ) .for example , fig .1 shows that the w(ca ii k ) of ngc6814 decreases by @xmath240% at the nucleus due to the fc . in the case of mrk348 , for the 15% fc contribution at @xmath45500derived above with the nuclear stellar population represented by the elliptical galaxy template , the fc contribution at the ca ii k line is 33% , while for mrk1210 the 28% contribution at @xmath45500 corresponds to 45% at the ca ii k line .such dilutions are comparable to that observed in ngc6814 and should thus be readily observed . the only possibility to have an elliptical galaxy template plus a fc at the nucleus and no radial dilution of the absorption lines is if the fc is _ extended _ and the fc to galaxy ratio remains constant throughout the inner regions . in this caseour extranuclear spectra would not correspond to the pure stellar population component , since they would be contaminated by the fc ( be it fc1 , fc2 or both ) . testingthis possibility will require further observational constraints .direct imaging in the uv could in principle reveal an extended continuum , as was observed in mrk573 ( pogge & de robertis 1993 ) .polarization maps would allow to test the presence of extended scattered light , as well as long - slit spectra obtained at several position angles .regardeless of its extension , the fc in mrk348 and mrk1210 is clearly dominated not by scattered light , but by fc2 .tran ( 1995c ) finds that 83% of the fc is due to fc2 in mrk348 and 76% in mrk1210 .the difference between the extranuclear templates of mrk348 and 1210 and an elliptical galaxy should thus be essentially due to fc2 . according to the synthesis results , this difference is due to young and intermediate age stars , which we therefore associate with fc2 .a conclusive detection of features from such stars would corroborate this idea .the wr feature in mrk1210 confirms the presence of young stars .uv spectroscopy could provide further evidence of this through the detection of siiv and civ absorptions , as found in mrk477 ( h97 ) . in the case of mrk348the evidence for young to intermediate age stars relies not on a direct detection , but on our population synthesis analysis .uv spectroscopy of this object will be crucial to test this result .uv spectropolarimetry would also be useful to test the contribution of young to intermediate age stars .while the older stellar components should contribute little to the uv emission , the young stars should extend into the uv range along with the scattered agn component .since the stellar flux is presumably unpolarized , we expect that mrk1210 ( and possibly mrk348 ) presents broad uv lines ( e.g. civ@xmath41550 and mgii@xmath42800 ) with a _ larger polarization _ than the adjacent continuum .in other words , the `` dilution by fc2 effect '' found by tran ( 1995c ) should extend into the uv .we have investigated the nature of the optical continuum and stellar population in the nuclear region of the seyfert 2 galaxies mrk348 , mrk573 , ngc1358 and mrk1210 using long - slit spectroscopic data . in a previous work ( paper i ) we have found that none of these galaxies present dilutions in the nuclear equivalent widths w s of the absorption lines as compared with those of the extranuclear spectra from the bulge of the same galaxy , as it would be expected if a featureless continuum was present at the nuclei in the proportions found in previous works ( 20%50% ) .we have tested one possibility to account for this lack of dilution : that the nuclear stellar population is the same as the extranuclear one from the bulge , and the featureless continuum contribution to the nuclear spectra is smaller than 10% ( and thus not producing enough dilution in the w s to be detectable with our method ) .we conclude that this hypothesis works for mrk348 , ngc1358 and mrk1210 .a larger stellar population contribution can solve several puzzles of seyfert 2 galaxies : ( 1 ) it raises the level of intrinsic polarization , bringing it closer to the values expected in the unified model ( miller & goodrich 1990 ) ; ( 2 ) it avoids the problem of overpredicting the strengths of broad lines in the direct spectra ( cid fernandes & terlevich 1992 , 1995 , heckman et al .1995 ) ; and ( 3 ) it allows a compatibilization of the polarizations measured for the broad lines and continua .this last point is implicit in our results , since we have shown that , for mrk 348 and mrk 1210 , the nuclear spectrum can be adequately fit with a stellar component plus the scattered continuum at the strengths inferred from spectropolarimetry ( tran 1995a , b , c ) . since these are exactly the problems which lead to the idea of a second featureless continuum component ( miller 1994 ) , our results suggest that fc2 is the result of an underestimation of the stellar contribution .our population synthesis analysis indicates that the larger stellar population contribution found for mrk348 and mrk1210 relative to that found in previous works is due to the presence of young and intermediate age stars ( @xmath30 myr ) , not accounted for by the more commonly used elliptical galaxy templates . in other words , the fc2 would be due to blue stars in these two cases . for mrk1210, the apparent detection of a `` wolf - rayet '' feature confirms this result .these findings , together with results from the recent study by h97 on mrk477 , suggest that young populations may be common in markarian seyferts , being responsible for at least part of the blue continuum in these galaxies ( which were discovered precisely due to their blue continuum ) .the above results suggest that star - formation is an important piece in the agn puzzle .we speculate that , if there is a causal link between star - formation and activity , the @xmath31 myr star formation time scale identified in this paper should be comparable to the duration of the nuclear activity cycle . since 100 myr is about 1% of the hubble time , a causal link between star formation and activity would imply that about 1% of the galaxies should be active .interestingly , this value seems to agree with current estimates ( e.g. huchra & burg 1992 ) .population synthesis for a complete sample of nearby seyferts could help putting tighter constraints on this link .we have also tested an elliptical galaxy template , as used in previous works . for mrk573, this template gives a better match to the nuclear stellar population than the extranuclear spectrum .the lack of dilution in this case could be due to the presence of an extended scattered continuum ( observed in uv images by pogge & de robertis , 1993 ) .mrk348 and mrk1210 also present an excess blue continuum when compared with an elliptical template .according to our w s measurements , this continuum should be extended , but according to the spectropolarimetry by tran ( 1995c ) , only a small part of it is polarized , so it is unlikely that it is dominated by scattered light .this result is thus consistent with our finding of an excess blue population in these galaxies relative to an elliptical galaxy . as a final remark, we note that , as all current fc2 indicators are critically dependent on the starlight correction procedure , unveiling the nature of fc2 will necessarily require a carefull examination of the stellar population in the central regions of agn . with this goal , in a forthcoming paper ( schmitt et al .1998 ) , we investigate the stellar content of a sample of 42 active and normal galaxies , via population synthesis as a function of distance from the nuclei , using the extensive data of paper i. we acknowledge fruitful discussions with laura kay , andrew s. wilson and roberto terlevich , and valuable suggestions by an anonymous referee .this research received partial support from the brazilian institutions cnpq , finep , fapergs and fapeu , and has made use of the nasa / ipac extragalactic database ( ned ) , which is operated by the jet propulsion laboratory , caltech , under contract with nasa .lccccc mrk348 & 45 ( 36 ) & 45 ( 50 ) & 5 ( 8) & 4 ( 5 ) & 1 ( 1)ngc1358 & 64 ( 55 ) & 34 ( 42 ) & 2 ( 3 ) & 0 ( 0 ) & 0 ( 0)mrk1210 & 53 ( 42 ) & 24 ( 25 ) & 21 ( 31 ) & 2 ( 2 ) & 0 ( 0)ic4889 & 79 ( 72 ) & 21 ( 28 ) & 0 ( 0 ) & 0 ( 0 ) & 0 ( 0 )

<S> we investigate the nature of the optical continuum and stellar population in the central kiloparsec of the seyfert 2 galaxies mrk348 , mrk573 , ngc1358 and mrk1210 using high signal - to - noise ratio long - slit spectra obtained along the radio axis or along the extended high excitation emission . </S><S> these four galaxies are known to have polarized continuum including polarized broad lines in mrk348 and mrk1210and previous studies indicate featureless continua contributions in the 2050% range at @xmath0 5500 . nevertheless , our measurements of the equivalent widths of absorption lines and continuum ratios as a function of distance from the nuclei show no decrease of the equivalent widths ( i.e. , no dilution ) nor bluening of the spectrum towards the nucleus , as expected if a blue featureless continuum was present at the nucleus in the above proportions . </S><S> we investigate one possibility to account for the lack of dilution : that the stellar population at the nucleus is the same as that from the surrounding bulge and dominates the nuclear light . by comparing the nuclear and the extranuclear spectra of each galaxy , </S><S> we conclude that this hypothesis works for mrk348 , ngc1358 and mrk1210 , for which we find stellar population contributions at the nucleus larger than 90% at all wavelengths . </S><S> our approach differs from that adopted in previous studies , where an elliptical galaxy template is used to represent the stellar population of the nucleus . </S><S> although the latter may be valid for some galaxies as , for example , mrk573in several cases the stellar population may be different from that of an elliptical galaxy . </S><S> we find that a larger stellar population contribution to the nuclear spectra can play the role of the `` second featureless continuum '' source ( fc2 ) inferred from previous studies . </S><S> in particular , stellar population synthesis shows that the nuclear regions of mrk348 and mrk1210 have important contributions of young to intermediate age stars ( 0@xmath1 ) , not present in templates of elliptical galaxies . in the case of mrk1210 , </S><S> this is further confirmed by the detection of a `` wolf - rayet feature '' in the nuclear emission - line spectrum . </S>

1207.3341

organic solar cells ( osc ) hold the potential to become a low - cost photovoltaic technology , produced in a roll - to - roll process and thus deserve serious consideration.@xcite therefore , the efficiency of organic solar cells still has to be improved and a proper understanding of the elementary physical processes is required .charge carrier recombination is one of the particularly crucial processes as it limits the electrical performance of osc .indeed , the polaron recombination dynamics fed into the continuity equation have been shown to deliver sufficient information to reproduce the complete j / v response of p3ht : pcbm solar cells at room temperature.@xcite recently , maurano _ etal ._ @xcite successfully predicted the open circuit voltage of osc based on four different polymer : fullerene compositions by studying nongeminate recombination .thereby , authors combined parameters obtained by static ( j / v ) and transient photovoltage / transient photocurrent ( tpv / tpc ) methods in the shockley equation.@xcite in order to justify the validity of this approach and/or its limitations , we i ) directly compared the ideality factor and dark saturation current density derived by the static and transient measurements and ii ) reproduced the photocurrent for p3ht : pcbm osc over the temperature range from 200 to 300 k at various light intensities . we demonstrate that identical shockley parameters can indeed be determined from transient and static measurements making se applicable to osc , if the charge carrier generation is voltage independent .deviations found at lower temperatures are attributed to voltage dependent charge carrier photogeneration due to the lack of thermal activation energy . extending the work of kirchartz _etal._,@xcite we analyzed the impact of charge carrier recombination on the ideality factors in detail , and derived the dominant nongeminate loss mechanism .the established model to describe the current voltage response of semiconductor p n junctions under illumination is based on the ideal se , @xmath2 with dark saturation current @xmath1 , elementary charge @xmath3 , ideality factor @xmath4 , thermal energy @xmath5 and the photogenerated current @xmath6.@xcite conventional p n junctions are characterized by well developed energy - bands where photogeneration yields delocalized charge carriers . in contrast , osc are based on excitonic materials , in which the photogeneration of quasi free polarons upon photon absorption is via bound precursor states and strongly depends on the active material .furthermore , bhj devices consist of donor and acceptor material phases across the whole volume , resulting in many spatially distributed heterojunctions instead of a single planar one . despite these differences to inorganic p n junctions , the se was successfully applied to organic solar cells based on different donor acceptor systems in the past.@xcite their application to devices based on mdmo - ppv : pcbm@xcite failed due to voltage dependent photocurrent generation @xmath7 . for osc based on p3ht : pcbm , used in this work , the polaron pair dissociation is reported to be independent@xcite or weakly dependent@xcite on voltage , providing a good starting point for our analysis .for simplicity we assumed an ideal diode in this study , neglecting any series resistance ( @xmath8 ) or leakages by shunts ( @xmath9 ) .no significant influence on the data was found by taking a series resistance into consideration , which is experimentally determined from the ohmic range of dark j / v - characteristics .( [ eqn : j_shockley ] ) solved for @xmath10 at a given light intensity with @xmath11 and for @xmath12 results in @xmath13 we note that in order to experimentally access the photogenerated charges by j / v measurements , we assumed a voltage independent generation current @xmath6 .bulk heterojunction ( bhj ) solar cells were prepared by spin coating a 35 nm thick layer of poly(3,4-ethylendioxythiophene):polystyrolsulfonate ( clevios p vp ai 4083 ) on indium tin oxide samples with post - annealing step of 130@xmath14c for 10 minutes in a water - free environment .the p3ht : pcbm 1:0.8 blend made from solutions of 25 mg / ml in chlorobenzene was spin coated in an inert atmosphere resulting in 100120 nm thick active layers .after annealing for 10 min at 130@xmath14c ca ( 3 nm)/al ( 100 nm ) were evaporated thermally on top of the organic layer .p3ht was purchased from rieke metals ( p200 , @xmath15 98 % regioregular ) , pcbm ( purity 99.5 % ) from solenne .all materials were used without further purification . prior to any additional measurements an oriel 1160 am1.5g solar simulator was used to perform illuminated j / v measurements of devices kept under inert glove - box atmosphere .the bhj solar cells showed a power conversion efficiency of approximately 3.3 % .the samples were transferred into a closed cycle optical cryostat for temperature dependent static and transient electrical studies .tpv measurements monitor the photovoltage decay upon a small optical perturbation during various constant bias light conditions .the voltage transient enable to determine the small perturbation charge carrier lifetime @xmath16 in dependence of the respective open circuit voltage due to the constant background illumination .tpc is measured with identical optical perturbation , but under short circuit conditions .further experimental details can be found in ref . , .tpv / tpc measurements were performed using a 10 w white light led ( seoul ) and focusing optics for bias light illumination and an attenuated nd : yag laser flash ( @xmath17 nm , @xmath18 80 ps pulse duration ) to generate an additional small amount of charges within the device .the voltage transient was probed using a 1.5 g@xmath19 impedance adapter to ensure open circuit conditions .both , voltage and current transients were acquired by a digital storage oscilloscope ( agilent infinium dso90254a ) .was varied by changing the illumination intensity . ]the results of temperature dependent tpv / tpc measurements are depicted in fig .[ fig : tau_n_voc ] .as can be seen , charge carrier density @xmath20 ( fig .[ fig : tau_n_voc ] ( a ) ) and small perturbation lifetime @xmath16 ( fig . [ fig : tau_n_voc ] ( b ) ) both exponentially depend on @xmath10 and can be described by @xmath21 with @xmath22 defined as the ideality factor of charge carrier density .and similarly , @xmath23 with the ideality factor of charge carrier lifetime @xmath24 .we note that the slopes in the semi - logarithmic representation of fig .[ fig : tau_n_voc ] were previously described by parameters @xmath25 and @xmath26,@xcite which are related to our dimensionless ideality factors @xmath22 and @xmath24 by @xmath27 as seen in fig .[ fig : tau_n_voc ] both slopes ( @xmath28 ) increase when the temperature is lowered . to describe the experimentally found polaron dynamics in organic photovoltaic devicesa generalized equation @xmath29 is often used.@xcite from tpv / tpc analysis , the decay order of the recombination rate @xmath30 , @xmath31 , and the small perturbation charge carrier lifetime @xmath16 can be experimentally determined . using these values the total charge carrier lifetime @xmath32 can be calculated , as it was shown in ref . ,@xmath33 where @xmath34 .using eqs .( [ eqn : taun ] ) and ( [ eqn : lambda ] ) , small perturbation lifetimes values can be eliminated , @xmath35 dividing eq .( [ eqn : n ] ) by eq .( [ eqn : tau ] ) , we obtain : @xmath36 where @xmath37 is defined as the recombination rate under illumination , depending on illumination intensity and @xmath38 the recombination rate in dark . solving eq .( [ eqn : ntau ] ) for @xmath10 yields a more generalized expression , @xmath39 with the recombination ideality factor @xmath40 defined as @xmath41 which is consistent with earlier representations of the ideality factor.@xcite comparing eq .( [ eqn : voc_tpv ] ) , derived by using relations empirically found in transient measurements ( eqs .( [ eqn : n]),([eqn : taun ] ) ) , and the right hand side of eq .( [ eqn : voc_shockley ] ) , the se solved for @xmath10 to describe steady - state j / v - measurements , the same equation structure becomes apparent . to compare both equations , firstthe equality of recombination rates and respective currents is motivated . at @xmath10 generation and recombination rates are equal , i.e. @xmath42 , which implies the generation current @xmath6 ( see eq .( [ eqn : voc_shockley ] ) ) is cancelled by the recombination current @xmath43 , defined as @xmath44 with the thickness of the active layer @xmath45 .likewise , the dark saturation current @xmath1 may be treated as thermally generated intrinsic charge carriers with the density @xmath46 and lifetime @xmath47 canceled by recombination in thermal equilibrium .hence , the respective loss current in the dark @xmath48 can be written as @xmath49 the above considerations provide a rationale to the interconnection between the se parameters derived diversely . fig .[ fig : comp ] ( a ) shows ideality factors @xmath4 and @xmath40 , as defined above , which were experimentally obtained as function of temperature by means of static ( j / v ) and transient ( tpv / tpc ) methods , respectively . to determine the static se parameters j / v characteristics were recorded for different bias lights .then , @xmath10 was plotted vs. @xmath6 and analyzed by eq .( [ eqn : voc_shockley ] ) to yield the ideality factor @xmath4 . within the measured temperature range @xmath40 and @xmath4correspond well to each other .similary , the dark saturation current @xmath1 and the dark recombination current @xmath48 , depicted in fig .[ fig : comp ] ( b ) , are almost equal for different temperatures .the ideality factor rises from 1.2 at room temperature to 1.5 at 170 k. a similar trend with temperature was observed in the past for organic solar cells based on different materials , although a stronger temperature dependence was obtained.@xcite and from illuminated j / v data @xmath4 , analyzed with the respective shockley equation ( see eq .( [ eqn : voc_shockley ] ) ) .( b ) respective dark saturation current densities in dependence on temperature .@xmath48 was calculated from tpv / tpc studies on the recombination dynamics in the dark with eq .( [ eqn : jloss0 ] ) and @xmath50 from illuminated j / v response at @xmath10 via eq .( [ eqn : voc_shockley ] ) . ] from the experimental observation shown in fig .[ fig : comp ] , the se equivalence for p3ht : pcbm based osc to eq .( [ eqn : voc_tpv ] ) , derived from studying charge carrier recombination by transient techniques , can be suggested .we first focus on the strong temperature dependence of the dark saturation current ( fig .[ fig : comp ] ( b ) ) . in fig .[ fig : j0_invnt ] , the dark saturation current @xmath48 determined by tpv / tpc is plotted versus @xmath52 , with the temperature dependent ideality factor @xmath53 from fig .[ fig : comp ] ( a ) .the linear fit of the semi - logarithmic plot is in excellent agreement with the measured data and is expressed by the relation @xmath54 the fit demonstrates that the dark saturation current follows the boltzmann distribution and allows to identify the effective band gap energy @xmath51 , which is proportional to the difference between homo and lumo energy levels of donor and acceptor , respectively .we found @xmath55ev for p3ht : pcbm . from the @xmath56 data ( not shown ) , a value of @xmath57 ev was evaluated according to ref . .vandewal et .al obtained the value of @xmath58 ev from the energy onset of the photocurrent generated by ct absorption,@xcite which agrees particularly well with our value .guan et al . found somewhat larger values ( @xmath59ev ) using ultraviolet and inverse photoemission spectroscopy.@xcite as discussed in ref ., energy values derived by photoemission spectroscopy can be seen as onsets of respective valence and conduction band . due to charge carrier relaxation into the tail states of the density of states , the techniques such as tpv / tpc and j / v - measurements are expected to reveal the effective bandgap @xmath51 from the relaxed density of occupied states instead . in order to estimate the effective band gap energy @xmath51 from eq .( [ eqn : j_0 ] ) . ] to complete the comparison of eq .( [ eqn : voc_shockley ] ) and eq .( [ eqn : voc_tpv ] ) we consider the photogenerated charge carrier density , being equal to the density of charges recombining at @xmath10 . in order to estimate the photogenerated charge carriers, we used the saturated current density @xmath60 under illumination in reverse bias ( @xmath61 ) .if the generation of charge carriers is voltage independent for p3ht : pcbm cells in the range of @xmath62@xcite and @xmath60 is not reduced by nongeminate recombination , we expect @xmath63 for comparison , eq .( [ eqn : jloss ] ) was used to calculate the recombination current @xmath43 under illumination at open circuit .the saturation current @xmath60 and the nongeminate recombination current @xmath43 at @xmath10 determined via tpv / tpc measurements for different illumination intensities from 0.1 sun to about 1.8 suns are shown in fig .[ fig : jloss_t ] . within experimental error, @xmath43 is in a very good agreement with @xmath60 for all light intensities and temperatures above 200 k. below this temperature , @xmath60 becomes smaller than @xmath43 , which seems to imply more charges recombining than being generated .this counterintuitive result can at least partly be explained by estimating the nongeminate recombination losses at @xmath61 according to ref . .we find 2% loss at room temperature and 5% at 170 k , revealing the limitations of the analysis as the generation current is underestimated .accounting for the nongeminate losses , a very small deviation of @xmath60 and @xmath43 remains , showing that the field dependence of the photogeneration is less than 10% between open and short circuit conditions even at low temperatures . in accordance with earlier findings,@xcite the photocurrent in p3ht :pcbm solar cells is at best slightly field dependent in this voltage range .( solid lines ) from j / v response compared to recombination current @xmath43 ( dashed lines ) measured by tpv / tpc for different temperatures and bias lights according to eq .( [ eqn : jloss_comp ] ) . for @xmath64200k a stronger deviation of both parameters becomes apparent , indicated by the dashed vertical line . ] , and lifetime , @xmath65 , as well as their sum , resulting in the inverse ideality factor @xmath66 .( b ) corresponding recombination parameter @xmath67 ( left ) and recombination order @xmath68 ( right ) . ] in eq .( [ eqn : voc_tpv ] ) the recombination ideality factor @xmath40 was defined . as mentioned above @xmath40is composed by the ideality factor for charge carrier density @xmath22 and charge carrier lifetime @xmath24 , respectively ( see eq .( [ eqn : n_r ] ) ) . in fig .[ fig : nr ] ( a ) , the inverse ideality factors derived from the transient experiments are plotted versus temperature . at 300 k , we found @xmath69 and @xmath70 , yielding a recombination ideality factor of @xmath71 .as shown already in fig .[ fig : comp ] ( a ) , the ideality factor from the j / v - characteristics under illumination was @xmath72 .although we focussed on the ideality factors measured under illumination , we state for comparison that the dark ideality factor was 1.30 , whereas kirchartz _etal._@xcite found a value of 1.6 .the recombination order , given by @xmath73 as pointed out above , is shown in fig .[ fig : nr ] ( b ) .the recombination order at 300 k of @xmath74 was slightly higher as compared to the values of 2.6 ( ref . ) or 2.75 ( ref . ) .based on the very concise approach offered by kirchartz _etal._,@xcite we used the experimentally determined ideality factors to consider the dominant recombination mechanism in the p3ht : pcbm solar cells under open circuit conditions .ideality factors higher than 1 in organic bulk heterojunction solar cells are commonly seen as an evidence for a dominant trap assisted recombination process ._ proposed the recombination of free charge carriers @xmath75 with trapped charge carriers @xmath76 in an exponential tail as dominant nongeminate loss mechanism .the corresponding recombination rate is @xmath77 , assuming symmetric electron and hole concentrations .using the relation , @xmath78,@xcite of free @xmath75 to trapped @xmath76 charge carriers they were able to calculate the characteristic ( urbach ) tail energy @xmath79 from experimental data@xcite by three routes : ( i ) by the voltage dependence of the extracted charge carrier concentration , which rewritten in our notation is given by the ideality factor @xmath22 , @xmath80 using their experimental @xmath81 .( ii ) from the ( in their case dark ) ideality factor , based on the relation by van berkel _etal._,@xcite @xmath82 which corresponds to our eq .( [ eqn : n_r ] ) , assuming @xmath83 and replacing @xmath22 by using eq .( [ eqn : eu_nn ] ) .( iii ) from the recombination order @xmath84 with an experimentally determined recombination order of 2.6.@xcite from ( i ) kirchartz _ etal ._ found @xmath85 mev , from ( ii ) @xmath86 mev at room temperature , whereas from ( iii ) @xmath87 mev can be calculated .this discrepancy was discussed in some detail in their publication.@xcite we considered a more general recombination rate , including the annihilation of free charge carriers with one another and free with trapped charge carriers , @xmath88 here , the overall charge carrier concentration is given as @xmath89 . for simplicity , we used the same recombination prefactor @xmath90 for both contributions @xmath91 and @xmath92 .usually the rate limiting step in nongeminate recombination is the finding of the actually localized charge carriers by a hopping process , which is reflected by a prefactor proportional to the mobility of the mobile charge carriers , based on langevin theory.@xcite however , although beyond the scope of this article , we point out that in the multiple - trapping - and - release approximation of hopping transport , the effective mobility @xmath93 is proportional to a trap - free mobility @xmath94 times the fraction of free to all charge carriers , @xmath95 . within this approach , @xmath96 and @xmath97are in principle equivalent .we ask the reader to bear in mind that this representation can serve as a first approximation , but neglects the field dependence in a hopping model . for our purpose here , however , eq . ( [ eqn : r ] ) is sufficient to describe the recombination process in terms of the ideality factors discussed above .the small - signal method tpv yields the effective lifetime of all charge carriers , @xmath32 . for charge extraction, we assumed that all charge carriers were extracted , in accordance with kirchartz _etal._@xcite thus , our experimentally determined effective recombination rate is @xmath98 ( c.f .( [ eqn : ntau ] ) ) . comparing our assumption for the recombination rate , eq .( [ eqn : r ] ) to our effective recombination rate , we find @xmath99 in order to test our assumption , we considered the voltage dependence of the constituents @xmath20 and @xmath32 of the effective recombination rate by investigating their ideality factors @xmath22 and @xmath100 , respectively . using the exponential tail model,@xcitewe also followed the three different routes outlined above . calculating the characteristic tail energy ( i )directly from @xmath22 , we determined @xmath101 mev at room temperature . by route ( ii ) , using our eq .( [ eqn : n_r ] ) instead of eq .( [ eqn : vanberkel ] ) , we again found 45 mev with either @xmath40 or ( illuminated ) @xmath4 .in contrast to kirchartz _etal._,@xcite , our result from ( ii ) is consistent with ( i ) .we point out that the authors had a rather high ideality factor of 1.6 which was determined under dark conditions , in contrast to our consistent values of @xmath102 .when using our dark ideality factor of 1.30 , we find 48 mev for the characteristic tail energy . deriving the tail energy from the recombination order , route ( iii ) , @xmath103 using eq .( [ eqn : eu_nn ] ) , yielded again a result compatible with the other two approaches , as essentially always the same equations are used for all routes within our consistent framework .in contrast to eq .( [ eqn : vanberkel ] ) , as used in ref ., deviations of @xmath100 from the value of 2 are considered .we point out that due to our focus on the interpretation of ideality factors here , which is only qualitative in view of the recombination rates , we are not able to make statements concerning the contributions from delayed recombination.@xcite for understanding the consistency of the charge carrier concentration and voltage dependence of the loss current in view of recombination order and ideality factor , see appendix [ appendix_jloss ] . the increase of @xmath22 with lower temperatures ( fig .[ fig : nr ] ( a ) ) also indicates a thermally activated process concerning the charge carrier concentration : the lack of thermal energy leading to a growing fraction of trapped charge carriers is compatible with the assumption of exponential tail states . calculating the characteristic urbach energy from the temperature dependent @xmath22 by using eq .( [ eqn : eu_nn ] ) , we found @xmath104 mev for 200 k , showing that an exponential tail may not be the precise shape of the density of trap states , but can serve as an approximation in a limited temperature range . within our assumption of the recombination rate , eq .( [ eqn : r ] ) , the effective lifetime is inversely proportional to the free charge carrier concentration @xmath75 ( eq . ( [ eqn : tau_for_r ] ) ) .therefore , we would expect that @xmath100 should equal 2 in accordance with @xmath22 for free charge carriers .however , this statement is only valid if the recombination prefactor @xmath90 is assumed to be voltage independent . while we can not determine the detailed reason for our experimental finding of @xmath69 , we point out that the assumption of @xmath105 may not hold true@xcite and that concentration gradients are disregarded . mev . ] in order to verify that the nongeminate recombination mechanism in annealed p3ht : pcbm solar cells is indeed due to losses of free carriers with one another and with carriers trapped in exponential tail states with @xmath85 mev , we reconsidered thermally stimulated current measurements presented previously.@xcite in this set of experiments , we did not account for trap states deeper than 400 mev .the distribution of the trap states is shown in a semilogarithmic plot ( fig .[ fig : exp - tail ] ) . neglecting the detailed shape, we approximated the energetic tail by an exponential urbach fit .our analysis yielded a characteristic energy @xmath106 mev , in accordance with a recent reconstruction of the density of trap states in p3ht : pcbm blends,@xcite and verifies our findings from above .the se parameters such as ideality factor @xmath4 and dark saturation current @xmath1 were derived either from static j / v - measurements or from the transient techniques ( tpv / tpc ) , were shown to coincide within experimental error for osc based on p3ht : pcbm .we directly compared the ideality factor and dark saturation current density experimentally and reproduced the saturation photocurrent for p3ht : pcbm osc over the temperature range from 200 to 300 k at various light intensities .recombination current determined under open circuit conditions is shown to be equal to the respective saturation current , which implies that the polaron pair dissociation is not significantly influenced by the electric field .a good agreement of static and transient approaches over a wide temperature range demonstrates the validity of the shockley model for osc , if the charge carrier photogeneration is voltage independent .additionally , the dark saturation current measured at different temperatures was used to determine the effective band gap of p3ht : pcbm blend to be in the range of @xmath107 ev , which is in good agreement to values from literature . considering the ideality factors, we found that nongeminate recombination of free with both free and trapped charge carriers in tail states is the dominant loss mechanism . using data from thermally stimulated current measurements, we verified that the charge carrier traps can indeed be approximated by an exponential trap distribution .the authors would like to thank t. kirchartz ( imperial college , london ) for helpful inspiration to parts of this work as well as d. rauh and j. lorrmann for discussions and reading the manuscript .a.f.s work was financed by the european commission in the framework of the dephotex project ( grant no .c.d . gratefully acknowledges the support of the bavarian academy of sciences and humanities .v.d.s work at the zae bayern is financed by the bavarian ministry of economic affairs , infrastructure , transport and technology .in eq . ( 16 ) of ref . , kirchartz _ etal ._ considered the carrier concentration dependent loss current at @xmath10 , @xmath108 and its connection to the voltage dependence of the carrier concentration , c.f .( [ eqn : n ] ) .

<S> the shockley equation ( se ) , originally derived to describe a p </S><S> n junction , was frequently used in the past to simulate current voltage ( j / v ) characteristics of organic solar cells ( osc ) . in order to gain a more detailed understanding of recombination losses </S><S> , we determined the se parameters , i.e. the ideality factor and the dark saturation current , from temperature dependent static j / v - measurements on poly(3-hexylthiophene-2,5-diyl)(p3ht):[6,6]-phenyl - c@xmath0 butyric acid methyl ester ( pcbm ) bulk heterojunction solar cells . as we show here , </S><S> these parameters are directly related to charge carrier recombination and become also accessible by transient photovoltage and photocurrent methods in the case of field - independent charge carrier generation . </S><S> although determined in very different ways , both se parameters were found to be identical . </S><S> the good agreement of static and transient approaches over a wide temperature range demonstrates the validity of the shockley model for osc based on material systems satisfying the requirement of field - independent polaron - pair dissociation . in particular , we were able to reproduce the photocurrent at various light intensities and temperatures from the respective nongeminate recombination rates . </S><S> furthermore , the temperature dependence of the dark saturation current @xmath1 allowed determining the effective band gap of the photoactive blend perfectly agreeing with the literature values of the energy onset of the photocurrent due to charge transfer absorption . </S><S> we also present a consistent model directly relating the ideality factor to recombination of free with trapped charge carriers in an exponential density of tail states . </S><S> we verify this finding by data from thermally stimulated current measurements . </S>

1310.6912

the strong interaction with photons as well as with electronic excitations makes excitons ideal particles for the processing of photonic information . within this approach, the information carried by photons is first converted to excitons , which are then manipulated by electric , optical , acoustic or magnetic fields and subsequently reconverted to photons for further information transmission @xcite .exciton manipulation profits from strong nonlinearity in excitonic systems mediated by their electric - dipole interactions , which can be explored for the realization of all - optical control devices @xcite .in addition , the interconversion between photons and excitons can be carried out coherently , i.e. , by transferring the phase information between the two systems .the latter becomes especially interesting when combined with macroscopically coherent phases of either exciton polaritons @xcite or indirect excitons ( ix ) @xcite , which provide a new platform for exciton - based information processing .yet , the coherent manipulation of these excitonic phases as well as their integration into optoelectronic circuits remains a great challenge .excitons are metastable with lifetimes normally in the _ ns_-range for conventional direct excitons ( dxs ) and in the _ps_-range for exciton polaritons .applications require particles with long lifetimes , which can be efficiently stored , manipulated and transported between different locations .such a requirement is met by indirect excitons ( ixs ) in bilayer semiconductor systems consisting of two quantum wells ( qws ) separated by a thin barrier , termed a double quantum well ( dqw ) . as illustrated in the inset of fig .[ fig1](a ) , the application of a transverse electric field @xmath0 forces the oppositely charged exciton constituents ( i.e. , electron and hole ) into different qws .the spatial charge separation enhances the ix radiative lifetime by orders of magnitude ( up to a @xmath1s - range ) as compared to dxs , while still maintaining the coulomb correlation between the electrons and holes .different studies have addressed the manipulation of ixs on time scales shorter than their lifetime .the trapping of ixs even down to the single particle level @xcite was demonstrated using a variety of potential landscapes created by optical @xcite , magnetic @xcite , electric @xcite or strain @xcite fields . applied electric fields , in particular ,can control the ix energy and lifetime via the quantum confined stark effect ( qcse ) induced by @xmath0 @xcite .proposals for all - optical control via remote dipolar interactions of ix fluids have also been put forward @xcite .recently , reference @xcite has introduced an exciton optoelectronic transistor , where the exciton flow ( over a few @xmath1 m ) induced by potential energy gradients has been exploited for the realization of simple operations on photonic signals , such as directional switching and merging @xcite .several groups have also demonstrated the transport of ixs over longer distances by diffusion @xcite , drift @xcite as well as by moving electrostatic lattices @xcite . finally , we have recently shown that ix can be efficiently transported over several hundreds of @xmath1 m by the moving ( and tunable ) strain modulation induced by a surface acoustic wave ( saw ) propagating on the dqw structure.@xcite previous reports have addressed the coupling of a reduced number of , in most cases , closely spaced ix systems .an interesting question is whether an approach can be envisaged to interconnect multiple remote ix sites , providing , therefore , a multi - functional and scalable technology for exciton - based optoelectronic circuits . in this paper, we introduce a novel concept for the scalable interconnection of remote ix systems based on a network of acoustic beams and electrostatic gates .the feasibility of the concept is demonstrated by an exciton acoustic multiplexer ( exam ) - an integrated multiport device for the storage and controlled transfer of ixs between any two nodes of a two - dimensional ( 2d ) array of ixs systems .the exam consists of a network of i / o ports interconnected by an array of transport channels driven by saws .each i / o port can be isolated from the array of transport channels by an exciton acoustic transistor ( exat ) . as a results, these ports can be used for storage of ixs as well as for their interconversion to photons .scalability , presently a main limitation for quantum computational systems , is ensured by the sublinear dependence of the exam dimensions on the number of interconnected nodes .the experimental transport dynamics between the nodes is well - reproduced by a theoretical model of self - interacting ix fluids in a dynamic potential landscape , which successfully demonstrates the ability to design , simulate and test complex ix devices .the exam concept is extensive and compatible with the well - established acoustic transport and manipulation of quantum excitations by saws @xcite . in the following section ( sec .ii ) , we describe the sample fabrication details as well as the optical techniques employed for the detection of ix transport and multiplexing .the experimental results are then presented in sec .[ results ] . here, we start with investigations of the dynamics of ixs under acoustic fields ( sec .[ acoustic_exciton_transport ] ) , which demonstrate that saw fields can capture ix packets and transport them over large distances .we then address the control of ix fluxes , which is carried out using exats ( sec .[ secexat ] ) . in sec .[ secexam ] , the flow control by exats is then combined with long - range acoustic transport to realize an ix multiplexer .finally , sec .[ discussions_and_conclusions ] discusses issues regarding the application of exams as a building block for integrated electro - optical devices based on ixs and summarizes the main conclusions of this work . to an interdigital transducer ( idt ) placed on a piezoelectric zno island .the saw amplitude is stated in terms of the nominal rf power p@xmath2 applied to the idt .the ix lifetime and energy are controlled by the bias voltage @xmath3 applied between the semitransparent top electrode ( ste ) and the doped substrate .the superimposed pl images compare the emission in the absence and presence of a saw .inset : energy band diagram of a dqw along the growth ( z ) direction showing the direct ( dx ) and indirect ( ix ) exciton transitions .( b ) ix transport by the moving strain modulation of the conduction ( cb ) and valence band ( vb ) in a dqw .e@xmath4 ( e@xmath5 ) denotes minimum ( maximum ) bandgaps induced by saw strain field along the x=[100 ] direction .( c ) spectral pl image of the transport channel at t=4.5 k. the superimposed spectrum was recorded at the ste edge .( d ) time - resolved pl trace recorded at t=3.8 k at the edge of the ste . ]the experiments were carried out on an ( al , ga)as field - effect heterostructure containing three sets of dqws grown on a @xmath6-doped gaas ( 001 ) wafer by molecular beam epitaxy .each dqw consists of two gaas qws with thickness of 14 and 17 nm ( qw@xmath7 and qw@xmath8 , respectively - cf .lower right inset in fig .1 ( a ) ) separated by a 4-nm thick al@xmath9ga@xmath10as tunnelling barrier .the dqws were positioned at a depth of 700 nm below the sample surface within an undoped 2.5-@xmath1m - thick gaas/(al , ga)as superlattice ( sl ) .the electric field @xmath0 is induced by a reverse bias @xmath3 applied across the schottky diode formed between a 10-nm - thick semitransparent titanium electrode ( ste ) on the sample surface and the n@xmath11-doped substrate , which serves as the common ground [ cf .[ fig1](a ) ] .the lower left inset shows the bending of the conduction ( cb ) and valence ( vb ) band edges along the growth ( z ) direction induced by @xmath0 .the excitons are locally excited underneath the ste area by a 780 nm pulsed diode laser ( average power density of approx .12 j/@xmath12 ) .the use of sls reduces the leakage photo - currents to values below 1 na ( for bias between -2 and 2 v and laser excitation power in the range up to a few tens of @xmath1w ) , thus ensuring that ixs are only formed by photoexcited carriers . the rayleigh saws ( with wavelength @xmath13=2.8 @xmath1 m corresponding to a saw frequency @xmath14 ghz at helium temperatures )were electrically excited along a @xmath15100@xmath16 surface direction by applying a radio - frequency ( rf ) signal to interdigital transducers ( idts ) deposited on piezoelectric zno islands .these saws do not carry a piezoelectric field outside the idt region , thus preventing field - induced exciton dissociation @xcite .their strain field creates , via the deformation potential interaction , the moving type - i modulation of the cb and vb band edges of the dqw @xcite displayed in fig .[ fig1](b ) . here , e@xmath4( e@xmath5 ) denotes the minimum ( maximum ) bandgap at the saw phases of largest tensile ( compressive ) strain .photoexcited ixs are captured close to the regions of e@xmath4 , which move with the acoustic velocity @xmath17 @xcite .the confinement of the electron and the hole at the same lateral ( i.e. , along the surface ) coordinate prevents exciton dissociation , thus making this approach qualitatively different from charge transport by the type - ii bandgap modulation induced by the piezoelectric saws .the depth of the dqw structures was selected in order to obtain a type - i potential with approximately equal modulation amplitudes for the conduction and valence bands.@xcite the transport of ixs by the saw strain field is tracked by imaging the photoluminescence ( pl ) emitted via exciton recombination along the saw transport path .the laser excitation was focused onto a 5 @xmath1 m ( 10 @xmath1 m ) spot on the top semitransparent electrode by a 10x ( 5x ) microscope objective .the light emitted along the saw transport channel was collected by the same objective , dispersed by a single grating spectrometer and imaged onto a cooled charge coupled device ( ccd ) camera with spatial resolution of 1 @xmath1 m ( 2 @xmath1 m ) .the two pl images superimposed on the device structure of fig . [ fig1](a ) compare the excitonic pl in the absence ( left plot ) and presence ( right plot ) of a saw . while in the former the emission is restricted to the region near the excitation spot , the latter shows substantial pl at the edge of the ste located approx .500 @xmath1 m away from the laser spot .the remote pl is attributed to the recombination of ixs transported by the saw at the edges of the ste , where the sudden reduction of the vertical field @xmath0 creates a potential barrier blocking the ix transport @xcite . as the ixsare pushed by the saw against this barrier their concentration increases while the lifetime reduces , thus resulting in a stronger pl . the long - range transport of ixs is confirmed by the spectral dependence of the pl along the transport channel displayed in fig .[ fig1](c ) .the pl around 1.53 ev is dominated by the recombination of spatially direct neutral ( dxs ) and charged ( dx@xmath18 ) exciton complexes associated with the two qws .this emission extends to distances of up to approximately 150 @xmath1 m from the generation point due to the expansion of hot excitons and free carriers created by the pulsed laser @xcite .in contrast , the remote emission near the ste edge only shows the red - shifted pl signal , which is attributed to the recombination of acoustically transported ix packets .the energy of this remote pl is determined ( via the qcse ) by the bias applied to the ste .the pl intensities integrated over the transport channel with and without a saw differ by less than 1% , demonstrating that the saw transports the excitons introducing negligible losses .moreover , by calculating the ratio between the integrated pl intensity for ixs at the excitation point in the absence of saw and at the recombination point for p@xmath2=10 dbm ( cf .[ fig1](c ) ) , one can estimate the exciton transport efficiency , which for a distance of 500 @xmath1 m is around 50% .the dynamics of the ix transport was probed by time - resolved pl measurements at the end of the saw transport channel , as illustrated in fig .[ fig1](d ) . from the delay of the pl pulse onsetwe calculate a maximum propagation velocity of [email protected] @xmath1m / ns , which is close to the saw group velocity of @[email protected] @xmath1m / ns . the width of the pl pulse is determined by the initial size of the ix cloud created by the pulsed laser excitation .we observe a slight broadening of the pl profile as well as the development of a tail with increasing transport distances .these features , which will be addressed in details elsewhere,@xcite are attributed to trapping sites along the transport path induced by potential fluctuations , which capture ix and delay their propagation .the transport within well - defined @xmath1m - sized packets over long distances and at constant velocity contrasts with conventional diffusion @xcite or drift - driven @xcite processes , where the ix fluid undergoes self - expansion .in addition , it enables synchronization with pulsed optical excitation by appropriately selecting the saw frequency .the saw - driven long - range ix flow can be efficiently controlled by the exat transistor placed on the transport path . the exat electrode configuration ( cf .[ exat](a ) ) is similar to the exciton optoelectronic transistor introduced in refs.@xcite . here, the voltage applied to the gate electrode ( g ) creates ( via the qcse ) a tunable energy barrier , which controls the acoustically driven ix flow between the source ( s ) and the drain ( d ) electrodes .in contrast to its optoelectronic counterpart , the ix flow through the exat is acoustically ( rather than diffusion or drift- ) driven with the flow direction defined by the saw propagation direction .note that , in an exat , the roles of the source and drain electrodes are interchangeable simply by reverting the saw propagation direction ( not shown ) . to simplify the notation , we denote by @xmath20 the port with the smaller area .m wide and separated by the same distance from the adjacent s and d regions .the lower color plots display pl images of the acoustic transport of ixs across the exat recorded in the ( b ) on state ( with biases v@xmath21=v@xmath22=v@xmath23=-1.4 v , cf .right inset ) and ( c ) off state ( v@xmath21=v@xmath22=-1.4 v , v@xmath23=2.5 v , cf . left inset ) . in both cases ,the ixs were driven by a saw generated using p@xmath2=2 dbm .the filled circles mark the position of the laser excitation spot .( d ) spatially resolved @xmath24i@xmath25 profiles along the transport channel in the on ( red curve ) and off ( blue curve ) states .( e ) and ( f ) display the corresponding time - averaged pl profiles calculated by a numerical model ( see text ) .the overlays indicate the positions of drain , gate and source electrodes . ]the switching action of the exat is demonstrated by the differential pl ( @xmath24i@xmath25 ) images in figs .[ exat](b ) and [ exat](c ) .these images display the changes induced by the saw and were obtained by subtracting pl images recorded in the presence and absence of the acoustic excitation . in both cases , excitons were optically excited on d , which was kept at the same bias voltage as s ( v@xmath21=v@xmath22=-1.4 v ) .note that the negative @xmath24i@xmath25 values around the excitation spot are due to the ix extraction by the saw and their transport along the saw propagation direction . in the on state ( fig .2(b ) ) , ixs can be easily transferred from d to s by applying a gate voltage v@xmath23=v@xmath21=v@xmath22 ( cf .right inset of fig .[ exat](b ) ) .the fringe electric field at the edges of the electrodes induce a potential barrier for ixs between s and g and between g and d , even when all electrodes are at the same potential .this barrier , however , is low in the on state and can be easily surmounted by the acoustic field .the transport , however , becomes completely blocked when the gate voltage is increased to v@xmath23=2.5 v ( off state in fig .the switching efficiency is quantified in fig .2(d ) , which compares @xmath24i@xmath25 profiles along the saw transport channel in the off and on states . note that in contrast to purely electrostatic transistor @xcite , which require a potential gradient between source and drain , the exat can be operated with these two ports at the same bias levels .the lower panels in fig .[ exat ] ( ( e ) and ( f ) ) compare the experimental results for the ix transport across an exat with a numerical modelling of the dynamics of the ix fluid .this model , which uses the experimental sample parameters , includes the mutual interactions between ixs and the external potential induced by the electrostatic gates and the saw .details of the model together with addition simulation results ( including videos demonstrating the exat operation ) are contained in the supplementary material sm accompanying the manuscript .the numerical analysis of the ix transport across the exat described in the sm show that the effect of the source , gate and drain edges on the exciton flow are negligible when the exat is in the on state .the good agreement between the experiment and modeling confirms that the device functions are well - predicted by our model of ix fluid dynamics in mobile potentials . the electrically controlled separation between the input and output ports ( i.e. , s and d electrodes ) in the exat allows us to prove that the acoustic field really transports both exciton constituents ( i.e. , both electrons and holes ) along the dqw structure .this question is relevant since the remote pl in ix transport experiments under a single ste can also be induced by the recombination of one type of carrier transported along the semiconductor channel with carriers of opposite polarity injected from the contacts ( as for the ix bright spots reported in @xcite ) . in order to address this issue, we have determined the absolute sensitivity of the setup , which is defined as the photon flux required for a given count rate measured on the ccd camera detecting the pl .the sensitivity calibration was carried out using a reference laser beam reflected on a metal film deposited on the sample surface . knowing the reflectance of the metal , it is possible to estimate the absolute reflected flux of photons .the ratio between the photon flux collected by the microscope objective in front of the sample and the ccd count rate gives the required calibration coefficient .the number of photons generated via the recombination of ixs was determined from the measured ccd count rate by taking into account the collection angle ( which depends on the refractive index of the sample layers and on the numerical aperture of the objective lens ) and the transmission of the semi - transparent metal electrodes .if the pl signal comes from the recombination of one type of carriers transported by the saw with carriers of opposite polarity electrically injected from the electric contacts , we expect the contact current to be given by @xmath26 , where @xmath27 is the detected photon flux and @xmath28 the electron charge .if the measured current is much less than @xmath29 , then at most of the electrons or holes forming the ixs must have been transported along the dqw structure by the saw . for the experimental conditions of fig .2b , where the laser beam is placed on the drain electrode , the measured photon flux in the source of the exat would require a source current @xmath30 na .this current is two orders of magnitude larger than the measured source current ( of 0.5 na ) under the same illumination conditions , showing that both electrons and holes must have been transported by the saw .similar results were obtained for the current in the drain electrode , where recombination takes place at the edge of the drain area .we now introduce the exam ( cf .micrograph of fig .[ fig2](a ) ) , a device capable of routing ix fluids between different input / output ports ( @xmath31 , labeled by the port index i=1, ... ,8 ) .each i / o port contains an idt ( not shown ) to generate an outgoing saw beam as well as an exciton acoustic transistor ( exat , cf .[ exat](a ) ) to control the ix flux between the exat source area and the common drain region , where multiplexing takes place . in each i/ o port , an input optical signal ( represented by a laser incident on the source electrode of exat@xmath32 in fig .[ fig2](a ) ) can be converted into ixs and stored in the exat source electrode .alternatively , stored ixs can be reconverted to photons by selecting an appropriate source voltage to reduce the ix recombination lifetime .the exat source acts , therefore , as a voltage - tunable location for photon - to - exciton interconversion as well as for ix storage .( saw@xmath33 or saw@xmath34 ) underneath the common drain .the dashed arrows indicate ix transfer between saw@xmath32 , saw@xmath33 and saw@xmath34 .the solid square labeled _ i _ denotes the intersection region of saw@xmath32 and saw@xmath33 ( or saw@xmath32 and saw@xmath34 ) .( b ) array of moving potential dots formed by saw interference at _ i_. ( c ) simulated time - averaged emission pattern within _i _ showing the ix transfer between saw@xmath32 and saw@xmath33 . ]ix transfer between two facing i / o ports ( e.g. , ports 1 and 4 in fig . [ fig2](a ) ) is carried out by simply turning the corresponding saw and exat ( i.e. , saw@xmath32 and exat@xmath32 ) on .the multiplexing function relies on the transfer of ixs between two orthogonal acoustic beams ( e.g. , from the vertical beam saw@xmath32 to the horizontal beam saw@xmath33 or saw@xmath34 ) at their intersection area ( _ i _ in fig .[ fig2](a ) ) .we have previously shown that the moving piezoelectric dots can efficiently transport uncorrelated electron - hole pairs between orthogonal piezoelectric saws @xcite . a similar concept using a moving array of strain dotsis applied here to transfer ixs between two non - piezoelectric saws .the moving dot array is created by the interference of the two saw beams . as seen in fig .[ fig2](b ) , the dots propagate with a velocity @xmath35 ( @xmath17 denotes the saw phase velocity ) in channels forming an angle of 45@xmath36 with respect to the propagation directions of the two interfering saws @xcite . due to their oblique propagation direction , the dots can capture the ixs from the input beam and transfer them to the other beam , as illustrated in the time averaged pl simulation of fig .[ fig2](c ) .the oblique lines show the propagation path of moving dots : a detailed numerical analysis of the transfer process including videos is presented in the sm .the multiplexing action is illustrated by the pl images of fig .figure [ fig3](a ) shows the direct transfer of ixs by saw@xmath32 from i / o@xmath32 to i / o@xmath37 , which terminates close to the gate of exat@xmath37 ( kept in the off state ) .the transfer of ixs from saw@xmath32 to either saw@xmath33 or saw@xmath34 is demonstrated in figs .[ fig3](b ) and [ fig3](c ) , respectively . when saw@xmath33 ( saw@xmath34 ) is switched on , ixs transported by saw@xmath32 to the intersection area are captured by the moving potential dots and transferred to saw@xmath33 ( saw@xmath34 ) . while the dot propagation paths are not spatially resolved in the experiment , their oblique propagation ( cf .[ fig2](c ) ) is clearly observed within the intersection areas of the saw beams in figs .[ fig3](b ) and [ fig3](c ) .the few ixs remaining within saw@xmath32 lead to the weak recombination observed in the lower part of this beam . from the pl image in fig .[ fig3](b ) , we determined that more than 90@xmath38 of the ixs are transferred from saw@xmath32 to saw@xmath33 and recombine near exat@xmath34 , thus demonstrating a high beam - to - beam transfer efficiency .these results demonstrate that ixs can be reversibly transfered from any input port to any output port by appropriately switching on at most three saw beams .the overhead time for switching will be established by the bandwidth of the idts and by the length of the saw channels . and between ( b ) saw@xmath32 and saw@xmath33 or ( c ) saw@xmath32 and saw@xmath34 .the output exat in ( a)-(c ) is in the off state .( c),(d ) control of ix flux by the gate electrode of exat@xmath39 .all saw beams are generated by applying @xmath40=10 dbm to the idts .the vertical ( horizontal ) dotted rectangle marks the transport channel formed by saw@xmath32 ( saw@xmath33 or saw@xmath34 ) . ]the pl images of figs .[ fig3](a)-[fig3](c ) have been recorded with the exats at the output ports in the off state , thus leading to the recombination of the transported ixs close to their gate electrodes .the combined action of long - range ix transport with electrostatic exat switching is exemplified in fig .[ fig3](d ) .the image was recorded under the same conditions as fig .[ fig3](c ) except that exat@xmath39 was turned on by changing its gate bias . under these conditions ,the ixs are transported to the transistor source area .flow control by the exat gate electrodes , which can isolate the sources from the common drain , becomes particularly important for the electric control of ix storage and recombination using the source bias .the total transport distance between the input and output ports in fig .[ fig3](d ) approaches 1000 @xmath1 m , thus yielding ix lifetimes exceeding 380 ns .these transport distances , which are well beyond the state of the art in the field , are limited by the length of the transport channel rather than by the acoustic transport efficiency ( see sec . [ acoustic_exciton_transport ] ) .in fact , ix lifetime can be increased by at least another order of magnitude , thus allowing much longer transport lengths with minimal signal loss and distortion , which is essential for the scalability of exciton - based circuitry .a further interesting feature of the saw - induced transport is that the ixs remain trapped within the 50 @xmath1m - wide beams ( outlined by vertical dotted rectangle in fig .[ fig3](a ) ) .the fabrication of the transport channels does not require lateral electrode structuring or the introduction of lateral interfaces , which can deteriorate the material s properties .moreover the exam concept is extensive to other systems and can be applied to interconnect an arbitrary number n of remote communication ports . in particular , that the maximum required acoustic transport length ( and ix lifetime ) does not limit the scalability since it increases only with @xmath41 . also , by decreasing the overlap length of the idt fingers , the width of the saw beams , and hence of the transport channels , can be further reduced . for a given length ,the minimum channel width will be ultimately determined by saw diffraction .the latter can be minimized by reducing @xmath13 or by shaping the saw beams using focusing idts .all these features make the acoustic transport a very attractive technique to interconnect remote exciton systems . in conclusion, the integrated exciton device presented in this work delivers a proof - of - concept for a scalable optoelectronic circuit capable of performing multiple electronic operations using ixs as operation medium .the exam is a building block , which can be easily integrated with other devices for on - chip optoelectronic applications .finally , the long - range ix transport by moving acoustic potentials can provide a versatile framework for the manipulation and coupling of other exciton phases , including ixs or exciton - polaritons condensates .* acknowledgements . *the authors thank m. ramsteiner for scientific discussions and s. rauwerdink and w. seidel for technical support in sample processing .we gratefully acknowledge financial support by dfg project no .sa 598/9 .j. kasprzak , m. richard , s. kundermann , a. baas , p. jeambrun , j. m. j. keeling , f. m. marchetti , m. h. szymaska , r. andr , j. l. staehli , v. savona , p. b. littlewood , b. deveaud , and l. s. dang , nature * 443 * , 409 ( 2006 ) .a. g. winbow , j. r. leonard , m. remeika , y. y. kuznetsova , a. a. high , a. t. hammack , l. v. butov , j. wilkes , a. a. guenther , a. l. ivanov , m. hanson , and a. c. gossard , phys .lett . * 106 * , 196806[4 ] ( 2011 ) .

<S> excitons , quasi - particles consisting of electron - hole pairs bound by the coulomb interaction , are a potential medium for processing of photonic information in the solid - state . </S><S> information processing via excitons requires efficient techniques for the transport and manipulation of these uncharged particles . </S><S> we introduce here a novel concept for the interconnection of multiple remote exciton systems based on the long - range transport of dipolar excitons by a network of configurable interconnects driven by acoustic wave beams . by combining this network with electrostatic gates , </S><S> we demonstrate an integrated exciton multiplexer capable of interconnecting , gating and routing exciton systems separated by millimeter distances . </S><S> the multiplexer provides a scalable platform for the manipulation of exciton fluids with potential applications in information processing . </S>

hep-ph9803322

in collider physics , clustering of the experimentally accessible hadronic final states is used to determine the underlying parton structure of events . in @xmath1 annihilation the widely known jade @xcite and durham @xcite jet algorithms have become indispensable in this process , permitting a wide range of important tests of qcd , allowing refined measurements of electro - weak physics with hadronic final states and being used in searches for new physics .recently a new jet clustering scheme , known as cambridge , has been introduced @xcite .this scheme is a modification of the original durham @xmath2-clustering scheme .the cambridge algorithm is designed to minimise the formation of spurious ` junk - jets ' , jets formed from a multitude of low transverse momentum particles , unrelated to the underlying parton structure . for all the above mentioned algorithms , clustering of the final stateis performed iteratively and is terminated at a clustering specific resolution scale , generically denoted by the resolution parameter @xmath0 . by changing the value of @xmath0, the final state is resolved into a varying number of jets .the cambridge algorithm involves three basic components in this iterative process .it uses an ordering variable , @xmath3 , a test variable , @xmath4 , and a recombination procedure . in jade - type jet clustering algorithms only two basic components are involved , since the ordering variable , @xmath3 , and the test variable , @xmath4 , are identical . in this notewe review the cambridge finder and discuss some of its experimental peculiarities . in terms of computingthis algorithm is more complex compared to the jade and durham algorithms . due to the distinction between test and ordering variables, the sequence of clustering now depends on the value of @xmath0 .we show that the jet multiplicity obtained with this algorithm is not monotonically decreasing for increasing @xmath0 , and that for some events it is impossible to resolve a certain jet multiplicity .therefore the concept of the ` transition values in @xmath0 ' has to be defined more precisely .the transition value at which the event classification changes from @xmath5-jets to @xmath6-jets , when going to larger values for @xmath0 , will subsequently be referred to as @xmath7value .next we developed a fast algorithm to obtain the transition values @xmath7 for the cambridge finder . using this algorithm, we compare results for monte carlo generated @xmath8 events between the durham and cambridge finder .we compare their performance in determining the size of the hadronization corrections . as another example, we determine the performance for hadronic decays of @xmath9 production at lep2 @xcite .finally we give our conclusions and cite an address to download our fortran code .in jade - type jet clustering algorithms one iteratively combines particles to form final state jets .first one introduces a ` test variable ' @xmath4 .the pair of two objects @xmath10 and @xmath11 with smallest value for @xmath4 is selected and its objects are combined or the iteration is terminated when @xmath12 for all pairs of objects . for thejade and durham algorithms , the test variables @xmath13 and @xmath14 are defined respectively as @xmath15 where @xmath16 and @xmath17 denote the energies of particles @xmath10 and @xmath11 and @xmath18 their opening angle .note that we normalise the values of @xmath13 and @xmath14 to the visible energy , @xmath19 , which is the sum of energies for all particles observed in the final state .the second ingredient is the recombination procedure .normally the @xmath20-scheme is taken , for which the four - momentum of the resulting object is simply the sum of the four - momenta of the two objects @xmath21 and @xmath22 .in contrast to this the cambridge algorithm involves three basic components to form the final state jets .the algorithm starts from a table of @xmath23 primary objects , which is the set of the particles four - momenta .it starts clustering the pair of particles with the smallest opening angle , using the ordering variable @xmath3 .the test variable @xmath14 , which is identical to the one for the durham algorithm , decides when the iterative procedure is stopped .it is subsequently denoted by @xmath4 .the algorithm proceeds as follows : 1 .if only one object remains , store this as a jet and stop .2 . select the pair of objects @xmath10 and @xmath11 that have the minimal value for their ordering variable , @xmath3 , with @xmath24 .3 . inspect the test variable @xmath4 . * if @xmath25 then combine @xmath10 and @xmath11 in a new object using the @xmath20-scheme . remove particles @xmath10 and @xmath11 from the table of objects that remain to be combined and add the new object with four - momentum @xmath26 . *if @xmath27 then store the object @xmath10 or @xmath11 with the smaller energy as a separated jet and remove it from the table .the higher energetic object remains in the table .removing the softer of two resolved objects , as described in the last step , is called _soft freezing_. it prevents the softer jet from attracting any extra particles , thereby reducing non - intuitive clustering effects .in order to test the various clustering algorithms , we generate monte carlo @xmath28 events at @xmath29 gev with the pythia event generator @xcite . the generation includes parton showering ( ` parton - level ' ) , and subsequent fragmentation and decays of the final state ( ` hadron - level ' ) .the parameters of the monte carlo event generator are adjusted in order to provide an optimal description of large samples of hadronic @xmath30 decay data @xcite . to illustrate the differences between the cambridge and durham finders we present in figure [ f : fourv ] the three - momenta of a typical event projected onto the @xmath31-plane .the underlying parton level is shown in the figure by the thick arrows and consists of a quark @xmath32 recoiling against a @xmath33 system , with the gluon being relatively soft . at the hadron level ,the event is clustered again to three final state jets , both with the durham and cambridge algorithm .the final jets are indicated by thick arrows , and the association of particles to the three jets is indicated by various line styles . in this example one clearly observes the positive effect of soft freezing on the hadronization corrections . in the cambridgealgorithm the soft gluon jet is separated and classified as a final state jet .most particles in the hemisphere are assigned in an intuitive way to the quark jet .the three final jets closely resemble the underlying parton structure .in contrast to this , in the durham algorithm more particles are clustered around the soft gluon , so that the gluon jet becomes even more energetic than the quark jet .it is obvious that in this example the final state found for the cambridge algorithm resembles the parton structure more than the durham algorithm . as function of @xmath0 for two montecarlo generated events .the dash - dotted line is for the durham algorithm , the full line for the cambridge algorithm .the points mark the @xmath7 transition values ._ , title="fig : " ] as function of @xmath0 for two monte carlo generated events .the dash - dotted line is for the durham algorithm , the full line for the cambridge algorithm .the points mark the @xmath7 transition values ._ , title="fig : " ] as an illustration of some of the peculiarities of the cambridge algorithm , in figure [ f : jetm ] we present the jet multiplicity as function of @xmath0 for two events . in the figures the dashed lines correspond to the durham algorithm , whereas the full lines correspond to the cambridge clusteringfor the durham algorithm , the jet multiplicity is decreasing monotonically for increasing @xmath0 .in addition , each event can be resolved into each jet - multiplicity @xmath34 , with @xmath35 . in the cambridgefinder the situation is a little more complex , as can be seen in the same figure .in the left plot of figure [ f : jetm ] an example is given where * the jet multiplicity is not monotonically decreasing for increasing @xmath0 .in this example , at the resolution @xmath36 , the jet - multiplicity decreases from 6 to 5 to 4 when @xmath0 increases , but then increases from 4 to 5 again . at @xmath37the multiplicity decreases from 5 to 4 . at @xmath38a situation occurs where the jet multiplicity does not change from being 4 , but the four final state jets change their four - momenta .the jet configuration of the two 5-jet states and three 4-jet state are all different . in the right plot of figure[ f : jetm ] we show an example where * it may not be possible to resolve the event into a certain @xmath5-jet final state .in this example , at @xmath39 , the event changes from being classified as a 6 jet event to a 3 jet event . for this event, it is impossible to choose a value for @xmath0 such that the event is resolved into a 4 or a 5 jet configuration . as a last ` peculiarity ' of the cambridge jet clustering we consider the particle to jet association . for the jade and durham algorithms , when crossing a transition value in @xmath0 towards higher values , two of the jetsare merged into one new jet while all other jets are left untouched .the resulting new jet consists of exactly all particles that belonged to the two merged jets , and the subjet history of jets can be traced unambiguously . in the cambridge algorithmthis need not be the case , since the sequence of recombination may be different for different values of @xmath0 .it thus can happen that the particle contents of a jet at a given value for @xmath0 does not match the sum of the particle contents of two resolved jets at lower values of @xmath0 . for many applicationsit is essential to obtain the transition values @xmath7 .for example , in previous studies of @xmath1 annihilation data the value of @xmath40 was analysed in order to obtain @xmath41 @xcite . in other studiesall events were classified as four @xcite or five @xcite jets and their angular correlations were studied in order to probe the non - abelian nature of qcd . in current studies at @xmath1 annihilation energies reached by the lep2 programme @xcite , events also have to be clustered to four jets in order to determine the @xmath42-boson characteristics in the hadronic decays of @xmath9 pairs .therefore , an algorithm to obtain all transition values @xmath7 with full information on the particle to jet association is highly desirable .in the jade and durham algorithm , the sequence of clustering of an event can be determined once and completely , and is independent of the value of @xmath0 . from this clustering information about jet multiplicities , four - momenta and jet - particle association ,can subsequently be retrieved for any value of @xmath0 .this is the strategy used in the ktclus@xcite and ykern @xcite packages .the final jet configuration is identical for all values of @xmath0between two subsequent transition values . at the transition value @xmath44 , the event flips from a @xmath45-jet to a @xmath5-jet configuration .the transition values for the jade and durham algorithms are ordered in @xmath0 .using the transition values one can select a value for @xmath0 such that the event is resolved into the required number of jets .in contrast to this , the clustering sequence in the cambridge algorithm depends on the value of @xmath0 because it distinguishes between ordering and testing variables .it is therefore no longer straightforward to calculate the transition values . in general , at the transition values @xmath46 the event can flip between a @xmath5-jet configuration to a @xmath6-jet configuration where @xmath5 and @xmath6 are not necessarily consecutive .as it is important to obtain the values for @xmath7 , it was suggested in @xcite to perform a binary search in @xmath0 to determine these transition values , by repeated evaluation of the clustering .this proposal is not completely satisfactory since such a search has an intrinsic limited precision , might skip over several transition values and becomes very computing time intensive .we have developed a method to determine the transition values of @xmath0 for the cambridge finder exactly , as follows . while performing the clustering at a particular value of @xmath0 , denoted by @xmath47 , we keep track of the maximum value of @xmath4 , between any two objects @xmath10 and @xmath11 encountered in this process , with @xmath4 being always smaller than @xmath47 . by construction this maximum value , which we denote by @xmath48 , is smaller than @xmath47 .we now note that for any value of @xmath49 , the cambridge algorithm will follow the same clustering sequence .only when the cluster algorithm is performed with a value @xmath0 smaller than @xmath48 , the condition @xmath27 is satisfied at least once more and the subsequent clustering sequence may change completely .the value @xmath48 is therefore one of the @xmath0 transition values .note that the clustering may also change completely for values of @xmath0 larger than @xmath47 .these observations can be utilised to scan the complete region of @xmath0 .we therefore start by clustering the complete event to a one - jet configuration by chosing @xmath50 in the first step . after this stepone iteratively repeats the clustering to calculate smaller and smaller values of @xmath48 at which the clustering changes , and thereby calculates smaller and smaller transition values . the process terminates if either the number of resolved jets equals the number of input objects or if the desired number of jets is resolved . to summarise : 1 .start with value @xmath51 and set @xmath52 .2 . perform the cambridge jet clustering for @xmath23 objects . during the clustering ,keep track of all values of @xmath4 between all objects @xmath10 and @xmath11 , and determine their maximum value , @xmath48 .3 . store the value of @xmath48 , the number of jets , @xmath5 , their four - momenta and the jet - particle association .the clustering for @xmath53 is now completely determined .4 . the algorithm stops if : * the number of resolved jets equals the number of input objects , @xmath54 .then the event is classified completely and the algorithm necessarily stops . * the desired number of jets or a preset lower limit in @xmath0 is reached , and the algorithm is stopped .set @xmath55 and go to step 2 .once this process has been performed , all information about the clustering is accessible without any appreciable additional computing time .the total amount of computing time is proportional to the desired jet - multiplicity .for example , to study four jet final states with the cambridge finder requires approximately four times as much computing time compared to the durham or jade algorithms .+ instead of the top - down approach for which the clustering starts at @xmath56 as explained above , a bottom - up approach is in principle also possible .one may implement the bottom - up approach by starting the clustering at the lowest value @xmath57 . for @xmath23given at the start of the clustering , one finds the pair of objects with smallest value @xmath3 ( corresponding to the pair closest in angle ) and determines the corresponding value for @xmath4 .then the two possible cases are considered : one in which the softer object is frozen , the other in which the two objects are combined . in both casesthe number of objects that remain to be combined is reduced by one .this combinatorical procedure is subsequently continued , and _ all _ possible clustering sequences are listed .the procedure terminates when only one object remains . from the corresponding values for @xmath4 , saved for each step, one can deduce the final transition values @xmath7 and the jet configuration associated to them .note that the number of possible clustering sequences is proportional to @xmath58 , which limits the practical use of the bottom - up approach .with the transition values @xmath7 defined both for the cambridge and durham algorithms , we compare , as an example , the values for @xmath40 . in @xcite similar studies have been performed to compare the performance of the durham and jade algorithms . in all the following we will define the region with the highest value for @xmath0as the nominal @xmath0 region . here , in figure[ f : mctest1]a we show the correlation of the cambridge and durham algorithms for @xmath40at the parton level , at the end of the pythia parton shower . for most of our generated events ,the obtained values for @xmath40 are identical for the cambridge and the durham algorithms ( approximately 75% of the events are found on the line in the figure ) . for a small fraction of events ,the value obtained with the cambridge algorithm is smaller compared to the value for the durham algorithm . at the hadron level ,as shown in figure [ f : mctest1]b , the values at low @xmath40 obtained using the cambridge algorithm are smaller for almost all events , but become similar for the two algorithms for increasing values of @xmath40 . at the hadron level ,approximately 15% of the events have identical values for @xmath40 for both algorithms .next , in figure [ f : mctest2]a and [ f : mctest2]b , we compare the hadronization corrections for the cambridge and durham algorithms .we present the correlation between the transition values @xmath40 calculated at the hadron and at the parton level , for both .the line indicates the ideal case for which equal values for @xmath40 at both levels are found . for the durham algorithmthe difference in @xmath40 at the parton and hadron level is small . when going to lower @xmath40 values , the distribution broadens and shifts toward smaller @xmath40 values at the hadron level .for the cambridge algorithm , at high values of @xmath40the parton and hadron level correlation is similar to the one for the durham algorithm .whereas , when going to smaller values for @xmath40 , the values at the hadron level get increasingly larger with respect to the parton level values .the width of the distribution is similar to that for the durham algorithm . in order to quantify the differences ,we calculated the mean of the logarithmic ratio of the @xmath40 values for the parton level and the hadron level : this value equals @xmath59 for the durham algorithm , and @xmath60 for the cambridge algorithm , which indicates that the overall hadronization corrections for the durham algorithm are @xmath61% smaller than for the cambridge algorithm .note however that the hadronization corrections do not only depend on the jet algorithm but also on the hadronization model used .the mean hadronization corrections can be studied more directly , as a function of @xmath0 , from plots as presented in figures [ f : mctest3]a and [ f : mctest3]b . in figure[ f : mctest3]a we show the mean of the logarithmic ratio of the values @xmath40 for the parton and hadron level , @xmath62 , as a function of the transition value at the parton level . when calculating the mean deviation between hadron and parton level for each value @xmath40 , the contribution to the hadronization corrections for many events may cancel . to exclude effects due to cancellation we present in figure [ f : mctest3]b the size of the hadronization corrections .we show the mean absolute difference of @xmath40 calculated on the parton and at the hadron level , @xmath63 , as a function of @xmath40calculated at the parton level . to compare the performance of the two jet finders we distinguish in figure [f : mctest3]a and [ f : mctest3]b three regions in @xmath40 , denoted by @xmath64 , @xmath65 and @xmath66 . in region @xmath64 , for values of @xmath40above @xmath67 , the hadronization corrections are small and comparable for both algorithms .a fraction of about 37% of our generated events belongs to this region .region @xmath65 is defined for values of @xmath40 between @xmath68 and @xmath67 . in this region differences between the two algorithmsthe mean deviation for the durham algorithm reaches a maximum of about 20% , and vanishes at @xmath69 , as can be seen in figure [ f : mctest3]a .however , figure [ f : mctest3]b shows that this decrease is due to cancellations and that the absolute hadronization corrections increase at @xmath69 . for the cambridge algorithm a different behaviouris observed .the mean deviation reaches a maximum of 100% for this algorithm , at @xmath70 , implying large hadronization corrections .the absolute hadronization corrections for the cambridge algorithm , as shown in figure [ f : mctest3]b , reach a maximum at about @xmath71 , and then decrease until the value for the durham algorithm is reached . in the whole region @xmath65 , where about 49% of our generated events can be found , hadronization corrections for the cambridge algorithm are significantly larger than for the durham algorithm . in region @xmath66 , for values of @xmath40 below @xmath68 ,the figures show that the hadronization corrections are large for both algorithms and that they increase rapidly towards smaller values of @xmath40 .the corrections for the cambridge algorithm are smaller compared to the durham algorithm .however , only about 14% of our generated events can be found in region @xmath66 .our analysis of the hadronization corrections for @xmath40 shows that the cambridge algorithm performs clearly better only in the region of low @xmath40 values ( region @xmath66 ) .this region contains 14% of the events and the corrections there are large for both algorithms . in all other regions the durham finder performs equally well ( region @xmath64 ) or even significantly better ( region @xmath65 ) .these two regions contain a fraction of 85% of our generated @xmath8 events .these basic tendencies of the hadronization corrections are also found when studying the transition value @xmath72 .note that for very low values of @xmath0 the approximations used for the implemention of qcd in pythia might not give a reliable description of the hadronization process .therefore , for very low values of @xmath0 the jets returned by the cambridge algorithm may correspond closer to the underlying parton structure than can be shown in these monte carlo studies .+ classical tests of qcd rely on relative production rates for multijet hadronic decays , defined as @xmath73 @xcite . in figures [ f : mctest4]a and [ f : mctest4]b we present the relative production rates for two , three , four , and five or more jet final states , for the hadron level and the parton level . for these figures we used the same set of @xmath8 events as before . in figure[ f : mctest4]a the performance for the durham algorithm is shown .for all @xmath0 values between one and approximately @xmath74 the hadron and parton level agree reasonably well . when going to smaller values of @xmath0 the curves for the two levels increasingly deviate . for the cambridge algorithmthe hadronization corrections are larger for the region in @xmath0 where the durham algorithm performs well , between @xmath0 of one and approximately @xmath74 . however , when going to lower values of @xmath0 the differences between hadron and parton level are , for most jet multiplicities , smaller than for the durham algorithm .to summarise our investigations of the hadronization corrections , we conclude that the durham algorithm provides smaller hadronization corrections for a large region in @xmath0 . in @xcite ,the hadronization corrections for the mean jet multiplicity , @xmath75 , was studied .there it was found that hadronization corrections for the cambridge and durham algorithms are small for values of @xmath76 .our figures [ f : mctest4]b show that for these values of @xmath0 the hadronization corrections for each jet production rate , @xmath77 , are sizable for the cambridge algorithm , whereas for the durham algorithm they are small .the small hadronization corrections found for the cambridge algorithm in the study of the mean jet rate @xmath78 are due to fortuitous cancellations in the individual jet production rates .as already indicated , the transition values @xmath7 in the cambridge algorithm need not be the transition between two consecutive jet - multiplicities . several intervals in @xmath0 may lead to the same jet - multiplicity , and they have in general different jet four - momenta . secondly , it need not always be possible to cluster the event to any required jet multiplicity . in order to determine the frequency that this might occur, we generated for figure [ f : impos ] @xmath28 monte carlo events with full hadronization , at @xmath79 gev .the full points present the fraction of events that have multiple regions in @xmath0 with the same jet multiplicity , as a function of the jet - multiplicity . for our generated events , for example , about 2.2% have multiple regions in @xmath0 that lead to a four - jet final state , albeit with different jet four - momenta . in the same figurethe open points show the fraction of events were the indicated number of jets could not be resolved .for example , in about 2.5% of events no four - jet configuration could be found .the figure shows that the fraction of impossible jet multiplicities and multiple jet multiplicities increases with increasing @xmath5 , reaches a maximum at around @xmath80 , and decreases again .the generated events have a mean total multiplicity of 44.2 , and 95% of the events have a multiplicity larger than 25 .note that both distributions are naturally limited by the input number of four - momenta in each event .as another example we generated hadronic decays of @xmath9 pairs at lep2 @xcite using pythia : @xmath81 , at @xmath82 gev .information about the kinematics of the two @xmath42 s can be obtained by forcing the hadronic final state to four jets . using the cambridge finder, we find that about 0.9% of the events have multiple regions in @xmath0 with four final state jets .for those events one therefore has the freedom to select the set of jets with the larger @xmath0 values , or the set with the smaller @xmath0values .clearly the selection which corresponds closer to the four primary partons is preferred . in figure[ f : fourj]a we compare the mean opening angle between the jets and primary four partons , and in figure [ f : fourj]b the mean absolute energy difference between the jets and the primary partons .it can be clearly seen that in both cases the resolution is better for the fraction of 99% of events in which only one four - jet configuration is found .for the small fraction of events where two four - jet configurations were found , the jet configuration with the lower values of @xmath0 matches the four primary partons better than the one with larger values of @xmath0 , which can be explained by the following observation . in the majority of events for which the cambridge algorithm returned two four jet configurations the appearance of hard gluon radiation in the parton shower was observed .detailed inspection revealed that the hard gluon , radiated from a quark - pair originating from one @xmath42 , points in the direction of a quark originating from the other @xmath42 .the configuration with the low value of @xmath0 correctly separates the gluon from this quark by the mechanism of _ soft - freezing _ , whereas they are merged for the configuration with the larger value of @xmath0 .the correspondence between partons and jets is therefore better in the configuration with the lower value of @xmath0 .in this note we review the cambridge jet clustering algorithm , as was recently introduced in @xcite .we show some of its particularities for monte carlo generated events .firstly , the algorithm may find several regions in @xmath0 with identical final state multiplicity , but different jet four - momenta .secondly , for some events it is impossible to resolve a certain jet multiplicity .both these properties are absent in the jade and durham algorithms .we propose a fast , new algorithm that is able to determine the transition values for @xmath0 , based on the yclus package .all transition values , jet multiplicities , jet four - momenta and the jet to particle associations are derived and stored , and can be subsequently inferred for all values of @xmath0 without any substantial additional computing time . using this algorithm we determine the hadronization corrections of @xmath8 generated events according to pythia , by comparing parton and hadron level values for @xmath40 , both for the durham and cambridge algorithms .this comparative study of the two algorithms is completed by a presentation of the relative jet production rates . for a large interval of @xmath0 valuesthe hadronization corrections for the cambridge algorithm are found to be significantly larger than for the durham algorithm .however , in the region of very small values of @xmath0 ( @xmath83 ) , the hadronization corrections are large , but better under control for the cambridge algorithm .note that for very low values of @xmath0 the reliability of the comparative monte carlo studies is limited due to the fact that for these values of @xmath0 the approximations used for the implementation of qcd in pythia might not give an appropriate description of the hadronization process .further , we present for the cambridge algorithm the fraction of events for which certain jet multiplicity could never be resolved , or could be resolved multiple times .four jet final states were explicitly studied in hadronic decays of @xmath9 events .the large fraction of events where just one four jet configuration was found has better energy and angular resolution than the small fraction of events with multiple four jet configurations .fortran code , containing our ckern routines to obtain the @xmath0 transition values , can be obtained from the world - wide web at http://wwwcn1.cern.ch/~stanb/ckern/ckern.html .we would like to thank s. bethke for help and inspiring discussions , as well as b. webber , yu . dokshitzer , s. moretti and z. trocsanyi for comments .we like to thank cern for its hospitality and in particular the opal collaboration for providing indispensable recourses . 99 jade collaboration , w. bartel et al .. lett . * b123*(1993)460 ; z. phys . *_ yu . l. dokshitzer , contribution cited in report of the hard qcd working group , proc .workshop on jet studies at lep and hera , durham , december 1990 ,_ j. phys . *_ yu . l. dokshitzer , g. leder , s. moretti and b. webber , _ jhep * 08*(1997)001 . _ the opal collaboration , k. ackerstaff et al .j. * c1*(1998)395 . _ t. sjstrand , _ comput .. commun . * 82*(1994)74 . _ opal collaboration , g. alexander et al . ,_ z. phys .* c69*(1996)543 ._ opal collaboration , m. z. akrawy et al . , _ phys .lett . * b235*(1990)389 , _ + mark - ii collaboration , s. komamiya et al .lett . , * 64*(1990)987 ._ delphi collaboration , p. abreu et al .* b414*(1997)401 , _+ aleph collaboration , d. decamp et al . ,_ z. phys . *c76*(1997)1 _ + opal collaboration , r. akers et al . , _z. phys . * c65*(1995)367 . _ opal collaboration , r. akers et al . , _ opal internal physics note * 188*(1995 )_ m. h. seymour , _z. physik * c62*(1994)127 : _ + ktclus program available at http://hepwww.rl.ac.uk/theory/seymour/ktclus/ opal collaboration , m. z. akrawy et al . ,_ z. phys . *c49*(1991)375 , _ + opal collaboration , p. acton et al . ,_ z. phys .* c55*(1992)1 ._ s. bethke , z. kunszt , d.e .soper and w.j .stirling , _ nucl .phys . * b370*(1992)310 ._ + s. bethke , z. kunszt , d.e .soper and w.j .stirling , _ erratum : jet clustering algorithms in @xmath1annihilation , hep - ph/9803267 . _

<S> jet clustering algorithms are widely used to analyse hadronic events in high energy collisions . recently </S><S> a new clustering method , known as ` cambridge ' , has been introduced . in this article </S><S> we present an algorithm to determine the transition values of @xmath0 for this clustering scheme , which allows to resolve any event to a definite number of jets in the final state . </S><S> we discuss some particularities of the cambridge clustering method and compare its performance to the durham clustering scheme for monte carlo generated @xmath1 annihilation events . </S><S> european laboratory for particle physics cern - ep/98 - 043 + march 11th , 1998 * the cambridge jet algorithm : + features and applications * stan bentvelsen and irmtraud meyer submitted to eur . </S><S> phys . </S><S> j. c </S>

1406.4852

regenerating codes were introduced by dimakis , godfrey , wu , wainwright and ramchandran @xcite .their main application is in large distributed storage systems where they lead to significant savings by optimizing the trade - off between storage size and repair bandwith . in a distributed storage system ( dss )an encoded file is stored on @xmath3 servers such that it can be recovered from any combination of @xmath4 servers .if a server fails it can be rebuilt by retrieving the information needed for its repair from any combination of @xmath5 other servers . an encoding scheme realizing these parametersis called an @xmath0 regenerating code . for background and details on distributed storage and regenerating codeswe refer to @xcite , @xcite .a common example is the use of a @xmath6 code to store four bits @xmath7 . by storing the pairs of bits @xmath8 on four different servers ( @xmath9 ) , the four bits @xmath7 can be recovered from the combined information on any two servers ( @xmath10 ) . andif a server fails it can be rebuilt by retrieving one bit from each of the remaining three servers ( @xmath11 ) . in particular , the first server can be rebuilt from the three bits @xmath12 and @xmath13 .an @xmath0 code comes with a secondary set of parameters @xmath14 . for a file of size @xmath15 , a part of size at most @xmath16is stored on a single server , and bandwith between a server and any of the @xmath5 servers helping in its repair is limited to @xmath17 .for the example , @xmath18 .the gains in a dss are obtained by using a total repair bandwith @xmath19 that is possibly larger than @xmath16 but much smaller than the file size @xmath15 .the challenge is , given @xmath0 , to optimize the trade - off between the storage @xmath16 per server and the repair bandwith @xmath17 between servers in order to store a file of size @xmath15 .for given parameters @xmath0 , the outer bound refers to the relation among the parameters @xmath14 .the outer bound can be interpreted as an upper bound on the file size @xmath15 , for given @xmath16 and @xmath17 , or as a lower bound for @xmath16 and @xmath17 , for a given file size @xmath15 . in the first case it is standard to scale to variables @xmath20 and @xmath21 , and in the second case to variables @xmath22 and @xmath23 . in this work ,we establish new outer bounds for exact - repair regenerating codes . in the exact - repair scenarioit is required that a server be rebuilt to its original form .the weaker requirement , known as functional repair , only requires that a server be rebuilt to a form that preserves the functionality of the dss .upper bounds for the file size under functional repair are piece - wise linear and take the form @xmath24 details of the bound and motivation behind the linear functions @xmath25 are recalled in section [ s:1 ] . for the values @xmath26 , @xmath27 for the four bit example with @xmath28the bound is sharp at both @xmath29 and @xmath30 .the vertex with @xmath31 minimizes @xmath16 and is called the msr point ( for minimum storage regenerating code ) .the vertex with @xmath32 minimizes @xmath17 and is called the mbr point ( for minimum bandwith regenerating code ) .both these points are achieved by exact - repair regenerating codes using the general construction in @xcite . using the notion of information flow graph and then applying results from network coding, the main result of @xcite shows that the bound ( [ bfr ] ) is sharp for regenerating codes under the functional - repair requirement .clearly , exact - repair implies functional - repair , and the outer bound ( [ bfr ] ) applies to exact - repair regenerating codes tian @xcite determined the rate region for a @xmath1 code and provided the first example of an exact - repair rate region that lies strictly within the functional - repair rate region . for a @xmath1 code @xmath33 the exact - repair regionis describeded by adding to ( [ b433 ] ) the new inequality @xmath34 to prove that all points in the new region can be achieved it suffices , using a standard time sharing argument , that the vertices with @xmath35 can be achieved .the first and the last are the msr and the mbr point .an example achieving @xmath36 is provided in @xcite .a different example is to encode eight bits @xmath37 as four triples @xmath38 the repair information to rebuilt the first server is given in parentheses . as part of their results , sasidharan , senthoor and kumar@xcite obtain the same inequality ( [ new ] ) .their main result however [ ibid . , theorem 1 ] is a non - vanishing gap between the functional - repair outer bound and the exact - repair outer bound for all codes with @xmath2 .our results include the following outer bound ( theorem [ thm - rs ] ) ._ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ for given @xmath39 , let @xmath40 be positive integers with @xmath41 .let @xmath42 be subsets of @xmath43 of size @xmath44 with empty intersection @xmath45 . then _ @xmath46 _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ for @xmath47 ,@xmath48 , it follows that @xmath49 .this difference is unbounded as @xmath50 goes to infinity . on the other hand ,the non - vanishing gap in @xcite remains bounded and is always less than @xmath17 .a second outer bound ( theorem [ thm - lm ] ) is obtained using a similar approach as in @xcite . for both outer bounds we give examples that illustrate the improvements over known bounds .we also show that the bounds can be combined for further improvements .the next section formulates the main problem .section [ s:2 ] presents the main arguments and how they are used in two different proofs for the rate region of a @xmath1-code .section [ s:4 ] proves theorem [ thm - rs ] .section [ s:3 ] builds on the approach used in @xcite and proves theorem [ thm - lm ] .it also contains a short proof for a non - vanishing gap of @xmath51 between the outer bounds for functional - repair and exact - repair .section [ s:5 ] illustrates how the results of the two previous sections can be used in combination .by an exact - repair regenerating code of type @xmath0 with secondary parameters @xmath14 we mean a collection of random variables @xmath52 , @xmath53 and @xmath54 that satisfy several entropy constraints .let @xmath55 denote the joint distributions @xmath56 and , for @xmath57 , let @xmath58 denote the joint distribution @xmath59 the entropy constraints are the following @xmath60 the interpretation for a distributed storage system is that @xmath52 is the file to be stored , @xmath61 is the part of the encoded file that is stored on server @xmath62 , and @xmath63 is the helper information provided by server @xmath62 to repair server @xmath64 . assuming uniform distributions for each of the variables , the conditions @xmath65 and @xmath66 describe the size of the underlying space for @xmath67 and @xmath63 , respectively .the condition @xmath68 says that the information stored on server @xmath62 is completely determined by the file @xmath52 , and similarly @xmath69 says that helper information provided by server @xmath62 is completely determined by information stored on server @xmath62 .finally , the access condition @xmath70 says that the file can be recovered from information stored on any @xmath4 servers , and similarly @xmath71 says that server @xmath64 can be rebuilt with helper information received from any @xmath5 remaining servers .clearly , for @xmath72 , @xmath73 and thus @xmath74 moreover , for @xmath75 , let @xmath76 be a partition with @xmath77 and @xmath78 , and let @xmath79 denote the joint distribution @xmath80 then @xmath81 and @xmath82 the upper bound @xmath83 applies to both the exact - repair and the functional - repair setting and is sharp for the latter @xcite .we will make use of the following .@xmath84 the single argument that we are aware of to improve the bound @xmath86 is to sum multiple copies of the bound for different choices of @xmath87 and to exploit nonzero mutual information among variables in the different copies . forthe @xmath1-code , one such sum of three copies is @xmath88 which corresponds to @xmath89 @xmath90 @xmath91 after adding a term @xmath92 , we can regroup the variables ( details are provided in section[s:4 ] ) .@xmath93 finally , @xmath94 leads to a savings of @xmath17 in the sum of the original three copies and thus to ( [ new ] ) .the same argument is at the core of the proof in @xcite .a different choice for the sum of three copies is @xmath95 which corresponds to @xmath96 @xmath97 @xmath98 first we regroup the variables to obtain ( details are provided in section[s:3 ] ) @xmath99 using @xmath100 and similarly @xmath101 leads to ( [ new ] ) .the final reduction follows ( * ? ? ?* equation ( 18 ) ) .a difference between the first choice and the second choice is that the first choice reduces the number of @xmath63 while the second choice reduces the number of @xmath61 at a cost of increasing the number of @xmath102 .we briefly point out the connection between regenerating codes and secret sharing . in secret sharing a sensitive messageis distributed over several parties such that only qualified combinations of parties can reconstruct the message . forthe distribution of a message @xmath52 using shares @xmath103 , @xmath104 let @xmath105 the minimal choice for @xmath106 ( resp . the maximal choice for @xmath107 ) is called the acceptance threshold ( resp . the rejection threshold ) .for the distribution of a file @xmath52 over servers @xmath103 , @xmath108 we use the conditions @xmath109 for a server @xmath103 , let @xmath110 be helper nodes .we add a second set of conditions @xmath111 these conditions express that a dss is similar to a two - layer secret sharing scheme where the condition on mutual information is relaxed from zero mutual infirmation to bounded mutual information .the condition of bounded mutual information is enforced to obtain efficient storage in the top - layer and efficient bandwith in the bottom layer ( rather than to reduce the information about the secret as in an actual secret sharing scheme ) .the two - layered secret scheme becomes a regenerating code if we enforce that a share @xmath112 is stored not on a dedicated second layer of servers but can be obtained from information on server @xmath61 .this is expressed by the conditions @xmath113 we add as conditions for the top layer @xmath114 the entropy conditions ( [ en1])([en3 ] ) imply the conditions ( [ eq : t1])([eq : wm ] ) and thus the optimization problem for regenerating codes reduces to a problem of share sizes for a special version of a two - layer secret sharing scheme .let @xmath4 and @xmath5 be fixed .we make use of three different configurations of random variables .minimal configurations appear in the upper bounds @xmath115 , for @xmath75 . for a partition of @xmath116 into subsets @xmath117 , @xmath52 and @xmath118,with @xmath119 @xmath120 @xmath121 @xmath122 with the interpretation of the random variables as edges in an information - flow graph, @xmath123 corresponds to a min - cut @xcite . for a partition of @xmath116 into susbets @xmath124@xmath125 @xmath52 and @xmath118 , the two configurations @xmath126 will be used for improvements of the min - cut bounds . for @xmath127 as in ( [ s3 ] ) , @xmath128 for @xmath127 as in ( [ s2 ] ) , @xmath129we illustrate each of the three cases for a partition of @xmath130 . putting a mark in position @xmath131 for @xmath132 and a mark in position @xmath133 for @xmath134, the configurations are represented by the diagrams @xmath135 ( a ) & & ( b ) & & ( c ) \end{array}\ ] ] @xmath136 the goal in the next sections is to collect several copies of type ( a ) , to break them into smaller pieces and to regroup them into a configuration of type ( b ) ( section [ s:4 ] ) , or type ( c ) ( section [ s:3 ] ) , or a combination of both ( section [ s:5 ] ) .the proofs for the improved outer bound ( [ new ] ) of a ( 4,3,3 ) code in section [ s:2 ] are special cases .the reduction @xmath94 uses ( [ r3 ] ) with @xmath137 .the reduction @xmath138 uses ( [ r2 ] ) with @xmath139 .assume that @xmath52 is uniformly distributed on a vector space @xmath117 of dimension @xmath15 , that the @xmath61 are uniformly distributed on vector spaces @xmath140 of dimension @xmath16 , and that the @xmath63 are uniformly distributed on vector spaces @xmath141 of dimension @xmath17 .we group the entropy conditions ( [ en1])-([en3 ] ) in a different way and give a vector space interpretation .@xmath142 condition ( [ c1 ] ) implies that there exists a linear map @xmath143 of rank @xmath16 with @xmath144 .together the maps define a linear encoder @xmath145 the image is a linear code of dimension @xmath15 and length @xmath146 .condition ( [ c2 ] ) implies that the generator matrix is of full rank @xmath15 on any submatrix of @xmath4 out of @xmath3 blocks of size @xmath16 .condition ( [ c3 ] ) implies that there exists a linear map of rank @xmath17 @xmath147 condition ( [ c4 ] ) implies that there exists a linear map @xmath148 and moreover that it factors as @xmath149 the factorization allows us to characterize a regenerating code of length @xmath150 through the structure of its parity check matrix .a linear code of dimension @xmath15 and length @xmath151 ( @xmath3 consecutive blocks of length @xmath16 ) represents a @xmath0 regenerating code with secondary parameters @xmath14 if @xmath152 for a code with @xmath150 , the construction of a regenerating code is equivalent to the construction of a square block matrix @xmath153 of size @xmath154 with blocks of size @xmath16 and with rank distribution @xmath155 such that columns in any @xmath156 blocks are independent ( equivalent to ( 1 ) ) . for the special case @xmath157 @xmath158 the last condition is automatically fulfilled . maximizing the rank@xmath15 is equivalent to minimizing the rank of the parity check matrix @xmath153 .for a @xmath1 code with @xmath159 we use @xmath160 the @xmath161 blocks are all equal to the identity matrix and thus the matrix has rank four and its row space is spanned by the first four rows .the code is the one used inthe introduction . for any @xmath162, there exists a code with@xmath163 and @xmath164 .thus @xmath165we restrict to the case @xmath150 .outer bounds obtained for @xmath0 codes apply to @xmath166 codes .let @xmath127 be a set of random variables , with each variable @xmath167 of the form either @xmath168 or @xmath169 .in the first case the entropy of @xmath170 is @xmath171 and in the second case it is @xmath172 the entropies @xmath16 and @xmath17 serve as weights for the random variables in @xmath127 and the weight of @xmath127 is defined as @xmath173 . by submodularity of the entropy function ,the weight of @xmath127 is an upper bound for the entropy of @xmath127 .[ chain ] let @xmath174 and @xmath175 be two sequences of sets of random variables such that for @xmath176 , @xmath177 then @xmath178 in particular , for @xmath179 , @xmath180 for @xmath181 , @xmath182 finally , sum the @xmath3 inequalities and apply the chain rule .[ thm - rs ] for given @xmath39 , let @xmath183 be positive integers with @xmath184 .let @xmath42 be subsets of @xmath185 of size @xmath186 with empty intersection @xmath45 . then @xmath187 for @xmath154 nodes@xmath116 , let @xmath188 , @xmath189 , and denote by @xmath118 the set @xmath185 .let @xmath190 be a partition of @xmath118 such that , for each @xmath62 , @xmath191 .the empty intersection of the @xmath140 guarantees that such a partition exists .we apply the proposition with suitable choices for @xmath174 and @xmath175 .let @xmath192 and let @xmath193 for each @xmath194 we choose @xmath195 such that @xmath196 is a minimal configuration and in particular @xmath197 .recall from section [ s:2 ] that a minimal configuration is determined by a partition @xmath198 . for @xmath199 ,we choose @xmath200 for @xmath201 , @xmath202 and for @xmath203 , @xmath204 . with these choices @xmath205 for @xmath176 . for @xmath47 , @xmath48, it follows that @xmath49 .this difference is unbounded as @xmath50 goes to infinity interpolation can be used to obtain similar estimates for other choices of parameters . the case @xmath1 .let @xmath206 .for @xmath207 and for @xmath208 , @xmath209 we list four minimal configurations and their division into @xmath210 @xmath211 the fourth row is used as an upper bound for @xmath92 but we can avoid it and use @xmath212 . then the bound becomes @xmath213 this improvement applies whenever the theorem is used with @xmath214 [ p1 ] for @xmath215 , @xmath216 for both the plus sign and minus sign we partition @xmath154 nodes into @xmath217 and we fix an ordering on each of @xmath218 , @xmath117 and @xmath118 . for the plus signwe choose the sets of size @xmath219 , @xmath220 and @xmath221 .let @xmath222 be the last element in @xmath117 and let @xmath223 be the last element in @xmath118 .we apply proposition @xmath224 with three minimal configurations . for each , we list @xmath225 , @xmath226 , @xmath227 and @xmath228 .@xmath229 for the minus sign we choose the sets of size @xmath230 , @xmath231 and @xmath221 . let @xmath232 be the last elements in @xmath117 and let @xmath223 be the last element in @xmath118 .we apply proposition @xmath224 with three minimal configurations . for each , we list @xmath225 , @xmath226 , @xmath227 and @xmath194 . @xmath233 [ e1 ] for a @xmath234 code we apply the theorem with @xmath235 , @xmath236. then @xmath237 this is less than the functional repair outer bound in the range @xmath238 .the gap reaches a maximum at @xmath239 where it lowers the bound @xmath240 by @xmath241 .we will compare this with other bounds in the next section .in graph terms , we consider the complete graph on @xmath154 vertices , with edges @xmath242 and @xmath243 .the variable @xmath63 connects node @xmath62 and node @xmath64 .the variable @xmath61 connects node @xmath62 with node @xmath62 or , after creating two copies of node @xmath62 , node @xmath62-in with node @xmath62-out .the sets @xmath127 define subgraphs with a block structure on the adjacency matrix .connections between nodes can be interpreted as channels of bandwith @xmath171 or @xmath66 . for a regenerating code with parameters @xmath0 the repair matrix for @xmath154 nodes is a square matrix of size @xmath154 that indicates which nodes function properly ( a nonzero entry in the position @xmath131 for node @xmath62 ) and which nodes help other nodes ( a nonzero entry in position @xmath133 for node @xmath62 helping node @xmath64 ) * ? ? ?a minimal configuration has @xmath244 diagonal entries and @xmath245 nonzero columns . for @xmath246 @xmath247 by replacing @xmath248 with @xmath16 and @xmath249 with @xmath17 , the configuration corresponds one - to - one to the adjacency matrix of an acyclic graph on @xmath154 vertices , with loops of weight @xmath16 and other edges of weight @xmath17 .it is common in this setting to think of a loop as an edge between two copies of the same node , an input node and an output node @xcite .the total weight of the edges , or the weight of the configuration , is given by @xmath250 the information stored in node @xmath62 is modeled by the random variable @xmath61 and the helper information from node @xmath62 to node @xmath64 by the random variable @xmath63 .the entropy of the random variables is bounded by @xmath251 and @xmath252 .the entropy of the message that a destination can recover from the configuration of nodes is bounded by the weight of the configuration .the minimal configurations give @xmath253 , the minimum taken over all @xmath254 .as observed in @xcite and developed in @xcite , variables @xmath255 , @xmath256 , in row @xmath257 of the configuration , all correspond to information from node @xmath257 , i.e to information from a source of bounded entropy @xmath251 . for @xmath258 large enough ,some nonzero mutual information among the @xmath255 , @xmath256 , is expected .this is captured by proposition 2 in @xcite .equation ( 18 ) in the next to last line of the proof of proposition 2 in @xcite states that the sum of the entropies @xmath259 , @xmath260 ( @xmath261 times ) and @xmath262 is stricly less than the trivial upper bound @xmath263 whenever @xmath264 .we use the claim with the singleton @xmath257 replaced by a set @xmath52 and adapt the proof .[ p0 ] let @xmath265 , @xmath266 , @xmath267 @xmath268 .@xmath269 combine equations ( [ p21 ] ) and ( [ p22 ] ) .@xmath270 for each @xmath271 , @xmath272 turning the proposition into an improved outer bound follows a standard procedure . in @xcite ,the improvement is applied , for given @xmath16 and @xmath17 , to the outer bound @xmath273 that is minimal , for the given @xmath16 and @xmath17 , among all @xmath25 , @xmath274 .this leaves open the possibility that the best overall outer bound for given @xmath16 and @xmath17 comes from improving a @xmath25 different from @xmath273 .for that reason we change the order and first collect a sequence of improved upper bounds and then address later which upper bound is optimal for which choice of @xmath16 and @xmath17 .other differences will be pointed out after the statement of the theorem and illustrated by an example . in the next thereom , the minimal configuration with @xmath244intact nodes and @xmath245 nodes being repaired refers to the configuration at the beginning of the section .the region of helper information @xmath63 is bounded by @xmath275 , @xmath276 .[ thm - lm ] let @xmath277 be a disjoint union of the helper information in a minimal configuration with @xmath244 intact nodes and @xmath245 nodes being repaired .for each @xmath278 , let @xmath279 , @xmath280 , @xmath281 .then @xmath282 let @xmath283 be a disjoint sum of the @xmath127 contributions to @xmath25 .@xmath284 fix a term @xmath278 and let @xmath285 , @xmath286 .@xmath287 @xmath288 [ e544 ] for the @xmath289 code , theorem [ thm - lm ] yields the following upper bounds .each upper bound is obtained as the average of the linear forms above it . below each upper boundis the range for @xmath290 where it is minimal among the given six upper bounds .@xmath291 \end{array } & \quad & \begin{array}{lll } { } \\ b_3 & { } \\[1.5ex ] \end{array } & \quad & \begin{array}{lll } b_2 \\ b_3 & b_2 ^ - \\[1.5ex ] \end{array } \\ \bar b \leq 4 \bar \alpha & & \bar b \leq 3 \bar \alpha + 1 & & \bar b \leq ( 7 \bar \alpha + 6 ) / 3 \\[1.5ex ] { \bar \alpha \in [ 0,1 ] } & & \bar \alpha \in [ 1,3/2 ] & & \bar \alpha \in [ 3/2,2 ] \\[3ex ] \begin{array}{lll } b_1 \\ b_3 & b_2^{- } \\b_3 & b_2^{- } & b_2^{- } \\[1.5ex ] \end{array } & \quad & \begin{array}{lll } b_1 \\ b_2 & b_1^{- } \\ b_3 & b_2^{- } & b_2^{- } \\[1.5ex ] \end{array } & \quad & \begin{array}{lll } b_1 \\ b_2 & b_1^{- } \\ b_2 & b_1^{- } & - \\[1.5ex ] \end{array } \\\bar b \leq ( 13 \bar \alpha + 14 ) / 6 & & \bar b \leq ( 11 \bar \alpha + 19 ) / 6 & & \bar b \leq ( 7 \bar \alpha + 22 ) / 5 \\[1.5ex ] \bar \alpha \in [ 2,5/2 ] & & \bar \alpha \in [ 5/2,37/13 ] & & \bar \alpha \in [ 37/13,4 ] \end{array}\ ] ] the functional repair outer bound @xmath292 attains its minimum in @xmath293 on the interval @xmath294 .we give a short proof that the exact repair outer bound is strictly less than the functional repair outer bound for all @xmath295 .let @xmath296 .we use theorem [ thm - lm ] to give a different proof of proposition [ p1 ] .[ p2 ] for @xmath215 , @xmath216 moreover , @xmath297 for @xmath215 , we apply the proposition with @xmath298 , @xmath299 : @xmath300 or with @xmath298 , @xmath301 : @xmath302 inequality ( 1 ) follows from @xmath303 and inequality ( 2 ) similarly from @xmath304 .together they proof @xmath305 on the interval @xmath306 now take the union of the intervals for @xmath215 .the next corollary follows from either proposition [ p1 ] in the previous section or from proposition [ p2 ] above .thus , methods in either section can be used to prove a non - vanishing gap between the functional - repair and exact - repair outer bounds . as a result , the exact repair capacity @xmath15 of an @xmath307 regenerating code satsfies @xmath308 on the interval @xmath309 .theorem [ thm - rs ] and theorem [ thm - lm ] give two different outer bounds for exact - repair regenerating codes . both are linear and are valid in the full range of @xmath310 . using the bounds with different choices for the parameters and then taking the minimum over all choices will produce a piece - wise linear upper bound for @xmath15 .we provide some details on using the bounds in combination for the cases @xmath311 and @xmath312 .figure 1 gives outer bounds for a @xmath234 code .they are presented as a trade - off between @xmath22 and @xmath23 and are based on theorem [ thm - lm ] . in this case , the graph for the trade - off shows the differences more clearly than the upper bound graph with @xmath313 as a function of @xmath310 .the lowest of the four outer bounds is the functional - repair outer bound @xmath314 next we apply the theorem with the choices of ( * ? ? ?* proposition 2 and theorem 1 ) , i.e. @xmath315 is a singleton and @xmath316 is fixed . next we allow choices with different @xmath316 and finally we allow @xmath52 of various sizes .we illustrate the improvements for @xmath317 this corresponds to the vertex @xmath318 in the functional - repair outer bound : @xmath319 , @xmath320 the various improvements are the following . in each casewe use theorem [ thm - lm ] . for @xmath321 and two choices with @xmath322 , as in (* proposition 2 and theorem 1 ) : @xmath323 for @xmath324 and choices @xmath325 ( @xmath326 ) and @xmath322 ( @xmath327 ) : @xmath328 for a choice with @xmath329 , let @xmath324 and @xmath330 ( @xmath326 ) and @xmath331 ( @xmath326 ) : @xmath332 or @xmath333 at @xmath334 , the gaps with the fractional - repair upper bound increase as @xmath335 .but none reaches the gap of @xmath336 that was found in example [ e1 ] , using theorem [ thm - rs ] : @xmath337 example [ e544 ] lists upper bounds for a @xmath289 code that are obtained with theorem [ thm - lm ] . in particular , @xmath338 here @xmath339 and @xmath340 .theorem [ thm - rs ] applied with @xmath341 and @xmath342 gives @xmath343 which yields no improvement to the bounds listed in example [ e544 ] .however , after exploiting nonzero mutual information among the @xmath194 , the theorem uses trivial estimates for each of the @xmath344 . in this case , together the @xmath195 contain five copies of @xmath345 , each with @xmath325 . using proposition [ p0 ]we can therefore improve the last bound to @xmath346 0r @xmath347 comparing ( [ b14 ] ) and ( [ b5 ] ) at @xmath348 ( @xmath349 ) , @xmath350 and ( [ b14 ] ) improves on the bounds in example [ e544 ] for @xmath351 korlakai vinayak rashmi , nihar b. shah , and p. vijay kumar .optimal exact - regenerating codes for distributed storage at the msr and mbr points via a product - matrix construction ., 57(8):52275239 , 2011 .nihar b. shah , k. v. rashmi , p. vijay kumar , and kannan ramchandran .regenerating codes for distributed storage networks . in _ arithmetic of finite fields_ , volume 6087 of _ lecture notes in comput ._ , pages 215223 .springer , berlin , 2010 .nihar b. shah , k. v. rashmi , p. vijay kumar , and kannan ramchandran . distributed storage codes with repair - by - transfer and nonachievability of interior points on the storage - bandwidth tradeoff ., 58(3):18371852 , 2012 .

<S> we address the open problem of establishing the rate region for exact - repair regenerating codes for given parameters @xmath0 . </S><S> tian determined the rate region for a @xmath1 code and found that it lies strictly within the functional - repair rate region . using different methods , sasidharan , senthoor and kumar </S><S> prove a non - vanishing gap between the functional - repair outer bound and the exact - repair outer bound for codes with @xmath2 . </S><S> our main results are two improved outer bounds for exact - repair regenerating codes . </S><S> they capture and then extend essential parts in the proofs by tian and by sasidharan , senthoor and kumar . </S><S> we show that the bounds can be combined for further improvements . </S>

0810.5693

the change of electric resistance upon varying magnetization direction is an old and well - known phenomenon@xcite with applications in spintronics.@xcite although the experimental observation of this anisotropic magnetoresistance ( amr ) is rather direct first accomplished as early as 1857 its theoretical understanding is far from being complete .it has long been clear that the phenomenon arises from the combined effects of magnetization and spin - orbit interaction .disregarding the crystalline anisotropic background , the magnetization - broken symmetry between two chosen directions and unequal resistivities along these has been described within different models . in transitionmetal ferromagnets , the anisotropy was ascribed to asymmetric scattering due to different parts of the spin - orbit interaction @xmath0 and the mechanisms considered were dubbed the @xmath1 model,@xcite @xmath2 model,@xcite or a combination of both.@xcite later , when computational power became sufficient for such task , ab initio calculations were performed@xcite for permalloy and reached a good agreement with experiments . however , no direct link between the ab initio and the model calculations listed above has been established , probably due to rather complex band structures involved . on the other hand ,such link between microscopic calculations and a simple physical model was recently found in the diluted magnetic semiconductor@xcite ( ga , mn)as whose band structure is much simpler . despite the long history of the amr research, the question has not been answered to date of how a rigorous transport formalism for anisotropic systems should be formulated . instead , the transport anisotropy has often been discussed only in terms of the asymmetry in scattering amplitudes between two states on the fermi surface .current availability of materials with relatively simple band structure motivates the quest for more precise theories of amr .the present article discusses how the semiclassical boltzmann equation should be solved in anisotropic systems , using an example of the model two - dimensional ( 2d ) electron system .this allows us to put the previous approximations on rigorous grounds .the conductivity of a given solid in the regime of linear response to the electric field @xmath3 can be evaluated within the semiclassical picture once we have found the distribution function satisfying the boltzmann equation . in the literature , this non - equilibrium distribution function is approximated in several ways .the relaxation time approximation ( rta ) relies on calculating the transport relaxation time @xmath4 from the scattering amplitudes @xmath5 between two states on the fermi surface using @xmath6\,.\ ] ] for isotropic systems , where @xmath7 depends only on the angle @xmath8 between @xmath9 and @xmath10 , the integral ( [ eq-15 ] ) does not depend on the direction of @xmath9 and the rta provides in fact the exact solution to the boltzmann equation@xcite .the scattering rate @xmath11 depends only on energy and it is constant on the whole fermi surface once the fermi energy is fixed . for anisotropic systems , keeping eq .( [ eq-15 ] ) in use produces @xmath11 that depends on the direction of @xmath9 .the non - equilibrium distribution function constructed utilising the rta and eq .( [ eq-15 ] ) may capture some aspects of the transport anisotropies but it is certainly not precise .this approximative approach underlies for example our previous calculations@xcite or those of mcguire and potter@xcite and we refer to it as to the `` @xmath11 approach '' .an improvement was proposed by schliemann and loss.@xcite in what we will call the `` @xmath12 approach '' , they use , according to their notation , eq .( [ eq-15 ] ) to calculate @xmath13 , and they provide an explicit formula for the non - equilibrium distribution in terms of this quantity and of @xmath14 in our article , we argue that in a general case the non - equilibrium distribution function can not be exactly calculated by just evaluating two integrals such as eqs .( [ eq-15],[eq-16 ] ) for each @xmath9-point of the fermi surface separately . instead, an integral equation must be solved that determines the whole non - equilibrium distribution at once . in section [ sec - ii ] , we describe this exact `` integral equation approach '' to transport in anisotropic 2d systems and then , in section [ sec - iii ] , we use a simple model system to illustrate how the procedure works . for this purposewe introduce the rashba hamiltonian combined with a scattering potential due to randomly distributed charged and ferromagnetically ordered impurities . in this model , the amr results from the spin - orbit coupled band structure and the broken time - reversal symmetry of the scattering potential.@xcite we explicitly calculate exact non - equilibrium distribution functions for several specific realizations of this model , starting from the ones with simple solutions and then proceeding to the more complex case . throughout section [ sec - iii ]we compare our distribution functions and amrs to results of the other two approximative approaches .section [ sec - iv ] concludes the main body of the article by discussing the relevance of our model calculations for the amr in magnetic semiconductors and by summarizing the key elements of the theoretical framework we have developed .the appendices contain details of our calculations and also a more thorough description of the @xmath11 and @xmath12 approaches .our central goal is to obtain the distribution function @xmath15 of a conductor displaced from equilibrium by a small bias represented by a weak homogeneous electric field @xmath16 .we start with the familiar form of the boltzmann equation in 2d @xmath17\ ] ] for a steady state of a spatially homogeneous system .derivation of this equation is described for instance in ref . .equation ( [ eq-01 ] ) is valid up to linear order in @xmath18 and it assumes small deviations of @xmath19 from the equilibrium distribution @xmath20 . the velocity @xmath21 is implied by the band dispersion @xmath22 , and @xmath23 is the charge of carriers .the scattering rate @xmath7 ( per unit area of the reciprocal space ) from the state @xmath9 to @xmath10 needs to be specified according to the microscopic origin of the scattering ; specific examples can be found in appendix [ sec - appa ] or in eq .( 38 ) of ref . .for now we only assume that the scattering is elastic , that is @xmath24 .focusing on the amr , we do not include anomalous terms@xcite like the coordinate shift related to the side jump in the anomalous hall effect into the right hand side of eq .( [ eq-01 ] ) .equation ( [ eq-01 ] ) can be applied to multi - band systems where @xmath9 is replaced by a compound index containing the wavevector and band index and the integral by integration over the wavevector and summation over bands .the solution to eq .( [ eq-01 ] ) is a function both of @xmath9 and @xmath16 .focusing first on the latter , we can write it as taylor series @xmath25 being interested only in the linear order of the electric field components @xmath26 , we will truncate the series after the third term .for the simplicity of notation , we will now assume that the band structure is isotropic , @xmath27 and @xmath28 .the anisotropy can still pervade into the transport via @xmath7 , for instance due to anisotropic scatterers .the more general boltzmann equation for anisotropic bands is treated in appendix [ sec - appg ] .we define two angles @xmath29 , @xmath30 as @xmath31 and @xmath32 and eq .( [ eq-02 ] ) becomes @xmath33 after the truncation , where @xmath34 and @xmath35 .the non - equilibrium distribution is now expressed in terms of two functions of @xmath29 which must , according to eq .( [ eq-01 ] ) with eq .( [ eq-03 ] ) inserted , fulfil @xmath36\cos\theta + \\ & & \label{eq-04 } \hskip-1 cm + \left[\bar{w}(\phi)b(\phi ) - \int d\phi'\ w(\phi,\phi')b(\phi ' ) \right]\sin\theta\,.\end{aligned}\ ] ] we define here @xmath37 , @xmath38 , and @xmath39 , where @xmath40 now includes the original transport scattering rate @xmath5 and also the density of states .the integral equation ( [ eq-04 ] ) with two variables @xmath41 can be decomposed into two independent integral equations @xmath42 \label{eq-05b } \sin\phi & = & \bar{w}(\phi)\ , b(\phi ) - \int d\phi ' w(\phi,\phi ' ) b(\phi')\end{aligned}\ ] ] whose solutions @xmath43 , @xmath44 inserted into eq .( [ eq-03 ] ) yield the exact solution of the boltzmann equation ( [ eq-01 ] ) up to the linear order in @xmath45 .the two decoupled inhomogeneous fredholm equations@xcite of the second kind ( [ eq-05a],[eq-05b ] ) can be most conveniently solved in terms of fourier series . for special choices of @xmath46 or band structure anisotropy ( see appendix [ sec - appg ] ) , the series may contain only few terms and reduce to an ansatz for @xmath47 such as eq .( 15 ) in ref . .we explain the general procedure how to solve eqs .( [ eq-05a],[eq-05b ] ) on three examples below .to illustrate how the above outlined procedure works , we choose the @xmath48 rashba hamiltonian@xcite in the basis of plane waves @xmath49 where @xmath50 are the pauli matrices , and @xmath51 is the rashba parameter .in addition to eq .( [ eq-06 ] ) , we consider scattering on dilute charged magnetic impurities@xcite described by the operator @xmath52 , @xmath53 that is impurities containing short range electric and ferromagnetically ordered magnetic potentials .the quantity @xmath54 is the ( dimensionless ) strength of the electric part , relative to the magnetic part , of the electro - magnetic scatterer whose magnetic moment was chosen to be along the @xmath55 direction .the magnitude @xmath56 and other aspects of this model are discussed in section [ sec - iv ] and appendices [ sec - appa ] and [ sec - appb ] .we now calculate the non - equilibrium boltzmann distribution function @xmath57 for this model in several special cases . to facilitate relevant comparison between the @xmath11 and @xmath12 approaches and the exact integral equation approach of section [ sec - ii ] , we calculate @xmath47 and evaluate the amr within the approximative approaches as well . the first special case of the model above concerns purely magnetic scatterers ( @xmath58 ) in the situation when the fermi energy cuts the spectrum of the rashba hamiltonian ( [ eq-06 ] ) precisely at the @xmath59 degeneracy point ( @xmath60 in fig . [ fig-02 ] ) .we further disregard this single point of the fermi surface and consider only the + band .this case offers the simplest way to explain the calculation of @xmath61 outlined in section [ sec - ii ] . the dimensionless scattering probability corresponding to @xmath52 of eq .( [ eq-07 ] ) is @xmath62\,.\ ] ] this result , including the dimensionful prefactor @xmath63 , is derived using the fermi golden rule in appendix [ sec - appa ] and [ sec - appb ] . although @xmath46 does not explicitly depend on the rashba parameter @xmath51 , the presence of the spin - orbit coupling , combined with the symmetry breaking scattering potential , has the crucial implication that @xmath7 depends on absolute values of angles @xmath29 and @xmath64 .this leads to the non - zero anisotropy of the magnetotransport , in contrast to the isotropic case in which @xmath46 depends only on the relative angle @xmath65 between the incoming and outgoing momenta . the total scattering probability , implied by eq .( [ eq-08 ] ) , reads @xmath66 note that despite the independence of @xmath67 on @xmath29 in the special case considered in this subsection the resulting relaxation times and conductivity are indeed anisotropic . ) .the three - dimensional plot shows energy dispersions ; its cross section along one ( arbitrary ) direction in the @xmath9-space is also shown .the spin textures for the pair of eigenstates @xmath68 and @xmath69 for each @xmath9 at the respective fermi surfaces with @xmath70 are shown on the top . unlike the diameters of the @xmath71 and @xmath72 fermi surfaces , the eigenstates ( spin textures ) are independent of the fermi energy @xmath73 , according to eq .( [ eq-35 ] ) . ]we will now look for the solution @xmath43 to eq .( [ eq-05a ] ) in the form of fourier series @xmath74 owing to the trivial form of @xmath46 ( of its fourier spectrum ) the integral in eq .( [ eq-05a ] ) can be readily calculated and eq .( [ eq-05a ] ) assumes the following form : @xmath75 the only non - zero coefficients in the fourier series ( [ eq-10 ] ) are therefore @xmath76 and @xmath77 .the solution of eq .( [ eq-05a ] ) then reads @xmath78 conservation of the number of particles requires @xmath76 to be zero . a completely analogous procedure applied to eq .( [ eq-05b ] ) yields a system of equations for coefficients @xmath79 , @xmath80 , @xmath81 which give @xmath82 the complete solution up to linear order in @xmath45 to the boltzmann equation ( [ eq-01 ] ) written using eq .( [ eq-03 ] ) is therefore @xmath83\,.\ ] ] let us now compare this result to the approximate approaches outlined in the introduction .the non - equilibrium distribution in the @xmath11 approach is ( see appendix [ sec - appc ] ) @xmath84\,,\end{aligned}\ ] ] while in the @xmath85 approach ( see appendix [ sec - appd ] ) , we obtain @xmath86\,.\end{aligned}\ ] ] distribution functions in eqs .( [ eq-14],[eq-17],[eq-18 ] ) are significantly different .to quantify the differences , we use these three distribution functions to calculate the amr , defined as @xmath87 and having the meaning of the ( relative ) difference in resistivity for current flowing parallel and perpendicular to the direction of the scatterer s magnetic moment , respectively .the conductivities are calculated from the current implied by the non - equilibrium distribution @xmath88 @xmath89 i.e. , as @xmath90 and @xmath91 where @xmath92 . the amr value of @xmath93 , obtained from the exact non - equilibrium distribution function in eq .( [ eq-14 ] ) , is markedly different from the results of the approximative approaches .the @xmath11 approach underestimates the amr by almost a factor of two ( @xmath94 ) , and the @xmath85 approach predicts even a wrong sign ( @xmath95 ) . before we proceed to comparing the three approaches on other realizations of our model disordered 2d system , let us make a remark about the distribution functions above . the non - equilibrium part of the distribution function in eq .( [ eq-17 ] ) was obtained as @xmath96 with @xmath97 and @xmath98 as derived in appendix [ sec - appc ] .analogous factorization of the bracket in eq .( [ eq-18 ] ) or eq .( [ eq-14 ] ) is not possible , reflecting the fact that no scalar relaxation time can be attributed to a given @xmath9state in these approaches .however , the @xmath99 approach still unambiguously assigns relaxation - rate - like quantities , a pair of ( not necessarily positive ) values @xmath100 , to each @xmath9-state , independent of the electric field direction ( determined by @xmath30 ; see appendix [ sec - appd ] ) .it remains an open question whether also the exact solution of the boltzmann equation , such as eq .( [ eq-14 ] ) , can be meaningfully interpreted in terms of @xmath30independent quantities related to scattering .we now extend results of the previous section by relaxing the condition @xmath58 , that is we consider the complete scatterer with electric and magnetic parts of its potential added up coherently , as defined by eq . ( [ eq-07 ] ) .the extension is straightforward although the algebra involved is richer than for the previous model .the dimensionless scattering probability @xmath101 and @xmath102 are @xmath103 \\[1 mm ] & & + \alpha(\sin\phi + \sin\phi ' ) \\[2 mm ] \bar{w}(\phi ) & = & \pi k(1+\alpha^2 + 2\alpha\sin\phi)\ , , \end{array}\ ] ] as shown in appendices [ sec - appa ] and [ sec - appb ] .note that @xmath102 is no longer constant , which is here the direct reason of the more complex algebra needed .we again look for the solution of eq .( [ eq-05a ] ) in the form of fourier series ( [ eq-10 ] ) and find that the higher order coefficients @xmath104 are now no longer zero . instead of eq .( [ eq-12 ] ) , we get a system of an infinite number of linear equations which is not surprising , given that eq .( [ eq-05a ] ) is an integral equation in its general form .this system of equations can be solved using a partitioning method , described in appendix [ sec - appe ] .herein , we segregate the variables into three groups : @xmath105 , @xmath106 , and @xmath107 .the first group must obey @xmath108 0 & = & ( 1+\alpha^2)a_{s1 } - \alpha a_{c2 } - \frac{1}{2}a_{s1}(1+\alpha^2)\ , , \end{array}\ ] ] and @xmath109 as in the previous subsection .equations ( [ eq-21 ] ) originate from comparing the coefficients in front of the @xmath110 and @xmath111 terms of eq .( [ eq-05a ] ) with eq .( [ eq-10 ] ) inserted .separate treatment of the other two infinite systems of equations , described in appendix [ sec - appe ] , yields @xmath112 together , eqs .( [ eq-21],[eq-22 ] ) form a closed system for @xmath77 , and @xmath113 which thus read @xmath114 the solution to eq .( [ eq-05a ] ) for @xmath115 is then @xmath116 for the evaluation of current and amr using eq .( [ eq-19 ] ) there is no need to know the higher order terms of @xmath47 ( by virtue of @xmath117 etc . ) . however , keeping _ all _ higher order terms of eq .( [ eq-10 ] ) in the derivation was necessary for obtaining the xscorrect form of eq .( [ eq-21 ] ) and also correct expressions for constants @xmath77 and @xmath113 at the end .we again repeat the same procedure for eq .( [ eq-05b ] ) , obtain @xmath44 , and finally we complete the calculation by writing down the non - equilibrium distribution function : @xmath118\end{aligned}\ ] ] for @xmath119 , while for @xmath120 the bracket is replaced by @xmath121\,.\ ] ] the dots symbolize @xmath122 , @xmath123 and higher order terms which as emphasised above do not contribute to the amr .the divergence of this expression for @xmath124 will be discussed in section [ sec - iv ] . evaluating the amr using the distribution function ( [ eq-24a],[eq-24b ] ) and eqs .( [ eq-57],[eq-19 ] ) amounts to comparing the coefficients in front of the @xmath125 and @xmath126 summands .we get @xmath127 we conclude the study of the single - band model by comparing this amr to the results of the approximate @xmath11 and @xmath12 approaches shown in fig .[ fig-01](a ) . while the @xmath11 approach can be regarded as only quantitatively inaccurate , as already suggested by the results of the previous subsection , the apparently more sophisticated @xmath128 approach yields remarkably large deviations from the exact amr .[ cols="^,^ " , ] we now consider the case when the fermi energy is above the @xmath59 degeneracy point of the rashba bands .let us first explicitly write down the scheme of sec .[ sec - ii ] for a two - band system . considering distribution functions of the + and @xmath72 bands , denoted by @xmath129 and @xmath130 , eq .( [ eq-01 ] ) is replaced by two coupled equations @xmath131\end{aligned}\ ] ] and @xmath132\,.\end{aligned}\ ] ] the scattering rate @xmath133 now also bears the indices of the initial ( @xmath134 ) and final ( @xmath61 ) band .we will abbreviate the equilibrium distributions @xmath135 by @xmath136 .assuming isotropic bands @xmath137 and @xmath138 , we seek a solution of eqs .( [ eq-26 ] ) in the form of @xmath139 \!\ !f_-(\phi,\theta)\ ! -\ ! f_0 \!=\ !-e{\cal e}v_- ( -\partial_\epsilon f_{0- } ) \big(a_-(\phi)\cos\theta + b_-(\phi)\sin\theta\big ) \end{array}\ ] ] and the four functions @xmath140 , @xmath141 must fulfil @xmath142\\ \cos\phi & = & \bar{w}_-(\phi)\ , a_-(\phi ) \\ & & \displaystyle \hskip-5 mm -\int d\phi ' \left[w_{--}(\phi,\phi ' ) a_-(\phi ' ) + w_{-+}(\phi,\phi ' ) a_+(\phi')\right ] \end{array } & \\[5 mm ] \label{eq-28b } \hskip-5mm&\begin{array}{rcl } \sin\phi & = & \bar{w}_+(\phi)\ , b_+(\phi ) \\ & & \displaystyle \hskip-5 mm -\int d\phi ' \left[w_{++}(\phi,\phi ' ) b_+(\phi ' ) + w_{+-}(\phi,\phi ' ) b_-(\phi')\right]\\ \sin\phi & = & \bar{w}_-(\phi)\ , b_-(\phi ) \\ & & \displaystyle \hskip-5 mm -\int d\phi ' \left[w_{--}(\phi,\phi ' ) b_-(\phi ' ) + w_{-+}(\phi,\phi ' ) b_+(\phi')\right ] \end{array } & \end{aligned}\ ] ] where @xmath143 and @xmath144 $ ] .note that eqs .( [ eq-28a ] ) are decoupled from eqs .( [ eq-28b ] ) .the dimensionless scattering probabilities for the complete electro - magnetic scattering operator given by eq .( [ eq-07 ] ) are @xmath145+\\ & & \hskip1.5 cm ( + /-) \alpha(\sin\phi + \sin\phi')\\[3 mm ] w_{-+}(\phi',\phi)/k & = & w_{+-}(\phi,\phi')/k = \\ & & \hskip-2 cm = \frac{1}{2}\big[1+\cos(\phi+\phi ' ) + \alpha^2\big(1-\cos(\phi-\phi')\big)\big]+\\ & & \hskip2.2 cm + \alpha(\sin\phi -\sin\phi ' ) \end{array}\ ] ] and @xmath146 . for simplicity, we assume that the constant @xmath63 ( and the density of states , as explained in appendix [ sec - appg ] ) is the same for both bands .this occurs in the rashba model ( [ eq-06 ] ) in the limit of @xmath147 and we will call this the @xmath148 limit . details of the derivation of eq .( [ eq-29 ] ) are given in appendix [ sec - appa ] and [ sec - appb ] .in a close analogy to the single - band case , equations ( [ eq-28a ] ) produce two coupled infinite sets of linear equations for coefficients of @xmath149 these may again be reduced to two coupled @xmath150 systems for variables @xmath151 using the partitioning method .mathematically , their solution @xmath152 leaves @xmath153 undetermined and , physically , particle number conservation again dictates that this constant is zero .terms in eq .( [ eq-30 ] ) containing higher multiples of @xmath29 are again not contributing to the current and to the amr but their coefficients can be evaluated within the partitioning procedure . applying the same procedure to eqs .( [ eq-28b ] ) leads to @xmath154 with @xmath155 the non - zero value of @xmath156 means that the scattering redistributes particles between the two bands . in another system , where the two bands would have different net spin polarization , such redistribution would correspond to the polarization of the particles by impurities .the overall particle number conservation nevertheless again requires @xmath157 .the two non - equilibrium distribution functions are now @xmath158 \end{aligned}\ ] ] for @xmath159 .the distribution function for @xmath120 is given by eq .( [ eq-31 ] ) with the term in the square brackets multiplied by @xmath160 .the explicit calculation outlined in appendix [ sec - appf ] shows that again all coefficients appearing in front of the cosine terms of eq .( [ eq-30 ] ) are non - zero .the infinite series , however , can be summed up and the complete exact non - equilibrium distributions fulfilling eqs .( [ eq-26 ] ) read @xmath161\end{aligned}\ ] ] for @xmath119 .the @xmath11 approach ( see appendix [ sec - appc ] ) leads to a similar but not identical approximative result @xmath162\,,\end{aligned}\ ] ] and the @xmath12 approach gives precisely the same result as eq .( [ eq-33 ] ) because @xmath163 vanishes in the two - band case ( see eq .( [ eq-45 ] ) in appendix [ sec - appd ] ) . remarkably , the amr calculated from the distribution functions of eqs .( [ eq-32 ] ) or ( [ eq-31 ] ) and of eq .( [ eq-33 ] ) , i.e. the exact and the two approximative results , comes out to be the same @xmath164 plot of this function is shown in fig .[ fig-01](b ) .for the two - band rashba model with small @xmath51 , we thus conclude that the discrepancy between the exact and approximative approaches remains only on the level of the complete non - equilibrium distributions ( in the higher order terms that do not contribute to the current ) .we speculate that the equal results for amr were not obtained by coincidence but because the @xmath148 system has a higher symmetry than the single - band model which is chiral .these symmetries are briefly commented in appendix [ sec - appd ] .let us start this discussion section with a remark on results shown in fig .[ fig-01](a ) .the amr takes on a singular value of @xmath165 at @xmath166 in all three approaches .this reflects the @xmath167 divergences of all non - equilibrium distribution functions ( see eq .( [ eq-24a ] ) for example ) .the origin of this divergence is as follows : the scattering operator @xmath52 in eq .( [ eq-07 ] ) with @xmath166 can annihilate one particular state @xmath68 on the fermi surface , as seen from eq .( [ eq-35 ] ) and the spin textures in fig .[ fig-02 ] .the state has its spin aligned parallel to the moment of the magnetic impurities , i.e. , along the @xmath168-axis . for the rashba modelthis implies that the @xmath9-vector of this state is parallel to the @xmath169-axis , more precisely @xmath170 . within the @xmath11 approach ,the fact that @xmath171 then implies that this state has an infinite ( transport ) relaxation time as dictated by eqs .( [ eq-37 ] ) and ( [ eq-40 ] ) of the appendix .consequent calculation focusing also on other states @xmath172 contributing to the current shows that this singularity is strong enough to produce @xmath173 when @xmath124 .the current ( and amr ) calculated for @xmath174 close to 1 are clearly inconsistent with the linear - response basis of our theory approach ( incorporated in eq .( [ eq-01 ] ) ) and are therefore not physically relevant . on the other hand ,the impurity operator ( [ eq-07 ] ) is idealised compared to realistic systems where the electric and magnetic part of @xmath52 will depend at least slightly differently on @xmath9 .this modification suffices to remove the singularity in conductivity . pointing our attention more towards experiments ,let us now discuss the relevance of the rashba model with dilute charged magnetic scatterers .our original motivation comes from the study of the diluted magnetic semiconductor@xcite ( ga , mn)as . mn atoms , when substituting for ga , introduce both the magnetic moments and holes to the material . the former via its @xmath175-electrons and the latter because their valence number is one less than that of ga .an electro - magnetic scatterer model as defined by eq .( [ eq-07 ] ) is therefore relevant to describe the mn atoms which constitute by far the most frequent source of scattering in ( ga , mn)as .indeed , it is possible to qualitatively explain trends for amr in ( ga , mn)as based on this model of scattering and by neglecting the exchange splitting of the ( ga , mn)as spin - orbit coupled valence band.@xcite the rashba model employed in this article provides arguably the simplest unpolarized spin - orbit coupled band structure in which the anisotropic scatterer mechanism fully determines the amr .the other two mechanisms , which are the anisotropy of the group velocity and the anisotropy of wavefunctions of the spin - split spin - orbit - coupled valence band and which only quantitatively modify the calculated amr in ( ga , mn)as , are completely absent in this model .the simplicity of the present model relies mostly in that it is 2d and it considers two rotationally - symmetric rather than six warped bands of ( ga , mn)as . the integral equation approach can be straightforwardly extended to ( ga , mn)as or other three - dimensional systems with more ( @xmath176 ) bands . however , the calculational complexity will be considerably higher ; the two functions @xmath177 of one variable will be replaced by @xmath178 functions of two variables ( two angles parametrizing the fermi surface in three dimensions ) .turning attention towards possible experiments , the calculations presented in this article are most relevant to asymmetric @xmath178-type heterostructures doped with magnetic donors.@xcite by changing the fermi level via doping , the effective strength @xmath54 of the electric part of the scatterer should change because the scattering amplitudes depend on the fermi wavevector which is a typical measure for involved momentum transfers.@xcite consequently , by polarizing the magnetic moments in - plane , the amr defined in eq .( [ eq-19 ] ) should be measurable and follow predictions shown in fig .[ fig-01](a ) .an alternative to doping by magnetic donors is to use an @xmath178-type heterostructure co - doped with magnetic impurities .experimental study of a iii - v or ii - vi heterostructure with dilute mn doping and heavy remote @xmath178-doping could be revealing . depending on the magnetic impurity character ( either acceptor or neutral ) , by varying the fermi level, we could again effectively change @xmath54 and/or interpolate between the single - band case ( @xmath60 ) and the two - band case ( @xmath179 ) .the challenge in this experiment would be to keep the scattering on mn the dominant ( or at least strong ) mechanism of relaxation . rather than these experimental suggestions , however , the main message of this paper should be of theoretical character .we have presented a framework to calculate exactly the conductivity in anisotropic systems within the semiclassical linear - response theory .this procedure was demonstrated on three simple and analytically solvable models .we found that in some special cases of high symmetry the previously employed approximate approaches may yield the same amr as our exact theory .in general , however , only the exact non - equilibrium solution to boltzmann equation of the form of an integral equation over the whole fermi surface , rather than of effective scattering rates at each individual @xmath9-point individually , provides a reliable account of the anisotropic transport .the work was funded through prmium academi and contracts number av0z10100521 , lc510 , kan400100652 , fon/06/e002 of , and kjb100100802 of of the czech republic , by onr through grant number onr - n000140610122 , by nsf under grant number dmr0547875 and by the namaste project ( fp7 grant no .it is our pleasure to thank maxim trushin for critical comments and for providing us some of his unpublished and copyrighted calculations , roman grill for fruitful discussions and vilm ha for his help with numerical checks of the presented results .we evaluate the scattering rates using the fermi golden rule .probability @xmath133 of transition between states @xmath180 and @xmath181 , induced by a perturbation described by time - independent operator @xmath52 , equals @xmath182 where @xmath183 is the energy of the final / initial state . considering many scatterers described by the operator @xmath52 distributed randomly with areal density @xmath184 ,the scattering rate per unit reciprocal space between the @xmath9 and @xmath10state equals @xmath185 within the lowest order of the born approximation ; contrary to the case of the anomalous hall effect@xcite , this order of the born approximation is sufficient for the calculation of the amr .note that the dimension of the scatterer strength @xmath56 is j@xmath186 , making @xmath187 dimensionless in the fourier space .finally , assuming isotropic parabolic dispersion @xmath188 , the density of states equals @xmath189 per spin , so that @xmath190 this is the definition of the dimensionful constant @xmath63 used in eq .( [ eq-08 ] ) and later on .its value determines the absolute value of conductivity but it cancels out in the definition of the amr , see eq .( [ eq-19 ] ) .we calculate the matrix elements of the scattering operator @xmath52 in eq .( [ eq-07 ] ) with respect to the basis @xmath191 where @xmath32 and @xmath192 is the system area .vectors @xmath68 and @xmath69 are the eigenstates of hamiltonian ( [ eq-06 ] ) with eigenvalues @xmath193 is illustrated in fig .[ fig-02 ] . the scattering operator @xmath52 in eq .( [ eq-07 ] ) is expressed in the basis of plane waves times spin up and spin down states .it does not depend on @xmath9 , @xmath10 so that it corresponds to short - range impurities ( @xmath194-scatterers ) . for @xmath195, this would be a non - magnetic charged impurity of strength @xmath196 , and for @xmath197 it is a purely magnetic impurity of strength @xmath56 . owing to the @xmath194-scatterer character of @xmath52 , the matrix elements of @xmath187 in the basis ( [ eq-35 ] ) depend on @xmath9 only through @xmath29 and not through @xmath198 .we take @xmath199 and get @xmath200\\ \langle \vec{k}'{-}|\alpha+\sigma_x|\vec{k}-\rangle & = & \frac{1}{2}\big [ ie^{i\phi } - ie^{-i\phi'}+\alpha(1+e^{i(\phi-\phi')})\big]\\ \langle \vec{k}'{+}|\alpha+\sigma_x|\vec{k}-\rangle & = & \frac{1}{2}\big [ ie^{i\phi } + ie^{-i\phi'}+\alpha(1-e^{i(\phi-\phi')})\big]\\ & = & \overline{\langle \vec{k}{-}|\alpha+\sigma_x|\vec{k}'+\rangle}\,.\end{aligned}\ ] ] taking the absolute values squared leads using eq .( [ eq-38 ] ) to eq .( [ eq-29 ] ) , to eq .( [ eq-20 ] ) ( the ++ element ) , and to eq .( [ eq-08 ] ) ( ++ element with @xmath58 ). non - equilibrium distribution function in an isotropic ( @xmath27 , and isotropic scatterer ) two - band system can be shown to be @xmath201 where the relaxation times for @xmath71 and @xmath72 bands may depend on @xmath9 only through energy @xmath202 .this fact , that for fixed energy the relaxation time as defined in eq .( [ eq-15 ] ) is constant , is a direct consequence of the scatterer isotropy @xmath203 . for clarity, we stress that in an @xmath178band system there are the total of @xmath204 scattering rates between pairs of bands , @xmath205\,,\ ] ] that combine into @xmath178 scattering times @xmath206 , one for each band , according to the matthiessen srule@xcite @xmath207 we note that @xmath208 measures the angle between @xmath209 and @xmath210 but given the isotropy of the band structure , @xmath211 and @xmath9 are parallel so that @xmath212 . equation ( [ eq-15 ] ) is a single - band variant of eq .( [ eq-40 ] ) for isotropic systems where @xmath213 drops out .in the @xmath11 approach , we simply evaluate eq .( [ eq-40 ] ) for @xmath214 and obtain @xmath9dependent @xmath11 .this is then inserted into the distribution function ( [ eq-39 ] ) , losing thereby its property @xmath215 . for the rashba model with @xmath148we get using eq .( [ eq-38 ] ) and appendix [ sec - appb ] the following @xmath216 the scattering rates in the two - band model are thus as simple as @xmath217 this result , plugged into eq .( [ eq-39 ] ) , produces the non - equilibrium distribution function ( [ eq-33 ] ) within the @xmath11 approach .the prescription for the non - equilibrium distribution function suggested by schliemann and loss@xcite can be summarized as follows : ( a ) evaluate the standard formulae ( [ eq-41],[eq-40 ] ) and denote the result as @xmath219 ; ( b ) calculate @xmath220 using formulae identical to eqs .( [ eq-40],[eq-41 ] ) save the replacement of the bracket in eq .( [ eq-40 ] ) by @xmath221 ; ( c ) write down the distribution function as @xmath222\,.\end{aligned}\ ] ] several remarks are in order .( i ) whenever @xmath223 vanishes , eq .( [ eq-44 ] ) simplifies to eq .( [ eq-39 ] ) of the @xmath11 approach .( ii ) this @xmath12 approach is suitable for the description of isotropic scatterers ( the amplitude depends only on the angle between @xmath9 and @xmath10 , the incoming and outgoing wave ) which however may exhibit an asymmetry ( or better chirality ) , i.e. scatter more clockwise than counterclockwise such as it is the case with skew scattering in the anomalous hall effect .( iii ) contrary to the statement of ref . , the distribution function ( [ eq-44 ] ) is not the exact solution to eq .( [ eq-26 ] ) for a general anisotropic system .the derivation of eq .( [ eq-44 ] ) presented in ref .is only valid if the expressions @xmath224 given by eq .( 27,28 ) of that reference are constant for each band ( i.e. @xmath29independent in our case ) .the most general distribution function this approach can therefore correctly capture must have the form @xmath225 while as the examples in section [ sec - ii ] show , the non - equilibrium distribution can have finer details than those of period @xmath226 in the angular variable @xmath29 ( and these details , when completely neglected , may even lead to wrong values of the constants @xmath227 , @xmath228 above ) .this original neglect of ref .was later corrected by one of its authors@xcite in the context of the specific hamiltonian considered.@xcite however , a general procedure for exact solution of the boltzmann equation was not given . in our specific model , as described by the scattering matrix elements of appendix [ sec - appb ] , we get @xmath229^{-1 } & = & k\pi\cos\phi\big((+/-)\alpha+\sin\phi\big ) \\[0 mm ] [ \tau^\perp_{(+-/-+)}(\phi)]^{-1}&= & k\pi\cos\phi\big((+/-)\alpha-\sin\phi\big)\ , . \end{array}\ ] ] for the two - band model , @xmath230 so that the @xmath12 approach reduces to the @xmath11 approach in line with the comment after eq .( [ eq-44 ] ) .the single - band case , however , has a finite @xmath231 so that the two approaches give different results .this is not surprising , since each rashba band has a chiral spin texture but both of them together form a non - chiral pair , provided they have both the same fermi @xmath232 ( as it happens for @xmath148 ) , see fig .[ fig-02 ] .the asymmetry of scattering expressed by @xmath163 thus vanishes in our two - band model . to obtain the non - equilibrium distribution ( [ eq-18 ] ) within the @xmath12 approach, we have to take @xmath233 of eq .( [ eq-43 ] ) , @xmath234 of eq .( [ eq-46 ] ) , insert them into eq .( [ eq-44 ] ) and expand @xmath235 , @xmath236 in terms of @xmath237 and @xmath238 .the actual infinite system of linear equations for variables @xmath239 appropriate for the _ single - band model _ assumes a structure suitable for partitioning if we perform the coordinate transformation @xmath240 .we will now solve the integral equation ( [ eq-05a ] ) using this coordinate ( and use @xmath241 throughout appendices [ sec - appe ] and [ sec - appf ] ) and transform the result back before we use it in eq .( [ eq-21 ] ) .the equation to be solved is now @xmath242 with @xmath243 + \\ & & + \alpha(\cos\ihp + \cos\ihp ' ) \\\bar{w}(\ihp)&= & \pi k ( 1+\alpha^2 + 2\alpha\cos\ihp)\,.\end{aligned}\ ] ] inserting @xmath244 into eq .( [ eq-53 ] ) , and comparing the coefficients at the constant , @xmath245 , @xmath246 , @xmath247 , @xmath248 , @xmath249 , @xmath250 , @xmath249 terms , we obtain the following infinite system of linear equations for @xmath251 ( in this order ) : @xmath252 * & * & * & \aa & 0 & 0 & \ldots & 0 & 0 & 0 & \ldots & * \\[0 cm ] * & * & * & 0 & 0 & 0 & \ldots & \aa & 0 & 0 & \ldots & * \\ \hline 0 & \aa & 0 & \!1+\aa^2\!\!\!&\aa & 0 & \ldots & 0 & 0 & 0 & \ldots & 0 \\ 0 & 0 & 0 & \aa & \!\!\!1+\aa^2\!\!\!&\aa & & 0 & 0 & 0 & \ldots & 0 \\ 0 & 0 & 0 & 0 & \aa & \!\!\!1+\aa^2\!\!\ ! & & 0 & 0 & 0 & \ldots & 0 \\ \vdots & & & \vdots & & & \ddots & \vdots & & & \vdots & \vdots \\\hline 0 & 0 & \aa & 0 & 0 & 0 & \!\ldots & 1+\aa^2 & \aa & 0&\ldots & 0\\[-2 mm ] 0 & 0 & 0 & 0 & 0 & 0 & \ldots & \aa & \ddots & & \vdots & 0 \\ \vdots & & & & & & & & & & \vdots & \vdots \end{array}\right)\ ] ] the double line separates the left and right - hand side of the equations .the twelve asterisks in the first three lines of the system ( [ eq-47 ] ) correspond to the @xmath150 system ( [ eq-21 ] ) , and the value of these coefficients will be unimportant within this appendix . it is apparent that the system ( [ eq-47 ] ) is almost block - diagonal .the partitioning method takes advantage of this structure and aims at solving three independent systems corresponding to groups @xmath253 , @xmath254 , and @xmath255 of the original variables .the basic idea is to treat the only non - zero element of the off - diagonal block as a right - hand - side term . in explicite terms , we rewrite for example the fourth and fifth equations of the system ( [ eq-47 ] ) @xmath256\cos 2\ihp & = & 0\\[0 cm ] [ \alpha\atr_{c2}+(1+\alpha^2)\atr_{c3}+\alpha\atr_{c4}]\cos 3\ihp & = & 0\end{aligned}\ ] ] as @xmath257 \alpha\atr_{c2}+(1+\alpha^2)\atr_{c3}+\alpha\atr_{c4 } & = & 0\,.\end{aligned}\ ] ] the system of all cosine - term equations of the system ( [ eq-47 ] ) ( starting with @xmath247 ) can now be solved as a function of @xmath258 .in other words , we are treating the central block of the matrix ( [ eq-47 ] ) .the still - infinite system to be solved is @xmath259 for the purposes of solving later the @xmath150 system ( [ eq-21 ] ) , we in fact need to know only a part of the solution , namely @xmath260 .to this end , linear algebra gives us a very quick answer .if we denote by @xmath261 the determinant of the infinite matrix left from the double line in ( [ eq-48 ] ) , and by @xmath262 the determinant of the analogous @xmath263 matrix , then if the system ( [ eq-48 ] ) were finite , @xmath264 where the numerator equals the determinant of the @xmath263 matrix left from the double line of ( [ eq-48 ] ) with first column replaced by the column right from the double line . considering @xmath265 , we immediatelly ( after transformation @xmath266 ) get @xmath267 as given in the first line of eq .( [ eq-22 ] ) .this answer is , however , not completely correct .the caveat of this procedure is that we should have been careful about taking the limit @xmath268 .it turns out that the limit is finite only for @xmath119 and then @xmath269 so that only in this case @xmath270 which is obviously equal to one .the determinant @xmath261 is infinite for @xmath120 and only @xmath271 remains finite , namely equal to @xmath160 as one can readily see from the explicit formula @xmath272 in conclusion , we find @xmath273 and the transformation back from @xmath241 to @xmath29 implies @xmath266 and @xmath274 .literally the same procedure works for the sine - term equations of the system ( [ eq-47 ] ) , i.e. the lower - right block .the only difference is now that @xmath275 and we use @xmath276 , @xmath277 .these two results , @xmath278 with the corresponding definitions of @xmath258 are summarized as eq .( [ eq-22 ] ) .the key feature needed for this partitioning method is that @xmath258 is a function only of @xmath279 ( @xmath280 ) and not of higher - order coefficients like @xmath281 . in this way , the @xmath150 system of equations ( [ eq-21 ] ) becomes closed after @xmath282 and @xmath283 have been inserted .finally , we stress , that if the coupling between the three subsystems had been neglected from the very beginning this amounts to setting to zero the four elements in the off - diagonal blocks in the system ( [ eq-47 ] ) the solution of the @xmath150 subsystem represented by the asterisks would have been different . in this way , even though @xmath122 and other higher terms do not contribute to the current calculated from the non - equilibrium distribution ( [ eq-03 ] ) , their complete neglect from the beginning may produce wrong coefficients in the @xmath110 and @xmath111 terms . in the case of two bands , we obtain two infinite systems of linear equations identical to the system ( [ eq-47 ] ) , one for variables with + index , another for those with @xmath72 index , see eq .( [ eq-30 ] ) .although the two systems are now coupled , the direct coupling exists only via variables @xmath284 , @xmath285 , @xmath286 corresponding to the upper left block .the partitioning method can therefore be independently carried out in the @xmath71 and @xmath72 sector .for all four infinite subsystems , of which the system ( [ eq-48 ] ) is one , we obtain the almost the same result @xmath287 for @xmath288 and with appropriate definition of @xmath258 for each subsystem , while @xmath289 obey eq .( [ eq-50 ] ) with @xmath290 replaced by @xmath291 .all coefficients in the series ( [ eq-30 ] ) are thus non - zero .nevertheless , eq . ( [ eq-30 ] ) can still be summed up using @xmath292 and a similar formula for sines .we now transform back from @xmath241 to @xmath29 , use @xmath293 for the cosine @xmath294 parts of eq .( [ eq-30 ] ) and @xmath295 for its sine parts , transform back @xmath296 , @xmath297 and finally get @xmath298 plugging the values of @xmath299 from eq .( [ eq-52 ] ) into eq .( [ eq-51 ] ) , repeating an analogous procedure for the @xmath300 s in eq .( [ eq-28b ] ) and inserting the results into eq .( [ eq-27 ] ) , we arrive at eq .( [ eq-32 ] ) .results of section [ sec - ii ] were derived for a special class of 2d systems where the band structure remains isotropic and the anisotropy is only introduced through the scatterer and the scattering rate @xmath5 . the results of eqs .( [ eq-05a],[eq-05b ] ) for single - band or of eqs .( [ eq-28a],[eq-28b ] ) for two - band system have to be slightly modified for anisotropic 2d band structure .the wavevectors @xmath301 of eq .( [ eq-01 ] ) or eqs .( [ eq-26 ] ) will still be bound to the fermi level @xmath73 but their magnitude now depends on @xmath29 .that is , we have @xmath302 and @xmath303 .we also tacitly assume that in each band and for each @xmath304 there is only one solution @xmath232 to @xmath305 .the calculation of @xmath46 , compared to what is done in appendix [ sec - appa ] , becomes @xmath306 \nonumber\end{aligned}\ ] ] the last expression should be understood as a function of @xmath29 , @xmath64 only ; the derivative and @xmath307 are to be taken at the fermi level , so that e.g. @xmath308 and @xmath309 .further , the expression @xmath310 in eq .( [ eq-01 ] ) is no longer simply @xmath311 .first of all , @xmath312 and moreover @xmath313 need not be parallel with @xmath9 . formally, we could replace @xmath314 in eq .( [ eq-04 ] ) by @xmath315 with @xmath316 defined by @xmath317 .single - band equations ( [ eq-05a],[eq-05b ] ) should be replaced by @xmath318 these two equations for @xmath43 and @xmath44 are still completely decoupled . with some luck, @xmath319 can be reasonably expanded in terms of cosines and sines of @xmath29 and higher multiples of @xmath29 but the @xmath320 and @xmath321 terms will most likely make an analytical solution of eq .( [ eq-55a ] ) impossible for realistic anisotropic fermi surfaces .the solution is , however , not difficult to obtain by numerical means .after discretization of the angular variable @xmath29 into @xmath178 steps , eq .( [ eq-55a ] ) constitutes an @xmath263 system of linear equations .once @xmath322 are known , the non - equilibrium distribution function is readily written as @xmath323\,.\end{aligned}\ ] ] note that the spectral function @xmath324 depends now both on @xmath232 and @xmath29 . a rather straightforward generalization of eq . ( [ eq-56 ] ) andthe appropriate pair of integral equations ( [ eq-55a],[eq-55b ] ) to multiband systems is possible .for instance , the analogy of the two coupled eqs .( [ eq-28a ] ) for anisotropic band structure reads @xmath325\\[2 mm ] \cos\xi_-(\phi ) & = & \bar{w}_-(\phi)\ , a_-(\phi ) - \int d\phi ' \times\\ & & \displaystyle \hskip-20 mm \times \left[\frac{v_-(\phi')}{v_-(\phi)}w_{--}(\phi,\phi ' ) a_-(\phi ' ) + \frac{v_+(\phi')}{v_-(\phi)}w_{-+}(\phi,\phi ' ) a_+(\phi')\right]\end{aligned}\ ] ] where @xmath326 are the fermi velocities of the two bands and the four quantities @xmath327 have to be calculated in the spirit of eq .( [ eq-54 ] ) .

<S> we present a procedure for finding the exact solution to the linear - response boltzmann equation for two - dimensional anisotropic systems and demonstrate it on examples of non - crystalline anisotropic magnetoresistance in a system with spin - orbit interaction . </S><S> we show that two decoupled integral equations must be solved in order to find the non - equilibrium distribution function up to linear order in the applied electric field . </S><S> the examples are all based on the rashba system with charged magnetic scatterers , a system where the non - equilibrium distribution function and anisotropic magnetoresistance can be evaluated analytically . </S><S> exact results are compared to earlier widely - used approximative approaches . </S><S> we find circumstances under which approximative approaches may become unreliable even on a qualitative level . </S>

1003.5555

time explicit schemes are prone to the courant - friedriech - lewy ( cfl ) stability condition which constrains the maximum time step one can use in numerical calculations .however , in some stellar physics problems , the physical solution evolves on time - scales which are much longer than this `` numerical '' time - scale .this often makes calculations too cumbersome to be tractable .it is then necessary to use an implicit scheme , which does not have such a limitation on the time step .this paper describes such an implicit hydrodynamic code .the first scientific problem to be adressed with the code is the role of convection in pulsating stars .this requires an implicit approach since the typical cfl time step ( for a moderate resolution ) is @xmath0 s , whereas the convection ( eddy turnover ) time - scale is @xmath1 s ( @xmath2 ) and typical pulsation periods are of the order of a few days ( @xmath3 s ) with a growth rate of the order of @xmath4 s.we solve numerically the standard hydrodynamic conservation equations with radiative transfer .we use the internal energy density equation and we treat radiation transport in the diffusion limit . this approximation is suitable for optically thick region but becomes inaccurate in the photosphere and optically thin regions . molecular viscosity is negligible in astrophysical flow , but we use artificial viscosity when robustness might be an issue , e.g. for the initial relaxation of stellar models or for shocks .the hydrodynamic equations are closed with an equation of state and opacities appropriate for the description of stellar interior structure .we use spherical coordinates and assume azimuthal symmetry , leaving @xmath5 as independent coordinates .the spatial discretization is done on a staggered grid , using a finite volume approach .fluxes are computed at cell interfaces with an upwind reconstruction scheme : we use either the donor cell method ( first order in space ) or the van leer method ( second order in space ) . for the implicit temporal discretization ,we use either the backward euler scheme ( first order in time ) or the crank - nicholson scheme ( second order in time ) .an implicit discretization results in a non - linear system which is solved by newton - raphson iterations .this requires the inversion of a linear system involving the jacobian matrix of the set of equations . here the jacobian matrix is computed by numerical differencing and we use a direct solver , mumps ( amestoy et al .2001 ) , which implements an efficient and robust lu decomposition .the variable time step is chosen on empirical ground , depending on the amount of changes of variables during a time step , as described for example in dorfi ( 1997 ) .for test purpose , we have also implemented in the code cartesian geometry ( 2d ) and the equation of state of a perfect gas with a constant adiabatic index @xmath6 .basic tests ( e.g. sedov , sod , barenblatt ) have been successfully done for code validation during its first stages of development , so we focus here on two tests more relevant to stellar physics problems such as convection or oscillations ..run parameters for the entropy bubble test [ cols="^,^,^,^,^",options="header " , ]we have presented a new multi - d hydrodynamic implicit code . as shown in section [ bubble ] ,high cfl simulations are efficient to study physical processes proceeding on time - scale larger than the cfl limit .one should always consider numerical results with care , as an inappropriate time step can give significantly wrong results .we have shown that the code is able to handle strong convection , at least in a simple setup .our first results on a realistic ( weakly convective ) stellar model are encouraging and provide very promising perspectives for the first multi - d simulations of convection in pulsating star , a challenge in the field of asteroseismology .p. r. amestoy , i. s. duff , j. koster and j .- y .lexcellent , simax , vol 23 , 1 , 15 , 2001 a. brandenburg , k.l .chan , a. nordlund , & r.f .stein , astron .nachr . 8 , 681 , 2005 b. dintrans & a. brandenburg , a&a , 421 , 775 , 2004 e.a .dorfi in r.j .leveque , d. mihalas , e.a .dorfi , e. muller , saas - fee advanced course 27 , swiss society for astrophysics and astronomy , 1997

<S> we report on the development of an implicit multi - d hydrodynamic code for stellar evolution . </S><S> we present two test - cases relevant for the first scientific goal of the code : the simulation of convection in pulsating stars . </S><S> first results on a realistic stellar model are also presented . </S>

astro-ph0401141

mass accretion onto a white dwarf , which may occur in a close binary system , plays an important role in its evolution .if the accretion rate is sufficiently high , heating of the outer layers of the accreting star leads to the ignition of nuclear burning . a c - o white dwarf accretes co - rich matter if the companion is also a c - o white dwarf ( i.e. , double c - o white dwarf system ) . whether the carbon is ignited in the stellar center or in the envelope of an accreting c - o white dwarfis crucial for the star s fate .if the ignition occurs at the center , it leads to a sn ia explosion .if carbon burning is ignited in the envelope , on the other hand , the flame propagates inward to the center converting the c - o white dwarf to an o - ne - mg white dwarf @xcite that is expected to eventually collapse to a neutron star @xcite . @xcite and@xcite have shown that carbon - ignition occurs in an outer layer if the c - o white dwarf accretes matter more rapidly than @xmath0 .when a double c - o white - dwarf system coalesces due to angular momentum loss by gravitational wave radiation and/or by a magnetic field , the less massive white dwarf is disrupted to become a thick disk around the more massive white dwarf @xcite .if the thick disk is turbulent , which is likely , the timescale of viscous angular momentum transport is very short @xcite so that the more massive white dwarf is expected to accrete co - rich matter at a rate close to the eddington limit from the thick disk .therefore , when a double c - o white - dwarf system coalesces , carbon burning is ignited in the envelope of the more massive white dwarf to transform it to an o - ne - mg white dwarf , thus preventing the occurrence of the sn ia explosion postulated by @xcite and @xcite . in the previous investigations on the carbon ignition , however , the effect of rotation was neglected despite the fact that the accreted matter should bring angular momentum to the white dwarf .it is important to see whether the above conclusion on the fate of a double white dwarf merger is unchanged when the effect of rotation is included .recently , @xcite investigated the effect of rotation on the evolution of white dwarfs that accrete co - rich matter .they assumed that all the angular momentum associated with the accreted matter was brought into the white dwarf .they claim that the combined effects of accretion and rotation induce expansion to make the surface zone gravitationally unbound and hence suppresses further accretion in the double white dwarf merger .they argued that the above effect makes the accretion rate smaller than the critical value for the occurrence of the off - center carbon ignition , and hence the white dwarf can grow up to the chandrasekhar mass to become a sn ia .in addition to the above assumption @xcite took into account the angular momentum loss via gravitational wave radiation , and obtained an evolutionary model leading to a central carbon ignition .in contrast to their assumption , @xcite and @xcite had shown that when the surface rotation rate is nearly critical , angular momentum is transported from the star back to the accretion disk ( see also @xcite ) and the star can continue to accrete matter as long as matter exists around the star .this indicates that we should let matter accrete onto the white dwarf keeping the surface rotation velocity nearly critical .those results were obtained by considering a central star at contact with a thin disk , in contrast of a thick disk expected in the case of a merging two white dwarfs .although the physics of thick disk accretion is poorly known , we expect that the thick disk is turbulent and the angular momentum would be exchanged as fast as in a thin disk .therefore , it is likely that the results of @xcite and @xcite are applicable also to thick disk accretion .based on this consideration , we assume that even after the accreting star is spun up close to the keplerian velocity at the surface , the star accretes matter at the initial rate while the surface velocity remains equal to the keplerian velocity . we note that under this assumption the total angular momentum of the accreting white dwarf can even decrease .recently , @xcite computed white - dwarf models accreting co - rich matter , taking into account the effect of rotation .the accretion rates they considered are lower than those adopted in our calculations .their assumption at the surface of the accreting white dwarf is somewhat different from ours .they assumed that the matter accretes onto the white dwarf without bringing angular momentum when the rotation velocity at the surface of the white dwarf exceeds the keplerian velocity . under their assumptionthe total angular momentum of the white dwarf never decreases and is higher than that of our models for a given mass .the initial models , carbon - oxygen white dwarfs of @xmath1 and @xmath2 , were obtained from zero - age helium main - sequence models by artificially suppressing nuclear reactions and converting chemical compositions to @xmath3 , where @xmath4 denotes the mass fraction of the elements heavier than oxygen , and by letting cool until the central temperature became @xmath5k .table [ tab_initialmodel ] lists physical characteristics of the initial models .adopted input physics are mostly the same as those used in saio & nomoto(1998 ) : opal opacity tables @xcite , neutrino emission rates by @xcite , and carbon - burning reaction rates by @xcite .keeping our evolutionary models spherical symmetric , we have included the effect of rotation only in the equation of hydrostatic equilibrium as @xmath6 where @xmath7 is the angular velocity of rotation .the factor @xmath8 comes from taking the mean of the radial component of the centrifugal force on a sphere @xcite .spatial and time variation of @xmath7 is determined by solving the conservation equation of angular momentum @xmath9_{m_r } = { \partial\over\partial m_r } \left[(4\pi\rho r^3)^2\nu{\partial\omega\over\partial m_r}\right],\ ] ] where @xmath10 stands for turbulent viscosity .the outer boundary condition @xmath11 was imposed at the photosphere , where @xmath12 is the keplerian angular velocity at the stellar radius @xmath13 .we have neglected the rotation of the initial models .we have adopted two kinds of formulas for the turbulent viscosity ; one from @xcite and another from @xcite .the turbulent viscosity formulated by @xcite includes the effect of baroclinic instability as well as shear instability .we will refer this case as case a ; the viscosity is given by @xmath14 where @xmath15 @xmath16 @xmath17 with @xmath18 being brunt - visl frequency defined as @xmath19 and @xmath20 the pressure scale height .for the second case , which is referred to as case b , we have included turbulent viscosity due to vertical shear instability formulated by @xcite but neglected the effect of meridional circulation ; we have used the following equation , @xmath21 where @xmath22 is the thermal diffusivity defined by @xmath23 .as we will see in the next section , the viscosity in case a is so efficient in transporting the angular momentum that the rotation becomes nearly uniform in the white - dwarf interior . in case b ,on the other hand , the angular momentum is slowly transported into the deep interior .we assume that the two models for the transport of angular momentum bracket what occurs in a real star .figure [ mass_radius ] shows the evolution of radius as a function of the total mass for c - o white - dwarfs that accrete co - rich matter at rates of @xmath24 and @xmath25y@xmath26 from the beginning of accretion to the off - center carbon ignition .the initial mass of the models is @xmath2 .wiggles seen in this figure are numerical artifacts .the dotted lines show the evolution of non - rotating models , the solid lines for rotating models in case a , and the dashed lines for rotating models in case b. in the early phase of accretion the radius of the white dwarf increases rapidly .this is caused by the fact that its envelope is heated by the accretion , and the centrifugal force of rotation lifts the envelope . the initial rapid expansion in case a models slows down when the angular momentum begins to be transferred back to the accretion disk ( fig.[ev_angmom ] below ) .such transition to a slower evolution did not occur in the models of @xcite , because they did not consider the possibility of transferring the angular momentum back to the disk , but did reduce the accretion rate . in later phase of accretion ,the radius begins to decrease .the decrease in radius is caused by the increase of the white - dwarf mass . during this phasethe total angular momentum is nearly constant or increases gradually .the lifting effect of rotation makes the radius of a rotating model larger than that of the non - rotating model with the same mass and the accretion rate .a short vertical increase in radius at the end of each evolution track indicates the occurrence of the off - center carbon ignition . for a given accretion rate ,the off - center carbon ignition occurs later in a rotating model than in the non - rotating model ; i.e. , the mass of the rotating white dwarf at the carbon ignition is larger than that for the corresponding non - rotating model .figure [ mr_t ] shows the temperature versus @xmath27 in the interior of models close to the off - center carbon ignition .these models have evolved from the initial model of @xmath2 with an accretion rate of @xmath28y@xmath26 .the solid and dashed lines are for rotating models , and the dotted line for the non - rotating model .the outer layers consisting of accreted matter ( @xmath29 ) have high temperatures heated by the gravitational energy release .the heat is conducted into the layers of the original white dwarf .the heat - conduction front , however , has not advanced much during the time between the start of accretion and the off - center ignition ( @xmath30y ) .rotating models have broader temperature peaks than the non - rotating model , because they accrete more mass and hence the time from the start of accretion to the carbon ignition is longer .the core temperature has increased from the initial value @xmath31 due to compression .table [ tab_sum ] lists the total mass @xmath32 and the central density @xmath33 of the model in which off - center carbon - ignition occurs , and the mass interior to the ignition layer @xmath34 for each case computed , where the values of masses are given in units of @xmath35 , @xmath36 means accretion rate in units of @xmath35y@xmath26 , and @xmath37 stands for the initial mass . the difference in @xmath32 between a rotating and the non - rotating model for a given accretion rate can be as large as @xmath38 .the value of @xmath34 is also larger in the rotating model .rotation delays the carbon ignition because the centrifugal force reduces the effective gravity to lift the outer layers and reduce the temperature there @xcite .the lifting effect of rotation appears also in the central density , @xmath39 , which is lower in the rotating model than in the non - rotating model with a similar mass .the case a models with @xmath40y@xmath26 did not ignite carbon .they began to expand before the maximum temperature in the outer layer reaches to the ignition temperature .table [ tab_sum ] shows that the results hardly depend on the initial mass of the white dwarf .the results confirm essentially the conclusion of the previous work done without including the effect of rotation that carbon is ignited in the outer layer of a c - o white dwarf if the accretion rate is fast enough .figure [ rel_omega ] shows runs of the ratio of angular velocity to its local critical value , @xmath41 , in the models at evolutionary stages around the carbon ignition for @xmath42 , @xmath43 and @xmath44 .( for the case a of @xmath45 , @xmath41 in the last model computed is shown . ) here @xmath46 is the local keplerian velocity . in all cases, @xmath41 increases rapidly toward the surface in the very superficial layer , which is due to a rapid increase in the radius of mass shell toward the surface . in casea the viscous transport of angular momentum is very efficient so that the rotation becomes nearly uniform ( figs .[ ang_freq_evol_3em6],[ang_freq_evol ] ) .therefore , the solid curves in figure [ rel_omega ] are almost parallel to the runs of @xmath47 .an exception is the outermost layer of the case a model for @xmath45 .since the radius of the model is rapidly increasing and hence the surface angular velocity (= keplerian velocity ) is rapidly decreasing , there is a steep outward decrease in the angular velocity of rotation in the outermost layer ( fig.[ang_freq_evol_3em6 ] below ) . in caseb the angular momentum is hardly transported into the original white dwarf ( @xmath48 ) .thus , the angular velocity at @xmath49 in the case b model of @xmath50 is considerably larger than in the corresponding case a model .this account for the result that @xmath32 and @xmath34 in case b with @xmath51 are slightly larger than those for case a. in other cases , however , the effect of rotation is generally stronger in case a models than in case b models .figure [ ang_freq_evol_3em6 ] shows runs of angular velocity of rotation @xmath52 in the interior of models at selected phases of evolution for @xmath40y@xmath26 ( @xmath53 ) .figure [ ang_freq_evol ] shows the same information for @xmath54 ( lower panel ) and @xmath55 ( upper panel ) .obviously , the angular momentum transport by the turbulent viscosity in case a is so efficient that @xmath52 is nearly constant in the interior . in the case a model with @xmath40y@xmath26 , @xmath52 varies steeply in the very superficial layer when the total mass is larger than @xmath56 ( fig .[ ang_freq_evol_3em6 ] ) . since @xmath7 at the stellar surfaceis always set to be the critical frequency @xmath57 ( @xmath58 ) , a steep increase in @xmath52 toward the surface indicates the star is rapidly contracting , and a steep decrease indicates a rapid expansion .the surface angular velocity changes so rapidly that the transport of angular momentum is not efficient enough to keep the angular velocity of rotation in the interior close to the surface value . the model contracts firstand then expands rapidly . in the interior of a case b modelthere is a peak in @xmath52 , which is produced by compression .the angular velocity at the peak is higher than that at the surface of the model .although @xmath7 decreases steeply outward near the surface , the specific angular momentum @xmath59 increases outward there because the distance from the center of a mass shell increases outward steeply near the surface . in early phases of accretion, there is a narrow layer where the specific angular momentum decreases outward just outside of the peak in @xmath7 at @xmath60 .but , the structure is always stable to the solberg - hiland instability , because the temperature gradient is positive there and hence the stabilizing effect of buoyancy exceeds the destabilizing effect of the negative gradient of the specific angular momentum distribution .figure [ ev_angmom ] shows the evolution of the total angular momentum of the models accreting matter at @xmath42 , @xmath43 and @xmath44. the higher the accretion rate , the lower the total angular momentum at a given total mass . in case a models ( solid lines ) the total angular momentum , @xmath61 , increases rapidly until a mass of @xmath62 is accreted .after the total angular momentum attains a peak , ( when the angular velocity in the interior becomes almost constant ) , it decreases as the total mass increases to @xmath63 , returning part of the accumulated angular momentum to the disk . for @xmath64 , @xmath61 increases but very slowly . in this phasemost of the angular momentum associated with the accreted matter is returned to the disk . to compare the evolutionary @xmath65 relations with those of simpler models ,we have computed uniformly and critically rotating steady models , in which the rate of entropy change @xmath66 is replaced with @xmath67 where @xmath68 ( e.g. , @xcite ) , and the nuclear reactions are artificially suppressed .the @xmath69 relation of the steady model for each accretion rate is shown by a dash - dotted line in fig [ ev_angmom ] .( detailed properties of the steady models will be presented in a future paper . )this figure shows that the gradual increase of @xmath61 with a increase in mass in the range @xmath64 occurs along the @xmath65 relation of the steady model with the same accretion rate , indicating the evolution in this phase occurs nearly homologously . the total angular momentum in a case b model is smaller than the case a model with the same mass and accretion rate .since the angular momentum transport in the case b models is less efficient then in the case a models , more angular momentum is returned to the accretion disk in the former case .the total angular momentum of a case a model is smaller than that of the uniformly and critically rotating white dwarf of the same mass without accretion ( uenishi , nomoto , hachisu 2003 ) .the difference is attributed to the fact that the critical angular velocity of rotation for the the accreting white dwarf is lower than that of the white dwarf without accretion , because the radius of the former is larger .when the mass of the white dwarf becomes larger than @xmath70 , the total angular momentum @xmath61 of the case a model of @xmath71y@xmath26 begins to decrease due to a rapid decrease in radius .the commencement of the rapid contraction roughly corresponds to the phase when the mass of the white dwarf exceeds the maximum mass of the steady model for the same accretion rate . at @xmath72 the radius and the total angular momentum become minimum and carbon burning is almost ignited at @xmath73 .however , at this point the model begins to expand and the total angular momentum begins to increase . due to the expansion, the maximum temperature in the interior decreases , and carbon ignition does not occur in this model .further computation became too difficult to continue .we note however that if the viscosity is switched to the case b viscosity , we can continue the calculation to have a massive ( say @xmath74 ) model having a high angular momentum , in which @xmath52 has a broad peak similar to those shown in @xcite .finally , we note that the ratio of the rotational kinetic energy to the absolute value of the gravitational potential energy defined as @xmath75^{-1}\ ] ] is always less than @xmath76 in our models .for such a small ratio , no dynamical instability occurs unless a very strong differential rotation in the deep interior of the star exists @xcite .the ratios @xmath77 of our models are smaller than the corresponding ratio denoted by @xmath78 in @xcite for the critically rotating model with a similar accretion rate , where @xmath78 is defined as @xmath79^{-1}.\ ] ] a rotating and accreting white dwarf has a large radius because the outer envelope ( in which only a small fraction of mass resides ) is expanded , while the inner layers contribute to the gravitational energy . therefore , @xmath80 is generally larger than @xmath81 for the rotating and accreting models .in fact , @xmath80 of our models can be as large as @xmath82 , which makes @xmath77 as small as @xmath83 .even if @xmath78 defined by equation ( [ gamma ] ) reached @xmath84 , @xmath77 can be as small as @xmath85 .since @xmath77 should be used for the stability criterion rather than @xmath78 , our models should be stable .we also note that the @xmath77 ratios of the models in @xcite are larger than those of our models .this is attributed to the fact that in the interior of their models there exist a broad peak of the angular velocity .taking into account the centrifugal force of rotation , we computed models of co - white dwarfs that accrete co - rich matter up to the carbon - ignition .we assumed that the surface angular velocity of rotation is equal to the keplerian angular velocity .we have found that carbon is ignited in an outer layer if the accretion rate is greater than @xmath86 .the rotation delays the ignition .the difference in the total mass at the ignition between rotating and non - rotating models can be as large as @xmath38 , depending on the assumed turbulent viscosity and the accretion rate ( see @xcite for the 2d models ) .when the double white dwarfs coalesce , the more massive white dwarf is expected to accrete co - rich matter at a rate close to the eddington limit ( @xmath87yr@xmath26 ) .we have found in this paper that such a high accretion rate leads to an off - center carbon ignition even if the effect of rapid rotation is included .once the ignition occurs , the carbon flame propagates through the interior to the stellar center due to conduction and the c - o white dwarf is converted to an o - ne - mg white dwarf peacefully @xcite . the result of our computation including the effect of rotation supports our previous conclusion that a double c - o white dwarf merger would _ not _ yield a sn ia .we are grateful to the referee for the constructive comments .this work has been supported in part by the grant - in - aid for coe scientific research ( 15204010 , 16042201 , 16540229 ) of the ministry of education , culture , science , sports , and technology in japan .+ @xmath98 & 1.0 & 1.35 & 1.36 & 9.15 & & & & 1.38 & 1.40 & 9.25 + @xmath98 & 0.8 & 1.35 & 1.36 & 9.13 & & & & 1.39 & 1.40 & 9.24 + + @xmath43 & 1.0 & 1.24 & 1.27 & 8.47 & 1.33 & 1.36 & 8.64 & 1.31 & 1.34 & 8.67 + @xmath43 & 0.8 & 1.25 & 1.28 & 8.48 & 1.33 & 1.37 & 8.64 & 1.31 & 1.35 & 8.66 + + @xmath99 & 1.0 & 1.05 & 1.12 & 7.82 & 1.07 & 1.15 & 7.83 & 1.08 & 1.16 & 7.87 + @xmath99 & 0.8 & 0.98 & 1.09 & 7.65 & 1.01 & 1.13 & 7.70 & 1.02 & 1.14 & 7.72 +

<S> we study the effect of stellar rotation on the carbon ignition in a carbon - oxygen white dwarf accreting co - rich matter . including the effect of the centrifugal force of rotation </S><S> , we have calculated evolutionary models up to the carbon ignition for various accretion rates . </S><S> the rotation velocity at the stellar surface is set to be the keplerian velocity . </S><S> the angular velocity in the stellar interior is determined by taking into account the transport of angular momentum due to turbulent viscosity . we have found that an off - center carbon ignition occurs even when the effect of stellar rotation is included if the accretion rate is sufficiently high ; the critical accretion rate for the off - center ignition is hardly changed by the effect of rotation . </S><S> rotation , however , delays the ignition ; i.e. , the mass coordinate of the ignition layer and the mass of the white dwarf at the ignition are larger than those for the corresponding no - rotating model . </S><S> the result supports our previous conclusion that a double - white dwarf merger would not be a progenitor of a sn ia . </S>

cond-mat0205127

this work is motivated by experiments @xcite on transport properties of the two dimensional electron system in high mobility si - mosfet s at small electron concentration @xmath0 .these experiments raised doubts about the applicability of the fermi liquid theory and the conventional theory of localization @xcite to the two dimensional disordered electron liquid at low temperatures .the aim of this article is to prove the existence of zero temperature phases of the two dimensional _ pure _ electron liquid in mosfet s which are intermediate between the fermi liquid and the wigner crystal .these phases exist in some interval of concentrations @xmath1 .the values of the critical concentrations @xmath2 and @xmath3 are estimated below .this phenomenon is due to a tendency for phase separation which originates from the existence of a first - order phase transition between the fermi liquid and the wigner crystal phases as a function of @xmath0 .the difference between the crystal - liquid phase transition in mosfet s and the usual first order phase transitions in neutral systems is the following .in neutral systems with first order phase transitions the energy of the surface between the phases is positive and the minimum of the free energy corresponds to a minimal surface area and to a global phase separation . in charged systems , like electrons on a positive frozen background ,global phase separation does not occur because of a large coulomb energy associated with a non - uniform distribution of electron density .the electron liquid in mosfet s , in a sense , is a system intermediate between these two limiting cases .similarly to the neutral systems with first order phase transitions , the electron liquid in mosfet s exhibits phase separation . on the other handthe surface energy of a minority phase droplet of a large enough radius turns out to be negative . as a result at different @xmath0there is a variety of intermediate phases in this system which are different both from the fermi liquid and from the wigner crystal .the electron system with phase separation demonstrates a number of unusual features . if @xmath4 , the state of the system corresponds to a small concentration of fermi liquid droplets embedded into the wigner crystal .the main difference between such a state and the usual wigner crystal is that it is not pinned by small disorder and can bypass obstacles , while the classical crystals at zero temperature are pinned by an infinitesimally small amount of disorder @xcite .phenomenologically this state of matter is similar to the supersolid phase proposed in @xcite for the case of @xmath5 and @xmath6 .the difference is that in our case the origin of droplets of liquid embedded in the crystal is classical electrostatic , whereas , in the case @xcite the existence of vacancies and interstitials in the ground state of quantum crystals is of quantum origin .if @xmath7 , then the state of the system corresponds to a small concentration of wigner crystal droplets embedded into the fermi liquid . at small concentrations and small temperatures , in principle , these droplets can be considered as quasiparticles .droplets of a minority phase interact at large distances via short - range dipole forces rather than via coulomb forces .this means that at @xmath8 and at small droplet concentration the system of such `` droplet quasiparticles '' should be in a liquid state similar to @xmath5 and @xmath6 which are also liquids at small densities .thus we can describe the system by two - fluid hydrodynamics .however , the statistics of these quasiparticles remains unknown . at zero temperaturethe one - dimensional boundary between the liquid and the solid is a quantum object itself .due to zero - point oscillations of its position there is a region where the wave function has a form which is intermediate between the fermi liquid and the wigner crystal .since the electron densities of the wigner crystal and the fermi liquid are slightly different the fluctuations of the position of the boundary is associated with the fact that the number of quasiparticles in the fermi liquid is not conserved . on the mean - field levelthis picture of droplet formation in the electron liquid in mosfet s is similar to the partial phase separations which occur in ferromagnetic films @xcite , charged polymers @xcite , neutron stars @xcite , doped manganites ( see for example @xcite ) , htc superconductors @xcite and 2-dimensional electron systems in the quantum hall regime @xcite .all these systems demonstrate a short - ranged tendency for phase separation which is thwarted by a long range coulomb interaction preventing global phase separation .the paper is organized as follows . in chapter 2we show that there is an interval of electron concentrations in which the system is unstable with respect to the phase separation .we also estimate the size of minority phase droplets embedded into the majority phase and the temperature and magnetic field dependence of the droplet concentration . in chapter 3we discuss transport properties of different nonuniform phases associated with the phase separation . in section 4we compare the theoretical and experimental results on transport properties of the low density electron gas in si mosfet s .in this chapter we show that a partial phase separation is a generic property of pure 2-d electron liquids in mosfet s .consider a two - dimensional electron liquid of density @xmath0 in a mosfet separated by a distance @xmath9 from a metallic gate .electrons interact via coulomb interaction while a global electric neutrality of the system is enforced by the metallic gate with a positive charge density @xmath10 .the energy density of the system per unit area @xmath11 is a sum of the energy density of the capacitor @xmath12 and the internal energy density of the electron liquid @xmath13 . in the case of a uniform electron distributionthe capacity per unit area is @xmath14 . at high electron densities@xmath15 the kinetic energy of electrons is larger than the potential energy and the interaction can be taken into account by a perturbation theory .( here @xmath16 is the electron bohr radius . ) in this case the system can be described by fermi liquid theory , the difference between the effective @xmath17 and the bare @xmath18 electron masses is small , and @xmath19 . on the other hand , in the opposite limit @xmath20 ( but still @xmath21 ) the potential coulomb energy of electrons is much larger than the kinetic energy and the ground state of the system is a wigner crystal with @xmath22 ( see , for example , @xcite ) .thus at zero temperature there is a critical electron concentration @xmath23 where the phase transition between the fermi liquid and the wigner crystal phases takes place . according to landau mean field theorythis transition is of the first order ( see for example @xcite ) .the @xmath0-dependence of the energy densities of the fermi liquid @xmath24 and the wigner crystal @xmath25 phases near the critical density @xmath23 is shown schematically in fig.1a . in the limit of small densities @xmath26 , due to the existence of the image charges in the gate, the interaction between adjacent electrons has a dipole character . in this case the ratio between the potential and the kinetic energy decreases as @xmath0 decreases .therefore , the small electron @xmath0 the electron system is a weakly interacting fermi liquid .thus we arrive at the conclusion that there exists another critical point @xmath27 which corresponds to a second wigner crystal - fermi liquid transition .the phase diagram of the electron system at @xmath8 is shown in fig.1b . if @xmath28 , than the system is in the liquid state at any value of @xmath0 . herethe factor @xmath29 is the result of numerical simulations @xcite . in the approximation when @xmath30 the qualitative picture of the phase transition is the same as the picture of any first order phase transition in neutral systems .in particular , there is an interval of electron densities @xmath1 shown in the fig.1a where there is a phase separation , which means that there is a spatially nonuniform distribution of the wigner crystal and fermi liquid phases coexisting in equilibrium . in the case of large @xmath9one can linearize @xmath31 near the point @xmath32 . as a result , we have @xmath33 where @xmath34 .one can get from eq.1 an estimate @xmath35 for the size of the interval of electron densities where the phase separation occurs .values of @xmath36 in various mosfet s range from of order one to 50 .the relative fractions of these phases @xmath37 and @xmath38 are determined by the maxwell rule . at @xmath39the fraction of the area occupied by the wigner crystal @xmath40 is small while in the case @xmath41 the fraction of the area occupied by the fermi liquid @xmath42 is small . @xmath43the compressibility of the system @xmath44 should exhibit jumps of order @xmath45 at points @xmath46 .the crucial difference between first order phase transitions in neutral systems and in the system of electrons in mosfet s arises when one considers shapes of the minority phases . in the case of neutral systemsthe surface energy density @xmath47 is positive .therefore in equilibrium the system should have a minimal area of the surface separating the phases , leading to global phase separation . on the other hand , in the three dimensional charged systemsthe global phase separation is impossible because of the large coulomb energy associated with the charge separation .it is possible , however , that in this case the electron system consists of bubbles and stripes of different electron density @xcite , provided the tendency for phase separation is strong enough .the situation in mosfet s is very different . on one hand , in the approximation when @xmath30 global phase separation is possible at an arbitrary value of @xmath48 . on the other hand ,it turns out that for large droplets of the minority phase the surface energy is _negative_. to prove this one has to take into account the finite size corrections to the standard formula for the capacitance @xcite @xmath49 where @xmath50 and @xmath9 are the capacitor area and thickness respectively and @xmath51 is the capacitor size .consider , for example , the case when @xmath40 .then @xmath37 can be determined by the maxwell rule in the approximation when the second term in eq.3 is neglected and @xmath30 .expanding @xmath52 with respect to the second term in eq.3 and taking into account also the microscopic surface energy we have an expression for the energy of the surface @xmath53 we assume that the wigner crystal phase embedded into the liquid consists of droplets of radius @xmath54 and concentration @xmath55 and take into account that inside the droplet @xmath56 .thus , at large @xmath54 the surface energy eq.4 turns out to be negative .we have to find a minimum of eq.4 at a given total area occupied by the minority phase , which gives us the characteristic size of the droplet @xmath57 with @xmath58 .a similar expression was obtained in @xcite for a different problem .the analogous calculation for the case @xmath42 gives the expression for the radius of liquid droplets embedded into the crystal which is identical to eq.5 . at the point of the transitionthe values of @xmath47 and @xmath59 are of the same order and at present nothing is known about the value of the dimensionless parameter @xmath60 . even the fact that @xmath61 is not proven .i would like to also note that in the case of the first order phase transitions which are close to the second order one we always have @xmath62 . in this articlewe assume that @xmath63 . to illustrate the physical meaning of this inequality we consider the case when 2-d electron liquid is compensated by a uniformly charged positive _ frozen _ background with a charge density @xmath10 .in this case the coulomb energy of a droplet associated with the phase separation is , roughly , @xmath64 times larger than in the mosfet s case .the most dangerous point with respect to the phase separation instability is @xmath32 ( see fig.1a ) .for example , let us compare the energies of the uniform liquid state with @xmath32 and a nonuniform state which contains two droplets embedded into the liquid .the first droplet is a liquid with electron concentration @xmath65 , while the second term is a crystal with electron concentration @xmath66 .suppose the droplets have the same radius @xmath67 .linearizing @xmath68 with respect to @xmath69 we estimate the energy difference @xmath70 between these two states as @xmath71 the first term in eq.6 corresponds to a decrease of the energy due to the phase separation .the second one corresponds to the positive coulomb energy associated with the nonuniform distribution of the electron density and the third term is the surface energy .a minimization of eq.6 with respect to @xmath69 gives us @xmath72 and @xmath73 the assumption @xmath74 means that @xmath75 in eq.7 is positive and that 2-d electron liquid on a _ frozen _ positive background does not exhibit a phase separation .( we would like to metion that if the value of @xmath76 is big enough a microscopic phase separation in the charged liquid on frozen positive background can take place even in 3d case .such situation has been considered in @xcite ) on the mean field level our problem is similar to @xcite . using this analogywe conclude that in the middle of the interval @xmath77 there is a stripe phase .the phase diagram of the system is shown schematically in fig.2 .the main difference with @xcite is the following . in @xciteall phase transitions between uniform , bubble and stripe phases are of the first order , whereas in our case the transitions between uniform ( fermi liquid and wigner crystal ) phases and the bubble phases are continuous .the transitions between the bubble phases and the stripe phase would be the first order one .however , such a transition would have an interval of concentrations where phase separation would take place . in this casethe presented above arguments could be repeated .thus we expect more complicated structures than bubbles and stripes phase to exist between the bubble and the stripe phases . since the complete solution of this problem remains to be found we indicated this in fig.2 by shaded lines .let us now estimate dependence of @xmath78 on the temperature @xmath79 and the magnetic field @xmath80 parallel to the film .it is determined by the corresponding dependence of the free energies for the fermi liquid and wigner crystal phases . at small @xmath79 and @xmath80one can neglect the @xmath79 and @xmath80 dependences of @xmath81 and we have the following expression for the free energy densities of the liquid and the wigner crystal phases @xmath82 where @xmath83 and @xmath84 are the entropies of the crystal and the liquid phases respectively , while @xmath85 and @xmath86 are the corresponding spin magnetizations per electron . as a result ,one can obtain how @xmath78 , and @xmath87 depend on @xmath79 and @xmath80 by making the following substitution in eqs.1,2 @xmath88 at small @xmath89 we have @xmath90 , where @xmath91 and @xmath92 are linear susceptibilities of the crystal and the liquid respectively .( here @xmath93 is the bohr magneton and @xmath94 is the fermi energy . ) at low temperature @xmath95 the spin susceptibility of the wigner crystal @xmath96 is much larger than the spin susceptibility of the fermi liquid .the entropy of the crystal @xmath97 is mainly due to the spin degrees of freedom and much larger than the entropy of the fermi liquid. thus @xmath37 increases linearly with @xmath79 and quadratically with @xmath80 , which means that both the temperature and the magnetic field parallel to the film drive the electron system toward the crystallization @xcite .( we assumed that the temperature is larger than the exchange energy between spins in the wigner crystal ) .these effects are known in the physics of @xmath98 as pomeranchuk effects . in the intermediate interval of magnetic fields@xmath99 spins in the wigner crystal are completely polarized while the fermi liquid is still in the linear regime . in this case@xmath37 increases linearly with @xmath80 . at high magnetic field @xmath100both fermi liquid and wigner crystal are spin polarized and @xmath101 saturates as a function of @xmath80 .we assume that @xmath102 and , therefore @xmath103 . on the other hand ,the spin entropy of the wigner crystal is frozen in this case . as a result , at @xmath104 the temperature dependence of @xmath78is suppressed significantly . in principle , at small enough concentrations and at small temperatures droplets of the minority phase embedded into the majority one should behave as quasiparticles . since the system is translationally invariant , they should be characterized by momentum ( or by quasi - momentum ) .the momentum coincides with the flux of mass .thus these quasiparticles carry a mass @xmath105 , a charge @xmath106 and a spin .the characteristic temperature of quantum degeneracy is @xmath107 .the value of @xmath105 depends on the mechanism of motion of the droplets , which in turn depends on whether the surface between the crystal and the liquid is rough or smooth .consider for example the case of wigner crystal droplet embedded into the liquid . in the case of a smooth surface , motion of the dropletis associated with a redistribution of the liquid mass on the distance of order @xmath54 . in this casewe can estimate the effective mass of the droplet as @xmath108 in the case of rough surfaces the motion of the droplet is associated with melting and crystallization of different parts of it . since the @xmath109 the liquid mass to be distributed and , consequently , effective mass of the droplet @xmath110 in this case is much smaller than eq.10 .droplets of the minority phase interact at large distances via short - range dipole forces rather than via coulomb forces . at small enough concentration of the droplets the amplitude of quantum ( or classical ) fluctuations of their positionsis larger than the typical distance between them .thus the liquid droplets are distributed uniformly over the whole crystal . in other words , at @xmath8 the system of such `` droplet quasiparticles '' should be in a liquid state similar to @xmath5 and @xmath6 which are also liquids at small densities .thus we can describe the system by two - fluid hydrodynamics . in this case statistics of the `` droplet quasiparticles '' becomes important .in this respect we would like to mention a difference between the droplets of the liquid embedded into the crystal and the droplets of the wigner crystal embedded into the liquid .the droplets of the liquid are topological objects which , in principle , are not different from vacancies or interstitials in quantum crystals @xmath5 and @xmath6 . in order to create such objects in wigner crystal onehas to add or to remove from the lattice an integer number of electrons .therefore , the liquid droplets have a definite statistics : they are either fermions or bosons @xcite .the main feature of the phase where there are droplets of liquid embedded into the crystal ( supersolid ) is it s ability to bypass static obstacles .in other words , unlike conventional crystals supersolids are not pinned by disordered potential of small amplitude . this will manifest itself in the finite conductivity of the system . from the phenomenological point of viewthis is very similar to the scenario of `` supersolid '' which has been introduced by a.f.andreev and i.m.lifshitz @xcite for quantum crystals of helium near the quantum melting point .they assumed that the crystals contain zero point defects ( vacancies or interstitials ) in the ground state and therefore the number of atoms and number of sites in the crystals are different .the difference with @xcite is that the origin of the negative surface energy eq.4 is purely classical .conversely , following @xcite the existence of point defects in the ground state could be of quantum origin .namely , the kinetic energy of the point defects can be larger than the energy required for their creation .thus , the supersolid phase @xcite can be considered as a particular case of a more general situation of the phase separation when the radius of liquid droplets embedded into the crystal is of order @xmath111 .this would mean that the surface energy is renormalized to a small ( or negative ) value .indications of the existence of such a phase have been reported in numerical simulations @xcite .the case of droplets of the wigner crystal embedded into the fermi liquid is different because they are not topological objects . in principlesuch droplets can contain an additional charge and spin which can be fractional or even irrational . a fundamental problem associated withthis fact is that statistics of such quasiparticles is _unknown_. to illustrate this point we consider a process of tunneling between two states : a state of uniform fermi liquid and a state when there is one crystalline droplet embedded into the fermi liquid .these two states have different total electron charge .thus the tunneling between these states is associated with a redistribution of this charge to ( and from ) the infinity .it is important that the action @xmath112 associated with this process in the pure two dimensional case is finite .one can estimate it in a way similar to @xcite . on distances larger than the droplet size @xmath54one can write the action in terms of the time - dependent electron density @xmath113 @xmath114^{2}}{c_{0}}\sim \int dt \frac{(em^{*})^{2}d}{m^{2}}\frac{1}{r^{2}(t)}\ ] ] here @xmath115 is the imaginary time and @xmath105 is given by eq.11 .we approximate that @xmath116 at @xmath117 and @xmath118 at @xmath119 .eq.12 corresponds to the potential energy contribution to the action . as usual , the contribution from the kinetic energy is of the same order . assuming that @xmath120 we get an estimate @xmath121 .thus , in principle , the wave function of the object is a coherent superposition of the wave functions of a uniform fermi liquid and a wigner crystal droplet . in this situationit is quite likely that the additional charge associated with such an object is not an integer .this is the reason why the nature of the ground state of the system remains unknown .the quantum melting of the phases , which are intermediate between the bubble and the stripe phases is even more complicated and we leave this question for further investigation .in this section i will consider cases when quantum statistics of the system of droplets of the minority phase is not important .the electron - electron scattering conserves the total momentum of the electron system and therefore does not contribute to the resistance of the system . to estimate it we have to consider the electron system in the presence of a random elastically scattering potential .the electron transport picture in the electron liquid with partial phase separation is quite rich . in particular , there is a region of electron concentrations where the hydrodynamics of the electron liquid is similar to the hydrodynamics of the liquid crystals @xcite . in this paperwe consider only cases where either there are crystalline droplets of small concentration embedded in the liquid ( @xmath42 ) , or there are liquid droplets with @xmath40 embedded into the crystal . in these situations , in principle , there are two types of current carriers in the system : electron quasiparticles and charged droplets of the minority phase . in this articlewe will ignore the contribution of the droplet motion to the charge transport . to illustrate the possible @xmath79 and @xmath122 dependence of the resistance we consider below only several limiting cases leaving a detailed analysis for future investigation .let us consider the case @xmath40 when crystalline droplets of small concentration are embedded into a fermi liquid .we will assume here that the wigner crystal droplets are either pinned by a small scattering potential , or have a short mean free path .we also assume that otherwise the impurities do not affect the thermodynamic properties of the system .the contribution to the resistance of the system from the scattering of quasiparticles on droplets has the form @xmath123 where @xmath124 is the fermi momentum of the fermi liquid , @xmath125 is the quasiparticle mean free path , and @xmath126 is the concentration of droplets of the wigner crystal .thus as followes from eq.1,2,9,13 at small @xmath79 the resistance of the electron system increases linearly in @xmath79 . at small@xmath80 it increases quadratically in @xmath80 , while in the intermediate interval of @xmath80 it increases linearly in @xmath80 .the saturation of the magnetoresistance as a function of @xmath80 takes place at @xmath127 when the electron fermi liquid gets polarized . at @xmath128 the spin entropy of the wigner crystalis frozen .therefore , as it has been discussed , @xmath129 and the resistance of the system do not have a significant @xmath79-dependence .the @xmath80 dependence of the resistance @xmath130 of the metallic phase at small @xmath79 is shown schematically in fig.3a .the @xmath79 dependences of @xmath131 at @xmath132 and @xmath127 are shown in fig.3b . eventually at high enough temperatures the crystalline droplets melt . since at this point@xmath133 the melting temperature @xmath134 is much smaller than the plasma frequency at the wave vector of order of the inverse inter electron distance .here @xmath135 is the ratio between the potential and the kinetic energies of electrons .let us now discuss the @xmath79-dependence of @xmath131 in this temperature interval .though in this case the liquid is not degenerate , it is strongly correlated .therefore the electron - electron scattering in the liquid is very effective and the local equilibrium is reached in a short time on a spatial scale of order @xmath136 . as a result , the flow of the electron liquid near an impurity can be considered in the framework of hydrodynamics . in the two - dimensional casethe moving electron liquid exerts a force on an impurity , which is given by the stokes formula @xmath137 , @xcite .here @xmath138 , @xmath139 and @xmath140 are the liquid hydrodynamic velocity , viscosity of the electron liquid and the impurity radius respectively . in a system with a finite concentration of impurities the logarithmic factor in the equation for @xmath141should be substituted for @xmath142 , where @xmath143 is the concentration of impurities .thus the resistance of the electron system has the form @xcite@xmath144 the viscosity of the strongly correlated liquid in the semi - quantum regime has been considered theoretically in @xcite for the case of liquid @xmath5 .it was conjectured that @xmath145 we can apply this result to the case of electron liquid in the semi quantum regime ( @xmath146 ) as well .thus we arrive at the conclusion that at high temperatures the resistance should decrease inversely proportional to @xmath79 and that it should have a maximum at @xmath147 .it is interesting to note that , as far as i know , the experimental data on the @xmath79-dependence of the viscosity of @xmath5 in this relatively high temperature region are unavailable . however , we can look at data for the viscosity of @xmath6 , which in this temperature interval is supposed to be similar to @xmath5 @xcite .though the experimental data for @xmath6 are in a reasonable agreement with eq.15 , we would like to mention that the viscosity of @xmath6 changes only by a factor of two in the temperature interval between @xmath94 and the evaporation point .let us consider the case @xmath148 , but @xmath133 .then at @xmath149 the electron system is a strongly correlated fermi liquid .the main feature of such a liquid is that at small distances and at small ( imaginary ) times it behaves like a solid @xcite .it has been suggested in @xcite that the cross - section of quasiparticle scattering on a short ranged impurity with a radius of order @xmath111 , is significantly enhanced by the electron - electron interaction .the nature of the enhancement becomes especially clear if we consider the interval of electron densities close to the critical point @xmath150 and the case when the fluctuations of the external potential have a relatively small amplitude .then the system can become split into the regions of a fermi liquid and a wigner crystal .i would like to mention that the linear @xmath79 and @xmath80 dependences of the resistance mentioned above are generic for strongly correlated electron system and are valid in this case as well .the fractions of volume occupied by the fermi liquid and the wigner crystal depend on @xmath0 and therefore the system should exhibit a percolation - type zero - temperature metal - insulator transition as @xmath0 decreases and the area occupied by the wigner crystal grows .there is however a significant difference with respect to the percolation transition , which originates from the fact that the position of the wigner crystal - fermi liquid boundary has quantum fluctuations .it is these fluctuations which determine the character of the electron transport near the transition point .the properties of the current carriers in this region are very different from properties of the fermi liquid quasiparticles .the significance of quantum fluctuations becomes even more clear if we consider quantum properties of the surface between the wigner crystal and the fermi liquid . at zero temperaturethe surface is a quantum object itself .there are at least two scenarios for the state of the surface at @xmath8 : it could be quantum smooth or quantum rough . in the first case the excitations of the surface are essentially the ralaigh surface waves which conserve the charge inside the droplets . in the second casethere is a new type of excitations at the surface : crystallization waves which do not conserve the total charge inside the droplets @xcite .the problem of quantum roughening has been discussed in the framework of the properties of the boundary between solid and liquid @xmath5 and @xmath6 @xcite . in the case of 2-d surfaces between a 3-d quantum liquid and crystalit has been argued that the surface is always quantum smooth @xcite . at the momentnothing is known about the state of the boundary between 2-dimensional liquid and solid .we would like to mention , however , that quantum effects ( including the quantum roughening ) are more pronounced in the case considered above because it is two - dimensional and because the jump of the electron density in this case is small .in any case , due to quantum fluctuations , there is a region near the boundary whose properties are intermediate between the liquid and the solid properties . in the conclusion of this section we would like to mention that the linear in @xmath79 increase of the resistance at small @xmath79 is a generic property of the model .in this subsection we present a short list of experimental results on the high - mobility two - dimensional electron liquid in si mosfet s @xcite which seem to be in contradiction with fermi liquid theory and with the conventional single particle localization theory of disordered 2-dimensional conductors @xcite .a. the electron system exhibits a `` transition '' as a function of @xmath0 from a metallic phase , where the resistance of the system saturates at low temperatures , to an insulating phase , where the resistance increases as @xmath79 decreases .the value of the critical concentration @xmath151 of the transition depends on the amount of disorder in the sample and corresponds to @xmath152 . here@xmath153 is the ratio between the electron potential and kinetic energies .b. at @xmath8 and for the electron concentration sufficiently close to the critical one , increasing the magnetic field @xmath80 parallel to the film drives the system toward the insulating phase @xcite .thus the critical metal - insulator concentration @xmath154 increases with @xmath80 . in the metallic phase ( @xmath155 ) and at small @xmath79 the system exhibits a big positive magnetoresistance as a function of @xmath80 .this magnetoresistance saturates at @xmath156 and @xmath157 @xcite .c. in the metallic phase at @xmath132 and @xmath158 the resistance @xmath131 significantly increases with increasing temperature .the characteristic value of @xmath159 at small @xmath79 is large and depends on the value of @xmath160 .d. if at @xmath127 the system is still in the metallic phase ( @xmath161 ) , the @xmath79-dependence of the resistance is much weaker than in the @xmath132 case @xcite .e. the value of @xmath162 decreases significantly as @xmath0 approaches @xmath163 . in this subsectionwe present a qualitative explanation of the experimental results @xcite a. _ the existence of the metal - insulator phase transition ._ the theoretical picture presented above involves a transition between the liquid and the crystal as a function of @xmath0 .therefore it can explain qualitatively the existence of the metal - insulator transition observed in the experiments .namely , the fractions of volume occupied by the fermi liquid and the wigner crystal depend on @xmath0 and therefore the system should exhibit a percolation - type zero - temperature metal - insulator transition as @xmath0 decreases and the area occupied by the wigner crystal grows .the transition takes place when the wigner crystal droplets overlap and block the electron transport through the fermi liquid area .the experimental values of @xmath164 correspond to @xmath165 . at presentit is difficult to say how close this value is to @xmath3 , or @xmath23 in the pure case .the critical value for the transition @xmath166 @xcite was obtained by numerical simulations .however it can not be applied to the case of electrons in si mosfet s because of the existence of two almost degenerate electron valleys .another reason for possible inapplicability of the results of @xcite to si mosfet s is that the calculations @xcite were restricted to the case of zero temperature while the experiments have been performed at temperatures larger than the spin exchange energy in the wigner crystal .thus the pomeranchuck effect has not been taken into account in @xcite .finally in the critical value of @xmath153 can be different in the disordered case .b. _ the positive magnetoresistance of the metallic phase in the magnetic field parallel to the film ._ the large positive magnetoresistance of the metallic phase in the parallel magnetic field is connected to the fact that @xmath167 and therefore the magnetic field parallel to the film drives the electron system toward the crystallization @xcite.(see eqs.1,2,9,13 ) .the magnetoresistance should saturate when @xmath127 and the electron fermi liquid is polarized .c. _ the temperature dependence of the resistance in the metallic phase ._ the significant increase of the resistance as a function of temperature can be explained naturally as a consequence of the pomeranchuk effect : the spin entropy of the wigner crystal is larger than the entropy of the fermi liquid and , therefore , the wigner crystal regions grow with increasing temperature . at high temperaturesthe droplets of crystal melt .it follows from eqs.13,15 that in this temperature range the resistance decreases with increasing @xmath79 .it is unclear at present whether the experiments support this picture .d. _ the temperature dependence of the resistance in the metallic phase at large @xmath80_. the pomeranchuk effect disappears when @xmath168 and electron spins are fully polarized . in this caseentropies of both the liquid and the solid are much smaller than the spin entropy of the crystal at @xmath132 .this means that in the leading approximation the areas occupied by the crystal and the liquid are @xmath79-independent .this explains the fact that in the metallic state at @xmath127 the @xmath79-dependence of the resistance is much smaller than in the case @xmath132 @xcite .( the ratio @xmath169 can be as big as @xmath170 ) .e. _ the @xmath0-dependence of @xmath162 ._ perhaps the most direct check of the concept of the fermi liquid which is close to crystallization is the measurement of the @xmath0-dependence of the magnetic field @xmath171 which polarizes the liquid . in the case of a non - interacting fermiliquid @xmath172 is a smooth function of @xmath0 .the problem of the @xmath0 dependence of the critical magnetic field @xmath171 in strongly correlated liquids near the crystallization point and the origin of the strong enhancement of the spin susceptibility has been discussed in the context of the theory of liquid @xmath5 @xcite .it has been pointed out that there are two different scenarios for the origin of the significant ( factor 15 ) enhancement of the spin susceptibility of @xmath5 near the crystallization point .a)the system is nearly ferromagnetic which means that it is close to the stoner instability . in this casethe linear spin susceptibility @xmath92 is large , but other coefficients @xmath173 in the expansion of the energy @xmath174 with respect to the spin magnetization @xmath175 are not small . here@xmath18 is an even integer . in this case@xmath176 , which is relatively large .\b ) the system is nearly solid . in this caseboth @xmath177 and other coefficients @xmath173 in the eq.15 decrease significantly as @xmath0 approaches the crystallization point @xmath23 . in this case@xmath178 is small . in the case of @xmath5the value of @xmath171 has never been measured . in the case of electrons in si - mosfet sit has been measured in @xcite .a dramatical decrease of @xmath171 compared to @xmath179 has been observed as @xmath0 approaches @xmath23 . in our opinionthese experimental results support the model of a nearly solid fermi liquid which is at @xmath133 . conversely it is unlikely that the system is close to the stoner instability . in this sectionwe compare the explanation presented above with another explanation given in @xcite .it is based on the fact that a single short range impurity in a metal creates friedel oscillations of the electron density . due to the electron - electron interaction the quasiparticles in the metalare scattered not only from the impurity but also from the modulations of the electron density . at finite temperaturethe friedel oscillations decay exponentially at distances larger than the coherence length of the normal metal @xmath180 . as a result , at low temperatures @xmath181 with @xmath182 @xcite .the exchange contribution to the resistance has not been taken into account in @xcite .it has been shown @xcite that in the presence of the exchange interaction at @xmath183 the quantity @xmath184 remains linear in @xmath79 . however , at @xmath185 the coefficient @xmath186 has negative sign , which is different from @xcite . on the other hand ,the expriments were performed in the regime @xmath187 .they yeald a positive value of the coefficient @xmath182 . at finite value of @xmath188the theory @xcite predicts , that the coefficient @xmath189 chenges it s sign agein and become positive . at relatively high temperatures@xmath190 one can neglect the interference corrections and the temperature dependence @xmath131 is determined by the corresponding dependence of the thermal velocity and the electron scattering cross - section in a nondegenerate gas . in this regime @xmath131 decreases with increasing @xmath79 @xcite .at this point we would like to mention that at low @xmath79 eqs.1,2,9,13 also predict the increase of the resistance linear in @xmath79 as well .eqs.14,15 also predict the existence of the maximum of @xmath131 at @xmath191 and decrease of the resistance at @xmath192 . thus, both the theory presented above and @xcite , in principle , could explain qualitatively the @xmath79-dependence of the resistance of the metallic state .the situation with the magnetoresistance in the parallel magnetic field is more delicate .strictly speaking , the interference corrections to the drude conductivity calculated in @xcite are relevant only at @xmath185 and at small @xmath79 and @xmath80 , when the effects are small . on the other hand , at @xmath127 , when the effects are large , the interference corrections are irrelevant , and the value of the magnetoresistance @xmath193 is determined by the @xmath80 dependence of the drude part of the resistance , which is due to the @xmath80 dependences of the fermi momentum and the scattering cross - section of quasiparticles .( this part of the magneto resistance has not been taken into account in @xcite ) .in this case a single electron theory yields a big and negative drude magnetoresistance in contradiction with the experimental fact that it is big and positive .in connection with this i would like to make several points .\a . at @xmath187 ,the diagrammatic calculations are not under control . in the framework of the conventional diagram technique it is difficult to account for all effects associated with the strong correlations at @xmath133 , including the giant renormalization of the electron scattering cross - section on impurities , the effects of phase separation , existence of the crystallization waves at the boundary between the two phases , and , finally , the wigner crystallization itself .the mechanism considered in @xcite can not explain the increase of the resistance as a function of @xmath80 and @xmath79 which is significantly larger than unity .this is because the amplitude of the potential created by the friedel oscillations of the density created by an impurity potential is smaller than the impurity potential itself .this theory also can not explain why the temperature dependence of the resistance @xmath131 is suppressed so dramatically by the magnetic field parallel to the film .on the other hand , the theory presented in this article can explain these facts .. these mechanisms of @xmath79 and @xmath80 dependence of the resistance are based on very different physics .this can be seen , for example , from the fact that all single electron interference phenomena including the friedel oscillations are smeared by finite temperature .conversely , the fraction of the wigner crystal increases with temperature .the amplitude of the friedel oscillations is suppressed significantly in the case when the scattering potential is a smooth function of coordinates on the scale of the electron wavelength .this is exactly what happens when the scattering cross - section is significantly renormalized by the fact that near a short range impurity there are crystalline droplets and the position of the crystalline surface exhibits quantum fluctuations .thus , in a sense , the mechanisms based on single electron interference and the mechanism based on quantum fluctuations of the solid - liquid boundary compete with each other . in order to distinguish between these two mechanismsone needs to perform experiments on samples with higher mobility , where effects considered in this article will be much larger than unity .finally we would like to mention that the theory @xcite may be relevant to experiments on the two dimensional electron system in gaas samples @xcite .we have shown that due to the existence of metallic gates in mosfet s the phase separation is a generic property of pure electron liquids . the proof is based only on the assumption about the existence of the first order phase transition between the uniform fermi liquid and the wigner crystal phases and on electrostatic properties of two dimensional electron system .this distinguishes the theory presented above from the theories @xcite , which attempted to explain the experiments using the fact that in the 2-d electron liquid there is a first order phase transition between the fermi - liquid and the wigner crystal which is destroyed by small disorder .this difference , however , manifests itself only at relatively small values of @xmath9 and at relatively small amplitude of the disorder .qualitative pictures of the @xmath79 and @xmath122-dependences of the resistance of the `` metallic '' phase are , roughly speaking , the same for the model presented above and for those considered in @xcite .it is an open question how a disorder of finite amplitude affects the results presented above . in some regimesthe system can demonstrate a glassy behavior characteristic for crystals in the presence of disorder .experimental indications of glassy behavior of the electronic system in si mosfet s have been reported in @xcite . in this paper we considered only bubble phases which exist near the critical concentrations @xmath3 and @xmath2 . in the interval @xmath194 the system ,will probably exhibit a sequence of quantum phase transitions .in particular , it is likely that at electron densities close to @xmath23 there is a stripe phase , which is similar to @xcite .in conclusion we would like to mention that the picture presented above is in many respects similar to the quantum critical point of strongly correlated electron systems considered in @xcite .in particular , the fermi liquid state with densities close to @xmath2 will demonstrate very large sensitivity to imperfections , which is characteristic for the `` almost critical '' quantum state @xcite .this work was supported by division of material sciences , u.s.national science foundation under contract no .we would like to thank a.f .andreev , e. abrahams , s.chakrovarty , a.efros , m.gershenson , s. kivelson , s. kravchenko , l. levitov , d. maslov , v. pudalov , b. shklovskii , m. sarachik , and s. vitkalov for useful discussions .kravchenko , g.v .kravchenko , j.e .furneaux , v.m .pudalov , m.d .iorio , phys .b. * 50 * , 8039 ( 1994 ) ; s. kravchenko , d. simonian , m.p .sarachik , w. mason , j.e .furneaux , phys .77 * , 4938 , ( 1996 ) ; d. popovic , a.b .fowler , s. washburn , phys .* 79 * , 1543 ( 1997 ) .sarachik , s.v .kravchenko , proc .usa * 96 * , 5900 ( 1999 ) ; e. abrahams , s.v .kravchenko , m.p .sarachik , rev .physics * 73 * , 251 ( 2001 ) .d. simonian , s.v .kravchenko , m.p .sarachik , v.m .pudalov , phys .* 79 * , 2304 ( 1997 ) .kravchenko d. simonian , m.p .sarachik , a.d .kent , v.m .pudalov , phys .b. * 58 * , 3553 ( 1998 ) .pudalov , g. brunthaler , a. prinz , g. bauer , jetp lett .* 65 * , 932 ( 1997 ) .pudalov , g. brunthaler , a. prinz , g. brauer , jetp lett . * 68 * , 442 ( 1998 ) .v.m.pudalov , m.diorio , s.v.kravchenko , j.w.campbell , phys.rev.lett .* 70 * , 1866 ( 1993 ) .k.m.metres , h.zeng , s.a.vitkalov , m.p.sarachik , t.m .klapwijk , phys.rev.*b * , rapid communication , * 63 * , r41101 , ( 2000 ) .a.a.shashkin , s.v.kravchenko , t.m.klapwijk , phys.rev.lett . * 87 * , 266402 ( 2001 ) a.a.shashkin , s.v.kravchenko,v.t.dolgopolov , t.m.klapwijk , pys.rev.lett. * 87 * , 086801 ( 2001 ) .t.okamoto , k. hosoya , s. kawaji , and a. yagi , phys.rev.lett .* 82 * , 3875 , ( 1999 ) .v. m. pudalov , g. brunthaler , a. prinz , and g. bauer , phys.rev.lett . *88 * , 076401 ( 2002 ) .s. a. vitkalov , hairong zheng , k. m. mertes , m. p. sarachik , t. m. klapwijk , phys.rev.lett . ,* 87 * , 086401 , ( 2001 ) .abrahams , p.w .anderson , d.c .licciardello , t.v .ramakrishnan , phys .* 42 * , 673 ( 1979 ) .gorkov , a.i.larkin , d.e .khmelnitskii , sov.phys.jetp ., * 30 * , 228 , ( 1979 ) .a.i.larkin , sov.phys.jetp * 31 * , 784 , 10070 ; a.i.larkin , yu.ovchinnikov , j.l.t.phys .* 34 * , 409 , ( 1979 ) .a.f.andreev , i.m .lifshitz , sov.phys .jetp , * 29 * 1107 , ( 1969 ) .t.garel , s.doniah , phys.rev .b , * 26 * , 325 , ( 1982 ) .m.seul , d.andelman , science , * 267 * , 477 , ( 1995 ) .i.a . nyrkiva , a.r.khohlov , m.dao , macromolecules , * 27 * , 4220 , ( 1994 ) .n.k . glendenning , `` compact stars '' springer , 2000 .a.moreo , s , yunoki , e , dagatto , science * 283 * , 2034 , ( 1999 ) .m.yu.kagan , d.i.khomskii , m.v.mostovoi , eur.phys.j . b. , * 12 * , 217 , ( 1999 ) .v.emery , s.kivelson , physica ( amsterdam ) * 209 * , 597 , ( 1993 ) .v.emery , s.kivelson , j. tranquada , proc .oganesyan v , kivelson s. fradkin e. , phys.rev .b. * 64 * , 195109 , ( 2001 ) .j.zaanen , o.gunnarson , phys.rev.b .* 40 * , 7391 , ( 1989 ) .koulakov , m.m .fogler , b.i.shklovskii , phys.rev.lett . * 76 * , 499 , ( 1996 ) ; m.m.fogler , a.a.koulakov , b.i.shklovskii , phys.rev .b * 54 * , 1853 , ( 1996 )mott `` metal - insulator transition '' , taylor and frenncis ltd , london , 1974 .p.m. chaikin , t.c.lubensky , `` principles of condensed matter physics '' , cambridge university press , 1995 .b. tanatar , d.m .ceperley , phys .b * 39 * , 5005 ( 1989 ) .l.d.landau , e.m.lifshitz , electrodynamics of continuous media , new york , pergamon , 1984 .s.kivelson , s.trugman , phys.rev .b. * 33 * , 3629 , ( 1986 ) .b. castaing , p. nozieres , j. de physique * 40 * , 257 ( 1979 ) .g. benenti , x. waintal and j.l .pichard , phys .lett . * 83 * , 1826 ( 1999 ) .l. levitov - s ; a. shytov , jetp.lett . * 66 * , 214 .landau , e.m .lifshitz , fluid mechanics ( pergamon press , 1984 ) .b.spivak , phys.rev.b . *64 * , 085317 , ( 2001 ) .m.hruska , b.spivak , phys.rev .b. * 65 * , 033315 , ( 2002 ) .andreev , yu.a .kosevich , sov .jetp * 50 * , 1218 ( 1979 ) .b. spivak , sov .jetp * 63 * , 760 ( 1986 ) .a.f.andreev , a.yu .parshin , sov.phys .jetp * 48 * , 763 , ( 1978 ) .d.fisher , j.weeks , phys.rev.lett . * 50 * , 1077 , ( 1983 ) .f.stern , s , das sarma , solid state electronics , * 28 * , 158 , ( 1985 ) . t. ando , a.b .fowler , f. stern , rev.mod.phys .* 54 * , 488 , ( 1982 ) .a.gold , v.t.dolgopolov , phys.rev.b , * 33 * , 1076 , ( 1986 ) .s. das sarma and e.h .hwang , phys .lett . * 83 * , 164 ( 1999 ) .v.t.dolgopolov , a.gold , jetp letters , * 71 * , 27 , ( 2000 ) .g.zala , b.n .narozhny , i.l.aleiner , phys .b * 64 * , 214204 , ( 2001 ) ; phys.rev.b . *65 * , 020201 , ( 2001 ) .j.yoon , c.c.li , d.shahar , d.c .tsui , m.shayegan , phys.rev.lett .* 84 * , 4421 , ( 2000 ) ; a.p.mills , jr ., a.p.ramires , l.n.pfeiffer , k.w.west , phys.rew.lett . *83 * , 2805 , ( 1999 ) .s.chakrovarty , s.kivelson , c.nayak , k.voelker , phylos .mag.79,859 , ( 1999 ) . v. m. pudalov , m. e. gershenson , h. kojima , cond - mat/0201001 .s. bogdanovic , d. popovic , cond - mat/0106545 ; j. jaroszynski , dragana popovic , t. m. klapwijk ; cond - mat/0205226 .r.b.laughlin , g.g.lonzarich , p.monthoux , d.pines ; adv . in phys .* 20 * , 361 , ( 2001 ) .s.sachdev,science , * 288 * , 480 , ( 2000 ) .j. ornstein , a. millis , science , * 268 * , 468 , ( 2000 ) .j. lorenzana , c. castellani , c. di castro phys . rev .* b 64 * , 235127 ( 2001 ) .

<S> we show that the existence of an intermediate phase between the fermi liquid and the wigner crystal phases is a generic property of the two - dimensional _ pure _ electron liquid in mosfet s at zero temperature . </S><S> the physical reason for the existence of these phases is a partial separation of the uniform phases . </S><S> we discuss properties of these phases and a possible explanation of experimental results on transport properties of low density electron gas in si mosfet s . </S><S> we also argue that in certain range of parameters the partial phase separation corresponds to a supersolid phase discussed in @xcite . </S>

cond-mat0109498

the resent discovery of superconductivity in sr@xmath2ruo@xmath3 has attracted much theoretical and experimental interest @xcite .muon spin rotation experiments show that the time - reversal symmetry is broken for the superconductor sr@xmath2ruo@xmath3 @xcite .knight - shift measurements show no change when passing through the superconducting state and is a clear evidence for spin triplet pairing state @xcite .also specific heat measurements support the scenario of line nodes within the gap as in the high @xmath4 cuprate superconductors @xcite . in tunneling experiments involving singlet superconductorsboth line nodes and time - reversal symmetry breaking can be detected from the v - like shape of the spectra and the splitting of the zero energy conductance peak ( zep ) at low temperatures respectively @xcite also the properties of ferromagnet - insulator - superconductor with triplet pairing symmetry have been analysed @xcite .it is found that the bound states are suppressed and hence the tunneling conductance peaks are eliminated with the increase of the exchange field . in this paperwe will use the bogoliubov - de gennes ( bdg ) equations to calculate the tunneling conductance of normal metal / triplet superconductor contacts , with a barrier of arbitrary strength between them , in terms of the probability amplitudes of andreev and normal reflection . for the triplet superconductor we shall assume three possible pairing states of two dimensional order parameter , having line nodes within the ruo@xmath2 plane , which break the time - reversal symmetry .the first two are the @xmath5 @xmath1-wave states proposed by hasegawa et al , @xcite having @xmath6 and @xmath7 symmetry respectively . the other one is called nodal @xmath0-wave state and has been proposed by dahm et al @xcite .this pairing symmetry has nodes as in the @xmath7 case .also we will consider the nodeless @xmath0-wave pairing state proposed by k. miyake and o. narikiyo @xcite .we consider the normal metal / insulator / superconductor junction shown in fig .[ fig1.fig ] .the geometry of the problem has the following limitations .the particles move in the @xmath8-plane and the boundary between the normal metal ( @xmath9 ) and superconductor ( @xmath10 ) is the @xmath11-plane at @xmath12 .the insulator is modeled by a delta function , located at @xmath12 , of the form @xmath13 .the temperature is fixed to @xmath14 k. we take the pair potential as a step function i.e. @xmath15 , where @xmath16 , and @xmath17 are spin indices .suppose that an electron is incident from the normal metal to the insulator with an angle @xmath18 .the electron like ( hole ) like quasiparticle will experience different pair potentials @xmath19 .the amplitudes for andreev and normal reflection are obtained by solving the bdg equations . using the matching conditions of the wave function at @xmath12 , @xmath20 and @xmath21 ,the andreev and normal reflection amplitudes @xmath22 are obtained @xmath23 @xmath24 the bcs coherence factors are given by @xmath25/2,\ ] ] @xmath26/2,\ ] ] and @xmath27 .the internal phase coming from the energy gap is given by @xmath28 $ ] , where @xmath29 ( @xmath30 ) , is the pair potential experienced by the transmitted electron - like ( hole - like ) quasiparticle .@xmath31 @xmath32 , since the cooper pairs have zero spin projection i.e. @xmath33 .the tunneling conductance , normalized by that in the normal state is given by @xcite @xmath34 according to the btk formula the conductance of the junction @xmath35 , for spin @xmath36 , is expressed in terms of the probability amplitudes @xmath37 , and @xmath38 as @xmath39 the transparency of the junction @xmath40 is connected to the barrier height @xmath41 by the relation @xmath42 where @xmath43 , denotes the strength of the barrier .the pairing potential is described by a @xmath44 form @xmath45 in terms of the @xmath46 vector .we consider the following pairing symmetries for sr@xmath2ruo@xmath3 .\a ) in the first @xmath5 @xmath1-wave state @xmath6 @xmath47 . this state has nodes at the same points as in the @xmath48-wave case .\b ) for the second @xmath5 @xmath1-wave state @xmath7 @xmath49 . this state has nodes at @xmath50 and has also been studied by graf and balatsky @xcite .\c ) in case of a nodal @xmath0-wave superconductor @xmath51 $ ] , with @xmath52 and @xmath53 , @xmath54 , and @xmath55 in order to have a node in @xmath56 @xcite . this state has nodes as in the @xmath7 state .\d ) in case of a nodeless @xmath0-wave superconductor , proposed by k. miyake , and o. narikiyo @xcite @xmath51 $ ] , with @xmath52 and @xmath53 , @xmath54 , and @xmath57 .this state does not have nodes .in fig . [ z10.fig ] we plot the tunneling conductance @xmath58 as a function of @xmath59 for @xmath60 , for different orientations ( a ) @xmath61 , ( b ) @xmath62 , ( c ) @xmath63 .the pairing symmetry of the superconductor is @xmath6 in fig . [ z10.fig]a , @xmath7 in fig .[ z10.fig]b , nodal @xmath0-wave , in fig .[ z10.fig]c , nodeless @xmath0-wave in fig .[ z10.fig]d .the temperature is fixed to @xmath64 .in all the cases we see the presence of a large residual density of states within the energy gap and peaks due to the sign change of the pair potential at discreet values of @xmath18 , for fixed @xmath65 .also the linear increase with @xmath66 is consistent with the presence on line nodes in the pairing potential .generally the spectra for the three pairing symmetries with line nodes depends strongly on the position of the nodes in the pairing potential and the orientation angle @xmath65 .the spectra for angle @xmath65 in the @xmath6 seen in fig .[ z10.fig]a case is identical to the spectra of @xmath7 in fig .[ z10.fig]b for angle @xmath67 , since the node positions for the two symmetries differ by @xmath68 .the nodal @xmath0-wave case in fig .[ z10.fig]c has the same nodal structure as the @xmath7 case and we see that the spectra for these two candidates are similar . for nodeless pairing states a subdap or a full gap opens in the tunneling spectra .this is seen in fig .[ z10.fig]d , for the nodeless @xmath0-wave pairing state .the peaks in the tunneling conductance are explained from the formation of bound states within the gap .the bound state energies @xmath69 , are given from the values of @xmath66 where the denominator of eqs .( [ ra],[rb ] ) vanishes . in this casethe andreev reflection coefficient is equal to unit , and the effect of the boundary is turned off .the corresponding equation is written as @xcite @xmath70 bound states are formed because the transmitted electron - like and hole - like quasiparticles feel different sign of the pairing potential .these bound states occur for a given orientation @xmath65 at discreet values of @xmath18 , as seen in fig .[ bs.fig ] where the bound state energy @xmath69 is plotted for @xmath71 as a function of @xmath18 for the various pairing states . for a given value of @xmath65 , the conductance @xmath72 at the angles @xmath18 where bound state occurs and the tunneling conductance @xmath58 is enhanced due to the normal state conductance @xmath40 .the residual values of the tunneling conductance within the gap is due to the bound states and the peaks are formed at energies where the number of bound states is increased . as seen in fig .[ bs.fig ] ( a ) for the pairing state @xmath73 for @xmath61 , at the energy in the interval @xmath74 only one bound state occurs . at @xmath75 two more bound statesare formed and the peak seen in fig .[ z10.fig ] ( a ) is due to these new bound states .the same happens for @xmath76 at @xmath77 and @xmath78 , where an increased number of bound states is formed , as seen in fig [ z10.fig ] ( a ) ( dotted line ) . also for @xmath79 in fig [ bs.fig ] ( a ) ( dashed line )the conductance peak is formed for energy close to @xmath80 , where two bound states are formed . in the @xmath81 pairing state the bound states and also the position of the peaks for an angle @xmath65 is the same as in the @xmath73 pairing state for the angle @xmath67 . in the nodal @xmath0-wave state seen in fig .[ bs.fig ] ( c ) the bound states are symmetric to the @xmath81 case with respect to @xmath82 .however this does not influences the energy levels which occur almost at the same position as in the @xmath81-wave case . in the nodeless@xmath0-wave pairing state , for @xmath76 close to @xmath83 we have a pair of new bound states and the peak seen in fig . [ z10.fig ] ( d ) is due to these new bound states .we calculated the tunneling conductance in normal metal / insulator / triplet superconductor using the btk formalism .we assumed nodal and nodeless pairing potentials , breaking the time - reversal symmetry .the residual values are due to the formation of bound states for each @xmath65 at discreet values of the angle @xmath18 for energies within the gap .the peaks occurs at the energies where an increasing number of bound states is formed .9 y. maeno , h. hashimoto , k. yoshida , s. nishizaki , t. fujita , j.g .bednorz , and f. lichtenberg , nature ( london ) * 372 * , 532 ( 1994 ) .luke , y. fukamoto , k.m .kojima , m.l .larkin , j. merrin , b. nachumi , y.j .uemura , y. maeno , z.q .mao , y. mori , h. nakamura , and m. sigrist , nature ( london ) * 394 * , 558 ( 1998 ) .k. ishida , h. mukuda , y. kitaoka , k. asayama , z.q .mao , y. mori , and y. maeno , nature * 396 * , 653 ( 1998 ) .s. nishizaki , y. maeno , and z. mao , j. phys .. jpn . * 69 * , 572 ( 2000 ) .blonder , m. tinkham , and t.m .klapwijk , phys .b * 25 * , 4515 ( 1982 ) .andreev , soviet phys .jetp * 19 * , 1228 ( 1964 ) .n. stefanakis , j. phys . : condens .matter * 13 * , 1265 ( 2001 ) .n. stefanakis , phys .b to appear ( 2001 ) .n. stefanakis , j. phys . : condens .matter * 13 * , 3643 ( 2001 ) .y. hasegawa , k. machida , and m. ozaki , j. phys .. jpn . * 69 * , 336 ( 2000 ) .t. dahm , h. won , k. maki , cond - mat/0006301(2000 ) .k. miyake , and o. narikiyo , phys .83 * , 1423 ( 1999 ) .graf , and a.v .balatsky , phys .b * 62 * , 9697 ( 2000 ) .

<S> we calculate the tunneling conductance spectra of a normal metal / insulator / triplet superconductor from the reflection amplitudes using the blonder - tinkham - klapwijk ( btk ) formula . for the triplet superconductor </S><S> we assume one special @xmath0-wave order parameter having line nodes and two two dimensional @xmath1-wave order parameters with line nodes breaking the time - reversal symmetry . </S><S> also we examine nodeless pairing potentials . </S><S> the tunneling peaks are due to the formation of bound states for each surface orientation at discrete quasiparticles trajectory angles . </S><S> the tunneling spectra can be used to distinguish the possible candidate pairing states of the superconductor sr@xmath2ruo@xmath3 . </S>

astro-ph0410262

the fsvs data set consists of 78 int wfc fields .each field is divided into four , corresponding to the 4 2kx4k ccds that form the wfc . for each field, we took one set of b , i and v images on a given night .several more images were taken in the v band on that night and on consecutive nights . typically , fields were observed in v 10 20 times within one week .exposure times were 10 min with a dead time between observations of 2 min .this observing pattern allows us to sample periodicity timescales from 2@xmath2(observing time + dead time ) ( i.e. 24 min ) up to the maximum time span of the observations ( i.e. a few days ) .the combination of the colour information and the variability allows us to distinguish between different types of variable systems .see groot et al .( 2003 ) for a full discussion on the fsvs data .in order to combine the colour with the variability information contained in the fsvs we must be able to measure the variability timescale associated with each target . because of the relatively few number of v observations per field ( between 10 and 20 ) we use the `` floating mean '' periodogram technique to obtain the characteristic variability timescale in each case . this method has been used successfully in planet searches ( cumming et al .1999 ) and to determine the orbits of subdwarf b binaries ( morales - rueda et al .2003 ) .the floating mean periodogram consists of fitting the data with a model composed of a sinusoid plus a constant of the form @xmath3 where @xmath4 is the frequency and @xmath5 is the observation time .this method corrects a failing of the lomb - scargle periodogram ( lomb 1976 ; scargle 1982 ) which subtracts the mean of the data and then fits a sinusoidal , which is incorrect for small numbers of points . to test whether the periodogram was able to recover the correct periods , as a first step , we simulated the brightness variations of a series of sources with variability periods between 24 min and the time span of the observations , using the time sampling of _ one of the fields _ , and generated fake lightcurves for each period .we used the floating mean periodogram to calculate the most probable variability timescale of each lightcurve , and plotted the calculated versus the real period .the result is a linear curve with a 45@xmath6 slope ( i.e. real periods are recovered successfully ) and a complicated error structure .examples of this test for two given variability amplitudes are presented in fig .1 . using a different time sampling ( which is equivalent to analysing a different field ) will generate a slightly different graph .a more systematic and complete test to check that the method works , and to calculate its period detection efficiency , consists in generating curves like those of fig . 1 for a complete range of values of variability timescale and amplitude , and object brightness ( which is directly related to the uncertainty in the brightness measured for each band ) .these calculations must be done for each time sampling in the data ( morales - rueda et al . in preparation ) .when we run the floating mean periodogram on the real data lightcurves we obtain their most likely variability timescale and the amplitude of that timescale . in the following colour - colour diagrams we show the point sources from the fsvs that show no variability ( top left panel ) , the point sources that show variability ( top right panel ) and the ratio between variable and non - variable point sources ( bottom panel ) .variability is determined in each case by calculating the @xmath7 of the light curve with respect to its average value .objects with @xmath7 above the 5-@xmath8 variability level are considered variable ( groot et al . 2003 ) .the ratio is presented in percentages .the main difference between the distribution of variable and non - variable objects is the excess of variable systems at colours b@xmath9v@xmath00 and v@xmath9i@xmath01. these sources will be mostly qsos but we also expect any cataclysmic variables ( cvs ) in the field to appear in that colour - colour region .the fraction of variable to non - variable objects along the main sequence seems fairly uniform , but more rigorous calculations are required before we can draw any conclusions .we also display the results in colour - colour diagrams for different ranges of variability timescales and amplitudes .the ranges have been selected so the first would include orbital periods corresponding to cvs ( variability scales up to 6 hours ) , the second rr lyr stars ( scales from 6 hours to 1 day ) , the third longer variability trends of cvs ( from 1 to 4 days ) , and anything else ( above 4 days ) .four different amplitude ranges are also chosen . in all colour - colour diagramswe have also plotted the 3-@xmath8 upper limit of the main sequence for clarity .we find no obvious correlation between the time scales of the variability and their amplitudes . in a preliminary analysis of the fsvs we find that , down to 24 mag , there are of the order of 500 objects in the variability range corresponding to cvs and 300 in the range corresponding to rr lyr stars .we find 62 sources showing longer variability periods .the number of point sources found in each variability and amplitude range is given in fig .these preliminary values have not been corrected for the period detection efficiency discussed in section 2 which might alter the results considerably .cumming , a. , marcy , g.w ., & butler , r.p . 1999 , , 526 , 890 groot , p.j . , vreeswijk , p.m. , huber , m.e . , et al . 2003 , , 339 , 427 lomb , n.r . , 1976 , , 39 , 447 morales - rueda , l. , maxted , p.f.l . , marsh , et al .2003 , , 338 , 752 scargle , j.d . , 1982 , , 263 , 835

<S> the faint sky variability survey ( fsvs ) is aimed at finding photometric and/or astrometric variable objects between 16th and 24th mag on time - scales between tens of minutes and years with photometric precisions ranging from 3 millimag to 0.2 mag . </S><S> an area of @xmath023 deg@xmath1 , located at mid and high galactic latitudes , was covered using the wide field camera ( wfc ) on the 2.5-m isaac newton telescope ( int ) on la palma . here </S><S> we present some preliminary results on the variability of sources in the fsvs . </S>

0910.0261

the sun is widely thought to be formed from material representative of local physical conditions in the galaxy at the time of its formation and to represent a standard chemical composition .homogeneity hypothesis _ has been often put in question because of many improvements in the observations techniques and data analysis . with the discovery of new planetary systems , the already known _sources ( stellar nucleosynthesis , stellar formation process ) have gained a new perspective and brought new questions .it is now a fact that stars with giant planets are , on average , richer in metals than those for which no planet was detected ( ( * ? ? ?* gonzalez 2006 ) and references therein ) .some authors suggested that this kind of anomaly may not only involve the content of heavy elements but also some light elements like li , c , n , and o ( ( * ? ? ? * ecuvillon et al .2006 ) and references therein ) .other authors found no difference in the abundance of light elements when comparing stars with and without planets ( see ( * ? ? ? * ecuvillon et al .2004 ) and references therein ) .the studies above are not conclusive yet .we need new tests , using more accurate and homogeneous data , with a larger number of stars .abundances of these elements in solar - like stars will bring new information that will surely help to distinguish the different stellar and planetary formation processes .this is the purpose of our work , in which we analyzed a sample of 200 stars to homogeneously measure the photospheric parameters and c abundances .our sample consists of 200 single solar - like stars in the solar neighborhood , observed in high signal - to - noise ratio ( @xmath1 200 ) and high resolution ( 42000 ) with the elodie spectrograph ( ( * ? ? ?* baranne et al . 1996 ) ) of the haute provence observatory ( france ) , and for which data are publicly available ( ( * ? ? ?* moultaka et al . 2004 ) ) .we obtained @xmath2 , [ fe / h ] , log(g ) , and @xmath3 ( through the excitation equilibrium of neutron iron and the ionization equilibrium between fe i and fe ii ) .we also performed spectral synthesis of some regions containing bands of the @xmath4 swan system ( e.g. at @xmath55165 ) , using the moog lte code ( sneden 2000 , http://verdi.as.utexas.edu/moog.html ) .the present work is still ongoing and we show here our first results : photospheric parameters and c abundances based on the intensity of the @xmath4 ( 0,0 ) band head at @xmath55165 .a comparison of the photospheric parameters here obtained to those published by other works having stars in common shows a good agreement between the samples .figure [ fig ] ( left ) shows the solar spectrum and synthetic spectra for different c abundances .the right panel shows the diagram [ c / fe ] versus [ fe / h ] , comparing stars with and without planets .it seems that stars with planets are slightly richer in c than field stars , but a larger number of planetary systems is required to confirm this possibility .the preliminary results on c abundances are presented here for a large number of nearby solar - like stars , based on homogeneous photospheric parameters obtained from spectra with high signal - to - noise ratio and high resolution .our analysis used public spectra from the elodie database , which represent about 90% of all data .the remaining 10% include many stars with detected planets and having many observations that shall be analyzed in the same way as soon as they become available to the scientific community , since they will contribute to more reliable conclusions .

<S> we have been analyzing a large sample of solar - like stars with and without planets in order to homogeneously measure their photospheric parameters and carbon abundances . </S><S> our sample contains around 200 stars in the solar neighborhood observed with the elodie spectrograph , for which the observational data are publicly available . </S><S> we performed spectral synthesis of prominent bands of c@xmath0 and c i lines , aiming to accurately obtain the c abundances . </S><S> we intend to contribute homogeneous results to studies that compare elemental abundances in stars with and without known planets . </S><S> new arguments will be brought forward to the discussion of possible chemical anomalies that have been suggested in the literature , leading us to a better understanding of the planetary formation process . in this work </S><S> we focus on the c abundances in both stellar groups of our sample . </S>

1605.09010

rate controlled queueing systems commonly arise from applications in communication systems , see e.g. @xcite and references therein . recently ,they have also been considered in modeling limit order books , see e.g. @xcite . a common approach to the study of such rate control problems when the system is in heavy trafficis through diffusion approximations . in a problem setting where there is interaction between servers / queues in that the rates associated with a particular queue and server can depend on the states of the other queues ,this approach leads to a stochastic control problem for @xmath0-dimensional reflected diffusions , where @xmath0 is the number of queues in the system .when @xmath0 is large such control problems are computationally intractable and in general this ` curse of dimensionality ' is unavoidable .however , when there are certain symmetries present and the interaction between queues is weak , in that each queue has @xmath1 affect on any other queue in the system , a natural approach is to consider , in addition to heavy traffic , another asymptotic regime where the number of queues @xmath0 approaches @xmath2 as well .such model settings arise in many applications , e.g. , cloud computing , live streaming , limit order books , customer service systems , etc . in many of these contexts the servers are strategic , for example , in customer service networks , servers respond to workload incentives ( see @xcite ) , and in the context of limit order books buyers and sellers place their orders in a strategic manner and interact weakly through their impact on the price distribution .we study one of the simplest forms of queuing networks , namely a collection of @xmath0 single server queues .the system is assumed to be _ critically loaded_. that is , for each queue , the traffic intensity , which is the ratio between the rate of arrivals to the rate of service is @xmath3 ( see @xcite ) .the arrival and service rates for each queue are @xmath4 . in absence of control and state dependence of rates, the analysis simply reduces to that of a single @xmath5 queue which in the heavy traffic can be approximated by a one dimensional reflected brownian motion . in the setting we consider , every server can exercise control on the arrival and service rates associated with its own queue .this control which is @xmath6 is of lower order compared to the overall rate but it can have significant impact on performance in the asymptotic regime we consider .the rates can depend on the state of the individual queue , furthermore a particular queue s state is influenced by the remaining queue states through their empirical measure .the control action for each server can use information on the history of queue lengths , arrival and processing times and control actions for all the queues in the system . we consider a rate control problem where each server aims to minimize its individual cost .although many different types of cost criteria can be considered , here for simplicity we consider a cost function over a finite time horizon .this cost function may depend on the individual queue lengths , the control action , and the empirical measure of all the queue states .a natural way to formulate optimality for such @xmath0-player games is through the notion of a ( near ) nash equilibrium .computing nash equilibria , even if one considers the simplified heavy traffic approximation in terms of reflected diffusions , is computationally a very challenging problem when @xmath0 is large .the goal of this work is to analyze an asymptotic formulation where simultaneously each queue approaches heavy traffic _ and _ the number of queues approaches @xmath2 .there is extensive literature on heavy traffic limits of rate controlled queuing systems with a fixed number of queues @xcite , however to the best of our knowledge this is the first work to study the asymptotics for controlled queues where in addition to heavy traffic , the number of queues approach infinity as well .we will show that the asymptotics in this regime is governed by a lasry - lions type mean field game ( mfg ) for reflected diffusions .we will use the solution of the mfg to construct an asymptotic nash equilibrium for the @xmath0-player queueing system .this equilibrium has a simple and appealing decentralized structure : each server bases his decision only upon the length of his own queue and a deterministic measure valued function obtained from the solution of the mfg .we also prove that the value of the above near nash equilibrium converges to the value function of the mfg as @xmath7 .in general closed form solutions for mfg of the form that arise here are not available and thus one needs numerical approximations . in our workwe study one such procedure that uses the markov chain approximation method ( @xcite ) and establish convergence of the scheme .the theory of mean field games was initiated a decade ago in the seminal work of lasry and lions @xcite , and huang , malham , and caines @xcite . in recent yearsthere has been a growing interest in this field . for recent theoretical developments and applications of this theorysee @xcite and references therein .mean field approximations for weakly interacting stochastic particles have a long history starting from the works of boltzmann , mckean , kac and others ( see @xcite and references therein ) . even in the context of queuing systems and communication networks ,there have been many works @xcite .another related branch is agent based models with mean - field interaction ( but without strategic agents ) , see e.g. @xcite .mfgs have also been used by queuing theorists as a tool in recent years see e.g. @xcite .in contrast to these works we consider an mfg associated with queuing systems in heavy traffic .we now make some comments on proof techniques . roughly speaking, solution of an mfg considered here can be viewed as the solution of a fixed point problem on the space of probability measures on certain path spaces ( see section [ sec3 ] ) . in order to characterize such solutions , in a setting where each agent s state evolution is described through a diffusion process ,lasry and lions @xcite studied wellposedness of two coupled nonlinear partial differential equations ; one is an equation of hamilton - jacobi - bellman ( hjb ) type while the second takes the form of a kolmogorov forward equation .a somewhat different approach is taken in recent works of carmona , delarue , and lacker @xcite .using probabilistic methods , authors characterize the mfg solution as a solution to certain forward backward stochastic differential equations .the paper @xcite is particularly relevant in that it considers a weak formulation of the mfg .one of our main results is theorem [ thm1 ] which identifies an asymptotic nash equilibrium for the @xmath0-player game as @xmath7 . in order to prove this result ,the main work is to prove suitable tightness properties and analyze and characterize the weak limit points of certain stochastic processes and random measures . herewe make use of a result from @xcite ( see lemma 4.2 therein ) which can be viewed as an extension of de finetti s theorem for sequences of random measures on certain path spaces . in proving suitable tightness properties of control processeswe consider a relaxed control formulation .our assumption on the uniqueness of the minimizer ( assumption [ assumption1](b ) ) ensures that extending the class of controls in this manner does not lower the cost .such relaxed control formulations have been widely used in stochastic control theory , see e.g. @xcite .more recently , they have also been invoked in the study of mean field games @xcite .other key ingredients needed for the proof are ( a ) regularity properties of the optimal feedback controls for certain stochastic control problems for reflected diffusions ; ( b ) weak formulation of stochastic control problems for reflected diffusions .we find that the approach based on hjb characterizations of value functions is particularly well suited for our problem setting .in particular , regularity results from @xcite for quasilinear pde with neumann boundary conditions give us the required estimates for obtaining the desired properties of the optimal feedback controls .use of hjb theory in our analysis is a feature that is common to the approach initiated by lasry and lions , however a point of departure is that instead of using kolmogorov forward equations we characterize probability distributions as unique weak solutions of suitable reflected diffusions . in this sense our approach is closer to @xcite although we do not make use of forward backward stochastic differential equations in our work . in the final section of this work we provide a numerical scheme for solving the mfg .viscosity solution methods for numerical approximations of mfg have been studied in @xcite .here we consider a probabilistic approach based on a markov chain approximation method @xcite .the convergence proof once more relies on weak convergence arguments of the form used in the proof of theorem [ thm1 ] . to the best of our knowledgethis is the first work to develop a numerical scheme based on markov chain approximations for an mfg problem . in summaryour main contributions are as follows .we * consider a rate control problem for large symmetric queuing systems in heavy traffic with strategic servers ; * introduce an mfg for controlled reflected diffusions , establish its solvability and prove unique solvability under additional conditions ( theorem [ thm_fixed ] ) ; * use the solution of mfg to construct an asymptotically optimal nash equilibrium for the @xmath0 player game ( theorem [ thm1 ] ) ; * develop a convergent numerical scheme to solve the mfg ( theorem [ thm : thm52 ] ) .the paper is organized as follows . in section [ sec2 ]we introduce the queueing model , the scaling regime and the cost criterion .next in section [ sec3 ] we introduce the mfg and present our main result on its solvability(theorem [ thm_fixed ] ) .section [ sec4 ] constructs an asymptotic nash equilibrium for the @xmath0-player game and the main result of this section is theorem [ thm1 ] which proves asymptotic optimality .finally , in section [ sec : nummfg ] we provide a numerical scheme for solving the mfg .we use the following notation . for @xmath8 and a function @xmath9 , @xmath10}\|f\|$ ] .for metric spaces @xmath11 denote by @xmath12 the space of @xmath13 valued continuous functions on @xmath14 .when @xmath15 , we abbreviate the notation to simply @xmath16 . the space @xmath17 : { { \cal s}})$ ] for a polish space @xmath18 will be equipped with the uniform topology .we will denote by @xmath19 : { { \cal s}})$ ] the space of @xmath20 valued functions that are right continuous and have left limits ( rcll ) defined on @xmath21 $ ] .this space is equipped with the usual skorohod topology . for @xmath22:{\mathbb{r}}^d)$ ] and@xmath23 $ ] , @xmath24 .we will denote by @xmath25 the space of real lipschitz functions on @xmath18 whose lipschitz norm is bounded by @xmath26 , namely the class of functions @xmath27 with @xmath28 where @xmath29 denotes the metric on @xmath18 . denote by @xmath30 the space of probability measures on @xmath18 .we endow @xmath30 with the topology of weak convergence of measures .convergence in distribution of @xmath20 valued random variable @xmath31 to @xmath32 will be denoted as @xmath33 . for @xmath34 , the space @xmath35:[0,l]))$ ]will be denoted as @xmath36 . for @xmath37 , the wasserstein distance of order @xmath38 on @xmath30 , where @xmath18 is a compact metric space ,is defined as @xmath39^{1/p } : \pi\in{{\cal p}}({{\cal s}}\times { { \cal s}})\;\;\text { with marginals $ \eta'$ and $ \eta$}\right\},\end{aligned}\ ] ] where @xmath40 .we denote by @xmath41\times [ 0,l])$ ] the space of functions @xmath42 that are continuously differentiable ( resp ., twice continuously differentiable ) with respect to ( w.r.t . ) the first ( resp ., second ) variable and are such that the function and the derivatives extend continuously to @xmath21\times [ 0,l]$ ] . for @xmath43\times [ 0,l])$ ], @xmath44 will denote the time derivative and the first two space derivatives of @xmath45 , respectively . for @xmath46, @xmath47 denotes the dirac measure at @xmath48 .we consider a symmetric @xmath0-server stochastic queueing system . each server @xmath49is associated with a queue with a finite capacity and is able to control the rates of the service and arrivals of jobs to its queue .the rates can also depend on the state of the queue and on the empirical measure of the states of all the @xmath0-queues .we will consider a regime where the arrival and processing rates are approximately the same and are of the same order as the number of queues in the system .we now describe our precise scaling regime and introduce the various processes that determine the evolution of the state of the system .fix @xmath50 and @xmath51 . here@xmath52 denotes the terminal time of our finite time horizon and @xmath53 $ ] will be the state space of the scaled queue length process .let @xmath54 be a compact subset of @xmath55 .we will denote by @xmath56 and @xmath57 the ( controlled ) arrival and service rates associated with queue @xmath58 .the rates will be @xmath6 perturbations of certain nominal ( uncontrolled ) @xmath4 overall rates .more precisely , we assume that there exist @xmath59 , and bounded and measurable functions @xmath60\times{{\cal p}}([0,l])\times[0,l]\times u \to { \mathbb{r}}$ ]such that @xmath61 where @xmath62 , @xmath63 is the size of the @xmath58-th queue at time @xmath64 , @xmath65 is the empirical distribution of the scaled queue lengths in the @xmath0-th system at time @xmath64 , and @xmath66 is the control that server @xmath58 exerecises at time @xmath64 .the term @xmath67 represents an expression of the form @xmath68 where @xmath69\times{{\cal p}}([0,l])\times[0,l]\times u \to{ \mathbb{r}}$ ] are measurable functions such that @xmath70 converges uniformly to @xmath71 as @xmath7 .additional conditions on @xmath72 and @xmath73 will be introduced in later sections .each server in the @xmath0-th system has a finite buffer of size @xmath74 and arriving jobs to a full buffer are lost .we assume that the system is in heavy traffic , namely @xmath75 in and in the definition of @xmath76 one can replace @xmath0 and @xmath77 by @xmath78 and @xmath79 , respectively , where @xmath80 as @xmath81 .however for simplicity of presentation we only consider the case @xmath82 .we now give a precise description of controlled stochastic processes of interest .for each fixed @xmath0 , let @xmath83 be a probability space on which are given unit rate independent poisson processes @xmath84 and @xmath85 , @xmath86 .roughly speaking , @xmath84 will correspond to the stream of jobs entering the @xmath58-the queue and @xmath85 to the jobs that leave the system after service .the evolution of the @xmath58-th controlled queue is given as follows .@xmath87,\end{aligned}\ ] ] where @xmath88 is the initial size of the @xmath58-th queue , @xmath89 are the arrival and departure processes respectively , where @xmath56 and @xmath57 are defined as in in terms of control processes @xmath90 .we will assume that @xmath91 are exchangeable for all @xmath0 .we require that the control processes are suitably non - anticipative .specifically , we assume that for some filtration @xmath92 on @xmath83 , @xmath93 is @xmath92-progressively measurable , @xmath94 , @xmath95 are @xmath92-adapted , and @xmath96 are @xmath92 martingales with quadratic variations @xmath97 and @xmath98 , @xmath99 $ ] , @xmath100 where @xmath101 if @xmath102 and @xmath71 otherwise . the process @xmath103 will be referred to as an _ admissible _ control and we denote the collection of all such controls by @xmath104 . with an abuse of terminology , for @xmath103 as above , we will refer to @xmath93 as an admissible control for the @xmath58-th player .from we have the following evolution equation for the scaled queue length processes . for @xmath99 $ ] @xmath105 where @xmath106 @xmath107 and @xmath108 represents a stochastic process @xmath109 satisfying for each @xmath58 @xmath110 , in probability , as @xmath111 .the above dynamics can equivalently be described in terms of a skorokhod map as we do below .let @xmath112 $ ] be the subset of @xmath19:{\mathbb{r}})$ ] consisting of all @xmath113 such that @xmath114 $ ] .[ def_skorokhod ] given @xmath115 $ ] we say the triplet of functions@xmath116 : { \mathbb{r}}^3)$ ] solve the skorohod problem for @xmath113 if the following properties are satisfied : \(i ) for every @xmath117 , \;{\varphi}(t)=\psi(t)+\zeta_1(t)-\zeta_2(t)\in [ 0,l]$ ] .\(ii ) @xmath118 are nonnegative and nondecreasing , @xmath119 , and @xmath120}1_{(0,l]}({\varphi}(s))d\zeta_1(s)=\int_{[0,t]}1_{[0,l)}({\varphi}(s))d\zeta_2(s)=0.\end{aligned}\ ] ] we denote by @xmath121 and refer to @xmath122 as the skorohod map .it is known that there is a unique solution to the skorohod problem for every @xmath123 $ ] and so the skorohod map in definition [ def_skorokhod ] is well defined .the skorohod map has the following lipschitz property ( see @xcite ) .[ lem_skorokhod ] there exists @xmath124 such that for all @xmath125)$ ] , @xmath126 the dynamics in can be described in terms of the skorokhod map as follows @xmath127.\end{aligned}\ ] ] we now introduce the cost function and the control problem studied in this work .the total expected cost for server @xmath58 associated with the initial condition @xmath128 @xmath129 and control @xmath130 is given by @xmath131,\ ] ] where @xmath132\times { { \cal p}}([0,l])\times[0,l ] \times u\to{\mathbb{r}}$ ] is the running cost , @xmath133\to{\mathbb{r}}$ ] is the terminal cost , and @xmath134\times { { \cal p}}([0,l])\to{\mathbb{r}}_+$ ] are the costs ( rewards ) associated with rejection of jobs and empty buffers , respectively .here @xmath135 , and @xmath136 are bounded measurable functions that will be required to satisfy additional conditions that will be introduced in section [ sec3 ] ( see assumption [ assumption1 ] ) . each player ( server )seeks to minimize its cost .a natural formulation of optimality for such @xmath0-player games is given in terms of a ( near ) nash equilibrium .computing near nash equilibria for such complex and large multi - player games is in general intractable and thus we instead consider an asymptotic formulation of the problem .a sequence of admissible controls @xmath137 is called an _ asymptotic nash equilibrium _ if for every player @xmath138 , and every sequence of admissible controls @xmath139 for that player , one has @xmath140 objective of this work is to show that , under conditions , an asymptotic nash equilibrium exists which can be numerically computed ( at least approximately ) by analyzing a related mfg . the main results are theorem [ thm_fixed ] ( solvability of mfg ) , theorem [ thm1 ] ( asymptotic optimality ) , and theorem [ thm : thm52 ] ( convergence of the numerical scheme ) .a natural approach for constructing asymptotic near nash equilibria for the above @xmath0-player game has emerged from the works of @xcite . starting point in this approach is to consider an mfg that formally corresponds to the limit of the above @xmath0-player games as @xmath7 . in this sectionwe give a precise description of this mfg in the current context and give our main results on existence and uniqueness of solutions .the basic idea in the formulation of the mfg is to approximate the scaled queue length process for a typical queue by a suitable drift - controlled reflected brownian motion .we next introduce this controlled process .let @xmath141 be a filtered probability space on which is given a one dimensional standard @xmath142-brownian motion @xmath143 .we will refer to the collection @xmath144 as a system and denote it by @xmath145 .given @xmath146 $ ] , @xmath23 $ ] , and @xmath147 , we denote by @xmath148 the collection of all pairs @xmath149 where @xmath150 is a @xmath54-valued @xmath151-progressively measurable process , @xmath152 is a @xmath53\times { \mathbb{r}}_+\times { \mathbb{r}}_+$ ] valued @xmath151-adapted continuous process such that , @xmath153 and @xmath154,\end{aligned}\ ] ] where @xmath155,\end{aligned}\ ] ] @xmath156 is the marginal of @xmath157 at time instant @xmath158 and @xmath159 .we now introduce the cost function in the mfg . given @xmath147 , @xmath146 $ ] , @xmath23 $ ] , and a system @xmath145 as above , let @xmath160 .define @xmath161.\label{eq : eq234}\end{aligned}\ ] ] we define the value function associated with the above cost as : @xmath162 [ def : mfg729 ] a * solution to the mfg * with initial condition @xmath146 $ ] is defined to be a @xmath147 such that there exist a system @xmath145 and an @xmath163 such that @xmath153 satisfies @xmath164 and @xmath165 if there exists a unique such @xmath157 , we refer to @xmath166 as the * value of the mfg * with initial condition @xmath48 . to find a solution of the mfg one usually follows the following iterative procedure : * for a fixed @xmath147 solve the stochastic control problem ( with @xmath167 ) , namely find a system @xmath145 and @xmath168 such that holds .denote by @xmath169 the law of @xmath32 where @xmath153 and write @xmath170 ( this is not precise since in general there may be more than one solution of the stochastic control problem in ) .* find the fixed point of the map @xmath171 , namely a @xmath172 for which @xmath173 .note that by definition , such a @xmath169 will be a solution of the mfg .we now analyze the mfg by following the above steps .the main result is theorem [ thm_fixed ] which gives existence of solutions of the mfg under suitable conditions and proves uniqueness of solutions under stronger conditions . for @xmath174 ,let @xmath175 be the collection of all @xmath147 such that @xmath176 let @xmath177 .fix a measure @xmath178 . for @xmath147 ,the function @xmath179 is a continuous function from @xmath21 $ ] to @xmath180)$ ] and with an abuse of notation we denote this continuous function once more as @xmath157 .as one might expect , the value function @xmath181 corresponds to the hamilton - jacobi - bellman ( hjb ) equation : @xmath182\times[0,l],\end{aligned}\ ] ] with the boundary conditions ( bc ) @xmath183,\end{aligned}\ ] ] where @xmath184 is the hamiltonian given as @xmath185 and @xmath186\times{{\cal p}}([0,l])\times [ 0,l]\times u\times { \mathbb{r}}\to { \mathbb{r}}$ ] is defined as @xmath187 we now introduce a key condition under which the above hjb equation characterizes the value function @xmath181 .[ assumption1]@xmath188 1. there exists @xmath189 such that for every @xmath190 $ ] , @xmath191)$ ] , @xmath192 $ ] , and @xmath193 , @xmath194 2 . for every @xmath195\times { { \cal p}}([0,l])\times [ 0,l]\times { \mathbb{r}}$ ], there is a unique @xmath196 such that @xmath197 from berge s maximum theorem ( see ( * ? ? ?* theorem 17.31 ) ) and part ( b ) of the above assumption it follows that @xmath198 is a continuous function on @xmath21\times { { \cal p}}([0,l])\times [ 0,l]\times { \mathbb{r}}$ ] .also note that implies that @xmath199 are bounded functions , in particular , @xmath200\times { { \cal p}}([0,l])\times [ 0,l]\times u}|b(t,\eta , x , u)| \doteq c_b< \infty .\label{eq : eq429}\ ] ] the first part of the condition implies hlder-1/2 continuity in @xmath64 when @xmath201 is replaced by @xmath202 for some @xmath203 .for example , if @xmath204 , for @xmath205 $ ] @xmath206 and @xmath207 .assumption [ assumption1 ] will guarantee that the value function is the unique classical solution of the hjb equation with bc ( see proposition [ lem2 ] ) .the assumption will also be used in the fixed point argument ( section [ sec3b2 ] ) , in the asymptotic analysis of the @xmath0-player game ( section [ sec4 ] ) , and for numerical approximations ( section [ sec : nummfg ] ) . using girsanov s theoremit is easily checked that given a measurable function @xmath208\times [ 0,l ] \to u$ ] , @xmath147 , and @xmath209\times [ 0,l]$ ] there is a unique weak solution @xmath210 to @xmath211 @xmath212 $ ] , namely there is a system @xmath213 on which is given a @xmath151-adapted continuous process @xmath214 solving the above equation and if @xmath215 is another system on which is given a @xmath216-adapted continuous process @xmath217 solving the above equation with @xmath218 replaced by @xmath219 then @xmath220 .note also that with @xmath221 , @xmath222 .we refer to the function @xmath223 as a feedback control and we call it an _ optimal _ feedback control for the stochastic control problem in if @xmath224 the following result which is a consequence of theorem 13.24 of @xcite says that the hjb equation with bc admits a unique classical solution which can be characterized as the value function @xmath225 .moreover , there exists an optimal feedback control with certain regularity properties .we follow the notation from @xcite .let for @xmath226 $ ] , @xmath227 be the collection of maps @xmath228 such that @xmath229 note that such a function can be continuously extended to @xmath21\times [ 0,l]$ ] and we denote the extension by the same symbol .also , let @xmath230 be the collection of continuous real functions @xmath113 on @xmath21\times [ 0,l]$ ] such that @xmath231 is twice continuously differentiable on @xmath232 for all @xmath233 , @xmath234 is continuously differentiable on @xmath235 for all @xmath236 , the functions @xmath237 are bounded on @xmath238 and the functions @xmath239 , @xmath240 are in @xmath241 .[ lem2 ] fix @xmath242 and suppose that assumption [ assumption1 ] holds. then @xmath243 and it is the unique solution of . furthermore , with @xmath244 as introduced in assumption [ assumption1 ] , the map @xmath245 is continuous and the feedback control @xmath246 is an optimal feedback control for for every @xmath233 .moreover , any optimal control @xmath247 for satisfies @xmath248 , @xmath249 almost surely ( a.s . ) , where @xmath250 denotes the lebesgue measure on @xmath251 $ ] .* proof : * from ( * ? ? ?* theorem 13.24 ) and the paragraph following its statement it follows that admits a solution in @xmath252 .we remark that the key conditions needed to appeal to this theorem are that @xmath253 and for each @xmath254 @xmath255,\ , x , x ' \in [ 0,l],\ , p , p ' \in [ -m , m],\\ t\neq t ' , x\neq x ' , p\neq p ' } } \frac{|h(t,\nu(t ) , x , p ) - h(t',\nu(t ' ) , x ' , p')| } { |t - t'|^{1/2 } + |x - x'| + |p - p'| } < \infty.\ ] ] both these conditions are easily seen to hold on using assumption [ assumption1 ] and the property that @xmath204 for some @xmath256 ( see e.g. ) .we now argue uniqueness and characterize the solution as the value function @xmath257 in .let @xmath258 be a solution of . fix @xmath209\times [ 0,l]$ ] , a system @xmath145 and @xmath259 .by an application of it s lemma and using we get @xmath260 & = { \mathbb{e}}[\psi(t , x(t - t))]\nonumber\\ & = \psi(t , x ) + { \mathbb{e}}\left[\int_0^{t - t } d\psi(s+t,0)dy(s)-\int_0^{t - t } d\psi(s+t , l)dr(s)\right]\nonumber\\ & \quad + { \mathbb{e}}\int_0^{t - t}\big[d_s \psi(s+t , x(s))+b(s+t,\nu(s+t),x(s),{\alpha}(s))d \psi(s+t , x(s))\nonumber\\ & \quad + \frac{1}{2}\sigma^2d^2\psi(s+t , x(s))\big]ds \nonumber\\ & \ge \psi(t , x)+{\mathbb{e}}\big[-\int_0^{t - t } f(s+t,\nu(s+t),x(s),{\alpha}(s))ds\nonumber\\ & \quad+\int_0^{t - t}y(s+t,\nu(s+t))dy(s ) -\int_0^{t - t}r(s+t,\nu(s+t))dr(s)\big ] .\label{eq : eq624}\end{aligned}\ ] ] hence , with @xmath153 , @xmath261\\\notag & \ge \psi(t , x).\end{aligned}\ ] ] since @xmath262 are arbitrary , we get that @xmath263 for all @xmath264\times [ 0,l]$ ] .let @xmath265 be as in the statement of the lemma with @xmath257 replaced by @xmath113 .let @xmath214 be a solution of with @xmath223 replaced by @xmath265 given on some system @xmath145 . then a similar calculation using it s formula but with @xmath266 shows that @xmath267 for all @xmath209\times [ 0,l]$ ] .this shows that @xmath268 and so @xmath225 is the unique @xmath230 solution of . also , we have that @xmath265 as in the statement of the lemma is an optimal feedback control .since @xmath269 , the continuity of the map @xmath270 is immediate from the continuity of the map @xmath271 that was noted below assumption [ assumption1 ] .we now show that @xmath272 is the unique optimal control for .fix an optimal control @xmath273 given on some system @xmath145 .we claim that @xmath274 , @xmath275-a.s . indeed , suppose that there is a set with a positive @xmath276-measure on which the equality fails .then by assumption [ assumption1].(b ) together with , it follows that holds with a strict inequality , which in turn implies that holds with a strict inequality . recalling that @xmath277, we arrive at a contradiction .we now turn to step ( ii ) in solving the fixed point problem .although for existence of a fixed point assumption [ assumption1 ] will suffice , in order to get uniqueness , we will need the following additional assumption .similar assumption has been used to argue uniqueness of fixed points in previous works on mfg ( see e.g. ( * ? ? ?* theorem 2.4 ) , ( * ? ? ?* section 3 ) , ( * ? ? ?* equation ( 17 ) ) , ( * ? ? ?* assumption ( u ) ) ) . fix @xmath279)$ ] .[ assumptionu ] for every @xmath280\times{{\cal p}}([0,l])\times[0,l]\times u$ ] , @xmath281 moreover , for every @xmath117 $ ] and @xmath191)$ ] , @xmath282 and @xmath207 satisfy the following monotonicity property @xmath283d(\eta-\eta')(x)&\ge 0,\\\notag \int_0^l ( g(\eta , x)-g(\eta',x))d(\eta-\eta')(x)&\ge 0 .\end{aligned}\ ] ] abusing notation , when assumption [ assumptionu ] holds , we will write @xmath284 , @xmath285 , and @xmath286 . [ thm_fixed ] under assumption [ assumption1 ] , there exists a solution of the mfg .if in addition assumption [ assumptionu ] holds then there is a unique mfg solution .the proof will appeal to schauder s fixed point theorem . since schauder s original work ( cf .@xcite ) , there have been several versions of this result .we now quote the version ( ( * ? ? ?* theorem 4.1.1 ) ) that will be used here .[ lem_schauder ] let @xmath287 be a non - empty convex subset of a normed space @xmath288 .let @xmath171 be a continuous mapping of @xmath287 into a compact set @xmath289 .then @xmath171 has a fixed point , namely there exists @xmath290 such that @xmath291 .* proof of theorem [ thm_fixed ] : * we will apply lemma [ lem_schauder ] to the space @xmath288 of finite signed measures on @xmath17:[0,l])$ ] which is equipped with the kantorovich - rubinstein norm @xmath292:[0,l])}f(\omega)d\nu(\omega)\right|\;:\;f\in\text{lip}_1({{\cal c}}([0,t]:[0,l]))\right\},\ ; \nu\in{{\cal b}}.\end{aligned}\ ] ] the distance driven by the norm , restricted to @xmath293 , coincides with the wasserstein s distance of order 1 ( see ( * ? ? ?* remark 6.5 ) ) , which due to compactness of @xmath53 $ ] generates the same topology on @xmath293 as that for weak convergence .we now introduce a mapping @xmath171 on the non - empty and convex set @xmath294 that satisfies the conditions stated in lemma [ lem_schauder ] .* definition of @xmath171 . * for @xmath242 ,let @xmath295 denote the optimal feedback control @xmath265 for ( with @xmath167 ) given through proposition [ lem2 ] .let @xmath296 denote the unique weak solution of with @xmath223 replaced with @xmath295 given on some system @xmath297 .define @xmath298 .* invariance of @xmath294 . * for @xmath242 and @xmath299 @xmath300 where @xmath301 , and the last inequality uses doob s maximal inequality .this shows that @xmath302 for all @xmath242 .we now show that @xmath303 is contained in a compact set in @xmath293 , i.e. @xmath303 is relatively compact .* relative compactness of @xmath303 . * for @xmath304:{\mathbb{r}})$ ] and @xmath305 ,let @xmath306:|t - s|\le \delta}|f(t)-f(s)|.\ ] ] then for @xmath307 and @xmath203 , similar to the estimate in , @xmath308 since @xmath309 as @xmath310 , we have from ( * ? ? ?* theorem 7.3 ) that @xmath311 is relatively compact in @xmath293 .* continuity of @xmath171 .* we now argue that @xmath171 is a continuous map on @xmath294 .consider a system @xmath297 and let @xmath312 be given as @xmath313 define for @xmath314 @xmath315.\end{aligned}\ ] ] let @xmath316 be such that @xmath317 .since @xmath318 , the above convergence implies that @xmath319 we now show that @xmath320 in @xmath321-measure . recall that @xmath244 is a continuous map .hence , in view of , for the desired convergence , it is sufficient to show that @xmath322=0.\end{aligned}\ ] ] the proof is a modification of the proof of theorem 2.1 in @xcite . applying it s lemma to @xmath323 for fixed @xmath324 and using from proposition [ lem2 ] the fact that @xmath325 solves , we have for every @xmath99 $ ] , @xmath326 let , @xmath327 then , and imply , @xmath328ds\\\notag & \qquad+\int_t^t[h(s,\nu^n(s),x^0(s),dv_{\nu}(s , x^0(s)))-h(s,\nu(s),x^0(s),dv_{\nu}(s , x^0(s)))]ds.\end{aligned}\ ] ] squaring both sides and then taking expectations gives @xmath329 + \sigma^2{\mathbb{e}}\left[\int_t^t(\delta d v^n(s))^2ds\right]\\\notag & \quad\le 2(t - t){\mathbb{e}}\big[\int_t^t[h(s,\nu^n(s),x^0(s),dv_{\nu^n}(s , x^0(s)))-h(s,\nu^n(s),x^0(s),dv_{\nu}(s , x^0(s)))]^2ds\big]\\\notag & \qquad + 2c^n(t)\\\notag & \quad\le 2c_b^2(t - t){\mathbb{e}}\big[\int_t^t(\delta dv^n(s))^2ds\big]+ 2c^n(t ) , \end{aligned}\ ] ] where @xmath330ds\big]^2.\end{aligned}\ ] ] letting @xmath331 , we get from that for every @xmath332 $ ] , @xmath333+\frac{\sigma^2}{2}{\mathbb{e}}\left[\int_t^t(\delta dv^n(s))^2ds\right]\le 4c^n(t ) .\end{aligned}\ ] ] we now show that @xmath334 . note that there exists @xmath335 such that the following inequality holds for all @xmath336 .@xmath337+{\mathbb{e}}\left[(y^0(t))^2\right]\sup_{0\le t\le t}(\delta y^n(t))^2+{\mathbb{e}}\left[(r^0(t))^2\right]\sup_{0\le t\le t}(\delta r^n(t))^2\\ & \quad+{\mathbb{e}}\left [ \int_0^t(\delta f^n(s))^2ds\right]+{\mathbb{e}}\left[\int_0^t(\delta b^n(s))^2ds \int_0^t ( dv_\nu(s , x^0(s)))^2ds\right].\label{eq : eq1259 } \ ] ] using the properties of the skorohod map ( lemma [ lem_skorokhod ] ) it follows that @xmath338<{\infty}.\ ] ] the convergence of the right side of to @xmath71 is now immediate from assumption [ assumption1](a ) , the boundedness of @xmath339 , and .thus from we have that @xmath340=0,\ ; { \lim_{n\to{\infty}}}{\mathbb{e}}[(\delta v^n(t-\delta))^2 ] = 0.\end{aligned}\ ] ] using the second convergence in and repeating the above argument for @xmath341 $ ] to the interval @xmath342 $ ] , we see that holds with @xmath52 replaced with @xmath343 . proceeding recursively in this manner we have .hence we have shown that @xmath344 using the above property we will now argue that @xmath345 as @xmath7 , completing the proof of continuity of @xmath171 .let for @xmath314 , @xmath346\times [ 0,l]\ ] ] and let @xmath347 be a probability measure on @xmath348 defined as @xmath349 d{{\mathbb p}}.\end{aligned}\ ] ] by girsanov s theorem @xmath350 .thus to show @xmath345 it suffices to argue that @xmath351 . we will in fact show that @xmath352 as @xmath7 , where @xmath353 is the relative entropy of @xmath354 with respect to @xmath355 , which due to pinsker s inequality ( see ( * ? ? ?* page 132 ) ) gives the convergence of @xmath355 to @xmath354 .let @xmath356\times [ 0,l].\ ] ] with this notation , @xmath357 also , noting that since @xmath358 is bounded @xmath359 , we have from cauchy - schwarz inequality @xmath360 ^ 2\right)^{1/2 } .\ ] ] next note that @xmath361 ^ 2 \\ & \quad = \frac{1}{\sigma^2 } \int_0^t { \mathbb{e}}[\delta \hat b_{n}(t , x^0(t))]^2 dt\\ & \quad = \frac{1}{\sigma^2 } \int_0^t { \mathbb{e}}\left[b(t , \nu(t ) , x^0(t ) , \alpha^0_{\nu}(t , x^0(t ) ) ) - b(t , \nu^n(t ) , x^0(t ) , \alpha^0_{\nu^n}(t , x^0(t ) ) ) \right]^2 dt.\end{aligned}\ ] ] the last term converges to @xmath71 from the boundedness and continuity of @xmath358 , and . similarly @xmath362 ^ 2 \to 0\ ] ] as @xmath111 . using the above two observations in we have @xmath352 and thus , as argued earlier , the proof of continuity of @xmath171 is complete .thus we have shown that @xmath171 is a continuous map on @xmath294 , which is a non - empty convex subset of the normed space @xmath288 , into a compact set @xmath363 .thus by the fixed point theorem in lemma [ lem_schauder ] , @xmath171 has a fixed point .+ the first results on unique solvability of a mfg go back to @xcite . since then uniqueness has been argued in various settings ( see e.g. @xcite ) .the proof given below uses arguments similar to those in ( * ? ? ?* section 7.3 ) , however for the sake of completeness we give the details .consider as before a system @xmath297 and let @xmath312 be given through .let @xmath364 .for @xmath365 , let @xmath366 and @xmath367 be given by and , respectively , with @xmath368 replacing @xmath369 .applying it s lemma to @xmath370 and recalling that @xmath371 solves with @xmath157 replaced with @xmath372 , we get @xmath373 where recalling the form of @xmath374 and @xmath136 from assumption [ assumptionu ] @xmath375 } dv_{\nu_1}(t,0 ) dy^0(t ) + \int_{[0,t ] } dv_{\nu_1}(t , l ) dr^0(t)\\ & = \int_{[0,t ] } y(t ) dy^0(t ) + \int_{[0,t ] } r(t ) dr^0(t).\end{aligned}\ ] ] observing that for @xmath23 $ ] @xmath376 where @xmath377 \hat b_{\nu_1}(t , x^0(t)),\ ] ] we have @xmath378 similarly , applying it s lemma to @xmath379 @xmath380 substracting from @xmath381 db^{\nu_1}(t)\\ & \quad + \int_0^t \left[h(t , \nu_1(t ) , x^0(t ) , \alpha^0_{\nu_1}(t ) , dv_{\nu_2}(t , x^0(t ) ) ) - h(t , \nu_2(t ) , x^0(t ) , dv_{\nu_2}(t , x^0(t)))\right ] dt,\end{aligned}\ ] ] where for @xmath365 , @xmath382 , @xmath99 $ ] .since under @xmath367 , @xmath383 is a standard brownian motion , taking expectation under the measure @xmath384 @xmath385dt\notag\\ & \qquad\qquad\qquad+g(\nu_1(t),x^0(t))-g(\nu_2(t),x^0(t ) ) .\big]\label{up3 } \ ] ] a similar calculation shows @xmath386dt \notag\\ & \qquad\qquad\qquad+g(\nu_1(t),x^0(t))-g(\nu_2(t),x^0(t ) ) \big].\label{up3b}\end{aligned}\ ] ] by the definition of the @xmath184 and the form of @xmath387 in assumption [ assumptionu ] we get , @xmath388 where the last equality uses the observation that since @xmath389 does not depend on @xmath201 , @xmath390 .therefore for all @xmath391 $ ] , @xmath392 similarly for all @xmath391 $ ] , @xmath393 applying the last two inequalities to and , we get @xmath394\big[&g(\nu_1(t),x^0(t))-g(\nu_2(t),x^0(t))\\\notag & \quad+\int_0^t[f_0(s,\nu_1(s),x^0(s))-f_0(s,\nu_2(s),x^0(s))]ds\big].\end{aligned}\ ] ] until now @xmath372 and @xmath395 were arbitrary measures in @xmath294 .suppose now that @xmath368 , @xmath365 , are fixed points of @xmath171 .then , for @xmath365 , @xmath396 and so for all @xmath397 $ ] , @xmath398 . in this case, using the inequalities in assumption [ assumptionu ] we get that the inequality can be replaced with equality .we claim that @xmath399 , @xmath400-a.s . indeed , suppose that there is a set with positive @xmath400-measure on which @xmath401 . then on this set and will hold with strict inequalities by assumption [ assumption1 ] ( b ) .since the measures @xmath402 , @xmath384 , and @xmath403 are equivalent , this will say that holds with a strict inequality as well , which contradicts the equality that was established above .this proves the claim .since @xmath404 does not depend on @xmath201 we conclude from the equality of @xmath405 and @xmath406 that @xmath407 , @xmath400-a.s . and thus by , @xmath408 .combining this with the fact that @xmath368 are fixed points of @xmath171 , we now have @xmath409 .the main result of this section is theorem [ thm1 ] .the main idea in the proof is to use a solution @xmath169 to the mfg ( which from theorem [ thm_fixed ] exists under condition [ assumption1 ] ) and the associated feedback control given by proposition [ lem2 ] in order to construct an admissible control @xmath410 for the @xmath0-player game .specifically , the control will be given in a feedback form through the following relation @xmath411 where @xmath76 is the corresponding scaled queue length under the feedback control .our main condition , in addition to assumption [ assumption1 ] , for @xmath412 to be an asymptotic nash equilibrium is the following .it in particular says that the drift function does not depend on the mean - field term .fix @xmath279)$ ] .[ assumption4 ] @xmath188 1 . for every @xmath117 $ ] , @xmath413)$ ], @xmath414 $ ] , and @xmath415 , one has , @xmath416 2 .there exists @xmath414 $ ] such that for every @xmath417 , @xmath418 as before , with an abuse of notation , we will write @xmath419 as @xmath420 when assumption [ assumption4 ] holds . recall the probability space @xmath421 from section [ sec2a ] . let for @xmath336 , @xmath23 $ ] and @xmath422 , @xmath423 be such that @xmath424 is an admissible control ( i.e. @xmath425 ) .the following is the main result of the section .[ thm1 ] suppose assumptions [ assumption1 ] and [ assumption4 ] hold .let @xmath169 , @xmath426 and @xmath427 , @xmath428 , @xmath336 be as introduced above .then @xmath429 theorem [ thm1 ] in particular says that for every @xmath430 , there is @xmath51 sufficiently large such that @xmath426 forms an @xmath431-nash equilibrium in the @xmath0-player game . the proof is given in the next three sections .first in section [ 4z ] ( proposition [ prop : tightemp ] ) we will prove the convergence of empirical measures of the scaled queue length processes under controls @xmath432 and @xmath433 to @xmath169 .next , in section [ sec4a ] ( proposition [ prop2 ] ) we will prove the first equality in and finally proposition [ prop3 ] in section [ sec4b ] will prove the inequality in .let for @xmath434 , @xmath435 be as defined below assumption [ assumption4 ] .let @xmath436 where @xmath437 is the controlled queue length process defined by with @xmath438 replaced with @xmath427 .the following result gives the convergence of @xmath439 to @xmath169 .[ prop : tightemp ] suppose assumptions [ assumption1 ] and [ assumption4 ] hold .then for every @xmath440 , @xmath439 converges in probability , in @xmath441:[0,l]))$ ] , to @xmath169 as @xmath111 . *proof : * without loss of generality we assume that @xmath442 . recall that for @xmath443 , @xmath437 is defined by with @xmath438 replaced with @xmath444 .define @xmath445,\ ] ] where @xmath446 are as in .define for @xmath447 , @xmath448 and let @xmath449 note that since by assumption @xmath450 are exchangeable and the controls @xmath451 are given in terms of the same feedback function @xmath244 for each @xmath452 , the processes @xmath453 are exchangeable . defining @xmath454 to be the zero process for @xmath455 we can regard , @xmath456 as a random variable with values in @xmath19 : ( { \mathbb{r}}^4)^{\otimes\infty})$ ] .we now argue the tightness of the sequence @xmath457 .it suffices to show for each @xmath58 , the tightness of @xmath458 in @xmath19 : { \mathbb{r}}^4)$ ] . since @xmath459 } \frac{{\lambda}^{n , i}(t,\omega)+ \mu^{n , i}(t,\omega)}{n } \doteq c_0 < \infty , \label{eq : eq858}\ ] ] the following two conditions are satisfied with @xmath460 equal to @xmath461 and @xmath462 for all @xmath463 [ a ] for each @xmath464 there exists a @xmath465 and @xmath466 with the property that for every family of stopping times @xmath467 ( @xmath468 being an @xmath142-stopping time on @xmath421 ) with @xmath469 , @xmath470 [ t@xmath471 for every @xmath64 in some dense subset of @xmath21 $ ], @xmath472 is a tight sequence of @xmath55 valued random variables .then by rebolledo s theorem ( see ( * ? ? ? * theorem 2.3.2 ) ) @xmath473 and @xmath474 are tight in @xmath19:{\mathbb{r}})$ ] for each @xmath58. also since the jumps of @xmath475 and @xmath476 are of size @xmath477 , these processes are @xmath478-tight ( namely all weak limit points are continuous a.s . ) . from boundedness of@xmath358 we see that @xmath479 is tight in @xmath17:{\mathbb{r}})$ ] . combining this with assumption [ assumption4](b )we see that @xmath480 is @xmath478-tight in @xmath19:{\mathbb{r}})$ ] .using now the continuity of the skorohod map ( lemma [ lem_skorokhod ] ) we have the desired tightness of @xmath458 .suppose now that , along some subsequence , @xmath481 converges to @xmath482 , in distribution , in @xmath19 : ( { \mathbb{r}}^4)^{\otimes\infty})$ ] .then @xmath483 : ( { \mathbb{r}}^4)^{\otimes\infty})$ ] a.s . and from ( * ? ? ?* lemma 4.2 ) and the exchangeability of @xmath453 it follows that @xmath484 is exchangeable and(along the subsequence ) , @xmath485 in @xmath19 : ( { \mathbb{r}}^4)^{\otimes\infty})\times { { \cal p}}({{\cal d}}([0,t]:{\mathbb{r}}^4))$ ] where @xmath486. we will now characterize the distribution of @xmath487 . from tightness of @xmath488 and@xmath489 argued above and it follows that @xmath490 uniformly on @xmath21 $ ] , in probability .also , from it follows that @xmath491 \to 0 \mbox { a.s . } \ ] ] as @xmath7 .thus from ( and the relation @xmath492 ) , for all @xmath463 , @xmath493 in probability , uniformly on @xmath21 $ ] , as @xmath7 . by standard martingale techniquesit now follows that @xmath494 in @xmath495:{\mathbb{r}}^{\infty})$ ] , where @xmath496 are mutually independent standard brownian motions .also , since for @xmath497 @xmath498,\ ] ] where @xmath499 we have from the continuity of @xmath358 ( assumption [ assumption1 ] ) , @xmath244 , and @xmath500 , for @xmath497 , @xmath501 once again using standard martingale arguments it follows that for @xmath502 , @xmath503 is independent of @xmath504 . from weak uniqueness property noted in section [ sec3b1 ] andthe fact that @xmath169 is a fixed point of @xmath171 we now have that @xmath505 has distribution @xmath169 for @xmath506 . using the fact that @xmath496 are mutually independent , a simple argument based on girsanov stheorem shows that @xmath487 are mutually independent as well .this characterize the distribution of @xmath507 as @xmath508 .we now have from , the definition of @xmath145 , and the law of large numbers that @xmath509 the result follows .@xmath278 [ rem : rem640 ] the above proof also shows that if @xmath510 , namely @xmath511 , then holds for all @xmath512 and the law of @xmath513 is @xmath508 . in this sectionwe prove the equality in .[ prop2 ] suppose assumptions [ assumption1 ] and [ assumption4 ] are satisfied .let @xmath514 be as in .then for all @xmath512 @xmath515 * proof : * without loss of generality we assume @xmath442 . from the proof of proposition [ prop : tightemp ] ( see remark [ rem : rem640 ] ) @xmath516 where the processes on the right side are given through with @xmath442 .let @xmath517.\ ] ] recall that @xmath387 and @xmath207 are bounded continuous functions and from proposition [ prop : tightemp ] we have that , for every @xmath23 $ ] , @xmath518 this shows that @xmath519\nonumber\\ & \quad \to { \mathbb{e}}\left[\int_0^t f(t,\bar \nu(t ) , \tilde q^1(t ) , \gamma(t , \tilde q^1(t)))dt + g(\bar \nu(t),\tilde q^1(t))\right].\label{eq : eq656}\end{aligned}\ ] ] also by continuity of @xmath374 and @xmath136 @xmath520 in @xmath495:{\mathbb{r}}^4)$ ] .it then follows ( cf .* lemma 2.4 ) ) @xmath521 as @xmath7 .also from lemma [ lem_skorokhod ] , @xmath522 \le c_s^2 { \mathbb{e}}\left [ \sup_{0\le s \le t } |\tilde \zeta^{n,1}(s)|^2\right].\ ] ] next note that @xmath523 where @xmath524 is as in .similarly , @xmath525 . combining these estimates @xmath526 < \infty\ ] ] which combined with and the boundedness of @xmath527 implies @xmath528 ^ 2 < \infty.\ ] ] combining this with the weak convergence in we have @xmath529\nonumber \\ & \quad \to { \mathbb{e}}\left[\int_0^ty(t,\bar \nu(t))d\tilde y^1(t)+ \int_0^tr(t,\bar \nu(t))d\tilde r^1(t)\right].\label{eq : eq908}\end{aligned}\ ] ] combining and and recalling from the optimality of @xmath244 that @xmath530,\end{aligned}\ ] ] we have the desired convergence @xmath531 .@xmath278 in this section we prove the inequality on the right side of .[ prop3 ] suppose assumptions [ assumption1 ] and [ assumption4 ] hold .let @xmath532 be as introduced at the beginning of section [ sec4 ] .then for each @xmath512 @xmath533 * proof : * as before , we assume without loss of generality that @xmath442 . we will need to argue the tightness of the control sequence @xmath534 in an appropriate space . for thisit will be convenient to consider a relaxed control formulation .consider the relaxation of the stochastic control problem in where the control space @xmath54 is replaced by @xmath535 , the drift function @xmath358 is replaced by the function @xmath536\times [ 0,l]\times { { \cal p}}(u ) \to { \mathbb{r}}$ ] defined as @xmath537\times [ 0,l]\times { { \cal p}}(u),\ ] ] and the running cost function @xmath387 is replaced by @xmath538\times { { \cal p}}([0,l])\times [ 0,l]\times { { \cal p}}(u ) \to { \mathbb{r}}$ ] , defined as @xmath539\times { { \cal p}}([0,l])\times [ 0,l]\times { { \cal p}}(u).\ ] ] also , the class of admissible controls @xmath540 is replaced by @xmath541 of pairs @xmath542 that are similar to pairs @xmath149 introduced above except that @xmath543 is @xmath535 valued rather than @xmath54 valued and holds with @xmath544 replaced with @xmath545 .the corresponding cost function @xmath546 is defined by with @xmath387 replaced by @xmath547 .the value function in this relaxed formulation , denoted as @xmath548 , is given by with @xmath549 replaced by @xmath550 .define the function @xmath551 by , replacing @xmath552 with @xmath553 .then , from assumption [ assumption1](b ) , @xmath554 this shows that @xmath257 and @xmath555 are both solutions of the pde - . in view of the uniqueness result from proposition [ lem2 ] , @xmath556 .let @xmath557 , @xmath23 $ ] and define @xmath558)$ ] as @xmath559 where @xmath560)$ ] is the space of finite measures on @xmath561 $ ] equipped with the topology of weak convergence .then we can rewrite @xmath562 } f(t,{\tilde\nu^{n}}(t ) , \tilde q^{n,1}(t ) , u ) \bar \beta^n_{{{\cal r}}}(du\,dt ) + g({\tilde\nu^{n}}(t),\tilde q^{n,1}(t))\\\notag & \quad-\int_0^ty(t,{\tilde\nu^{n}}(t))d\tilde y^{n,1}(t)+\int_0^tr(t,{\tilde\nu^{n}}(t))d\tilde r^{n,1}(t)\big]\ ] ] and @xmath563}b(s , \tilde q^{n,1}(s ) , u ) \bar \beta^n_{{{\cal r}}}(du\,ds ) + { \tilde a^{n , i}}(\cdot)-{\tilde d^{n , i}}(\cdot ) + o(1)\right).\end{aligned}\ ] ] from proposition [ prop : tightemp ] , @xmath564 .also , the arguments of the same proposition show that @xmath565 } b(s , \tilde q^{n,1}(s ) , u ) \bar \beta^n_{{{\cal r}}}(du\,ds ) \right\}_{n\in { \mathbb{n}}}\ ] ] are @xmath478-tight in @xmath19 : { \mathbb{r}}^3)$ ] .furthermore , since @xmath566 $ ] is compact and @xmath567 ) = t$ ] , the sequence @xmath568 is tight in @xmath569)$ ] .suppose now that along a subsequence ( labeled once more as @xmath570 ) the sequence in along with @xmath571 converges in distribution to @xmath572 then from the lipschitz property of @xmath358 ( assumption [ assumption1](a ) ) we have , for @xmath99 $ ] , @xmath573 } b(s , \tilde q^1(s ) , u ) \bar \beta_{{{\cal r}}}(du ds)\\ & = \int_{0}^t b_{{{\cal r}}}(s , \tilde q^1(s ) , \beta_{{{\cal r}}}(s ) ) ds,\end{aligned}\ ] ] where @xmath574 is obtained by disintegrating @xmath575 , i.e. @xmath576 .also as in the proof of proposition [ prop : tightemp ] it can be argued that @xmath577 is a standard brownian motion and thus we can conclude as in the proof of @xmath578.\ ] ] once again , by a standard martingale argument , one can argue that for @xmath579 , @xmath580 is independent of @xmath581 ) : 0\le s'\le s,\ , a \in { { \cal b}}(u)\right\}.\ ] ] thus denoting by @xmath582 the probability space on which the limit processes are defined , @xmath583 is a system and @xmath584 .exactly as in the proof of proposition [ prop2 ] we see that the convergence in holds . also using the weak convergence @xmath585 and the lipschitz property of @xmath387 in assumption [ assumption1 ], we have @xmath586 } f(t,{\tilde\nu^{n}}(t ) , \tilde q^{n,1}(t ) , u)\bar \beta^n_{{{\cal r}}}(du\ , dt ) + g({\tilde\nu^{n}}(t),\tilde q^{n,1}(t))\right]\nonumber\\ & \quad \to { \mathbb{e}}\left[\int_0^t f(t,\bar \nu(t ) , \tilde q^1(t ) , u)\bar \beta_{{{\cal r}}}(du\ , dt ) + g(\bar \nu(t),\tilde q^1(t))\right].\label{eq : eq504}\end{aligned}\ ] ] combining the above convergence properties , we have as @xmath111 , @xmath587 since the above holds for an arbitrary convergent subsequence of processes in and the sequence @xmath571 , the result follows .construction of an asymptotic nash equilibrium for the queuing system in section [ sec2a ] requires the solution of the mfg introduced in section [ sec3 ] . in general , tractable closed form expressions for the mfg solution are not available and thus one needs suitable numerical approximations . in this sectionwe will use the markov chain approximation method ( @xcite ) to construct numerical solutions of the mfg .the main result of the section is theorem [ thm : thm52 ]. we will assume here that assumption [ assumption1 ] is satisfied . in additionwe will make the following assumption .fix @xmath146 $ ] .[ rem : rem629 ] recall from section [ sec3b2 ] that assumption [ assu : uniqmfg ] is satisfied if in addition to assumption [ assumption1 ] , assumption [ assumptionu ] holds .also , from assumptions [ assumption1 ] and [ assu : uniqmfg ] it follows from arguments of section [ sec3b1 ] ( see proof of proposition [ lem2 ] ) that there is a continuous map @xmath588 \times [ 0,l ] \to u$ ] such that if there exist a system @xmath145 and an @xmath589 such that @xmath590,\ ] ] @xmath591 , and @xmath592 then @xmath593 , @xmath594-a.s .fix @xmath596 such that @xmath597 is an integer multiple of @xmath598 .the parameter @xmath598 is the approximation parameter .denote the @xmath598-grid @xmath599 by @xmath600 .this is a discretized version of the state space @xmath53 $ ] .since @xmath71 and @xmath597 are reflecting barriers for the state process @xmath32 in section [ sec3a ] , we will consider two types of transition steps for the approximating chain .the first , which occurs when the chain is away from the boundary , will be referred to as the _ rate control step _ and the second occurs at the end points @xmath601 and @xmath602 and is referred to as the _ reflection step_. also , define @xmath610 this will be used to define the continuous time interpolation of the controlled markov chain .note that @xmath611 as @xmath612 .one can verify that for any @xmath613 the following _ local consistency _ conditions ( cf .@xcite ) hold for every @xmath604 .@xmath614 * reflection step .* such a step occurs only when @xmath615 . for every @xmath117 $ ] , @xmath413)$ ] , and @xmath603 let @xmath616 we will now define a sequence of controlled markov chains @xmath617 , one for each value of the discretization parameter @xmath618 , where @xmath619 as @xmath620 .we denote @xmath621 and @xmath622 .we will assume that @xmath623 and @xmath624 are integers for each @xmath625 .* define @xmath632 , @xmath633 and let @xmath634 be a fixed element of @xmath54 .* having defined for @xmath635 time instants @xmath636 and random variables @xmath637 with values in @xmath638 and @xmath54 respectively , let @xmath639 . *choose the control @xmath640 for the @xmath0-th step that is a @xmath54 valued @xmath641 measurable random variable and let @xmath642 be such that the conditional distribution of @xmath642 given @xmath643 is @xmath644 where @xmath645 denotes the marginal distribution of @xmath646 at time instant @xmath64 . also define @xmath647 where the indicator in the above definition will ensure that when we do a continuous time interpolation of the chain , reflection steps ` occur instantaneously ' .* define for @xmath648 , @xmath649 .let @xmath650}$ ] be the continuous stochastic process which is linear on @xmath651 $ ] and equals @xmath652 at @xmath653 , for @xmath654 .let @xmath655 be the probability law of @xmath656 .* define for @xmath657 , @xmath658 and @xmath659 recursively , using the above steps by replacing @xmath660 with @xmath661 .* some auxiliary processes . * we will now introduce sequences of processes that will be useful for the convergence analysis .along with the piecewise linear process @xmath662 , consider the piecewise constant processes @xmath663,\ ] ] where @xmath664 let @xmath665 and for every @xmath117 $ ] let @xmath6661_{\{x^{k}_j\notin\{-h^k , l+h^k\}\}},\\\label{eq : eq907 } b^{k}(t)&\doteq\frac{1}{\sigma}\sum_{j=0}^{n^{k}(t)-1}\big(x^{k}_{j+1}-x^{k}_j-{\mathbb{e}}[x^{k}_{j+1}-x^{k}_j\;|\;{{\cal f}}^{k}_j]\big)1_{\{x^{k}_j\notin\{-h^k , l+h^k\}\}},\\\notagy^{k}(t)&\doteq\sum_{j=0}^{n^{k}(t)-1}(x^{k}_{j+1}-x^{k}_j)1_{\{x^{k}_j =- h^k\ } } = h^k\sum_{j=0}^{n^{k}(t)-1}1_{\{x^{k}_j =- h^k\}},\\\notagr^{k}(t)&\doteq\sum_{j=0}^{n^{k}(t)-1}(x^{k}_{j}-x^{k}_{j+1})1_{\{x^{k}_j = l+h^k\ } } = h^k\sum_{j=0}^{n^{k}(t)-1}1_{\{x^{k}_j = l+h^k\}}.\end{aligned}\ ] ] one can verify that the following representation holds @xmath667.\end{aligned}\ ] ] also , from and it follows that , on the set @xmath668@xmath669=b(t^{k}_n,\nu^k(t^{k}_n),x^{k}(t^{k}_n),{\alpha}^{k}(t^{k}_n))\delta^{k}+o((h^k)^\rho\delta^{k}),\\\notag&{\mathbb{e}}\left[\left[x^{k}_{n+1}-x^{k}_n-{\mathbb{e}}[x^{k}_{n+1}-x^{k}_n\;|\;{{\cal f}}^{k}_n ] \right]^2\mid { { \cal f}}^{k}_n \right]= \delta^{k}+o((h^k)^\rho\delta^{k}).\end{aligned}\ ] ] define @xmath670 .then since @xmath671 for each fixed @xmath64 is a @xmath672 stopping time , we have by optional sampling theorem that @xmath673 is a @xmath674 martingale . using assumption [ assumption1 ]it now follows that @xmath675 where @xmath676 and @xmath677 are @xmath678-adapted processes that satisfy @xmath679 * cost function for the mdp . * for @xmath626 and the admissible control @xmath680 used to construct the @xmath625-th controlled markov chain , define the associated cost @xmath681.\end{aligned}\ ] ] we note that in construction [ const : const1 ] , @xmath682 was given recursively as the law of @xmath683 .however , in a similar way one can define controlled markov chains associated with an arbitrary sequence @xmath684 in @xmath36 .the cost associated with the @xmath625-th chain in such a sequence with some control @xmath685 is defined similarly and will be denoted as @xmath686 .the cost function in can be written in terms of the original discrete time controlled markov chain and the control sequence @xmath687 .since this is a finite state discrete time markov decision problem , a solution can be found by standard dynamic programming principle arguments .we denote the optimal control sequence by @xmath688 and denote the interpolated continuous time control process as @xmath689. we will continue to denote the corresponding state , reflection and noise processes as @xmath690 , however the corresponding law of the optimal state process ( with piecewise linear interpolation ) @xmath662 will be denoted as @xmath691 to note its dependence on the choice of the control sequence .we can now present the main result of this section .the statement of the theorem and its proof use relaxed controls .recall the definitions of @xmath692)$ ] , @xmath693 , @xmath694 , and @xmath695 , given in the proof of proposition [ prop3 ] . define @xmath569)$ ]valued random variable @xmath696 as @xmath697 [ thm : thm52 ] suppose assumptions [ assumption1 ] and assumption [ assu : uniqmfg ] hold .then the sequence @xmath698 introduced in the above paragraph converges in distribution to @xmath699 in @xmath19:{\mathbb{r}}^4)\times { { \cal m}}(u \times [ 0,t])$ ] and @xmath700 in @xmath36 where the limit processes defined on some probability space @xmath701 satisfy the following .( a ) : : @xmath143 is a @xmath702 ) : s \le t , a \in { { \cal b}}(u)\}$ ] brownian motion and so @xmath703 is a system .( b ) : : disintegrating @xmath704 , the following relationship holds a.s .@xmath705.\ ] ] ( c ) : : @xmath706 .( d ) : : @xmath707 , the unique solution of the mfg .( e ) : : the pair @xmath708 and @xmath709 . in particular , with @xmath223 as in remark [ rem :rem629 ] , @xmath710 .* proof : * since many steps in the proof are quite standard we will only provide details where appropriate . using properties and of the controlled transition probability kernel it can be argued ( cf .proof of ( * ? ? ? * theorem 9.4.1 ) ) that @xmath711 are tight in @xmath19:{\mathbb{r}}^2)$ ] . using this tightness property along with the continuity of the skorohod map ( lemma [ lem_skorokhod ] )it now follows that @xmath712 is tight in @xmath19:{\mathbb{r}}^4)$ ] .in fact since @xmath619 as @xmath629 , the processes are @xmath478-tight .next note that , since @xmath713 ) = t$ ] and @xmath54 is compact , the sequence @xmath714 is tight in @xmath569)$ ] .also , recalling the definition of the interpolated processes it can be checked that @xmath715 combining this with the tightness of @xmath716 and the fact that @xmath717 gives the relative compactness of @xmath718 in @xmath36 .suppose now that along a subsequence ( relabeled again as @xmath719 ) @xmath720 and @xmath721 by standard martingale methods it follows that @xmath143 is a @xmath722 brownian motion ( see e.g. proof of ( * ? ? ? * theorem 9.4.1 ) ) proving part ( a ) of the theorem .using and , @xmath723 converges , along with the above processes , in distribution to @xmath724 } b(s , \nu(s ) , x(s ) , u ) m(du\ , ds ) = \int_{0}^{\cdot } b_{{{\cal r}}}(s , \nu(s ) , x(s ) , m_s ) ds,\ ] ] where @xmath725 . using the continuity property of the skorohod map we now get ( b ). clearly @xmath726 .we will now argue that @xmath727 this will prove that @xmath157 solves the mfg and thus from assumption [ assu : uniqmfg ] we will get part ( d ) for this limit point . in view of remark [ rem : rem629 ] it will also show ( e ) for the limit point . since the convergent subsequence was arbitrary , we will get the convergence asserted in the statement of the theorem and complete the proof .* proof of .* for the rest of the proof we consider the subsequence along which the convergence in and holds .using arguments similar to those in proposition [ prop2 ] it can be checked that @xmath728 next , let @xmath729 for some system @xmath730. we will now show that for every @xmath731 there is a sequence of controls @xmath732 for the controlled markov chains defined as in construction [ const : const1 ] but with the sequence @xmath733 in the construction replaced by the sequence @xmath734 introduced above theorem [ thm : thm52 ] , such that @xmath735 note that from the optimality property of @xmath736 @xmath737 combining the above inequality with and we now get that @xmath738 since @xmath739 , the system @xmath740 , and @xmath729 are arbitrary we have , completing the proof of the theorem .we now prove . using arguments as in the proof of ( * ? ? ?* theorem 10.3.1 ) it can be shown that there is a @xmath741 such that @xmath742 as @xmath743 and for every @xmath430 there is a system @xmath744 and @xmath745 with the following properties * @xmath746 satisfies the following equation for @xmath23 $ ] . @xmath747 * for some @xmath748 , @xmath749 is piecewise constant on intervals of the form @xmath750 , @xmath751 .for some finite set @xmath752 , @xmath753 takes values in @xmath754 for every @xmath397 $ ] . * for some @xmath755 , for each @xmath756 @xmath757 where for suitable @xmath758 , @xmath759 $ ] is a measurable function such that @xmath760 is continuous on @xmath761 for every @xmath762 .* letting @xmath763 , @xmath764 , as @xmath765 , @xmath766 in @xmath19:{\mathbb{r}}^4)\times { { \cal m}}(u\times [ 0,t])$ ] .* @xmath767 .we will now use the piecewise constant control @xmath768 to construct a collection of control sequences @xmath732 as stated above and for which @xmath769 note that since @xmath770 as @xmath765 , this will prove and complete the proof of the theorem .the construction is carried out as follows .* define @xmath771 , @xmath633 and let @xmath772 be a fixed element of @xmath754 . *having defined for @xmath635 time instants @xmath773 and random variables @xmath774 with values in @xmath775 and @xmath754 respectively , let @xmath776 .* choose the control @xmath777 for the @xmath0-th step that is a @xmath754 valued @xmath778 measurable random variable as follows : * * if @xmath779 and @xmath780 both lie in @xmath781 for some @xmath138 , set @xmath782 . * * if @xmath783 for some @xmath784 , choose @xmath785 according the the conditional distribution @xmath786 where @xmath787 is defined as in , @xmath788 , and @xmath789 is as in .* let @xmath790 be such that its conditional distribution given @xmath791 is @xmath792 where @xmath691 is as introduced above theorem [ thm : thm52 ] .also define @xmath793 now define processes @xmath794 as below .exactly as in the first part of the proof we now have that @xmath712 is @xmath478-tight in @xmath19:{\mathbb{r}}^4)$ ] and the sequence @xmath795 , where @xmath796 is tight in @xmath569)$ ] . arguing as before , if along a further subsequence the convergence holds ( with @xmath696 replaced with @xmath797 ) then parts ( a ) and ( b ) as in the statement of theorem [ thm : thm52 ] are satisfied with @xmath157 as in . using the continuity property of @xmath798 and the fact that the control is piecewise constant with values in a finite set it follows that ( cf .proof of ( * ? ? ?* theorem 10.5.2 ) ) @xmath799 has the same distribution as @xmath800 . by unique solvability of , that follows from the lipschitz property of @xmath358 ( assumption [ assumption1 ] ) and the lipschitz property of the skorohod map ( lemma [ lem_skorokhod ] ) we now have that @xmath801 has the same law as @xmath802 for every limit point of the chosen further subsequence .since the chosen further subsequence was arbitrary , this proves the weak convergence of @xmath803 to @xmath804 along the subsequence fixed above and arguing once again as in the proof of proposition [ prop2 ] we have the convergence of costs as in , completing the proof of the theorem .

<S> we consider a large queueing system that consists of many strategic servers that are weakly interacting . </S><S> each server processes jobs from its unique critically loaded buffer and controls the rate of arrivals and departures associated with its queue to minimize its expected cost . </S><S> the rates and the cost functions in addition to depending on the control action , can depend , in a symmetric fashion , on the size of the individual queue and the empirical measure of the states of all queues in the system . in order to determine an approximate nash equilibrium for this finite player game we construct a lasry - lions type mean - field game ( mfg ) for certain reflected diffusions that governs the limiting behavior . under conditions , </S><S> we establish the convergence of the nash - equilibrium value for the finite size queuing system to the value of the mfg . </S><S> in general closed form solutions of such mfg are not available and thus numerical methods are needed . </S><S> we use the markov chain approximation method to construct approximations for the solution of the mfg and establish convergence of the numerical scheme . </S><S> * ams classification : * 60k25 , 91a13 , 60k35 , 93e20 , 60h30 , 65m12 , 60f17 . * </S><S> keywords : * heavy traffic limits , queuing systems , strategic servers , mean - field games , rate control , reflected diffusions . </S>

nucl-th9608030

halo nuclei have been extensively investigated in the latest years both from the theoretical and experimental point of view . fora review see for instance @xcite and references therein .the general properties of light nuclear halos have succesfully been revealed by means of few - body models , dividing the degrees of freedom into the approximately frozen ( core ) and active ( halo ) .special attention has been paid to three - body borromean bound systems @xcite , where none of the two - body subsystems is bound .the most prominant examples are @xmath3he ( @xmath4he+n+n ) and @xmath0li ( @xmath5li+n+n ) both extensively discussed in a general theoretical framework @xcite .the most detailed source of experimental information about the structure of these nuclei is the momentum distributions of the particles resulting from the fragmentation reactions @xcite . for one particle removal reaction with a high energy beam we can consider that the removed particle in the halo projectile is instantaneously captured by the target without disturbing the remaining two .this reaction picture is known as sudden approximation , and has been proved to be fairly accurate in the description of such halo fragmentation reactions @xcite . for a correct interpretation of the experimental data it is essential to include the interaction between the two non - disturbed particles after the fragmentation , especially when low - lying resonances are present @xcite .detailed and consistent incorporation of this final state interaction ( fsi ) into the model is necessary as seen by its influence on the momentum distributions @xcite .neutron momentum distributions are very much affected by the final state interaction , but also core momentum distributions are significantly influenced .a polarized beam interacting with a target is not equally sensitive to all the components of the wave function .it is then possible by means of this tool to observe signals of the less dominant components , and in this way to investigate finer details of the nuclear structure of the projectile .the purpose of this paper is to investigate the effects of polarized beams in fragmentation reactions of fast three - body borromean halo nuclei .the case where the spin projection of one of the particles is measured in the final state will also be considered .we introduce several useful observables and show the kind of information we can extract from them . in particular we consider neutron removal from the polarized @xmath0li projectile .the resulting neutron momentum distributions are sensitive to the amount of @xmath6-state admixture in the neutron - core relative state .the correct description of the structure and reactions of three - body halo nuclei is an indispensable prerequisite for the present investigation .these general procedures are discussed in previous papers @xcite which we shall use extensively .this paper is organized as follows : in section ii we formulate the theoretical model and the method used to describe the fragmentation reaction including definition of the polarization observables .the numerical results obtained for @xmath0li are reported in section iii and in section iv we give a summary and the conclusions .we collect the rather complicated analytic expressions of the differential momentum distributions for different reactions in appendices a and b.the fragmentation reaction is described in the sudden approximation , which assumes that one of the three particles in the projectile is instantaneously removed by the target without disturbing the remaining two .this approximation is valid for a high - energy three - body halo beam interacting with a small target through short range interactions .therefore we assume a light target where the coulomb dissociation process only contributes marginally .the transition matrix in the sudden approximation is given as @xmath7 where @xmath8 is the final state wave function after break - up and @xmath9 is the initial wave function of the three - body projectile .we now must specify these wave functions , compute the matrix element and integrate @xmath10 over all the non observed variables .the initial wave function describes the bound three - body halo system whereas the final state wave function describes the two - body continuum state specified by the experimental conditions . the projectile three - body wave function @xmath9 with inclusion of the proper quantum numbers and variables is denoted by @xmath11 , where @xmath12 and @xmath13 are the total spin and its projection .the coordinates @xmath14 and @xmath15 are the usual jacobi coordinates @xcite , where @xmath14 is drawn between the two particles surviving after the fragmentation .the wave function is expanded in terms of a complete set of hyperspherical harmonics @xmath16 , where the quantum number @xmath17 usually is called the hypermoment , @xmath18 and @xmath19 are the orbital angular momenta associated with @xmath14 and @xmath15 , and @xmath20 is resulting from the coupling of these angular momenta .the variables @xmath21 , @xmath22 , @xmath23 , and @xmath24 are the hyperspherical variables obtained from the jacobi coordinates @xmath14 and @xmath15 @xcite .then @xmath25 is written as @xmath26^{j m } \ ; , \label{totrot } \end{aligned}\ ] ] where the spin function @xmath27 is obtained by coupling the spins of two of the particles to @xmath28 which in turn is coupled to the spin @xmath29 of the third particle resulting in the total spin @xmath30 .the radial functions @xmath31 and the expansion coefficients @xmath32 are calculated numerically by solving the faddeev equations in coordinate space @xcite .details about this procedure can be found in @xcite .the final state wave function @xmath33 in the plane wave approximation in the center of mass frame of the projectile is given by @xmath34 where @xmath35 for i=1,2 and @xmath36 are the spin wave functions respectively for the two undisturbed particles and the third removed particle . here@xmath37 and @xmath38 are the total spin and its projection on an axis .furthermore @xmath39 and @xmath40 are the total and relative momentum of the two remaining particles in the final state , @xmath41 is the distance between the two remaining particles , and @xmath42 is the distance between the center of mass of this two - body system and the removed particle . the description in terms of plane waves in eq.([e5 ] ) implies that all the interactions between the particles are neglected in the final state wave function .however , in our approximation only one of the particles is removed from the other two , and the interaction between the non - disturbed particles can not be neglected . in a recent work @xcite we investigated the influence of this final state interaction ( fsi ) on the momentum distributions , and observed that they are crucial for reproduction of the experimental momentum distributions . in eq.([e5 ] ) introduction of the final state interaction amounts to substitution of the plane wave , @xmath43 , describing the two - body system in the final state by the appropriate _ distorted _ two - body wave function @xmath44 . when the two - body interaction do not mix the two - body states with different spin @xmath28 and relative orbital angular momentum @xmath18, we can expand this two - body wave function ( @xmath44 ) in partial waves as in @xcite .these assumptions are usually strictly valid .the only exception arises from the tensor interaction .however the resulting mixing is often very small and therefore insignificant in the present context .we then have @xmath45 with @xmath46 where @xmath47 is the momentum associated with @xmath14 , @xmath48 is the spin projection of @xmath28 , @xmath49 is the radial and @xmath50 the angular distorted wave function . the angles @xmath23 and@xmath51 define the direction of @xmath14 and @xmath47 , respectively .the dependence on the spins @xmath52 and @xmath53 of the two particles surviving after the fragmentation reaction has been omitted for notational simplicity .the radial wave functions @xmath54 are obtained numerically by solving the schrdinger equation with the appropriate two - body potential @xmath55 @xmath56 where the effective radial potential is obtained by integration over the angles @xmath23 @xmath57 and @xmath58 is an arbitrary normalization mass .the calculation of the transition matrix requires now the overlap of the initial three - body wave function in eq.([totrot ] ) and the final state wave function @xmath59 , where @xmath60 is the momentum associated with the @xmath15 coordinate . since @xmath61we then find @xmath62 for simplicity we do not specify the dependence of @xmath63 on the spins @xmath52 , @xmath53 and @xmath29 of the three particles involved in the reaction .the analytic form of @xmath64 is @xmath65 where @xmath66 . in eq.([tm0 ] ) we have introduced the hyperspherical coordinates in momentum space @xmath67 , @xmath68 , and the numerical function @xmath69 is given by @xmath70 \label{ifun}\end{aligned}\ ] ] where @xmath71 , @xmath72 is a jacobi polynomial , and @xmath73^{1/2}.\ ] ] the cross section or momentum distribution is now obtained by squaring the transition matrix , and subsequently averaging over initial states and summing up over final states . if we consider the possibility of having a polarized beam and measuring the spin projection of the two particles in the final state we can then write @xmath74 where @xmath75 , @xmath76 , and @xmath77 give the occupation probability of each magnetic substate in the initial and final states .when the beam is unpolarized and there is no detection of final spin projections these probabilities take the values @xmath78 , @xmath79 , and @xmath80 , and we obtain the unpolarized expression for the momentum distributions @xcite .the volume element @xmath81 is written @xmath82 where @xmath83 , @xmath84 , and @xmath85 are the polar and azimuthal angles associated to @xmath51 .the integration of eq.([mom ] ) over @xmath86 and @xmath87 can be done analytically . by further numerical integration over @xmath88 and @xmath89, we get the one - dimensional relative momentum @xmath90 distribution of the remaining particles . by integrating over @xmath88 and @xmath91 ,we get instead the two - dimensional relative momentum ( @xmath89 ) distribution .it should be noted that we have not specified any coordinate system and the axes @xmath14 , @xmath15 , and @xmath92 are therefore completely arbitrary .thus , in the sudden approximation the longitudinal and transverse momentum distributions are identical . after substituting eqs.([tm ] ) and ( [ tm0 ] ) into ( [ mom ] ) , and integrating analytically over @xmath86 and @xmath87 we obtain the expressions for the three - differential momentum distributions that are collected in appendix a. three different casesare considered : \(i ) eq.([rel1 ] ) describes the process with 100% polarized beam where the polarization of one of the particles in the final state is measured , i.e. @xmath93 , @xmath94 , @xmath95 , \(ii ) eq.([rel2 ] ) describes the process with 100% polarized beam without detection of the final spin states , i.e. @xmath96 , @xmath97 , @xmath95 , \(iii ) eq.([rel3 ] ) describes the process with unpolarized beam without detection of the final spin states ,i.e. @xmath98 , @xmath97 , @xmath95 . to compare with the experimental data the momentum distributionsshould be referred to the center of mass frame of the projectile .to do this we construct the momentum @xmath99 of one of the particles in the final state relative to the center of mass of the projectile as a linear combination of @xmath47 and @xmath60 , i.e. @xmath100 where @xmath101 refer to the two remaining particles .if the jacobi coordinate @xmath14 goes from particle 1 to particle 2 , then @xmath102 and @xmath103 take the values @xmath104@xmath105 when we compute the momentum distributions of particle 1 and 2 , respectively .since the volume element @xmath106 we obtain the relation @xmath107 the only additional complication in this coordinate system is that the numerical function @xmath69 now depends on the relative angle @xmath108 between @xmath99 and @xmath60 .only two angular integrations can then be performed analytically , and three integrations must be done numerically in order to get the one - dimensional and two - dimensional momentum distributions of one of the particles in the final state relative to the center of mass of the projectile . in appendixb we show the analytic expressions for the four - differential momentum distributions relative to the center of mass of the projectile for the same three cases as specified in appendix a. they are given by eqs.([cm1 ] ) , ( [ cm2 ] ) and ( [ cm3 ] ) , respectively . we shall denote the one and two - dimensional momentum distributions when the beam is unpolarized and the spin projection of the particles is not measured in the final state by @xmath109 and @xmath110 , respectively .the distributions are referred to the center of mass of the three - body projectile , and obtained by integration over the unobserved quantities of eq.([cm3 ] ) in appendix b. in the same way we shall denote the one and two - dimensional momentum distributions when the beam is 100% polarized ( with spin projection @xmath13 ) and the spin projection of the particles in the final state is not measured by @xmath111 and @xmath112 , respectively .these distributions are obtained by integration of eqs.([cm2 ] ) in appendix b. finally , @xmath113 and @xmath114 will denote the one and two - dimensional momentum distributions for a 100% polarized beam ( with spin projection @xmath13 ) , and when the spin projection of one of the particles in the final state is measured to be @xmath38 .these distributions are obtained by integration of eqs.([cm1 ] ) in appendix b. we define now the following asymmetries @xmath115 @xmath116 it is easy to check that the @xmath117 term in eq.([cm1 ] ) and the @xmath118 term in eq.([cm2 ] ) are independent of @xmath13 and @xmath38 , and equal to the totally unpolarized momentum distribution eq.([cm3 ] ) .this means that the deviations from zero of the asymmetries eqs.([as1 ] ) and ( [ as2 ] ) are a direct consequence of the polarization of the beam and the measure of the spin projection of one of the particles after the fragmentation .direct examination of eq.([cm2 ] ) permits us to observe that if only @xmath37-waves were involved in the initial and final wave functions ( @xmath119 and @xmath120 ) only the @xmath121=0 term would contribute to the momentum distribution .in other words , the asymmetry eq.([as1 ] ) would vanish .therefore , the deviation from zero of this asymmetry is a signal of non - vanishing components different from s - waves in the wave function .this conclusion is not true for the asymmetry eq.([as2 ] ) , since the momentum distribution eq.([cm1 ] ) has additional non - vanishing terms even when only @xmath37-waves are involved .in this section we apply the procedure described above to two - neutron halo nuclei , and in particular to @xmath0li .we consider neutron removal reactions , and the neutron-@xmath5li interaction is include in the final state .we only included @xmath37- and @xmath6-waves in the calculations , since components with higher angular momenta are negligibly small .we use the two - body potentials described in ref.@xcite , i.e. @xmath122\ ] ] @xmath123 , \label{pots } \\ v_{nc}^{(l)}&=&(v_l + v_{so}^{(l ) } \mbox{\bf l}_{nc } \cdot \mbox{\bf s}_{nc } ) \exp \left [ - ( r / b_{nc})^2 \right ] \label{potl } \ ; , \end{aligned}\ ] ] the nucleon - nucleon potential @xmath124 is made of gaussians containing central ( @xmath125 ) , spin splitting ( @xmath126 ) , spin - orbit ( @xmath127 ) and tensor ( @xmath128 ) parts where the parameters are fitted to low energy @xmath37- and @xmath6-wave nucleon - nucleon scattering data ( @xmath129 and @xmath130 are the spins of the two neutrons , @xmath131 is the two - neutron relative orbital angular momentum , @xmath132 , and @xmath133 is the usual tensor operator ) . in the neutron - neutron subsystem more than 98% of the probablityis found in @xmath37-waves .the neutron - core potential @xmath134 is also made of gaussians adjusted to reproduce the experimental binding energy and root mean square radius of the three - body projectile ( @xmath135 kev @xcite and @xmath136 fm @xcite , respectively , for @xmath0li ) .the different waves are independently fitted , allowing us to adjust separately the energy positions of the virtual @xmath37-states and the resonances in the neutron - core subsystem . for simplicitythe spin splitting term ( introduced by the parameter @xmath137 ) has been included only for @xmath37-waves in the neutron - core interaction .the expression we use for the wave function of the three - body halo projectile is given by eq.([totrot ] ) , where the radial functions @xmath31 and the coefficients @xmath138 are computed numerically .to illustrate some of the general aspects of the three - body wave functions , we show in fig .[ 1 ] the @xmath139 function for @xmath0li .this function dominates the expansion in eq.([totrot ] ) as it contains more than 95% of the norm of the total wave function .the neutron - core interaction has been fitted to produce a low - lying virtual @xmath37-state at 100 kev and the lowest @xmath6-resonance at 0.5 mev . by use of the spin - orbit parameter ( @xmath140 ) we can move the higher - lying p - resonances up or down andthereby modify the @xmath6-wave content in the neutron-@xmath5li subsystem .four different situations of varying @xmath6-wave content are shown , i.e. 4% ( solid line ) , 18% ( short - dashed line ) , 26% ( dotted line ) and 32% ( long - dashed line ) .we observe that the more @xmath6-wave content in the neutron-@xmath5li subsystem the narrower the radial wave function .this result can be attributed to the centrifugal barrier , that keeps the neutrons closer to the core for @xmath6-waves . when the final state interaction is neglected in the calculation ,the momentum distributions are given by the square of the fourier transform of the three - body wave function in eq.([totrot ] ) .therefore a larger @xmath6-wave content in the neutron - core subsystem will produce a broader momentum distribution .details about calculations of momentum distributions following this method have been presented in @xcite. nevertheless , the differences between the three - body wave functions in fig . [ 1 ] are not very large and the corresponding momentum distributions would also be very similar . in the inset of fig .[ 1 ] we display the same radial three - body wave function @xmath139 for a neutron - core potential producing a @xmath6-resonance at 100 kev ( solid line ) and a virtual @xmath37-state at 100 kev ( dashed line ) .again we find that the low - lying @xmath6-resonance produces a narrower radial wave function , and therefore a broader momentum distribution ( without fsi ) . in this casethe difference between the wave functions is larger and as a consequence the momentum distributions for a low - lying @xmath6-resonance and a low - lying virtual @xmath37-state will also be clearly distinguishable . to analyze the effect of the inclusion of the final state interaction we investigate the behaviour of the distorted wave function in eq.([2bexp ] ) .first we consider a neutron - core potential with a virtual @xmath37-state at 100 kev , and investigate how the corresponding continuum radial @xmath37-state wave functions behave for different energies @xmath141 of the neutron - core system .when the final state interaction is neglected this radial s - state wave function is simply the bessel function @xmath142 , which in fig .[ 2a ] ( the dashed line ) is compared to the distorted wave functions ( the solid lines ) for different values of @xmath143=0.1 mev , 1 mev and 5 mev . at large distances ( not shown in the figure )the oscillations continue and have the same amplitude .the difference from the bessel function is only significant at small distance and this difference increases in general when @xmath143 approaches the energy of the virtual state .( note , however , the singularity at the origin for vanishing @xmath143 . ) therefore for small values of @xmath144 the overlap in eq.([tran ] ) is larger than in eq.([tm ] ) and since the normalization is the same in all cases , the momentum distribution becomes narrower by inclusion of final state interactions .the inset in fig .[ 2a ] shows the same plot as in the main part , but for a neutron - core potential with a virtual @xmath37-state at 1 mev .the general conclusions are the same , the smaller @xmath143 the larger the effect of the final state interaction .however , in this case the difference with the bessel function is clearly smaller ( note the different vertical scales ) , and therefore also the influence of the final state interaction is smaller . we can then conclude that the inclusion of final state interaction in @xmath37-waves makes the momentum distribution narrower , and the smaller the energy of the virtual @xmath37-state the larger the effect of the final state interaction . in fig .[ 2b ] we show the same quantities as in fig .[ 2a ] , but now for @xmath6-waves . in the main part of the figure we consider a neutron - core potential with a @xmath6-resonance at 0.1 mev .again the dashed line is the radial part of the @xmath6-wave without final state interaction ( equal to the bessel function @xmath145 ) .when we solve the schrdinger equation for a @xmath6-wave and when @xmath141 is far from the resonance energy ( we have taken @xmath146 mev ) the difference with the bessel function is very small , and the effect of the final state interaction is almost negligible . when @xmath143 approaches the energy of the resonance the @xmath6-wave becomes more and more localized , in such a way that it gets a maximum for @xmath143 equal to the energy of the resonance ( 0.1 mev in this case ) . as a consequence , the overlap in eq.([tm ] ) is maximum for @xmath147 , and the momentum distributions are enhanced around that value of @xmath144 .the inset of fig .[ 2b ] shows the same as in the main part for @xmath148 = 1 mev .the behaviour of the @xmath6-wave is identical to the previous one .the wave is very much localized and gets its maximum value for @xmath149 .however , the maximum is now clearly smaller than in the previous case of @xmath148 = 0.1 mev , and therefore also the effect on the momentum distributions is less pronounced .therefore , the inclusion of the final state interaction in the @xmath6-wave will produce an increase in the momentum distributions around @xmath150 .this increase is especially important when the neutron - core potential has a low - lying @xmath6-resonance , and again , due to the normalization , makes the momentum distributions narrower .summarizing , the inclusion of final state interaction highly modifies the momentum distributions for @xmath37- and @xmath6-waves .this change is clearly more significant for low - lying virtual @xmath37-states and low - lying @xmath6-resonances . for s - waves the distributions are enhanced at the origin ( @xmath151 ) , and for p - waves the enhancement is produced at @xmath152 . in both casesthe momentum distributions become narrower by inclusion of the final state interaction . before concluding this subsection we must emphasize that the momentum distributions discussed so far were expressed in terms of @xmath144 , i.e. the relative momentum between the two remaining particles . from eqs.([mcm ] ) - ( [ ab2 ] ) we see that the momentum of the neutron relative to the center of mass of the projectile is very close to @xmath47 ( the @xmath103 coefficient is small due to the large mass of the core ) .therefore the above conclusions about the effect of the final state interaction roughly remain unchanged for the neutron momentum distributions relative to the center of mass of the projectile .for the core momentum distribution the @xmath153 coefficient in eq.([mcm ] ) is large , and the effect of the final state interaction is much smaller when we refer momentum distributions of the core to the center of mass of the projectile .the reason is simply that the center of mass of the core is fairly close to the center of mass of the total three - body system .experimental data suggest that the neutron unbound @xmath1li system has a @xmath6-resonance around 0.5 mev @xcite and a low - lying ( @xmath154 0.2 mev ) virtual @xmath37-state or @xmath6-resonance @xcite . in all the computations reported in this workthe neutron - core potential is fitted to produce a @xmath6-resonance at 0.5 mev in @xmath1li . for the lowest - lying statewe investigate several possibilities in order to analyze the dependence of the polarization observables on the properties of the lowest - lying state . in fig .[ 3a ] we show the asymmetry @xmath155 for neutron momentum distributions after a neutron removal @xmath0li fragmentation reaction . a low - lying state ( s- or p - wave ) is present in @xmath1li at 50 kev ( solid line ) , 100 kev ( dashed line ) or at 150 kev ( dot - dashed line ) . for low - lying @xmath6-resonanceswe observe for all the three energies two peaks in the asymmetry .the first peak at low momentum corresponds to the lowest - lying @xmath6-resonance .the second peak is in all the three cases generated by the @xmath6-resonance at 0.5 mev .the connection between the position of the different minima and maxima of the asymmetry and the energy of the @xmath6-resonance may be seen from the relations @xmath156 and @xmath157 , where @xmath158 is the relative neutron - core momentum , @xmath148 is the energy of the resonance and @xmath159 is the reduced mass of the neutron - core subsystem .the different p - resonance energies in the figure correspond to values @xmath158 of 10 mev / c , 14 mev / c , and 17 mev / c , respectively and an energy of 0.5 mev corresponds to @xmath160 30 mev / c .as discussed in section 2 , the momentum distributions are enhanced around the resonance value of @xmath144 , that is very close to the momentum of the neutron relative to the center of mass of the three - body projectile .this increase in the momentum distributions produces the oscillations in the asymmetry . as a consequence ,when the energy of the resonance increases the peak in the asymmetry is displaced towards higher momenta .since we plot the asymmetries as functions of the longitudinal component of the neutron momentum relative to the center of mass of the projectile ( that is not exactly @xmath158 , but very close ) , the maxima and minima of the asymmetry do not exactly coincide with these numbers .however , the connection to the energy of the resonances in the neutron - core subsystem is very clear . inspecting the asymmetry in fig .[ 3a ] where a low - lying virtual @xmath37-state is present in @xmath1li we observe that the first peak completely disappears .this fact was expected , since in section 2 we showed that the asymmetry eq.([as1 ] ) vanishes when only @xmath37-waves are involved in the wave functions . in other words, the asymmetry @xmath161 does not feel the presence of virtual @xmath37-states in the neutron - core subsystem , and it is only sensitive to the structure of the non zero angular momnetum resonances . since only s - waves are changed for the three different cases considered , the asymmetries are essentially indistinguishible .let us now consider the case where the spin projection of one of the particles in the final state is measured .in particular we assume that the spin projection of the core is measured to be 3/2 , and construct the asymmetry @xmath162 for neutron momentum distributions in a neutron removal reaction .the result is shown in fig .[ 4a ] for the cases corresponding to fig .again , for low - lying @xmath6-resonances there are two peaks in the asymmetry , one of them connected to the energy of the low - lying resonance and the other peak is related to the resonance at 0.5 mev .for low - lying virtual @xmath37-states the first peak disappears , and only the peak associated with the @xmath6-resonance at 0.5 mev remains .now the @xmath37-waves also contribute to the asymmetry , but the contribution is localized in the region close to the origin where the curves now are different for the three energies of the low - lying virtual @xmath37-state .therefore the asymmetries shown in fig .[ 4a ] display the same features as those of fig .the main difference is that the amplitude of the oscillations of the asymmetries clearly is larger in the asymmetries shown in fig .[ 4a ] .we have now concluded that the asymmetries are sensitive essentially only to the non - zero angular momentum partial waves in the initial and final state wave function .thus we also expect sensitivity to the content of the different waves in the two - body subsystem . in particular , for @xmath0li the asymmetries should depend on the @xmath6-wave content in @xmath1li subsystem ( the neutron - neutron subsystem contains in any case almost exclusively @xmath37-waves ) . in fig .[ 5a ] we show the same asymmetry , @xmath155 , as in fig. [ 3a ] , but now assuming a virtual @xmath37-state at 100 kev and a @xmath6-resonance at 0.5 mev in @xmath1li . by use of the spin - orbit parameter @xmath163 in the neutron - core potential in eq.([potl ] ) we modify the @xmath6-wave content in the neutron-@xmath5li subsystem .the solid line , dashed line , and dot - dashed line are the resulting calculations with 4% , 18% , and 32% of @xmath6-wave content in the @xmath1li wave function . in the three cases we observe the peak in the asymmetryaround 30 mev / c corresponding to the @xmath6-resonance at 0.5 mev .the second peak in the asymmetry corresponds to the next @xmath6-resonance , closer to the first peak for a larger @xmath6-wave content , since the increase in the @xmath6-wave content is obtained by decreasing the energies of the @xmath6-resonances .we observe that the lower the @xmath6-wave content the smaller the amplitudes of the oscillations of the asymmetry , see the solid line corresponding to the @xmath6-wave content of 4% .this is related to the fact that the asymmetry must vanish when the @xmath6-waves are totally absent , however , it should be mentioned that the denominator of the asymmetry in eq.([as1 ] ) is the unpolarized momentum distribution , that is known to be narrower for lower @xmath6-wave contents @xcite .it is then possible to find oscillations with similar amplitude for different values of the @xmath6-wave content . in fig .[ 6a ] we show the same asymmetry , @xmath164 , as in fig . [ 5a ] , i.e. one - dimensional neutron momentum distributions in a neutron removal reaction . as in fig .[ 4a ] the spin projection of the core is assumed to be measured to have a value of 3/2 in the final state .again , in this case the asymmetry shows essentially the same features as in fig .[ 5a ] , but with the advantage of larger amplitudes .the values of the asymmetries can vary now between 0.8 and more than 1.3 .all the asymmetries we have discussed so far are obtained from one - dimensional neutron momentum distributions .the same conclusions can be obtained from the corresponding two - dimensional neutron momentum distributions shown for completeness in figs .[ 3b ] , [ 4b ] , [ 5b ] , [ 6b ] .first we note that when only the beam is polarized the asymmetries are not sensitive to the structure of the virtual @xmath37-states in the two - body subsystems . when only the energy of the @xmath37-states is modified in the @xmath1li wave function the asymmetries do not change .second , the minima and maxima of the asymmetries provide information about the @xmath6-resonance structure in the @xmath1li two - body subsystem .third , the asymmetries are clearly dependent on the @xmath6-wave content of the @xmath1li wave function .fourth , the asymmetries obtained assuming that the spin projection of the core has been measured in the final state ( 3/2 in our case ) show a clearly larger effect than those obtained assuming that only the beam is polarized . in principlethe same asymmetries could also be obtained from core momentum distributions and from neutron momentum distributions after core break - up .however these two cases are not especially interesting .the former because core momentum distributions are very little affected by the final state interaction , and then the asymmetries do not show very much structure .the latter because after core break - up the two neutrons survive in the final state , and the three - body wave function written in the jacobi set where the @xmath14 coordinate is drawn between the two neutrons contains almost exclusively s - waves , to which the asymmetries are not sensitive .we have formulated a detailed model to investigate momentum distributions of particles emerging after nuclear break - up reactions of fast polarized three - body halo nuclei .the three main ingredients are : \(i ) the three - body description of the initial nuclear wave function .this assumes that the particles or clusters are inert throughout the process and only distances larger than the cluster radii can be well described .such models have been successfully applied to various light nuclei .\(ii ) the sudden approximation where the target nucleus instantaneously removes one of the particles from the three - body projectile without disturbing the other two .this approximation is very well suited for short range interactions and fast processes where the beam velocity is much larger than the velocity of the intrinsic motion .the radioactive beams are available with high energies exceedingly well fulfilling this criterium .\(iii ) the final state relative wave function for the two remaining particles must include effects of their mutual interaction .the resulting distorted wave function may differ substantially from the plane wave solution when no two - body interaction is present .these effects are crucial when virtual s - states or higher angular momentum resonances occur at energies below about 1 mev in the final state two - body system .it is rather difficult to construct a theoretical system where assumption ( i ) is valid without the necessity of using ( iii ) .the reason is that the three - body structure only is appropriate when the the particles are fairly loosely bound , since they otherwise would overlap and destroy the cluster property . as a consequencethe two - body subsystems must have fairly weak effective attractions yet they must be sufficiently strong to create a three - body bound state when working together . the result is low - lying continuum two - body structure .the same two - body interactions are therefore responsible for both the initial three - body bound state and for the distorted final state wave function . a consistent treatment of initial and final stages of the process is therefore essential and our model meets this crucial requirement . in other wordsthe two - body continuum structure is decisive for both the three - body halo structure and the momentum distributions after fragmentation .we first describe the method to compute the momentum distributions in nuclear break - up processes .the overlap between initial and final state wave function is the essential quantity .the initial bound state three - body halo wave function is obtained by solving the faddeev equations by use of an adiabatic hyperspherical expansion technique .this method has been tested previously and shown to be very efficient for ground states or low - lying excited states with small two - body relative orbital angular momenta .the final state distorted two - body continuum wave function is expanded in partial waves and the related radial wave functions are calculated numerically .they are normalized at large distances to coincide with the plane wave solutions .the resulting overlap matrix with different angular momentum projections is then available .the cross sections or the observable momentum distributions are then essentially only the absolute square of the overlap integrated over unobserved quantities . these expressions contain numerically computed quantities .they appear to be rather complicated especially due to the use of the coordinate system where the center of mass of the three - body projectile is at rest and the dedicated goal of computing momentum distributions for polarized projectiles .first we specify the two - body interactions appropriate for the @xmath0li system considered as two neutrons surrounding the @xmath165li - core .the neutron - neutron interaction is parametrized to reproduce low - energy scattering properties and then maintained throughout the present investigation .the neutron - core interaction is adjusted to reproduce the measured @xmath0li binding energy and root mean square radius .the remaining free parameters are used to place virtual s - states and p - resonances at various energies close to the threshold resulting in different @xmath0li - structures and different momentum distributions .the final state two - body s - wave functions all have maxima at the origin .the highest value occurs when the scattering energy equals to the energy of the virtual s - state .the state is therefore peaked around zero and this localization is strongest around the energy of the virtual state .the p - state two - body wave function is peaked at an energy corresponding to the p - resonance energy . without resonances and virtual statesthe wave functions are simply bessel functions with the related oscillations and without these pronounced peak structures .neutron momentum distributions after neutron removal are computed for polarized @xmath0li and when the polarization of the final two - body state is measured .several polarization observables are constructed and expressed in terms of quantities , called asymmetries , deviating from zero only when the momentum distributions depend on the polarization .several continuum structures of @xmath1li are considered , but in all cases we maintain the binding energy and root mean square radius of @xmath0li .the inclusion of the final state interaction is an essential ingredient in the computation , determining the behaviour of the asymmetries .in fact , when they are not included the structure in the polarization observables completely disappears .the core momentum distributions are little affected by the final state interaction , and therefore the asymmetries obtained from them exhibit very little structure . from the asymmetries constructed from neutron momentum distributionswe have observed the following features : if only the beam is polarized the asymmetries vanish when only s - waves contribute .this makes the asymmetries obtained from a core break - up reaction from @xmath0li not relevant , because almost exclusively s - waves are involved in this fragmentation process .much more interesting are the asymmetries obtained from neutron momentum distributions after neutron removal reactions .the rest of the waves plays an important role , and the asymmetries become a perfect tool to investigate the @xmath166-resonance structure of the neutron - core subsystem . these observables then contain information about the continuum structure of the two - body subsystems .in fact low - lying resonances ( p - resonances in the @xmath0li case ) in the neutron - core system show up as an extremum at the momentum value corresponding to the resonance energy .an additonal measurement of the two - body final state polarization enhances the amplitudes of the asymmetries , but otherwise leaves essentially unchanged the structure of the asymmetry .in this appendix we give the analytic expressions for the relative momentum distributions of the two particles in the final state .we consider three different cases : \(i ) 100% polarized beam and spin projection of one of the particles in the final state measured : @xmath167 \(ii ) 100% polarized beam and no spins projections measured in the final state : this amounts to multiply the previous expression by @xmath168 and sum over @xmath169 .the main consequence of the summation is that the index @xmath170 is zero , and then @xmath171 .then we get the expression @xmath172 \(iii ) unpolarized beam and no spins projections measured in the final state : multiplying by @xmath173 and summing over @xmath13 we get the totally unpolarized momentum distribution . from the summation we obtain that @xmath118 , and the analytic expression is @xmath174in this appendix we give the analytic expressions for the momentum distributions of the two particles in the final state relative to the center of mass of the incident three - body system .the same three cases as in the previous appendix are considered : if we assume that one of the particles in the final state , say particle 1 , has infinite mass , then the momentum distribution of particle 2 relative to the three - body center of mass should coincide with the momentum distribution relative to particle 1 .in other words , eqs.([rel1 ] ) , ( [ rel2 ] ) , and ( [ rel3 ] ) should be recovered by integrating over @xmath182 eqs.([cm1 ] ) , ( [ cm2 ] ) , and ( [ cm3 ] ) , respectively . in this case @xmath183 , see eq.([ab2 ] ) , @xmath184 and therefore @xmath185 and @xmath186 .now @xmath187 and @xmath188 are independent of @xmath182 and the integration over @xmath182 can easily be done analytically leading directly to eqs.([rel1 ] ) , ( [ rel2 ] ) , and ( [ rel3 ] ) .zhukov , b.v .danilin , d.v .fedorov , j.m .bang , i.j .thompson , and j.s .vaagen , phys .231 ( 1993 ) 151 c. bertulani , l.f .canto , and m.s .hussein , phys .226 ( 1993 ) 281 p.g .hansen , a.s .jensen , and b. jonson , annu .nucl . part .45 ( 1995 ) 591 p.g .hansen and b. jonson , europhys . lett . 4 ( 1987 )409 l. johannsen , a.s .jensen , and p.g .hansen , phys .b244 ( 1990 ) 357 d.v .fedorov , a.s .jensen , and k. riisager , phys .c49 ( 1994 ) 201 t. kobayashi , o. yamakawa , k. omata , k. sugimoto , t. shimoda , n. takahashi and tanihata , phys .lett . 60 ( 1988 ) 2599 r. anne _ et al .b250 ( 1990 ) 19 n.a .et al . _ ,69 ( 1992 ) 2050 n.a .et al . _ ,c51 ( 1995 ) 3116 m. zinser _et al . _ ,phys . rev. lett . 75 ( 1995 )1719 t. nilsson _et al . _ ,. lett . 30 ( 1995 )19 f. humbert _et al . _ ,b347 ( 1995 ) 198 m.v .zhukov , l.v.chulkov , d.v .fedorov , b.v .danilin , j.m .bang , j.s .vaagen and i.j .thompson , j. phys .g20 ( 1994 ) 201 a.a .korsheninnikov and t. kobayashi , nucl .a567 ( 1994 ) 97 i.j .thompson and m.v .zhukov , phys .c49 ( 1994)1904 m.v. zhukov and b. jonson , nucl .a589 ( 1995 ) 1 , f. barranco , e. vigezzi , and r.a .broglia , phys .b319 ( 1993 ) 387 e. garrido , d.v .fedorov and a.s .jensen , phys .c53 ( 1996 ) 3159 , and submitted for publication d.v .fedorov , a.s .jensen , and k. riisager , phys .c50 ( 1994 ) 2372 e. garrido , d.v .fedorov and a.s .jensen , phys .c51 ( 1995 ) 3052 a. cobis , d.v .fedorov , and a.s .jensen , submitted for publication r.g .newton , scattering theory of waves and particles , second edition ( springer - verlag , new york , 1982 ) , p.444 b.m .et al . _ ,71 ( 1993 ) 4124 i. tanihata_ et al . _ , phys .b287 ( 1992 ) 307 b.m .et al . _ ,c49 ( 1994 ) 279 r.a .et al . _ ,c47 ( 1993 ) r2439 h.g .et al . _ , z. phys .a344 ( 1993 ) 381

<S> momentum distributions of particles from fast three - body halo fragmentation reactions with polarized beams and outgoing fragments are analyzed by use of the sudden approximation . </S><S> the final state interaction between the two non - disturbed particles is considered . </S><S> we first give a general and detailed description of the method . </S><S> we introduce observable quantities that emphasize the two - body correlations in the initial three - body structure as well as in the final two - body system . using neutron removal from @xmath0li as an example we investigate the dependence of the polarization observables on the properties of the low - lying resonances and virtual states of the @xmath1li subsystem . </S><S> these observables are very sensitive to the @xmath2-waves in the wave functions , and they provide detailed information on the resonance structure of the neutron - core subsystem . -1.2 cm 25.60.gc , 25.60.-t , 24.70.+s , 21.45.+v </S>

1612.02052

gauge theories constitute the backbone of the standard model of particle interactions .gauge theories exist in several different phases that are naturally classified according to the force measured between static sources .knowledge of the phase diagram proves crucial when investigating extensions of the standard model both for particle physics and cosmology .a special class of gauge theories are the ones that are fundamental according to wilson @xcite , meaning that they possess a complete ( in all couplings ) ultraviolet ( uv ) fixed point either of non - interacting ( asymptotically free @xcite ) or of interacting nature ( asymptotically safe @xcite ) .complete asymptotically safe quantum field theories were discovered only very recently @xcite , widening the horizon of fundamental theories that can be used for novel phenomenological applications @xcite beyond the traditional asymptotically free paradigm @xcite .the thermal properties of completely asymptotically safe field theories were elucidated in ref .@xcite . herewe focus our attention on asymptotically free gauge theories featuring gauge and fermion degrees of freedom that develop an infrared ( ir ) interacting fixed point .we henceforth push forward our program to systematically understand , in a rigorous manner , the dynamics of these theories at zero @xcite and non - zero matter density @xcite , by analysing their conformal viscous behavior as a function of the number of flavors .because of the perturbative nature of the theories investigated here , along the full energy range , our investigation of their viscous properties is also much better controlled than for qcd - like theories .this is so because at very high energies the theory is non - interacting and at very low energies the theory reaches an ir perturbative fixed point .furthermore , the value of the gauge coupling at the ir fixed point can be made arbitrarily small by changing the number of flavors and colors of the theory .this allows us to consistently truncate the perturbative expansion within the range of convergence of the theory .we henceforth determine the conformal behavior , as a function of the number of flavors , for the shear viscosity - to - entropy density ratio and the fermion - number diffusion coefficient . by adapting the results of ref .@xcite we learn that , as we decrease the number of flavors below the loss of asymptotic freedom , their ir fixed point values decrease .furthermore , for a given number of flavors within the perturbative conformal window both coefficients decrease with decreasing temperature ( once we multiply the diffusion coefficient by the temperature ) from their infinite value in the deep uv down to the value at the ir fixed point . we represent the results for three colors as a function of the number of flavors , but to the order investigated here the results are similar for any other fermion representation .we organise this paper as follows . in sec .[ sec : g2 ] we shortly review the theory , introduce the notation , and provide the salient zero and non - zero temperature properties .this is followed by the determination of the transport coefficients in sec .[ sectioniii ] . herewe will comment on our findings and finally conclude in sec .[ conclusions ] .our starting point is a generic asymptotically free gauge theory with @xmath2 dirac flavors transforming according to the representation @xmath3 of the underlying gauge group .the relevant group - normalization factors are : @xmath4 = t[r ] \delta^{ab}\ ; , \qquad t^a_r t^a_r = c_2[r ] \mathbf{1}\ ; , \ ] ] where @xmath5 is the @xmath6-th group generator in the representation @xmath3 and @xmath7 $ ] .we denote with @xmath8 $ ] the dimension of the representation .the quantities @xmath9 $ ] and @xmath10 $ ] are related via the identity @xmath10 d[r ] = t[r ] d[g]$ ] .we summarise useful group theory factors in table [ factors ] ..relevant group factors for the representations used throughout this paper . however , a complete list of all the group factors for any representation and the way to compute them is available in table ii and the appendix of ref.@xcite . [cols="^,^,^,^",options="header " , ] because of the overall @xmath11 dependence of the shear viscosity it is convenient to normalise it to the entropy density .the so constructed ratio reads at a generic fixed point @xmath12 } \ ; , \label{etaoversfp}\ ] ] with @xmath13 and @xmath14 calculable definite positive and smooth functions of the number of colors and flavors , with @xmath15 the value of the coupling at the fixed point . as expected at non - interacting fixed points , such as the uv fixed point, the ratio diverges . on the other hand at the interacting ir fixed pointthe ratio approaches a finite value controlled by a small non - vanishing @xmath16 . in the left panel of fig .[ etas_nf ] we plot @xmath17 as function of the number of flavors , for fermions in the fundamental representation with @xmath18 .when decreasing the number of flavors below the asymptotically free boundary , where the shear viscosity diverges , we observe a dramatic decrease while still remaining much above the bound @xmath19 conjectured by ads / cft @xcite .it is natural to expect that , as we further decrease the number of flavors , the ir ratio further decreases to reach a minimum value at the lower boundary of the conformal window .below this critical number of flavors we expect the onset of chiral symmetry breaking and the theory in the deep ir becomes a theory of non - interacting pions with again a divergent value of this quantity . in the right panel of fig .[ etas_nf ] we present the temperature dependence of the shear viscosity over the entropy density for several values of @xmath2 .the quantity @xmath20 depends on the temperature over a reference scale @xmath21 via the gauge coupling .the reference energy scale is chosen to be the one for which the @xmath22 function displays a minimum occurring between the trivial uv and interacting ir fixed points .the energy scale @xmath21 is therefore defined by : @xmath23 for @xmath24 , for which the theory does not display an ir perturbative fixed point , @xmath21 is taken to be the scale at which the one - loop gauge coupling diverges as function of the temperature .the ratio @xmath20 decreases as we decrease the temperature for different values of the number of flavors within the conformal window .however for @xmath25 we observe that a minimum develops around @xmath26 .this happens because for this value of @xmath2 there is a temperature for which @xmath27 , which corresponds to a mininum for the @xmath28 function .evaluated at the ir fixed point , as a function of the number of flavors , for fermions in the fundamental representation with @xmath18 colors .right panel : @xmath29 as function of the temperature over the rg scale @xmath21 for different values of @xmath2 in the conformal window and one outside corresponding to @xmath24 , for @xmath30 colors .although @xmath31 still displays a potential ir fixed point the ir dynamics of @xmath29 can not be accessed perturbatively .the horizontal line at the bottom is the conjectured ads / cft bound.,title="fig:",scaledwidth=46.0% ] evaluated at the ir fixed point , as a function of the number of flavors , for fermions in the fundamental representation with @xmath18 colors .right panel : @xmath29 as function of the temperature over the rg scale @xmath21 for different values of @xmath2 in the conformal window and one outside corresponding to @xmath24 , for @xmath30 colors .although @xmath31 still displays a potential ir fixed point the ir dynamics of @xmath29 can not be accessed perturbatively .the horizontal line at the bottom is the conjectured ads / cft bound.,title="fig:",scaledwidth=46.0% ] we now move our attention to another relevant transport quantity , the fermion - number diffusion coefficient . the diffusion coefficient for the net number density of the fermion flavor @xmath6 is given in ref .@xcite and reads , at the next - to - leading - log level : @xmath32 } \biggl [ \sum_b^{f \bar{f } h } t[r_b ] \lambda_b + \frac{3 \pi^2}{8 } c_2[r_a ] \biggr]^{-1 } \frac{t^{-1}}{g^4 \ln(b\ , t/ m_d)}\ ; , \label{diff_coeff}\ ] ] where the sum extends over all particle species @xmath33 that the fermion species @xmath6 can scatter with in the process @xmath34 , mediated by a gauge boson .particles and antiparticles are counted separately , and the same goes for the helicity states : this means that we have to count a factor of four for every dirac fermion , and a factor of two for gauge bosons .furthermore , @xmath35 if the particle @xmath33 is a fermion , and @xmath36 if it is a boson .@xmath37 is a numerical coefficient , whose values relevant for the cases studied in this paper are reported in table [ ab ]. we can specialize eq . to our theory with @xmath38 gauge symmetry and @xmath2 fermions , all in the same representation @xmath3 .we obtain : @xmath39 } \biggl [ 4 n_f t[r ] + 4 n + \frac{3 \pi^2}{8 } c_2[r ] \biggr]^{-1 } \frac{t^{-1}}{g^4\ln(b \ , t/ m_d ) } \;.\ ] ] at very low energies , where the coupling is frozen at the fixed - point value @xmath40 , the dimensionless quantity @xmath41 can be plotted as a function of the number of flavors .this is represented in the left panel of fig.[d_nf ] , for the case of fermions in the fundamental representation and @xmath30 . as a function of the number of flavors , for the case of fermions in the fundamental representation and @xmath30right panel : @xmath42 as a function of the temperature , for different values of @xmath2 and @xmath30 ., title="fig:",scaledwidth=46.0% ] as a function of the number of flavors , for the case of fermions in the fundamental representation and @xmath30 .right panel : @xmath42 as a function of the temperature , for different values of @xmath2 and @xmath30 ., title="fig:",scaledwidth=46.0% ] as for the case of the shear viscosity - to - entropy density ratio , we observe that @xmath43 diverges as @xmath40 approaches the origin when increasing the number of flavors towards the asymptotic freedom boundary . as for the shear viscosity - to - entropy density ratio , in the right panel of fig .[ d_nf ] we also plot @xmath42 as a function of temperature for different values of the number of flavors in the conformal window and for @xmath44 .we determined the shear viscosity - to - entropy density ratio and the fermion - number diffusion coefficient within the perturbative regime of the conformal window for gauge - fermion theories .our formalism is valid for any fermionic matter representation , while the physical results , which are expected to hold generically , were elucidated via a three - color gauge theory as functions of the number of flavors in the fundamental representation .we observed that when the number of flavors decreases from the value at the loss of asymptotic freedom both the shear viscosity to entropy density as well as the fermion - number diffusion coefficient measured at the ir fixed point dramatically decrease .furthermore , for a given number of flavors within the perturbative conformal window both coefficients decrease ( albeit not monotonically for @xmath25 ) with the temperature from their divergent value in the uv down to the value at the ir fixed point .more specifically we discovered that the next - to - leading - log results are trustworthy down to almost 15 flavors .the ratio @xmath20 at the ir fixed point drops significantly when going from 16.25 to 15 flavors showing that a modest change in the number of flavors dramatically affects the dynamics of the theory encoded in the transport coefficients .higher - order corrections are needed to reach lower values of @xmath2 within the conformal window for the transport coefficients .in contrast , at zero temperature one observes that four and five loops are almost enough to cover the whole conformal window @xcite .although unproven it is reasonable to expect that the minimum of @xmath20 as function of temperature in qcd lies below the lowest value of @xmath20 obtained at the bottom of the conformal window , and therefore lower than the one obtained near 15 flavors .we thank guy d. moore for providing the numerical values of the constants @xmath45 and @xmath37 given in table ii .the work of f.s . and a.t .is partially supported by the danish national research foundation under the grant dnrf:90 .f. sannino and i. m. shoemaker , phys .d * 92 * , no . 4 , 043518 ( 2015 ) doi:10.1103/physrevd.92.043518 [ arxiv:1412.8034 [ hep - ph ] ] .d. h. rischke and f. sannino , phys .d * 92 * , no .6 , 065014 ( 2015 ) doi:10.1103/physrevd.92.065014 [ arxiv:1505.07828 [ hep - th ] ] .

<S> we present the conformal behavior of the shear viscosity - to - entropy density ratio and the fermion - number diffusion coefficient within the perturbative regime of the conformal window for gauge - fermion theories . </S><S> + [ .1 cm ] _ preprint : cp@xmath0-origins-2016 - 052 dnrf90 _ </S>

1604.01794

given a fluid velocity field @xmath1 , a passive tracer satisfies the linear advection equation @xmath2 where the scalar field @xmath3 denotes the concentration of the tracer at time @xmath4 and @xmath5 is its initial concentration at time @xmath6 .@xcite pointed out that laminar unsteady velocity fields can , over time , develop complex tracer patterns consisting of ever smaller scales .this observation has inspired the successful development of many stirring protocols to enhance mixing in engineered devices ( see , e.g. , @xcite ) .systematic classification of mixing efficiency of fluid flow , however , is relatively recent .this classification was initiated by @xcite who derived rigorous bounds on the mixing efficiency of velocity fields with a prescribed stirring energy or stirring power budget .a notable outcome of their program is the rather remarkable discovery of a finite - energy velocity field ( @xmath7 ) that achieves perfect mixing in finite time @xcite .it was shown later , however , that any such velocity field must have infinite viscous dissipation , i.e. @xmath8 @xcite .besides this fundamental limitation , the implementation of mathematically obtained optimal stirring strategies in a mixing device is not always feasible due to , for instance , geometric constraints .the problem is more acute in natural fluid flow ( such as geophysical flows or the blood stream ) over which we have virtually no control . in light of the above discussion, the natural question is : \(q ) _ given an unsteady velocity field , what is the optimal initial tracer pattern that leads to the most homogeneous mixture after a prescribed finite time ?_ in spite of its importance , this question has received relatively little attention .@xcite carried out a case study where the effect of the tracer injection location in a kenics mixer is examined .they find that , at least for short time horizons , the mixing efficiency depends significantly on the injection location .a similar case study is carried out by @xcite who studied the mixing efficiency of five different initial tracer patterns in a two - dimensional nonlinear model , known as the sine flow .@xcite addressed an analogous question : the asymptotic mixing of passive tracers advected under a steady velocity field where the tracer is injected continuously into the flow via source terms . through a variational approach , they determined the optimal distribution of the sources ( also see @xcite , for related numerical results ) . here, we address the finite - time mixing of passive tracers advected by fully unsteady velocity fields , as formulated in .specifically , we seek the optimal initial condition @xmath5 that leads to the most homogeneous mixture after a given finite time . to the best of our knowledge , a rigorous method for determining this optimal initial condition is missing .problem ( q ) can , in principle , be formulated and solved as an infinite - dimensional optimization problem , where the optimal initial condition coincides with the minimizer of an appropriate cost functional .such minimizers are typically obtained by iterative methods of adjoint - based optimization @xcite .this is , however , computationally prohibitive since it requires the backward - time integration of an adjoint partial differential equation ( pde ) at each iteration . here, we show that under reasonable assumptions , the problem reduces to a finite - dimensional one that can be readily solved at a relatively low computational cost . to obtain this finite - dimensional reduction, we assume that tracer blobs smaller than a small , prescribed length - scale @xmath0 are considered completely mixed for all practical purposes .this assumption , that is made precise in section [ sec : theory ] , results in a natural galerkin truncation of the perron frobenius ( pf ) operator associated with the advection equation .we show that the optimal initial condition @xmath5 then coincides with a singular vector of the truncated pf operator .our results complement the transfer operator - based methods for detecting finite - time coherent sets in unsteady fluid flows ( @xcite ; also see @xcite ) .coherent sets refer to subsets of the fluid which exhibit minimal deformation under advection and therefore inhibit efficient mixing of tracers with the surrounding fluid .our aim here is the opposite , namely , initially large - scale structures that under advection deform mostly into small - scale filaments .the remainder of the paper is organized as follows . in section [ sec : prelim ] , we introduce some basic notation and definitions .section [ sec : theory ] contains our main results and section [ sec : numerics ] details their numerical implementation . in section [ sec : results ] , the results are demonstrated on two examples .consider an unsteady , incompressible velocity field @xmath1 defined over a bounded open subset @xmath9 where @xmath10 or @xmath11 for two- and three - dimensional flows , respectively .the trajectories @xmath12 of the fluid particles satisfy the ordinary differential equation @xmath13 where @xmath12 denotes the time@xmath4 position of the particle starting from the initial position @xmath14 at time @xmath6 .if the velocity field is sufficiently smooth , there exists a two - parameter family of homeomorphisms @xmath15 ( the flow map ) such that @xmath16 for all times @xmath4 and @xmath17 . as our interest here is in finite - time mixing ,we restrict our attention to a prescribed finite time interval @xmath18 $ ] of interest .the flow map @xmath19 takes the initial position @xmath14 of a fluid particle at time @xmath6 to its final position at time @xmath20 .since the finite time interval is fixed , we drop the dependence of the flow map on @xmath6 and @xmath20 , and write @xmath21 for notational simplicity .let @xmath3 denote the concentration of a passive tracer , i.e. @xmath22 satisfies equation .since the passive tracer is conserved along fluid trajectories , we have @xmath23 for all @xmath24 . note that since the flow map is a homeomorphisms , the inverse @xmath25 is well - defined .equation motivates the definition of the perron - frobenius ( pf ) operator .the perron frobenius operator associated with the flow map @xmath26 is the linear transformation @xmath27 such that , for all @xmath28 , @xmath29 [ def : pf ] the evolution of passive tracers can be described by the action of the pf operator on their initial conditions .more specifically , for the passive tracer @xmath22 described above , we have @xmath30 for all @xmath31 ( cf .equation ) .we point out that there is a more general definition of the pf operator applicable to non - invertible dynamics ( see definition 3.2.3 of @xcite ) . in the special case where the flow map @xmath21 is invertible and volume - preserving ,the general definition is equivalent to definition [ def : pf ] above ( corollary 3.2.1 in @xcite ; @xcite ) . for incompressible flow ,the pf operator is a unitary transformation with respect to the @xmath32 inner product @xmath33 . as a consequence ,the @xmath34-norm @xmath35 of the tracer remains invariant under advection .furthermore , the spatial average of the tracer is an invariant . without loss of generality, one can assume that this spatial average vanishes , @xmath36 @xcite .there has been several attempts to detect coherent structures in unsteady fluid flows using approximations of the pf operator @xcite .@xcite puts these approaches on a mathematically rigorous basis by composing the pf operator with diffusion operators .the resulting diffusive pf operator is compact and has a well - defined singular value decomposition ( svd ) .@xcite shows that a singular vector , corresponding to the largest non - unit singular value of the diffusive pf operator , can reveal minimally dispersive subsets of the fluid that remain coherent and thereby inhibit mixing ( also see @xcite ) .our goal here , however , is the opposite as we seek passive tracer initial conditions that mix most efficiently with their surrounding fluid .given an initial tracer distribution , a reasonable mixer will generically deform the tracer through stretching and folding of material elements such that , over time , it develops ever smaller length scales .it is , therefore , desirable to release the tracer initially into smallest possible scales . in practice ,the initially available range of scales into which the tracer may be released is limited to relatively large scales .we denote this large length - scale limit by @xmath37 ( see figure [ fig : schem_scales ] , for an illustration ) .it is left then to the fluid flow to transform the initially large - scale blobs of tracer to small filaments through a stretch - and - fold mechanism . ) and the mixed scales ( smaller than @xmath0 ) . ] on the other hand , we assume that there is a small length scale threshold @xmath38 , under which the tracer is considered , for all practical purposes , completely mixed .an efficient mixer , therefore , transfers the tracer distribution from large , initially available scales @xmath39 to the mixed scales @xmath40 . in the next section, we make these statements precise .we consider a set of functions @xmath41 forming a complete , orthonormal basis for the space of square integrable functions @xmath32 .that is @xmath42 and for any @xmath43 there are constants @xmath44 such that @xmath45 .we also assume that there is a length scale @xmath46 associated to each function @xmath47 , and that they are ordered such that the sequence @xmath48 is decreasing .in other words , the length scale associated with the function @xmath47 decreases as @xmath49 increases .such a basis can be taken , for instance , to be fourier modes or wavelets @xcite . with this basis, we can mathematically model the subspace of initial conditions @xmath50 .the subspace @xmath50 consists of all scalar functions @xmath5 whose smallest length scale is larger than or equal to @xmath37 .since the basis @xmath41 is ordered , there is a positive integers @xmath51 such that @xmath52 the subspace of unmixed length scales @xmath53 can be modeled similarly using a basis @xmath54 for @xmath32 .we assume that this basis is also orthonormal , complete and associated with a decreasing sequence of length scales .the subspace @xmath53 consists of all scalar functions whose smallest length scale is larger than or equal to the unmixed length scale @xmath0 . therefore , there is @xmath55 such that @xmath56 note that the bases @xmath54 and @xmath57 can be taken to be identical , but this is not necessary here .we denote the orthogonal complement of @xmath53 by @xmath58 . in terms of the basis functions , we have @xmath59 where the overline denotes closure in the @xmath34 topology .the space @xmath58 consists of functions that only contain the mixed scales , that is scales smaller than @xmath0 ( see figure [ fig : schem_scales ] ) . given an initial condition @xmath60 for the tracer , its advected image @xmath61 at the final time can potentially contain all length scales @xmath46 .the flow redistributes the ` energy ' budget of the tracer among various scales in such a way that the @xmath34-norm is conserved , i.e. , @xmath62 a tracer is better mixed if more of its energy budget is transfered to the mixed scales @xmath63 . therefore , we seek optimal initial conditions @xmath60 such that the energy budget of its image @xmath64 is mostly stored in the mixed scales , maximizing @xmath65 . to make these statements more precisewe introduce the following truncation of the pf operator .we define the truncated pf operator @xmath66 as the linear map @xmath67 , where @xmath68 is the orthogonal projection onto the @xmath69-dimensional subspace @xmath53 .we also define the remainder operator @xmath70 as @xmath71 .[ def : pftrunc ] it follows from parseval s identity that @xmath72 ( see equation ) .the quantity @xmath73 represents the portion of the energy budget of the tracer that remains unmixed after advection to the final time @xmath20 .the quantity @xmath74 , on the other hand , represents the portion of the tracer that is completely mixed .we , therefore , seek initial conditions @xmath60 that maximize the mixed energy budget @xmath74 .since the truncated pf operator @xmath75 is a linear transformation between finite - dimensional vector spaces @xmath50 and @xmath53 , it can be represented by a matrix @xmath76 .more specifically , for any @xmath60 , there are scalars @xmath77 and @xmath78 such that @xmath79 the matrix @xmath80 maps @xmath81 into @xmath82 , that is @xmath83 .it follows from elementary linear algebra that the entries @xmath84_{ij}$ ] of the matrix @xmath80 are given by @xmath85_{ij}=\langle \mathcal p\phi_j,\psi_i\rangle_{l^2 } , \quad i\in\{1,2,\cdots , n\},\quad j\in\{1,2,\cdots , n\}. \label{eq : kmatrix}\ ] ] with this prelude , we can now state our main result .consider the function spaces @xmath50 and @xmath53 and their associated truncated pf operator defined above .the solution of @xmath86 with the maximum taken over all @xmath60 with @xmath87 , is given by @xmath88 , where @xmath89 is a right singular vector of the truncated pf matrix corresponding to its smallest singular value .[ thm : main ] since @xmath90 , maximizing @xmath91 is equivalent to minimizing @xmath92 . since @xmath5 belongs to the subspace @xmath50 , the initial condition @xmath5 and its image @xmath93 can be expressed by the series with @xmath94 . denoting the standard euclidean norm by @xmath95, we have @xmath96 and @xmath97 .therefore , @xmath98 the minimum on the right hand side is well - known to coincide with the smallest singular value of the matrix @xmath80 @xcite .the minimum is attained at the corresponding right singular vector of the matrix @xmath80 .this completes the proof .once the pf matrix @xmath80 is formed , the evaluation of the optimal initial condition @xmath99 , from the above theorem , is straightforward .we outline the computation of the truncated pf matrix @xmath80 in section [ sec : numerics ] . note that if the matrix @xmath80 is not full - rank , there are initial conditions @xmath5 of the form with @xmath100 , such that @xmath101 .such initial conditions result in ` perfect mixing ' since their advected image @xmath102 belongs entirely to the mixed scales @xmath40 , i.e. , @xmath103 . in the examples studied in section [ sec : results ] , such perfect finite - time mixing was not observed . in other words ,the matrices @xmath80 are full - rank in these examples .we emphasize that the truncated pf operator @xmath75 is _ not _ used as an approximation of the full pf operator @xmath104 . instead ,the truncation @xmath75 followed naturally from the physical assumption that length scales @xmath40 are completely mixed .as is clear from equation , to evaluate the truncation @xmath75 , one still needs to utilize the full pf operator to evaluate the terms @xmath105 .numerical computation of the optimal initial condition @xmath99 relies on the scale - dependent bases @xmath106 and @xmath107 . for completeness, we discuss two such bases : the fourier basis and the haar wavelet basis . since the examples considered in section [ sec : results ] below are defined on equilateral two - dimensional domains , @xmath108\times [ 0,l]$ ] , we focus on this special case .the generalization to the rectangular domain and to the three - dimensional case is straightforward . for periodic boundary conditions, it is natural to use the fourier basis to define the spaces @xmath50 and @xmath53 . the orthonormal fourier basis associated with the two - dimensional domain @xmath108\times [ 0,l]$ ] consist of functions @xmath109 $ ] where @xmath110 denotes the wave vector .the length scale associated to each fourier mode is inversely proportional to the wave number , @xmath111 .we take the space of available initial scalar fields to be the functions whose fourier modes contain at most a prescribed wave number @xmath112 , i.e. , @xmath113 : { \mathbf}k=(k_x , k_y)\in \mathbb z^2 , |k_x|\leq k_i , |k_y|\leq k_i\right\}. \label{eq : vi_ki}\ ] ] similarly , the space of unmixed scales @xmath114 is the functions whose fourier modes contain at most a prescribed wave number @xmath115 , i.e. , @xmath116 : { \mathbf}k=(k_x , k_y)\in \mathbb z^2 , |k_x|\leq k_\nu , |k_y|\leq k_\nu\right\}. \label{eq : vnu_knu}\ ] ] more generally , one could define the space @xmath50 ( and similarly @xmath53 ) with independent upper bounds @xmath117 and @xmath118 on the wave - numbers @xmath119 and @xmath120 , respectively .since the domain is equilateral , and for simplicity , we choose the same upper bounds in both directions , @xmath121 .since the tracer concentration is real - valued , the complex conjugate basis functions in @xmath50 and @xmath53 are redundant .also , the basis with @xmath122 ( corresponding to constant functions ) is unnecessary since we assumed that the tracer has zero mean . excluding these redundant functions , the effective dimension of the vector spaces @xmath50 and @xmath53 are @xmath123 and @xmath124 , respectively .while the above fourier basis is a convenient choice , it restricts its applicability to the periodic boundary conditions .more general boundary conditions can be handled with an alternative basis , such as haar wavelets .such wavelet bases have the added advantage that they can be localized in space in addition to scale .this property renders wavelets particularly attractive in applications where the tracer can only be released into a subset of the fluid domain @xmath125 due to geometric or design constraints .contrast this with the global nature of the fourier basis . , @xmath126 .the domain is @xmath127\times[0,1]$ ] ] here , we consider the haar wavelet basis . for completeness, we briefly review the construction of this basis in two dimensions .denote the one - dimensional haar scaling function with @xmath128 and the corresponding wavelet with @xmath129 where @xmath130 is the indicator function of the set @xmath131 . by dilations and translations ,we obtain @xmath132 @xmath133 where @xmath134 for a domain of size @xmath135 . the collection of the wavelets @xmath136 forms an orthogonal basis for mean - zero functions in @xmath137)$ ] @xcite .the integer @xmath49 determines the size of the support of @xmath136 ( or @xmath138 ) which is @xmath139 . since the wavelets with larger @xmath49 resolve finer structures ( or smaller length scales ) , the integer @xmath49 is referred to as the scale of the wavelet .the integer @xmath140 , on the other hand , introduces a translation in the support of each wavelet , introducing a space dependence at each scale @xmath49 .the functions @xmath138 and @xmath136 serve as the building blocks of multidimensional wavelet bases @xcite .for instance , a complete orthonormal basis for mean - zero functions in @xmath32 , with @xmath108\times [ 0,l]$ ] , is formed by the set of functions @xmath141 where @xmath142 @xmath143@xmath144 [ eq : wavelet_2d_02 ] the prefactor @xmath145 ensures that each basis function is of unit norm . the integer @xmath49 determine the scale in both @xmath146 and @xmath147 directions , while the integers @xmath148 and @xmath149 introduce the corresponding translations .figure [ fig : wavelets ] shows three examples of the two - dimensional wavelet functions with @xmath150 .the construction of two - dimensional wavelet bases from one - dimensional wavelets is not unique . for an alternative wavelet basissee , e.g. , chapter 10 of @xcite . using the wavelet basis , we define the subspace of initial conditions @xmath50 as @xmath151 where the integer @xmath152 sets the initially available length scales .roughly speaking , the wavelet subspace @xmath153 contains tracer blobs of size @xmath154 or larger .similarly , we define the subspace of unmixed length scales by @xmath155 containing the unmixed tracer blobs of size @xmath156 or larger . for given positive integers @xmath152 and @xmath157, we have @xmath158 and @xmath159 .recall that the basis functions @xmath160 spanning the domain @xmath50 of the truncated pf operator @xmath75 need not to be identical to the basis functions @xmath161 spanning its range @xmath53 . as a result, the fourier - based subspaces and can be used in conjunction with the wavelet - based subspaces and . in the following ,we consider examples with both fourier - based and wavelet - based subspaces and for defining the domain @xmath50 . for the range @xmath53 ,however , we only consider the fourier - based subspace in order to achieve speedup in the computations by taking advantage of the fast fourier transform ` fftw ` .+ once the choice of bases is made , the truncated pf matrix can be computed by evaluating the integral , @xmath85_{ij}=\langle \mathcal p\phi_j,\psi_i \rangle_{l^2}:=\int_{\mathcal d}(\mathcal p\phi_j)({\mathbf}x)[\psi_i({\mathbf}x)]^\ast{\mathrm d}{\mathbf}x , \label{eq : kmatrix_2}\ ] ] where @xmath162 denotes the complex conjugation .we approximate this integral using the standard trapezoidal rule @xcite .to ensure the accuracy of the approximation , the results reported in section [ sec : results ] are computed using a dense uniform grid @xmath163 of @xmath164 collocation points over the domain @xmath108\times [ 0,l]$ ] .the terms @xmath105 are computed from the definition of the pf operator ( definition [ def : pf ] ) , i.e. , @xmath165 for any @xmath166 .( see @xcite , for more accurate numerical methods ) .as the first example , we consider the time - periodic sine flow @xcite .this model is simple enough to unambiguously demonstrate our results , yet it can exhibit complex dynamics with simultaneous presence of chaotic mixing and coherent vortices .the sine flow has a spatially sinusoidal velocity field on the domain @xmath167\times [ 0,1]$ ] with periodic boundary conditions .the temporal period of the flow is @xmath168 for some @xmath169 . during the first @xmath170 time units , the velocity field is @xmath171 and switches instantly to @xmath172 for the second @xmath170 time units .this process repeats iteratively .starting from @xmath173 initial conditions distributed uniformly over the domain @xmath174\times [ 0,1]$ ] .the domain is periodic in both directions . ]the sine flow generates a reversible map @xmath175 that , over one period , maps points @xmath176 to @xmath177 . the inverse of the map @xmath175 is given explicitly by @xcite @xmath178 figure [ fig : pmap ] shows @xmath179 iterations of this map with @xmath180 launched from a uniform grid of initial conditions .the map @xmath181 has two hyperbolic fixed points located at @xmath182 and @xmath183 whose tangle of stable and unstable manifolds creates a chaotic mixing region .in addition , the map has two elliptic fixed points located at @xmath184 and @xmath185 .these elliptic fixed points are surrounded by invariant kolmogorov arnold - moser ( kam ) tori with quasi - periodic motion that inhibit mixing @xcite . for the sine map , with @xmath50 being the fourier - based subspace defined in .four optimal initial conditions with @xmath186 , @xmath187 , @xmath188 and @xmath189 are shown in the top panel .the range of the truncated pf operator @xmath75 is the fourier - based subspace @xmath53 with @xmath190 .the bottom panel shows their corresponding advected image @xmath191 under @xmath179 iterations of the sine map .all figures show the entire domain @xmath127\times [ 0,1]$ ] . ]it is known that mixing is more efficient around the hyperbolic fixed points due to their tangle of stable and unstable manifolds @xcite .the kam regions , in contrast , form islands of coherent motion that inhibit efficient mixing of passive tracers .therefore , it is desirable to release the tracer blobs around the hyperbolic fixed points , avoiding the kam region . here , we examine whether the optimal initial condition given by theorem [ thm : main ] agrees with this intuitive assessment . for the sine map .four optimal initial conditions with @xmath192 , @xmath187 , @xmath188 are shown in the top panel .the bottom panel shows their corresponding advected images @xmath191 under @xmath179 iterations of the sine map .the range of the truncated pf operator @xmath75 is the fourier - based subspace @xmath53 with @xmath190 .all figures show the entire domain @xmath127\times [ 0,1]$ ] . ] for the finite - time analysis , we consider the flow under @xmath179 iterations of the sine map , i.e. we set the flow map @xmath193 .first , we consider the fourier - based initial subspace @xmath50 defined in .figure [ fig : optic ] shows the optimal initial conditions obtained from theorem [ thm : main ] with @xmath194 and @xmath189 . for all parameter values @xmath195 ,the optimal initial condition consists of two prominent blobs centered at the hyperbolic fixed points @xmath182 and @xmath183 . for @xmath186 ,only very large scales are available for the distribution of the tracer blob and therefore some intersection with the kam region is inevitable .as the number of available wave numbers @xmath195 ( or equivalently , available initial length scales ) increases the blobs become more concentrated at the hyperbolic fixed points . even for @xmath196 , the optimal initial condition has very small but non - zero concentration in the kam regions .this is due to the global nature of the fourier modes which inhibits the perfect localization around the hyperbolic fixed points .the wavelet - bases subspace does not suffer from this drawback .figure [ fig : optic_wavelet ] , for instance , shows three optimal initial conditions in this wavelet - based subspace . for @xmath192 , where only the largest scales are available ,intersection with the kam region is inevitable ( similar to the case of @xmath186 in figure [ fig : optic ] ) .as the smaller scales become available , the optimal initial condition @xmath99 concentrates around the hyperbolic fixed points with no concentration at the kam regions . the results in figures [ fig : optic ] and [ fig : optic_wavelet ]are computed using the fourier - based subspace @xmath53 with @xmath190 . to ensure the insensitivity of the results to perturbations , we recomputed them by varying the cut - off wavenumber in the interval @xmath197 and obtained almost identical optimal initial conditions . as the second example , we consider a fully unsteady flow obtained from a direct numerical simulation of the two - dimensional navier stokes equation , @xmath198 with the dimensionless viscosity @xmath199 and a band - limited stochastic forcing @xmath200 .the flow domain is the box @xmath201\times[0,2\pi]$ ] with periodic boundary conditions .a standard pseudo - spectral code with 2/3 dealiasing was used to numerically solve the navier stokes equations ( see section 6.2 of @xcite for further computational details ) . .( b ) the vorticity field at the final time @xmath20 .( c ) the forward - time ftle field corresponding to the time interval @xmath18 $ ] . here , @xmath175 is @xmath202 time units . ] starting from a random - phase initial condition , we numerically integrate the navier stokes equation .after @xmath203 time units the flow has reached a statistically steady turbulent state with reynolds number @xmath204 .we set this time as the initial time @xmath6 for the mixing analysis .the final time instance is set to @xmath20 with @xmath205 .figures [ fig:2dturb_vort](a , b ) show the vorticity fields at these initial and final times . as is typical of two - dimensional turbulence, the flow contains several coherent vortices that exhibit minimal material deformation over the time interval @xmath18 $ ] @xcite .these coherent vortices are signaled by the islands of small finite - time lyapunov exponent ( ftle ) shown in figure [ fig:2dturb_vort](c ) .the ftle field is computed as @xmath206/(2t)$ ] for all @xmath31 , with @xmath207 being the largest eigenvalue of the cauchy green strain tensor @xmath208^\top{\mathrm d}{\varphi}({\mathbf}x)$ ] , and @xmath209 denoting the jacobian of the flow map @xcite . outside the coherent vorticesthe flow is mostly chaotic , dominated by the stretching and folding of material lines .( upper panel ) in the turbulent flow with @xmath194 and @xmath189 .their corresponding advected images @xmath191 at the final time @xmath20 are shown in the lower panel . in all four cases shown here ,the mixed wavenumber is @xmath190 ( see equation ) .all panels show the entire domain @xmath201\times [ 0,2\pi]$ ] . ]next we compute the optimal initial conditions @xmath99 . unlike the sine map ,the preimages @xmath210 are not explicitly known here .we numerically evaluate the preimages by integrating the ode backwards in time from the final time @xmath20 to the initial time @xmath6 , for each initial condition @xmath166 .this numerical integration is carried out by the fifth - order runge - kutta scheme of @xcite .since the velocity field @xmath211 is stored on a discrete spatiotemporal grid , it needs to be interpolated for the particle advection . here, we use cubic splines for the spatial interpolation of the velocity field together with a linear interpolation in time . for the optimal initial concentrations @xmath212 .the optimal initial conditions for @xmath186 , @xmath187 , @xmath188 and @xmath189 are shown in figure [ fig:2dturb ] . blue line ( circles ) : the mix - norm @xmath213 for the initial concentrations @xmath214 .[ fig : mixnorm ] ] figure [ fig:2dturb ] shows the optimal initial tracer patterns @xmath99 for @xmath194 and @xmath189 , which belong to the corresponding fourier - based subspaces @xmath50 as defined in equation . as opposed to the simple model considered in section [ sec : sinemap ] ,the optimal tracer patterns here have fairly complicated structures .this is to be expected as the turbulent flow itself has a complex spatiotemporal structure . ideally , the tracer should concentrate outside the coherent vortices to achieve better mixing .similar to the sine flow , for @xmath186 , where only the very large scales are available for the release of the tracer , there is some inevitable overlap between the coherent vortices and the tracer .this results in the visibly unmixed blobs in the advected tracers @xmath191 shown in the lower panel of figure [ fig:2dturb ] .theorem [ thm : main ] , however , guarantees that the optimal initial condition @xmath99 is such that the unmixed blobs are minimal . as smaller scalesbecome available ( @xmath215 ) , the intersection of the high initial tracer concentration and the coherent vortices becomes smaller , leading to a more homogeneous mixture after advection to the final time @xmath20 .( upper panel ) in the turbulent flow with @xmath216 and @xmath217 .their corresponding advected images @xmath191 at the final time @xmath20 are shown in the lower panel . in all four cases shown here ,the mixed wavenumber is @xmath190 ( see equation ) .all panels show the entire domain @xmath201\times [ 0,2\pi]$ ] . ] to quantify the mixture qualities , we compute the mix - norm of the advected tracers proposed by @xcite .this mix - norm is the sobolev @xmath218 norm , @xmath219 where the hat sign denotes the fourier transform .@xcite proposed the alternative sobolev norm @xmath220 for quantifying the mixture quality .the motivation for using such sobolev norms is that the density of homogeneous mixtures are concentrated at ever smaller scales or equivalently larger wave numbers @xmath221 . as a result ,more homogeneous mixtures have smaller mix - norms .the mix - norm @xmath222 is shown in figure [ fig : mixnorm ] for the optimal initial conditions @xmath223 with @xmath224 .as @xmath195 increases , more homogeneous mixtures are obtained , as is also visible in figure [ fig:2dturb ] .for comparison , we also show the mix - norm @xmath213 for the non - optimal initial conditions @xmath225 .the non - optimal initial conditions result into a larger mix - norm , showing that they do not mix as well as the optimal initial conditions @xmath99 do .figure [ fig : optic_wavelet ] shows the optimal initial conditions found in the wavelet - based subspace .their mix - norms exhibit a similar behavior as the one shown in figure [ fig : mixnorm ] .the design of mixing devices has primarily been concerned with the stirring protocols that enhance mixing .the optimality of these protocols are limited by design constraints and fundamental physics @xcite . on the other hand , for a given stirring protocol, the final mixture quality also depends on the initial configuration of the tracer .the optimal initial condition for the release of the tracer has received far less attention . here , we proposed a rigorous framework for determining the optimal initial tracer configuration to achieve maximal mixing under finite - time passive advection .we showed that , under reasonable assumptions , the problem reduces to a finite - dimensional optimization problem .the optimal initial condition then coincides with a singular vector of a truncated perron frobenius ( pf ) operator .this truncation is not an approximation of the infinite - dimensional pf operator ; it rather follows naturally from our simplifying assumption that the tracer blobs smaller than a prescribed critical length scale @xmath0 are completely mixed .we discussed two numerical implementations of the optimization problem using fourier modes and haar wavelets . while the fourier modes are convenient for the spatially periodic flows considered here , the wavelets are more suitable for handing more complicated geometries and boundary conditions .wavelets also allow for optimal initial conditions that are local in both space and scale .the space localization is crucial in many applications where the tracer can only be released into a subset of the flow domain duo to geometric constraints .we restricted our attention here to ideal passive tracers .future work will expand the framework to account for diffusion and the presence of sinks and sources .diffusion , in particular , dictates a dissipative length scale @xmath0 for mixed blobs which , in the absence of diffusion , was prescribed here in an ad - hoc manner .+ * acknowledgments : * i would like to thank daniel karrasch for pointing out the correct terminology in definition [ def : pf ] and bringing a number of relevant references to my attention .i am also grateful to charles doering , gary froyland , george haller and christopher miles for their comments on this manuscript .m. dellnitz , g. froyland , and o. junge .the algorithms behind gaio set oriented numerical methods for dynamical systems . in _ergodic theory , analysis , and efficient simulation of dynamical systems _ , pages 145174 .springer , 2001 .m. farge , k. schneider , and n. kevlahan .non - gaussianity and coherent vortex simulation for two - dimensional turbulence using an adaptive orthogonal wavelet basis ._ phys . fluids _ , 110 ( 8):0 21872201 , 1999 . g. froyland and k. padberg - gehle .almost - invariant and finite - time coherent sets : directionality , duration , and diffusion . in _ ergodic theory , open dynamics , and coherent structures _ , pages 171216 .springer , 2014 .o. gubanov and l. cortelezzi .sensitivity of mixing optimization to the geometry of the initial scalar field . in l.cortelezzi and i. mezi , editors , _ analysis and control of mixing with an application to micro and macro flow processes _ , pages 369405 .springer , vienna , 2009 .e. lunasin , z. lin , a. novikov , a. mazzucato , and c. r. doering .optimal mixing and optimal stirring for fixed energy , fixed power , or fixed palenstrophy flows ._ j. math ._ , 530 ( 11):0 115611 , 2012 .t. okabe , b. eckhardt , j .- l . thiffeault , and c. r. doering .mixing effectiveness depends on the source sink structure : simulation results . _ journal of statistical mechanics : theory and experiment _ , 20080 ( 07):0 p07018 , 2008 .n. santitissadeekorn , g. froyland , and a. monahan .optimally coherent sets in geophysical flows : a transfer - operator approach to delimiting the stratospheric polar vortex ._ physical review e _ , 820 ( 5):0 056311 , 2010

<S> the efficiency of a fluid mixing device is often limited by fundamental laws and/or design constraints , such that a perfectly homogeneous mixture can not be obtained in finite time . here </S><S> , we address the natural corollary question : given the best available mixer , what is the optimal initial tracer pattern that leads to the most homogeneous mixture after a prescribed finite time ? for ideal passive tracers , </S><S> we show that this optimal initial condition coincides with the right singular vector ( corresponding to the smallest singular value ) of a suitably truncated perron - frobenius ( pf ) operator . </S><S> the truncation of the pf operator is made under the assumption that there is a small length - scale threshold @xmath0 under which the tracer blobs are considered , for all practical purposes , completely mixed . </S><S> we demonstrate our results on two examples : a prototypical model known as the sine flow and a direct numerical simulation of two - dimensional turbulence . evaluating the optimal initial condition through this framework only requires the position of a dense grid of fluid particles at the final instance and their preimages at the initial instance of the prescribed time interval . as such </S><S> , our framework can be readily applied to flows where such data is available through numerical simulations or experimental measurements . </S>

hep-ph0205249

one of the major challenges in qcd is the description of high energy scattering phenomena . in the high center - of - mass energy @xmath2 and in the perturbative domainwhen @xmath3 the scattering amplitude is obtained by the summation of diagrams leading in @xmath4 @xcite . at this level , when @xmath5 is frozen , the dependence of the resulting cross section on the energy is governed by the power law @xmath6 where @xmath1 is the bjorken variable .the critical exponent @xmath7 ( @xmath8 ) is provided by the minimum of the eigenvalue function @xmath9 of the bfkl evolution kernel .the conceptual problem in this approach is the fact that at sufficiently high center - of - mass energies the bfkl amplitude violates the froissart unitarity bound .this means that the validity of this approach is strongly limited and it has to be modified at very small @xmath1 in order to guarantee the unitarity of the resulting cross section .the solution to this problem can be provided by including the recombination effects , which are likely to occur at very small values of @xmath1 @xcite . by decreasing the value of @xmath1 at fixed gluon virtuality @xmath10 ,the density of partons becomes so large that they start to overlap . in this casethe gluon splitting process must be supplemented by a competing gluon recombination . in terms of evolutionthe master equations become nonlinear with an additional quadratic term which suppress the growth of the amplitude with energy and restore the unitarity .there have been extensive studies on this problem , see @xcite , @xcite-@xcite , which result in the nonlinear evolution .one of the important outcomes of these studies is the existence of the saturation scale @xmath11 which is a characteristic scale at which the parton recombination effects become important .in particular case of the balitsky - kovchegov equation @xcite the existence of such scale has yet another important impact on the picture of the bfkl evolution . the diffusion into the infra - red , which is the charactersitic property of bfkl evolution is strongly limited due to the existence of the saturation scale @xcite . in fact , in the regime when the gluon transverse momenta @xmath12 , the solution to the nonlinear equation @xcite becomes a function of only one combined variable @xmath13 @xcite-@xcite . in the regime of high momenta ,@xmath14 the parton density is small and the evolution is governed by a normal linear equation . in this talkwe present the analytical and numerical analysis of the balitsky - kovchegov @xcite equation which is a nonlinear evolution equation in the leading @xmath4 limit .we illustrate the emergence of the saturation scale and scaling and show that it leads to the suppression of the infra - red diffusion .we also consider the case with additional nll effects such as kinematical constraint and running coupling .the results presented in this talk have been obtained in the collaboration with k. golec - biernat and l. motyka . for the details of the calculation the readeris referred to @xcite .the balitsky - kovchegov equation @xcite has been derived as an evolution equation for the dipole - nucleus amplitude in the dipole picture by a summation of multiple pomeron exchanges in the leading @xmath15 level and in the large @xmath16 limit .the resulting evolution equation reads @xmath17 where @xmath18 , and the linear term is determined by the bfkl kernel @xmath19 the function @xmath20 is the dipole - nucleus amplitude for the scattering of the dipole with transverse size @xmath21 at impact parameter @xmath22 and at rapidity @xmath23 . in the linear approximation , when each dipole scatters only once off the nucleus , the bfkl equation in the dipole picture is obtained . the non - linear term in ( [ eq : kov ] ) takes into account multiple scatterings and is essentially determined by the triple pomeron vertex @xcite in the large @xmath16 limit .( [ eq : kov ] ) unitarizes the bfkl pomeron in the sense that at @xmath24 and @xmath25 fixed , @xmath26 thus the power - like rise with energy for the bfkl pomeron is tamed @xcite .for the subsequent analysis we shall assume the approximation of the big nucleus , i.e. @xmath27 which allows us to factorize the impact parameter dependence in eq.([eq : kov ] ) .we also consider the spherical symmetric solutions in @xmath21 and transform the equation ( [ eq : kov ] ) into the momentum space by performing the fourier transform @xmath28 where @xmath29 is the bessel function . in this casethe following equation is obtained @xmath30 and the action of the bfkl kernel is given by @xmath31 where now @xmath32 and @xmath33 are the transverse momenta of the exchanged gluons in the bfkl ladder .in order to study the diffusion and scaling properties of the solution we shall consider the function @xmath34 . in the case of the linear bfkl equation this function is a gaussian concentrated around some initial scale @xmath35 and with width increasing with rapidity , leading to a diffusion . in fig.[fig : distr ] we illustrate this distribution for the case of the solution to the linear bfkl and in the nonlinear balitsky - kovchegov equation for different choices of rapidity . as an initial conditionwe have chosen a delta function @xmath36 . in the case of the linear bfkl evolutionthe solution is always peaked at @xmath37 and exhibits the well known diffusion pattern , in which the momentum distribution has an increasing width with growing rapidity .in the nonlinear case however the solution behavies quite differently .the peak of the distribution @xmath38 moves towards higher @xmath32 values as the rapidity increases and the solution becomes washed out from the @xmath39 regime . only in the region@xmath40 it coincides with the linear evolution .the impact of unitarization of the bfkl pomeron on the infra - red behaviour can be also visualised by studying the properties of the following normalised distribution @xmath41 and by performing the projection of this function onto the ( @xmath42 ) plane , fig .[ fig : proj ] . again , for small @xmath23 , when the non - linearity in the bk equation is negligible , the re - normalized solutions ( [ eq : norm ] ) of the bfkl and the bk equations coincide . with increasing y ,when the non - linear effects become important , the difference between them in the region of small @xmath32 becomes fully visible .note that in certain region of ( @xmath42 ) in fig .[ fig : proj ] the contours become parallel straight lines .this means that @xmath43 in this region is a function of the combination @xmath44 instead of @xmath32 and @xmath23 separately .this is referred as the geometric scaling property .we note , that this scaling holds in the regime where @xmath45 and becomes violated when @xmath46 .this suggests that we can identify the saturation scale @xmath47 with @xmath38 @xmath48 with the exponential dependence on rapidity governed by the value of the scaling parameter @xmath49 .the solution to eq .( [ eq : newkov ] ) has the same scaling property as the function @xmath50 namely @xmath51 provided that @xmath52 .we have checked that this condition is satisfied .we have checked that the scaling coefficient @xmath49 ( [ eq : satscale ] ) is a universal quantity and it does not depend on the type of the input distribution whereas the normalisation @xmath53 is determined by the initial conditions .the transition between the scaling and linear regime can be perhaps best seen in fig .[ fig : trans ] where the function @xmath54 is plotted as a function of the scaling variable @xmath55 for different values of the rapidity @xmath23 .the scaling behaviour is represented by a common line to the left of the point @xmath56 .the slow departure from scaling for @xmath57 is clearly visible . in this sensethe line @xmath58 is only an approximation characterizing the position of the transition region in the @xmath59-plane .however , the choice based on the position in @xmath32 of the maximum of @xmath60 as a function of @xmath23 is the most natural one . from the numerical solutionwe have extracted the value of the scaling parameter and found @xmath61 which is in agreement with the estimates of @xcite for the balitsky - kovchegov equation and also with the previous work @xcite .the geometric scaling is also the property of the golec - biernat and wsthoff saturation model @xcite which succesfully described the data on inclusive and diffractive proton structure function . in this modelit is assumed that the whole energy dependence of the dipole cross section @xmath62 is driven through the saturation scale @xmath47 ( or saturation radius @xmath63 ) .it then results in an approximate scaling property of the data for the total photon - proton cross section @xmath64 such scaling law was found in the small-@xmath1 dis data @xcite .the balitsky - kovchegov equation has been derived in the leading @xmath15 level , therefore it is important to study the impact of the subleading corrections . although it has been argued @xcite that the unitarity corrections might become important before next - to - leading ones , the study of the latter ones is crucial due to the large numerical value of the subleading series .after the evaluation of the nll contribution to the bfkl amplitude has been performed @xcite it turned out that the correction is very large and need to be resummed @xcite in order to stablize the final result . we have tested two important types of the nll corrections : the inclusion of running of the coupling @xmath65 and the so called kinematical constraint . in the case of linear bfklthe running of the coupling poses serious problems due to the existence of the landau pole and thus it is necessary to regularise @xmath65 at small scales .this results in a strong dependence of the solution on the cut - off ( or freezing ) parameter @xmath35 .the intercept of the bfkl pomeron turns out to be dominated by the values of @xmath66 and instead of the typical diffusion pattern one has a factorized behaviour of the solution @xmath67^{\nu}$ ] at large rapidities .the distribution of the gluon momenta is dominanted by the virtualities in the infrared regime @xmath68 . in the case of the balitsky - kovchegov equation with the running couplingone has still problem of regularisation around the landau pole , however the solution is itself much more stable with respect to the details of the phenomenological regularisation .it turns out , that since the saturation effects are very strong in the regime of small values of @xmath32 , they tend to decrease the rapid rise of the amplitude for the values of running coupling evaluated at scales around the cutoff scale @xmath35 .the rapidity - dependent saturation scale , @xmath47 shifts the momentum distribution out of the infrared regime into the perturbative domain .this phenomenon can be best visualized by means of similar contour plots as before , fig.[fig : conrun ] . in the linear bfkl case oneobserves that the solution initially undergoes the diffusion pattern , slightly asymmetric due to corrections coming from the running coupling , and then suddenly it drops into the small scales regime @xmath69 where it exhibits factorized ( in @xmath23 and @xmath70 ) behaviour .on the contrary , the balitsky -kovchegov equation which initially coincides with bfkl , also undergoes some form of transition but then the distribution moves away from the infrared regime due to the generation of the saturation scale .we have estimated the saturation scale in the case of balitsky - kovchegov equation with running coupling . by taking the ansatz that the local exponent @xmath71 takes the form @xmath72 where @xmath73 we have found that the saturation scale is given by the solution to the differential equation @xmath74 with the initial condition @xmath75 and @xmath76 chosen in the region where scaling sets in .the solution takes the form @xmath77 where @xmath78 .it follows that the local exponent @xmath79 decreases with increasing rapidity , and @xmath80 for very large @xmath23 .such dependence is indeed seen in the numerical analysis . in fig .[ fig : scalerun ] we illustrate scaling in the case with running coupling by showing the function @xmath81 plotted versus @xmath55 for different values of rapidity .@xmath47 is given by formula ( [ eq : satscalerc ] ) with the initial condition @xmath82 .the overlapping curves at low values of the scaling variable clearly indicate that for @xmath83 scaling is satisfied to a very good accuracy , thus justifying our ansatz ( [ eq : satscalerc ] ) for the saturation scale .we have also tested the impact of the so called kinematical constraint onto the solution of balitsky - kovchegov equation and found that though qualitative features are the same ( with generation of the saturation scale and scaling ) the absolute numerical value of the solution is strongly decreased and also the saturation scale is shifted towards smaller value of gluon momenta . to summarize , the diffusion into infrared in the case of the balitsky - kovchegov equation is strongly suppressed due to the emergence of the saturation scale which can be approximately identified with the maximum of the momentum distribution @xmath34 .the solution to the nonlinear equation has the property of the geometric scaling in the regime where @xmath83 whereas in the case when @xmath57 the solution enters the linear regime . in the case of runningcoupling the nonlinear equation turns out to be more stable as compared with the linear bfkl evolution .the sensitivity to the treatment of the infra - red region is much smaller than in the linear case due to the appearance of the saturation scale .in contrast to the bfkl equation with the running coupling , the large @xmath23 asymptotics of the dipole scattering amplitude arising from the bk equation is governed by gluon momenta in the perturbative domain .this research was supported in part by the eu framework tmr programme , contract fmrx - ct98 - 0194 and by the polish committee for scientific research grants nos .kbn 2p03b 120 19 , 2p03b 051 19 , 5p03b 144 20 .99 l. n. lipatov , _ sov .* 23 * ( 1976 ) 338 ; + e. a. kuraev , l. n. lipatov , v. s. fadin , _ sov .. jetp _ * 45 * ( 1977 ) 199 ; + i. i. balitsky , l. n. lipatov , _ sov ._ * 28 * ( 1978 ) 338 .l. mclerran and r. venugopalan , _ phys ._ * d49 * ( 1994 ) 2233 ; _ phys ._ * d49 * ( 1994 ) 3352 ; _ phys ._ * d50 * ( 1994 ) 2225 ; + r. venugopalan , _ acta phys .polon . _ * b30 * ( 199 ) 3731 ; + e. iancu , a. leonidov and l. mclerran , _ nucl.phys ._ * a692 * ( 2001 ) 583 ; + e. ferreiro , e. iancu , a. leonidov and l. mclerran , hep - ph/0109115 .v. s. fadin , m. i. kotsky and r. fiore , _ phys .lett . _ * b 359 * , 181 ( 1995 ) ; + v. s. fadin , m. i. kotsky and l. n. lipatov , budkerinp-96 - 92 , hep - ph/9704267 ; v. s. fadin , r. fiore , a. flachi and m. i. kotsky , _ phys . lett . _* b 422 * , 287 ( 1998 ) ; + v. s. fadin and l. n. lipatov , _ phys .lett . _ * b 429 * , 127 ( 1998 ) ; + g. camici and m. ciafaloni , _ phys . lett . _* b 386 * , 341 ( 1996 ) ; _ phys .* b 412 * , 396 ( 1997 ) [ erratum - ibid .b * 417 * , 390 ( 1997 ) ] ; _ phys .* b 430 * , 349 ( 1998 ) .m. ciafaloni , d. colferai , phys .b * 452 * ( 1999 ) 372 ; + m. ciafaloni , d. colferai and g. p. salam , phys .d * 60 * ( 1999 ) 114036 ; + g. p. salam , _ acta phys ._ * b 30 * ( 1999 ) 3679 ; + g. altarelli , r.d .ball and s. forte _ nucl .phys . _ * b575 * ( 2000 ) 313 ; _ nucl . phys . _ * b599 * ( 2001 ) 383 ; + r.s .thorne _ phys .rev . _ * d64 * ( 2001 ) 074005 .

<S> in this talk the results of the analytical and numerical analysis of the nonlinear balitsky - kovchegov equation are presented . </S><S> the characteristic bfkl diffusion into infrared regime is suppressed by the generation of the saturation scale @xmath0 . </S><S> we identify the scaling and linear regimes for the solution . </S><S> we also study the impact of subleading corrections onto the nonlinear evolution . * </S><S> saturation at low @xmath1 and nonlinear evolution * + a. m. stato + + and + </S>

astro-ph0002515

if future telescopes like the ska are developed with sensitivities up to 100 times greater than present sensitivities , it is quite likely that current regulations will not provide the necessary protection against interference .there is a range of experiments ( eg redshifted hydrogen or molecular lines ) which require use of arbitrary parts of the spectrum , but only at a few locations , and at particular times , suggesting that a very flexible approach may be beneficial .other experiments require very large bandwidths , in order to achieve enough sensitivity .as shown in figure [ alloc ] , presently only 1 - 2% of the spectrum in the metre and centimetre bands is reserved for passive uses , such as radio astronomy ( morimoto 1993 ) . in the millimetre band ,much larger pieces of the spectrum are available for passive use , but the existing allocations are not necessarily at the most useful frequencies .current regulations alone will be inadequate , we need technology as well as regulation . we can not ( and do not want to ) impede the telecommunications revolution , but we can try to minimise its impact on passive users of the radio spectrum and maximise the benefits of technological advances . further information on many of the topics discussed below is available on http://www.atnf.csiro.au / ska / intmit/.it is important to be clear of what we mean when we talk about interference .radio astronomers make passive use of many parts of the spectrum legally allocated to communication and other services . as a result , many of the unwanted signals are entirely legal and legitimate .we will adopt the working definition that interference is any unwanted signal entering the receiving system .interfering signals vary a great deal in their source and nature .this naturally leads to different mitigation approaches .local sources of interference include things internal to telescope instruments , networking for it systems , and general and special purpose digital processors in the observatory .interference compliance testing , shielding , separate power circuits , minimising nearby equipment are key steps that need to be taken to minimise this kind of interference .external interference may arise from fixed or moving sources .not all methods of mitigation apply to both : in fact methods that work well for fixed sources , may not work at all for moving sources , due to problems like side lobe rumble .interference may be naturally occurring or human generated .examples of naturally occurring interference include : the ground , sun , other bright radio sources , and lightening .human generated interference may come from broadcast services ( eg tv , radio ) , voice and data communications ( eg mobile telephones , two - way radio , wireless it networks ) , navigation systems ( eg gps , glonass ) , radar , remote sensing , military systems , electric fences , car ignitions , and domestic appliances ( eg microwave ovens ) ( goris 1998 ) .the vast majority of these operate legally with in their allocated bands , regulated by national authorities and the itu ( international telecommunications union ) .however there are sources of interference , such as the iridium mobile communications systems , whose signals leak into bands protected for passive use . in this case , these interfering signals are @xmath0 times stronger than the signal from the early universe . in the case of australia, there is a single communications authority for whole country and therefore for the whole continent . as a resultthere is a single database containing information on the frequency , strength , location , etc of every licensed transmitter ( sarkissian 2000 ) .a key point therefore , is that the modulation schemes and other characteristics of the vast majority of these signal are known .their effect on radio telescopes is not only predictable , but can be modelled and used to excise the unwanted signals .radio astronomy could deal with most terrestrial interfering signals , by moving to a remote location , where the density and strength of unwanted signals is greatly reduced . as shown in figure [ forte] , this is getting more and more difficult , but there are still some possibilities .however with the increasing number of space borne telecom and other communications systems in low ( and mid ) earth orbits , a new class of interference mitigation challenges are arising - radio astronomy can run , but it can not hide !the are several new aspects introduced into the interference mitigation problem by this and they include : rapid motion of the transmitter on satellites , more strong transmitters in dish side lobes and possibly in primary beam , and different spectrum management challenges , because no place on earth is free from interference from the sky .undesired interfering signals and astronomy signals can differ ( be orthogonal ) in a range of parameters , including : frequency , time , position , polarisation , distance , coding , positivity , and multi path .it is extremely rare that interfering and astronomy signals do not possess some level of orthogonality in this @xmath1 dimensional parameter space .we therefore need to develop sufficiently flexible signal processing systems to take advantage of the orthogonality and separate the signals .this is of course very similar to the kinds of problems faced by mobile communication services , which are being addressed with smart antennas and software radio technologies .examples in radio astronomy to date include the use of the time or frequency phase space or even better both , as in pulsar studies wherer the time / frequency dispersion relation can be used , and the requirement that signals are positive in very low frequency studies .antenna arrays could take advantage of the position and distance ( curvature of wavefront ) phase space .human generated interference is normally polarised , so unpolarised astronomical signals can be observed by measuring the unpolarised component @xmath2 ( r. fisher , private communication . there is no silver bullet for detecting weak astronomical signals in the presence of strong undesired naturally occurring or human generated signals .spectral bands allocated for passive use provide a vital window , which can not be achieved in any other way .it is important to characterise the rfi so that the number , strength , band width , duty cycle , spatial and frequency distributions , and modulation and coding schemes can all be used to advantage in modelling and mitigating rfi . doing this at low frequenciesgives greater sensitivity due to the effects of harmonic content and ease of propagation . in order to do this, the telescope and instruments must be calibrated to provide the best possible characterisation of interfering and astronomy signals .a / d converters must be fast enough to give sufficient bandwidth , with a sufficient number of bits so that both strong and weak signals are well sampled .there are a range of techniques that can make passive use of other bands possible and in general these need to be used in a progressive or hierarchical way .@xmath3*remove at source * is obviously best , but that is often not possible , + @xmath3*regulation * providing radio quiet frequencies or regions , + @xmath3*negotiation * with owners users can lead to win - win solutions , for example replacing nearby radio links with underground fibres , removes interference and improves voice and data connectivity for users , + @xmath3*avoid interference * by choosing appropriate locations with terrain screening or radio quiet zones , + @xmath3*move to another frequency * , + @xmath3*screening * to prevent signals entering the primary elements of receivers , + @xmath3*far side lobes * of primary and secondary elements must be both minimised and well characterised , + @xmath3*minimise coherent signals * through out the array and thereby allow the natural rejection of the array to deal with the incoherent signals , + @xmath3*front end filtering * , using for example high temperature super conducting filters with high q to reject strong signals in narrow bands , before they cause saturation effects .+ @xmath3*high dynamic range linear receivers * to allow appropriate detection of both astronomy ( signals below the noise ) and very strong interfering signals , + @xmath3*notch filters * ( analog , digital or photonic ) to excise bad spectral regions , + @xmath3*clip * samples from data streams to mitigate burst type interference , + @xmath3*decoding * to remove signals with complex modulation and multiplexing schemes .blanking of period or time dependent signals is a very successful but simple case of this more general approach , + @xmath3*cancellation * of undesired signals , before correlation using fixed and adaptive signal processing ( harris , 2000 ) , + @xmath3*post correlation cancellation * of undesired signals , taking advantage of phase closure techniques ( sault 2000 ) + @xmath3*parametric techniques * allow the possibility of taking advantage of known interference characteristics to excise it ( ellingson , 2000 ) , + @xmath3*adaptive beam forming * to steer one or more nulls onto interfering sources .this is equivalent to cancellation , but it provides a way of taking advantage of the spatial orthogonality of astronomy and interfering signals , + @xmath3*use of robust statistics * in data processing to minimise the effects of outliers . in most applications of signal processing , there is a strong trend towards the use of digital techniques , as well as photonic techniques .the fundamental reason for this is that digital and photonic devices have cost curves which are evolving much more rapidly than traditional analog systems .in addition to that , they open up new techniques and offer substantial reduction in computational effort in many cases .the inherent immunity of photonic approaches to radio interference also creates functional advantages ( minasian 2000 ) .astronomers are joining these trends for exactly the same reasons .the jury is still out on what the appropriate balance or mixture of these techniques will be .adaptive rejection algorithms can be either constrained or unconstrained . constrainedalgorithms generally incorporate either a model or a copy of the desired or interfering signal , which is used control the adaption .for example it may be constrained so that only those signals with a certain coding or chip sequence are removed ( ellingson , 2000 ) .most astronomy signals are expected to be pure noise , so one could envisage a constraint that rejects all non noise like signals . in the case of unconstrained adaption , some algorithms ( predictive adaptive algorithms ) simply assume that the interference is much stronger than the signal and just use previous data samples to predict the following data samples for cancellation ( harris , 2000 ) .if that assumption can not be made , another approach is to block the desired signal and let the unconstrained algorithm work on the remaining signals .this is often done using a blocking matrix , which can be thought of as an operator that applies a set of complex weights which block certain signals , while passing everything else .advantages of this approach are that it can deal with multiple interfering signals which are changing in time and space , without affecting the signal to noise of the desired signal .a key ingredient of constrained adaptive algorithms is a reference channel that maximises interference to noise ratio for the ensemble of interferers .one way of achieving this is using additional omni - directional antennas or arrays which at least matches the gain of the side lobe response of the main array .of all the approaches listed above , the nulling or cancellation systems ( may be adaptive or predictive ) are the most likely to permit the observation of weak astronomy signals that are coincident in frequency or space with undesired signals .there is an important space - time duality with cancelling algorithms .any algorithm that works in the time domain can also be applied in the space domain .these techniques have been used extensively in communications , sonar , radar , medicine and others ( widrow & stearns 1985 , haykin 1995 ) .radio astronomers have not kept pace with these developments and in this case need to infuse rather than diffuse technology in this area .a prototype time based cancellation system developed at nrao ( shown in figure [ fig-2 ] ) has demonstrated 70db of rejection on the lab bench and 30db of rejection on real signals when attached to the 140 foot at green bank ( barnbaum & bradley 1998 ) .adaptive nulling systems are being prototyped by nfra in the netherlands ( van ardenne , these proceedings ) .combined space - time approaches have been used to cancel interference in gps receivers ( trinkle , 2000 ) . however , in all cases , the application in the presence of real radio astronomy signals is yet to be demonstrated and their effects on the weak astronomy signals needs to be quantified .a good prospect for doing this in the near future is recording baseband data from existing telescopes , containing both interfering and astronomy signals and simulating the receiver system in software ( bell et al .1999 ) . a number of algorithms can then be implemented is software and assessed relative to each other .08fisher2.ps beam forming and adaptive nulling as are not necessarily being done sequentially , but rather in parallel .while there are some sequential schemes ( genetic algorithms for example ) , most approaches simultaneously solve for the coefficients that give the desired beams and nulls .one can think of this as an optimisation problem .for example , in the minimum variance beam former , the `` goal '' is to minimise output power , and the `` constraint '' is maintain constant gain in a certain direction .the goal forces the nulls onto all interfering signals not coming from the direction of the astronomical source , and the constraint protects the beam gain ( ellingson , 2000 ) . of course, it is only protected for one direction , so there is still shape distortion . in general for an n element array , you can form up to n-1 nulls .however if more control over the main beam shape is required , one may use other beam formers , which form a smaller number of nulls , and use more degrees of freedom to control the main beam shape . a physical interpretation of why you can form nulls without wreckingthe beam might go as follows : phased arrays with many elements ( not equally spaced ) , have lots of nulls , and they are all over the sky once you get a reasonable distance from the main beam .imagine changing the coefficients a little to get the closest one on to an interferer .very little variation in the coefficients is required .since the difference is so small , the main beam hardly notices .the other nulls will shift around , of course , because they are sensitive to small changes in the coefficients .close to the main beam , the nulls are further apart , so you need a bigger variation in the coefficients to nudge the closest null into place - hence the increased distortion in the main beam in this case .it may be necessary to record the weights applied to generate the nulls , so that the beam shape changes can be calibrated out later ( cram , 2000 ) .real time systems permit full recovery of temporal information in the signal required .real time systems have been well studied and numerous examples can be found in other fields such as : radar , sonar , communications , defence anti jam , speech processing , and medicine .for example , there are existing systems which are capable of nulling up to 7 simultaneous moving jammers . in radio astronomywe operate in a totally different regime in which the astronomical signals are weak and noise like .we only wish to measure the time averaged statistical properties of the signals .for example in aperture synthesis the time averaged coherence between two antennas .since we do nt have to recover the signal modulation , radio astronomy does not have to use the real time algorithms developed for communications and radar .in such post correlation systems ( sault 2000 ) the information is only contained in the statistical properties of the signal which may vary slowly in time , frequency , space or direction . both the signal of interest and the interference obey phase and amplitude closure relations .this results in an over determined set of equations which form a closed set which can be used to self calibrate the array for both the source and the interferer .we conjecture that : * _ `` post correlation processing of time averaged signals can achieve the same rfi rejection as in real - time algorithms and that self calibration ( phase closure ) techniques provide powerful additional constraints''_*. bell j. , et al .1999 `` software radio telescope : interference mitigation atlas and mitigation strategies '' , in perspectives in radio astronomy : scientific imperatives at cm and m wavelengths ( dwingeloo : nfra ) , edited by : m.p .van haarlem & j.m . van der hulst .

<S> we describe the nature of the interference challenges facing radio astronomy in the next decade . </S><S> these challenges will not be solved by regulation only , negotiation and mitigation will become vital . </S><S> there is no silver bullet for mitigating against interference . </S><S> a successful mitigation approach is most likely to be a hierarchical or progressive approach throughout the telescope and signal conditioning and processing systems . </S><S> we summarise some of the approaches , including adaptive systems . # </S><S> 1= 08#1=0.8 </S>

astro-ph0105293

the existence of an intense ultraviolet background ( uvb ) radiation , inferred from observations of qso absorption spectra ( gunn & peterson 1965 ; bajtlik , duncan & ostriker 1988 ) , is likely to have prominent impacts on galaxy formation . while its origin could be attributed to qsos or young galaxies appeared at high redshifts ( e.g. , couchman 1985 ; miralda - escude & ostriker 1990 ; sasaki & takahara 1994 ; fukugita & kawasaki 1994 ) , the uvb once produced photoionizes the intergalactic medium and alters the subsequent growth of cosmic structures .these secondary structures emerged in the uvb are further responsible for producing the radiation field , which can in turn affect the development of next generation objects .it is therefore crucial to determine quantitatively the consequences of radiative feedback on the formation of early generation ( hereafter population iii ) objects , in order to link the build - up of structures with the thermal history of the universe .it has long been suggested that the formation of subgalactic objects is suppressed via photoionization and heating caused by the uvb ( umemura & ikeuchi 1984 ; ikeuchi 1986 ; rees 1986 ; bond , szalay & silk 1988 ; efstathiou 1992 ; babul & rees 1992 ; murakami & ikeuchi 1993 ; chiba & nath 1994 ; zhang , anninos & norman 1995 ; thoul & weinberg 1996 , among others ) .most of previous studies , however , assumed for simplicity that the medium is optically thin against the external radiation .several attempts have been made recently to take into account radiative transfer of photons ( e.g. , kepner , babul & spergel 1997 ; tajiri & umemura 1998 ; abel , norman & madau 1999 ; barkana & loeb 1999 ; razoumov & scott 1999 ; gnedin 2000b ; susa & umemura 2000a ; nakamoto , umemura & susa 2001 ; ciardi et al. 2001 ) . in our previous papers ( kitayama & ikeuchi 2000 , hereafter paper i ; kitayama et al .2000 , hereafter paper ii ) , we have studied the evolution of spherical clouds exposed to the uvb , solving explicitly radiative transfer of ionizing photons and hydrodynamics .we have found that the collapse of objects with circular velocities below @xmath8 km s@xmath9 can be prohibited completely even if self - shielding of gas is taken into account . we have also noted that a spherical cloud above this threshold undergoes run - away collapse in the dark matter potential and does not settle into hydrostatic equilibrium . for plausible intensities and spectra of the uvb , the central region of such a cloudis shown to get self - shielded against photoionization before shrinking to the rotation barrier .a question still remains as to whether star formation can take place efficiently in the cloud collapsing under the uvb .a key element in this context is molecular hydrogen ( h@xmath0 ) . in order for stars to form, the gas needs to radiate energy efficiently and cool down to temperatures well below @xmath10k . in the gas of primordial composition without metals , cooling processes at this temperatureare almost solely dominated by rotational - vibrational line excitation of h@xmath0 ( e.g. , peebles & dicke 1968 ; hirasawa 1969 ; matsuda , sato & takeda 1969 ) .formation and destruction of h@xmath0 , however , are very sensitive to the presence of a radiation field . for instance , h@xmath0 molecules are easily dissociated by photons in the so - called lyman - werner ( lw , 11.213.6 ev ) bands ( stecher & williams 1967 ; haiman , rees & loeb 1997 ; ciardi et al 2000b ) or via collisions with ions . on the other hand , the gas - phase formation of h@xmath0 is promoted by an enhanced ionized fraction in the post - shock gas ( shapiro & kang 1987 ; kang & shapiro 1992 ) or in weakly photoionized media ( haiman , rees & loeb 1996b ) . in addition , self - shielding taking place during dynamical collapse can also aid h@xmath0 cooling ( susa & kitayama 2000 ) .it is thus by no means trivial in what circumstances the uvb has positive or negative feedback on star formation in population iii objects . in this paper , by including h@xmath0 chemistry consistently with radiative transfer of uv photons , we extend our previous analyses ( papers i and ii ) and study the fate of population iii objects exposed to the external uvb . assuming spherical symmetry , we solve the radiative transfer equation , non - equilibrium chemical reactions including h@xmath0 formation and destruction , and dynamics of baryonic gas and dark matter .particular attention is paid to the possibilities of both collapse and h@xmath0 cooling in the presence of uv radiation fields .we explore the evolution of objects with wide ranges of masses , including those with the virial temperature @xmath11k ( @xmath12 km s@xmath9 ) , which have been left out in our previous analyses neglecting h@xmath0 formation .a similar problem was investigated earlier by kepner et al .( 1997 ) and haiman , abel & rees ( 2000 ) , assuming that collapsed clouds are in approximate hydrostatic equilibrium .as mentioned above and shown explicitly in this paper , the gas in collapsed objects is far from static and its physical state changes highly dynamically .we also note that our analysis is complementary to that of susa & umemura ( 2000a , b ) , who explored the pregalactic pancake collapse in plane - parallel geometry and studied the fate of objects far above the jeans scale .we pay particular attention to the development of smaller objects near the jeans scale , which are likely to evolve spherically to a first - order approximation .a crucial distinction between the earlier works and ours is that the gas density rises rapidly along with the spherical collapse , in proportion to the cube of scale length , providing greater chances of self - shielding and h@xmath0 cooling .this paper is organized as follows .section 2 describes the numerical model used in the paper .we present the simulation results in section 3 and discuss their implications on galaxy formation in section 4 .finally , section 5 summarizes our conclusions . throughout the paper ,we assume the density parameter @xmath13 , the cosmological constant @xmath14 , the hubble constant @xmath15/(100 km s@xmath9 mpc@xmath16 , and the baryon density parameter @xmath17 .we simulate the evolution of a spherically symmetric cloud exposed to the external uvb .the cloud is a mixture of baryonic gas and dark matter with the mass ratio of @xmath18 .the basic equations are summarized in section 2.1 . of paper i. one modificationis that the acceleration term due to the cosmological constant @xmath19 is added to the right hand side of momentum equations .they are solved with the second - order - accurate lagrangian finite - difference scheme described in bowers & wilson ( 1991 ) and thoul & weinberg ( 1995 ) .shocks are treated with an artificial tensor viscosity ( tscharnuter & winkler 1979 ) .the number of mass shells is @xmath20 for baryonic gas and @xmath21 for dark matter .our code reproduces accurately the similarity solutions for the adiabatic accretion of collisional gas and the pressureless collapse around a point - mass perturbation ( bertschinger 1985 ) .we have also checked that our results are essentially unchanged when the shell numbers are doubled or halved . at each time - step of the simulation , we solve self - consistently the radiative transfer equation , chemical reactions , and energy equation in the following three steps . repeating these procedures , the radiation field and the internal state of the gas are determined iteratively until the abundance of each species and the internal energy in each mesh converge within an accuracy of 0.1% .first , the direction / frequency - dependent radiative transfer is worked out for ionizing photons with energies @xmath22 ev , with the method devised by tajiri & umemura ( 1998 ) , in which both absorption and emission by hydrogen are explicitly taken into account . at each radial point , angular integration of the transfer equation is done over at least 20 bins in @xmath23 , where @xmath24 is the angle between the light ray and the radial direction .this is achieved by handling 4001,000 impact parameters for light rays .self - shielding of h@xmath0 against the lw photons ( 11.213.6 ev ) is evaluated from the self - shielding function of draine & beltoldi ( 1996 ) , considering the direction dependence of the h@xmath0 column density as in kepner et al .the gas is assumed to be optically thin against the photons with @xmath25ev , which only contribute to destructions of h@xmath26 and h@xmath27 .secondly , non - equilibrium chemical reactions are solved in each mesh for the species e , h , h@xmath28 , h@xmath26 , h@xmath0 , h@xmath27 .hydrogen molecules form mainly via the creation of the intermediates h@xmath26 and h@xmath27 ( eqs [ 1.1][1.4 ] of shapiro & kang 1987 ) .we neglect helium for the following reasons : ( i ) it only changes the ionization degree of the gas up to @xmath29 ( osterbrock 1989 ; susa & umemura 2000a ) ; ( ii ) helium line cooling is unimportant at the typical temperatures ( @xmath30k ) of the objects presently studied ; and ( iii ) photoionization heating of helium does not alter our main results as discussed in section [ sec : signif ] .unless stated explicitly , reaction rates are taken from the recent compilation of galli & palla ( 1998 ) .the coefficients for photoionization of h and photodissociation of h@xmath0 are computed from the results of the radiative transfer mentioned above .photodetachment of h@xmath26 and photodissociation of h@xmath27 are also taken into account using the cross sections given in tegmark et al .( 1997 ) and stancil ( 1994 ) for the photons with @xmath31 ev and @xmath32 ev , respectively .these two processes , particularly the former , can suppress h@xmath0 formation by destroying its intermediates .finally , the energy equation is solved including uv heating and radiative cooling due to collisional ionization , collisional excitation , recombination , thermal bremsstrahlung , compton scattering with the cosmic microwave background ( cmb ) radiation , and rotational - vibrational excitation of h@xmath0 .the uv heating rate is obtained directly from the results of the radiative transfer .the atomic cooling rates are taken from fukugita & kawasaki ( 1994 ) and molecular cooling rates from galli & palla ( 1998 ) .for comparison with previous approaches , we also perform simpler calculations in the following two ways .one assumes at each time - step an optically thin medium and ionization equilibrium among e , h and h@xmath28 , neglecting h@xmath0 , h@xmath26 and h@xmath27 .the other solves radiative transfer of ionizing photons , but neglects h@xmath0 , h@xmath26 and h@xmath27 .the latter corresponds to the analysis of paper ii .hereafter we refer to the former as the ` optically thin calculation ' , the latter as the ` no h@xmath0 calculation ' , and the one described in the previous paragraphs as the ` full calculation ' . unless otherwise stated , the full calculation is adopted .the external uvb is assumed to be isotropic and to have either a power - law or black - body spectrum : @xmath33 where @xmath34 is the boltzmann constant .we adopt the spectral index @xmath35 and the effective temperature @xmath36k , to mimic the spectra of qsos and stars , respectively .the above spectrum is normalized to give the intensity @xmath37 , in units of @xmath38erg s@xmath9 @xmath39 str@xmath9 hz@xmath9 , at the lyman limit of hydrogen ( @xmath40ev ) . fig .1 illustrates the model spectra described above with @xmath41 .also shown for reference are those with @xmath42 and @xmath43k . in the present analyses ,the ratio between the fluxes below and above the lyman limit is a key parameter to control the radiative feedback on the gas evolution ( e.g. , ciardi , ferrara & abel 2000a ) . to help quantitative comparisons among different spectral shapes , we list in table 1 the photon - number - weighted average intensities of these spectra : @xmath44 where the average is taken over the energies relevant for destruction of both h@xmath0 and h@xmath45 ( 0.7413.6 ev ) , dissociation of only h@xmath0 ( 11.213.6 ev ) , and ionization of h ( 13.6@xmath3 ev ) .notice that for a given value of @xmath37 the spectrum with @xmath36k has considerably less ionizing ( @xmath46ev ) photons and more h@xmath0 and h@xmath45 destroying photons ( @xmath47ev ) than the one with @xmath35 .lccc + + spectrum & 0.7413.6 ev & 11.213.6 ev & 13.6@xmath3 ev + + @xmath35 & 6.0 & 1.1 & 0.15 + @xmath42 & @xmath48 & 1.7 & @xmath49 + @xmath36 k & @xmath50 & 3.8 & @xmath51 + @xmath43 k & 0.39 & 1.0 & @xmath52 + observations of the proximity effect in the ly@xmath53 forest suggest @xmath54 at redshift @xmath55 ( bajtlik et al .1988 ; bechtold 1994 ; giallongo et al . 1996; cooke , espey & carswell 1997 ) , but its value is still highly uncertain at other redshifts .recent cmb power spectrum measurement by maxima has been used to constrain the redshift of reionization to be @xmath56 at the 2@xmath7 confidence level ( schmalzing , sommer - larsen & gtz 2000 ) . in this paper, we fiducially fix the onset of the uvb at @xmath57 and consider at @xmath58 constant @xmath37 of an arbitrary value . to take into account uncertainties in the reionization history, we also examine some cases with @xmath59 and @xmath60 , and evolving @xmath37 of the form : @xmath61 & 5 \leq z \leq 20 \\ 1 & 3 \leq z \leq 5 \\ { \left(\frac{1+z}{4}\right)}^4 & 0 \leq z \leq 3 , \\\end{array } \right .\label{eq : uvevl}\ ] ] where the evolution at @xmath62 is roughly consistent with recent models of reionization of the universe ( ciardi et al 2000b ; umemura , nakamoto & susa 2000 ) .initial and boundary conditions are similar to those in sections 2.3 .and 2.4 . of paper i.in the following , we give their brief summary and state any modifications and additions .we start the simulations when the overdensity of a cloud is still in the linear regime .the initial overdensity profile for both baryon and dark matter is @xmath63 , where @xmath64 is the comoving wave number , and the central overdensity @xmath65 is fixed at 0.2 . only the growing mode is considered . the outer boundary is taken at the first minimum of @xmath66 , i.e. , @xmath67 , within which the volume averaged overdensity @xmath68 vanishes .the characteristic mass of a cloud @xmath69 is defined as the baryon mass enclosed within the first zero of @xmath66 , i.e. , @xmath70 , which is 0.363 of the total baryon mass ( haiman , thoul & loeb 1996a ) .we define the central collapse redshift @xmath71 and the cloud collapse redshift @xmath72 , respectively , as epochs at which the inner - most gas shell and the shell enclosing @xmath69 would collapse to the center in the absence of thermal pressure .these two redshifts are related by @xmath73 ( strictly speaking , this relation is correct in the einstein - de sitter universe , but it still gives a reasonable approximation in the present case at @xmath74 ) .we will often characterize a cloud by its circular velocity @xmath75 and virial temperature @xmath76 , defined in terms of @xmath72 and @xmath69 as @xmath77 @xmath78 where @xmath79 is the mean molecular weight in units of the proton mass @xmath80 , and @xmath81 is the mean overdensity of a virialized object predicted from the spherical infall model ( peebles 1980 ; kitayama & suto 1996 ) . using circular velocity or virial temperature to denote the size of an object , we can minimize the dependences of our simulation results on assumed cosmological parameters .the initial abundance of each species is taken to be its cosmological post - recombination value ( e.g. , galli & palla 1998 ) , i.e. , @xmath82 , @xmath83 , @xmath84 , and @xmath85 , where @xmath86 is the fraction of the species i with respect to the total number of hydrogen atoms .our results are insensitive to the precise values of initial abundances . in the optically thin and no h@xmath0 calculations ,@xmath87 is assumed before the gas is exposed to the uvb or experiences shocks . in this paper, we mainly focus on the evolution of the central gas of a cloud until it reaches the rotation radius specified by the dimensionless spin parameter at turn - around @xmath88 : @xmath89 where @xmath90 is the turn - around radius of the shell , and we adopt a median for the spin parameter , @xmath91 ( efstathiou & jones 1979 ; barns & efstathiou 1987 ; warren et al . 1992 ) .the above radius defines the scale below which the gas is supposed to attain rotational support and the approximation of spherical symmetry breaks down .the overdensity of the gas exceeds @xmath92 when it reaches this radius .the simulation is halted when the innermost gas shell has fallen below this radius and is regarded as ` collapsed ' . on the other hand, a dark matter shell falling to the center is rebounced at the radius @xmath93 of the innermost gas shell , to mimic shell crossings . at the outer boundary ,we assume that the gas is in pressure equilibrium , i.e. , the pressure outside the cloud @xmath94 is equal to that of the outermost gas shell .this is different from the free boundary condition ( @xmath95 ) adopted in papers i and ii . while the inner structure of a simulated cloud is hardly affected by this difference , we adopt the former in this paper so that the cloud envelope maintains to follow the hubble expansion throughout its evolution .the results of the full calculations are first compared with those of previous approaches , to illustrate the significance of incorporating radiative transfer and h@xmath0 formation in our analyses .the most noticeable changes are in the evolution of a low mass object with @xmath96k .[ fig : rtxl ] shows the time evolution of the central part of a cloud with @xmath97 ( @xmath98 ) and @xmath99 km s@xmath9 , corresponding to @xmath100k .both the dynamics and the thermal state are shown to change completely under different treatments of the radiative processes . in the optically thin calculation ,the cloud is photoionized to the center and evaporated promptly by a little radiation with @xmath101 and @xmath35 . in the no h@xmath0 calculation ,the external radiation is attenuated by absorption , but contraction of the gas is halted by thermal pressure as it lacks coolant at @xmath102k . in the full calculation , on the other hand , the cloud center is kept self - shielded against the uvb and h@xmath0 molecules in excess of @xmath103 are produced , allowing the gas to collapse and cool efficiently to @xmath104k .details of the h@xmath0 formation and destruction processes in the uvb will be discussed in section [ sec : dyn ] . for a high mass object well above the scale @xmath105k , in contrast ,radiative transfer and h@xmath0 formation appear to have smaller impacts as far as the dynamics is concerned . fig .[ fig : rtxh ] shows that such a cloud can collapse even in a stronger uv field with @xmath41 and @xmath35 , regardless of treatments of radiative processes .this is because gravitational potential of the cloud is much greater than thermal energy attained by photoionization ( @xmath106k ) and atomic cooling is efficient enough to enable the collapse .nevertheless , the final thermal state of the gas is still modified to a great extent in a high mass collapse , as indicated by the differences in the temperature and electron fraction at the collapse in fig .[ fig : rtxh ] . to see this point more clearly ,[ fig : prof1 ] compares the internal structure of the same objects at the collapse of the central gas .the density profile commonly exhibits that the central gas is undergoing run - away collapse ( panel a ) , as it cools efficiently and contracts nearly at the free - fall rate under the dark matter potential .the radiative transfer effects , however , produce a self - shielded neutral region at @xmath107kpc ( panel b ) , whereas @xmath108 is simply determined by ionization equilibrium with uniform radiation in the optically thin case .the gas well inside this neutral region can cool down to @xmath109k once h@xmath0 cooling is taken into account ( panel c ) . in the absence of h@xmath0 molecules , the gas can cool only to @xmath110k by atomic cooling .rapid increase of density due to run - away collapse is in fact essential for self - shielding and h@xmath0 cooling , as will be discussed throughly in the forthcoming sections . to further corroborate the above results , fig .[ fig : prof2 ] shows more detailed structures in the full calculation .while h@xmath0 molecules are destroyed in the ionized outer envelope , they form efficiently in the self - shielded central region ( panel a ) . roughly speaking, the gas becomes optically - thick against ionizing ( @xmath111ev ) photons at @xmath112 @xmath39 and the lw ( @xmath113ev ) photons at @xmath114 @xmath39 , where @xmath115 and @xmath116 are respectively the hi and h@xmath0 column densities measured from the outer boundary .both of these conditions are amply satisfied at @xmath117 kpc ( panel b ) . the uv heating rate is reduced significantly in this optically thick region ( panel c ) .photons emitted in the cloud interior contribute at most @xmath118 to the total heating rate for the @xmath35 spectrum , though their contribution can be much larger for a softer spectrum ( tajiri & umemura 2000 ) .cooling to temperatures below @xmath3k is possible once h@xmath0 cooling overtakes uv heating . at @xmath119k ( @xmath120 kpc ) ,the timescale of h@xmath0 formation becomes shorter than that of destruction as recombination lags behind the temperature decrease ( kang & shapiro 1992 ; corbelli , galli & palla 1997 ) , and chemical reactions are out of equilibrium ( panel d ) .the cooling timescale , however , becomes even shorter as temperature drops , and @xmath121 is frozen out at @xmath4 at the cloud center .this abundance is comparable to that achieved commonly in a metal - deficient post - shock layer in the absence of radiation ( shapiro & kang 1987 ; ferrara 1998 ; susa et al .1998 ) .if helium is further incorporated into the present calculations , the largest impact will be via heating by photons with @xmath122 ev . for given uv intensity @xmath37 , hi number density @xmath123 , and hi column density @xmath115, the heating rate will be enhanced by the addition of helium , in the case of @xmath35 , as @xmath124 where @xmath125 is the total heating rate in units of erg s@xmath9 @xmath126 of both h and he , and @xmath127 is that of the pure hydrogen gas .these rates are computed as in the appendix of paper i , which are in good agreement with more rigorous treatment .equation ( [ eq : he ] ) indicates that the total heating rate tends to increase by @xmath128 once helium with the primordial abundance ( @xmath129 by number ) is taken into account .this is much smaller than the overall variations in the heating and cooling rates in a cloud shown in fig .[ fig : prof2](c ) .in fact , we have checked that an additional heat input of @xmath130 does not considerably alter either the time evolutions in figs [ fig : rtxl ] and [ fig : rtxh ] or the internal structures in figs [ fig : prof1 ] and [ fig : prof2 ] .though temperature of the diffuse outer envelope can increase by as much as a factor of 2 by helium heating ( abel & haehnelt 1999 ) , the evolution of the central high - density region , in which our main interests reside , is little influenced .helium heating will be of less importance for softer spectra .hence we neglect helium in our analyses . in summary ,the above results justify our present approach and confirm the importance of including radiative transfer of uv photons , h@xmath0 formation / destruction , and hydrodynamics , in studying the development of population iii objects in the uvb .the evolution of uv irradiated clouds are studied in more detail in the following .[ fig : rtxj ] displays how the time evolution of a low mass object ( @xmath99 km s@xmath9 ) depends on the radiation intensity . in the absence of the uvb ( @xmath131 ) , the cloud decoupled from the hubble expansion is first shock - heated and then cools radiatively . at @xmath132 , in contrast, the gas is photoionized and evaporated even if radiative transfer is taken into account .once the gas is photoionized and heated , h@xmath0 molecules are destroyed by a large amount . at lower intensity ( @xmath101 ) , as was also shown in fig .[ fig : rtxl ] , the cloud can collapse and cool down by h@xmath0 . at even weaker intensity ( @xmath133 ) , the radiation is attenuated almost completely and the evolution resembles that in the no uv case .note that the h@xmath0 fraction increases after the onset of the uvb at @xmath101 , compared to the no uv case .this is indeed the case in which the formation of h@xmath0 via h@xmath26 and h@xmath27 is promoted by an enhanced ionized fraction in a weakly photoionized medium ( haiman et al .1996b ) .[ fig : rtxaj ] illustrates that the shape of the radiation spectrum also influences greatly the thermal and dynamical evolution .compared to the @xmath35 case , the black - body spectrum with @xmath134k is softer and has much less ionizing photons with energies above @xmath135ev ( see fig . 1 and table 1 ) . for the same values of @xmath37 as in fig .[ fig : rtxj ] , the black - body uvb therefore has weaker impacts on the evolution of radius , temperature , and the ionization degree . on the other hand , h@xmath0molecules are destroyed by a larger amount at the onset of the uvb , because a greater number of photons with energies below @xmath135ev are available to destroy h@xmath0 and its intermediates , h@xmath26 and h@xmath27 .nevertheless , as the gas contracts , it becomes optically thick against the lw photons and the final h@xmath0 fraction of @xmath136 is still achieved . for a high mass object well above the scale @xmath105k , the collapse is hardly suppressed even under the strong uvb , but its thermal evolution remains sensitive to the uvb . as shown in fig .[ fig : rtxhj ] , for @xmath137 and @xmath35 , the gas is promptly heated to @xmath110k and h@xmath0 molecules are destroyed mainly via collisions with h@xmath28 . at @xmath132 , on the other hand , the h@xmath0 abundance is enhanced compared to the no uv case by an increased degree of ionization , as already mentioned . in either case, the h@xmath0 fraction at the collapse again exceeds @xmath138 . to illustrate more clearly how self - shielding and radiative cooling operate in a collapsing cloud , we plot in figs [ fig : core1 ] and [ fig : core2 ] the temperature and fractions of hi and h@xmath0 at the cloud center as a function of its density . as the gas contracts and density rises , its state changes rather suddenly from the hot ( @xmath139k ) ionized` hii phase ' , to the warm ( @xmath140k ) neutral ` hi phase ' , and the cold ( @xmath102k ) neutral ` h@xmath0 phase ' .such transitions are analogous to those pointed out by kepner et al .( 1997 ) , except that they are induced here by the dynamical contraction of the gas itself rather than the decline of the external uv intensity .the transition from the hii phase to the hi one takes place when the cloud becomes optically thick against ionizing photons .the gas finally enters the h@xmath0 phase once the cooling timescale becomes shorter than the other relevant timescales , such as those of free - fall and h@xmath0 formation / destruction .these figures confirm that @xmath121 tends to converge to @xmath4 after the h@xmath0 phase is achieved .figs [ fig : core1 ] and [ fig : core2 ] further indicate that quantitative details of the transitions among the above phases depend sensitively on the intensity and spectrum of the uvb . to examine this more directly , we plot in fig .[ fig : nj ] the critical densities for hi shielding @xmath141 and h@xmath0 cooling @xmath142 , defined as those at which the central hi fraction exceeds 0.5 and temperature falls below 5000k , respectively . inwhat follows , we regard the gas as in the hii , hi , and h@xmath0 phases , if the density is in the range @xmath143 , @xmath144 , and @xmath145 , respectively .though fig .[ fig : nj ] is plotted for particular choices of @xmath75 and @xmath72 , basic features are similar in other cases .a possible change is that for smaller masses and large @xmath37 the collapse is completely prohibited ( see fig .[ fig : jv ] ) and the critical densities simply disappear . fig .[ fig : nj ] shows that @xmath141 increases with @xmath37 for a given spectral shape , i.e. , the higher density is required to shield the stronger ionizing flux . at small @xmath37, @xmath141 falls below the turn - around density , indicating that photoionization is kept attenuated throughout the cloud evolution .this feature is also apparent in figs [ fig : core1 ] and [ fig : core2 ] , in which @xmath108 is kept high at weak uv intensities .compared to the @xmath35 case , the @xmath36k spectrum yields lower values of @xmath141 , because there are less ionizing ( @xmath146 ev ) photons for a given @xmath37 . on the other hand, @xmath147 behaves in a more complex manner , depending on the spectral shape . for @xmath35 ,it is not a monotonic function of @xmath37 , but has a minimum at moderate intensities @xmath148 .formation of h@xmath0 molecules is enhanced maximally at these intensities . at weaker intensities @xmath149 the resultalmost coincides with that of the no uv case , whereas at stronger intensities @xmath150 h@xmath0 formation is suppressed to higher densities . for @xmath36k , in contrast , a greater fraction of photons below 13.6 ev is present to suppress h@xmath0 formation .the threshold @xmath142 is therefore higher than that of @xmath35 for a given @xmath37 .this is the reason why the convergence of @xmath121 is rather slow in fig .[ fig : core2 ] . in either case ,[ fig : nj ] assures that the h@xmath0 phase is realized at the center of a contracting cloud before it is rotationally supported , as long as @xmath37 is not far above the level @xmath151 observed at @xmath152 . the above critical densities @xmath141 and @xmath147 at @xmath153 with @xmath35 are in rough agreement with analytical estimations by tajiri & umemura ( 2000 ) for an isothermal sphere ; @xmath154 .this confirms , in the case of @xmath35 , that @xmath155 is essentially the density at which the number of recombinations in a cloud exceeds that of ionizing photons .similarly , @xmath147 at @xmath153 corresponds to the density at which the h@xmath0 cooling rate overcomes the uv heating rate attenuated at the cloud center . for a much softer spectrum ( @xmath36k ) ,the deviations from such simplified estimations become larger , because of the greater significance of emitted photons and the lw photons .we have performed the simulations for a variety of parameters , such as the intensity and spectrum of the uvb , and the circular velocity and collapse redshift of a cloud .the results are summarized in the following .[ fig : jv ] shows a @xmath156 diagram on collapse and cooling of a cloud with @xmath157 ( @xmath98 ) .the final state of the central gas is classified into three categories : ( i ) it collapses and cools to @xmath158k , i.e. the h@xmath0 phase is realized ; ( ii ) it collapses but does not cool to @xmath159k ; and ( iii ) it is prohibited to collapse within the present age of the universe .also plotted for reference is the relation @xmath160 , where @xmath161 is the maximum equilibrium temperature defined semi - analytically in section 3.3 of paper ii .this relation roughly corresponds to the jeans condition at the maximum expansion of the central gas and provides a reasonable estimation for the collapse threshold in the optically thin limit .the predictions of the full calculations are distinct from the optically thin results in two ways .one is that the collapse of low mass objects with @xmath96k ( @xmath12 km s@xmath9 ) is possible at small uv intensities .the other is that the central gas in collapsed clouds can cool efficiently to temperatures well below @xmath3k , as long as @xmath37 is not far above the observed level @xmath151 .the suppression of h@xmath0 cooling occurs at @xmath162 for @xmath35 and @xmath163 for @xmath134k , mainly due to photodetachment of h@xmath26 as well as photodissociation of h@xmath0 . to quantify this effect ,we indicate in fig .[ fig : jv ] the average intensity of the external uvb defined by equation ( [ eq : jb ] ) at 0.7413.6 ev .the suppression of h@xmath0 cooling is shown to take place at @xmath164 in massive clouds for both spectra . given circular velocity and collapse redshift of a cloud , one can compute the critical radiation intensity for collapse and h@xmath0 cooling .[ fig : jcrit ] exhibits , as a function of the central collapse redshift @xmath165 , the intensity above which h@xmath0 cooling is prevented completely in a cloud , i.e. , the final state of the central gas belongs to the category ( ii ) or ( iii ) defined above .the critical intensity is in general an increasing function of redshift , i.e. , gas density .it also increases with circular velocity but converges at @xmath166 km s@xmath9 and @xmath167 km s@xmath9 for @xmath35 and @xmath168k , respectively. the convergence takes place once photoheating no longer affects the cloud evolution and only photodestruction of h@xmath0 and h@xmath26 becomes important .the converged intensities therefore have nearly identical average values below the lyman limit @xmath169 for both spectra. once converged , the critical intensities tend to drop slightly for larger @xmath75 ( panel b ) because the electron fraction of the gas is lower in a cloud with higher hydrogen column densities . to explore the possibilities of different epochs of reionization, we have added to fig .[ fig : jcrit ] the results of runs with @xmath170 and 50 .if @xmath165 is sufficiently close to @xmath171 , the central gas is already contracting when it is exposed to the uvb , and hence the stronger intensity is required to prevent collapse and cooling .the difference is greater for the harder spectrum and the smaller cloud mass , for which suppression of cooling is mainly via photoevaporation and depends highly on the initial dynamical state of the gas . for the softer spectrum and the larger cloud mass ,on the other hand , the suppression is regulated to a greater extent by photodestruction of h@xmath0 and h@xmath26 , and is less sensitive to the initial dynamical state .after all , fig .[ fig : jcrit ] indicates that there is a greater chance for the gas to cool even in a cloud well below the scale @xmath172 km s@xmath9 at high redshifts , provided that the intensity is comparable to or less than @xmath173 .alternatively , given a specific model on the evolution of the uvb , one can deduce the critical circular velocity or mass for collapse and cooling .[ fig : vz ] shows such quantities under which h@xmath0 cooling is completely suppressed in an object with the cloud collapse epoch @xmath72 .the uvb greatly prohibits cooling in clouds with @xmath174 km s@xmath9 or @xmath175 at low redshifts .the threshold values , however , depend highly on the time history and spectra of the uvb . for an evolutionary model given by equation ( [ eq : uvevl ] ) , the critical mass drops by several orders of magnitude at @xmath176 ( @xmath177 ) , enabling cooling in objects as small as @xmath178 .[ fig : vz](b ) also exhibits the mass and collapse redshift of the 1,2,3@xmath7 density perturbations in a cold dark matter ( cdm ) model with @xmath179 .this is a specific example of the most successful cosmological model which is consistent with a number of current observations ( e.g. , wang et al . 2000 ) .[ fig : vz ] shows that the uvb mainly suppresses collapse and cooling of @xmath180 objects with @xmath175 in the @xmath181cdm model .it is rather difficult to suppress cooling in higher-@xmath7 objects , unless the radiation intensity becomes much higher than the level observed at @xmath152 .if different values of cosmological parameters are adopted , the collapse criterion in terms of circular velocity remains practically unchanged as long as @xmath182 .thus , the critical mass shown in fig .[ fig : vz](b ) scales with cosmological parameters merely according to equation ( [ eq : vc ] ) .our calculations predict that the formation of objects with @xmath183 km s@xmath9 is greatly suppressed by the uvb at low redshifts , whereas much smaller objects ( @xmath184k , @xmath185 km s@xmath9 ) can collapse and cool at high redshifts ( @xmath176 , @xmath177 ) .this is mainly because of three reasons : ( i ) intensity of the uvb tends to decline at high redshifts prior to complete reionization of the universe ; ( ii ) self - shielding of the gas is stronger at higher redshifts owing to higher gas density ; and ( iii ) the gas density in jeans unstable clouds rises rapidly along with run - away collapse , further promoting self - shielding and h@xmath0 formation .what are the fates of low mass objects that are able to cool and collapse in the presence of the uvb ?if the cooled gas can promptly turn into stars , they are likely to constitute ` building blocks ' of present - day luminous structures .while such objects may be detected directly by future instruments such as _ ngst _ , they may also have links to observed subgalactic objects ( e.g. , pascarelle et al .1996 ) or dwarf galaxies .in contrast , those prevented to collapse by the uvb will become gravitationally unbound and spread into intergalactic medium .they might contribute to ly@xmath53 absorption lines in the qso spectra ( e.g. , umemura & ikeuchi 1984 ) .these speculations on the bifurcation of low mass objects are illustrated schematically in fig .[ fig : mbif ] .the critical baryonic mass in the case of evolving @xmath37 ( eq . [ [ eq : uvevl ] ] ) with @xmath35 is used to distinguish the uncollapsed ` expanding clouds ' from the others .furthermore , we expect that the collapsed objects may be identified , mainly based on their distinct star formation histories , as at least two separate populations : the lower - mass ( @xmath186 ) , higher-@xmath7 ( @xmath187 ) systems and the higher - mass , lower-@xmath7 ones .the former population can be a plausible candidate of dwarf spheroidal / elliptical galaxies for the following reasons .first , star formation in these objects is expected to be active at redshifts higher than @xmath188 , consistent with the fact that the majority of local group dwarfs show a sharp decline in the star formation rate at some @xmath189 gyr ago ( mateo 1998 ; gnedin 2000a ) .second , their gravitational potential is compatible with small velocity dispersions ( @xmath190 km s@xmath9 ) of local group dwarfs .third , they are apt to be dark matter dominated and metal - poor since the gas infall and star formation are suppressed particularly at the outer envelope .fourth , our simulations indicate that the central gas in these objects can cool efficiently by h@xmath0 in the course of spherical contraction before reaching the rotation barrier .galaxy formation is thus likely to proceed rapidly in a dissipationless manner and lead to the formation of spheroids rather than disks . on the other hand, the latter population will exhibit signs of more recent star formation and contribute significantly to the global star formation rate of the universe .the clustering of these low @xmath7 objects is expected to be rather weak .they might thus have been identified as a part of faint blue objects ( e.g. , tyson 1988 ; cowie et al .1988 ; ellis 1997 ) or star - forming dwarfs ( e.g. , thuan & martin 1981 ; lee et al . 2000 ) .this is in line with the idea proposed by babul & rees ( 1992 ) and extended by babul & ferguson ( 1996 ) and kepner et al .one distinction is that the impacts of the decline of the uv intensity at low redshifts turn out to be modest .this is because the collapse is delayed by the kinetic energy of the gas attained during the peak of the uv intensity at @xmath191 ( although the threshold circular velocity does decrease with time , the threshold mass continues to rise due to a factor @xmath192^{-3/2}$ ] ) .the present spherical collapse simulations are mainly applicable to low mass objects near the jeans scale .much larger objects , in contrast , are likely to first undergo sheet - like pancake collapse and evolve in a different manner from the spherical case . they would not end up with the run - away collapse as gravity of the sheetis readily overwhelmed by thermal pressure .in fact , susa & umemura ( 2000a , b ) argued that the pancakes collapsing in the uvb can bifurcate , depending upon the degree of self - shielding , into observed early and late type galaxies .both their results and ours suggest that the uvb could play an essential role in regulating the star formation history of the universe .recent three - dimensional simulations show that a number of subclumps tend to arise from the fragmentation process in a parent cloud ( e.g. , bromm , coppi & larson 1999 ; abel , bryan & norman 2000 ) .the criteria obtained in the current calculations are applicable to each of such locally collapsing subclumps . as a net effect to the parent cloud ,the gas clumpiness is likely to enhance self - shielding against the external uvb , since the number of photons necessary to ionize the gas increases roughly by a factor @xmath193 ( gnedin & ostriker 1997 ; ciardi et al .2000b ) .major uncertainties in our predictions are regarding the feedback from the sources formed in an collapsing object .while our simulations can probe directly the cloud evolution until its central collapse , the evolution thereafter will be influenced by several additional processes .for instance , omukai & nishi ( 1999 ) argued that even a single ob star can photodissociate surrounding hydrogen molecules within a static cloud with @xmath184k . on the other hand ,ferrara ( 1998 ) pointed out that blow - away processes during supernova explosions can enhance the fraction of hydrogen molecules in surrounding media .such explosions may also disrupt low mass objects and induce metal injections into intergalactic space ( e.g. , dekel & silk 1986 ; nishi & susa 1999 ; ferrara & tolstoy 2000 ) .it still remains uncertain to what extent these processes affect successive gas infall and cooling in a dynamically contracting object whose central overdensity exceeds @xmath92 .as this issue is beyond the scope of the present paper , we will investigate it separately in our future publications .we have studied the formation of population iii objects exposed to the external uv radiation .radiative transfer of photons , h@xmath0 formation and destruction processes , and hydrodynamics are incorporated explicitly in our analyses and all shown to play essential roles in the development of population iii objects .although the uvb does suppress the formation of low mass objects , the negative feedback turns out to be weaker than previously suggested in the following two respects .first , the cut - off scale of the collapse drops significantly below @xmath194k ( @xmath195 km s@xmath9 ) at weak uv intensities ( @xmath2 ) , due to both self - shielding of the gas and h@xmath0 cooling .their importance becomes greater at higher redshifts . at stronger intensities, objects as large as @xmath196 km s@xmath9 can be photoevaporated and prohibited to collapse , in agreement with previous investigations based on the optically thin approximation ( e.g. , thoul & weinberg 1996 ) .high mass objects well above this scale can collapse almost irrespectively of the uvb by atomic cooling .secondly , a spherical cloud that has become jeans unstable in the uvb tends to undergo run - away collapse and does not settle into hydrostatic equilibrium . as a result ,even if the gas is ionized and/or h@xmath0 molecules are destroyed once by radiation at low densities ( e.g. , haiman et al . 2000 ) , the radiation is attenuated and h@xmath0 can re - form during the dynamical collapse .the critical densities at which self - shielding and efficient h@xmath0 cooling take place depend sensitively on the spectrum and intensity of the external uvb . for a qso - like ( @xmath35 )spectrum , h@xmath0 formation is even enhanced by a little radiation with @xmath148 ( haiman et al .1996b ) . for a stellar - like ( @xmath36k ) spectrum ,h@xmath0 formation is hindered to a greater extent by the photons with energies below 13.6 ev ( haiman et al .1997 ) . in any case, the collapsing gas is shown to cool efficiently to temperatures well below @xmath3 k before rotationally supported , as long as the radiation intensity is comparable to or less than @xmath197 observed at @xmath152 .the h@xmath0 fraction in the cooled gas reaches the level @xmath4 , comparable to that achieved commonly in the metal - deficient post - shock layer ( shapiro & kang 1987 ; ferrara 1998 ; susa et al .1998 ) .our results imply that star formation can take place in low mass objects collapsing in the uvb .the threshold baryon mass for the complete suppression of collapse and star formation is @xmath198 at @xmath199 but drops significantly at higher redshifts . in a conventional @xmath181cdm universe ,it coincides roughly with that of the 1@xmath7 density fluctuations at @xmath176 .objects near and above this threshold can thus constitute building blocks of present - day luminous structures , and may also have links to dwarf spheroidal / elliptical galaxies and faint blue objects .these results indicate that the uvb can play a key role in regulating the star formation history of the universe .we thank kuniaki masai , taishi nakamoto and yukiko tajiri for discussions , and the referee , andrea ferrara , for helpful comments .tk and hs acknowledge support from research fellowships of the japan society for the promotion of science for young scientists and research grant of university of tsukuba research project , respectively . this work was partially carried out with facilities at the center for computational physics at university of tsukuba .999 abel t. , haehnelt m. g. , 1999 , apj , 520 , l13 abel t. , norman m. l. , madau p. , 1999, apj , 523 , 66 abel t. , bryan g. l. , norman m. l. , 2000 , apj , 540 , 39 babul a. , rees m. j. , 1992 , mnras , 255 , 346 babul a. , ferguson h. c. , 1996 , apj , 458 , 100 bajtlik s. , duncan r. c. , ostriker j. p. , 1988 ,apj , 327 , 570 barkana r. , loeb a. , 1999 , 523 , 54 barns j. , efstathiou g. , 1987 , apj , 319 , 575 bechtold j. , 1994 , apjs , 91 , 1 bertschinger e. , 1985 , apjs , 58 , 39 bond j. r. , szalay a. s. , silk j. , 1988 , apj , 324 , 627 bowers r. l. , wilson j. r. , 1991 , numerical modeling in applied physics and astrophysics , boston , jones & bartlett bromm v. , coppi p. a. , larson r. b. , 1999 , apj , 527 , l5 chiba m. , nath b. b. , 1994 , apj , 436 , 618 ciardi b. , ferrara a. , abel t. , 2000a , apj , 533 , 594 ciardi b. , ferrara a. , governato f. , jenkins a. , 2000b , mnras , 314 , 611 ciardi b. , ferrara a. , marri s. , raimondo g. , 2001 , mnras , in press cooke a. j. , espey b. , carswell r. f. 1997 , mnras , 284 , 552 corbelli e. , galli d. , palla f. , 1997 , apj , 487 , l53 couchman h. m. p. 1985 , mnras , 214 , 137 cowie l. l. , lilly s. j. , gardner j. p. , mclean i. s. , 1988 , apj , 332 , l2 dekel a. , silk j. , 1986 , apj , 303 , 39 draine b. t. , bertoldi f. , 1996 , apj , 468 , 269 efstathiou g. , 1992 , mnras , 256 , 43 efstathiou g. , jones b. j. t. , 1979 , mnras , 186 , 133 ellis , r. , 1997 , ara&a , 35 , 389 ferrara a. , 1998 , apj , 499 , l17 ferrara a. , tolstoy e. , 2000 , mnras , 313 , 291 fukugita m. , kawasaki m. 1994 , mnras , 269 , 563 galli d. , palla f. , 1998 , a&a , 335 , 403 giallongo e. , cristiani s. , dodorico s. , fontana a. , savaglio s. , 1996 , apj , 466 , 46 gnedin n. y. , 2000a , apj , 535 , l75 gnedin n. y. , 2000b , apj , 535 , 530 gnedin n. y. , ostriker j. p. , 1997, apj , 486 , 581 gunn j. e. , peterson b. a. 1965 , apj , 142 , 1633 haiman z. , thoul a. a. , loeb a. , 1996a , apj , 464 , 523 haiman z. , rees m. j. , loeb a. , 1996b , apj , 467 , 522 haiman z. , rees m. j. , loeb a. , 1997 , apj , 476 , 458 haiman z. , abel t. , rees m. j. , 2000 , apj , 534 , 11 hirasawa t. , 1969 , prog . theor .phys . , 42 , 523 ikeuchi s. , 1986 , apss , 118 , 509 kang h. , shapiro p. , 1992 ,apj , 386 , 432 kepner j. , babul a. , spergel n. , 1997 , apj , 487 , 61 kitayama t. , ikeuchi s. , 2000 , apj , 529 , 615 ( paper i ) kitayama t. , suto y. , 1996 , apj , 469 , 480 kitayama t. , tajiri y. , umemura m. , susa h. , ikeuchi s. , 2000 , mnras , 315 , l1 ( paper ii ) lee j. c. , salzer j. j. , law d. a. , rosenberg j. l. , 2000 , apj , 536 , 606 mateo , m. , 1998 , ara&a , 36 , 435 matsuda t. , sato h. , takeda h. 1969 , prog . theor .phys . , 42 , 219 miralda - escud j. , ostriker j. p. 1990, apj , 350 , 1 murakami i. , ikeuchi s. , 1993 , apj , 409 , 42 nakamoto t. , umemura m. , susa h. , 2001 , mnras , 321 , 593 nishi r. , susa h. , 1999 , apj , 523 , l103 omukai k. , nishi r. , 1999 , apj , 518 , 64 osterbrock d. e. , 1989 , astrophysics of gaseous nebulae and active galactic nuclei , mill valley , university science books peebles , p. j. e. , 1980 , the large - scale structure of the universe , princeton univ .press , princeton peebles p. j. e. , dicke r. h. , 1968 , apj , 154 , 891 pascarelle s. m. , windhorst s. p. , keel w. c. , odewahn s. c. , 1996 , nature , 383 , 45 razoumov a , o. , scott d. , 1999 , mnras , 309 , 287 rees m. j. , 1986 , mnras , 218 , 25p sasaki s. , takahara f. , prog .phys . , 1994 , 91 , 699 schmalzing j. , sommer - larsen j. , gtz m. , 2000 , apj , subitted ( astro - ph/0010063 ) shapiro p. r. , kang h. , 1987 , apj , 318 , 32 stancil , p. c. , 1994 , apj , 430 , 360 stecher t. p. , williams d. a. , 1967 , apj , 149 , l29 susa h. , kitayama t. , 2000 , mnras , 317 , 175 susa h. , umemura m. , 2000a , apj , 537 , 578 susa h. , umemura m. , 2000b , mnras , 316 , l17 susa h. , uehara h. , nishi r. , yamada , m. , 1998 , prog ., 100 , 63 tajiri y. , umemura m. , 1998 , apj , 502 , 59 tajiri y. , umemura m. , 2000 , apj , submitted tegmark m. , silk j. , rees m. j. , blanchard a. , abel t. , palla f. , 1997 , apj , 474 , 1 thoul a. a. , weinberg d. h. , 1995 , apj , 442 , 480 thoul a. a. , weinberg d. h. , 1996 , apj , 465 , 608 thuan t. x. , martin g. e. , 1981 , apj , 247 , 823 tscharnuter , w. m. , winkler , k. -h . , 1979 , comp .18 , 171 tyson j. a. , 1988 , aj , 96 , 1 umemura m. , ikeuchi s. , 1984 , prog . theor .phys . , 72 , 47 umemura m. , nakamoto t. , susa h. , 2000 , in kawasaki m. , sato k. , eds , the birth and evolution of the universe , universal academy press , tokyo , in press wang l. , caldwell r. r. , ostriker j. p. , steinhardt p. j. , 2000 , apj , 530 , 17 warren m. s. , quinn p. j. , salmon j. k. , zurek w. h. , 1992 , apj , 399 , 405 zhang y. , anninos p. , norman m. l. , 1995 , apj , 453 , l57

<S> we explore possibilities of collapse and star formation in population iii objects exposed to the external ultraviolet background ( uvb ) radiation . assuming spherical symmetry , we solve self - consistently radiative transfer of photons , non - equilibrium h@xmath0 chemistry , and gas hydrodynamics . </S><S> although the uvb does suppress the formation of low mass objects , the negative feedback turns out to be weaker than previously suggested . </S><S> in particular , the cut - off scale of collapse drops significantly below the virial temperature @xmath1k at weak uv intensities ( @xmath2 ) , due to both self - shielding of the gas and h@xmath0 cooling . </S><S> clouds above this cut - off tend to contract highly dynamically , further promoting self - shielding and h@xmath0 formation . for plausible radiation intensities and spectra </S><S> , the collapsing gas can cool efficiently to temperatures well below @xmath3k before rotationally supported and the final h@xmath0 fraction reaches @xmath4 . </S><S> our results imply that star formation can take place in low mass objects collapsing in the uvb . </S><S> the threshold baryon mass for star formation is @xmath5 for clouds collapsing at redshifts @xmath6 , but drops significantly at higher redshifts . in a conventional cold dark matter universe , </S><S> the latter coincides roughly with that of the 1@xmath7 density fluctuations . </S><S> objects near and above this threshold can thus constitute ` building blocks ' of luminous structures , and we discuss their links to dwarf spheroidal / elliptical galaxies and faint blue objects . </S><S> these results suggest that the uvb can play a key role in regulating the star formation history of the universe . </S><S> -5 mm = = = = = = = = # 1 # 1 # 1 # 1 @mathgroup@group @mathgroup@normal@groupeurmn @mathgroup@bold@groupeurbn @mathgroup@group @mathgroup@normal@groupmsamn @mathgroup@bold@groupmsamn = `` 019 = ' ' 016 = `` 040 = ' ' 336 = " 33e = = = = = = = = # 1 # 1 # 1 # 1 = = = = = = = = [ firstpage ] cosmology : theory diffuse radiation galaxies : formation radiative transfer molecular processes </S>

1703.03625

this note is concerned with the following differential equation driven by a @xmath3-dimensional fractional brownian motion ( fbm in the sequel ) @xmath4 with hurst parameter @xmath5 : @xmath6 , \\y_{0}&=&y \in \rr^{d}. \nonumber\end{aligned}\ ] ] assuming that the collection of vector fields @xmath7 belongs to @xmath8 and @xmath9 sits in @xmath10 , the theory of rough paths gives a framework allowing to get existence and uniqueness results for equation .in addition , the unique solution @xmath11 in the rough paths sense has @xmath12-hlder continuity for all @xmath13 .the reader is referred to @xcite for further details . in this paper , we are interested in the numerical approximation of equation based on a discretization of the time parameter @xmath14 . for simplicity , we are considering a finite time interval @xmath15 $ ] and we take the uniform partition @xmath16 on @xmath15 $ ] .specifically , for @xmath17 we have @xmath18 , where we denote @xmath19 .our generic approximation is called @xmath20 , and it starts from the initial condition @xmath21 . in order to introduce our numerical schemes , we shall also use the following notation : [ not : iterated - vector - field ] let @xmath22 and @xmath23 be two smooth vector fields defined on @xmath24 .we denote by @xmath25 the vector @xmath26 , that is , for @xmath27 we have @xmath28^{kl}=\partial_{x_{l}}u^{k}(x)$ ] . with the same matrix convention ,the vector field @xmath29 is defined as @xmath30^{k } = \sum_{l=1}^{d } \partial_{x_{l}}u^{k}(x ) \, v^{l}(x)$ ] . with those preliminaries in mind ,the most classical numerical scheme for stochastic equations is the so - called euler scheme ( or first - order taylor scheme ) , which is recursively defined as follows on the uniform partition : @xmath31 where @xmath32 is defined as @xmath33 for a function @xmath34 .however , it is easily seen ( see e.g @xcite for details ) that the euler scheme is divergent when the hurst parameter @xmath35 is less than @xmath36 . to obtain a convergent numerical approximation in this rough situation ,higher - order terms from the taylor expansion need to be included in the scheme . having the rough paths construction in mind ,the simplest method of this kind is the milstein scheme , or second - order taylor scheme .it can be expressed recursively as : @xmath37 where we have used notation [ not : iterated - vector - field ] and where @xmath38 designates the second - order iterated integral of @xmath4 ( see section [ section3.1 ] for a proper definition ) .this numerical approximation has first been considered in @xcite , and has been shown to be convergent as long as @xmath39 , with an almost sure convergence rate @xmath40 . here and in the following @xmath41 represents an arbitrarily small constant .an extension of the result to @xmath42th - order taylor schemes and to an abstract rough path with arbitrary regularity is contained in @xcite ; see also @xcite for the optimized @xmath42th - order taylor scheme when @xmath1 .the @xmath42th - order taylor schemes of the form are , however , not implementable in general .this is due to the fact that when @xmath43 the terms @xmath44 can not be simulated exactly and have to be approximated on their own .we now mention some contributions giving implementable versions of for stochastic differential equation .they all rely on some cancellation of the randomness in the error process @xmath45 related to our standing equation . _( i ) _ the first second - order implementable scheme for has been introduced in @xcite .it can be expressed in the following form : @xmath46 this scheme has been shown to have convergence rate of order @xmath47 , and the proof relies on the fact that is the 2nd - order taylor scheme for the wong - zakai approximation of equation .the approximation has been extended in @xcite to a third - order scheme defined as follows : @xmath48 thanks to a thorough analysis of differences of iterated integrals between two gaussian processes , a convergence rate @xmath49 has been achieved for the scheme .one should also notice that @xcite handles in fact very general gaussian processes , as long as their covariance function is regular enough in the @xmath50-variation sense ._ ( ii ) _ a different direction has been considered in @xcite , where the following first - order ( meaning first - order with respect to the increments of @xmath4 ) scheme has been introduced : @xmath51 this approximation is called modified euler scheme in @xcite . as has been explained in @xcite ,the modified euler scheme is a natural generalization of the classical euler scheme of the stratonovich sde to the rough sde .for this reason , we will call the euler scheme from now on . as the reader might also see from relation ,one gets the euler scheme from the second - order taylor scheme by changing the terms @xmath44 into their respective expected values .note that since the euler scheme does not involve products of increments of the underlying fbm , its computation cost is much lower than those of and . in spite of this cost reduction , an exact rate of convergence@xmath52 has been achieved in @xcite .the asymptotic error distributions and the weak convergence of the scheme have also been considered , and those results heavily hinge on malliavin calculus considerations .notice however that the results in @xcite are restricted to the case @xmath53 .having recalled those previous results , the aim of the current paper is quite simple : we wish to extend the results concerning the euler scheme to a truly rough situation .namely , we will consider equation driven by a fractional brownian motion @xmath4 with @xmath54 .for this equation , we show that the euler scheme maintains the rate of convergence @xmath55 , which is the same as the third - order implementable scheme in .we also obtain some asymptotic results for the error distribution of the numerical scheme , which generalize the corresponding results in @xcite to the case @xmath56 .more specifically , we will prove the following results ( see theorems [ thm7.3 ] and [ thm9.1 ] for more precise statements ) .[ thm : cvgce - scheme ] let @xmath11 be the solution of equation , and consider the euler approximation scheme @xmath20 defined in . then _( i ) _ for any arbitrarily small @xmath57 , the following almost sure convergence holds true : @xmath58 } |y_{t}-y^{n}_{t}| \rightarrow 0 \quad \text { as } n\rightarrow \infty;\end{aligned}\ ] ] _ ( ii ) _ the sequence of processes @xmath59 converges weakly in @xmath60)$ ] to a process @xmath61 which solves the following equation : @xmath62 where @xmath63 is a @xmath64-valued brownian motion with correlated components , independent of @xmath4 .let us highlight the fact that our approach does not rely on the special structure of the numerical scheme and does not require the analysis of the wong - zakai approximation .in fact , it provides a general procedure for studying time - discrete numerical approximations of rdes , including the implementable schemes and we just mentioned , the backward euler scheme , the crank - nicolson scheme and its modifications , taylor schemes and their modifications introduced in @xcite and so on .let us now explain briefly our strategy in order to prove theorem [ thm : cvgce - scheme ] : @xmath65 the fact that we are replacing the terms @xmath44 in by their expectations forces us to thoroughly analyze some discrete weighted sums in the second chaos of @xmath4 .this will be a substantial part of our computations .@xmath66 then we have to get some uniform bounds ( in @xmath42 ) for the numerical scheme and the related linear rdes like .this will be done thanks to some discrete time rough paths type techniques , which replace the malliavin calculus arguments invoked in @xcite .a special attention has obviously to be paid in order to handle the terms @xmath67 in .also note that the numerical scheme will be reduced to a differential equation driven by a process in the second chaos ( see ) .hence we can not rely on the integrability of the malliavin derivative of @xmath20 , which is a linear rdes driven by the scheme process .@xmath68 one of the main ingredient of our proofs is to consider @xmath69 as a rough path , where @xmath70 is a constant less than @xmath71 .this property will be proved by analyzing some crossed integrals between the 3 components of the process @xmath69 .@xmath72 eventually , our limit theorems are obtained by considering the asymptotic behavior of a rough linear equation describing the evolution of the error @xmath45 , the center of which is a weighted - variation term , or in our point of view , a `` discrete '' rough integral .if we compare the above ingredients with those in @xcite , one can see that the gap between the rough and the non - rough situations is substantial .new ideas are required for every step ( a ) to ( d ) .this corresponds to the steps where the gap between the rough and the non rough situations is particularly large . among the ingredients we have alluded to above ,the asymptotic behavior of weighted variations has been received a lot of attention in recent works ; see e.g. @xcite .our approach to this problem relies on a combination of rough paths and malliavin calculus tools , and might have an interest in its own right ; see theorem [ prop9.2 ] for the precise statement . indeed , with respect to the aforementioned results, it seems that we can reach a more general class of weights .we are also able to consider the variations for multi - dimensional fbms , thanks to a simple approximation argument on the simplex .eventually let us stress the fact that , though we have restricted our analysis to equations driven by a fractional brownian motion here for sake of simplicity , we believe that our results can be extended to a general class of gaussian processes whose covariance function satisfies reasonable assumptions ( such as the ones exhibited in @xcite ) .in this case , if @xmath73 denotes the centered gaussian process at stake , we expect the numerical scheme to become : @xmath74 where @xmath75 is a deterministic constant defined by @xmath76 and where we have assumed that @xmath73 has stationary increments .notice however that ( similar to what is done in @xcite ) one can prove that in general the euler scheme is divergent when one considers an equation driven by a fractional brownian motion with hurst parameter @xmath77 .our analysis is thus restricted to a rough signal with hlder regularity greater than @xmath78 . hereis how our paper is structured : in section [ section2 ] , we recall some results from the theory of rough paths , and prove a discrete version of the sewing map lemma . in section [ section3 ] ,we consider the fractional brownian motion as a rough path and derive some elementary results . in section [ section4i ]we first develop some useful upper - bound estimates , and then we introduce a general limit theorem on the asymptotic behavior of weighted random sums . in section [ section4 ]we consider the couple @xmath79 as a rough path and show that it is uniformly bounded in @xmath42 . in section [ section5 ] we show that the euler scheme @xmath20 is convergent , and we derive our first result on the rate of strong convergence of @xmath20 .we also derive some estimates on the error process @xmath45 .section [ section6 ] is devoted to an elaboration of the estimates for the error process under some new conditions .this leads , in section [ section7 ] , to consider the rate of strong convergence of the euler scheme again , improving the results obtained in section [ section5 ] up to an optimal rate . in section [ section9 ]we prove our main result on the asymptotic error distribution of @xmath20 . in the appendixwe prove some auxiliary results .[ [ notation ] ] * notation * + + + + + + + + + + let @xmath80 be a partition on @xmath15 $ ] .take @xmath81 $ ] .we write @xmath82 for the discrete interval that consists of @xmath83 s such that @xmath84 $ ] .we denote by @xmath85)$ ] the simplex @xmath86 ; \, t_{1}\leq \cdots\leq t_{k } \}$ ] .in contrast , whenever we deal with a discrete interval , we set @xmath87 . for @xmath88we denote @xmath89 , @xmath90 . throughout the paper we work on a probability space @xmath91 .if @xmath73 is a random variable , we denote by @xmath92 the @xmath93-norm of @xmath73 . the letter @xmath94 stands for a constant which can change from line to line , and @xmath95 denotes the integer part of @xmath96 .this section is devoted to introduce the main rough paths notations which will be used in the sequel .we refer to @xcite for further details .we shall also state and prove a discrete sewing lemma which is a simplified version of an analogous result contained in @xcite . in this subsection, we introduce some basic concepts of the rough paths theory .let @xmath97 , and call @xmath98 a fixed finite time horizon .the following notation will prevail until the end of the paper : for a finite dimensional vector space @xmath99 and two functions @xmath100,{{\mathcal v}})$ ] and @xmath101),{{\mathcal v}})$ ] we set @xmath102 we start with the definition of some hlder semi - norms : consider here two paths @xmath103 , \rr^{m})$ ] and @xmath104 ) , ( \rr^{m})^{\otimes 2})$ ] .then we denote @xmath105 , \ga}:=\sup_{(u , v)\in{{\mathcal s}}_{2}([s , t])}\frac{|\delta x_{uv}|}{|v - u|^{\ga } } , \qquad\qquad \|\xx\|_{[s , t ] , 2\ga}:= \sup_{(u , v)\in{{\mathcal s}}_{2}([s , t])}\frac{| \xx_{uv}| } { |v - u|^ { 2\ga } } , \ ] ] where we stress the fact that the regularity of @xmath106 is measured in terms of @xmath107 .when the semi - norms in are finite we say that @xmath108 and @xmath106 are respectively in @xmath109 , \rr^{m})$ ] and @xmath110 ) , ( \rr^{m})^{\otimes 2})$ ] . for convenience ,we denote @xmath111 , \ga } $ ] and @xmath112 , 2\ga } $ ] . with this preliminary notation in hand, we can now turn to the definition of rough path .[ def : rough - path ] let @xmath103 , \rr^{m})$ ] , @xmath104 ) , ( \rr^{m})^{\otimes 2})$ ] , and @xmath97 .we call @xmath113 a ( second - order ) @xmath12-rough path if @xmath114 and @xmath115 , and the following algebraic relation holds true : @xmath116 where we have invoked for the definition of @xmath117 . for a @xmath12-rough path @xmath118, we define a @xmath12-hlder semi - norm as follows : @xmath119 an important subclass of rough paths are the so - called _ geometric @xmath12-hlder rough paths_.a geometric @xmath12-hlder rough path is a rough path @xmath120 such that there exists a sequence of smooth @xmath24-valued paths @xmath121 verifying : @xmath122 we will mainly consider geometric rough paths in the remainder of the article . in relation to , notice that when @xmath108 is a smooth @xmath24-valued path , we can choose @xmath123 it is then easily verified that @xmath124 , with @xmath106 defined in , is a @xmath12-rough path with @xmath125 .in fact , this is also the unique way to lift a smooth path to a @xmath12-rough path .recall now that we interpret equation in the rough paths sense .that is , we shall consider the following general rough differential equation ( rde ) : @xmath126 where @xmath127 and @xmath128 are smooth enough coefficients and @xmath108 is a rough path as given in definition [ def : rough - path ] .we shall interpret equation in a way introduced by davie in @xcite , which is conveniently compatible with numerical approximations .[ def : diff - eq - davie ] we say that @xmath11 is a solution of on @xmath15 $ ] if @xmath129 and there exists a constant @xmath130 and @xmath131 such that @xmath132 for all @xmath133)$ ] , where we recall that @xmath134 is defined by . notice that if @xmath11 solves according to definition [ def : diff - eq - davie ] , then it is also a controlled process as defined in @xcite .namely , if @xmath11 satisfies relation , then we also have : @xmath135 where @xmath136))$ ] .we can thus define iterated integrals of @xmath11 with respect to itself thanks to the sewing map ; see proposition 1 in @xcite .this yields the following decomposition : @xmath137 for all @xmath133)$ ] and @xmath138 .in other words , the signature type path @xmath139 defines a rough path according to definition [ def : rough - path ] .we can now state an existence and uniqueness result for rough differential equations .the reader is referred again to @xcite for further details . in this subsectionwe derive a discrete version of the sewing map lemma which will play a prominent role in the analysis of our numerical scheme .let @xmath145 be a generic partition of the interval @xmath15 $ ] for @xmath146 .for @xmath147 , we denote by @xmath148 the discrete interval @xmath149 .we also label the following definition for further use .[ def : discrete - increments ] we denote by @xmath150 the collection of functions @xmath151 on @xmath152 with values in a banach space @xmath153 such that @xmath154 for @xmath155 . similarly to the continuous case ( relations and ) , we define the operator @xmath156 and some hlder semi - norms on @xmath157 as follows : @xmath158 for @xmath159 and @xmath160 , we also set @xmath161 the space of functions @xmath159 such that @xmath162 is denoted by @xmath163 we can now turn to our discrete version of the sewing map lemma .this result is inspired by @xcite , but is included here since our situation is simpler and leads to a straightforward proof .[ lem2.4 ] for a banach space @xmath164 , an exponent @xmath131 and @xmath165 as in definition [ def : discrete - increments ] , the following relation holds true : @xmath166 take @xmath167 .let @xmath168 , @xmath169 be partitions on @xmath170 defined recursively as follows : set @xmath171 and @xmath172 .given a partition @xmath173 on @xmath174 , @xmath175 , we can find @xmath176 such that @xmath177 we denote by @xmath178 the partition @xmath179 .for @xmath180 , we also set @xmath181 and we observe that @xmath182 , and @xmath183 , where the last relation is due to the fact that @xmath184 . with those preliminaries in hand , we can decompose @xmath185 as follows : we write @xmath186 now , according to the definition of @xmath187 , we have : @xmath188 where the first inequality follows from and the second from . applying the above estimate of @xmath189 to ,we obtain @xmath190 dividing both sides of the above inequality by @xmath191 and taking supremum over @xmath192 in @xmath193 , we obtain the desired estimate .in this section we briefly recall the construction of a rough path above our fbm @xmath4 .the reader is referred to @xcite for further details . in the second part of the section, we turn to some estimates for the lvy area of @xmath4 on a discrete grid , which are essential in the analysis of our scheme .let @xmath194 be a standard @xmath3-dimensional fbm on @xmath15 $ ] with hurst parameter @xmath195 .recall that the covariance function of each coordinate of @xmath4 is defined on @xmath196)$ ] by : @xmath197}.\ ] ] we start by reviewing some properties of the covariance function of @xmath4 considered as a function on @xmath198))^{2}$ ]. namely , take @xmath199 in @xmath15 $ ] and set @xmath200}.\end{aligned}\ ] ] then , whenever @xmath201 , it can be shown that the integral @xmath202 is well - defined as a young integral in the plane . furthermore ,if intervals @xmath203 $ ] and @xmath204 $ ] are disjoint , we have @xmath205 here and in the following , we denote @xmath206 using the elementary properties above , it is shown in ( * ? ? ?* chapter 15 ) that for any piecewise linear or mollifier approximation @xmath207 to @xmath4 , the geometric rough path @xmath208 converges in the @xmath12-hlder semi - norm to a @xmath12-geometric rough path @xmath209 ( given as in definition [ def : rough - path ] ) for @xmath210 .in addition , for @xmath43 the covariance of @xmath211 can be expressed in terms of a @xmath212-dimensional young integral : @xmath213 it is also established in ( * ? ? ? * chapter 15 ) that @xmath214 enjoys the following integrability property .[ prop : integrability - signature ] let @xmath209 be the rough path above @xmath4 , and @xmath215 . then there exists a random variable @xmath216 such that @xmath217 , where @xmath218 is defined by . we now specialize to a situation where @xmath219 and @xmath220 are disjoint intervals such that @xmath221 . in this case relationenables to write : @xmath222 where @xmath223 is the measure given by .note that the left - hand side of converges to @xmath224 as @xmath225 .therefore , the quadruple integral in converges as @xmath225 .this implies that the quadruple integral exists and identity still holds when @xmath226 .having relation in mind , let us label the following definition for further use .[ def3.2 ] denote by @xmath227}$ ] the set of step functions on an interval @xmath228 \subset [ 0,t]$ ] .we call @xmath229}$ ] the hilbert space defined as the closure of @xmath230 } $ ] with respect to the scalar product @xmath231 } , \mathbf{1}_{[s , t ] } \rangle_{\mathcal{h}_{[a , b ] } } & = & r\big(\begin{matrix } u & v \\ s & t \end{matrix } \big ) .\end{aligned}\ ] ] in order to alleviate notations , we will still write @xmath232}$ ] when @xmath228=[0,t]$ ] .notice that the mapping @xmath233 } \rightarrow \delta b_{st}$ ] can be extended to an isometry between @xmath229}$ ] and the gaussian space associated with @xmath234\}$ ] .we denote this isometry by @xmath235 .the random variable @xmath236 is called the ( first - order ) wiener integral and is also denoted by @xmath237 . owing to the fact that @xmath238 throughout this article, we have the following identity : @xmath239 } } & = & \big\| d_{h } s^{\frac12 - h } ( d^{\frac12 -h}_{t- } u^{h-\frac12 } \phi ( u ) ) ( s ) \big\|_{l^{2}([a , b])},\end{aligned}\ ] ] where @xmath240 is a constant depending on @xmath35 and @xmath241 is the right - sided fractional differentiation operator ; see ( 5.31 ) in @xcite. with the help of it is easy to derive the following relation for @xmath242 and @xmath243 : @xmath244 ) } \leq \|h\|_{\mathcal{h}_{[a , b ] } } \leq k_{2}(\sup_{t\in [ a , b]}h(t ) + \|h\|_{c^{\ga } ( [ a , b])})\ , .\end{aligned}\ ] ] indeed , the lower - bound inequality can be obtained by the hardy - littlewood inequality , while the upper - bound estimate follows from the definition of the fractional derivative . in order to generalize relations to a more general situation , recall that for @xmath245}$ ] we have @xmath246}$ ] such that @xmath247 in @xmath229}$ ] .we denote by @xmath248 the extension of @xmath249 on @xmath15 $ ] such that @xmath250 on @xmath228 $ ] and @xmath251 on @xmath15 \setminus [ a , b]$ ] .then @xmath252 } h_{n } db $ ] is a cauchy sequence in @xmath253 , and thus so is @xmath254 in @xmath255 .we denote the limit of @xmath248 by @xmath256 .it is easy to see that @xmath257 satisfies @xmath258 } = h$ ] , @xmath258^{c } } = 0 $ ] , and @xmath259}}$ ] .[ lem3.1 ] take @xmath260}$ ] and @xmath261}$ ] , where @xmath228 $ ] and @xmath262 $ ] are disjoint subintervals of @xmath15 $ ] such that @xmath263 .then the following identity holds true : @xmath264 } f \delta b \int_{[c , d ] } g \delta b \big ) = \me\big ( \int_{0}^{t } f^{e } \delta b \int_{0}^{t } g^{e } \delta b \big ) = \int_{[a , b ] } \int_{[c , d ] } f_{t } g_{s } \mu(ds\ , dt),\end{aligned}\ ] ] where @xmath223 is the measure defined by . take @xmath265 } $ ] and @xmath266}$ ] such that @xmath267 in @xmath229}$ ] and @xmath268 in @xmath269}$ ] .then we have @xmath270 on the other hand , take @xmath271,[c , d])>\kappa>0 $ ] , then owing to we have @xmath272}\int_{[c , d ] } |f_{t } | \cdot |g_{s}| \mu(ds\,dt ) & \leq & h(1 - 2h)\kappa^{2h-2 } \|f^{e}\|_{l^{1}([0,t ] ) } \|g^{e}\|_{l^{1}([0,t ] ) } .\end{aligned}\ ] ] in particular , the left hand side of is finite . since @xmath273}\int_{[c , d ] }f_{n}(t ) g_{n}(s ) \mu(ds\,dt ) , \end{aligned}\ ] ] we can write @xmath274}\int_{[c , d ] } f ( t ) g ( s ) \mu(ds\,dt ) \nonumber \\ & & = \int_{[a , b]}\int_{[c , d ] } ( f_{n } -f ) ( t ) g_{n } ( s)\mu(ds\,dt ) + \int_{[a , b]}\int_{[c , d ] } f ( t ) ( g_{n } -g)(s ) \mu(ds\,dt).\end{aligned}\ ] ] applying to the right - hand side of and taking into account that @xmath275)}\rightarrow 0 $ ] and @xmath276)}\rightarrow 0 $ ] as @xmath277 , which follow from , we obtain @xmath278}\int_{[c , d ] } f_{t } g_{s } \mu(ds\,dt ) \quad \text{as } \quad n\rightarrow \infty.\end{aligned}\ ] ] the identity then follows from and and the uniqueness of the limit of @xmath279 .let @xmath280 be an enhanced fractional brownian motion as in the previous subsection .we now go back to the discrete interval @xmath170 considered in section [ section2.2 ] .we denote by @xmath281 the process on @xmath282 such that @xmath283 for @xmath284 , where @xmath285 if @xmath286 and where we recall that @xmath287 .[ lem4.1 ] for @xmath288 defined as in , we have the following estimate @xmath289 where @xmath290 is a constant depending on @xmath50 , @xmath35 and @xmath291 , and @xmath292 denotes the @xmath293-norm .we only consider the case @xmath294 .the case @xmath43 can be considered similarly . since @xmath295 is a random variable in the second chaos of @xmath4 ,some hyper - contractivity arguments show that it suffices to consider the case @xmath296 in . on the other hand ,it is clear that @xmath297 is equal to @xmath298 in distribution , where @xmath299 .so we are reduced to estimate the following quantity : @xmath300 a classical result in @xcite shows that @xmath301 ; see also theorem 7.4.1 in @xcite . applying this relation to weobtain the estimate .the following result provides a way to find a uniform almost sure upper - bound estimate for a sequence of stochastic processes .[ lem4.2 ] let @xmath302 be a sequence of stochastic processes such that @xmath303 for all @xmath304 , where @xmath305 is a constant depending on @xmath50 . then for @xmath306 and @xmath57 , we can find an integrable random variable @xmath307 independent of @xmath42 such that @xmath308 take @xmath304 such that @xmath309 .the garsia - rodemich - rumsey lemma ( see @xcite ) implies that @xmath310 taking expectation on both sides and taking into account the inequality we obtain @xmath311 & \leq & k_{p } \int_{0}^{t}\int_{0}^{t } \frac { \me \left [ |x^{n}_{u } - x^{n}_{v } |^{p}\right ] } { |u - v|^{2+p\ga } } dudv \\& \leq & k_{p } n^{-p \al } , \end{aligned}\ ] ] and the last inequality can be recast as : @xmath312&\leq & k_{p } n^{-p\kappa}.\end{aligned}\ ] ] we now choose @xmath50 such that @xmath313. therefore the above estimate implies that : @xmath314 \leq \me\big [ \sum_{n \in \nn } \| n^{\al-\kappa } x^{n } \|_{\ga}^{p } \big ] \\\leq k_{p}\sum_{n \in \nn } n^{-p\kappa}<\infty.\ ] ] in partitular , we obtain that @xmath315 is an integrable random variable . by taking @xmath316, we obtain the desired estimate for @xmath317 .[ remark3.6 ] let @xmath12 be a parameter such that @xmath318 .starting from relation and taking into account the fact that @xmath319 for all @xmath320 , it is readily checked that the increment @xmath321 introduced in lemma [ lem4.1 ] satisfies @xmath322 for all @xmath323 and @xmath324 . by considering the linear interpolation of @xmath321 on @xmath15 $ ], inequality also holds for all @xmath325 ) $ ] . owing to lemma [ lem4.2 ], we can thus find an integrable random variable @xmath326 such that for any @xmath327 we have @xmath328 weighted random sums via the rough path approach ------------------------------------------------ in this section , we derive some useful upper - bound estimates for weighted random sums related to @xmath4 .in the second part of the section , we prove a general limit theorem , which is our main result of this section .we now derive some estimates for weighted random sums . as has been mentioned in the introduction, these results only require the weight function to satisfy some proper regularity conditions .[ prop3.4 ] let @xmath34 and @xmath329 be processes on @xmath330 such that @xmath331 and @xmath332 , where @xmath333 .we define an increment @xmath151 on @xmath334 by : @xmath335 then the following estimate holds true : @xmath336 it is clear that @xmath337 .in addition , the following relation is readily checked , where we recall that @xmath338 is defined by : @xmath339 therefore we have @xmath340 .since we have assumed @xmath341 , we can invoke the discrete sewing map lemma [ lem2.4 ] , which yields : @xmath342 the proposition then follows immediately .[ remark3.8 ] the riemann - stieltjes sum @xmath343 in proposition [ prop3.4 ] can be thought of as a @xmath344-valued `` discrete '' young integral .one can also consider @xmath93-valued `` discrete '' young integrals in a similar way by viewing @xmath34 and @xmath329 as functions with values in @xmath93 .this will lead us to an @xmath93-estimate of @xmath345 . precisely , suppose that @xmath34 and @xmath329 are processes such that @xmath346 and @xmath347 for all @xmath304 .then we have @xmath348 in the sequel , we consider an application of proposition [ prop3.4 ] to third - order terms in our taylor expansion for equation . towards this aim , we first need the following estimate in @xmath293 .they are somehow reminiscent of the estimates for triple integrals in @xcite , though our main focus here is on cumulative sums of triple integrals .[ lem11.2 ] let @xmath4 be an @xmath349-valued fractional brownian motion with hurst parameter @xmath350 .for a fixed set of coordinates @xmath351 , we define two increments @xmath352 and @xmath353on @xmath354 as follows : @xmath355 then the following estimate is valid for @xmath356 : @xmath357 when @xmath358 we have @xmath359 and the estimate follows from the classical results in @xcite . in the followingwe consider the case when @xmath360 , @xmath361 are not all equal. let us further reduce our problem .first , since the fbm has stationary increment it suffices to prove the lemma for @xmath362 .furthermore , by self - similarity of the fbm , we have the following equation in distribution : @xmath363 as a last preliminary step , note that @xmath364 takes values in the third chaos of @xmath4 , on which all @xmath365-norms are equivalent .hence our claim boils down to prove : @xmath366 we now focus on this inequality .we first consider the case when @xmath367 and @xmath368 are different from each other . in order to prove relation ,write @xmath369 for the sum @xmath370 above , thanks to the independence of @xmath371 , @xmath372 and @xmath373 , one can apply lemma [ lem3.1 ] twice in order to get : @xmath374 in particular , since reveals that @xmath223 is a negative measure whenever @xmath238 , we have @xmath375 moreover , since @xmath376 and @xmath377 , the following bound is easily checked : @xmath378 applying and to the right - hand side of , we obtain .assume now that @xmath379 .then @xmath380 where we have set @xmath381db^{l}_{u } , \quad\text{and}\quad \zeta^{2,l}_{k , k+1 } = \int_{k}^{k+1 } ( u - k)^{2h } db^{l}_{u}.\ ] ] we now treat the terms @xmath382 and @xmath383 similarly to the case of different indices @xmath384 .namely , we decompose @xmath385 as in .then in the same way as for and , we can show that : @xmath386 one can thus easily show that @xmath387 satisfies the inequality .the same argument can be applied to @xmath388 , which yields for the case @xmath379 .we let the patient reader check that the same inequality holds true in the case @xmath389 , resorting to a stochastic fubini lemma .this completes the proof .[ remark3.9 ] the estimate of @xmath387 obtained in the proof of lemma [ lem11.2 ] implies that @xmath390 this inequality will be used below in order to prove lemma [ cor3.13 ] .we can now deliver a path - wise bound on weighted sums of the process @xmath364 .[ lem3.11 ] consider the increment @xmath364 defined by .let @xmath34 be a process on @xmath15 $ ] such that , for any @xmath391 , there exists a random variable @xmath392 such that @xmath393 .then for any @xmath57 we have the estimate @xmath394 where @xmath392 is an integrable random variable independent of @xmath42 .consider @xmath395 , and observe that the following decomposition holds true : @xmath396 where increment @xmath329 has been defined by .in addition , thanks to lemma [ lem11.2 ] , we obtain : @xmath397 where the last inequality is due to the fact that @xmath398 . here@xmath13 and @xmath57 .applying lemma [ lem4.2 ] to @xmath329 we thus get : @xmath399 is a random variable independent of @xmath42 .now observe that a direct application of proposition [ prop3.4 ] ( notice that @xmath400 in the relation below ) enables to write : @xmath401 where @xmath402 is another integrable random variable independent of @xmath42 . plugging and into the right - hand side of, we obtain the desired estimate .we now consider the case of a weighted sum involving a wiener integral with respect to @xmath4 .[ cor3.13 ] let @xmath34 be as in lemma [ lem3.11 ] and @xmath13. then the following estimate holds true : @xmath403 the corollary is a direct application of proposition [ prop3.4 ] and is similar to the proof of lemma [ lem3.11 ] .the details are omitted .we turn to controls of weighted sums in cases involving rougher processes .they provide our first instances where we apply rough path methods for weighted sums , as announced in the introduction . in the following, @xmath99 and @xmath404 stands for some finite dimensional vector spaces .[ prop3.6 ]let @xmath34 , @xmath329 be two processes defined on @xmath15 $ ] with values in @xmath99 and @xmath405 , respectively , and @xmath406 be an increment defined on @xmath407)$ ] with values in @xmath404 .assume that there is a constant @xmath94 and an exponent @xmath408 such that the following conditions are met for @xmath409)$ ] and all @xmath410 : @xmath411 we also suppose that @xmath406 satisfies @xmath412 for @xmath356 and any @xmath410 , where @xmath413 is such that @xmath414 .then we have @xmath415 and @xmath416 which are valid for @xmath356 and all @xmath410 . furthermore , set @xmath417 then we have the estimate @xmath418 we start by proving inequality .to this aim , set @xmath419 for @xmath356 , and consider @xmath410 .we decompose the increment @xmath420 into @xmath421 , where @xmath151 is the increment defined in and @xmath422 is defined by : @xmath423 then it is immediate from and that @xmath424 in order to bound the increment @xmath151 , we note that @xmath425 .let us now calculate @xmath338 : for @xmath426 , it is readily checked that : @xmath427 therefore invoking and again , we get : @xmath428 where @xmath429 , and where by assumption we have @xmath131 .hence , owing to the discrete sewing lemma [ lem2.4 ] ( applied to the banach space @xmath430 ) we have : @xmath431 where @xmath432 designates the @xmath223-hlder norm for @xmath433-valued functions .putting together our estimates and on @xmath422 and @xmath151 , inequality is proved . in the followingwe derive our second claim .the method is similar to the proof of , so that it will only be sketched for sake of conciseness .we resort to the following decomposition : @xmath434 where @xmath435 in order to bound @xmath436 we change the order of summation , which allows to exhibits some terms of the form @xmath437. then we let the reader check that inequality can be applied directly .as far as @xmath438 is concerned , notice again that @xmath439 .it is then readily seen , as in the previous step , that our estimate boils down to a bound on @xmath440 .furthermore , @xmath441 can be computed as follows : @xmath442 we can thus resort to , and in order to get : @xmath443 the proof is now completed as for relation .[ remark4.7 ] in proposition [ prop3.6 ] the weighted sum @xmath444 is viewed as a @xmath93-valued `` discrete '' rough integral .similarly as in remark [ remark3.8 ] , a @xmath344-valued `` discrete '' rough path can also be considered .proposition [ prop3.6 ] can be applied to the sum @xmath445 of lvy area increments of @xmath4 .this is the contents of the corollary below .[ lem9.1 ] let @xmath446 , @xmath34 and @xmath329 be as in proposition [ prop3.6 ] .let @xmath445 be the process defined by , considered as a path taking values in @xmath447 .then the following estimates hold true for @xmath448 : @xmath449 and @xmath450 furthermore , set @xmath451 , then we have @xmath452 take @xmath453 and @xmath454 .it follows from lemma [ lem4.1 ] and lemma [ lem : sum - delta - b - f ] that @xmath406 satisfies the conditions in proposition [ prop3.6 ] .in addition , if @xmath455 , the condition @xmath456 is trivially satisfied .the corollary then follows immediately from proposition [ prop3.6 ] and taking into account the decomposition @xmath457 .the estimate of @xmath458 follows directly from relation .take two uniform partitions on @xmath15 $ ] : @xmath459 and @xmath460 , @xmath461 for @xmath462 and @xmath463 .let @xmath464 be such that @xmath465 . in the following , we set for each @xmath466 $ ] : @xmath467our main result in this section is a central limit theorem for sums weighted by a controlled process @xmath34 .[ prop9.2 ] let the assumptions in proposition [ prop3.6 ] prevail , and suppose that @xmath468 where `` @xmath469 '' stands for convergence of finite dimensional distributions and @xmath470 is a brownian motion independent of @xmath4 . set @xmath471 and suppose that @xmath472 for @xmath473 and for an arbitrary @xmath57 . set @xmath474 and @xmath475 .then the following relation holds true : @xmath476 there are several possible generalizations of the statement of theorem [ prop9.2 ] .if one has the convergence of @xmath477 in @xmath93 instead of the weak convergence , then by a similar proof one can show that @xmath478 converges to @xmath479 in @xmath365 . on the other hand ,similarly to what has been mentioned in remark [ remark4.7 ] , if conditions , and are replaced by the corresponding almost sure upper - bound conditions , then one can show that theorem [ prop9.2 ] still holds true .[ remark4.12 ] in the case @xmath480 and @xmath481 , conditions , and are reduced to @xmath482 and @xmath483 .theorem [ prop9.2 ] then recovers the central and non - central limit theorem results in @xcite and @xcite . according to our proof of theorem [ prop9.2 ] , in general the limit of the `` riemann sum '' @xmath484 is independent of the choices of the representative points .in the situation of remark [ remark4.12 ] this fact can be proved directly from the expression of @xmath484 . by definition of the f.d.d .convergence , it suffices to show the following weak convergence for @xmath485 $ ] : @xmath486 _ step 1 : a coarse graining argument . _consider an extra parameter @xmath487 and take @xmath488 to be the uniform @xmath489-step partition of @xmath15 $ ] .we make the following decomposition : @xmath490 where @xmath491 let us first handle the convergence of @xmath492 : by letting @xmath493 and taking into account the convergence , and then letting @xmath494 , we easily obtain the weak convergence : @xmath495 therefore , in order to prove our claim , we are reduced to show that for @xmath466 $ ] @xmath496 _ step 2 : first - order approximation of @xmath34 . _ let @xmath464 be such that @xmath465 .we compare the two sets : @xmath497 defined by and @xmath498 .it is easy to see that @xmath499 .it is also readily checked from conditions and that @xmath500 and @xmath501 .a simple use of cauchy - schwarz inequality thus yields : @xmath502 for any @xmath503 .in particular , holds for @xmath504 .therefore , in order to show , it is sufficient to show that @xmath505 where @xmath506 now in order to get relation , consider the following decomposition for @xmath507 : @xmath508 where the increments @xmath509 , @xmath510 , @xmath511 and @xmath512 are defined by @xmath513 and @xmath514 it follows from that @xmath515 on the other hand , it follows from and that @xmath516 where recall that we denote @xmath517 and @xmath518 . similarly , applying and we obtain :@xmath519 it follows immediately from , and that @xmath520 in view of and taking into account the decomposition , in order to show it suffices to show that @xmath521 _ step 3 : study of @xmath522 ._ we will see that @xmath523 can be considered as a discrete `` young '' integral in @xmath524 in the sense of remark [ remark3.8 ] ( see also proposition [ prop3.4 ] ) , which then leads to the convergence .namely , starting from the expression of @xmath522 , let us first consider the `` weight - free '' sum @xmath525 where @xmath526 and recall that @xmath527 is defined in .observe that can be recast as @xmath528 for all @xmath529 . according to, we have @xmath530 then our assumption and the bound ensures that we are in a position to apply proposition [ prop3.4 ] .this immediately yields our claim , which concludes the proof .in this section , we consider a continuous time interpolation of the euler scheme @xmath20 given by .namely , we introduce a sequence of processes @xmath20 indexed by @xmath15 $ ] in the following way : for @xmath531 we set @xmath532 where we recall that @xmath533 designates the uniform partition of the interval @xmath15 $ ] .the remainder of the section is devoted to get some uniform bounds on @xmath20 , and then to prove that the couple @xmath534 can be lifted as a rough path . throughout the section ,we assume that @xmath535 and @xmath536 .[ lem5.1 ] let @xmath20 be the process defined by the euler scheme .take @xmath318 .then for all @xmath537 , the following relations are satisfied : latexmath:[\ ] ] the estimate also holds true for @xmath1147 , and the proof is similar . in addition , for @xmath1167 the relation shows that @xmath1168 and @xmath1169 which yields for @xmath1170 : @xmath1171 gathering our bounds and for @xmath1172 , @xmath1146 and @xmath1147 , it is readily checked from our decompositions and that for @xmath1126 we have : @xmath1173 furthermore , using the stationarity of the increments of @xmath445 and @xmath4 , plus the equivalence of @xmath93-norms in finite chaos , we obtain from the desired estimate . moreover , the estimate holds true for other @xmath1174 .the proof is similar and is omitted .since @xmath4 has stationary increment , it suffices to prove the lemma for @xmath362 . as in the proof of proposition [ prop9.2 ], we consider the sum @xmath1176 for @xmath1126 .the other cases can be considered similarly .let @xmath1177 , @xmath709 be the quantities defined in , , . bywe have @xmath1178 applying and taking into account lemma [ lem4.1 ] yields @xmath1179 on the other hand , it follows from and that for @xmath1170 we have @xmath1180 the lemma then follows by applying and to .deya , a. ; neuenkirch , a. and tindel , s. ( 2012 ) . a milstein - type scheme without lvy area terms for sdes driven by fractional brownian motion .henri poincar probab .stat . _ * 48 * , no . 2 , 518 - 550 .friz , peter k. ; gess , benjamin ; gulisashvili , archil ; riedel , sebastian .the jain - monrad criterion for rough paths and applications to random fourier series and non - markovian hrmander theory .* 44 * ( 2016 ) , no .1 , 684738 .nourdin , i. ; nualart , d. and tudor , c. ( 2010 ) .central and non - central limit theorems for weighted power variations of fractional brownian motion .henri poincar probab .stat . _ * 46 * no . 4 , 1055 - 1079 .

<S> in this article , we consider the so - called modified euler scheme for stochastic differential equations ( sdes ) driven by fractional brownian motions ( fbm ) with hurst parameter @xmath0 . </S><S> this is a first - order time - discrete numerical approximation scheme , and has been introduced in @xcite recently in order to generalize the classical euler scheme for it sdes to the case @xmath1 . </S><S> the current contribution generalizes the modified euler scheme to the rough case @xmath0 . </S><S> namely , we show a convergence rate of order @xmath2 for the scheme , and we argue that this rate is exact . we also derive a central limit theorem for the renormalized error of the scheme , thanks to some new techniques for asymptotics of weighted random sums . </S><S> our main idea is based on the following observation : the triple of processes obtained by considering the fbm , the scheme process and the normalized error process , can be lifted to a new rough path . </S><S> in addition , the hlder norm of this new rough path has an estimate which is independent of the step - size of the scheme . </S>

hep-th9903054

there has been a great progress in string theory of the d1-d5 brane system with momentum modes along the string direction(s@xmath1 ) .this gives us a 5d black hole(m@xmath2 ) with three charges @xmath3 .the first progress was achieved in the bekenstein - hawking entropy@xcite .apart from the success of counting the microstates of a 5d black hole through d - brane physics , a dynamical consideration becomes an important issue@xcite .this is so because the semiclassical absorption cross section ( greybody factor ) for a test field arises as a consequence of its potential barrier surrounding the horizon .that is , this is an effect of spacetime curvature .more precisely , it is worth noting that a semiclassical absorption cross section can be derived from a solution to the differential equation of a test field(most often , @xmath4 ) on the supergravity side .the ads / cft correspondence states that string theory in the ads is dual to a conformal field theory(cft ) defined on its remote boundary of ads@xcite .the semiclassical limit of spacetime physics is related to the large n limit of the dual cft .a 5d black hole(m@xmath5s@xmath6t@xmath7 ) becomes ads@xmath8s@xmath9t@xmath7 near horizon but with an asymptotically flat space(afs)@xcite .recently this is called an ads@xmath0 bubble in afs and corresponds to the dilatonic solution@xcite . in this casethe matching procedure is crucial for obtaining an absorption coefficient and here we need to match an ads@xmath0 bubble to afs . on the other hand , one obtains ads@xmath8s@xmath9t@xmath7 as the other solution to the type iib string theory .this is an exact ads@xmath0 with asymptotically ads@xmath0 and corresponds to the non - dilatonic solution@xcite .further it includes the btz black hole@xcite .we point out that the near horizon limit of a 5d black hole yields the btz black hole , whereas this solution accomodates to the btz black hole spacetime as a whole .it was understood that the greybody factor calculation makes sense when one finds an asymptotically flat region as in sec.ii@xcite .hence it may not be possible in an ads@xmath0 because there is no asymptotic state corresponding to particle at infinity of ads@xmath0 .however , the authors in@xcite calculated the greybody factor for a free scalar and the dilaton both in m@xmath5s@xmath6t@xmath7(an ads@xmath0 bubble ) and ads@xmath8s@xmath9t@xmath7(an exact ads@xmath0 ) within the type iib supergravity . here in the exact ads@xmath0 calculation we choose the non - normalizable modes to obtain the greybody factor .this corresponds to the ads@xmath0-ads@xmath0 matching procedure .this expression denotes shorthand for a certain choice of boundary conditions where non - normalizable modes inject flux into ads@xmath0 .it turns out that two results of a free scalar are exactly the same . and the results for the dilaton are the same upto a factor 3 .these show a close relationship between two approaches .more recently , vacuum correlators of the dual cft@xmath10 were expressed as truncated n - point functions for the non - normalizable modes in ads@xmath2 .one can interpret this result as an s matrix of an exact ads@xmath2 arising from a limit of scattering from an ads@xmath2 bubble@xcite .this supports that our calculation based on the non - normalizable modes is correct . in this paperwe will show that the s - matrix of an ads@xmath0 bubble can be derived partly from an exact ads@xmath0 space .this is one of the current issues in the ads / cft correspondence . for this purpose, we investigate the close relationship between the potential and absorption cross section in an ads@xmath0 bubble and an exact ads@xmath0 . comparing an ads@xmath0 bubble with an exact ads@xmath0leads to an assumption that the potential of an exact ads@xmath0 is the left - hand side of an ads@xmath0 bubble . for this study, we introduce the \{in } and \{out}-state pictures for an ads@xmath0 bubble . for the exact ads@xmath0 study, one needs the \{in}-state as well as non - normalizable modes .further , we introduce the test fields for scattering analysis .these are in an ads@xmath0 bubble : a free scalar(@xmath11 ) which , in the decoupling limit , relates to an ( 1,1 ) operator @xmath12 in the holographically dual theory ; two fixed scalars(@xmath13 ) to ( 2,2 ) , ( 3,1 ) , and ( 1,3 ) operators ; two intermediate scalars(@xmath14 ) to ( 1,2 ) and ( 2,1 ) operators . on the exact ads@xmath0 side ,the test fields are a free scalar(@xmath15 ) , the dilaton(@xmath16 ) , an intermediate scalar(@xmath17 ) , and the tachyon(@xmath18 ) .the organization of this work is a follows .sec.ii is devoted to analyzing the scattering from an ads@xmath0 bubble within a 5d black hole .this corresponds to a conventional scattering study .we study the scattering of the test fields in an exact ads@xmath0 in sec .iii . in this casewe are careful for obtaining the absorption coefficient because we can not define the asymptotically flat space .finally we discuss our results in sec .initially we introduce all perturbing modes in a 5d black hole background .it is pointed out that in s - wave calculation the fixed scalars are physically propagating modes and other fields belong to be redundant modes@xcite .hence we choose two fixed scalars @xmath19 and a free scalar@xmath20 as the relevant modes .we begin with the 5d black hole with three charges , @xmath21 where @xmath22 here the radii are related to the boost parameters @xmath23 and the charges @xmath24 as @xmath25 hence the d - brane black hole depends on the four parameters @xmath26 .the event horizon ( outer horizon ) is clearly at @xmath27 . when all three charges are nonzero , the surface @xmath28 becomes a smooth inner horizon ( cauchy horizon ) .when at least one of the charges is zero , the surface @xmath28 becomes singular .the extremal case corresponds to the limit of @xmath29 with the boost parameters @xmath30 , keeping the charges @xmath24 fixed .we are interested in the limit of @xmath31 , which is called the dilute gas limit .this is so because this limit corresponds to the near - outer horizon .here we choose @xmath32 , and @xmath33 with a finite @xmath34 .this corresponds to the near - extremal black hole and its thermodynamic quantities ( energy , entropy , hawking temperature ) are given by @xmath35 , \label{energy - next}\\ & & s_{\rm next } = { 4 \pi^3 r_0 \over \kappa_5^ 2 } r_1 r_5\cosh \alpha_n , \label{entropy - next}\\ & & { 1 \over t_{\rm h , next } } = { 2 \pi \over r_0 } r_1 r_5 \cosh \alpha_n , \label{temperature - next}\end{aligned}\ ] ] where @xmath36 is the 5d gravitational constant .the above energy and entropy are those of a gas of massless 1d particles . in this casethe temperatures for left and right moving string modes are given by @xmath37 this implies that the ( left and right moving ) momentum modes along the string direction are excited , while the excitations of d1-anti d1 and d5-anti d5-branes are suppressed .the hawking temperature is given by their harmonic average @xmath38 for a free scalar @xmath39 , the linearized equation @xmath40 in the background of ( [ metric-5d ] ) leads to@xcite @xmath41 \phi = 0 .\label{eq - free}\ ] ] the s - wave(@xmath42 ) linearized equation for the fixed scalars takes the form@xcite @xmath43 \phi_{\pm}=0 , \label{eq - swave}\ ] ] where one gets @xmath44 for @xmath45 and @xmath46 for @xmath47 . here@xmath48/3 $ ] .considering @xmath49 , for @xmath50 and introducing a tortoise coordinate @xmath51@xcite , then the equation takes the form @xmath52 here we take @xmath53 and @xmath54 for simplicity . in the dilute gas limit @xmath55is given by @xmath56 , \label{potential - nu}\ ] ] @xmath57 , \label{potential - lambda}\ ] ] @xmath58 , \label{potential - phi}\ ] ] where @xmath59 we note that @xmath60 depends on two parameters ( @xmath61 ) as well as the energy@xmath62 . as ( [ eqom ] ) stands , it is far from the schrdinger - type equation .the @xmath63-dependence is a matter of peculiar interest to us compared with the schwarzschild black hole potentials @xmath64@xcite .this makes the interpretation of @xmath60 as a potential difficult . as is shown in fig.[fig - f-1 ] , this arises because @xmath65 is very large as @xmath66 for @xmath67 in the near - horizon . in order for @xmath60 to be a potential ,it is necessary to take the low energy limit of @xmath68 .it is suitable to be @xmath69 . and @xmath70 is of order @xmath71 and thus it can be ignored in comparison to the remaing ones .now we can define a potential @xmath72 .hence , in the low energy limit(@xmath68 ) , eq.([eqom ] ) becomes as the schrdinger - type equation .further the last terms in ( [ potential - nu])-([potential - lambda ] ) are important to compare each other . after the partial fraction, these lead to @xmath73 the last term of a free scalar in ( [ potential - phi ] ) with @xmath74 keeps the first terms in ( [ par - far1])-([par - far2 ] ) only .one finds immediately the sequence @xmath75 here @xmath76 denotes the s - wave(@xmath42 ) free scalar and @xmath77 the free one with @xmath74 .this is also observed from the graphs of potential in fig.[fig - potential ] with @xmath78 . because the shape of their potentials takes nearly the same form ( @xmath79 ) , these give us nearly the same reflection coefficient @xmath80 and absorption one @xmath81 .for example , in the low energy limit of @xmath68 , @xmath82 take nearly the zero - absorption cross section .furthermore all potentials go to zero , as @xmath83 approaches infinity .this implies the existence of the asymptotic states outside an ads@xmath0 bubble .the size of an ads@xmath0 bubble is from @xmath84 to @xmath85 .we are interested in the scattering of the test fields off @xmath86 .it is well known that the scattering analysis is usually done by choosing a coordinate such as @xmath87 .it is always possible to visualize the black hole as presenting an effective potential barrier ( or well ) to an incoming test wave .one expects that some of the incident wave will be irreversibly absorbed by the black hole , while the remaining fraction will be scattered back to the infinity . in this scatteringwe can calculate the reflection and transmission(absorption ) coefficients@xcite . as is shown in fig.[fig - potential - afs - tor ], we note that all potential barriers @xmath86 take nearly the symmetric forms around @xmath88 when they are rewritten by a tortoise coordinate @xmath89@xcite .also they are localized at @xmath88 . since @xmath90 as @xmath91 ,one finds @xmath92 asymptotically(@xmath93 ) the solution is given by @xmath94 considering the time - part of @xmath95 , the first is an incoming wave(@xmath96 ) and the last is an outgoing wave(@xmath97 ) . near the horizonit is purely incoming(@xmath96 ) as @xmath98 we call this type of solution as @xmath99 and the corresponding vacuum state is defined as @xmath100 . in order to study the hawking radiation , we introduce another boundary condition . asymptotically the wave is purely outgoing(@xmath97 ) @xmath101 but near the horizon it has both outgoing(@xmath97 ) and incoming(@xmath96 ) parts as @xmath102 we call this type of solution as @xmath103 and its vacuum state is defined as @xmath104 .the vacuum states @xmath105 and @xmath106 form two different bases of which any state can be expanded in terms of the other .these are two different fock space vacuum states and @xmath107 and @xmath103 are related to each other by the bogoliubov transformation , @xmath108 the computation of hawking shows in a semi - classical approximation that the thermal radiation from the black hole with temperature @xmath109 is given by@xcite @xmath110 with an absorption cross section @xmath111 .note that if @xmath112 is a constant , @xmath113 is the same as that of a black body .typically , @xmath112 is not constant but varies .the deviations from the black body spectrum have earned it the name `` greybody factor '' . herewe define the s - matrix from @xmath114 and @xmath115 as@xcite @xmath116 here an incident wave(@xmath117 ) of unit amplitude from @xmath118 gives rise to @xmath119 and @xmath120 . on the other hand ,an incident wave(@xmath121 ) of unit amplitude from @xmath122 gives rise to @xmath123 and @xmath124 .the relation between these is given by @xmath125 @xmath126 @xmath127 the above relations establish the symmetry and unitarity of the s - matrix in the ads@xmath0 bubble scattering .actually @xmath128 can be calculated from the backscattering of an incident wave @xmath129 off the potential @xmath86 .it is not easy to find out the absorption amplitude @xmath130 directly in the complicated potentials such as @xmath86 . in this caseone use the flux @xmath131 of the incoming wave to obtain the absorption coefficient @xmath132 where @xmath133 are the fluxes at the horizon(infinity ) . in this waywe can calculate the semiclassical absoprtion cross section .first let us consider a free scalar .using @xmath134 , the wave equation ( [ eq - free ] ) can be rewritten as @xmath135 where @xmath136 \simeq \left ( { \omega \over 4 \pi t_h } \right ) ^2 \label{coef - q}.\end{aligned}\ ] ] here @xmath137 means both the dilute gas limit(@xmath31 ) and low energy limit(@xmath68 ) .then one has @xmath138 . in order to compare ( [ near - eq ] ) with the hypergeometric equation, one has to transform it into the pole - free equation . with an unknown constant @xmath139, we find the ingoing mode at the horizon @xmath140 where @xmath141 the large @xmath83-behavior ( @xmath142 ) of @xmath143 can be obtained from the ( @xmath144 ) transformation rule for the hypergeometric functions as@xmath145 for the fixed scalars(@xmath13 ) , considering both ( [ par - far1 ] ) , ( [ par - far2 ] ) and the near - horizon condition of @xmath146 , one finds that ( [ eq - swave ] ) leads to ( [ eq - free ] ) with @xmath74 .thus one can obtain their near - horizon behaviors from @xmath77 .let us first consider a free scalar . in the far region , we introduce @xmath147 and @xmath148 and then ( [ eq - free ] ) leads to @xmath149 \breve \phi = 0 .\label{far - eq}\ ] ] the solution is given by the bessel function when @xmath44 is not an integer @xmath150 , \label{far - sol}\ ] ] where @xmath151 are unknown constants .from the large @xmath152-behavior(@xmath153 ) , one finds the asymptotic states @xmath154 and its incoming flux @xmath155 the small @xmath152-behavior(@xmath156 ) of @xmath157 is @xmath158 .\label{far - small - sol}\ ] ] on the other hand , the asymptotic behavior of the s - wave fixed scalars ( @xmath159 ) is governed by @xmath160 \breve \phi_\pm = 0 \label{far - eq - swave}\ ] ] with @xmath161 .its solution is given by @xmath162 .\label{far - sol - swave}\ ] ] here @xmath163 are unknown constants .here we consider the geometry of an exact ads@xmath0(ads@xmath187s@xmath186 t@xmath7 ) as the other solution to the type iib string theory@xcite .this corresponds to the non - dilatonic solution .this geometry can be led the btz black hole spacetime as a whole by the periodic identification .a ten - dimensional minimally coupled scalar satisfies @xmath188 @xmath189 can be decomposed into @xmath190 then eq.([eq - minimal ] ) leads to @xmath191 with @xmath192 .the @xmath193 case contains both the dilaton(@xmath16 ) and a free scalar(@xmath15 ) with @xmath74 .the @xmath194 case corresponds to an intermediate scalar ( @xmath17 ) and @xmath195 leads to tachyon(@xmath18 ) . herethe btz black hole spacetime is given by@xcite @xmath196 with @xmath197 @xmath198 .the mass , angular momentum , angular velocity at horizon and area of horizon are @xmath199 further one finds the relation between the btz and a 5d black holes as @xmath200 its s - wave equation with @xmath201 takes the form @xmath202 \psi(\rho ) = 0 .\label{eq - decoupled}\end{aligned}\ ] ] defining @xmath203 and then ( [ eq - decoupled ] ) takes the form @xmath204\tilde \psi = 0 , \label{eq - dec - tilde}\ ] ] where the prime(@xmath205 ) denotes the differentiation with respect to @xmath206 . in order to obtain the schrdinger - type equation, we introduce the tortoise coordinate @xmath207 as@xcite @xmath208 . \label{tortoise}\ ] ] we note that @xmath209 , while @xmath210 for outside horizon . on the other hand , in a 5d black holeone finds that @xmath211 is mapped into @xmath87 .then ( [ eq - dec - tilde ] ) leads to @xmath212 where the potential is given by @xmath213 .\label{potential - tortoise}\ ] ] four potential graphs(@xmath214 ) in an exact ads@xmath0 background are shown in fig.[fig - potential - psi ] .the parameters are chosen as @xmath215 .these are all monotonically increasing functions with @xmath206 , contrast to @xmath216 for an ads@xmath0 bubble .this shows a peculiar property of ads@xmath0 , which in this spacetime the asymptotic states can not be defined .but the tachyon potential(@xmath217 ) is a monotonically decreasing function .this field couples to the minimal weight primary operator with ( 1/2,1/2)@xcite .also this satisfies both the stability condition for ads@xmath0 and the dirichlet boundary condition@xcite , which are clearly related to a shape of its potential . in the near - horizon(@xmath218 ) , ( [ eq - sch ] ) reduces to @xmath219 which leads to the plane - wave solution @xmath220 here the first term is an outgoing mode(@xmath97 ) and the second is an incoming mode(@xmath96 ) .now let us discuss the asymptotically ads@xmath0-behavior .near the timelike boundary(@xmath221 ) , one finds @xmath222 { \rho^2 \over r^4 } \tilde \psi_\infty = 0 .\label{eq - boundary}\ ] ] here we introduce the relation between @xmath206 and @xmath207 @xmath223 if @xmath224 is very small and @xmath225 , one finds @xmath226 using ( [ eq - rho ] ) , ( [ eq - boundary ] ) leads to @xmath227 { \tilde \psi_\infty \over { \rho^*}^2 } = 0 .\label{eq - bnd - tortoise}\ ] ] its solution takes the form @xmath228 finally we have @xmath229 for the s - wave free scalar(@xmath230 ) , @xmath231 takes the form(@xmath232 , constant ) and for the dilaton field(@xmath233 ) , one finds ( @xmath234 ) .we find ( @xmath235 ) for a conformally coupled scalar(@xmath236)@xcite .for the tachyon(@xmath195 ) , one has ( @xmath237 ) and for an intermediate scalar(@xmath238 ) , one finds ( @xmath239 ) . instead of the plane - wave from , hereone finds the power - law behaviors of @xmath207@xcite .the non - normalizable modes are found as @xmath240 for the dilaton and @xmath241 for an intermediate scalar , which diverge as @xmath207 approaches the timelike bounadry(@xmath242 ) .the positive powers of @xmath207 all belong to be the normalizable modes , which converge at spatial infinity .we note that the tachyon takes only the normalizable modes@xcite .this can be easily conjectured from its shape of potential @xmath217 .the s - wave free scalar takes a constant - behavior at infinity .this makes an difficulty in dividing @xmath231 into an ingoing and outgoing modes at spatial infinity@xcite .actually it is impossible to define an ingoing wave and an outgoing wave at spatial infinity of an exact ads@xmath0 .instead in the asymptotically ads@xmath0 space it contains the normalizable as well as the non - normalizable modes. the latter will play an important role in calculation of the absorption coefficient @xmath243 .first we note that @xmath207 covers only the left - hand side(@xmath210 ) of the whole space . in the case of where the black hole geometry is asymptotically flat as a 5d black hole ,the tortoise coordinate @xmath89 goes from @xmath244 to @xmath245 .hence this is similar to the infinite string problem in which the initial data propagates towards left and right indefinitely@xcite .the initial data no longer enjoys this privilege when the background is asymptotically ads@xmath0 because the tortoise coordinate @xmath207 goes from @xmath244 to 0 only .one may consider this as the semi - infinite string problem(or a finite cavity with reflecting walls in ads@xmath2@xcite ) .then the dirichlet or neumann boundary condition at spatial infinity(@xmath246 ) is required to formulate the problem appropriately .however , we take a different point of view to attack an asymptotically ads@xmath0 problem .this is based on the observation of the shape of potential and the global structure of an exact ads@xmath0@xcite .we first construct the potential @xmath247 by replacing @xmath206 in ( [ potential - tortoise ] ) with @xmath248 . the potentials in fig.[fig - potential - tortoise ]look like the half of the ads@xmath0-bubble potentials .especially , we observe that the free field potentials @xmath249 and @xmath250 in fig.[fig - potential - tortoise ] take nearly the same form as in @xmath251 and ( @xmath252 ) in the region of @xmath253 in fig.[fig - potential - afs - tor ] .furthermore the penrose diagram of an exact ads@xmath0(@xmath210 ) is an half of the would - be whole diagram in @xmath254@xcite . in this workwe assume that the potential of an exact ads@xmath0 is the left - hand side one of an ads@xmath0 bubble .the important thing is to calculate the absorption cross section in the background of an exact ads@xmath0 .considering the @xmath255-state picture , it is not hard to calculate the absorption coefficient @xmath243 .although the @xmath256-state can not be defined , we can derive @xmath243 from the backscattering of a test field @xmath15 off @xmath257 . if we choose the boundary condition appropriately ,the of potential fig.[fig - potential - tortoise ] is enough to calculate the absorption coefficient , regardless of the would - be right - hand side(@xmath258 ) . in this backscattering process ,the key point is to use an appropriate matching procedure between a near ads@xmath0 and an asymptotically ads@xmath0 .we remind the reader that @xmath246 is a timelike bounadry and thus information enter or exit from it .this is an exact middle point if one assumes a whole space of @xmath254 .thus requiring the conventional boundary condition may lead to a wrong result in deriving the absorption coefficient . instead of the dirichlet condition of @xmath259 , one may use the non - normalizable modes .the non - normalizable mode is a divergent quantity at @xmath246 but its flux is finite at @xmath246 .also it corresponds to specifying another boundary condition at spatial infinity . on the other hand ,if one use the normalizable modes which satisfy the dirichlet boundary condition , one may not succeed in obtaining the absorption coefficient in an exact ads@xmath0 . in order to calculate the semiclassical absorption cross section, we have to solve the exact differential equation eq.([eq - btz ] ) with an appropriate boundary condition . since it is difficult to solve eq.([eq - btz ] ) directly , one has to use the matching prodecure between the near - horizon(@xmath260 ) ads@xmath0 and the far - region(@xmath261 ) ads@xmath0 .here the matching point resides on @xmath262 in fig.[fig - potential - tortoise ] . in the far region eq.([eq - btz ] ) becomes @xmath263 with a dimensionless variable @xmath264 .one easily finds the far - region solution @xmath265 \label{sol - far}\ ] ] with two unknown constants @xmath266 .the first term is a divergent quantity at @xmath267 but behaves well in the interior region .this corresponds to the non - normalizable modes and is coupled to the boundary operator @xmath12 at infinity .the second one is the normalizable mode and propagates in the bulk .this can be used to construct two , three and four - point functions . although one can not define an ingoing flux at infinity of ads@xmath0, one can calculate the total flux .the flux at spatial infinity is given by @xmath268 we note here that for the tachyon with @xmath195 , @xmath269 .this is because the tachyon takes only the normalizable modes which are to be zero at infinity .thus we exclude it from our analysis . in order to obatin the near - horizon behavior , we introduce the new variable @xmath270 @xmath271 .then eq.([eq - btz ] ) leads to @xmath272 where @xmath273 in the case of the s - wave propagation with @xmath201 , the near - horizon ads@xmath0 equation ( [ eq - hyper ] ) leads exactly to the near - horizon equation ( [ near - eq ] ) of a 5d black hole .explicitly , the relationship between these is given by @xmath274 in this sense , ( [ near - eq ] ) is called an ads@xmath0 bubble .the ingoing wave is given by the hypergeometric function @xmath275 where @xmath276 here @xmath137 means @xmath277 .the coressponding flux is @xmath278 with @xmath279 .the absorption coefficient will be taken as @xmath280 in order to obtain @xmath281 and @xmath282 , we use the matching procedure .it is important to remember that the present spacetime is an exact ads@xmath0 .thus we have to match the near - ads@xmath0 with asymptotically ads@xmath0 to find an absorption coefficient .we know the far - region behavior of ( [ sol - hyper ] ) .this can be found from the @xmath144 for the hypegeometric function @xmath283 , \label{sol - hyp - far}\end{aligned}\ ] ] where @xmath284 matching ( [ sol - far ] ) with ( [ sol - hyp - far ] ) in the far - region(@xmath285 ) leads to @xmath286 considering @xmath287 , one finds @xmath288 even for @xmath42 case .hence we can neglect @xmath282 in favor of @xmath281 .this amounts to taking the flux of the non - normalizable modes .the key point is an ads@xmath0-ads@xmath0 matching in this backscattering process .then the absorption coefficient is approximately given by @xmath289 the absorption cross section for ads@xmath8s@xmath290 with @xmath201 leads to @xmath291 with @xmath292 . in the low energy limit @xmath68, it turns out that the 6d cross sections for an exact ads - theory take the same form as ( [ sigma - phi0 ] ) and ( [ sigma - nu ] ) @xmath293 this 6d result is derived from ads@xmath8s@xmath290 . in order for this to match with the cross section of a 5d black hole , it needs to introduce a compactified circle(s@xmath1 ) in m@xmath5s@xmath6t@xmath7 . in this case one finds @xmath294 with a radius of s@xmath1(@xmath295 ) .we note that @xmath77 , @xmath296 and @xmath46 give us slightly different cross sections in an ads@xmath0 bubble , whereas these(@xmath297 ) do not make any distinction in an exact ads - theory .it seems that the s - matrix can not be extracted from the anti - de sitter space even in a limit@xcite .this is based on the fact that in an exact ads@xmath0 the asymptotic states can not be defined , due to the timelike bounadry and the periodicity of geodesics .however , the authors in @xcite showed that the correlation functions of the dual cft@xmath10 to ads@xmath2 are considered as the bulk s - matrices .the vacuum correlators @xmath298 of the cft@xmath10 are expressed as truncated n - point functions convolved against the non - normalizable modes .these can be interpreted as an s - matrix for an exact ads@xmath2 arising from a limit of scattering from an ads@xmath2 bubble in asymptotically flat space . in this work , we show that the s - matrix of an ads@xmath0 bubble can be derived from an exact ads@xmath0 in the dilute gas and low energy limits .we confirm this from the calculation of the absorption cross section .this originates from the fact that the near - horizon equations for an ads@xmath0 bubble ( [ near - eq ] ) and an exact ads@xmath0 ( [ eq - hyper ] ) are the same form , but they have different boundary conditions at infinity . in the ads@xmath0 bubble - calculation , one uses the ads@xmath0-afs matching to obtain the absorption coefficient @xmath299 . on the other hand , in the exact ads@xmath0-calculation we use both the \{in}-state picture and the non - normalizable mode to obtain the absorption coefficient @xmath243. this amounts to taking the ads@xmath0-ads@xmath0 matching .the s - wave greybody factor of a free scalar of an ads@xmath0 bubble has exactly the same form as that of an exact ads@xmath0 . for the dilatonwe find the same form of cross section @xmath300 but with @xmath301 for an ads@xmath0 bubble and @xmath302 for an exact ads@xmath0 .let us compare our results with the others .the general formular for the gerybody factor is derived from the vacuum two - point function @xmath303 of a boundary operator @xmath12 in the effective string@xcite and boundary cft@xmath304 approaches@xcite .these give us the same result for a free scalar but for the dilaton , @xmath301 as in an ads@xmath0 bubble .consequently , the two - point correlator provides us the greybody factor in the dilute gas and low energy limits .this quantity takes exactly the same form in the cft@xmath304 and ads@xmath0 bubble approaches .further , in the exact ads@xmath0 approach one finds the same form of the greybody factor .this means that the s - matrix can be derived from an exact ads@xmath0 space .it is obvious that the conformal limit of the gauge theory(cft@xmath304 ) corresponds with scattering from an exact ads@xmath0 .finally we present here a scattering picture in an exact ads@xmath0 and compare it with the scattering of an ads@xmath0 bubble in afs .a. strominger and c. vafa , physb379 * , 99 ( 1996 ) , hep - th/9601029 ; c. callan and j. maldacena , nucl . phys . *b472 * , 591 ( 1996 ) , hep - th/9602043 ; g. horowitz and a. strominger , phys . rev .lett . * 77 * , 2368 ( 1996 ) , hep - th/9602051 .a. dhar , g. mandal , and s. wadia , phys . lett . *b388 * , 51 ( 1996 ) , hep - th/9605234 .s. das , g. gibbons , and s. mathur , phys .* 78 * , 417 ( 1997 ) . c. callan , s. gubser , i. klebanov , and a. tseytlin , nuclb489 * , 65 ( 1997 ) , het - th/9610172 ; i.r .klebanov and m. krasnitz , phys . rev .* d55 * , r3250 ( 1997 ) . m. krasnitz and i. klebanov , phys .rev . * d56 * , 2173 ( 1997 ) , hep - th/9703216 ; m. taylor - robinson , hep - th/9704172 ; h. w. lee , y. s. myung , j. y. kim , phys . rev . * d58 * , 104006(1998 ) , hep - th/9708099 .j. maldacena , adv .theor . math .* 2 * , 231(1998 ) , hep - th/9711200 ; s.s .gubser , i.r . klebanov and a.m. polyakov ,. lett . * b428 * , 105(1998 ) , hep - th/9802109 ; e. witten , adv .* 2 * , 253(1998 ) , hep - th/9802150 .s. hyun , hep - th/9704005 ; k. sfetsos and k. skenderis , nucl .* b517 * , 179 ( 1998 ) , hep - th/9711138 ; s. de alwis , phys . lett . *b435 * , 31 ( 1998 ) , hep - th/9804019 .v. balasubramanian , s. giddings , and a. lawrence , jhep * 9903 * , 001 ( 1999 ) , hep - th/9902052 .w. lee and y. s. myung , phys . rev . *d58 * , 104013 ( 1998 ) , hep - th/9804095 ; h.w .lee , n.j .kim , and y.s .myung , hep - th/9805050 .m. banados , m. henneaux , c. teitelboim , and a. zanelli , phys . rev . *d48 * , 1506(1993 ) . j. maldacena and a. strominger , phys . rev . *d56 * , 4975 ( 1997 ) , hep - th/9702015 ; t. regge and j. a. wheeler , phys . rev . * 108 * , 1403 ( 1957 ) ; c. v. vishveshwara , phys . rev . * d1 * , 2870 ( 1970 ) ; f. j. zerilli , phys . rev* 24 * , 737 ( 1970 ) ; o. j. kwon , y. d. kim , y. s. myung , b. h. cho and y. j. park , phys . rev . *d34 * , 333 ( 1986 ) .r. dijkgraaf , h. verlinde and e. verlinde , nucl . phys . *b371 * , 269(1992 ) ; k. sfetsos , nucl . phys . * b389 * , 424(1993 ) ; j. y. kim , h. w. lee and y. s. myung , phys . rev .* d50 * , 3942 ( 1994 ) ; y. s. myung , phys . lett . *b362 * , 46 ( 1995 ) .s. chandrasekhar , _ the mathematical theory of black hole _ ( oxford univ . press , new york , 1983 ) .s.w . hawking , comm .phys . * 43 * , 199(1975 ) ; n. birrell and p. davies , _ quantum fields in curved space _ ( cambridge university press , 1982 , n.y . ) .mathur , nucl .* b514 * , 204(1998 ) , hep - th/9704156 .chan and r.b .mann , phys . rev . *d55 * , 7546(1997 ) . j. maldacena and a. strominger , jhep * 9812 * , 005(1998 ) , hep - th/9804085 .lee , n.j .kim , and y.s .myung , phys .b441 * , 83(1998 ) , hep - th/9803227 .v. balasubramanian , p. krauss , and a. lawrence , phys . rev . *d59 * , 046003 ( 1999 ) , hep - th/9805171 .d. birmingham , i. sachs and s. sen , phys .lett . * b413 * , 281(1997 ) ; h.w .lee , n.j .kim and y.s .myung , phys . rev .* d58 * , 084022(1998 ) , hep - th/9803080 .l. susskind , hep - th/9901079 .j. polchinski , hep - th/9901076 .gubser , phys . rev .* d56 * , 7854(1997 ) .e. teo , phys . lett . *b436 * , 269(1998 ) .

<S> we investigate the close relationship between the potential and absorption cross section for test fields in an ads@xmath0 bubble(a 5d black hole ) and an exact ads@xmath0 . </S><S> there are two solutions in type iib string theory : an ads@xmath0 bubble corresponds to the dilatonic solution , while an exact ads@xmath0 is the non - dilatonic solution . in order to obtain the cross section for an ads@xmath0 bubble </S><S> , we introduce the \{out}-state scattering picture with the ads@xmath0-afs matching procedure . for an exact ads@xmath0 , </S><S> one considers the \{in}-state scattering picture with the ads@xmath0-ads@xmath0 matching . here </S><S> the non - normalizable modes are crucially taken into account for the matching procedure . </S><S> it turns out that the cross sections for the test fields in an ads@xmath0 bubble take the same forms as those in an exact ads@xmath0 . </S><S> this suggests that in the dilute gas and the low energy limits , the s - matrix for an ads@xmath0 bubble can be derived from an exact ads@xmath0 space . </S>

1510.02202

electronic spins associated with color center defects in silicon carbide ( sic ) show promise as a potential component for solid - state quantum technologies due to their combination of long coherence times,@xcite room - temperature operation,@xcite and a host material for which mature growth@xcite and fabrication protocols exist.@xcite the ability to engineer the local photonic environment of optically active solid - state qubits through the use of microfabrication techniques is crucial for their scalable application to the field of quantum information.@xcite in contrast to the most widely studied forms of sic ( 4h and 6h ) , the availability of cubic 3c - sic as a single crystal heteroepitaxial layer on silicon opens up the possibility of combining the favorable properties of sic defect qubits with the fabrication capabilities available in iii - v and silicon - on - insulator semiconductor systems .this may enable on - chip , scalable architectures for generating,@xcite routing,@xcite manipulating,@xcite and detecting@xcite single photon emission from defect qubits in sic .these capabilities can be combined with small mode volume optical cavities @xcite @xcite @xcite @xcite to utilize strong purcell enhancements@xcite @xcite for efficient single photon generation or to use cavity qed protocols to realize a spin - photon interface @xcite as a node in an optically connected ` quantum network'.@xcite @xcite + to date , only modest quality factors ( ` q ' ) of up to 1,550 have been demonstrated for small mode volume ( ` @xmath0 ' ) cavities ( @xmath0 .17ex@xmath1 ) operating at wavelengths relevant for coupling to known defect qubit optical transitions in 3c - sic.@xcite nevertheless , low- and modest - q microcavities can be used to improve defect qubit optical readout and control schemes .more specifically , microcavities can be used to compensate for on - chip power coupling limitations @xcite to facilitate optical coherent control schemes,@xcite generate on - chip frequency conversion,@xcite and improve rates of absorption and fluorescence.@xcite @xcite @xcite in this work , we utilize cavity resonant excitation to increase the photoluminescence ( pl ) count rate , optically detected magnetic resonance ( odmr ) signal intensity , and rate of spin state initialization for ky5 defect spin ensembles incorporated into 3c - sic photonic crystal cavities .furthermore , we use these signal improvements and techniques to study inhomogeneity in the ky5 defect spin and optical properties and extract estimates of the defects sublevel transition rates .+ the ky5 point defect is a color center in 3c - sic that has been previously demonstrated to exhibit spin and optical properties that are similar to the negatively charged nitrogen vacancy ( nv ) center in diamond .@xcite it produces an optical emission band consisting of a zero phonon line ( zpl ) around 1118 nm and a red - shifted phonon sideband extending out to approximately 1300 nm and , like the nv center , its @xmath2 spin ground state can be initialized and read out optically .spin coherence times ( @xmath3 ) in excess of 20 @xmath4s have been demonstrated for the ground - state spin sublevels of ky5 ensembles .these states can be manipulated with resonant microwave pulses on fast time scales ( tens of nanoseconds ) even up to room temperature , making ky5 centers a viable candidate for a defect - based spin qubit .the ky5 center can be controllably generated within the 3c - sic crystal lattice through a combination of radiation damage and subsequent annealing and has been tentatively identified as a neutral divacancy.@xcite due to the availability of 3c - sic as a heteroepitaxial film on silicon , photonic structures with three - dimensional optical confinement can be fabricated by removing the underlying substrate through conventional silicon wet and dry etch processes .the use of on - chip photonic cavities to enhance defect qubit excitation is analogous to the use of macroscopic enhancement cavities to generate large intracavity optical fields for applications such as high sensitivity absorption spectroscopy in atomic gases@xcite or efficient non - linear optical frequency conversion . in general , for a fixed incident excitation power , the local electric field intensity inside a microcavity scales as @xmath5 where @xmath6 is the input power coupling efficiency .more specifically , temporal coupled - mode theory can be used to calculate the response of a photonic crystal cavity subject to far field excitation by an externally incident gaussian beam , yielding the expression : + @xmath7 + where @xmath8 is the cavity electric field intensity maximum , @xmath9 is the incident beam electric field intensity maximum , @xmath10 is the wavelength of the cavity mode , @xmath11 is the refractive index of the medium surrounding the cavity , @xmath12 is the incident beam waist , and @xmath13 is the cavity index of refraction ( .17ex2.64 for 3c - sic).@xcite alternatively , finite - difference time - domain simulations can be used to directly calculate the steady state response of the cavity to continuous wave excitation .figure [ fig : fig1](a ) shows the excitation geometry used for simulating the local field enhancement generated in a 300 nm thick l3 cavity for an incident gaussian electric field intensity with unity amplitude and beam waist equal to our experimentally measured value of .17ex1 @xmath4 m .the resulting simulated electric field intensity enhancement of 193.48 agrees within 1% of the value of calculated from eq .1 using the input parameters extracted from simulations for the l3 cavity design . see @xcite for a comparison of the local field enhancements and cavity parameters for different structure designs and a more detailed discussion of the simulations . + to observe these cavity enhanced field intensities , we measured photonic crystal cavities that were designed to exhibit fundamental modes tuned to the inhomogeneously broadened zpl of the ky5 centers incorporated into the cavity [ around 1118 nm , see fig .[ fig : fig1](b ) ] .the cavities consisted of l3 and h1 structures fabricated in 300 nm thick 3c - sic films that were implanted with carbon ions at an energy of 110 kev and annealed at 750@xmath14 c for 30 minutes in order to generate ensembles of ky5 defects .details of the sample design , fabrication , and characterization are presented in ref .we primarily focused on l3 designs to observe cavity - enhanced optical excitation due to their greater degree of coupling to far field gaussian modes as compared to the h1 cavities , which were better suited for improved narrowband collection of off - resonantly excited defect pl ( 10 - 20 times improvement over unpatterned thin films ) .the measured l3 structures exhibited q .17ex900 with simulated mode volumes of .68 @xmath15 and far field coupling efficiencies to external , free space gaussian modes of 9.6%.@xcite + all measurements were performed in a home - built scanning confocal microscope with an integrated helium flow cryostat at 20k .a 1064 nm diode laser was used for off - resonant excitation and a tunable ( 1090 - 1180 nm ) , narrow linewidth ( < 300 khz ) littman - metcalf diode laser was used for resonant photoluminescence excitation and cross - polarized resonant scattering spectroscopy .samples were mounted with the cavity axis at 45 degrees with respect to the incident laser polarization to allow for the use of the latter technique for control measurements . collected pl andreflected laser light passed back through the objective and were detected using an ingaas ccd array , a low noise femtowatt photoreceiver , or a superconducting nanowire single photon detector ( snspd ) .see @xcite for a detailed description of the experimental apparatus and control measurements .+ we determined the wavelength of the cavity resonances by measuring the pl spectrum of emitters localized within the cavity under off - resonant 1064 nm excitation as shown in fig .[ fig : fig1](b ) .figure [ fig : fig1](b ) also depicts the excitation and pl collection wavelength ranges of the cavity - enhanced photoluminescence excitation ( ceple ) spectroscopy measurement . in order to observe enhanced excitation of the ky5 defects using the cavity mode, we tuned the excitation wavelength to the cavity resonance peak where the light is absorbed by exciting the zpl transitions and collect red - shifted sideband pl . to determine the total pl signal count rate increase with respect to the unpatterned , released thin film , we performed a series of spatial scans over the photonic crystal cavity area and compared the overall cavity pl count rate to that of the surrounding thin film .figure [ fig : fig2](a ) shows a scanning electron microscope ( sem ) image of the l3 cavity and fig .[ fig : fig2 ] ( b ) and ( c ) show a pair of scanning confocal pl images of a 12 @xmath4 m by 12 @xmath4 m l3 photonic crystal cavity with excitation wavelengths as designated in the fig .[ fig : fig2](d ) .the photonic crystal extent is delineated by the white dashed lines and the bright spot in its center is fluorescence originating from the cavity .figure [ fig : fig2](d ) shows the excitation wavelength dependence of the pl count rate originating from the cavity location .the excitation wavelength - dependent count rate matched the cavity mode spectrum and the peak exhibited an approximately factor of 5 increase over off - resonant excitation .figure [ fig : fig2](e ) shows a line cut of the pl map corresponding to the red dashed line in fig .[ fig : fig2](c ) . + at the excitation wavelength corresponding to the cavity resonance , the pl count rate was approximately @xmath16 30 times higher than the pl count rate from the unpatterned thin film , where we have defined @xmath17 as the ratio of the resonantly excited cavity pl count rate to the thin film pl count rate at the same excitation wavelength and power . for h1 designs, we observed a lower maximum @xmath18 13 due to a smaller degree of input coupling for an incident gaussian beam .scattered laser light added a negligible contribution to the measured signal ( see ref .while we excited cavity modes tuned to the ky5 defect zpl wavelength range , this same approach can be applied to applications that require efficient off - resonant excitation@xcite and would be particularly beneficial for excitation wavelengths that overlap weakly with the defects absorption spectrum .+ aside from increasing the overall luminescence count rate , cavity resonant excitation can also be used to improve measurements on single emitters or ensembles by exciting a significantly smaller sample volume than a standard objective lens configuration .the excitation volume of a thin film is greatly reduced as compared to that of bulk material because the excitation volume dimension along the optical axis is set by the thickness of the thin film rather than the diffraction limit ( .17ex10 @xmath4 m for a .7 na objective at @xmath10=1.1 @xmath4 m ) . furthermore, for our optical configuration , the cavity provides a further 12.4-fold reduction in the sample excitation volume as compared to the thin film , resulting in an overall 705-fold reduction as compared to bulk material.@xcite this reduction in the excitation of localized states in the proximity of isolated single emitters within the surrounding material can improve the performance of single photon sources by reducing background fluorescence@xcite or charge - induced spectral diffusion.@xcitethe observed signal improvements provide a means to greatly increase the odmr signal amplitude in order to probe the ky5 center s spin - dependent electronic structure or for sub - diffraction limit , on - chip sensing applications . in order to perform spin - dependent measurements on defects within the cavity structure, we performed an additional fabrication step that adds a 10 nm/300 nm ti / au metallization layer to the sample surface for applying intense local microwave fields to the sample .figure [ fig : fig3 ] ( a ) shows an optical image of an array of released 3c - sic films patterned with photonic crystal cavities next to a 50 @xmath4 m wide microstrip positioned 50 @xmath4 m away from the structures . due to the robustness of the approximately 40 @xmath4 m by 40 @xmath4 m freestanding films , the metallization can be applied prior to or after the membrane release step without the need for critical point drying .we used on - chip microstrips in order to apply sufficiently intense microwave fields to achieve coherent spin manipulation ( rabi oscillations ) on time scales faster than the ky5 defects @xmath19 of .17ex 50 ns.@xcite + figure[ fig : fig3](b ) compares the ky5 odmr signal ( @xmath20 ) with the excitation beam incident on the released 3c - sic thin film ( black line and dots ) and the photonic crystal cavity ( red line and dots ) for the same optical power at zero magnetic field .the overall odmr signal increase matches the pl count rate increase observed for this cavity ( @xmath21 20 ) , confirming that the pl count rate increases under resonant excitation originate from defect pl rather than scattered excitation .similar improvements were observed for pulsed odmr signals , as depicted in fig .[ fig : fig3](c ) . by synchronizing pulsed optical excitation to polarize and readout the spin ensemble with pulsed microwaves for spin manipulation , we observed rabi oscillations between the @xmath22 and @xmath23 spin sublevels of the defects ground states for emitters excited via the cavity mode or in the unpatterned thin film .for pulsed odmr measurements , signal is greatly increased under resonant excitation due to a combination of both higher pl count rates and faster optical pumping .+ ceple spectroscopy on the inhomogeneously broadened zpl of ensembles can greatly increase measurement signal intensities and reduce the number of measured emitters to facilitate studies of the optical and spin inhomogeneity that we observe for defects in 3c - sic thin films.@xcite @xcite @xcite for high - emitter - density samples ( @xmath24 ions @xmath25 implantation dose ) like those measured in the fig .[ fig : fig3 ] , we observed variations in the defect ground - state zero field splitting ( d @xmath26 1.32 ghz ) of up to 10 mhz and variations in the ensemble odmr linewidth by almost a factor of 2 as the excitation wavelength is varied within the inhomogeneously broadened ensemble zpl ( fwhm of 28.2 nm ) . for lower - density samples ( @xmath27 ions @xmath25 dose ) ,the 10 - 20 times improvement of collection efficiency @xcite for defects emitting into the cavity mode reveals pl lines as narrow as 25 ghz within the inhomogeneously broadened zpl ( 28.2 nm fwhm @xmath26 6.7 thz ) .these emission lines were also observed with ceple spectroscopy for significantly lower excitation powers ( 1.5 @xmath4w vs. 3.2 mw ) and higher pl count rates ( .17ex1,400 cts / s vs. .17ex200 cts / s ) as compared to narrowband pl collection of the emission line using a scanning monochromator.@xcite exciting and collecting pl from a reduced mode volume facilitates the isolation of a smaller number of spectral lines within a given wavelength range because this spectral fine structure and its polarization can vary throughout the thin film , likely due to variations in the local crystalline environment .in addition to providing increased signal intensities , the excitation enhancement provided by the resonant cavity mode can also be used to increase the rate of optically - induced spin polarization in the ky5 defects ground state . to observe this process , we measured the defects time - dependent pl intensity in response to variable - length laser pulses with interleaved microwave @xmath28 pulses driving the @xmath22 to @xmath29 ground - state spin sublevel transitions .a series of optical pulses with a variable length from 250 to 1500 ns were applied to excite the ky5 defects from the ground state to the excited state where they experience either spin - conserving radiative recombination or spin - dependent relaxation to the ground - state spin sublevels predominantly via the intersystem crossing ( isc ) transitions.@xcite as a result of this relaxation pathway , optical excitation achieves ground - state spin polarization at a rate that depends on the optical excitation pulse length and field intensity .relaxation through the isc transitions also provides a means for reading out the spin state : for a period of time immediately after the turn - on of the optical pulse , the ky5 defects will fluorescence more or less brightly depending on the ground - state spin polarization immediately prior to the optical pulse . accordingly , we collected the pl for a fixed time period of 100 ns immediately after the turn - on of the optical pulse to obtain a measure of the spin polarization generated by the previous pulse .prior to half of the optical pulses , we applied a microwave @xmath28 pulse to invert the ground - state spin polarization , which was then re - pumped to the steady state polarization by the subsequent optical excitation .we measured this difference in pl ( pl@xmath30pl@xmath31 ) between optical pulses that immediately followed a @xmath28 pulse and those that did not to obtain a measure of the ground - state spin polarization .the optical excitation was turned off for at least 500 ns prior to microwave manipulation and readout to allow for population within the isc levels to fully relax to the ground state .see @xcite for pulse sequences and further details .+ the same model used to explain the pl dynamics of the nv center in diamond can be applied to model the expected dynamics.@xcite figure [ fig : fig4](a ) shows the results of numerical simulations of the time - dependent difference in pl between an nv center that has been initially prepared in the @xmath22 state and the @xmath29 states for two excitation rates differing by a factor of 3 .the figure also depicts the measurement scheme described above and the approximately exponential decay of the pl difference curves for laser pulse lengths longer than the collection time ( red lines ) .this behavior corresponds to an approximately exponentially saturating ground - state spin polarization and pl difference signal .we observed corresponding behavior when comparing the laser pulse length - dependent pl difference signal for ky5 defects excited within the thin film with those excited using the photonic crystal cavity mode , as shown in fig .[ fig : fig4](b ) .figure [ fig : fig4](b ) also shows the results of numerical simulations of this difference signal for varying laser pulse lengths .the observed saturation behavior matches the predicted dynamics for intersystem crossing rates that differ from the nv center , as determined from a nonlinear least squares fit to a numerical model that includes the expected excitation enhancement ( @xmath8/@xmath9 .17ex96 ) for this structure geometry.@xcite the enhanced excitation rate for defects within the cavity structure generates a faster rate of ground - state spin initialization corresponding to a shorter time constant @xmath32 . the time constant associated with this approximately exponential saturation of the pl difference signal decreases from 500 @xmath33 55 ns to 280 @xmath33 20 ns .this approximately factor of 2 difference in the time constant for optical pumping is relatively small as compared to the overall pl count rate increase ( .17ex20x ) because for high excitation rates , the spin initialization rate becomes limited by the transition rate to the intersystem crossing levels rather than the rate of optical cycling . as a control measurement to verify the spin - dependent origin of these dynamics , we removed the microwave @xmath28 pulse and observed no difference signal .@xcite + these measurements provide the first direct observation of spin - dependent pl dynamics in sic and support previous assumptions that these defects exhibit optical dynamics similar to the nv center in diamond .furthermore , we corroborated the variable length laser pulse results using tcspc methods and observed a spin - dependent pulsed odmr contrast of 2.75% . by combining this value with the numerical modeling, we estimate an intrinsic spin - dependent pulsed odmr contrast of 4.9% for ideal measurement parameters( for details , see ref .while this value is lower than typically observed for single nv centers , it is similar to what has been previously observed for high density nv center ensembles and single defects in sic .in conclusion , we have demonstrated resonant excitation of 3c - sic photonic crystal cavities with integrated defect spins for large pl and odmr signal enhancements and increased spin initialization rates .our analysis shows that our present cavity designs are capable of achieving localized optical field intensities that can be enhanced by a factor of almost 200 relative to an incident gaussian beam .this value could be further improved with concurrent optimization of cavity q and coupling to the gaussian excitation mode.@xcite for measurements of ensembles of defects , these small mode volumes and excitation intensity enhancements may facilitate on - chip applications that are limited by inefficient optical coupling or poor spectral overlap of the excitation source and defect absorption bands .these applications include spin ensemble - based sensing techniques using on - chip optical traps,@xcite enhanced absorption for hole - burning experiments,@xcite or for enhancing the signal of spectrally distinct sub - ensembles in order to study inhomogeneous broadening.@xcite for applications involving single defects , cavity resonant excitation can provide enhanced optical stark shifts,@xcite compact , on - chip single photon frequency conversion,@xcite or reduced excitation of background impurities that are detrimental to single photon source performance.@xcite @xcite additionally , we have provided evidence that defects in 3c - sic exhibit photodynamics similar to those of the nv center in diamond .these results underscore the benefits of fabricating devices with integrated defect spin states in heteroepitaxial 3c - sic as a means to incorporate defect qubits into scalable device architectures for applications in the field of quantum information and sensing .+ + we thank david christle , bob buckley , and joerg bochmann for helpful discussions .this work was supported by the afosr qumpass muri fa9550 - 12 - 1 - 004 and nsf dmr-1306300 .a portion of this work was done in the uc santa barbara nanofabrication facility , part of the nsf funded nnin network .we acknowledge support from the center for scientific computing from the cnsi , mrl : an nsf mrsec ( dmr-1121053 ) and nsf cns-0960316 .39ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwosecondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ] + 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty http://dx.doi.org/10.1038/nmat4144 [ * * , ( ) ] http://dx.doi.org/10.1038/nature10562 [ * * , ( ) ] http://dx.doi.org/10.1142/s0129156406004016 [ * * , ( ) ] http://books.google.com/books?id=byhgjumpu_sc[__ ] , emis processing series ( , ) link:\doibase 10.1557/mrs.2013.19 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.110.037402 [ * * , ( ) ] http://stacks.iop.org/1367-2630/13/i=5/a=055025 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.111.213603 [ * * , ( ) ] http://dx.doi.org/10.1038/srep01901 [ * * , ( ) ] link:\doibase 10.1364/ol.39.001768 [ * * , ( ) ] link:\doibase 10.1364/oe.21.032623 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.4906923 [ * * , ( ) ] link:\doibase 10.1021/acs.nanolett.5b02542 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.109.033604 [ * * , ( ) ] http://dx.doi.org/10.1038/ncomms7173 [ * * , ( ) ] http://dx.doi.org/10.1038/nature07127 [ * * , ( ) ] @noop * * , ( ) \doibase http://dx.doi.org/10.1063/1.4890083 [ * * , ( ) ] link:\doibase 10.1364/oe.18.024163 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.3571446 [ * * , ( ) ] link:\doibase 10.1364/oe.17.022689 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.4839235 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.112.160802 [ * * , ( ) ] link:\doibase 10.1103/physrevapplied.3.014006 [ * * , ( ) ] http://dx.doi.org/10.1038/ncomms2854 [ * * , ( ) ]\doibase http://dx.doi.org/10.1016/j.physb.2009.08.124 [ * * , ( ) ]( , ) pp .http://dx.doi.org/10.1038/nphys3291 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.2193811 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.1507838 [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/0022-2313(92)90113-n [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/j.jcrysgro.2013.08.005 [ * * , ( ) ] link:\doibase 10.1103/physrevb.74.104303 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.3337096 [ * * , ( ) ] link:\doibase 10.1364/oe.18.016064 [ * * , ( ) ] link:\doibase 10.1039/c3lc50879j [ * * , ( ) ] http://stacks.iop.org/0022-3719/17/i=8/a=002 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.107.266403 [ * * , ( ) ]

<S> * the identification of new solid - state defect qubit candidates in widely used semiconductors has the potential to enable the use of nanofabricated devices for enhanced qubit measurement and control operations . </S><S> in particular , the recent discovery of optically active spin states in silicon carbide thin films offers a scalable route for incorporating defect qubits into on - chip photonic devices . </S><S> here we demonstrate the use of 3c silicon carbide photonic crystal cavities for enhanced excitation of color center defect spin ensembles in order to increase measured photoluminescence signal count rates , optically detected magnetic resonance signal intensities , and optical spin initialization rates . </S><S> we observe up to a factor of 30 increase in the photoluminescence and odmr signals from ky5 color centers excited by cavity resonant excitation and increase the rate of ground - state spin initialization by approximately a factor of two . </S><S> furthermore , we show that the small excitation mode volume and enhanced excitation and collection efficiencies provided by the structures can be used to study inhomogeneous broadening in defect qubit ensembles . </S><S> these results highlight some of the benefits that nanofabricated devices offer for engineering the local photonic environment of color center defect qubits to enable applications in quantum information and sensing . * </S>

1603.04585

quarkonia are flavorless bound states composed of combinations of quarks and their antiquarks .since discovery of @xmath2 meson in 1974 , many new quarkonia states have been detected .the quarkonia systems consist of charmonium and bottomonium . due to the large massthere are no toponium bound states and no light quark - antiquark states because of the mixture of the light quarks in experiments .the large mass difference between charm and bottom quarks prevents them from mixing .the heavy quark bound states may provide key tools for understanding the interactions between quarks , new hadronic production mechanisms and transitions , the magnitude of the ckm matrix elements and also analyzing the results of heavy - ion experiments .the @xmath3 suppression in ultra relativistic heavy - ion collisions was first suggested as a signal of the formation of a quark - gluon plasma ( qgp ) .however , most recently , attentions have shifted to the bottomonium states due to that they are more massive than charmonium states .the bottom quarks and anti - quarks are relatively rare within the plasma , so the probability for reproduction of the bottomonium states through recombination is much smaller than for charm quarks .consequently , it is expected that the bottomonium system to be a cleaner probe of the qgp than the charmonium system .hence investigations on the properties of bottomonium systems can help us get useful information not only about the nature of the @xmath4 systems , but also on the existence of qgp . since the discovery of quarkonium states , qcd sum rule technique as one of the most powerful nonperturbative tools to hadron physics @xcite has played an important role in understanding the quarkonia spectrum . in order to find missing states one should know their physical properties to develop a successful search strategy .clearly significant progress in understanding of quarkonium production can not be reached without detailed measurements of the cross sections and fractions of the quarkonia .thereby , form factors and decay widths of quarkonia become significant for completing quarkonia spectrum . completingthe bottomonium spectrum is a crucial validation of theoretical calculations and a test of our understanding of bottomonium states in the context of the quark model .bottomonium states are considered as great laboratories to search for the properties od qcd at low energies . in this connection, we investigate the decay properties of tensor @xmath5 meson as one of the important members of the bottomonia to the heavy @xmath6 meson in the present work .the @xmath6 meson with @xmath7 is the only meson consists of two heavy quarks with different flavors . yet ,other possible @xmath6 states ( the scalar , vector , axial - vector and tensor ) have not been observed , however many new @xmath6 species are expected to be produced at the large hadron collider ( lhc ) in the near future .taking into account the two - gluon condensate corrections , the transition form factors of the semileptonic @xmath8 decay channel are calculated within the three - point qcd sum rule .we use the values of transition form factors to estimate the decay rate of the transition under consideration at all lepton channels .the interpolating current of @xmath5 with quantum numbers @xmath9 contains derivatives with respect to space - time .so , we start our calculations in the coordinate space then we apply the fourier transformation to go to the momentum space . to suppress the contributions of the higher states and continuum , we apply a double borel transformation .the article is organized as follows .we derive the qcd sum rules for the transition form factors in section ii .last section is devoted to the numerical analysis of the obtained sum rules , estimation of the decay rates at all lepton channels and concluding remarks .the semileptonic @xmath10 decay is based on @xmath11 transition at quark level whose effective hamiltonian can be written as @xmath12 where @xmath13 is the fermi weak coupling constant and @xmath14 is element of the cabibbo - kobayashi - maskawa ( ckm ) mixing matrix .after sandwiching the effective hamiltonian between the initial and final states , the amplitude of this transition is obtained in terms of transition matrix elements .these matrix elements will be parameterized in terms of transition form factors later . in order to start our calculations , we consider the three - point correlation function @xmath15 where @xmath16 is the time - ordering operator and @xmath17 is the transition current .to proceed we also need the interpolating currents of the initial and final mesons in terms of the quark fields , which are given as @xmath18\ ] ] and @xmath19 where the covariant derivative @xmath20 denotes the four - derivative with respect to @xmath21 acting on two sides simultaneousely and it is defined as @xmath22,\ ] ] with @xmath23 and @xmath24 here , @xmath25 are the gell - mann matrices and @xmath26 denote the external gluon fields . according to the method used , the correlation function in eq.([correl.func .] ) is calculated in two different ways . in physical or phenomenological sidewe obtain it in terms of hadronic parameters such as masses and decay constants . in qcd or theoretical sidewe evaluate it in terms of qcd degrees of freedom like quark masses as well as quark and gluon condensates via operator product expansion ( ope ) .the qcd sum rules for form factors are obtained by equating the above representations to each other . after applying a double borel transformation ,the contributions of the higher states and continuum are suppressed .the hadronic side of the correlation function is obtained by inserting complete sets of intermediate states into eq.([correl.func . ] ) .after performing the four - integrals over @xmath21 and @xmath27 , we get @xmath28 where @xmath29 denotes the contribution of the higher states and continuum . to go further, we need to know the following matrix elements : @xmath30 @xmath31\big\}~,\ ] ] @xmath32 and @xmath33 where @xmath34 , @xmath35 , @xmath36 and @xmath37 are transition form factors , @xmath38 is the polarization tensor associated with the @xmath39 tensor meson ; @xmath40 and @xmath41 are leptonic decay constants of @xmath39 and @xmath6 mesons , respectively , @xmath42 and @xmath43 .putting all matrix elements given in eqs .( [ matrixel1a ] ) , ( [ matrixel1b ] ) , ( [ kayb2current ] ) and ( [ eq : bccurrent ] ) into eq.([phys1 ] ) , the final representation of the correlation function on physical side is obtained as @xmath44q_{\mu}g_{\alpha\beta } \nonumber \\ & - & \frac{2}{3}\big[\delta'b_{+}(q^2)+k(q^2)(\delta+4m_{\chi_{b2}}^2)\big ] p_{\mu}g_{\alpha\beta } \nonumber \\ & - & i(\delta-4m_{\chi_{b2}}^2 ) h(q^2)\varepsilon_{\lambda\eta\beta\mu}p_{\lambda}p_{\eta}q_{\alpha}+\mbox{other structures}\big \}+ ... ,\end{aligned}\ ] ] where @xmath45 and @xmath46 note that we represented only the structures which we will use to find the corresponding form factors .meanwhile , we used the following summation over the polarization tensors to obtain eq .( [ phen2 ] ) @xmath47 where @xmath48 the next step is to calculate the qcd side of the correlation function in deep euclidean region , where @xmath49 and @xmath50 via ope . placing the explicit expressions of the interpolating currents into the correlation function and contracting out all quark pairs via wick s theorem, we obtain @xmath51+\left[\beta\leftrightarrow\alpha\right]\bigg\},\nonumber\\\end{aligned}\ ] ] where @xmath52 is the heavy quark propagator and it is given by @xcite @xmath53 where @xmath54 or @xmath55 quark .replacing the explicit expression of the heavy quark propagators in eq .( [ correl.func.2 ] ) and applying integrals over @xmath21 and @xmath27 , we find the qcd side as @xmath56 here @xmath57 with @xmath58 are the perturbative parts which are expressed in terms of double dispersion integrals as @xmath59 where the spectral densities are defined as @xmath60 $ ] .the spectral densities corresponding to four different structures shown in eq .( [ eq : qcdside ] ) are obtained as @xmath61}{16\pi^2 } , \nonumber \\\rho_2(s , s',q^2)&= & \int_{0}^{1}dx \int_{0}^{1-x}dy~~\frac { 3\big[m_c(3 - 4x-2y)+m_b(-1 + 4x+2y)\big]}{8\pi^2 } , \nonumber \\\rho_3(s , s',q^2)&= & -\int_{0}^{1}dx \int_{0}^{1-x}dy~~\frac{3\big[m_c(1 - 2y)+m_b(1 + 2y)\big]}{8\pi^2}\end{aligned}\ ] ] and @xmath62 the function @xmath63 in nonperturbative parts are calculated from a similiar manner and by considering the two - gluon condensates contributes .having calculated both the physical and ope sides of the correlation function , now , we match them to find qcd sum rules for form factors . in the borel schemewe get @xmath64 + \widehat{{\cal b}}_{m^2}\widehat{{\cal b}}_{m'^2}\pi_1^{nonpert}(q^2)\bigg\ } , \nonumber \\b_{-}(q^2)&=&-\frac{12(m_b+m_c)}{f_{\chi_{b2 } } f_{b_{c}}m_{b_{c}}^2 m_{\chi_{b2}}{\delta'}}e^{{m_{\chi_{b2}}^2}/{m^2 } } e^{{m_{b_{c}}^2}/{m'^2 } } \nonumber \\ & \times&\bigg\{\int^{s_0}_{4m_b^2}ds \int^{s_{0}^{'}}_{(m_b+m_c)^2}ds ' e^{\frac{-s}{m^2}}e^{\frac{-s'}{m'^2}}\rho_2(s , s',q^2)\theta[l(s , s',q^2)]+\widehat{{\cal b}}_{m^2}\widehat{{\cal b}}_{m'^2}\pi_2^{nonpert}(q^2 ) \nonumber\\ & -&\frac{\delta}{\delta'}k(q^2)\bigg\ } , \nonumber \\b_{+}(q^2)&=&-\frac{12(m_b+m_c)}{f_{\chi_{b2}}f_{b_{c}}m_{b_{c}}^2m_{\chi_{b2}}{\delta'}}e^{{m_{\chi_{b2}}^2}/{m^2}}e^{{m_{b_{c}}^2}/{m'^2 } } \nonumber \\ & \times&\bigg\{\int^{s_0}_{4m_b^2}ds \int^{s_{0}^{'}}_{(m_b+m_c)^2}ds ' e^{\frac{-s}{m^2}}e^{\frac{-s'}{m'^2}}\rho_3(s , s',q^2)\theta[l(s , s',q^2)]+\widehat{{\cal b}}_{m^2}\widehat{{\cal b}}_{m'^2}\pi_3^{nonpert}(q^2 ) \nonumber \\ & -&\frac{\delta+4m_{\chi_{b2}}^2}{\delta'}k(q^2)\bigg\ } \nonumber \\\end{aligned}\ ] ] and @xmath65 + \widehat{{\cal b}}_{m^2}\widehat{{\cal b}}_{m'^2}\pi_4^{nonpert}(q^2)\bigg\ } \nonumber \\ & & { } \end{aligned}\ ] ] where @xmath66 and @xmath67 are borel mass parameters ; and @xmath68 and @xmath69 are continuum thresholds in the initial and final channels . here , @xmath70 is the step function and @xmath71 is given by @xmath72 the functions @xmath73 are written as @xmath74 where @xmath75 are very lengthy functions and we do not present their explicit expressions here .to numerically analyze the sum rules obtained for the form factors , we use the meson masses from pdg @xcite .considering the fact that the results of sum rules considerably depend on the quark masses , decay constants and gluon condensate , we use the values of these parameters from different sources .for the quark masses we take into account all the pole values and those obtained at @xmath76 scheme from the refs .@xcite . for @xmath41 , we consider all the values predicted using different methods in refs .@xcite . in the case of @xmath40, we use the only value exists in the litreature , i.e. @xmath77 @xcite . for the gluon condensate , we also use its value from different sources @xcite calculated via different approaches . from the sum rules for the formfactors it is also clear that they contain extra four auxiliary parameters , namely the borel parameters @xmath66 and @xmath67 as well as continuum thresholds @xmath68 and @xmath69 .the general criteria is that the physical quantities like form factors should be independent of these parameters .therefore , we need to determine their working regions such that the form factors weakly depend on these parameters . to find the borel windows , we require that the higher states and continuum contributions are sufficiently suppressed and the perturbative parts exceed the nonperturbative contributions and the series of ope converge . as a result we get the windows : @xmath78 and @xmath79 .the continuum threshold @xmath80 and @xmath69 are not completely arbitrary but they are related to the energy of the first excited states with the same quantum numbers as the interpolating currents of the initial and final channels . in this work the continuum thresholds are chosen in the intervals @xmath81 and @xmath82 . the dependence of form factors @xmath83 and @xmath84 , as examples , on borel parameters @xmath66 and @xmath67 at @xmath85 are plotted in figures [ kmsq ] and [ bplusmsq ] . ) as a function of the borel mass parameter @xmath66 at fixed values of @xmath80 , @xmath86 and @xmath67 . right : k(@xmath85 ) as a function of the borel mass parameter @xmath67 at fixed values of @xmath80 , @xmath86 and @xmath87.,title="fig:",width=302 ] ) as a function of the borel mass parameter @xmath66 at fixed values of @xmath80 , @xmath86 and @xmath67 . right : k(@xmath85 ) as a function of the borel mass parameter @xmath67 at fixed values of @xmath80 , @xmath86 and @xmath87.,title="fig:",width=302 ] as a function of the borel mass parameter @xmath66 at fixed values of @xmath68 , @xmath86 and @xmath67 . right : @xmath88 as a function of the borel mass parameter @xmath67 at fixed values of @xmath68 , @xmath86 and @xmath66.,title="fig:",width=302 ] as a function of the borel mass parameter @xmath66 at fixed values of @xmath68 , @xmath86 and @xmath67 . right : @xmath88 as a function of the borel mass parameter @xmath67 at fixed values of @xmath68 , @xmath86 and @xmath66.,title="fig:",width=302 ] from these figures we see that form factors show overall weak dependence on the borel mass parameters .the behaviors of the form factors @xmath83 , @xmath84 , @xmath89 and @xmath90 in terms of @xmath91 are shown in figures ( 3 - 6 ) . in these figures ,the red triangles show the qcd sum rule predictions , yellow - solid line denotes the prediction of fit function obtained using the central values of the input parameters and the green band shows the uncertainty due to errors of input parameters .note that to obtain the central values , we consider the average values of input parameters discussed above , however , to calculate the uncertainties we consider all errors of these parameters from different sources previously quoted .as it is seen from these figures the sum rules results are truncated at some points . hence , to enlarge the region to whole physical region we need to find some fit functions such that their results coincide well with the qcd sum rules predictions at reliable regions .for this reason we show the @xmath91 dependence of form factors including both the sum rules and fit results in figures ( 3 - 6 ) .our numerical calculations reveal that the following fit function well defines the form factors under consideration : @xmath92,\ ] ] where the values of the parameters , @xmath93 , @xmath94 and @xmath95 obtained at @xmath96 and @xmath97 for @xmath98 transition are presented in table [ formfactors ] .our final purpose in this section is to obtain the decay width of the @xmath98 transition at all lepton channels .the differential decay width for this transition is obtained as @xmath99 @xmath100 \nonumber\\ & + & 2re[k(q^2)b^*_{+}(q^2)]\delta'\bigg[-q^4+m_{b_{c}}^2(m_{l}^2+q^2)-m_{\chi_{b2}}^2(m_{l}^2+q^2)\bigg ] \nonumber\\ & -&2re[b_{-}(q^2)b^*_{+}(q^2)]\delta'm_{l}^2q^2(m^2_{b_{c}}-m^2_{\chi_{b2 } } ) \nonumber\\ & + & |k(q^2)|^2\bigg[m_{b_{c}}^4(m_{l}^2+q^2)+m_{\chi_{b2}}^4(m_{l}^2+q^2)+q^4(m_{l}^2+q^2)-2m_{b_{c}}^2 \nonumber\\ & \times&(m_{\chi_{b2}}^2+q^2)(m_{l}^2+q^2)+m_{\chi_{b2}}^2q^2(m_{l}^2 + 5q^2)\bigg ] \nonumber\\ & + & 3|h(q^2)|^2\delta'm_{\chi_{b2}}^2q^2(m_{l}^2+q^2)-2re[k(q^2)b^*_{-}(q^2)]\delta'm_{l}^2q^2\bigg\}.\end{aligned}\ ] ] after performing integration over @xmath91 in eq.([decaywidth ] ) in the interval @xmath101 , we obtain the total decay widths as presented in table [ numdecaywidth ] for different leptons .@xmath102 the errors belong to the uncertainties coming from the determination of the working regions for auxiliary parameters as well as those of the other input parameters .the orders of decay rates at all lepton channels show that these transitions are accessible at lhc in near future . in summary, we have calculated the transition form factors for the semileptonic @xmath98 transition using qcd sum rule technique .we took into account the two - gluon condensate contributions as nonperturbative effects .we used these form factors to estimate the order of decay widths at all lepton channels .the order of decay width reveal that these transitions can be seen at lhc in near future .any comparison of the experimental results with our predictions can provide us with essential knowledge on the nature of the tensor @xmath5 state . ) as a function of @xmath91 at @xmath103 , @xmath104 , @xmath96 and @xmath105 .the red triangles show the qcd sum rule predictions , yellow - solid line denotes the prediction of fit function obtained using the central values of the input parameters and the green band shows the uncertainty due to errors of input parameters ., width=377 ] but for h(@xmath91 ) ., width=377 ] but for @xmath36 . ,width=377 ] but for @xmath37 ., width=377 ]the authors declare that there is no conflict of interest regarding the publication of this paper .l. j. reinders , h. rubinstein , s. yazaki , hadron properties from qcd sum rules " , phys .rept . * 127 * , 1 ( 1985 ) .k. a. olive et al .( particle data group ) , review of particle physics " , chin .c * 38 * , 090001 ( 2014 ) .h. abramowicz et al .[ h1 and zeus collaborations ] , `` combination and qcd analysis of charm production cross section measurements in deep - inelastic ep scattering at hera '' , eur.phys.j .c * 73 * , 2311 ( 2013 ) , arxiv:1211.1182 [ hep - ex ] .h. abramowicz et al .[ zeus collaboration ] , `` measurement of beauty and charm production in deep inelastic scattering at hera and measurement of the beauty - quark mass '' , jhep * 1409 * , 127 ( 2014 ) , arxiv:1405.6915 [ hep - ex ] .s. narison , summary on @xmath106 and precise @xmath107 from heavy - light qcd spectral sum rules " , nucl .b - proceedings supplements * 234 * , 187 ( 2013 ) , 16th inter . conf . in qcd , arxiv:1209.2925 [ hep - ph ] .n. carrasco et al .[ european twisted mass collaboration ] , `` up , down , strange and charm quark masses with @xmath108 twisted mass lattice qcd '' , nucl .b * 887 * , 19 ( 2014 ) , arxiv:1403.4504 [ hep - lat ] .a. ali , a.y .parkhomenko , a.v .rusov,precise calculation of the dilepton invariant - mass spectrum and the decay rate in @xmath111 in the sm " , phys .d * 89 * , 094021 ( 2014 ) , arxiv:1312.2523 [ hep - ph ] .p. colangelo , a. khodjamirian , `` qcd sum rules , a modern perspective '' , at the frontier of particle physics / handbook of qcd , edited by m. shifman ( world scientific , singapore ) , 3 , 1495 , ( 2001 ) , arxiv : hep - ph/0010175 .m. j. baker , j. bordes , c. a. dominguez , j. penarrocha , k. schilcher , b meson decay constants @xmath113 , @xmath114 and @xmath115 from qcd sum rules " , jhep * 1407 * , 032 ( 2014 ) arxiv:1310.0941 [ hep - ph ] .b. colquhoun et al . , b - meson decay constants : a more complete picture from full lattice qcd " , phys .d * 91 * , 114509 ( 2015 ) , arxiv:1503.05762 [ hep - lat ]. s. narison , improved @xmath116 and @xmath41 from qcd laplace sum rules " , int .phys . a * 30 * , 20 , 1550116 ( 2015 ) , arxiv:1404.6642 [ hep - ph ] .e. bagan , h. g. dosch , p. gosdzinsky , s. narison and j. m. richard , hadrons with charm and beauty " , z. phys .c * 64 * , 57 ( 1994 ) , arxiv : hep - ph/9403208 .b. l. ioffe , qcd at low energies " , prog .phys . * 56 * , 232 ( 2006 ) , arxiv : hep - ph/ 0502148 . c. a. dominguez ,l. a. hernandez and k. schilcher , determination of the gluon condensate from data in the charm - quark region " , jhep * 1507 * , 110 ( 2015 ) , arxiv:1411.4500 [ hep - ph ] .r. horsley et al . , wilson loops to 20th order numerical stochastic perturbation theory " , phys .d * 86 * , 054502 ( 2012 ) , arxiv:1205.1659 [ hep - lat ] .b. chakraborty et al . , high - precision quark masses and qcd coupling from @xmath118 lattice qcd " , phys .d * 91 * , 054508 ( 2015 ) , arxiv:1408.4169 [ hep - lat ] . c. a. dominguez , l. a. hernandez , k. schilcher and h. spiesberger , chiral sum rules and vacuum condensates from tau - lepton decay data " , jhep * 1503 * , 053 ( 2015 ) , arxiv:1410.3779 [ hep - ph ] . b. v. geshkenbein , calculation of gluon and four - quark condensates from the operator expansion " , phys .d * 70 * , 074027 ( 2004 ) , arxiv : hep - ph/0309122. h. g. dosch , m. jamin , s. narison , baryon masses and flavour symmetry breaking of chiral condensates " , phys .b * 220 * , 251 ( 1989 ) .

<S> the form factors of the semileptonic @xmath0 decay are calculated using the qcd sum rule approach . </S><S> the results obtained are then used to estimate the decay widths of this transition in all lepton channels . </S><S> the orders of decay rates indicate that this transition is accessible at lhc for all lepton channels . # 1#2#3 @xmath1 # </S><S> 1#2#3 @xmath1 0 ^ 0 5_5 o _ ^0 _ pacs number(s):11.55.hx , 14.40.pq , 13.20.he </S>

1206.5951

there is currently much interest for electromagnetically induced transparency ( eit ) in rydberg atoms @xcite , in particular for the cooperative atom - light and photon - photon dynamics found in rydberg systems @xcite and for their possible application , e.g. , in quantum information processing @xcite , in the production of correlated photon pairs @xcite , and in metrology @xcite .the present work is motivated by a recent experimental study of eit in a 3-level rydberg system with radio frequency ( rf ) modulation @xcite . in this experiment ,the absorption on the @xmath0@xmath1 transition of @xmath2rb was probed by a weak laser field in the presence of both an rf field and a strong laser field resonant with the transition between the intermediate @xmath3 state and the highly polarizable @xmath4 state .the rf field was too low in frequency for driving multiphoton transitions between the lower , intermediate or upper states and other bound states .however , it was strong enough to stark - shift the upper state periodically by tens of mhz , which resulted into the eit features in the probe absorption spectrum shifting and splitting into multiple eit sidebands .the existence of absorption sidebands is a well known feature of ac - modulated 2-level systems @xcite , including 2-level atomic or molecular systems @xcite .the effect of an oscillating electric field on molecular absorption lines with quadratic stark shift was described by townes and merritt in 1947 @xcite : lines whose frequency width is much larger than the frequency of modulation adiabatically follow the oscillations of the field i.e. , a state of dipole polarizability @xmath5 in a non - resonant ac field @xmath6 is simply displaced in energy by @xmath7 ^ 2/2 $ ] at time @xmath8 .however , lines whose frequency width is of the same order or smaller than the modulation frequency split into manifolds of components displaced in energy from their zero - field limit by @xmath9 , @xmath10 , and the intensity of these components varies with @xmath11 , @xmath12 and @xmath13 like @xmath14 .analogous shifts and sidebands are also found in systems governed by an hamiltonian unitarily equivalent to that governing ac - modulated 2-level atomic or molecular systems , such as ac - driven superconducting flux qubits @xcite .interactions with other fields will in general introduce additional @xmath11-dependent stark shifts . for this reason ,the separation between adjacent sidebands of an ac modulated 2-level system is not exactly @xmath15 and the intensities of the absorption lines do not exactly follow a @xmath14 law . the correct separations and scalings can be obtained by combining floquet theory and van vleck perturbation theory @xcite . however , these deviations from the results of townes and merritt are significant only when the interaction coupling the two states of the system is sufficiently strong . perhaps for this reason , they do not appear to have been previously considered in an atomic and molecular physics context @xcite .in many ac - modulated 2-level atomic systems of practical interest , one of the two states is much more polarizable than the other and the frequency widths of both the state and the coupling laser field are small compared to the modulation frequency . in these conditions, it may be tempting to ignore the manifold structure of the less polarizable state , retain that of the more polarizable state , and model the ac - modulated two - level system as a many - level system in which the less polarizable state interacts with a manifold of independent sideband states .we refer to this approximate description as the @xmath16-level approximation .it leads to the results described by townes and merritt @xcite in the limit of a weak interaction between the states and makes it possible to calculate the absorption spectrum rapidly at a minimal cost .however , as is shown below , this approximation breaks down for strong coupling between the two states : in order to obtain the correct results , the sideband structure of _ both _ states must be properly taken into account .this can be done , e.g. , by carrying out a full floquet analysis of the ac - modulated 2-level system @xcite , or by solving the optical bloch equations with time - dependent modulation .the sidebands of the less polarizable state here originate from the coupling of this state to the floquet manifold formed by the other state , not from a direct effect of the ac - modulation .nonetheless , as is illustrated below , they may manifest by well defined , additional resonance structures in absorption spectra .we refer to these as induced sidebands . in this paper, we investigate the changes in the sideband structure of the absorption spectrum arising from the interaction with the coupling laser field , both in two - level and three - level systems .the absorption spectrum is calculated by solving the optical bloch equations and is interpreted by reference to the floquet quasienergy spectrum .the @xmath16-level approximation is compared to the exact results throughout the paper . for simplicity ,we first focus on ac - modulated two - level systems ( sec .[ sec:3 ] ) . the coupling with a lower state is then introduced .the probe absorption in an rf - modulated 3-state ladder system and the appearance of induced sidebands is studied in sec .[ sec:4 ] .in particular , we give examples of sidebands shift in eit spectra for both cold and thermal ensembles . we conclude with a summary of our main findings ( sec .[ sec:5 ] ) .we use atomic units throughout the rest of this paper , except where indicated otherwise .we first concentrate on the rf dynamics of the subsystem formed by the intermediate and upper states of our 3-level ladder rydberg system , states @xmath17 and @xmath18 .we thus assume , for the time being , that neither @xmath17 nor @xmath18 interact with the lower state of the system , @xmath19 . making the rotating wave approximation for the laser field and passing to slowly - varying variableswould then reduce the time - dependent schrdinger equation to @xmath20 if the rf field was absent . @xmath21 and @xmath22 are respectively the rabi frequency @xcite ( which we assume real ) and the detuning of the control laser field coupling the states @xmath17 and @xmath18 in the ladder system : denoting the zero - field energies of these two states by @xmath23 and @xmath24 and the angular frequency of the control laser by @xmath25 , @xmath26 with @xmath27 .the state vector of the system , @xmath28 , is a linear superposition of the bare states @xmath17 and @xmath18 and the functions @xmath29 and @xmath30 are the respective probability amplitudes : @xmath31 we write the electric field component of the rf field as @xmath32 we assume that the angular frequency @xmath33 is so much smaller than @xmath34 , and so much smaller than the transition frequencies between @xmath17 and @xmath18 and any other unperturbed state , that @xmath17 and @xmath18 evolve adiabatically under the effect of this field .hence , we replace the time - dependent bare states @xmath35 and @xmath36 by the adiabatic states @xcite @xmath37\,{d}t ' \right ) |b[{\cal e}(t)]\rangle\ ] ] and @xmath38\,{d}t ' \right ) |c[{\cal e}(t)]\rangle.\ ] ] in these expressions , @xmath39\rangle$ ] and @xmath40\rangle$ ] denote the state vectors which develop adiabatically from the unperturbed states @xmath17 and @xmath18 when a static electric field is turned on fromn 0 to @xmath41 , and @xmath42 $ ] and @xmath43 $ ] are the corresponding eigenenergies of the stark hamiltonian .the rf field is assumed to be weak enough that , in sufficiently good approximation , @xmath44=w_b-\alpha_b [ { \cal e}(t)]^2/2 $ ] and @xmath45=w_c-\alpha_c [ { \cal e}(t)]^2/2 $ ] , and moreover that @xmath46\rangle= @xmath47\rangle= @xmath48 and @xmath49 are defined by first order perturbation theory , and @xmath50 and @xmath51 are the static dipole polarizability of the states @xmath17 and @xmath18 . for the applications we have in mind , @xmath52 , so that one can take @xmath44\approx w_b$ ] and @xmath46\rangle \approx @xmath18 is significantly perturbed by the rf field .moreover , we are also considering systems for which @xmath17 and @xmath49 have the same parity and therefore are not directly coupled to each other by the control laser field . in these conditions , the time - dependent schrdinger equation for the rf - dressed system becomes @xmath53\end{pmatrix } \begin{pmatrix}c_{b}\\c_{c}\end{pmatrix}\label{eq : ht}\ ] ] where @xmath54 eq .( [ eq : ht ] ) is a system of linear differential equations with periodic coefficients , which is amenable to the floquet description studied in the following section .the effect of relaxation , which is not taken into account in eq .( [ eq : ht ] ) , will be considered in sec.[subsec:3b ] . by virtue of the floquet theorem @xcite , any solution of eq .( [ eq : ht ] ) can be written as a superposition of fundamental solutions of the form [ eq : fsol ] @xmath55 where the quasienergy @xmath56 and the coefficients @xmath57 and @xmath58 are constants .( the fundamental angular frequency is @xmath59 , not @xmath33 , because the time - dependence entirely arises from the periodic stark shift of the upper level , which varies at twice the rf frequency . )replacing @xmath29 and @xmath30 by @xmath60 and @xmath61 turns eq .( [ eq : ht ] ) into a time - independent system of algebraic equations for the coefficients @xmath57 and @xmath58 , namely [ eq : ceq ] c_c , n-2n_c_b , n & = _ c_b , n,[eq : ceq1 ] + ( _ c-_-2n_)c_c , n+c_b , n & + -(c_c , n-1+c_c , n+1 ) & = _ c_c , n.[eq : ceq2 ] the quasienergies are those values of @xmath56 for which this system has a non - trivial solution . in matrix form , eqs .( [ eq : ceq1 ] ) and ( [ eq : ceq2 ] ) read @xmath62 where @xmath63 is the column vector @xmath64 and @xmath65 the possible values of the quasienergy are the eigenvalues of @xmath66 , which we denote by @xmath67 .each one corresponds to a particular dressed state of the system , i.e. , to a solution of the schrdinger equation of the form @xmath68 where the state vectors @xmath69 are time - independent .it follows from the above that @xmath70 .we thus have two equivalent descriptions of the dressed states of the system , namely one in terms of the time - dependent coefficients @xmath71 and @xmath72 and one in terms of the time - independent vector @xmath73 formed by the harmonic components of @xmath74 .( here and in the following , we use the notation @xmath75 to represent a column vector of state vectors . )the vector @xmath76 , solution of eq .( [ eq : mf ] ) with @xmath77 , is the representation of @xmath78 in the basis of the bare floquet states @xmath79 and @xmath80 thus @xcite @xmath81 the hamiltonian matrix @xmath66 has a four - block structure .its upper and lower diagonal blocks , @xmath82 and @xmath83 respectively , are square matrices representing the hamiltonians of the floquet manifolds spawned by the bare states @xmath17 and @xmath18 .the non - zero elements of the off - diagonal blocks arise from the coupling of the states @xmath17 and @xmath18 by the control field , while the off - diagonal elements @xmath84 of @xmath83 arise from the coupling of the state @xmath18 with itself via the absorption or stimulated emission of two rf photons .the rf field thus mixes the different vectors @xmath85 with each other .by contrast , the matrix @xmath82 is diagonal since we neglect the dressing of the state @xmath17 by the rf field .it is useful to diagonalize the matrix @xmath83 , which can be done by transforming the basis from @xmath86 to @xmath87 , with @xmath88 and @xmath89 where @xmath90 is the @xmath91-th order bessel function of the first kind @xcite . under this change of basis ,the matrix representing the hamiltonian becomes @xmath92 where @xmath93 , @xmath94 , denotes @xmath95 .neither the quasienergies nor the diagonal elements of the hamiltonian matrix are affected by this change of basis .we see that in the absence of the control field , i.e. , for @xmath96 , the vectors @xmath97 and @xmath98 describe the floquet sidebands of the states @xmath17 and @xmath18 , respectively . when @xmath99 , the sidebands of @xmath100 are coupled to those of @xmath101 by the off - diagonal blocks of @xmath102 , the strength of the coupling between the @xmath11-th side band of @xmath100 and the @xmath103-th side band of @xmath101 being proportional to @xmath104 @xcite .( a ) the manifold structure of the dressed 2-level system .the energy separation between adjacent sidebands is @xmath59 .the energy of state @xmath18 is stark shifted by the rf field by an amount @xmath105 .the effective detuning from resonance of the laser field coupling the two states is thus @xmath106 .( b ) the manifold structure in the @xmath16-level approximation : the sidebands of state @xmath17 are neglected and the interaction between the two manifolds reduces to the interaction of the single state @xmath17 with the @xmath16 states forming the @xmath101 manifold . ,width=321 ] @xmath66 and @xmath102 reduce to the same diagonal matrix in the absence of control field , i.e. , for @xmath107 .it follows from eq .( [ eq : mhf2 ] ) that their spectrum is a double comb of quasienergies : when @xmath107 , any quasienergy @xmath108 is equal either to @xmath109 or to @xmath110 for some value of @xmath11 , where @xmath111 the corresponding dressed states , @xmath78 , are either @xmath97 or @xmath112 , respectively .each of the quasienergies is thus associated with an energy sideband of either state @xmath17 or state @xmath18 .the sidebands of @xmath17 are located at the energies @xmath113 , @xmath114 , and those of @xmath18 at the energies @xmath115 , @xmath114 see fig .[ fig : fig2](a ) . as should be expected when @xmath107 , the sideband energies @xmath116 and @xmath117 do not depend on the frequency of the control laser .however , the quasienergies of the @xmath110-manifold of quasienergies change with the detuning @xmath118 , with the consequence that this manifold crosses the ( @xmath119-independent ) @xmath109-manifold at @xmath120 , @xmath121 ( color online ) the quasienergy spectrum , ( a ) and ( b ) , and the average population in the upper state at large times ( @xmath122 , as defined in section [ subsec:3b ] ) , ( c ) , in an rf - dressed 2-level system .@xmath123 mhz , @xmath124 mhz , @xmath125 mhz , @xmath126 mhz and @xmath127 .( a ) and dark pink curve in ( c ) : exact results .( b ) and light purple curve in ( c ) : results obtained in the @xmath16-level approximation .maxima in the population transferred to the upper state occur at the detunings at which the quasienergy curves cross or anticross each other .differences in the positions and amplitudes of these maxima between the exact results and the @xmath16-level approximation are visible in ( c ) .the crosses represent one of the quasienergy curves predicted by eqs .( [ eq : qepn1 ] ) and ( [ eq : qepn2 ] ) ., width=321 ] the quasienergy spectrum keeps that double comb structure for @xmath128 : any quasienergy @xmath108 is equal either to @xmath129 or to @xmath130 for some value of @xmath11 , with @xmath131 @xmath132 and @xmath133 being such that @xmath134 and @xmath135 in the limit @xmath136 .thus @xmath137 and @xmath138 if the control field is sufficiently weak. however , when @xmath99 , the @xmath129-manifold and the @xmath130-manifold are no longer degenerate at @xmath120 , @xmath121 instead , as illustrated by fig .[ fig : fig3](a ) , these two manifolds anticross each other in sequences of avoided crossings . the gaps and positions of these avoided crossing are determined by the off - diagonal blocks of the hamiltonian matrix @xmath102 .if @xmath139 , i.e. , if the rf field is vanishingly weak , @xmath140 , and these off - diagonal blocks are the unit matrix multiplied by @xmath141 . in this case , the system is equivalent to an infinite set of non - interacting two - level systems .one then has @xmath142 where the upper signs apply for @xmath143 and the lower signs for @xmath144 . the states interchange character at @xmath145 .however , since in the absence of the rf field floquet harmonic components of different @xmath11 values are not coupled to each other , the crossings between the @xmath100 and the @xmath101 manifolds at @xmath146 ( @xmath147 ) are true crossings rather than avoided crossings .the off - diagonal elements of the off - diagonal blocks are normally non - zero if the rf field is not vanishingly weak , but they rapidly decrease away from the diagonal when @xmath148 since @xmath149 then goes rapidly to zero when @xmath150 increases . if one neglects these off - diagonal elements , eqs .( [ eq : qep1 ] ) and ( [ eq : qep2 ] ) then generalize to @xmath151 and @xmath152 , with _ b , n^(1)&= ( _ b , n^(0)+_c , n^(0 ) ) + & = ( _ c- _ ) + 2 n _ [ eq : qenern1 ] and _ c , n^(1)&= ( _ b , n^(0)+_c , n^(0 ) ) + & = ( _ c- _ ) + 2 n_. [ eq : qenern2 ] the upper signs apply for @xmath153 and the lower signs for @xmath154 .the control laser field thus lifts the degeneracy between the states @xmath155 and @xmath156 at @xmath157 and creates an energy gap approximately equal to @xmath158 between @xmath129 and @xmath130 . in fig .[ fig : fig3](a ) , this interaction results in conspicuous avoided crossings at @xmath159 mhz .the two manifolds formed by the quasienergy curves @xmath160 cross each other at the detunings at which @xmath161 , @xmath162 , i.e. , at @xmath163 with @xmath164 therefore these true crossings do not occur exactly at @xmath165 , as would be the case if @xmath96 , but instead at detunings shifted towards the avoided crossings at @xmath166 . expanding the square root function to lowest order in @xmath167 , eq .( [ eq : sideband ] ) gives @xmath168 this result can also be obtained using perturbation theory .the degeneracy of the @xmath160 and @xmath169 quasienergies at @xmath170 is lifted by the off - diagonal terms of the off - diagonal blocks of the hamiltonian matrix @xmath102 .for instance , the avoided crossings visible at @xmath171 mhz and 30 mhz in fig .[ fig : fig3](a ) originate from the terms @xmath172 in these off - diagonal blocks , which couple the @xmath97 states to the @xmath173 states . treating each pair of intersecting quasienergies as an isolated two - level system in the vicinity of their intersection and neglecting the terms in @xmath174 with @xmath175 in @xmath102 yields the improved approximation @xmath176 and @xmath177 , with @xmath178 and @xmath179 as in the above , the signsshould be chosen so that @xmath180 and @xmath181 for @xmath136 .one of the resulting quasienergy curves is represented by crosses in fig .[ fig : fig3](a ) .as seen from the figure , it reproduces the exact quasienergies well .the only significant differences occur when @xmath182 with @xmath183 , where the approximate quasienergy manifolds @xmath184 and @xmath185 cross rather than anticross each other. the approximation can be improved by taking into account more off - diagonal components of the off - diagonal blocks of @xmath102 and further iterating the process leading to eqs .( [ eq : qepn1 ] ) and ( [ eq : qepn2 ] ) .the effective rabi frequency for two quasienergy curves intersecting at @xmath182 is @xmath186 .the gap of the corresponding avoided crossing is therefore a decreasing function of @xmath103 when , as we have been assuming throughout this discussion , @xmath187 . treating the crossings at @xmath188 separately from that at @xmath189is justified if the former are sufficiently far apart from the latter , namely if @xmath190 .given that @xmath191 when @xmath192 , this condition can also be written @xmath193 the iterative method outlined above fails when the inequality ( [ eq : cond1 ] ) is violated . the crossings between the manifoldsmay then occur at different values of @xmath118 than predicted by eqs .( [ eq : sideband ] ) and ( [ eq : sideband2 ] ) .finally , we note that for weak fields the quasienergies obtained by the iterative method are consistent with those predicted by perturbation theory , namely , for @xmath194 @xcite , @xmath195 as seen above , the bare state @xmath17 turns into a manifold of sideband states when the system is described in the floquet formalism .the different states of this manifold are not directly coupled to each other by the rf field under our assumption that the dressing of state @xmath17 by this field is negligible ; however , they interact with each other indirectly , through their coupling with the harmonic components of state @xmath18 by the control laser field . neglecting the manifold structure of state @xmath17amounts to setting @xmath196 for @xmath197 in eq .( [ eq : fsol1 ] ) and to reducing the hamiltonian matrix @xmath198 to the matrix @xmath199 we refer to this approximation as the @xmath16-level approximation , in view of the fact that the matrix @xmath200 is effectively the hamiltonian of a system consisting of a single state @xmath201 interacting with a manifold of @xmath16 independent states @xmath112 , @xmath202 see fig .[ fig : fig2](b ) . as mentioned in the introduction, this approximation has often been made in studies of the interaction of low - lying states with rf - dressed rydberg states .the corresponding quasienergies are the eigenvalues @xmath203 of the matrix @xmath204 .an example of the resulting spectrum is shown in fig .[ fig : fig3](b ) . proceeding as in sec .[ subsubsec:3a2 ] , one finds that for @xmath96 the corresponding quasienergies are equal either to @xmath205 or to @xmath206 for some value of @xmath11 , whereas for @xmath99 and @xmath187 , they are approximately equal either to @xmath207 , to @xmath208 , or to @xmath209 . in the latter case ,the @xmath101-manifold intersects the @xmath100-quasienergy curve at values of @xmath118 for which @xmath210 for some value of @xmath11. however , these intersections normally occur at different detunings than found in sec .[ subsubsec:3a2 ] , since in general @xmath211 when @xmath197 . instead , the crossings occur at @xmath212 where @xmath213 comparing this result with eq .( [ eq : sideband2 ] ) , we see that within the @xmath16-level approximation the shift of the crossings from their zero-@xmath167 positions is about half the correct value @xcite .being entirely based on eq .( [ eq : ht ] ) , the analysis developed in sec .[ subsec:3a ] neglects both the frequency width of the control laser and the energy width of the states @xmath17 and @xmath18 .the corresponding time evolution of the system is purely oscillatory , with population transferred to - and - fro between @xmath17 and @xmath18 without any damping .the evolution would be a pure rabi flopping of characteristic angular frequency @xmath214 in the absence of the rf field .the oscillation is multimode in the presence of the latter , with fourier components separated in angular frequency by integer multiples of @xmath59 .( [ eq : qenern1 ] ) and ( [ eq : qenern2 ] ) approximately give the corresponding characteristic frequencies as @xmath215^{1/2}/2 + 2p\omega_{\rm rf}$ ] , @xmath216 relaxation modifies this picture . taking into account the natural width of state @xmath18 , @xmath217 , and the linewidth of the control laser , @xmath218 , and assuming that @xmath18 can de - excite only to @xmath17 , the time evolution of the systemis described by the following optical bloch equations : _ bb= & _ c_cc+ ( _ bc-_cb),[eq : hrho1 ] + _ cc=&-_c_cc+ ( _ cb-_bc),[eq : hrho2 ] + _ bc=&-_c/2+_c+i_bc + & + ( _ bb-_cc).[eq : hrho3 ] ( recall that at the moment we study the system formed by states @xmath17 and @xmath18 in isolation from state @xmath19 . we therefore assume , for the time being , that @xmath17 does not de - excite to a lower state .the finite lifetime of state @xmath17 will be taken into account in the analysis of the full 3-level ladder system , in sec.[sec:4 ] . )decoherence dampens the dynamical evolution of the system until the oscillation is purely driven by the rf field . in this steady state , which is reached for @xmath122 , the elements of the density matrix oscillate at integer multiples of the fundamental angular frequency @xmath219 about a constant cycle - average . in the @xmath16-level approximation, we treat each side band of state @xmath18 as an independent bound state and describe the evolution of the system by optical bloch equations based on the hamiltonian matrix @xmath204 , namely _ bb= & _ c_k _ kk+ _ k ( j_k_bk - j_k_kb),[eq : hrhon1 ] + _ bk= & -_bk + & + ( _ bb-_jj_j_jk),[eq : hrhon2 ] + _ jk= & -[_c-2i(j - k)_rf]_jk + ( j_k_jb - j_j_bk),[eq : hrhon3 ] with the indexes @xmath220 and @xmath221 running over all the states forming the @xmath101-manifold . in this approach ,the population in state @xmath18 , @xmath222 , is defined as @xmath223 , or equivalently , as @xmath224 .in contrast to eqs .( [ eq : hrho1][eq : hrho3 ] ) , which predict oscillatory populations and coherences , eqs .( [ eq : hrhon1][eq : hrhon3 ] ) lead to a constant density matrix in the long time limit .we assume that the atom is initially in state @xmath17 and calculate the population transferred to state @xmath18 , @xmath225 , and its temporal average in the steady state , @xmath226 .we define the latter as @xmath227 with @xmath228 .we thus set @xmath229 to 1 and all the other elements of the density matrix to zero at time @xmath230 and numerically solve either eqs .( [ eq : hrho1][eq : hrho3 ] ) or eqs .( [ eq : hrhon1][eq : hrhon3 ] ) from @xmath230 to @xmath122 . how the population transferred to state @xmath18 varies with the detuning @xmath22 depends both on the value of @xmath231 and on whether the inequality ( [ eq : cond1 ] ) is fulfilled or not. avoided crossings between quasienergy curves can be treated in isolation when this inequality is fulfilled , and adjacent resonances do not overlap significantly due to the natural width of the upper level when @xmath232 .the variation of @xmath226 in these conditions is examplified by fig .[ fig : fig3](c ) .the results represented respectively by a dark pink curve and by a light purple curve were obtained by solving eqs .( [ eq : hrho1][eq : hrho3 ] ) and eqs .( [ eq : hrhon1][eq : hrhon3 ] ) . as seen from the figure , the avoided crossings in the dressed state spectrum are associated with strong enhancements of population transfer .the origin of these enhancements is easily understood given that @xmath217 and @xmath218 are both sufficiently small that the evolution of the system is dominated by its dressed states dynamics rather than by relaxation .for the parameters considered , the dressed atom can be described as a simple two - state system in the vicinity of each avoided crossing . sufficiently far from any crossing ,the dressed states are close to either one of the @xmath97 states or one of the @xmath112 states , with little admixture between the two . at a crossing , however , the interacting dressed states are approximately equal superpositions of @xmath17 and @xmath18 .population transfer is therefore enhanced at each crossing , typically over the range of values of @xmath22 for which the relevant @xmath100- and @xmath101-dressed states are significantly admixed with each other . treating the dressed atom as a pure two - state system in the vicinity of eachcrossing leads to a simple expression for the population transferred to the upper state , i.e. , for @xmath233 @xcite , @xmath234+\omega_{\text{c}}^2j_{n}^2}.\ ] ] thus @xmath226 exhibits power - broadened sidebands whose height and width are modulated by the square of the bessel function factors @xmath235 .when @xmath236 is larger than the argument of these bessel functions , these enhancements decrease in amplitude and their width tends to the natural width of the upper level , @xmath217 , for increasing sideband orders . for the parameters of fig .[ fig : fig3](c ) , this analysis predicts that the @xmath194 , @xmath237 and @xmath238 enhancements have an heigth of 0.50 , 0.46 and 0.08 , respectively , and a full width at half maximum of 13.3 , 3.6 and 1.1 mhz . as can be seen from the figure ( dark purple curve ) , these values are in agreement with the numerical results , although the @xmath194 sideband somewhat overlaps the others .( the inequality ( [ eq : cond1 ] ) is fulfilled but marginally here , since @xmath239 . )the picture is the same in the @xmath16-level approximation for the parameters considered .the discrepancy in the position of the enhancements mirrors the differences in the position of the crossings in the quasienergy maps .it can be largely eliminated by a simple @xmath221-dependent shift in the @xmath240 factor multiplying @xmath241 in eq .( [ eq : hrhon2 ] ) @xcite .the comparison shows that for the parameters considered in fig .[ fig : fig3 ] , the upper state effectively turns into a manifold of states well described within the @xmath16-level approximation . as mentioned in sec .[ subsubsec:3a2 ] , increasing @xmath167 eventually results in a break down of this approximation as the states forming the floquet manifold will start to interact with each other . for sufficiently strong control fields ,the resonance sidebands overlap each other and merge with the zeroth order resonance .( color online ) the population in the upper state at @xmath242 ( first row ) , @xmath243 ( second row ) and @xmath244 ( third row ) , and its time average ( bottom row ) , in the steady state regime ( @xmath122 ) .@xmath245 mhz and @xmath127 in all cases .( a ) : @xmath246 mhz , @xmath247 mhz and @xmath248 mhz .( b ) : @xmath249 mhz , @xmath247 mhz and @xmath248 mhz .( c ) : @xmath249 mhz , @xmath250 mhz and @xmath248 mhz .( d ) : @xmath249 mhz , @xmath250 mhz and @xmath251 mhz .@xmath245 mhz and @xmath127 in all cases .@xmath222 is multiplied by a factor of 0.025 in column ( b ) ., width=396 ] how large @xmath252 is relative to @xmath219 also impacts on the temporal evolution of the system . when @xmath253 , the atom interacts with the laser field as if state @xmath18 were a manifold of stationary states . in the opposite limit, the rf field changes slowly on the time scale on which the states @xmath17 and @xmath18 interact with each other .this interaction then occurs as if the rf field was static at any instant . in this case, the stark shift due to the rf field is expected to make the population in the state @xmath18 oscillate periodically .columns ( a ) and ( b ) of fig .[ fig : fig4 ] compare instantaneous and cycle - average values of @xmath222 in the steady state regime for two different values of @xmath167 . in column ( a ) ,this value is insufficiently small for the @xmath16-state picture to hold well , which leads to a clear oscillation in the population in the state @xmath18 . in column ( b ) , @xmath255 , which fulfills the condition ( [ eq : cond2 ] ) : @xmath222 is almost constant in time .the resonance sidebands are also much better marked than in column ( a ) because of the much smaller power broadening .the population in the state @xmath18 also depends on the time scale of the relaxation mechanisms .well defined enhancement sidebands are not expected unless spontaneous decay is slow compared to the oscillation of the rf field , i.e. , unless @xcite @xmath256 this inequality also expresses the requirement that adjacent sidebands , which are separated roughly by @xmath219 , do not overlap due to the the natural linewidth of the resonance , @xmath257 .the role of the lifetime of the state @xmath18 is illustrated by column ( c ) of fig .[ fig : fig4 ] , which was calculated for the same parameters as column ( b ) apart for @xmath257 : while @xmath258 in column ( b ) , @xmath259 in column ( c ) .in the latter case the sidebands partly overlap each other and the population in the state @xmath18 oscillates markedly .finally , column ( d ) of fig .[ fig : fig4 ] shows the time evolution of @xmath222 for the same parameters as in column ( c ) apart that @xmath59 is much smaller than both @xmath231 and @xmath22 .the @xmath16-level picture is now invalid and there is no sign of sidebands . instead, the profile of @xmath222 is a single peak of width @xmath217 following the instantaneous position of state @xmath18 as this state is periodically stark - shifted by the rf field @xcite .( a ) schematic of the rf - dressed three - level rydberg system .the probe laser ( angular frequency @xmath260 ) couples the states @xmath261 and @xmath262 , while the control laser ( angular frequency @xmath119 ) couples the states @xmath262 and @xmath263 .the energy of the latter is modulated by the rf field ( angular frequency @xmath33 ) .the detunings of these two laser fields from the respective transition frequencies are @xmath264 and @xmath118 .the state @xmath18 decays to the state @xmath17 , and the state @xmath17 to the state @xmath19 .( b ) the manifold structure of the states @xmath17 and @xmath18 induced by the rf field .the energy separation between adjacent sidebands is @xmath219 .the rf field also shifts the energy of state @xmath263 by an amount @xmath105 ; the effective detuning of the control field is thus @xmath265 ., width=321 ] we now take into account the coupling of state @xmath17 to state @xmath19 by the probe laser field and allow state @xmath17 to de - excite to state @xmath19 , which is assumed to be stable ( fig .[ fig : fig1 ] ) .we proceed as in sec .[ subsec:3a0 ] and , ignoring relaxation at first , write the state vector of the atom as @xmath266 with the amplitudes @xmath267 , @xmath29 , @xmath30 satisfying the equation @xmath268\end{pmatrix}\begin{pmatrix}c_{a}\\c_{b}\\c_{c}\end{pmatrix}.\ ] ] following the same analysis as in sec . [ subsec:3a ] , we formulate the problem in terms of the floquet hamiltonian @xmath269 where a= & ( , 2_,0,-2_, ),[eq : ch3fa ] + _ = & ( , , , , ) , [ eq : ch3fb ] and @xmath270 generalizing the discussion of last section ,the corresponding floquet states are described by time - independent vectors @xmath78 which can be written in terms of the vectors @xmath271 and of the vectors @xmath272 and @xmath85 defined by eqs .( [ eq : repb ] ) and ( [ eq : repc ] ) : @xmath273 as before , the @xmath274-block of @xmath275 can be diagonalized by a change of basis transforming the matrix @xmath66 into the matrix @xmath102 defined by eq .( [ eq : mhf2 ] ) .the change transforms the matrix @xmath275 into the matrix @xmath276 differing from @xmath275 by the replacement of @xmath66 by @xmath102 .it does not affect the spectrum of quasienergy .making the @xmath16-level approximation further reduces @xmath276 to @xmath277 the optical bloch equations for the 3-state system with rf - modulation can be written as [ eq : emotion ] _ aa=&_b_bb+(_ab-_ba),[eq : emotion1 ] + _ bb=&-_b_bb+_c_cc-(_ab-_ba ) + & + ( _ bc-_cb),[eq : emotion2 ] + _ cc=&-_c_cc-(_bc-_cb),[eq : emotion3 ] + _ ab=&-(i_++_p)_ab + & -(_bb-_aa)+_ac,[eq : emotion4 ] + _ bc=&-(i_-2i_c^2_t++_p+_c)_bc + -&(_cc-_bb)-_ac,[eq : emotion5 ] + _ ac=&-(i_p + i_c -2i_c^2_t++_c)_ac + & + _ ab-_bc.[eq : emotion6 ] as in sec .[ sec:3 ] , @xmath231 and @xmath218 denote the natural width of state @xmath18 and the linewidth of the control laser . @xmath278 and @xmath279 , which were not introduced in sec .[ sec:3 ] , are the corresponding quantities for state @xmath17 and the probe laser ( fig .[ fig : fig1 ] ) .correspondingly , the optical bloch equations in the @xmath16-level approximation now read _ aa= & _ b _ bb + ( _ ab-_ba ) , [ eq : hrhon31 ] + _ bb= & -_b _ bb - ( _ ab-_ba ) + & + _ c_k _ kk+ _ k ( j_k_bk - j_k_kb),[eq : hrhon32 ] + _ ab= & -(_b 2+_p+ i _ ) _ ab+ ( _ aa-_bb ) + & + _ k j_k _ ak , [ eq : hrhon33 ] + _ ak= & -_ak + & -_bk + _ ab , [ eq : hrhon34 ] + _ bk= & -_bk + & + ( _ bb-_j j_j_jk ) -_ak , [ eq : hrhon35 ] + _ jk= & -[_c-2i(j - k)_rf]_jk + ( j_k_jb - j_j_bk),[eq : hrhon36 ] with the indexes @xmath220 and @xmath221 running over all the components of the floquet manifold of state @xmath18 .these equations are to the hamiltonian @xmath280 what eqs .( [ eq : emotion1][eq : emotion6 ] ) are to the hamiltonians @xmath281 and @xmath282 . as in the 2-state case , the solutions of eqs .( [ eq : hrhon31][eq : hrhon36 ] ) tend to constants in the long time limit , whereas those of eqs.([eq : emotion1][eq : emotion6 ] ) tend to a steady state in which the elements of the density matrix oscillate at a fundamental angular frequency @xmath283 . in either case, the absorption coefficient for the probe beam is given by the equation @xmath284,\ ] ] where @xmath285 is the wave number of the probe field , @xmath286 is the number density of the atoms forming the medium , @xmath287 is the matrix element of the dipole operator between the states @xmath19 and @xmath17 , @xmath288 is the polarization vector of the probe field , @xmath289 is the permittivity of free space , and @xmath290 denotes the average value of @xmath291 in the stationary regime . for a cold atomic ensemble , @xmath292 , with @xmath293 for a thermal ensemble wherethe doppler effect must be taken into account , @xmath290 is the average of @xmath294 over the maxwellian distribution of velocity of the atoms .( color online ) the quasienergy spectrum , ( a ) and ( b ) , and the absorption profile of the probe field , ( c ) , in an rf - dressed 3-level system .@xmath295 mhz , @xmath296 mhz , @xmath297 mhz , @xmath298 mhz , @xmath299 mhz , @xmath300 mhz , @xmath301 mhz and @xmath302 mhz .( a ) and thick dark blue curve in ( c ) : exact results .( b ) and thin light pink curve in ( c ) : results in the @xmath16-level approximation . in ( a ), the crosses identify @xmath132 and the circles @xmath133 , @xmath303 and @xmath304 .the other oblique lines correspond to @xmath129 , @xmath305 .the horizontal line corresponds to @xmath306 .one may note small differences in the position of the crossings between ( a ) and ( b ) . because of the significant stark shift of the upper state , the control laser is off resonance although @xmath298 , which explains why the absorption profile of the probe field does not exhibit eit dips ., width=321 ] a representative section of the quasienergy spectrum of the full floquet hamiltonian is shown in fig .[ fig : fig5](a ) for a case where the control laser is sufficiently weak that the inequality ( [ eq : cond1 ] ) is fulfilled .the horizontal line interrupted by avoided crossings indicates the quasienergy of the zeroth order floquet sideband of state @xmath19 .the oblique lines arise either from the sidebands of the state @xmath17 or from those of the state @xmath18 , as indicated in the caption .the @xmath100- and @xmath101-quasienergy manifolds run parallel to each other in fig .[ fig : fig5](a ) , although they intersect in fig .[ fig : fig3](a ) , because the parameter varied in fig .[ fig : fig5 ] is the detuning of the probe field while that varied in fig .[ fig : fig3 ] is the detuning of the control field .the latter is actually set to zero in these calculations , which means , in view of the stark shift of state @xmath18 , that the control laser is off resonance .the quasienergy spectrum obtained in the @xmath16-level approximation is shown in fig .[ fig : fig5](b ) , for comparison . the absorption profile for a cold atom ensembleis given in fig .[ fig : fig5](c ) .( the quantity represented in fig .[ fig : fig5](c ) is @xmath307 , which is proportional to the absorption coefficient if the atomic motion can be neglected . the maxima of @xmath307 correspond to maxima of absorption . )the dressed states crossing the horizontal lines all contain the bare state @xmath17 i.e. , at least one of the @xmath57 coefficients is non - zero in eq .( [ eq : newpsi ] ) .therefore each of these crossings corresponds to a resonant coupling between this state and the bare state @xmath19 and may give rise to an enhancement of absorption of the probe field coupling @xmath19 to @xmath17 .[ fig : fig5](c ) shows a large enhancement of absorption at the crossing involving the zeroth order sideband of state @xmath17 ( at @xmath308 ) , and smaller enhancements at the other crossings .the enhancement at the crossings with the @xmath309 sideband of @xmath18 ( at 85 mhz ) is hardly visible , though , and the enhancements associated with the crossings of the @xmath309 and @xmath310 sidebands of @xmath17 ( at 50 and 100 mhz , respectively ) are too small to be seen .these enhancements are weaker compared to those involving the zeroth - order sidebands both because of the scaling of @xmath21 by the @xmath235 factors ( @xmath311 is much smaller than 1 when @xmath197 for the parameters of fig .[ fig : fig5 ] ) and because of the larger energy difference between the respective sidebands ( a larger energy difference means a smaller admixture of @xmath17 into the sidebands of @xmath18 ) .however , as will be seen shortly , significant enhancements may occur at the crossings with the @xmath305 sidebands of @xmath100 for stronger control fields . comparing fig .[ fig : fig5](a ) to fig .[ fig : fig5](b ) and the thick blue line to the thin pink line in fig .[ fig : fig5](c ) , we see that the @xmath16-level approximation holds well in the present case , as could have been expected since the condition ( [ eq : model ] ) is fulfilled . because the side bands of the @xmath100-state are neglected in the @xmath16-level approximation , the @xmath100-quasienergy manifold of fig .[ fig : fig5](a ) reduces to a single quasienergy curve in fig .[ fig : fig5](b ) .however , the @xmath101-manifold is practically the same in the two spectra .the absorption profile is also very well reproduced in the @xmath16-level approximation for the parameters considered : the two sets of results are almost identical apart for very small differences in the position of the enhancements ( noticeable only for the enhancement at @xmath312 mhz on the scale of the figure ) and very small differences in their amplitude .we now increase @xmath313 to 30 and 60 mhz , keeping the other parameters the same as in fig .[ fig : fig5 ] .the resulting absorption profiles are shown in figs .[ fig : fig6](a ) and [ fig : fig6](b ) , respectively .as expected from the previous discussion , the @xmath16-level approximation deteriorates as the strength of the control field increases .it is still in rough agreement with the exact result in the case of fig .[ fig : fig6](a ) , for which @xmath21 is smaller , but not much smaller , than @xmath283 , although there are differences in the position and the amplitudes of the enhancements .however , when the condition ( [ eq : model ] ) is more strongly violated , the @xmath16-level approximation breaks down completely .there is little agreement with the exact result in fig .[ fig : fig6](b ) . an interesting feature of fig . [ fig : fig6](a ) is the presence of a small enhancement of absorption at @xmath314 mhz .this enhancement is not present in the @xmath16-level approximation .it coincides with the crossing of @xmath19 with the @xmath309 sideband of @xmath17 in the quasienergy spectrum , at @xmath315 above the main absorption peak ( which is concomitant with the crossing of the 0-th order sideband of @xmath17 . )this feature is more prominent in fig . [fig : fig6](b ) , at @xmath316 mhz ; the @xmath310 sideband of @xmath17 is also ( barely ) visible , at about 80 mhz at this higher power of the control laser .we refer to such sidebands as induced sidebands " , as they arise from sidebands of @xmath17 induced by the coupling of this state with the sidebands of @xmath18 by the control field .( recall that the rf field is assumed to be too weak to dress state @xmath17 directly : the @xmath100-manifold originates from the coupling of @xmath17 to the @xmath101-manifold by the control field , not from the coupling of @xmath17 to itself by the rf field . ) that the sidebands of @xmath19 play no role in this absorption profile is shown by the good agreement between the exact results ( thick blue curves ) and the results obtained when these sidebands are neglected altogether ( green dashed curves ) . before closing, we briefly comment on whether the conditions eq . ( [ eq : cond1 ] ) and ( [ eq : cond2 ] ) are sufficient to garantee the validity of the @xmath16-level approximation for a three - state system .recall that these conditions mean that the oscillations of the rf field are faster than the time scales over which the system evolves under the effect of the control laser field and the spontaneous decay of state @xmath18 .however , the probe field introduces an additional time scale in the problem : if this field is excessively strong , it will disturb the dynamics of the @xmath100-@xmath101 system too rapidly for the state @xmath18 to show a manifold structure . in this case, the floquet sidebands of state @xmath19 induced by its coupling to state @xmath18 can no longer be ignored .given that the effective rabi frequency for the coupling of @xmath19 to @xmath18 is approximately given by @xmath317 @xcite , we expect that the condition @xmath318 ) and ( [ eq : cond2 ] ) for the @xmath16-level approximation to be valid .should the inequality ( [ eq : cond3 ] ) not be fulfilled , then it is likely that the floquet sidebands of @xmath19 need to be taken into account , as well as those of @xmath17 and @xmath18 .( color online ) the probe absorption in the same rf - dressed 3-level system as in fig .[ fig : fig5 ] but for stronger control fields for which the @xmath16-level approximation is inaccurate .@xmath319 mhz in ( a ) and 60 mhz in ( b ) .thick blue solid curves : results obtained by solving the full optical bloch equations , eqs .( [ eq : emotion1][eq : emotion6 ] ) .thin pink solid curves : results obtained in the @xmath16-level approximation , eqs .( [ eq : hrhon31][eq : hrhon36 ] ) . dashed green curves : results obtained by adding the sidebands of state @xmath17 , but not those of state @xmath19 , to the @xmath16-level approximation .the enhancements associated with crossings between the lower state and the intermediate state manifold are identified by the letter @xmath100 between the lower state and the rydberg state manifold by the letter @xmath101 . , width=321 ] ( color online ) probe absorption vs. control laser detuning in an rf - dressed 3-level system , for three different control rabi frequencies. the parameters of the system are the same as in figs .[ fig : fig5 ] and [ fig : fig6 ] , but here the probe field is on - resonance ( @xmath320 ) and @xmath22 varies .@xmath313 is 8 mhz in ( a ) , 30 mhz in ( b ) and 60 mhz in ( c ) .thick blue solid curves : results obtained by solving the full optical bloch equations , eqs .( [ eq : emotion1][eq : emotion6 ] ) .thin pink solid curves : results obtained in the @xmath16-level approximation , eqs .( [ eq : hrhon31][eq : hrhon36 ] ) . dotted green curves : absorption profile without rf field ( the energy of the rydberg state was shifted by @xmath105 to facilitate the comparison with the other results ) .as @xmath21 increases , the eit transparency window in ( a ) expands into the sidebands region and the eit absorption dips change into absorption peaks .the width of the features is modulated by the square of the bessel function factors @xmath235 ., width=321 ] we still assume that the doppler effect is negligible , as in the last section .however , we now set @xmath321 to 0 and vary @xmath22 rather than set @xmath22 to 0 and vary @xmath321 .if the rf field was absent , the control laser field would couple the states @xmath17 and @xmath18 resonantly at @xmath322 , which would be accompanied by a dip in im@xmath323 $ ] ) .( this is the well known eit absorption window at @xmath320 @xcite . ) in the presence of the rf field , this dip is displaced to @xmath324 , due to the ac stark shift of @xmath18 , and splits into multiple sidebands .eit with rf modulation is illustrated by fig .[ fig : fig7 ] .these results were calculated for the same values of @xmath325 , @xmath105 , @xmath33 , @xmath278 , @xmath231 , @xmath279 and @xmath218 as in figs .[ fig : fig4 ] and [ fig : fig5 ] .the probe rabi frequency , @xmath326 , increases from 8 mhz in fig .[ fig : fig7](a ) to 30 mhz in fig .[ fig : fig7](b ) and to 60 mhz in fig .[ fig : fig7](c ) .as seen from the figure , the rf field creates transmission sidebands on either side of the main eit feature .since @xmath321 is set to zero and the frequency of the control field is varied , these sidebands occur at the values of @xmath22 at which the @xmath101-quasienergy manifold intersects the zeroth - order quasienergy sideband of the state @xmath17 . in the weak control case of fig .[ fig : fig7](a ) , they manifest as narrow dips regularly spaced by @xmath59 @xcite . apart for minor differences in the depth , the @xmath16-state approximation reproduces these features very well ( the thick blue curve to the thin pink curve ) .the absorption profile is thus well explained by the model in which the upper state simply splits into a comb of floquet states , each one interacting with the intermediate state independently from the others .however , this picture changes when @xmath21 approaches or exceeds @xmath59 see , respectively , fig .[ fig : fig7](b ) and ( c ) : the eit dips broaden and shift towards the zeroth - order sideband when @xmath21 increases , and the absorption profile first acquires a peak - and - trough structure and then changes into a series of absorption peaks ( rather than absorption dips ) superimposed on a slowly varying background .these peaks occur at the detunings at which state @xmath19 is resonant with the dressed state formed by the states @xmath17 and @xmath18 coupled by the control field .the same changes are observed in the @xmath16-level approximation , at least for the parameters considered , but the position of the absorption features is more and more incorrect .the green dotted curves represent the absorption profile calculated without the sidebands of state @xmath18 ( but taking its stark shift into account ) . comparing these results to the thick blue curve in fig .[ fig : fig7](c ) shows that the suppression of eit in narrow ranges of frequencies at large control powers arises from the interplay between these sidebands .( color online ) the probe absorption in the same rf - dressed 3-level system as in fig .[ fig : fig7 ] but here with doppler averaging and plotted against probe detuning .the probe and control laser beams are assumed to be colinear and counterpropagating .@xmath313 is 8 mhz in ( a ) and 30 mhz in ( b ) .@xmath322 in both graphs .thick green curves : results obtained by solving the full optical bloch equations , eqs .( [ eq : emotion1][eq : emotion6 ] ) .thin black curves : results obtained in the @xmath16-level approximation , eqs .( [ eq : hrhon31][eq : hrhon36 ] ) . the higher power of the control field in ( b ) results in a broadening and a shift of the eit dips and to inaccuracies in the @xmath16-level approximation . ,width=321 ] let us assume that the two laser beams are colinear and counterpropagating .compared to an atom at rest , the probe and control angular frequencies for an atom moving with a velocity component @xmath327 in the direction of the control laser beam are shifted respectively by @xmath328 and @xmath329 , where @xmath330 and @xmath331 are the wave numbers in the laboratory frame . in the reference frame of the atom , the probe detuning is therefore upshifted to @xmath332 and the control detuning downshifted to @xmath333 , where @xmath264 and @xmath118 are the detunings in the laboratory frame .the absorption of the probe beam at @xmath264 and @xmath118 will thus be determined by the velocity average of the steady - state coherence @xmath334 calculated at the doppler - shifted detunings .we denote this average by @xmath335 . for atoms of mass @xmath103 at a temperature @xmath336 , @xmath337 with @xmath338 where @xmath339 is the boltzmann constant .[ fig : fig8 ] shows the probe absorption calculated using the @xmath16-level approximation ( thin black curves ) or calculated directly from eqs .( [ eq : emotion1][eq : emotion6 ] ) ( thick green curves ) , as a function of the probe detuning , for two different values of @xmath167 .the other parameters are the same as those used in fig .[ fig : fig5 ] . in the doppler average, we assume a cloud of @xmath340rb atoms at a temperature of 40 @xmath341c , and probe and control wavelengths of 780 and 480 nm , respectively .we see that the @xmath16-level approximation is accurate in the case of figure ( a ) , for which the control power is relatively low . for small values of @xmath167, each absorption minimum corresponds to eit in the velocity class which simultaneously satisfies the two conditions @xmath342 and @xmath343 . for a fixed value of @xmath118, the corresponding absorption dips occur at the probe detunings @xmath344 adjacent eit dips are thus separated in probe detuning by @xmath345 in these conditions .increasing @xmath21 both widens the eip dips through power broadening and shifts their positions .the latter effect is due to the shift in the sidebands of @xmath18 arising from their coupling with state @xmath17 .when this shift is non - negligible , the second of the above resonance conditions must be replaced by @xmath346 , where @xmath347 is an @xmath21-dependent shift , and the absorption dips occur at the probe detunings @xmath348 as seen from the figure , the @xmath16-level approximation becomes inaccurate when @xmath349 is increased from 8 to 30 mhz : at the higher control power there is a clear difference both in the position and in the depth of the eit dips between the absorption profiles calculated with and without making this approximation .( correcting the @xmath16-level approximation for the sidebands of the intermediate state , still neglecting the sidebands of the lower states , restores the agreement with the exact results . )to conclude , we have investigated the absorption spectrum of two- and three - level model atomic systems with ac - modulated quadratic stark shift , both in the weak coupling regime and beyond this regime . although we specifically consider the case of a rydberg system modulated by a radio - frequency field , our results are generic .the different regimes investigated can be defined in terms of the modulation amplitude of the upper state energy , which is characterized by the quantity @xmath105 defined by eq .( [ eq : starksigma ] ) , and in terms of the angular frequency of the rf field , @xmath33 , of the rabi frequency of the control field , @xmath21 , and of the natural width of the upper state , @xmath217 .when the conditions @xmath350 , @xmath351 and @xmath352 are all met , the quasienergy curve crossings in the floquet spectrum ( and the associated structures in the absorption spectrum ) can be treated in isolation of each other .the system then behaves as if the most polarizable state is effectively a manifold of sideband states and the approximation we refer to as the @xmath16-level approximation is accurate .this approximation amounts to neglecting the manifold structure of the less polarizable states and model the ac modulated system as a many - level system in which these states interact with a manifold of independent sideband states spawned by the most polarizable state .increasing the strength of the control field to @xmath353 shifts the position of the absorption sidebands and changes their amplitude , in agreement with the perturbative analysis developed in a different context in refs.@xcite and @xcite and with the iterative approach outlined in sec .[ subsubsec:3a2 ] .the @xmath16-level approximation becomes inaccurate as @xmath354 increases , in particular in predicting sidebands shifts half too small .however , this approximation can be brought into agreement with the correct positions of the sidebands by a simple change in the optical bloch equations ( or the corresponding hamiltonian ) . in the strong coupling regime , where @xmath355 , the @xmath16-level approximation fails unless it is corrected by including the intermediate - state floquet manifold in the model . for sufficiently large values of @xmath354 , the probe absorption spectrum in a 3-level ladder system develops sidebands induced by the coupling of the intermediate state to the polarizable upper state , in addition to the sidebands manifesting in the weak coupling regime . for certain combinations of the parameters ,this spectrum changes from one exhibiting sharp eit dips to one exhibiting sharp absorption peaks as @xmath354 increases from @xmath356 to @xmath355 .we thank c. s. adams for useful discussions at the early stages of this work and for his useful comments on the manuscript .we also thank the epsrc and the dpst programme of the thai government for financial support .these departures do not manifest in the results reported in ref .however , measurements in the same ladder eit system than in ref . @xcitebut for higher rydberg states and lower rf frequencies found sidebands separations differing from the usual @xmath15 spacing ( m. bason and c. s. adams , unpublished ) . formally , and ignoring relaxation .the system formed by an atom exposed to both a laser field and an rf field can be described within the generalized multi - frequency floquet theory [ s .- i chu , adv .phys . * 73 * , 739 ( 1989 ) ; m. drr , r. m. potvliege , d. proulx , and r. shakeshaft , phys . rev . a * 44 * , 574 ( 1991 ) ] . however , this general approach would be unnecessarily complicated here , since for the intensities and frequencies of interest the atom can be accurately represented as a two - level system with a periodic perturbation of the energy of one of the two states. the signs in eqs .( [ eq : repb ] ) , ( [ eq : repb ] ) and ( [ eq : psiexp ] ) are chosen in such a way that the eigenstates of @xmath357 belonging to the zero-@xmath21 quasienergies @xmath109 and @xmath110 defined in sec .[ subsubsec:3a2 ] are , respectively , @xmath358 and @xmath359 , not @xmath360 and @xmath361 .perturbation theory gives @xmath362 with @xmath363 $ ] .since @xmath364 is approximately half the correct shift , the @xmath16-level approximation can be improved by replacing @xmath365 by @xmath366 in the diagonal elements of the matrix @xmath200 , or equivalently by replacing @xmath367 by @xmath368 in eqs .( [ eq : hrhon1][eq : hrhon3 ] ) .

<S> the effect of an ac modulation on a 2-level or 3-level system is studied theoretically . </S><S> the absorption spectrum is calculated by solving the optical bloch equations and is interpreted by reference to the floquet quasienergy spectrum . </S><S> the dependence of the absorption sidebands on the intensity of the coupling laser field in rydberg systems submitted to a radio - frequency ( rf ) field is analysed in detail . </S><S> it is shown that for sufficiently strong coupling fields additional sidebands appear in the probe absorption spectrum in ladder 3-level systems . </S><S> these additional sidebands are induced by the coupling of the intermediate state to the floquet manifold spawned by the upper state under rf modulation . </S>

1509.08872

microphases supersede simple gas - liquid coexistence when short - range inter - particle attraction is frustrated by long - range repulsion ( salr ) .the resulting structures are both elegant and remarkably useful @xcite .block copolymers @xcite , for instance , form a rich array of periodic structures , such as lamellae , gyroid @xcite and exotic morphologies @xcite , whose robust assembly enables industrial applications in drug delivery @xcite and nanoscale patterning @xcite , among others . because microphase formation constitutes a universality class of sort @xcite , many other systems either exhibit or share the potential to form similar assemblies @xcite . in the latter category ,colloidal suspensions are particularly interesting .the relative ease with which interactions between colloids can be tuned indeed suggests that a broad array of ordered microphases should be achievable @xcite . yet , in experiments @xcite only amorphous gels and clusters have been observed in systems ranging from proteins @xcite to micron - scale beads @xcite . a variety of explanations have been advanced to explain the difficulty of assembling periodic microphases in colloids , including a glass - like dynamical slowdown upon approaching the microphase regime @xcite , the existence of an equilibrium gel phase @xcite , and the dynamical arrest of partly assembled structures due either to particle scale sluggishness @xcite or competition between morphologies @xcite . in order to obtain a clearer physical picture of these effects and thus hopefully guide experimental microphase ordering , a better understanding of the relationship between equilibrium statics and dynamics is needed .insights from theory and simulation would be beneficial , but both approaches face serious challenges . on the one hand ,theoretical descriptions , such as density - functional theory @xcite , self - consistent field theory @xcite , random - phase approximation @xcite and others @xcite , capture reasonably well the microphase structures , but corresponding dynamical descriptions are more limited @xcite . on the other hand ,the dynamics of particle - based models has been extensively studied by simulations @xcite , but our thermodynamic grasp of these models is rather poor @xcite . in this letter , we introduce the components needed to study the thermodynamic behavior of microscopic , microphase - forming models and thus help clarify the interplay between equilibrium ordering and sluggish dynamics . . [fig : fig4 ] of lamellae depends only weakly on temperature in this regime , where @xmath0 ( dots ) , @xmath1 ( squares ) and @xmath2 ( triangles ) .( b ) decay of @xmath3 at @xmath4 ( dashed ) fitted to an ising critical form @xmath5(solid ) with @xmath6 up to @xmath7 .we estimate @xmath8 .( c ) the average cluster size at the cmc , @xmath9 decreases as @xmath10 increases ( dots ) .the line is a guide for the eye .( d ) equilibrium fcc - cluster crystal occupancy , @xmath11 ( solid symbols ) , and average fluid cluster size , @xmath12 ( empty symbols ) , at @xmath1 ( squares ) and @xmath13 ( triangles ) .because @xmath14 at the cluster fluid fcc - cluster crystal transition , most clusters in the fluid must shrink for crystallization to proceed.,scaledwidth=50.0% ] , ( b ) @xmath15 , ( c ) @xmath16 , and ( d ) @xmath17 mc sweeps of an initially equilibrated simulation with @xmath18 at @xmath19 and @xmath20 . the distinction between particles that remain within @xmath21 of their initial conditions ( red ) and those that have moved beyond it ( blue ) illustrates the network dynamics .surface particles decorrelate more rapidly than core particles , and network nodes ( yellow circles ) reorganize more slowly than network edges . ]the square - well - linear ( swl ) model we study here has a schematic interaction form that can be smoothly transformed into that of diblock copolymers and other microphase - forming models .its radial pair interaction @xmath22 includes hard - sphere volume exclusion @xmath23 at the particle diameter @xmath24 , which sets the unit of length , as well as a salr contribution @xmath25 the square - well attraction strength @xmath26 , which sets the unit of energy , is felt up to @xmath27 ; beyond that point repulsion of strength @xmath28 takes over and decays linearly .note that choosing @xmath29 places the system well above the lifshitz point , @xmath30 , for the prototypical values @xmath31 and @xmath32 used here @xcite ( see appendix sec . [ section : ti ] ) .we simulate systems containing between @xmath33 and @xmath34 particles under periodic boundary conditions at fixed temperature @xmath35 ( the boltzmann constant is set to unity ) , fixing either pressure @xmath36 or volume @xmath37 ( and thus number density @xmath38 ) . for each state point, we perform between @xmath39 and @xmath40 monte carlo ( mc ) sweeps , which include non - standard mc moves @xcite , in order to obtain equilibrium configurations of the different phases studied ( see appendix sec .[ section : simulationdetails ] ) .the results presented here have been first equilibrated and then averaged over simulations at least five times longer than the structural relaxation time .obtaining equilibrium information about microscopic salr particle - based models requires going beyond the common free energy techniques used for simulating gas , liquids and crystals @xcite , because these techniques fail to account for the fluctuating occupancy of periodic microphase features @xcite .the problem is similar to that encountered in multiple - occupancy crystals @xcite and lattices with vacancies @xcite .we thus consider an expanded differential form for the helmholtz free energy per particle @xcite @xmath41 in which the standard thermodynamic contributions , including the entropy per particle @xmath42 , are complemented with a field @xmath43 that is conjugate to the microphase occupancy @xmath44 ( see appendix sec. [ section : occupancyandreferencefield ] ) .( this last quantity is generally proportional to the number of particles per period , but , for convenience , its specific definition here depends on the phase symmetry , e.g. , area density @xmath45 for lamellae of periodicity @xmath46 , line density for cylinders and average cluster size @xmath47 for cluster crystals . ) because in the thermodynamic @xmath48 limit @xmath43 must vanish at equilibrium , optimal finite - size estimates have @xmath49 .standard simulation schemes can not , however , directly minimize this function because of the incommensurability between the mesoscale patterns and the simulation box in finite size systems @xcite . hence , we first obtain the constrained free energy , @xmath50 , of a given microphase morphology at a given ( @xmath51 ) state point and fixed @xmath44 through a two - step thermodynamic integration ( ti ) scheme : ( i ) from an ideal gas to a liquid of hard spheres under a modulated field ; ( ii ) from this last state to swl particles without field .the resulting constrained free energy is then optimized with respect to @xmath44 ( fig .[ fig : fig1 ] ) ( see appendix sec .[ section : ti ] ) .the common tangent construction is used to obtain the coexistence boundaries between different phases and thus the overall phase diagram ( fig . [fig : fig4 ] ) .as expected , at high @xmath10 the system is disordered , while at low @xmath10 equilibrium microphases form .four different ordered microphase morphologies are identified for @xmath52 : face - centered cubic ( fcc ) cluster crystal , cylindrical , double gyroid and lamellar phases ( see depictions fig .[ fig : fig4 ] and symmetry details in appendix sec .[ section : occupancyandreferencefield ] .although a landau functional calculation for simple microphase formers suggests that a body - centered cubic ( bcc ) cluster crystal phase might also form @xcite , we found this structure to be only metastable in our system .the absence of the bcc symmetry suggests that the effective repulsion between clusters is harsher than @xmath53 @xcite , which a hamaker - like calculation confirms ( see appendix sec .[ section : ucc ] ) .this morphology thus appears to be more sensitive than others to the form of the interaction potential .the other microphase morphologies considered , i.e. , o70 @xcite , p surface @xcite , ordered bicontinuous double diamond @xcite and perforated lamellae @xcite , were all found to be unstable within the regime studied .the highest temperature at which periodic microphases melt is the weakly first - order , order - disorder transition ( odt ) @xcite .this transition replaces the second - order gas - liquid critical point for systems beyond the lifshitz point , i.e. , for @xmath54 @xcite ( ( see appendix sec. [ section : lifshitz ] ) ) .melting of the periodic lamellae at @xmath55 is monitored by the decay of the order parameter @xmath56 where @xmath57 is the low-@xmath58 maximum of the structure factor , @xmath59 . in our model, this transition occurs roughly halfway through the lamellar regime , at @xmath60 . because @xmath45 , and thus @xmath46 , is fairly independent of temperature in this regime ( fig .[ fig : fig2](a ) ) , we use the @xmath0 value of @xmath61 to study the decay of @xmath3 .simulation results indicate that although away from the transition the order parameter behaves nearly critically , @xmath3 vanishes discontinuously at @xmath8 ( fig .[ fig : fig2](b ) ). mechanistically , upon going through the transition lamellae become increasingly flexible , giving rise to a percolated network , as observed in diblock copolymers @xcite . at low temperatures , clustersform upon increasing density even before the onset of periodic microphase ordering , as reported in prior simulations and experiments @xcite .the fluid equation of state allows us to locate the onset of clustering , which is akin to determining the critical micelle concentration ( cmc ) in a surfactant system ( see appendix sec .[ section : cmc ] ) .increasing @xmath10 along this line decreases the average cluster size @xmath9 ( fig .[ fig : fig2](c ) ) , and the last hints of a cmc vanish around @xmath62 ( see appendix sec .[ section : cmc ] ) . even within the fluid of clusters ,the intra - cluster cohesion is relatively weak , resulting in the clusters _ internal _ structure to also be fluid - like .this behavior contrasts with the crystallites observed in systems with shorter attraction ranges @xcite , but lowering temperature may also lead to internally - ordered clusters in this system . in spite of their internal fluidity ,the clusters are not generally spherical , and their asphericity increases with @xmath63 ( see appendix sec .[ section : clustersize ] ) . for @xmath64, they even become wormlike and eventually percolate @xcite ( see appendix sec . [ section : clustersize ] ) , which gels the system ( see below ) _ before _ the first - order transition into the periodic microphase regime is reached .this behavior is similar to that observed in refs .@xcite , but contrasts with that of ref .@xcite , where the percolating network was instead associated with incompletely ordered cylinder or lamellar phases . for @xmath65 , by contrast , cluster elongation is preempted by the microphase regime .although this last transition is reminiscent of the crystallization of purely repulsive particles , the clusters in the fluid phase are larger ( fig .[ fig : fig2](d ) ) and display a much wider range of sizes and morphologies than those in the fcc - cluster crystal @xcite .the fcc - cluster crystal assembly is thus expected to be more intricate than simple nucleation and growth .we finally consider the percolated regime observed at temperatures above the periodic microphase regime . at @xmath66the system behaves like a regular fluid , but as @xmath10 approaches @xmath55 the structural relaxation grows increasingly complex , even under the strongly non - local mc sampling we use here ( fig . [fig : fig41 ] ) : ( i ) particles at the surface are lot more mobile than those in the core @xcite , and ( ii ) edges of the network reorganize much faster than its nodes . being in equilibrium , the system does not age , but mc sampling is nonetheless arduous ( see appendix sec . [sec : simulationdetails ] ) , making the system gel - like .this multi - timescale dynamics , in particular the sluggish relaxation of network nodes , may contribute to the difficulty of assembling microphases in colloids @xcite . note that other mechanisms slowing down the dynamics could also emerge below @xmath55 , including competing microphase morphologies @xcite and spinodal - like arrest @xcite , but a systematic study of these non - equilibrium effects is left for future consideration .we have developed a ti - based simulation method for solving the phase diagram of arbitrary continuous - space microphase - forming models .our solution of the prototypical swl model presents the periodic microphase sequence cluster crystal , cylindrical , double gyroid and lamellar phases of systems described by a comparable landau functional @xcite , our search for ordered phases , however , was not exhaustive , hence other stable morphologies are possible .more importantly , we have clarified the thermodynamic interplay between fluids of spherical and wormlike clusters , the equilibrium percolating fluid ( gel - like ) , and periodic microphases .this distinction is essential for separating equilibrium from non - equilibrium effects in the dynamical arrest of microphase formers @xcite .it is also essential for guiding experiments with salr - like colloidal interactions , whose precise form can vary with system density @xcite .indeed , colloidal experiments have thus far only identified equilibrium cluster fluids and gels @xcite .whether the challenge of assembling ordered microphases in colloids could be surmounted by tuning the properties of these disordered regimes or by identifying alternate assembly pathways remains , however , an open question .we acknowledge many stimulating discussions about this project over the years , in particular with a. ciach , e. del gado , d. frenkel , p. royall and s. yaida .we acknowledge support from the national science foundation grant no .nsf dmr-1055586 .the following material contains additional methodological details about the study of the periodic microphase and cluster fluid and percolated regimes as well as a description of the monte carlo sampling scheme .high free - energy barriers separate periodic microphases with different morphologies , making direct equilibration unachievable in that regime .we instead prepare the system into a given morphology and calculate its free energy at a given state point .we then compare the free energy of the different phases in order to identify the equilibrium structure .general arguments suggest that particles with isotropic salr interactions form periodic cluster crystal , cylindrical , double gyroid and lamellar phases @xcite .we thus consider these morphologies as well as less common ones , including o70 @xcite , perforated lamellar @xcite , ordered bicontinuous double diamond @xcite and simple gyroid @xcite .in order to obtain the equilibrium free energy of a given morphology at a given state point , we define a specific measure of lattice occupancy and a reference modulated field ( table [ tab : occ ] ) .the former is a parameter with respect to which the free energy is optimized , which is necessary because the equilibrium periodicity of finite - size systems may be incommensurate with the prepared simulation box .the latter is chosen so as to obtain a continuous thermodynamic integration path from a reference ideal system whose free energy is known .the field must thus break the same symmetry as the periodic phase , while remaining simple to formulate .when possible , we use combinations of trigonometric functions with a wavevector @xmath67 to control the periodicity @xmath46 of the field . for example , for the lamellar phase , which is periodic in one dimension, we use a simple sinusoidal function , for which lattice occupancy is measured by taking the particle number density per unit area of lamella , @xmath45 . note that @xmath68 , where @xmath69 is the total number of particles in one period , and @xmath70 is the area of a lamella .the double gyroid morphology , which is a bicontinuous phase that emerges from a minimal surface problem , is , however , more intricate to describe . in that case , the field that couples to the particle distribution is a minimal surface function .p1.83 cm p3.2 cm c c p 8.1 cm morphology & measure of occupancy & @xmath44 symbol & typical range & reference field , @xmath71 + lamellar & average area density per layer & @xmath45 & @xmath72 & @xmath73 + cylindrical & average line density per cylinder & & @xmath74 & @xmath75 \cos\left[k \left ( \frac{x}{2}-\frac{\sqrt{3}y}{2 } \right)\right]\end{aligned}$ ] + fcc - cluster crystal & average number of particles per cluster & @xmath47 & @xmath76 & @xmath77 + bcc - cluster crystal & average number of particles per cluster & @xmath47 & @xmath76 & @xmath78\cos[k(x - y)]\cos[k(y+z)]\cos[k(y - z)]\right . \\& \left.\ \cos[k(z+x)]\cos[k(z - x)]\right|^{1/2 } \end{aligned}$ ] + double gyroid & average number of particles per unit cell . & & @xmath79 & @xmath80 & , \\\mathrm{where\quad } g(x , y , z)=&\cos(kx)\sin(ky)+\cos(ky)\sin(kz ) \\&+\cos(kz)\sin(kx)\end{aligned } $ ] + the free energy of the periodic microphases is determined by a two - step thermodynamic integration ( ti ) scheme .the reference state consists of low - density hard spheres under an external modulated field ( hsf ) , whose free energy is @xmath81 where @xmath82 is the free energy of an ideal gas of number density @xmath63 .the thermal de broglie wave length , @xmath83 , is set to unity without loss of generality .the first ti step brings the system density from @xmath84 up to the targeted density as @xmath85 while keeping the field fixed . at low densities the first two virial coefficientsare used to facilitate numerical integration ( sect .[ section : virial ] ) .note that for the lamellar and cylindrical phases , a virtual harmonic spring @xmath86 with stiffness @xmath87 is also applied to the periodic directions of the box .this spring ensures that the final configuration preserves the targeted lattice occupancy . because the integrand is here independent of the system energy , the integration is unaffected by this constraint .the second ti step , which brings hsf to fully - interacting swl particles , follows the linear alchemical transformation , @xmath88 , which gives @xmath89 for a gauss lobatto quadrature with @xmath90 points is used to minimize the numerical integration error .summing the various contributions gives the free energy of a system constrained to a given morphology and lattice occupancy@xmath91 the first two virial coefficients for hsf , @xmath92 and @xmath93 , correspond to the intercept and the initial slope , respectively , of the integrand for the first ti step . because simulation results suffer from error magnification at low @xmath63 , these two quantities are particularly useful .@xmath92 is evaluated by integrating over particle positions @xmath94 under the external field @xmath71 , @xmath95 where for hard spheres the mayer function is simply @xmath96 , with @xmath97 the heaviside theta function .the expression for lamellae can be simplified as @xmath98+\sin { k_zz}}}{l_z\left(\int_0^{l_z}{\mathrm{d}}z e^{-\beta\mathcal{f}_0\sin{k_zz}}\right)^2},\ ] ] but for the other fields the full expression is numerically evaluated by monte carlo integration .we can similarly obtain and evaluate @xmath99 sample results are given in fig .[ fig : tb3 ] . , @xmath100 from simulations ( dots ) and from the virial expansion with @xmath101 and @xmath102 ( dashed line ) .third virial coefficient , @xmath93 , for the reference field of ( b ) the lamellar phase for interlayer @xmath103 , @xmath104 , @xmath105 , @xmath106 , @xmath107 and @xmath108 ( from bottom to top ) , and ( c ) the fcc - cluster crystal phase for intercluster @xmath109 , @xmath110 , @xmath111 and @xmath112 ( from top to bottom ) ., title="fig:",scaledwidth=32.0% ] , @xmath100 from simulations ( dots ) and from the virial expansion with @xmath101 and @xmath102 ( dashed line ) .third virial coefficient , @xmath93 , for the reference field of ( b ) the lamellar phase for interlayer @xmath103 , @xmath104 , @xmath105 , @xmath106 , @xmath107 and @xmath108 ( from bottom to top ) , and ( c ) the fcc - cluster crystal phase for intercluster @xmath109 , @xmath110 , @xmath111 and @xmath112 ( from top to bottom ) ., title="fig:",scaledwidth=32.0% ] , @xmath100 from simulations ( dots ) and from the virial expansion with @xmath101 and @xmath102 ( dashed line ) .third virial coefficient , @xmath93 , for the reference field of ( b ) the lamellar phase for interlayer @xmath103 , @xmath104 , @xmath105 , @xmath106 , @xmath107 and @xmath108 ( from bottom to top ) , and ( c ) the fcc - cluster crystal phase for intercluster @xmath109 , @xmath110 , @xmath111 and @xmath112 ( from top to bottom ) ., title="fig:",scaledwidth=32.0% ] microphases form in systems with salr interactions for which the repulsion is sufficiently strong . for weak repulsion, a simple gas - liquid binodal is found in their stead . for anisotropic repulsionthe transition between the two regimes produces a tricritical lifshitz point @xcite , but for isotropic repulsion the order - disorder transition into the microphase regime is weakly first - order .although the lifshitz point then formally disappears @xcite , some authors have elected to keep the same vocable to describe the transition from a gas - liquid critical point to microphase formation @xcite .we follow this convention . in order to ensure that our swl potential falls on the microphase - forming side of the lifshitz point , we estimate its location with respect to the repulsion strength , @xmath113 .standard gibbs ensemble monte carlo ( gemc ) simulations with two simulation boxes , initially each with @xmath114 , for @xmath40 steps locate the gas - liquid coexistence regime ( fig .[ fig : potentials ] ) .( black ) , @xmath115 ( red ) , 0.02 ( green ) and 0.05 ( blue ) .( b ) gemc simulation results ( squares ) for different @xmath113 show the accompanying evolution of the critical temperature @xmath116 ( dots ) .lines are fit to eq . .note that for @xmath117 the results are consistent with those of ref .@xcite , and that for @xmath118 , the critical temperature @xmath119 is quite close to @xmath8 for @xmath29.,title="fig:",scaledwidth=48.0% ] ( black ) , @xmath115 ( red ) , 0.02 ( green ) and 0.05 ( blue ) .( b ) gemc simulation results ( squares ) for different @xmath113 show the accompanying evolution of the critical temperature @xmath116 ( dots ) .lines are fit to eq . .note that for @xmath117 the results are consistent with those of ref .@xcite , and that for @xmath118 , the critical temperature @xmath119 is quite close to @xmath8 for @xmath29.,title="fig:",scaledwidth=48.0% ] fitting the coexistence results with @xmath120 where @xmath121 and @xmath122 are fit parameters and @xmath6 is the three - dimensional ising universality class critical exponent , provides an estimate of the gas - liquid critical temperature @xmath123 and density @xmath124 . the results for @xmath125 are given in fig .[ fig : potentials ] , but @xmath126 no gas - liquid coexistence regime can be detected .we thus estimate the lifshitz point , @xmath127 , which indicates that that our swl model at @xmath29 is well within the microphase - forming regime .standard approaches for determining the critical micelle concentration ( cmc ) implicitly rely on the existence of ideal - gas - like scaling in both the particle and the cluster fluid regimes of the equation of state .such an ideal scenario only exists for systems in which the cluster fluid regime has a fairly broad density range of stability and for which the distribution of cluster sizes , @xmath128 , is narrow and roughly independent of particle concentration .because neither of these conditions is met here , we develop a more general scheme for estimating the cmc . in the spirit of the classical definition of the cmc , which situates the crossover as the point of most abrupt change in the physico - chemical properties of the system @xcite, we take the minimum when it exists of @xmath129 this function measures deviations of the equation of state from the ideal gas behavior .the pronounced peak it exhibits in cluster - forming systems ( see fig . 3 of the main text ) indicates a rapid change in the system properties . in practice , we determine the cmc by first fitting the numerical results for the equation of state to @xmath130 where @xmath131 and @xmath132 are the monomer and cluster number density , respectively .note that the second virial coefficients @xmath133 and @xmath134 are fitted to the simulation data , but we directly calculate @xmath135[e^{\beta\varepsilon}-1 ] + 6\int_{\lambda\sigma}^{\kappa\sigma}r^2\left[e^{-\beta\xi\varepsilon(\kappa - r/\sigma)}-1\right]{\mathrm{d}}r\right\}.\ ] ] in the rest of this section , we demonstrate that the above approach reduces to the standard scheme for the ideal scenario @xcite .we first consider the classical derivation . for a monodisperse distribution of cluster sizes ,i.e. , @xmath136 , we can write a chemical equilibrium between individual particles and clusters , @xmath137 with the equilibrium constant @xmath138 recalling that the total concentration of material @xmath63 is conserved , i.e. , @xmath139 , and defining @xmath140 gives @xmath141 where the explicit dependence of @xmath142 on @xmath63 is dropped in order to simplify the notation . this equation can be rewritten as @xmath143 at the cmc , typical values for @xmath144 range from @xmath145 . because @xmath146 is typically much larger than that, the second term can be neglected .hence , for @xmath147 we get @xmath148 this last result is the classical definition of the cmc . using the scheme proposed here for the ideal scenario , we write for @xmath147 @xmath149 defining the cmc as the minimum of @xmath150 , we then obtain @xmath151 where @xmath152 can be obtained from the law of mass action in eq .( [ eq : massaction ] ) , @xmath153 for @xmath154 , we finally obtain @xmath155 and thus @xmath156 in the ideal scenario , we finally recover eq .( [ eq : cmc ] ) . by analogy with soft spheres , we expect that a sufficiently steep effective cluster - cluster interaction , @xmath157 , should result in their crystallization to a fcc - cluster crystal phase @xcite . by assuming that the particle density inside spherical clusters is constant and that two clusters can not overlap , we obtain @xmath158 where @xmath159 is the probability distribution function of a particle at distance @xmath160 from the center of the cluster @xmath161 .defining the particle - cluster interaction @xmath162 and using hamaker s method @xcite , we get for clusters of radius @xmath163 @xmath164 where @xmath165}{r}$ ] . using the integration setup described in fig .[ fig : ucc ] , a particle within cluster @xmath166 at a distance @xmath167 has an interaction energy with cluster @xmath168 , @xmath169 . integrating all the particle positions in @xmath166, we thus obtain the cluster - cluster interaction @xmath170 results for different cluster sizes @xmath171(fig .[ fig : ucc ] ) suggest that the effective cluster - cluster interaction decays faster than @xmath172 and thus falls within the stability regime for the fcc crystal phase ., at the cmc for @xmath1 ( red ) , @xmath2 ( blue ) , @xmath19 ( purple ) and @xmath173 ( green ) .the distribution steadily shifts towards smaller cluster sizes as @xmath10 increases .( b ) pdf of cluster sizes at @xmath2 for @xmath174 ( blue ) , @xmath175 ( red ) , @xmath176 ( green ) and @xmath177 ( black ) .the distribution becomes increasingly broad with increasing density .( c ) total ( gray ) , non - spherical ( dark gray ) and spherical ( black ) pdf of cluster sizes at @xmath174 and @xmath2 , using @xmath178 .recall that the average of @xmath179 is @xmath12.,title="fig:",scaledwidth=32.0% ] , at the cmc for @xmath1 ( red ) , @xmath2 ( blue ) , @xmath19 ( purple ) and @xmath173 ( green ) .the distribution steadily shifts towards smaller cluster sizes as @xmath10 increases .( b ) pdf of cluster sizes at @xmath2 for @xmath174 ( blue ) , @xmath175 ( red ) , @xmath176 ( green ) and @xmath177 ( black ) .the distribution becomes increasingly broad with increasing density .( c ) total ( gray ) , non - spherical ( dark gray ) and spherical ( black ) pdf of cluster sizes at @xmath174 and @xmath2 , using @xmath178 .recall that the average of @xmath179 is @xmath12.,title="fig:",scaledwidth=32.0% ] , at the cmc for @xmath1 ( red ) , @xmath2 ( blue ) , @xmath19 ( purple ) and @xmath173 ( green ) .the distribution steadily shifts towards smaller cluster sizes as @xmath10 increases .( b ) pdf of cluster sizes at @xmath2 for @xmath174 ( blue ) , @xmath175 ( red ) , @xmath176 ( green ) and @xmath177 ( black ) .the distribution becomes increasingly broad with increasing density .( c ) total ( gray ) , non - spherical ( dark gray ) and spherical ( black ) pdf of cluster sizes at @xmath174 and @xmath2 , using @xmath178 .recall that the average of @xmath179 is @xmath12.,title="fig:",scaledwidth=32.0% ] a broad range cluster shapes and sizes is observed in the cluster fluid regime . in fig . 2 of the main text only average valuesare reported , but in fig .[ fig : clusterdistribution ] , we consider the full cluster size distribution . in order to separate spherical from non - spherical clusters, we also define the anisotropy parameter @xmath180 @xmath181 from the eigenvalues @xmath182 of the inertia tensor of each cluster .a perfectly spherical object would have @xmath183 . breaking down the probability distribution between spherical andnon - spherical clusters using an arbitrary threshold @xmath184 suggests that a continuum of deformations from sphericity is observed ( fig .[ fig : clusterdistribution ] ) . note that this behavior is rather different from whatis observed in the fcc - cluster phase , where not only the average cluster size , but also the size and shape variances are all reduced .for systems of size @xmath185 ( purple ) , @xmath186 ( blue ) , @xmath33 ( red ) , @xmath187 ( black ) , @xmath188 ( green ) and @xmath18 ( brown ) .the dashed line indicates the mid - point of the percolation probability , @xmath189 .( b ) extraction of the percolation transition , @xmath190 ( solid dots ) , for @xmath191 ( black ) , @xmath192 ( red ) and @xmath193 ( purple ) from finite - size scaling ( empty dots ) .the solid line is a fit of eq . to the numerical results.,title="fig:",scaledwidth=48.0% ] for systems of size @xmath185 ( purple ) , @xmath186 ( blue ) , @xmath33 ( red ) , @xmath187 ( black ) , @xmath188 ( green ) and @xmath18 ( brown ) .the dashed line indicates the mid - point of the percolation probability , @xmath189 .( b ) extraction of the percolation transition , @xmath190 ( solid dots ) , for @xmath191 ( black ) , @xmath192 ( red ) and @xmath193 ( purple ) from finite - size scaling ( empty dots ) .the solid line is a fit of eq . to the numerical results.,title="fig:",scaledwidth=48.0% ] at low @xmath10 , spherical clusters first turn into wormlike clusters and then form a disordered percolating network upon increasing @xmath63 , while outside the range of stability of periodic microphases . changing the cluster fluid morphology toa percolated netowrk dramatically slows down phase - space sampling , which must thence be supplemented by parallel tempering . in order to clearly delineate the hard - to - sample regime , we determine the percolation transition @xmath190 by finite - size scaling of the mid - point of the percolation probability @xmath194 where @xmath195 is the universal critical scaling for three - dimensional standard percolation ( fig .[ fig : percolation ] ) @xcite .the intercept of the fit gives the thermodynamic , @xmath48 , percolation transition .results are given in the inset of fig .4 of the main text .note that although the percolation transition can be formally defined for any @xmath10 , for @xmath196 , the impact of percolation on monte carlo sampling is found to be negligible .the inter - particle bonding is then comparable to thermal excitations .in order to optimize phase space sampling , we use different sets of monte carlo moves for each of the phases . *all phases use single - particle local moves following a metropolis scheme .the maximal displacement is tuned , such that the acceptance ratio is kept between @xmath197 .* all phases also use two types of non - local moves : ( i ) @xmath198 of the moves are system - wide random displacements , and ( ii ) @xmath198 of the moves are aggregation volume biased ( avb ) moves , which specifically displace a particle to the surface of another @xcite .the avb `` in '' region is set to the attraction radius , i.e. , @xmath199 , while the `` out '' region is the rest of the system .* because the first ti step is done at constant pressure , volume changes must be made .for the lamellar morphology , the cartesian directions perpendicular to the lamellae fluctuate together , independently of the third direction . for the cylindrical phase ,the ratio between the directions parallel to the cylinder is kept fixed , in order to maintain the cylindrical periodicity .for the other phases , a cubic symmetry of the box is maintained . *to accelerate volume sampling of fcc - cluster phase in this same context , we use a cluster volume move algorithm .the move consists of two changes ( i ) the volume change of equilibrium lattice sites and ( ii ) the changes between particle , cluster and this lattice sites , see sec .[ section : clustervolumemove ] .* we use @xmath200pt$ ] simulation and histogram reweighting to determine the coexistence regime near the cluster fluid , fcc - cluster crystal cylindrical triple point @xcite . for each morphology , we iterate the @xmath40 monte carlo steps with @xmath201 replicates , in order to ensure sufficient samplings of each configurations . * in the cluster fluid regime , @xmath202 of the displacements attempts are virtual move monte carlo , which enable cluster displacements @xcite .as the virtual moves have a significant computational overhead , we only use them in the non - percolated cluster fluid regime , where the increase efficiency warrants it . * near and within the percolated fluid regime , parallel tempering is used with attempts at exchanging configurations occuring @xmath202 of the displacements .temperature intervals are @xmath203 for @xmath204 and @xmath205 for @xmath206 . in order to obtain an ergodic sampling , for @xmath207the temperature chain goes up to @xmath208 , while for @xmath209 the chain goes up to @xmath210 . note that in the cluster fluid regime , numerical convergence is checked by determining the free energy by two different approaches for each state point : ( i ) widom insertion at @xmath208 @xcite , followed by ti over @xmath10 @xcite ; and ( ii ) ti to a targeted @xmath10 , followed by a ti over @xmath63 .the pressure is then determined from the virial , as described in sec .[ sec : pressure ] .all phase boundaries are determined by common tangent construction from the free energy results . in constant volume simulations , the pressure for the swl model can be determined from its virial @xmath211(\lambda\sigma)^3 + \xi\varepsilon\int_{\lambda\sigma^+}^{\kappa\sigma}g(r)r^2{\mathrm{d}}r . \right\},\end{aligned}\ ] ] recalling that for @xmath212 the radial pair distribution and the cavity functions are equal , i.e. , @xmath213 . in constant pressure simulations of the fcc - cluster phase , a generalization of the scheme in described in ref .@xcite is used to enhance sampling .the logarithmic volume move of the lattice sites @xmath214 is then accompanied by changes to the particle positions @xmath215\right)^{\frac{1}{d(n-1)}}.\ ] ] the acceptance rule is then @xmath216,1 \right\}.\ ] ] 68ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ] + 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop ( ) @noop * * , ( ) @noop * * , ( ) in http://link.springer.com/10.1007/978-3-319-00354-2[__ ] , ( , , ) pp . @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) , @noop * * , ( ) , @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) , @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , ) @noop * * , ( ) @noop * * , ( ) @noop * * ( ) @noop _ _ ( , ) @noop _ _ ( , )

<S> periodic microphases universally emerge in systems for which short - range inter - particle attraction is frustrated by long - range repulsion . </S><S> the morphological richness of these phases makes them desirable material targets , but our relatively coarse understanding of even simple models hinders controlling their assembly . </S><S> we report here the solution of the equilibrium phase behavior of a microscopic microphase former through specialized monte carlo simulations . the results for cluster crystal , cylindrical , double gyroid and lamellar ordering qualitatively agree with a landau - type free energy description and reveal the nontrivial interplay between cluster , gel and microphase formation . </S>

1702.05890

the htl ( hard thermal loop)resumed propagators , i.e . this htl resumed technique is introduced by braaten and pisarski @xcite , to cure some serious problem of gauge theories at finite temperature , such as ir singularity and gauge dependent results , when the bare propagators ( vertices ) are used .some gauge independent physical quantities are given with the htl resumed propagators in the calculation , other than the bare propagators at finite temperature .the htl approximation is the high temperature limit .if the momentum of the propagators are soft at finite temperature , we should use the htl resumed propagators .multi - loop htl diagrams can give a low order contribution in the coupling constant at finite temperature , which should be resumed .because of the htl resummation , the medium effects are taken into account , such as the debye screening caused by the color charges of the qgp .the htl resummation technique represents a great progress compared to the bare perturbation theory at finite temperature .photon production is a good hard probe to study the formation and evolution of the quark - gluon plasma in the experiment of the heavy ion collisions . for the leading order photon emission rate , due to the near - collinear singularities , bremsstrahlung and inelastic pair annihilation give the leading order contribution as two - to - two particle processes@xcite . by powercounting , these two processes with multiple collinear soft gluon scattering still contribute at the leading order . sum these infinite class diagrams , and the result can give an @xmath1 suppression , which limits the coherence length of the emitted radiation .this is the lpm effect ( landau - pomeranchuck - migdal)@xcite,@xcite,@xcite,@xcite .the htl resumed gluon propagator in the coulomb gauge is used to handle the situation of the soft gluons .the light cone gauge is one of non - covariant gauges @xcite,@xcite , and is also ghost - free .when the multiple gluon emission is calculated in the light cone gauge , the differential cross section with @xmath2-gluon emission in the leading pole approximation has a simple ladder structure @xcite . in the light conegauge interference terms do not contribute to the leading order .these nice properties simplify the calculation .however , the light cone gauge has its disadvantage , such as the spurious singularity of @xmath3 , the renormalization and so on . in the experiment of the heavy ion collisions ,the longitudinal momentum of the parton is very large , the transverse momentum is very small , and it is suitable to calculate in light cone gauge . for considering the medium effects in the light cone gauge , we should work out the transverse and longitudinal parts of the gluon htl self energy in the light cone gauge and derive the htl resumed gluon propagator .the remainder of the paper is organized as follows . in sec.iiwe review the htl resumed gluon propagator in the coulomb gauge . in sec.iiithe htl resumed gluon propagator in the light cone gauge is worked out , and then we calculate and analyze the transverse and longitudinal parts of the gluon htl self energy in the light cone gauge . viathe htl resumed gluon propagator we show the transverse and longitudinal spectral functions and the equations of the dispersion relation .in sec.iv we calculate the quark self energy in the htl approximation , and show the htl resumed quark propagator gauge independent . in sec.v we calculate the damping rates of the hard quark and gluon with this htl resumed gluon propagator in the light cone gauge , and demonstrate in general case we can get the same result in the light cone gauge and the coulomb gauge .our conclusion is given in sec.vi .at zero temperature , covariant gauge have a definite advantage over non - covariant gauges such as the coulomb gauge or axial gauges .calculations are simplified considerably due to lorentz invariance , and the renormalization program can be implemented in practice only in covariant gauge . at finite temperature ,lorentz invariance is broken because the heat bath defines a privileged frame , and renormalization program is of secondary importance , so that non - covariant gauges may present useful alternatives to covariant gauge@xcite .the htl resumed gluon propagator has been derived in the coulomb gauge @xcite,@xcite .the htl gluon self energy @xmath4 is expressed as the transverse part and the longitudinal part .the gluon self energy is given by @xmath5 where the transverse projection tensor @xmath6 , the longitudinal projection tensor @xmath7 , and the four vector @xmath8 are defined as @xmath9 the axial vector is @xmath10 which specifies the thermal rest frame .the inverse propagator for general @xmath11 in the coulomb gauge is @xmath12 where @xmath11 is a arbitrary gauge parameter .the inverse propagator reduces in the limit @xmath13 to @xmath14 @xmath15 can also be written as @xmath16 where @xmath17 and@xmath18 are the transverse and longitudinal propagators : @xmath19 the htl resumed gluon propagator in the coulomb gauge @xcite is @xmath20 by calculating , we can find @xmath21 so the htl resumed gluon propagator in the coulomb gauge can be simplified into @xmath22 where @xmath23 is a unit vector on the direction of @xmath24 , @xmath25 and @xmath26 .the longitudinal and transverse gluon htl self energy @xcite are @xmath27\notag\;,\\ % \pi_{t}(p)&=&\frac{m_{d}^{2}}{2}\frac{p_{0}^{2}}{p^{2}}\left[1-(1-\frac{p^{2}}{p_{0}^{2}})\frac{p_{0}}{2p}\left[\ln|\frac{p_{0}+p}{p_{0}-p}|-i\pi\theta(p^{2}-p_{0}^{2})\right]\right]\;,\end{aligned}\ ] ] where the gluon screening mass @xmath28 and @xmath29 is the step function .the imaginary parts in the above equations correspond to the landau damping , which means that one particle is emitted from the thermal medium and absorbed by the medium . in the static limit , @xmath30 , the longitudinal htl self energy @xmath31 which means the debye screening of the gluon in the plasma .however , in the static limit , the transverse htl self energy @xmath32 which shows no static magnetic screening .the self - energy tensor @xmath33 is symmetric in @xmath34 and @xmath35 and satisfies @xmath36in this section , we derive the htl resumed gluon propagator in the light cone gauge , and compute the transverse and longitudinal parts of gluon htl self energy in the real time formalism . in the static limit, we discuss @xmath37 in the light cone gauge and the coulomb gauge .we obtain the pole terms and the cut terms of the transverse and longitudinal spectral functions .the light cone gauge is one of axial type gauges and non - covariant gauges @xcite,@xcite , @xmath38 the axial vector in the light cone gauge is @xmath39 the bare gluon propagator in the light cone gauge is @xmath40 here we use the mandelstam - leibbrandt(ml ) prescription of @xmath41 instead of the usual principal - value prescription , @xmath42 where @xmath43 .the usual principal - value prescription of @xmath0 leads to some serious problems , such as violating power counting and other basic criteria , when we calculate the integral of the loop diagram@xcite . in the real time mechanism ,the time of the field goes from @xmath44 to @xmath45 .the contour can be deformed in order to include the real time axis by going first from @xmath44 to @xmath46 above the real time axis and then back to @xmath45 below real time axis .so we have double degrees of freedom , one exists above the real time axis and the other one exists below the real time axis .we get the propagator in the real time formalism @xcite,@xcite , which is a @xmath47 matrix , @xmath48 the bose - einstein distribution function @xmath49is @xmath50 for fermions , we have @xmath51\;,\end{aligned}\ ] ] the fermi - dirac distribution function @xmath52 is @xmath53 we use the keldysh representation in real time formalism @xcite,@xcite .the retarded propagator , advanced propagator and symmetric propagator for bosons are @xmath54\;.\end{aligned}\ ] ] the inverse relation for bosons is @xmath55\notag\;,\\ % & & \delta_{12}=\frac{1}{2}[\delta_{s}(k)+\delta_{a}(k)-\delta_{r}(k)]\notag\;,\\ % & & \delta_{21}=\frac{1}{2}[\delta_{s}(k)-\delta_{a}(k)+\delta_{r}(k)]\notag\;,\\ % & & \delta_{22}=\frac{1}{2}[\delta_{s}(k)-\delta_{a}(k)-\delta_{r}(k)]\;.\end{aligned}\ ] ] for fermions , in the keldysh representation , we only replace the bose - einstein distribution function @xmath49 by the fermion - dirac distribution function @xmath56 in the symmetric propagator , and the retarded propagator and advanced propagator are the same as that of bosons , @xmath57\;.\ ] ] the inverse relation in eq .is also applicable for fermions .it has been checked by explicit computation in different gauges ( covariant , coulomb , temporal ) whose axial vectors are all the same , @xmath58 , that the gluon htl self energy does not depend on the choice of gauge . however , the axial vector in light cone gauge is different , that brings about some changes .a massive boson gives rise to the longitudinal polarization state , so that the boson self energy is separated into the longitudinal and transverse parts @xcite .the massive gluon self energy in the light cone gauge is made up of the transverse and longitudinal parts , @xmath59\;.\ ] ] the transverse projection tensor is @xmath60 the longitudinal projection tensor is @xmath61\;.\ ] ] the four - vector @xmath8 is @xmath62 @xmath63 and @xmath64 satisfy the following relation , @xmath65 the axial vector @xmath66 in the light cone gauge is defined in eq ., @xmath67 @xmath7 and @xmath6 are the longitudinal and transverse projection tensors with respect to the axial vector @xmath66 in the light cone gauge in eq .. because the axial vector in the light cone gauge in eq .is different from the axial vector in the coulomb gauge in eq ., so the longitudinal and transverse projection tensors @xmath7 and @xmath6 are different from those in the coulomb gauge .finally , these differences make the expression of the htl resumed gluon propagator in the light cone gauge changed .the inverse propagator in the light cone gauge in the limit @xmath68 is @xmath69t_{p}^{\mu\nu}+\big[-p^{2}+\frac{1}{n_{p}^{2}}\pi_{l}(p)\big]l_{p}^{\mu\nu}\;.\end{aligned}\ ] ] apply the two equations @xmath70 and we can get the htl resumed gluon propagator in the light cone gauge @xmath71 where we do not consider the terms containing the gauge parameter @xmath11 . when @xmath72 and @xmath73 , the htl resumed gluon propagator turns back to the bare gluon propagator in eq .. due to the relation in eq ., , , , the transverse and longitudinal gluon htl self - energies are given by @xmath74\pi^{\mu\nu}(p),\\ % \frac{1}{n_{p}^{2}}\pi_{l}(p)&=&-l_{p\mu\nu}\pi^{\mu\nu}(p)=\left[\frac{p_{\mu}p_{\nu}}{p^{2}}-\frac{n_{\mu}p_{\nu}+n_{\nu}p_{\mu}}{n\cdot p}+\frac{n_{\mu}n_{\nu}p^{2}}{(n\cdot p)^{2}}\right]\pi^{\mu\nu}(p)\;.\end{aligned}\ ] ] @xmath4 is the sum of the quark loop , the gluon loop and the gluon tadpole in the light cone gauge .multiply @xmath4 by the projection tensors @xmath75 and @xmath76 , and we can calculate @xmath77 and @xmath78 in the htl approximation . + the quark loop in fig.[quark - loop ] can be expressed as @xmath79\int\frac{d^{4}k}{(2\pi)^{4}}\mathrm{tr}[\gamma^{\mu}f(k)\gamma^{\alpha}f(k - p)]\;,\end{aligned}\ ] ] where @xmath80 is the active quark flavors and @xmath81 is the bare quark propagator . the retarded self energy in the real time formalism @xcite , @xcite is expressed as @xmath82-\mathrm{tr}[\gamma^{\mu}f_{12}(k)\gamma^{\alpha}f_{21}(k - p)]\big]\notag\\ % & & = -\frac{i}{2}g^{2}n_{f}\int\frac{d^{4}k}{(2\pi)^{4}}\mathrm{tr}[\gamma^{\mu}\cancel{k}\gamma^{\alpha}(\cancel{k}-\cancel{p})][\tilde{\delta}_{11}(k)\tilde{\delta } _ { 11}(k - p)-\tilde{\delta}_{12}(k)\tilde{\delta}_{21}(k - p)]\;,\end{aligned}\ ] ] where the rtf green function @xmath83 refers to the component of the propagator in the real time formalism in eq .. multiply @xmath84 by the transverse projection tensor @xmath85 , we can get the transverse self energy @xmath77 , @xmath86\big[\tilde{\delta}_{11}(k)\tilde{\delta}_{11}(k\!-\!p)\!-\!\tilde{\delta}_{12}(k)\tilde{\delta}_{21}(k\!-\!p)\big]\;.\end{aligned}\ ] ] using the relation in eq ., we can obtain @xmath87\notag\\ & & = \frac{1}{2}\big[\tilde{\delta}_{s}(k - p)\tilde{\delta}_{r}(k)+\tilde{\delta}_{a}(k - p)\tilde{\delta}_{s}(k)\big]\;,\end{aligned}\ ] ] where the minus sign in front of the term @xmath88 comes from the vertex of the type 2 fields @xcite . the @xmath89 integral of @xmath90 and @xmath91reduce to zero .replace @xmath92 by @xmath93 in the first term and using @xmath94 , and this expression can be simplified further on , @xmath95\tilde{\delta } _ { s}(k)\tilde{\delta } _ { a}(k - p)\notag\\ % & = & \frac{i}{4}n_{f}g^{2}\int\frac{\mathrm{d}^{4}k}{(2\pi)^{4}}8\left[k\!\cdot\ ! p\!-\!2\frac{n\!\cdot\ !p\!+\!\frac{(n\!\cdot \!k)^{2}}{(n\!\cdot\ ! p)^{2}}p^{2}\right](-2\pi i)\delta(k^{2})[1\!-\!2f(k_{0})]\frac{1}{(k\!-\!p)^{2}\!-\!isgn(k_{0}\!-\!p_{0})\epsilon}\;.\end{aligned}\ ] ] in the calculation , we use the htl approximation , i.e . high temperature limit .the internal momentum @xmath92 is hard , and the external momentum momentum @xmath96 is soft . the transverse part of the quark loop in the light cone gauge in the htl approximation from eq . is obtained by @xmath97\tilde{\delta } _ { s}(k)\tilde{\delta } _ { a}(k\!-\!p)\notag\\ % & = & \frac{1}{12}n_{f } g^{2}t^{2}\left[\frac{(p_{0})^{2}}{p^{2}}-\frac{p_{0}p^{2}}{2p^{3}}\ln\frac{p_{0}+p+i\epsilon}{p_{0}-p+i\epsilon}\right]\;.\end{aligned}\ ] ] similarly , the longitudinal part of the quark loop in the light cone gauge in the htl approximation is @xmath98\tilde{\delta } _ { s}(k)\tilde{\delta } _ { a}(k\!-\!p)\notag\\ % & = & -\frac{1}{6}n_{f}g^{2}t^{2}\left[\frac{p^{2}}{p^{2}}-\frac{p_{0}p^{2}}{2p^{3}}\ln\frac{p_{0}+p+i\epsilon}{p_{0}-p+i\epsilon}\right]\;.\end{aligned}\ ] ] + the gluon loop and the gluon tadpole in fig.[gloun - loop and tadpole ] are expressed as @xmath99 where @xmath100 , @xmath101and @xmath102 are all the bare gluon propagators . below are the tensors of the bare gluon propagators in the light cone gauge @xmath103 the three gluon vertexes are @xmath104\notag\;,\\ % v^{\beta\gamma\alpha}(k ,p - k ,- p)&=&gf^{cdb}\big[g^{\beta\gamma}(2k - p)^{\alpha}+g^{\gamma\alpha}(2p - k)^{\beta}+g^{\beta\alpha}(-p - k)^{\gamma}\big]\;.\end{aligned}\ ] ] the four gluon vertex is @xmath105\;.\ ] ] the transverse part of the gluon self energy in the htl approximation from eq .is @xmath106g(k)g(k\!-\!p)\notag\\ % % & & = \frac{1}{6}c_{a}g^{2}t^{2}\left[\frac{(p_{0})^{2}}{p^{2}}-\frac{p_{0}p^{2}}{2p^{3}}\ln\frac{p_{0}+p+i\epsilon}{p_{0}-p - i\epsilon}\right]\notag\\ % & & + \frac{i}{4}c_{a}g^{2}\int\frac{d^{4}k}{(2\pi)^{4}}\left[8\frac{n\!\cdot\ ! p}{n\!\cdot\!(k\!-\!p)}(k^{2}\!-\!p^{2})\!-\!8\frac{n\!\cdot\ ! p}{n\!\cdot\!k}(k^{2}\!-\!2k\!\cdot\ !p)\right]g(k)g(k\!-\!p)\;.\end{aligned}\ ] ] the last two light cone terms in the forth line come from the the tensors of the gluon propagators @xmath107 and @xmath108/((n\cdot ( k - p))$ ] .replace @xmath92 by @xmath93 in the first light cone term , and the sum of the two light cone terms is @xmath109 at zero temperature , the divergence of the integral calculation of one loop diagram in fig.[gloun - loop and tadpole ] has been renormalized successfully . herewe only consider the contribution at finite temperature .use the ml prescription of @xmath110 in eq .. by calculating with the contour integral , we find there is no divergence , and the power of the part is @xmath111 order , which can be ignored in the result .the proof is in the appendix .similarly , the longitudinal part of the gluon self energy in the htl approximation is obtained by @xmath112g(k)g(k\!-\!p)\notag\\ % & \!=\!&-\frac{1}{3}c_{a}g^{2}t^{2}\left[\frac{p^{2}}{p^{2}}-\frac{p_{0}p^{2}}{2p^{3}}\ln\frac{p_{0}+p+i\epsilon}{p_{0}-p+i\epsilon}\right]+\frac{i}{2}c_{a}g^{2}\int\frac{d^{4}k}{(2\pi)^{4}}\frac{1}{p^{2}}\left[2\frac{n\!\cdot\ ! p}{n\!\cdot\ !( k - p)}\!-\!2\frac{n\!\cdot\ ! p}{n\!\cdot\!k}\right]\;.\end{aligned}\ ] ] there are two light cone terms in the third line . by calculating , the integral with the light cone terms is zero .@xmath113=0\;.\ ] ] during the calculation , we set the momentum @xmath24 is on the positive direction of @xmath114 axis .@xmath115 in eq .is on the negative direction of @xmath114 axis .the angle between @xmath116 and @xmath24 is @xmath117 .here we use the kelydsh representation in the real time formalism and consider the @xmath118 contribution . in the htl approximation ,the internal momentum @xmath92 is soft , and the external momentum @xmath96 is hard .adding up the results from the longitudinal and transverse parts of the quark loop in fig.[quark - loop ] , the gluon loop and the gluon tadpole in fig.[gloun - loop and tadpole ] in the htl approximation , we obtain the following expression , @xmath119\notag\;,\\ % \frac{1}{n_{p}^{2 } } \pi_{l}(p)&\!=\!&-\frac{1}{3}(c_{a}+\frac{1}{2}n_{f})g^{2}t^{2}\left[\frac{p^{2}}{p^{2}}-\frac{p_{0}p^{2}}{2p^{3}}\ln\frac{p_{0}+p+i\epsilon}{p_{0}-p+i\epsilon}\right]\;.\end{aligned}\ ] ] the factor @xmath120 in the coefficient @xmath121 stems from that these integrals @xmath122 and @xmath123 which have different distribution functions .we find we get the same result of the transverse and longitudinal htl gluon self energy in the light cone gauge and the coulomb gauge , although this two gauges have different projection tensors . in the static limit @xmath30, the longitudinal and transverse parts of gluon htl self energy in the light cone gauge reduce to @xmath124 in the static limit @xmath30 , the longitudinal and transverse parts of the gluon htl self energy in the coulomb gauge reduces to @xmath125 we think the axial vector @xmath126 in the light cone gauge is rotated with respect to the axial vector @xmath58 in the coulomb gauge , so that it gives rise to some changes . in the static limit , @xmath127 of the gluon htl self energy in the coulomb gauge in eq .is @xmath128 where the axial vector @xmath58 in eq .. because the axial vector @xmath66 in the light cone gauge is different from @xmath66 in the coulomb gauge , we compare @xmath37 in the light cone gauge with @xmath37 in the coulomb gauge . in the static limit , @xmath37 of the gluon htl self energy in the light conegauge in eq .is @xmath129 where the axial vector @xmath126 in eq .. the external momentum @xmath130 , @xmath131 is on the @xmath114 axis .the result is the same as that in the coulomb gauge .the htl resumed gluon propagator in the light cone gauge is @xmath132}+\frac{\frac{n^{\mu}n^{\nu}p^{2}}{(n\cdot p)^{2}}}{p^{2}+m_{d}^{2}\left[\frac{p^{2}}{p^{2}}-\frac{p_{0}p^{2}}{2p^{3}}\ln\frac{p_{0}+p+i\epsilon}{p_{0}-p+i\epsilon}\right]}\;.\end{aligned}\ ] ] the transverse and longitudinal spectral functions are expressed as @xmath133+\beta_{t}(p)\;,\notag\\ % \rho_{l}(p)&=&2\pi z_{l}\mathrm{sgn}(p_{0})\big[\delta(p_{0}-w_{l})+\delta(p_{0}+w_{l})\big]+\beta_{l}(p)\;.\end{aligned}\ ] ] the spectral function @xmath134 is made up of the pole term and the cut term @xmath135 . for @xmath136 space - like , i.e @xmath137, the function @xmath138 generates the imaginary part , @xmath139 so the cut terms of the transverse and longitudinal part of the htl resumed gluon propagator in the light cone gauge are obtained by @xmath140\bigg]^{2}\!+\!\frac{1}{16}m_{d}^{4}\pi^{2}\frac{(p_{0})^{2}p^{2}p^{2}}{p^{6}}}\notag\;,\\ % & & \beta_{l}(p)\!=\!\frac{-\pi m_{d}^{2}\frac{p_{0}}{p}\theta(p^{2}\!-\!p_{0}^{2})}{\bigg [ p^{2}\!+\!m_{d}^{2}\big[1\!-\!\frac{p_{0}}{2p}\ln|\frac{p_{0}\!+\!p}{p_{0}\!-\!p}|\big]\bigg]^{2}\!+\!\frac{1}{4}m_{d}^{4}\pi^{2}\frac{(p_{0})^{2}}{p^{2}}}\;,\end{aligned}\ ] ] where @xmath29 is the step function , and the debye screening mass @xmath141 . via the expression ofthe htl resumed gluon propagator in the light cone gauge , the transverse dispersion relation is given by @xmath142=0\;,\ ] ] where @xmath143 is the solution of the above transverse dispersion relation .the longitudinal dispersion relation is given by @xmath144=0\;,\ ] ] where @xmath145 is the solution of the above longitudinal dispersion relation .obviously the transverse and longitudinal parts have the same dispersion relation as that of the htl resumed gluon propagator in the coulomb gauge .however , this two kinds of gauge have different expression of the transverse and longitudinal projection tensor , and then have different expression of the htl resumed gluon propagator . due to the same dispersion relation , for more analyses you can refer to @xcite .the residue for the transverse part is @xmath146_{p_{0}=\omega_{t}(p)}\bigg)^{-1}\notag\\ % & & = \frac{\omega_{t}(\omega_{t}^{2}-p^{2})}{m_{d}^{2}\omega_{t}^{2}-(\omega_{t}^{2}-p^{2})^{2}}\;.\end{aligned}\ ] ] the residue for the longitudinal part is @xmath147_{p_{0}=\omega_{l}(p)}\bigg)^{-1}\notag\\ % & & = \frac{\omega_{l}(\omega_{l}^{2}-p^{2})}{p^{2}(p^{2}-m_{d}^{2}-\omega_{l}^{2})}\;.\end{aligned}\ ] ] about the transverse and longitudinal residues you can find more discussion in @xcite too .below is the proof to tell the reason why we can get the same dispersion relation .from the famous ward identity , @xmath148 we can get the below relation , @xmath149 where the external momentum @xmath130 , @xmath150 , and @xmath24 is on the positive direction of z axis . using the above relation ,the longitudinal part of the gluon htl self energy can become @xmath151 where @xmath7 is the longitudinal projection tensor in the light cone gauge in eq ., @xmath37 is the longitudinal part of the htl gluon self energy in the coulomb gauge in eq . , and the axial vector in the light cone gauge @xmath126 .the result tells us the longitudinal part of the gluon htl self energy in the light cone gauge is the same as that in the coulomb gauge .similarly , the transverse part of the gluon htl self energy can become @xmath152\pi_{\mu\nu}(p)\notag\\ % & & = -\frac{1}{2}m_{d}^{2}+\frac{1}{2}\frac{p^{2}}{(p_{3})^{2}}\pi^{00}(p)\;.\end{aligned}\ ] ] where @xmath6 is the transverse projection tensor in the light cone gauge in eq .. so the transverse part of the gluon htl self energy in the light cone gauge is the same as that in the coulomb gauge . in the proof , we use the famous ward identity .@xmath24 is on the positive direction of the z axis , and @xmath115 is on the negative direction of z axis .we can get the same transverse and longitudinal parts of the gluon htl self energy in the two gauges , and finally get the same dispersion relation of the transverse and longitudinal parts .+ the quark self energy in fig.[quarkselfenergy ] can be expressed as @xmath153 where @xmath154 is the color factor , @xmath155 is the bare quark propagator , and @xmath156 is the bare gluon propagator . the retarded quark self energy in the real time formalism is expressed as @xmath157\;.\end{aligned}\ ] ] using the relation in eq . , we get @xmath158\notag\\ % & & = \frac{1}{2}[\tilde{\delta}_{r}\delta_{s}+\tilde{\delta}_{s}\delta_{r}]\;,\end{aligned}\ ] ] where @xmath159 and @xmath160 respectively represent the green functions of quark and gluon.the terms @xmath161 and @xmath162 are both zero temperature parts , and we neglect them here. the quark self energy is decomposed into two parts , @xmath163\big]\;,\notag\\ % & & b(p_{0},p)=\frac{1}{4p^{2}}\big[p^{2}\mathrm{tr}[\gamma_{0}\sigma_{r}]-\gamma_{0}\mathrm{tr}(\cancel{p}\sigma_{r})\big]\;.\end{aligned}\ ] ] with the above relation , we can do the following calculation , @xmath164=ic_{f}g^{2}\int\frac{\mathrm{d}^{4}k}{(2\pi)^{4}}\big[-8k\cdot p+\frac{n\cdot p}{n\cdot k}(-8k^{2}+16k\cdot p)\big]\frac{1}{2}\big[\tilde{\delta}_{r}(p - k)\delta_{s}(k)+\tilde{\delta}_{s}(p - k)\delta_{r}(k)\big]\;.\end{aligned}\ ] ] replace @xmath92 by @xmath93 in this term @xmath165 , this expression becomes @xmath166=ic_{f}g^{2}\int\frac{\mathrm{d}^{4}k}{(2\pi)^{4}}\bigg[\big[8(k\cdot p - p^{2})+\frac{8n\cdot p}{n\cdot(k - p)}(k^{2}-p^{2 } ) \big]\frac{1}{2}\tilde{\delta}_{s}(k)\delta_{r}(p - k)\notag\\ % & & + \big[-8k\cdot p+\frac{n\cdot p}{n\cdot k}(-8k^{2}+16k\cdot p)\big]\frac{1}{2}\tilde{\delta}_{r}(p - k)\delta_{s}(k)\bigg]\notag\\ % & & = ic_{f}g^{2}\int\frac{\mathrm{d}^{4}k}{(2\pi)^{4}}\bigg[\big[4(k\cdot p - p^{2})\tilde{\delta}_{s}(k)\delta_{r}(p - k)-4k\cdot p\tilde{\delta}_{r}(p - k)\delta_{s}(k)\big]\notag\\ % & & + \big[\frac{4n\cdot p}{n\cdot(k - p)}(k^{2}-p^{2 } ) \tilde{\delta}_{s}(k)\delta_{r}(p - k)+\frac{n\cdot p}{n\cdot k}(-4k^{2}+8k\cdot p)\tilde{\delta}_{r}(p - k)\delta_{s}(k)\big]\bigg]\notag\\ % & & = 4m_{f}^{2}\;,\end{aligned}\ ] ] in the htl approximation , we can prove there is no spurious divergence from the light cone terms in the forth line , and these finite terms is power suppressed than the covariant terms , so we ignore the light cone terms . in the same way, we can get @xmath167=c_{f}g^{2}\int\frac{\mathrm{d}^{4}k}{(2\pi)^{4}}8[-k_{0}+k_{0}\frac{n\cdot p}{n\cdot k}-\frac{n_{0}}{n\cdot k}(k^{2}-k\cdot p)]\frac{1}{2}\big[\tilde{\delta}_{r}(p - k)\delta_{s}(k)+\tilde{\delta}_{s}(p - k)\delta_{r}(k)\big]\notag\\ % & & = c_{f}g^{2}\int\frac{\mathrm{d}^{4}k}{(2\pi)^{4}}\bigg[4(k_{0}-p_{0})\tilde{\delta}_{s}(k)\delta_{r}(p - k)-4k_{0}\tilde{\delta}_{r}(p - k)\delta_{s}(k)\notag\\ % & & + 4\big[(k_{0}-p_{0})\frac{n\cdot p}{n\cdot(k - p)}+\frac{n_{0}}{n\cdot(k - p)}(k^{2}-k\cdot p)\big]\tilde{\delta}_{s}(k)\delta_{r}(p - k)\notag\\ % & & + 4\big[k_{0}\frac{n\cdot p}{n\cdot k}-\frac{n_{0}}{n\cdot k}(k^{2}-k\cdot p)\big]\tilde{\delta}_{r}(p - k)\delta_{s}(k)\bigg]\notag\\ % & & = 2m_{f}^{2}\frac{1}{p}\ln\frac{p_{0}+p+i\epsilon}{p_{0}-p+i\epsilon}\;.\end{aligned}\ ] ] in the htl approximation , it can be proved that there is no spurious divergence from the light cone terms in the third and forth lines , and these finite terms are power suppressed than the covariant terms , so we ignore these light cone terms too .so the result shows the quark htl self energy is independent of the light cone gauge , and it is the same as that of covariant gauge @xcite . with the quark htl self energy, we can derive the same quark resumed propagator .the htl resumed quark propagator is @xmath168\;,\end{aligned}\ ] ] where the effective quark mass @xmath169 in qcd , and the notation @xmath170 .the damping rates of the heavy fermion have been done @xcite,@xcite . in a similar way, we use the htl resumed gluon propagator in the light cone gauge to calculate the damping rates of the hard quark and gluon in the static limit . by above analyses, we know the htl resumed gluon propagator in the coulomb gauge and the light cone gauge have the same denominator and different projection tensors in the nominator .but we can prove in general case , using the htl resumed gluon propagator in the two gauges get the same result of the damping rates of the hard quark and gluon .the quark self energy in fig.[quarkdamping ] is expressed as + @xmath171\;,\ ] ] where @xmath172 is the htl resumed gluon propagator in the light cone gauge , and @xmath173 is the bare quark propagator . in the real time formalism ,the retarded quark self energy is expressed as @xmath174[\tilde{\delta}_{11}(q - p)\delta_{11}^{\mu\nu}(p)-\tilde{\delta}_{12}(q - p)\delta_{12}^{\mu\nu}(p)]\;.\end{aligned}\ ] ] using the relation of the keldysh representation in eq . , we have @xmath175\;.\ ] ] by power counting the leading contribution comes from the first term in the bracket .the symmetric propagator of the soft gluon in the light cone gauge is @xmath176[1 + 2n_{b}(p_{0})]\;.\ ] ] the internal momentum @xmath96 is soft , @xmath177 , @xmath178 using the below equation , we can calculate the imaginary part of the quark self energy , @xmath179=-\pi\mathrm{sgn}(q_{0}-p_{0})\delta[(q - p)^{2}]\;.\ ] ] in the integral , we use the @xmath180 function to integrate out @xmath181 , @xmath182=\frac{1}{2pq}\delta[\cos\theta-\frac{p_{0}}{p}+\frac{p^{2}}{2pq}]\approx\frac{1}{2pq}\delta[\cos\theta-\frac{p_{0}}{p}]\;,\ ] ] where the term @xmath183 , which we can ignore .the transverse projection tensor in the light cone gauge is @xmath184 where the internal soft momentum @xmath185 . by calculatingwe can find the relation @xmath186 , so we have @xmath187 the longitudinal and transverse spectral functions in the static limit @xmath188 are @xmath189 where we find no static magnetic screening . with above equations , we can calculate the imaginary part of @xmath190 $ ] , @xmath191\big]\notag\\ % & & = -4\pi^{2}c_{f}g^{2}\int\frac{\mathrm{d^{4}}p}{(2\pi)^{4}}\big[4q_{\perp}^{2}\rho_{t}(p)+4\frac{(n\cdot q)^{2}p^{2}}{(n\cdot p)^{2}}\rho_{l}(p)\big]n_{b}(p_{0})\mathrm{sgn}(q_{0}-p_{0})\delta[(q - p)^{2}]\notag\\ % & & = -\frac{1}{2\pi}c_{f}g^{2}tq[1 + 2\ln\frac{1}{g}]\;.\end{aligned}\ ] ] the damping rate for the hard quark is @xmath192\big]\notag\\ & & = \frac{1}{4\pi}c_{f}g^{2}t[1 + 2\ln\frac{1}{g}]\;.\end{aligned}\ ] ] the first term in the bracket comes from the longitudinal contribution of the htl resumed gluon propagator , and the second term comes from the transverse contribution . in the transverse partthere is an ir - cutoff , which results from the magnetic mass of the order @xmath193 .+ we can express the gluon loop in fig.[gluondamping ] as @xmath194 where @xmath195 is the htl resumed gluon propagator in the light cone gauge in eq ., @xmath196 is the bare gluon propagator in the light cone gauge , and the three gluon vertexes @xmath197 and @xmath198 are given in eq.. using the relation of the keldysh representation in , we have @xmath199\;.\end{aligned}\ ] ] similarly , by power counting the second term gives the leading contribution .the three gluon vertexes are simplified into @xmath200\notag\;,\\ % & & v_{\beta\gamma\alpha}(p , q\!-\!p,\!-\!q)\approx gf^{cdb}[-q_{\alpha}g_{\beta\gamma}+2q_{\beta}g_{\gamma\alpha}-q_{\gamma}g_{\alpha\beta}]\;,\end{aligned}\ ] ] where the external momentum @xmath201 is hard , and the internal momentum @xmath96 is soft , which is ignored in the above expression .the nominator of the bare propagator in the light cone gauge @xmath202 is @xmath203 in the calculation , we have @xmath204 where the external hard momentum @xmath201 is on shell , @xmath205 . due to the equation , we can simplify the calculation .now we calculate the imaginary part of the transverse part @xmath206 , @xmath207n_{b}(p_{0})\mathrm{sgn}(q_{0}-p_{0})\delta[(q - p)^{2}]\notag\\ % & & = -\frac{1}{8\pi}c_{a}g^{2}tq[1 + 2\ln\frac{1}{g}]\;.\end{aligned}\ ] ] the damping rate for the hard gluon is @xmath208}=\frac{1}{16\pi}c_{a}g^{2}t [ 1 + 2\ln\frac{1}{g}]\;.\ ] ] the first term in the bracket stems from the longitudinal contribution of the htl resumed gluon propagator , and the second term stems from the transverse contribution . for the transverse partwe take an ir - cutoff too .we work out the result about the damping rates of the hard quark and gluon in the static limit .below we can prove the general case that the expression about the transverse and longitudinal projection in the above calculation of the damping rates is the same in the two gauges . via this function@xmath209 $ ] , we have the relation about @xmath181 , @xmath210 where the internal soft momentum @xmath130 , the angle @xmath117 is arbitrary .when @xmath150 , we have @xmath211 where @xmath117 is the angle between @xmath24 and @xmath212 .the expression of the longitudinal part in the calculation of the damping rate is @xmath213 when @xmath214 , we have @xmath215 the expression of the longitudinal part in the calculation is @xmath216 we can find , in the calculation we have the same expression about the longitudinal projection in the two gauges .the transverse projection tensor of the htl resumed gluon propagator in the coulomb gauge is @xmath217 by calculating , we have @xmath218 where the internal soft momentum @xmath130 .so we can get the same transverse projection tensors in the two gauges .we have the simple expression about the transverse projection tensor , @xmath219 here we show the transverse and longitudinal projection tensors in the above calculation of the damping rates are the same in the two gauges when the transverse momentum of the internal momentum @xmath96 is zero .so we can get the same result about the damping rates of the hard quark and gluon in the two gauges .in this paper , we have derived the htl resumed gluon propagator in the light cone gauge , and presented the results for the transverse and longitudinal htl gluon self energy in the light cone gauge .we show the quark htl energy is independent of the light cone gauge , and get the same htl resumed quark propagator .we calculate the damping rates of the hard quark and gluon in the static limit with this propagator .although the longitudinal and transverse expression of the gluon htl self energy both have the light cone terms with @xmath110 , there is no divergence in the transverse and longitudinal parts , which are different from the case at zero temperature . by calculating ,we can find we obtain the same transverse and longitudinal gluon htl self energies in the light cone gauge and the coulomb gauge .but in the two gauges we have different transverse and longitudinal projection tensors .we think the axial vector @xmath126 have the longitudinal part and is rotated with respect to the axial vector @xmath58 in the coulomb gauge , which brings about changes .correspondingly , in the static limit , we compare the component @xmath37 of the gluon htl self energy in the light cone gauge with @xmath37 in the coulomb gauge , and find the result in the light cone gauge is the same as that in the coulomb gauge .we show the transverse and longitudinal spectral functions of the htl resumed gluon propagator in the light cone gauge and the transverse and longitudinal dispersion relation .however , in the light cone gauge , the transverse and longitudinal projection tensors are both based on the the axial vector @xmath66 , and they have different expression from the transverse and longitudinal projection tensors in the coulomb gauge , so the expression of the htl resumed gluon propagator in the light cone gauge is different from that in the coulomb gauge . with the htl resumed gluon propagator in the light cone gauge ,we calculate the damping rates of the hard on shell quark and gluon in the static limits .we demonstrate in general case , we can get the same result about the damping rates in the two gauges .at zero temperature , this kind of the integral in eq . has divergence , which has been renormalized successfully . however , we find this integral does not have the divergence in the htl approximation .we use the contour integral to substitute the frequency sum of boson at finite temperature @xcite , @xmath220{\infty}{\sum}f(p_{0}=2n\pi ti)=\frac{1}{2\pi i}\int_{-i\infty}^{i\infty}\mathrm{d}p_{0}\frac{1}{2}[f(p_{0})+f(-p_{0})]+\frac{1}{2\pi i}\int_{-i\infty+\epsilon}^{i\infty+\epsilon}\mathrm{d}p_{0}[f(p_{0})+f(-p_{0})]\frac{1}{e^{\beta k_{0}}-1}\;.\end{aligned}\ ] ] this equation has the zero temperature part and the finite temperature part , and we only consider the the finite temperature part here . below we prove this integral at finite temperature does not have the divergence in eq . ,@xmath221 where @xmath222 , the angle @xmath117 is the angle between @xmath116 and @xmath24 , and the axial vector @xmath126 .the first term in the bracket has poles at @xmath225 and @xmath226 , and the residues of the two poles do not have the term of @xmath227 , so there is no divergence at @xmath228 .the second term has poles at @xmath225 , @xmath229 and @xmath230 .the residues with the two poles of @xmath225 and @xmath229 contain the terms of @xmath227 , but the sum of the two residues is finite when @xmath231 , @xmath232 the second term in the bracket has poles at @xmath225 and @xmath230 , and the residues of the two poles do not have the term of @xmath234 , so there is no divergence at @xmath235 .the first term has poles at @xmath225 , @xmath236 and @xmath226 .the residues with the two poles of @xmath225 and @xmath236 contain the terms of @xmath234 , but the sum of the two residues is finite when @xmath237 , @xmath238

<S> the expression of the htl resumed gluon propagator in the light cone gauge is derived . in the real time mechanism , using the mandelstam leibbrant prescription of @xmath0 , we calculate the transverse and longitudinal parts of the gluon htl self - energy and prove the transverse and longitudinal parts do not have divergence . </S><S> we also calculate the quark self energy in the htl approximation , and find it gauge independent . in the static limit we calculate the damping rates of the hard quark and gluon with this htl resumed gluon propagator . </S>

0902.2934

inflation is considered as the standard paradigm for providing an explanation for the observed spatial flatness , homogeneity and isotropy on large scales as well as for the observed perturbations on smaller scales @xcite . from the point of view of particle physics , realizing inflation requires an extension of the standard model ( sm ) . in order to connect inflation to particle physics motivated extensions of the sm , for example to grand unified theories ( guts ) ( see e.g. @xcite ) or to models towards solving the flavor puzzle @xcite , hybrid - type inflation models @xcite are particularly suitable . since in general high energies close to the gut scale are involved in models of this type , they are typically embedded in a supersymmetric ( susy ) framework .however , when trying to embed f - term hybrid inflation ( or inflation in general ) in local susy , i.e. supergravity theories , a severe problem arises : a general sugra theory seems to almost inevitably generate a large mass for the inflaton of the order of the hubble scale @xmath2 which results in a slow - roll parameter @xmath3 , spoiling inflation .this problem is referred to as the @xmath0-problem of sugra inflation @xcite .several approaches to solve it have been discussed in the literature so far : for instance , since the fields involved in hybrid inflation typically have values well below the planck scale @xmath4 , one can expand the khler potential in powers of field values over @xmath4 and finally adjust ( tune ) the expansion parameters to protect the inflaton direction from obtaining large mass corrections ( see e.g. @xcite ) .a different approach is to use a so - called heisenberg symmetry @xcite , which involves the inflaton as well as a modulus field . here, the flatness of the inflaton potential is guaranteed by the symmetry .it has been shown in @xcite how this modulus field can be stabilized successfully during inflation in a novel class of models .the third approach towards solving the @xmath0-problem is to apply a nambu - goldstone - like shift symmetry @xcite which can in principle protect the inflaton part of a complex scalar field against obtaining sugra mass corrections .however , it has turned out that in conventional ( `` standard '' ) models of susy hybrid inflation @xcite , the inflaton direction always develops a large tachyonic mass . to cure this instability , a coupling between the inflaton and the moduli sector has been considered , which however again leads to severe problems for realizing inflation as discussed in @xcite . in this paper we propose a new model of sugra hybrid inflation where a nambu - goldstone - like shift symmetry is applied to guarantee the flatness of the inflaton potential .in contrast to the `` standard '' models of hybrid inflation in sugra , our model has the novel property of having @xmath5 during inflation . due to this property ,potentially dangerous couplings between the inflaton and the moduli sector are absent .in addition , the problem of having a large tachyonic mass at tree - level for the inflaton direction is also cured in our model .we point out that this model forms part of a larger class of models which have the same desirable property .furthermore , we note that in this class of models the right - handed sneutrino is a natural candidate for being the inflaton @xcite . within the proposed class of models we address a second challenge to sugra models of hybrid inflation , namely the consistency between the predicted spectral index and the latest wmap results @xcite , which favor a value of @xmath1 for small tensor - to - scalar ratio @xmath6 .in fact , `` standard '' susy hybrid inflation models with flat inflaton potential at tree - level predict a spectral index above @xmath7 @xcite .since this is slightly outside the @xmath8-range given by wmap , it raises the question whether a lower spectral index can be realized . in this context, an expansion of the khler potential has been considered and it has been shown that a lowering of @xmath9 to @xmath1 can be achieved @xcite by a small term which , however , already introduces a curvature of the potential at tree - level .another possibility , pointed out in @xcite , is that a cosmic string component could allow for a more blue - tilted power spectrum of the cosmic microwave background ( cmb ) .other approaches suggest , e.g. , to involve field values well above the planck - scale @xcite . herewe propose and analyze a new possibility to lower the spectral index @xmath9 in sugra hybrid inflation models based on a new term in the khler potential involving the waterfall field and the field which gives rise to the vacuum energy by its f - term . while the inflaton potential remains flat at tree - level, the new term changes the quantum loop corrections and therefore the form of the scalar potential .we show that without any tuning of parameters , the spectral index can be lowered and the wmap best - fit value for @xmath9 can be realized .the paper is organized as follows : in section [ the model ] , we describe the model given in terms of the superpotential and khler potential and the chiral superfield content .additionally , the shift symmetry in the inflaton sector is introduced .section [ effective potential ] is dedicated to the investigation of the loop - corrected scalar potential and the tree - level mass spectrum to show that all other directions in field space are stable during inflation .section [ predictions ] contains the calculation of observables obtained from the effective potential and finally , in section [ summary and conclusions ] , we summarize and present our conclusions .to solve the @xmath0-problem in sugra hybrid inflation using a nambu - goldstone - like shift symmetry , we propose the following minimal setup described by the superpotential @xmath10 with @xmath11 being the superfield which contains the inflaton , and the khler potential @xmath12 the gauge singlet supermultiplet @xmath11 contains the slow - rolling inflaton condensate as the pseudoscalar component which will turn out to be a tree - level flat direction .@xmath13 is a static , so - called driving field which is fixed at zero during inflation and only contributes the large vacuum energy density during inflation by its f - term .@xmath14 is the waterfall field which also stays at zero during inflation , but gets destabilized when the inflaton reaches a critical value .both @xmath13 and @xmath14 are also gauge singlets in our minimal model .the form of the superpotential is motivated by the model of @xcite and the classes of models considered in @xcite , where the right - handed sneutrino has been identified as the inflaton field .it has the desirable property of having vanishing superpotential @xmath15 both during and after inflation , and all the derivatives vanish , except for the one @xmath16 , as we will discuss below .as an aside , we would like to note at this point that in `` conventional '' sugra models of hybrid inflation the superpotential consists of only the first term @xmath17 in eq .( [ tribridsuperpotential ] ) and the superfield @xmath13 contains the inflaton . the khler potential for @xmath13 is chosen to be minimal , i.e. @xmath18 .large sugra contributions to the inflaton mass get cancelled by this special choice of @xmath19 , however not protected by a fundamental symmetry . to use a shift symmetric khler potential in combination with the above superpotential to protect the flatness of the inflaton direction ,the scalar potential would always contain a tachyonic direction for one of the components of the complex inflaton field .one possibility to stabilize it could be the introduction of a coupling to another sector ( i.e. to a modulus field ) .however , as has been argued in @xcite , such a coupling is very problematic for successfully realizing inflation . on the contrary ,the model we propose is free of these problems . turning to the khler potential , it is symmetric under the `` shift '' of @xmath11 by @xmath20 where @xmath21 is an arbitrary dimensionful parameter .shift symmetry invariance of the khler potential requires it to be a function @xmath22 . for generality , and since all field values are well below the planck scale , we use an expansion in the absolute values squared of the fields @xmath14 and @xmath13 and in even powers of @xmath23 , where higher order operators are suppressed by suitable powers of some scale @xmath24 .the scale @xmath24 can also be viewed as the generation scale of the effective operators , and can well be below the planck scale @xmath25 .the fact that the shift symmetry of the khler potential provides a solution to the @xmath0-problem in sugra inflation can be seen by looking at the full ( f - term ) scalar potential given by to unity . ]@xmath26\,,\ ] ] where the definition @xmath27 has been used .the indices @xmath28 denote different scalar fields and the lower indices on the superpotential or khler potential denote the derivatives with respect to the scalar component of the chiral superfield or their conjugate where a bar is involved .the inverse khler metric is defined as @xmath29 .the @xmath0-problem is the tendency that in sugra an inflaton mass of the order of the hubble scale is typically generated , which spoils inflation . because of the shift symmetry the exponential sugra contribution in eq .is independent of the imaginary part of the @xmath11 field , identified as the inflaton direction .this solves the usual @xmath0-problem in sugra inflation . in conventional hybrid inflation models@xmath30 during inflation and this can also induce a large tachyonic mass for the inflaton field at tree - level from the @xmath31 term in eq . .this problem is automatically avoided in our setup due to the special property of having @xmath15 during inflation .furthermore , this property also eliminates certain couplings between the moduli sector and the inflaton which have turned out to be problematic for successfully realizing inflation @xcite .the khler metric is given by the second derivatives w.r.t .the fields and their conjugates . in the limit @xmath32 , it diagonalizes and in @xmath33-basis is given by @xmath34 with @xmath35 , we recover canonically normalized kinetic terms for @xmath36 , @xmath37 and @xmath38 .the shift symmetry , together with the form of the superpotential , protects the imaginary part of @xmath11 from obtaining any mass term at tree - level .finally , the shift symmetry is slightly broken by the effective operator @xmath39 of the superpotential in eq .( [ tribridsuperpotential ] ) .while the inflaton potential is still flat at tree - level , this term gives rise to a slope of the inflaton potential at loop - level , as will be discussed in section [ effective potential ] .we note that the minimal superpotential of eq .( 1 ) may be justified by introducing additional symmetries and spurion fields , as outlined in the appendix .we would also like to note that the above example model can be generalized to a whole class of models , along the line of @xcite , by allowing for variants of the form of the superpotential .furthermore , a specific ( shift symmetric ) khler potential may be chosen instead of the expansion we have used here for generality .the main point we would like to emphasize is that in our class of models with @xmath40 ( but @xmath41 ) during inflation , a shift symmetry naturally solves the @xmath0-problem , and that in the same class of models several problems of earlier approaches , related to tachyons or to dangerous couplings to moduli fields , are automatically avoided .the full f - term scalar potential is given by eq . .in the following , we will assume that the fields @xmath42 and @xmath43 have already settled to their minima at @xmath44 .this is justified because , as we will show below , the fields can easily have masses larger than the hubble scale @xmath2 . from eq .we see that , in the minimum , we have @xmath45 such that all fields are already canonically normalized .the vacuum energy density is then given by the ( f - term ) tree - level potential which reads @xmath46 and is flat in @xmath47-direction .using the supercosmology code @xcite , the scalar and pseudoscalar mass matrices are calculated to be diagonal and the mass spectrum is given by @xmath48 the directions in field space different from the inflaton are stable provided that their masses are larger than the hubble scale .this requirement leads to the constraints @xmath49 and @xmath50 .for the fermions , we calculate the sugra mass matrix from @xmath51 this is also diagonal in the minimum and has only one non - vanishing eigenvalue during inflation . the fermion mass matrix squared in @xmath33-basisis obtained to be @xmath52 since for both the scalars and pseudoscalars , as well as for the fermions , only the waterfall sector @xmath14-components contribute @xmath47-dependent masses , we only consider their contributions to the one - loop potential .we denote their masses by @xmath53\,,\\ m^2_{\text{p}}&=2\,\kappa^2m^2\,\left[x+m^2\left(1-\frac{\kappa_{sh}}{\lambda^2}\right)/2 + 1\right]\,,\\ m^2_{\text{f}}&=2\,\kappa^2m^2\,x\ , , \end{split}\ ] ] where we have defined the new variable @xmath54 hence , the critical value of @xmath47 at which the waterfall field destabilizes can be calculated from @xmath55 and is given by @xmath56 the coleman - weinberg one - loop contribution @xcite to the effective potential using the squared masses in eq. reads @xmath57\ , , \end{split}\ ] ] where @xmath58 is the renormalization scale which we fix to @xmath59 .the complete effective scalar potential we consider thus reads @xmath60 for a particular choice of parameters , we have plotted the potential in fig .[ shiftpot ] .note , that the shape of the potential is of hilltop - type , generated purely by radiative corrections . in order to achieve at least 60 e - folds of inflation, the initial condition for the inflaton field is that it has to start between @xmath61 , about @xmath62 e - folds before the end of inflation , and the top of the hill ( i.e. the local maximum of the potential ) . we note that , although certain initial conditions have to be satisfied for inflation to start , this does not impose any tuning since the relative amount of field space where the initial conditions are suitable is quite extended in our model , about @xmath63 ( c.f .[ shiftpot ] ) .graphical illustration of the one - loop effective potential for @xmath47 with typical values of the field 60 e - folds before the end of inflation @xmath61 and at the critical value @xmath64 where inflation ends .the parameter choice in this example was @xmath65 , @xmath66 , @xmath67 and @xmath68 . ]in order to make predictions for the observable quantities in the cmb radiation , we have to integrate the equations of motion ( eoms ) for @xmath47 from @xmath64 to the value @xmath61 where the relevant scales leave the horizon , i.e. about @xmath62 e - folds before the end of inflation . in slow - roll approximation ,the eoms read @xmath69 the slow - roll parameters are defined as @xmath70 in terms of the slow - roll parameters , the scalar spectral index @xmath9 , the tensor - to - scalar ratio @xmath6 and the running of the scalar spectral index @xmath71 are given by @xmath72 in addition , the amplitude of the curvature perturbations can be obtained from @xmath73 where in our case @xmath74 from eq . .all of the above quantities have to be evaluated at the field value @xmath61 , about @xmath62 e - folds before the end of inflation .the scale @xmath75 is normalized to fit the observation of the amplitude , which at @xmath76% cl is measured to be @xmath77 .fixing the parameters @xmath78 and @xmath58 as above , we vary @xmath79 for different values of @xmath80 while keeping fixed @xmath66 . the results for the observables calculated in this way are displayed in figs .[ paramspace ] and [ paramspace2 ] .observables @xmath9 and @xmath6 for @xmath66 and @xmath81 ( full line ) , @xmath65 ( dashed line ) , @xmath82 ( dotted dashed line ) , @xmath83 ( dotted line ) depending on @xmath84 .the shaded region in the first plot highlights the latest wmap @xmath8-range of the spectral index.,title="fig : " ] + observables @xmath9 and @xmath6 for @xmath66 and @xmath81 ( full line ) , @xmath65 ( dashed line ) , @xmath82 ( dotted dashed line ) , @xmath83 ( dotted line ) depending on @xmath84 .the shaded region in the first plot highlights the latest wmap @xmath8-range of the spectral index.,title="fig : " ] note , that for a relatively wide and generic range of the parameter @xmath79 , the predicted spectral index @xmath85 lies within the @xmath8-range @xmath86 as taken from the latest wmap data @xcite .the tensor - to - scalar ratio is very small @xmath87 as typical for hybrid and small - field models of inflation and easily satisfies the observational upper bound @xmath88 ( 95% cl ) . for a running of the spectral index , the current observational evidence is still not considered as significant , @xmath89 ( 68% cl ) .our prediction @xmath90 does not lie within this range , which we do not consider as a problem for our model due to the low accuracy of the measurement .we note that the scale @xmath75 obtained from the cobe - normalization for the amplitude of the power spectrum predicts @xmath91 , close to the gut scale . in this simplest type of effective single - field model, we do not expect large non - gaussianities .running of the spectral index @xmath92 and the scale @xmath75 for the same choice of parameters as in fig .[ paramspace].,title="fig : " ] + running of the spectral index @xmath92 and the scale @xmath75 for the same choice of parameters as in fig .[ paramspace].,title="fig : " ] +in this paper , we have proposed a solution to the @xmath0-problem in supergravity ( sugra ) hybrid inflation using a nambu - goldstone - like shift symmetry within a new class of models ( see eqs .( [ tribridsuperpotential ] ) , ( [ kaehlerpotential ] ) and generalizations discussed in the text ) . the flatness of the tree - level inflaton potential is ensured by shift symmetry invariance of the khler potential , while a small symmetry breaking term in the superpotential gives rise to a slope of the potential at loop - level .due to the special property of the class of models that during inflation @xmath93 , with @xmath11 being the superfield which contains the inflaton , the appearance of tachyonic directions in the inflaton potential is avoided and certain dangerous couplings between the moduli sector and the inflaton are eliminated .the right - handed sneutrino provides an interesting particle physics candidate for being the inflaton .we have furthermore proposed a new possibility to lower the spectral index @xmath9 in sugra hybrid inflation models based on a new term in the khler potential involving the waterfall field and the field which gives rise to the vacuum energy by its f - term .while the inflaton potential remains flat at tree - level , the new term changes the quantum loop corrections and therefore the form of the scalar potential can become of hilltop - type .this leads to a reduced @xmath9 compared to the generic prediction of @xmath94 in `` conventional '' susy models of hybrid inflation .we have shown that with the new term in the khler potential the spectral index can be lowered without any tuning of parameters and the wmap best - fit value @xmath1 can be realized .the authors would like to thank steve f. king and mar bastero - gil for discussions and comments on the draft .the authors acknowledge partial support by the dfg cluster of excellence `` origin and structure of the universe '' .in this appendix we would like to outline how the simple superpotential and khler potential of eqs . ( 1 ) and ( 2 ) could be realized in an explicit model and , in particular , how additional unwanted terms ( for instance a superpotential term @xmath95 ) can be avoided .we therefore assume three additional superfields @xmath96 , @xmath97 and @xmath98 which are heavy and , at the time of inflation , have already acquired vevs in their bosonic components .we next impose the discrete symmetries @xmath99 , @xmath100 , @xmath101 as well as an r - symmetry under which the superpotential carries charge @xmath102 .the field content of the toy model ( which should be understood as a `` proof of existence '' only ) together with the assigned charges are displayed in table [ symmetries ] . with these symmetries ,the following terms in the superpotential ( up to operators of dimension 6 ) and khler potential are allowed : @xmath103 since we have assumed the fields @xmath96 , @xmath97 and @xmath98 as heavy and already at their minima , we can neglect them for the dynamics of inflation and just insert their vevs . realizing that for a given complex ( spatially constant ) vev @xmath104 we can always perform global phase redefinitions of the fields to make @xmath104 real, we see that we recover the simple superpotential and khler potential of eqs .( 1 ) and ( 2 ) , apart from the additional operator @xmath105 . since @xmath106 during inflation , however , this term has no effect on the inflationary dynamics . 00 for textbook reviews on inflationsee : a. r. liddle and d. h. lyth , `` cosmological inflation and large - scale structure , '' _ cambridge , uk : univ .( 2000 ) 400 p _; a. d. linde , `` particle physics and inflationary cosmology , '' [ arxiv : hep - th/0503203 ] ; v. mukhanov , `` physical foundations of cosmology , '' _ cambridge , uk : univ .( 2005 ) 421 p_. r. jeannerot , `` inflation in supersymmetric unified theories , '' phys .d * 56 * , 6205 ( 1997 ) [ arxiv : hep - ph/9706391 ] .s. antusch , s. f. king , m. malinsky , l. velasco - sevilla and i. zavala , `` flavon inflation , '' phys .b * 666 * ( 2008 ) 176 [ arxiv:0805.0325 [ hep - ph ] ] .a. d. linde , `` axions in inflationary cosmology , '' phys .b * 259 * , 38 ( 1991 ) .e. j. copeland , a. r. liddle , d. h. lyth , e. d. stewart and d. wands , `` false vacuum inflation with einstein gravity , '' phys .d * 49 * , 6410 ( 1994 ) [ arxiv : astro - ph/9401011 ] .a. d. linde and a. riotto , `` hybrid inflation in supergravity , '' phys .d * 56 * , 1841 ( 1997 ) [ arxiv : hep - ph/9703209 ] .m. dine , l. randall and s. d. thomas , `` supersymmetry breaking in the early universe , '' phys . rev .* 75 * , 398 ( 1995 ) [ arxiv : hep - ph/9503303 ] .v. n. senoguz and q. shafi , `` new inflation , preinflation , and leptogenesis , '' phys .b * 596 * ( 2004 ) 8 [ arxiv : hep - ph/0403294 ] .s. antusch , m. bastero - gil , s. f. king and q. shafi , `` sneutrino hybrid inflation in supergravity , '' phys .d * 71 * , 083519 ( 2005 ) [ arxiv : hep - ph/0411298 ] .m. bastero - gil , s. f. king and q. shafi , `` supersymmetric hybrid inflation with non - minimal kaehler potential , '' phys .b * 651 * , 345 ( 2007 ) [ arxiv : hep - ph/0604198 ] .m. k. gaillard , h. murayama and k. a. olive , `` preserving flat directions during inflation , '' phys .b * 355 * , 71 ( 1995 ) [ arxiv : hep - ph/9504307 ] .s. antusch , m. bastero - gil , k. dutta , s. f. king and p. m. kostka , `` solving the @xmath0-problem in hybrid inflation with heisenberg symmetry and stabilized modulus , '' jcap * 0901 * , 040 ( 2009 ) [ arxiv:0808.2425 [ hep - ph ] ] .m. kawasaki , m. yamaguchi and t. yanagida , `` natural chaotic inflation in supergravity , '' phys .lett . * 85 * , 3572 ( 2000 ) [ arxiv : hep - ph/0004243 ] . m. yamaguchi and j. yokoyama , `` new inflation in supergravity with a chaotic initial condition , '' phys .d * 63 * , 043506 ( 2001 ) [ arxiv : hep - ph/0007021 ] .p. brax and j. martin , `` shift symmetry and inflation in supergravity , '' phys .d * 72 * , 023518 ( 2005 ) [ arxiv : hep - th/0504168 ]. g. r. dvali , q. shafi and r. k. schaefer , `` large scale structure and supersymmetric inflation without fine tuning , '' phys . rev. lett . * 73 * , 1886 ( 1994 ) [ arxiv : hep - ph/9406319 ] .p. brax , c. van de bruck , a. c. davis and s. c. davis , `` coupling hybrid inflation to moduli , '' jcap * 0609 * , 012 ( 2006 ) [ arxiv : hep - th/0606140 ] .s. c. davis and m. postma , `` successfully combining sugra hybrid inflation and moduli stabilisation , '' jcap * 0804 * , 022 ( 2008 ) [ arxiv:0801.2116 [ hep - th ] ] .e. komatsu _ et al ._ [ wmap collaboration ] , `` five - year wilkinson microwave anisotropy probe ( wmap ) observations : cosmological interpretation , '' arxiv:0803.0547 [ astro - ph ] .g. hinshaw _ et al . _[ wmap collaboration ] , `` five - year wilkinson microwave anisotropy probe ( wmap ) observations : data processing , sky maps , & basic results , '' arxiv:0803.0732 [ astro - ph ] . c. pallis ,`` kaehler potentials for hilltop f - term hybrid inflation , '' jcap * 0904 * , 024 ( 2009 ) [ arxiv:0902.0334 [ hep - ph ] ] .r. a. battye , b. garbrecht and a. moss , `` constraints on supersymmetric models of hybrid inflation , '' jcap * 0609 * , 007 ( 2006 ) [ arxiv : astro - ph/0607339 ] .s. clesse and j. rocher ,`` avoiding the blue spectrum and the fine - tuning of initial conditions in hybrid inflation , '' phys .d * 79 * , 103507 ( 2009 ) [ arxiv:0809.4355 [ hep - ph ] ] .k. freese , j. a. frieman and a. v. olinto , `` natural inflation with pseudo - nambu - goldstone bosons , '' phys .. lett . * 65 * , 3233 ( 1990 ) .j. p. hsu and r. kallosh , `` volume stabilization and the origin of the inflaton shift symmetry in string theory , '' jhep * 0404 * , 042 ( 2004 ) [ arxiv : hep - th/0402047 ] .r. kallosh and s. prokushkin , `` supercosmology , '' arxiv : hep - th/0403060 .s. r. coleman and e. j. weinberg , `` radiative corrections as the origin of spontaneous symmetry breaking , '' phys .d * 7 * , 1888 ( 1973 ) .m. u. rehman , q. shafi and j. r. wickman , `` hybrid inflation revisited in light of wmap5 , '' arxiv:0901.4345 [ hep - ph ] .

<S> we propose a solution to the @xmath0-problem in supergravity ( sugra ) hybrid inflation using a nambu - goldstone - like shift symmetry within a new class of models . </S><S> the flatness of the tree - level inflaton potential is ensured by shift symmetry invariance of the khler potential , while a small symmetry breaking term in the superpotential gives rise to a slope of the potential at loop - level . in the proposed class of inflation models , </S><S> potentially dangerous couplings between the inflaton and the moduli sector are avoided . </S><S> we also discuss under which conditions the predicted spectral index can be in agreement with the best - fit - value of the latest wmap observation @xmath1 , corresponding to a hilltop - type inflaton potential at loop - level . </S>

1404.6993

social dynamics have been studied through statistical physics techniques in the last twenty years and are now part of the new branch of physics called sociophysics @xcite .this recent research area uses tools and concepts of the physics of disordered matter and more recently from the network science to describe some aspects of social and political behavior @xcite . from the theoretical point of view , opinion models are interesting to physicists because they present correlations , order - disorder transitions , scaling and universality , among other typical features of physical systems @xcite. one of the most studied models of opinion formation is the majority - rule model @xcite .it considers a population of @xmath0 agents , each carrying one of two possible opinions @xmath1 or @xmath2 .a group of @xmath3 agents ( where @xmath3 is an odd number ) is picked at random and all agents in the group adopt the state of the local majority .these rules are repeated until the system reaches a final state of consensus , where all agents are @xmath1- or @xmath2-supporters . in this workwe introduce a mechanism in the majority - rule model that limits the capacity of persuasion of the majorities , and study the impact of this limitation on the opinion formation .this work is organized as follows . in section 2we present the model , define its microscopic rules and discuss the numerical results .our conclusions are presented in section 3 .we consider a population of @xmath4 agents with opinions @xmath1 or @xmath2 concerning a given subject .the initial fraction of @xmath1-supporters is @xmath5 at @xmath6 .each agent @xmath7 has its own `` convincing power '' ( cp ) , a real number @xmath8 initially sorted uniformly inside a given interval @xmath9 $ ] .a group of @xmath10 agents is randomly chosen .if there is a majority of two , say @xmath11 and @xmath7 , in favor of one opinion and the average convincing power @xmath12 of this pair is larger than that of the third agent , say @xmath13 , this last flips his opinion and the winning pair @xmath14 increases its convincing powers , i.e. , we update @xmath15 and @xmath16 .otherwise , nothing occurs .one time step in the model consists of the application of those rules @xmath0 times .notice from the microscopic rules that the agents capacities of persuasion are limited , compared with the traditional pure majority - rule : only if @xmath17 the majority succeeds in convincing the minority within the tossed @xmath3-group .in addition , the updating of the cps models the tendency for a given agent who succeeds in persuading many others to increase his persuasion ability @xcite . this new ingredient introduces a historical long - term memory into the dynamical evolution which turns it non - trivial , with its non - ergodicity exhibited since beginning .agents by any kind of global average .] thus , the inclusion of the above rules is a simple form to capture this real - world characteristic .notice also that the distribution of cps ( uniform inside @xmath9 $ ] at beginning or any other future form ) can be freely translated by a constant displacement ( as a whole ) , and thus can include even negative numbers .this is due to the fact that only the relative difference among the convincing powers ( @xmath18 and @xmath19 ) matters . in this workwe will present results for @xmath20 , but the behavior of the model does not depend qualitatively on the value of @xmath21 .one can start studying the time evolution of one of the two fractions of opinions @xmath1 or @xmath2 , represented by ising variables @xmath22 and @xmath23 , respectively . since the system is symmetric around @xmath24 , in fig .[ fig1 ] we exhibit the time evolution of the fraction @xmath25 of @xmath1-opinions for typical values of @xmath26 and population size @xmath27 . in opposition to what occurs in the pure majority - rule model @xcite , the initial minority can grow , eventually dominating all the population . for the case@xmath28 , fig .[ fig1 ] ( a ) , @xmath29 of the population starts with opinion @xmath2 ( @xmath23 ) .after few time steps the number of individuals with the minority opinion @xmath1 ( @xmath30 ) decreases , but the majority opinion @xmath2 does not reach @xmath31 of the population . in this case , the system remains in this state near the consensus with @xmath32 ( and @xmath33 ) for a long time , but after that the number of individuals with @xmath22 starts to increase and suddenly grows until the consensus state with @xmath34 .this behavior occurs for all simulated samples ( 5 for the plot in fig .[ fig1 ] , thousands in total ) . in the case @xmath35, one can see in fig .[ fig1 ] ( b ) that there is an oscillation with respect to the majority opinion : initially the opinion @xmath1 ( @xmath22 ) is the minority one , but after some time steps a reversion occurs and the opinion @xmath2 ( @xmath36 ) becomes the ( second ) minority . in this case , the state @xmath23 is shared by a very small fraction of the population , less than the previous @xmath1-minority ( near @xmath37 , see fig .[ fig1 ] ( b ) ) .this new global @xmath2-minority opinion grows and dominates all the population near @xmath38 .the first @xmath1-minority is not sufficiently small to win the final consensus , in opposition to what occurs with the second , smaller @xmath2-minority .this kind of oscillation and reversion of the majority opinion occurs three times in the case @xmath39 , as exhibited in fig .[ fig1 ] ( c ) . of @xmath1-opinions for population size @xmath27 .results are for different initial fractions of @xmath1-supporters , namely @xmath28 ( a ) , @xmath35 ( b ) and @xmath39 ( c ) .arrows indicate which ( @xmath1 or @xmath2 ) final consensus is reached .each curve represents a single realization of the dynamics.,title="fig:",scaledwidth=40.0% ] of @xmath1-opinions for population size @xmath27 .results are for different initial fractions of @xmath1-supporters , namely @xmath28 ( a ) , @xmath35 ( b ) and @xmath39 ( c ) .arrows indicate which ( @xmath1 or @xmath2 ) final consensus is reached .each curve represents a single realization of the dynamics.,title="fig:",scaledwidth=40.0% ] + of @xmath1-opinions for population size @xmath27 .results are for different initial fractions of @xmath1-supporters , namely @xmath28 ( a ) , @xmath35 ( b ) and @xmath39 ( c ) .arrows indicate which ( @xmath1 or @xmath2 ) final consensus is reached .each curve represents a single realization of the dynamics.,title="fig:",scaledwidth=40.0% ] - and @xmath2- distributions of the convincing powers for @xmath28 and population size @xmath27 , for distinct time steps @xmath40 .the average value @xmath41 of the whole distribution is translated to the origin in all plots ( its original value is written ) . @xmath1- and@xmath2- averages are indicated by spikes .observe , already for @xmath42 , the sloped behavior of the lower distribution indicating the presence of a strong selection pressure among the minority , absent among the majority , upper distribution .that is why opinion @xmath1 eventually dominates the population , in spite of being the initial minority.,title="fig:",scaledwidth=30.0% ] - and @xmath2- distributions of the convincing powers for @xmath28 and population size @xmath27 , for distinct time steps @xmath40 .the average value @xmath41 of the whole distribution is translated to the origin in all plots ( its original value is written ) . @xmath1- and@xmath2- averages are indicated by spikes .observe , already for @xmath42 , the sloped behavior of the lower distribution indicating the presence of a strong selection pressure among the minority , absent among the majority , upper distribution .that is why opinion @xmath1 eventually dominates the population , in spite of being the initial minority.,title="fig:",scaledwidth=28.5% ] + - and @xmath2- distributions of the convincing powers for @xmath28 and population size @xmath27 , for distinct time steps @xmath40 .the average value @xmath41 of the whole distribution is translated to the origin in all plots ( its original value is written ) . @xmath1- and@xmath2- averages are indicated by spikes .observe , already for @xmath42 , the sloped behavior of the lower distribution indicating the presence of a strong selection pressure among the minority , absent among the majority , upper distribution .that is why opinion @xmath1 eventually dominates the population , in spite of being the initial minority.,title="fig:",scaledwidth=30.0% ] - and @xmath2- distributions of the convincing powers for @xmath28 and population size @xmath27 , for distinct time steps @xmath40 .the average value @xmath41 of the whole distribution is translated to the origin in all plots ( its original value is written ) . @xmath1- and@xmath2- averages are indicated by spikes .observe , already for @xmath42 , the sloped behavior of the lower distribution indicating the presence of a strong selection pressure among the minority , absent among the majority , upper distribution .that is why opinion @xmath1 eventually dominates the population , in spite of being the initial minority.,title="fig:",scaledwidth=28.5% ] + - and @xmath2- distributions of the convincing powers for @xmath28 and population size @xmath27 , for distinct time steps @xmath40 .the average value @xmath41 of the whole distribution is translated to the origin in all plots ( its original value is written ) . @xmath1- and@xmath2- averages are indicated by spikes .observe , already for @xmath42 , the sloped behavior of the lower distribution indicating the presence of a strong selection pressure among the minority , absent among the majority , upper distribution .that is why opinion @xmath1 eventually dominates the population , in spite of being the initial minority.,title="fig:",scaledwidth=30.0% ] - and @xmath2- distributions of the convincing powers for @xmath28 and population size @xmath27 , for distinct time steps @xmath40 .the average value @xmath41 of the whole distribution is translated to the origin in all plots ( its original value is written ) . @xmath1- and@xmath2- averages are indicated by spikes .observe , already for @xmath42 , the sloped behavior of the lower distribution indicating the presence of a strong selection pressure among the minority , absent among the majority , upper distribution .that is why opinion @xmath1 eventually dominates the population , in spite of being the initial minority.,title="fig:",scaledwidth=28.5% ] + - and @xmath2- distributions of the convincing powers for @xmath28 and population size @xmath27 , for distinct time steps @xmath40 .the average value @xmath41 of the whole distribution is translated to the origin in all plots ( its original value is written ) . @xmath1- and@xmath2- averages are indicated by spikes .observe , already for @xmath42 , the sloped behavior of the lower distribution indicating the presence of a strong selection pressure among the minority , absent among the majority , upper distribution .that is why opinion @xmath1 eventually dominates the population , in spite of being the initial minority.,title="fig:",scaledwidth=30.0% ] - and @xmath2- distributions of the convincing powers for @xmath28 and population size @xmath27 , for distinct time steps @xmath40 .the average value @xmath41 of the whole distribution is translated to the origin in all plots ( its original value is written ) . @xmath1- and@xmath2- averages are indicated by spikes .observe , already for @xmath42 , the sloped behavior of the lower distribution indicating the presence of a strong selection pressure among the minority , absent among the majority , upper distribution .that is why opinion @xmath1 eventually dominates the population , in spite of being the initial minority.,title="fig:",scaledwidth=28.5% ] the final victory of the ( sufficiently small ) minority is a consequence of the limited persuasion considered in the majority - rule model , promoting selection among the minority group but not among the majority , a darwinian bottleneck as explained below .let s take the case @xmath28 for which we have verified that the lower values of the fraction of @xmath1-opinions are close to @xmath43 , i.e. , @xmath44 of the population are favorable to opinion @xmath2 ( @xmath23 ) .however , as agents forming the groups are randomly chosen , we have in this case many groups of @xmath10 individuals with @xmath23 , which does not change their convincing powers .this fact explains why the system is kept for a long time in a quasi - stationary state . during this period ,individuals with minority opinion , when rarely chosen to form a majority in a @xmath3-group , can persuade individuals with @xmath23 , upgrading their convincing powers . a tiny minority opinion ( @xmath1 , in this case )group is nucleated with high convincing powers .they can not be persuaded to change to the majority opinion ( @xmath2 ) , and eventually dominate the whole population .one can better understand the behavior presented in fig .[ fig1 ] if we follow the time evolution of the agents cps @xmath1- and @xmath2-distributions . in fig .[ fig2 ] we exhibit a typical realization for @xmath28 and @xmath27 agents , at certain time steps @xmath40 .the initial majority is favorable to opinion @xmath2 ( 90,000 agents , see fig .[ fig2 ] ) . at @xmath6we have a uniform distribution of cps with average @xmath45 .after 1 time step one can see that the cp distribution for the current majority ( @xmath2-supporters ) remains similar as at @xmath6 , no selection .but the cp distribution for the current minority ( @xmath1-supporters ) changes : a selection towards high cp is at work , the quoted darwinian bottleneck . in evolutionary biology ,populations suffering a drastic decrement in size tend to have a set of few individuals with high reproductive activities .this leads to a corresponding decrement in genetic diversity , the founder effect . here, a small group of @xmath1-supporters ( a minority among the minority ) consistently increase their convincing powers , surpassing that of all @xmath2-supporters .this small group can no longer be persuaded , and keeps the minority alive .finally , an inversion of the majority opinion occurs between @xmath46 and @xmath47 , until the consensus with @xmath1-opinion is reached . - and @xmath2- distributions of convincing powers for @xmath35 and population size @xmath27 . at beginning , the darwinian bottleneck selection among the @xmath1-minorityis not so strong , as can be observed by comparing the slope of the minority distribution at @xmath42 with the previous figure .the stronger the bottleneck ( smaller minority ) , the stronger its selection pressure . here , even under a weaker selection than in fig .[ fig2 ] , the minority also succeeds in becoming majority between @xmath48 and @xmath49 .but now the new @xmath2-minority is much smaller , the darwinian bottleneck selection is now strong enough to guarantee a further majority inversion between @xmath50 and @xmath51 .eventually @xmath2-consensus is settled.,title="fig:",scaledwidth=30.0% ] - and @xmath2- distributions of convincing powers for @xmath35 and population size @xmath27 . at beginning ,the darwinian bottleneck selection among the @xmath1-minority is not so strong , as can be observed by comparing the slope of the minority distribution at @xmath42 with the previous figure .the stronger the bottleneck ( smaller minority ) , the stronger its selection pressure . here , even under a weaker selection than in fig .[ fig2 ] , the minority also succeeds in becoming majority between @xmath48 and @xmath49 .but now the new @xmath2-minority is much smaller , the darwinian bottleneck selection is now strong enough to guarantee a further majority inversion between @xmath50 and @xmath51 . eventually @xmath2-consensus is settled.,title="fig:",scaledwidth=28.5% ] + - and @xmath2- distributions of convincing powers for @xmath35 and population size @xmath27 . at beginning ,the darwinian bottleneck selection among the @xmath1-minority is not so strong , as can be observed by comparing the slope of the minority distribution at @xmath42 with the previous figure .the stronger the bottleneck ( smaller minority ) , the stronger its selection pressure . here , even under a weaker selection than in fig .[ fig2 ] , the minority also succeeds in becoming majority between @xmath48 and @xmath49 .but now the new @xmath2-minority is much smaller , the darwinian bottleneck selection is now strong enough to guarantee a further majority inversion between @xmath50 and @xmath51 . eventually @xmath2-consensus is settled.,title="fig:",scaledwidth=30.0% ] - and @xmath2- distributions of convincing powers for @xmath35 and population size @xmath27 . at beginning ,the darwinian bottleneck selection among the @xmath1-minority is not so strong , as can be observed by comparing the slope of the minority distribution at @xmath42 with the previous figure .the stronger the bottleneck ( smaller minority ) , the stronger its selection pressure . here , even under a weaker selection than in fig .[ fig2 ] , the minority also succeeds in becoming majority between @xmath48 and @xmath49 .but now the new @xmath2-minority is much smaller , the darwinian bottleneck selection is now strong enough to guarantee a further majority inversion between @xmath50 and @xmath51 . eventually @xmath2-consensus is settled.,title="fig:",scaledwidth=28.5% ] + - and @xmath2- distributions of convincing powers for @xmath35 and population size @xmath27 . at beginning ,the darwinian bottleneck selection among the @xmath1-minority is not so strong , as can be observed by comparing the slope of the minority distribution at @xmath42 with the previous figure .the stronger the bottleneck ( smaller minority ) , the stronger its selection pressure . here , even under a weaker selection than in fig .[ fig2 ] , the minority also succeeds in becoming majority between @xmath48 and @xmath49 .but now the new @xmath2-minority is much smaller , the darwinian bottleneck selection is now strong enough to guarantee a further majority inversion between @xmath50 and @xmath51 . eventually @xmath2-consensus is settled.,title="fig:",scaledwidth=30.0% ] - and @xmath2- distributions of convincing powers for @xmath35 and population size @xmath27 . at beginning, the darwinian bottleneck selection among the @xmath1-minority is not so strong , as can be observed by comparing the slope of the minority distribution at @xmath42 with the previous figure .the stronger the bottleneck ( smaller minority ) , the stronger its selection pressure . here , even under a weaker selection than in fig .[ fig2 ] , the minority also succeeds in becoming majority between @xmath48 and @xmath49 .but now the new @xmath2-minority is much smaller , the darwinian bottleneck selection is now strong enough to guarantee a further majority inversion between @xmath50 and @xmath51 . eventually @xmath2-consensus is settled.,title="fig:",scaledwidth=28.5% ] + - and @xmath2- distributions of convincing powers for @xmath35 and population size @xmath27 . at beginning , the darwinian bottleneck selection among the @xmath1-minority is not so strong , as can be observed by comparing the slope of the minority distribution at @xmath42 with the previous figure .the stronger the bottleneck ( smaller minority ) , the stronger its selection pressure . here , even under a weaker selection than in fig .[ fig2 ] , the minority also succeeds in becoming majority between @xmath48 and @xmath49 .but now the new @xmath2-minority is much smaller , the darwinian bottleneck selection is now strong enough to guarantee a further majority inversion between @xmath50 and @xmath51 . eventually @xmath2-consensus is settled.,title="fig:",scaledwidth=30.0% ] - and @xmath2- distributions of convincing powers for @xmath35 and population size @xmath27 . at beginning ,the darwinian bottleneck selection among the @xmath1-minority is not so strong , as can be observed by comparing the slope of the minority distribution at @xmath42 with the previous figure .the stronger the bottleneck ( smaller minority ) , the stronger its selection pressure . here , even under a weaker selection than in fig .[ fig2 ] , the minority also succeeds in becoming majority between @xmath48 and @xmath49 .but now the new @xmath2-minority is much smaller , the darwinian bottleneck selection is now strong enough to guarantee a further majority inversion between @xmath50 and @xmath51 .eventually @xmath2-consensus is settled.,title="fig:",scaledwidth=28.5% ] in fig .[ fig3 ] we exhibit the time evolution of the cps for @xmath35 . as in the previous case ,an initial minority supports @xmath1-opinion , but now it is not a so small minority as for @xmath28 , fig .as time goes by , this minority becomes majority and later minority again , eventually becoming extinct . for @xmath39 (not shown ) , the initial majority is slightly larger than the initial minority . at beginning , the darwinian selection bottleneck is weaker yet than in both previous cases .the consequence is the occurrence of three majority inversions .after the first inversion , the new minority becomes smaller than the initial one , but not small enough . only after the second inversionthe minority becomes small enough to guarantee a further , definitive inversion . by gradually increasing @xmath52 towards @xmath53, one can observe 4 inversions , then 5 , 6 , etc , forming an infinite series .similar behavior can be observed in biological models of two - species competition , in which the species with smaller population size is , under certain conditions , less susceptible to extinction than the more populous species @xcite .-consensus ( @xmath54 ) when the initial fraction of @xmath1-supporters is varied , for some population sizes @xmath0 .notice the symmetry with respect to the vertical and horizontal ( black ) lines.,scaledwidth=55.0% ] to summarize the results , we plot in fig .[ fig4 ] the final fraction @xmath55 of simulations which show absorbing states with @xmath1-consensus , when the initial fraction @xmath52 of @xmath1-supporters is varied , for some population sizes @xmath0. one can observe regions ( plateaus ) were the probability of @xmath1-consensus is @xmath56 and some others where this probability is @xmath57 .observe that these plateaus become ever more narrow as the limiting case @xmath24 is approached .also , in the thermodynamic limit ( @xmath58 ) the smooth curves that separate the plateaus will present a straight - step , vertical behavior , i.e. the final absorbing state depends only on @xmath52 , not on the specific initial distribution of convincing powers .furthermore , the above - discussed majority oscillations occur more times as @xmath52 approaches @xmath53 .to conclude , we considered a limited - persuasion mechanism in the majority - rule model @xcite .thus , agents with a local majority opinion in a given group can be persuasive only if their average capacity of persuasion is larger than the minority s conviction .this limitation provokes the appearance of a selection rule acting in favor of the minority , a darwinian bottleneck which opens the possibility of the initially minority opinion dominating the population in the long - time limit , provided it is sufficiently small .these findings suggest that such behaviors , typically observed in problems with biological interest like two - species competition ones @xcite , can occur also in sociological systems as the human dynamics of opinion formation .the authors are grateful to ronald dickman for fruitful suggestions , including the title . financial support from the brazilian funding agenciesfaperj , capes and cnpq is also acknowledged .

<S> street demonstrations occur across the world . in rio de janeiro , june / july 2013 , they reach beyond one million people . a wrathful reader of _ o globo _ , </S><S> leading newspaper in the same city , published a letter @xcite where many social questions are stated and answered yes or no . </S><S> these million people of street demonstrations share opinion consensus about a similar set of social issues . </S><S> but they did not reach this consensus within such a huge numbered meetings . </S><S> earlier , they have met in diverse small groups where some of them could be convinced to change mind by other few fellows . suddenly , a macroscopic consensus emerges . </S><S> many other big manifestations are widespread all over the world in recent times , and are supposed to remain in the future . </S><S> the interesting questions are : 1 ) how a binary - option opinion distributed among some population evolves in time , through local changes occurred within small - group meetings ? and </S><S> 2 ) is there some natural selection rule acting upon ? here , we address these questions through an agent - based model </S><S> . keywords : opinion dynamics , collective phenomenon , agent - based model , computer simulation </S>

1209.5219

in recent years much of the effort in theoretical physics has gone into the study of the observed present accelerated expansion of the universe first reported in @xcite through observational data from type ia supernovae .subsequent work on type ia supernovae @xcite , the cosmic microwave background @xcite , and baryon acoustic oscillations @xcite , all support the initial observations that the expansion of the universe is accelerating .this late - time acceleration of the universe is driven by a fluid / field generically called dark energy .very little is known about dark energy . within the context of string theorythere is a model for the _ early - time _ accelerated expansion of the universe associated with inflation .this string - theory - motivated model for inflation is called dirac - born - infield ( dbi ) inflation @xcite , and it is driven by the open string sector through dynamical dp - branes .dbi inflation is a special case of k - inflation models @xcite .it was originally thought that dbi inflation models would yield large non - gaussian perturbations which could be used to verify or falsify these models and by extension to test string theory @xcite .however , subsequent work has shown that this may not be the case , and that the simplest dbi models are effectively indistinguishable from standard field - theoretic slow - roll models of inflation @xcite . in the present workwe examine variants of these dbi models as a mechanism , not for the early - time acceleration of inflation , but for the observed late - time acceleration .our dbi scalar field will play the role of dark energy .the action for our scalar dbi field is taken to have the form found in ref .@xcite . @xmath0 ~,\ ] ] where we have assumed that the scalar dbi field , @xmath1 , is spatially homogeneous so that its spatial derivatives can be ignored .this is in accord with the fact that dark energy seems to be very homogeneously distributed .note that @xmath1 has a nonstandard kinetic energy term which yields a standard kinetic energy term if one expands the square root to first order in @xmath2 .for a pure ads@xmath3 geometry with radius @xmath4 , the warped tension @xmath5 is the d3-brane tension with @xmath6 , where @xmath7 is the string coupling , @xmath8 is the inverse string tension , and @xmath9 is the t hooft coupling in the ads / cft correspondence .@xmath10 is the potential arising from interactions with ramond - ramond fluxes or with other sectors @xcite .here we take the potential to be quadratic , @xmath11 , and the associated d - brane is in the anti - de sitter throat @xcite . in a spatially flat friedmann - robertson - walker ( frw )metric with scale factor @xmath12 , it can be shown that the energy density @xmath13 and the pressure @xmath14 of the dbi scalar field are given by @xmath15 where @xmath16 has the form of a lorentz boost factor , @xmath17 in this paper we analyze this system of a dbi dark energy field interacting with dark matter in terms of late - time scaling solutions .such models are different from the original work in ref .@xcite which studied the dynamics of a dbi field plus a perfect fluid but with no interaction between the dbi field and the perfect fluid. in the course of our analysis of this model of dbi dark energy interacting with dark matter , we find that for certain parameters there are late - time attractor solutions or fixed points in the phase space of the parameters .the organization of the rest of the paper is as follows . in sec .ii , we write down two possible phenomenological interactions between the dbi dark energy scalar field with the dark matter fluid . in sec .iii , we analyze the autonomous equations in terms of the relevant variables of dbi dark energy interacting with dark matter .we find the fixed points in the phase - space flow of these variables .we also discuss the stability ( i.e. stable fixed point , unstable fixed point or saddle fixed point ) of fixed points and determine if the fixed points correspond to late - time accelerated expansion or not .of particular importance will be stable fixed points which lead to late - time accelerated expansion . finally , we summarize our results in sec .cosmological evolution is thought to be largely dominated by dark energy and dark matter .dark energy gives a gravitationally repulsive effect while dark matter is gravitationally attractive .usually there is no interaction between these two components such as in the model in ref.@xcite where graded lie algebras were used to give a unified theory with both dark energy and dark matter , but without any interaction between these two components .there has been work such as the two measure cosmological model of ref .@xcite where there is some interaction between the dark energy and dark matter components of the model .however , since the gravitational effects of dark energy and dark matter are opposite ( i.e. , gravitational repulsion versus gravitational attraction ) and since dark energy appears to be very homogeneously distributed , while dark matter clumps around ordinary matter , one expects that any interaction between these two dark components of the universe would be weak . in this paperwe will consider models where there is an interaction between dark energy and dark matter .the dark energy component will come from the dbi scalar field in eq . with energy density @xmath18 and pressure @xmath14 , andthe dark matter component will come from a fluid with an equation of state @xmath19 . considering a spatially flat friedman - robertson - walker background with scale factor @xmath12 , and allowing for creation / annihilation between the dbi scalar field and the dark matter fluid at a rate @xmath20 ,we can write down the equations for @xmath13 and @xmath21 as @xmath22 \dot{\rho}_{m}+3h\rho_{m } & = & + q\label{eq05 } ~.\end{aligned}\ ] ] here @xmath23 is the hubble rate with derivatives with respect to cosmological time , @xmath24 , being indicated by a dot .the dbi scalar field and the dark matter create / decay into one another via the common creation / annihilation rate @xmath25 .@xmath20 represents the interaction between these two fields .although at this point this interaction is generic , one can say that if @xmath26 dark energy converts to dark matter , and if @xmath27 dark matter is converted to dark energy @xcite . since there is no fundamental theory which specifies a coupling between dark energy and dark matter , our coupling models will necessarily be phenomenological , although one might view some couplings as more physical or more natural than others . in this paperwe consider two types of coupling : @xmath28 \mathrm{model ~ ii}\quad & q&= \alpha h \rho_{m}\label{eq07}\end{aligned}\ ] ] where @xmath29 and @xmath30 are dimensionless constants whose sign determines the direction of energy transfer . for positive values of the parameters@xmath31 ( @xmath26 ) there is a transfer of energy from dbi dark energy to dark matter ; for negative values of the parameters @xmath32 ( @xmath27 ) there is a transfer of energy from dark matter to dbi dark energy . the interaction given in modeli may be motivated within the context of scalar - tensor theories @xcite where similar interaction terms can be found .generalizations of this model allow for @xmath33 and more general forms of @xmath10 ( e.g. , refs .interactions of the form given by model ii have been considered in ref .@xcite which used @xmath34 and in ref .@xcite , which used @xmath35 .the equation for the rate of change of the hubble parameter is @xmath36\ ] ] the hubble parameter is subject to the constraint @xmath37 in this work we use the units such that @xmath38 , where @xmath39 is newton s gravitational constant .we define the fractional density of the dbi dark energy and dark matter via @xmath40 and @xmath41 , with the condition that @xmath42 which comes from . the modified klein - gordon equation , which follows from eqs . , , and and gives the evolution of the dbi scalar field , takes the form @xmath43 where @xmath44 and @xmath45 .eqs . , , and give a closed system of equations that determines the dynamics of the dbi dark energy scalar field , @xmath1 , interacting with dark matter .in order to find the fixed points of this system and to study the late - time attractor behavior of these two models we introduce the following set of dimensionless variables similar to those used in ref .@xcite : @xmath46 with @xmath47 and @xmath48 defined in this way we can recover the variables for the scalar field model originally proposed in ref .@xcite by taking the limit @xmath49 .the variable @xmath47 roughly corresponds to the the kinetic energy of the dbi field , while @xmath48 roughly corresponds to the potential energy of the dbi field .in addition to these dynamical variables @xmath47 and @xmath48 we introduce a third variable @xmath50 which is connected with the brane tension @xmath51 .taking the inverse of @xmath16 makes the final equations more compact .thus we have exchanged the three variables @xmath1 , @xmath52 , and @xmath51 for @xmath47 , @xmath48 , and @xmath53 . in terms of these variables ,the friedmann constraint from eq . can be expressed as @xmath54 the equation of state of the dbi dark energy is given by @xmath55 finally , we introduce two sets of variables related to the potential , @xmath10 , and the brane tension , @xmath51 .the first set , @xmath56 and @xmath57 , are defined as @xmath58 the second set of variables , @xmath59 and @xmath60 , are given by @xmath61 the relationship between the two sets of variables and is given by @xmath62 combining the above definitions , the evolution equations for @xmath47 , @xmath48 and @xmath63 can be written as the following autonomous system : @xmath64 & & { } \label{eq17}\\[0.5\baselineskip ] \frac{\mathrm{d}y}{\mathrm{d}n}= & & -\frac { \tilde{\lambda}_{1 } y^{2}\sqrt{3(1-\tilde{\gamma}^{2})}}{2 } + \frac{3y}{2}\left\{1+\tilde{\gamma}x^{2}-y^{2}\right\}\label{eq18}\\[0.5\baselineskip ] \frac{\mathrm{d}\tilde{\gamma}}{\mathrm{d}n}=&&\frac{\tilde{\gamma}(1-\tilde{\gamma}^{2})}{\sqrt{(1+\tilde{\gamma } ) } } \!\!\left[3\sqrt{(1+\tilde{\gamma})}\!\!-\!\!\sqrt{3(1-\tilde{\gamma})}\frac{y}{x^{2}}\!\left [ \tilde{\lambda}_{1 } y^{2 } + \tilde{\lambda}_{2 } x^{2}\right]\right.\nonumber\\[0.5\baselineskip ] & & \qquad\qquad\quad\quad\left.+\frac{q}{x^{2}\sqrt{3(1+\tilde{\gamma})}h^{3}}\!\right]\label{eq19}\end{aligned}\ ] ] where @xmath65 and @xmath66 .there are also two equations for @xmath67 and @xmath68 @xmath69 \frac{\mathrm{d } \lambda_{2}}{\mathrm{d}n}=&&-x \lambda_{2}^{2}\sqrt{3\tilde{\gamma}(1+\tilde{\gamma } ) } \left\{\frac{tt_{,\,\phi\phi}}{t^{2}_{,\,\phi}}-1\right\ } \ ; .\label{eq21}\end{aligned}\ ] ] since @xmath70 and @xmath71 can be expressed in terms of @xmath72 , @xmath73 , @xmath74 from eqs . - ,thus equations eqs . andare not independent equations and we only need to solve three equations for @xmath47 , @xmath48 , and @xmath63 .the time evolution equation for @xmath75 given in eq .can be rewritten by differentiating the hubble parameter with respect to @xmath76 , which yields @xmath77 the total effective equation of state for the dbi scalar field plus dark matter can be written as @xmath78 in the next section we will investigate the two different models , given by eqs . and , near critical points ( @xmath79 , @xmath80 ) .near a critical point the scale factor of the frw space - time takes the form @xmath81 in order to have accelerated expansion( i.e. @xmath82 ) the above equation requires that @xmath83 . combining eq . with eq . , and recalling that @xmath66 , the hubble parameter evolution equation becomes @xmath84 the energy balance equations and for model i and model ii are independent of @xmath75 when expressed in terms of the variables @xmath85 and @xmath86 , where @xmath87 .thus the hubble parameter evolution equation is not needed for these particular interacting models , and the phase space of both models is a two - dimensional phase space involving @xmath47 and @xmath48 .in this section we find the critical or fixed points of the autonomous system eqs . - and perform a stability analysis of these fixed points .we are looking for the late - time attractor structure of this system of a dbi scalar field interacting with dark matter via an energy exchange given by @xmath20 .the fixed points for eqs . - are found by setting @xmath88 and solving the resulting three algebraic equations for the critical @xmath89 and @xmath90 .additionally , we will focus on the case when @xmath91 [ from eq . this implies @xmath92 or @xmath93 [ from eq .this implies @xmath94 .thus @xmath63 is constant and the autonomous system reduces to only two dynamical variables : @xmath47 and @xmath48 . after finding the fixed points we study their stability with respect to small perturbations , @xmath95 and @xmath96 , about the critical points @xmath79 , @xmath80 .explicitly , these take the form @xmath97 substituting eq . into eqs . and , andkeeping terms up to first order in @xmath95 and @xmath96 , leads to a system of first - order differential equations of the form @xmath98 where @xmath99 is a @xmath100 matrix that depends on @xmath79 and @xmath80 . to study the stability around the fixed pointsone calculates the eigenvalues of @xmath99 .we denote these eigenvalues as @xmath101 and @xmath102 , and for every critical point @xmath103 there is an associated eigenvalue pair @xmath104 .the stability of the critical point @xmath103 is then determined by its associated eigenvalue pair @xmath104 in the following way : ( i ) if @xmath105 and @xmath106 , then the critical point is stable ; ( ii ) if @xmath107 and @xmath108 , then the critical point is unstable ; ( iii ) if @xmath109 and @xmath108 or @xmath107 and @xmath106 , then one has a saddle point ; ( iv ) if the determinant of the matrix @xmath99 is negative and the real parts of @xmath101 and @xmath102 are negative , then one has a limit cycle . for our two models , model i and model ii , all the fixed points fall into cases ( i ) , ( ii ) , or ( iii ). none of the fixed point we found are limit cycles . recalling that we are taking @xmath63 to be a nondynamical constant with a value of @xmath111 or@xmath112 our autonomous system - for model i becomes @xmath113}{2}\nonumber\\[0.5\baselineskip ] & & \qquad+\frac{3x}{2}\left\{\tilde{\gamma}\left(x^{2}-1\right)-y^{2}\right\}\label{eq27}\\[0.5\baselineskip ] & & \frac{\mathrm{d}y}{\mathrm{d}n}= -\frac { \tilde{\lambda}_{1 } y^{2}\sqrt{3(1-\tilde{\gamma}^{2})}}{2 } + \frac{3y}{2}\left\{1+\tilde{\gamma}x^{2}-y^{2}\right\}\label{eq28}\end{aligned}\ ] ] the dynamics of this autonomous system is determined by the parameters @xmath29 and @xmath59 .the fixed points for eqs . and are obtained by setting @xmath114 and @xmath115 and solving the resulting algebraic equations for @xmath116 and @xmath117 .there are six fixed points and these are presented in the first two columns of table [ tab01 ] . [ cols="^,^,^,^,^,^,^,^",options="header " , ] in this subsection we move on to the analysis of the late - time attractor structure of model ii as given by the autonomous system in eqs . - .the results for the eight fixed points of model ii are summarized in tables [ tab03 ] and [ tab04 ] .the behavior of the dynamics of the dbi scalar field interacting with dark matter via @xmath118 depends on the values of the parameters @xmath30 and @xmath59 .we found that there are nontrivial `` scaling solutions '' where @xmath47 , @xmath48 , and @xmath53 are finite constants depending on the model parameters @xmath30 and @xmath59 .tables [ tab03 ] and [ tab04 ] , show that in the interacting model ii , for ultrarelativistic " fixed points ( d1 ) and ( d2 ) , @xmath101 is always positive , while @xmath102 can be either positive or negative depending on the value of @xmath30 . in particular , ( d1 ) and ( d2 ) are saddles point for @xmath119 and are unstable for @xmath120 . for the fixed points ( d3 ) and( d4 ) @xmath101 is always negative while @xmath121 can be positive or negative depending on the value of @xmath30 and @xmath59 .in particular , ( d3 ) and ( d4 ) are stable , fixed points when @xmath122 and are unstable points for @xmath123 .these fixed points , ( d3 ) and ( d4 ) , are the only critical points from model ii which can give rise to an accelerated expansion for the universe. accelerated expansion will occur for ( d3 ) and ( d4 ) if @xmath124 .thus the fixed points ( d3 ) and ( d4 ) are good candidates for the late - time attractor solution under the requirement that the parameters @xmath30 and @xmath59 are such that these points are stable and that they meet the conditions for accelerated expansion .the @xmath125 phase - space flow of the points ( d1 ) - ( d4 ) are shown in fig .[ fig04 ] .model ii : the phase plane for the dbi dark energy with interaction to dark matter corresponding to @xmath126 for @xmath127 , @xmath128 .the late - time attractors , ( d3 ) and ( d4 ) , are stable fixed points which lead to accelerated expansion.,scaledwidth=50.0% ] for the nonrelativistic " ( i.e. , @xmath129 ) fixed points ( e1 ) and ( e2 ) , @xmath101 is always positive , whereas @xmath102 can be either positive or negative depending on the value of @xmath30 . if @xmath130 then @xmath131 and ( e1 ) and ( e2 ) are saddle points . if @xmath132 , then @xmath133 and ( e1 ) and ( e2 ) are unstable fixed points . for the points ( e3 ) and ( e4 ) , @xmath101 and @xmath102 can be either positive or negative depending on the value of @xmath30 .these points are stable for @xmath134 , are saddle points for @xmath135 , and are unstable nodes for @xmath130 .thus none of the points ( e1 ) - ( e4 ) are stable , and from table [ tab04 ] none of these fixed points lead to accelerated expansion .thus all of these points are not phenomenologically viable . only points ( d3 ) and ( d4 ) satisfy the criteria of accelerated expansion ( for values of @xmath124 ) and stability ( for certain values of @xmath30 _ and _ @xmath59 ) . here , in contrast to the case for model i , the coupling between dark energy and dark matter plays a significant role in the _ stability _ of the fixed points .although fixed points ( d3 ) and ( d4 ) would still exist without the coupling between dark energy and dark matter characterized by the parameter @xmath30 , the stability of these points does depend crucially on @xmath30 and therefore on the coupling between dark energy and dark matter . from table[ tab04 ] one can see that if one sets @xmath136 ( i.e. , the maximum value for which one still gets accelerated expansion ) , then @xmath30 is restricted as @xmath137 solving this gives the restriction that @xmath138 only for this range of @xmath30 ( and also one needs @xmath124 ) does one get both a stable critical point and accelerated expansion .in this paper we studied the dynamics of dark energy ( in the form of a dirac - born - infeld scalar field ) interacting with dark matter ( in the form of a fluid ) in standard , flat frw cosmology . for the scalar potential and the brane tension of the dbi scalar field we took power - law functions .we analyzed two different models for the interaction between dark energy and dark matter .starting with the general interaction equation and we considered @xmath139 ( i.e. , model i ) and @xmath118 ( i.e. , model ii ) .for each model of the dark energy - dark matter interaction we were interested in fixed points which had accelerated expansion ( i.e. , @xmath140 ) _ and _ for which the fixed points were stable . for model i , from tables[ tab01 ] and [ tab02 ] one can see that points ( a1 ) , ( a2 ) , and ( b2 ) satisfy these two conditions. however , none of these points had any dependence on the coupling parameter @xmath29 between the dark energy and dark matter .thus for model i there was no effect of adding the coupling between dark energy and dark matter or a model which only had a dbi scalar field .thus while not ruled out , model i is not really of interest since the coupling to dark matter does not lead to any different result from simply having a dbi scalar field . for modelii we find two stable fixed points which have late time acceleration - points ( d3 ) and ( d4 ) - as can be seen in tables [ tab03 ] and [ tab04 ] .in addition from table [ tab04 ] one can see that whether or not one has accelerated expansion depends on @xmath141 , which from eqs . and depends on the scalar field potential and the tension , but does not depend on the coupling between the dbi field and the dark matter fluid .thus the addition of a coupling between dark energy , in the form of a dbi scalar field and the dark matter fluid does not appear to contribute to the existence of accelerated expansion for points ( d3 ) and ( d4 ) in model ii .a similar conclusion was reached for model i where the stable fixed points with accelerated expansion did not depend on the parameter @xmath29 which was a measure of the coupling between dark energy and dark matter for this model . however , for model ii one can see , by looking at the stability " column of table [ tab04 ] that whether or not a given fixed point is stable _ does _ depend on @xmath30 and therefore on the coupling between the dbi field and the dark matter fluid . in particular in order for the critical points ( d3 ) and ( d4 ) to lead to accelerated expansion _ and _ to be stable one needs to restrict the coupling parameter @xmath30 between the dbi dark energy and dark matter fluid via eq . .although we have only examined two specific types of couplings between dark energy ( in the guise of a dbi scalar field ) and dark matter , we will tentatively advance some general conclusions about such models : \(i ) the existence of fixed points with accelerated expansion does not depend on the coupling between the dark energy and the dark matter .one can make a general argument to support this conclusion .dark matter will fall off like @xmath142 ^{-3}$ ] , while if the dark energy is to lead to late - time acceleration it will act like a cosmological constant which falls off like @xmath142 ^0 \rightarrow$ ] constant . at late timesthe @xmath142 ^0 $ ] behavior will always dominate the @xmath142 ^{-3}$ ] behavior .\(ii ) while the coupling to dark matter does not seem to play a role in the existence of fixed points with accelerated expansion it can play a role in their stability . for example , in model ii the stability of the fixed points ( d3 ) and ( d4 ) , which had accelerated expansion , depended on the dark energy - dark matter coupling parameter @xmath30 .however for model i none of the fixed points stability depended on @xmath29 , the coupling parameter for this model .thus it seems that whether coupling between dark energy and dark matter is important to the stability of the fixed points depends crucially on the type of coupling one chooses . the work of ny has been supported by the thailand center of excellence in physics ,the work of ss was supported by a russian foundation for basic research grant no .11 - 02 - 01162 and a fulbright scholars grant , and the work of ds is supported by a daad grant .we would like to thank dr .burin gumjudpai for discussions and guidance .we also thank prof .eduardo guendelman , prof .edouard b. manoukian , dr . piyabut burikham , dr . khampee karwan and dr.suppiya siranan for discussions and comments .finally , we would like to acknowledge with thanks for department of physics , california state university , fresno , usa , for its kind hospitality .m. quartin , m. o. calvao , s. e. joras , r. r. r. reis , and i. waga , jcap 0805 , 007 ( 2008 ) , 0802.0546 . ; l. p. chimento , a. s. jakubi , d. pavon , andw .zimdahl , phys .d67 , 083513 ( 2003 ) , astro - ph/0303145 . ; a. p. billyard and a. a. coley , phys .d61 , 083503 ( 2000 ) , astro - ph/9908224 . ; j. valiviita , e. majerotto , and r. maartens , jcap 0807 , 020 ( 2008 ) , 0804.0232 . ;c. g. boehmer , g. caldera - cabral , r. lazkoz , and r. maartens , phys .d78 , 023505 ( 2008 ) , 0801.1565 . ;guo , n. ohta , and s. tsujikawa , phys .d76 , 023508 ( 2007 ) , astro - ph/0702015 . ; c. wetterich , astron .301 , 321 ( 1995 ) , hep11 th/9408025 . ;l. amendola , phys .d60 , 043501 ( 1999 ) , astroph/ 9904120 . ; w. zimdahl and d. pavon , phys .b521 , 133 ( 2001 ) , astro - ph/0105479 .; g. r. farrar and p. j. e. peebles , astrophys .j. 604 , 1 ( 2004 ) , astro - ph/0307316 . ; r .- g .cai and a. wang , jcap 0503 , 002 ( 2005 ) , hepth/ 0411025 . ; g. olivares , f. atrio - barandela , and d. pavon , phys .d71 , 063523 ( 2005 ) , astro - ph/0503242 . ; h. m. sadjadi and m. alimohammadi , phys .d74 , 103007 ( 2006 ) , gr - qc/0610080 . ; c. quercellini , m. bruni , a. balbi , and d. pietrobon ( 2008 ) , 0803.1976 . ; j .- h .he , b. wang , and e. abdalla ( 2008 ) , 0807.3471 .chen , y. gong , and e. n. saridakis ( 2008 ) , 0812.1117 . ; n. dalal , k. abazajian , e. e. jenkins , and a. v. manohar , phys .87 , 141302 ( 2001 ) , astroph/ 0105317 . ; e. majerotto , d. sapone , and l. amendola ( 2004 ) , astroph/ 0410543 . ; h. garcia - compean , g. garcia - jimenez , o. obregon , and c. ramirez , jcap 0807 , 016 ( 2008 ) , 0710.4283 . ; m. r. setare and e. c. vagenas ( 2007 ) , 0704.2070 . ; n. j. nunes and d. f. mota , mon . not .368 , 751 ( 2006 ) , astro - ph/0409481 .r. bean and j. magueijo , phys .b 517 , 177 ( 2001 ) [ arxiv : astro - ph/0007199 ] ; r. bean , phys . rev .d 64 , 123516 ( 2001 ) [ arxiv : astro - ph/0104464 ] ; l. amendola , c. quercellini , d. tocchini - valentini and a. pasqui , astrophys. j. 583 , l53 ( 2003 ) [ arxiv : astro - ph/0205097 ] ; d. comelli , m. pietroni and a. riotto , phys .b 571 , 115 ( 2003 ) [ arxiv : hep - ph/0302080 ] ; g. r. farrar and p. j. e. peebles , astrophys . j. 604 , 1 ( 2004 ) [ arxiv : astro - ph/0307316 ] ; m. b. hoffman , arxiv : astro - ph/0307350 ; u. franca and r. rosenfeld , phys . rev .d 69 , 063517 ( 2004 ) [ arxiv : astro - ph/0308149 ] ; r. fardon , a. e. nelson and n. weiner , jcap 0410 , 005 ( 2004 ) [ arxiv : astro - ph/0309800 ] ; f. vernizzi , phys . rev .d 69 , 083526 ( 2004 ) [ arxiv : astro - ph/0311167 ] ; x. j. bi , p. h. gu , x. l. wang and x. m. zhang , phys . rev .d 69 , 113007 ( 2004 ) [ arxiv : hep - ph/0311022 ] ; s. lee , k. a. olive and m. pospelov , phys .d 70 , 083503 ( 2004 ) [ arxiv : astro - ph/0406039 ] ; a. w. brookfield , c. van de bruck , d. f. mota and d. tocchini - valentini , phys .96 , 061301 ( 2006 ) [ arxiv : astro - ph/0503349 ] ; t. koivisto , phys . rev .d 72 , 043516 ( 2005 ) [ arxiv : astro - ph/0504571 ] ; h. wei and r. g. cai , phys . rev .d 72 , 123507 ( 2005 ) [ arxiv : astro - ph/0509328 ] ; r. mainini and s. bonometto , jcap 0706 , 020 ( 2007 ) [ arxiv : astro - ph/0703303 ] ; a. fuzfa and j. m. alimi , phys . rev .d 75 ( 2007 ) 123007 [ arxiv : astro - ph/0702478 ] ; r. bean , e. e. flanagan and m. trodden , arxiv:0709.1124 [ astro - ph ] ; ibid .arxiv:0709.1128 [ astro - ph ] ; t. gonzalez and i. quiros , arxiv:0707.2089 [ gr - qc ] .w. zimdahl and d. pavon , phys .b 521 , 133 ( 2001 ) [ arxiv : astro - ph/0105479 ] ; l. p. chimento , a. s. jakubi , d. pavon and w. zimdahl , phys . rev .d 67 , 083513 ( 2003 ) [ arxiv : astro - ph/0303145 ] ; j. d. barrow and t. clifton , phys .d 73 ( 2006 ) 103520 [ arxiv : gr - qc/0604063 ] ; h. m. sadjadi and m. alimohammadi , phys . rev .d 74 , 103007 ( 2006 ) [ arxiv : gr - qc/0610080 ] .

<S> we study the dynamics of dirac - born - infeld ( dbi ) dark energy interacting with dark matter . </S><S> the dbi dark energy model considered here has a scalar field with a nonstandard kinetic energy term , and has potential and brane tension that are power - law functions . </S><S> the new feature considered here is an interaction between the dbi dark energy and dark matter through a phenomenological interaction between the dbi scalar field and the dark matter fluid . </S><S> we analyze two different types of interactions between the dbi scalar field and the dark matter fluid . </S><S> in particular we study the phase - space diagrams of and look for critical points of the phase space that are both stable and lead to accelerated , late - time expansion . in general </S><S> we find that the interaction between the two dark components does not appear to give rise to late - time accelerated expansion . </S><S> however , the interaction can make the critical points in the phase space of the system stable . whether such stabilization occurs or not depends on the form of the interaction between the two dark components . </S>

1204.2504

e. n. lorenz in @xcite demonstrated numerically the existence of certain three - dimensional flows that have a complicated behaviour .the _ lorenz flow _ has a saddle fixed point with a one - dimensional unstable manifold and an infinite set of periodic orbits whose closure constitutes a global attractor of the flow .as it is often done in dynamics , one can attempt to understand the behaviour of a three - dimensional flow by looking at the first return map to an appropriately chosen two - dimensional section . in the case of the lorenz flow, it is convenient to choose the section as a plane transversal to the local stable manifold , and , therefore , intersecting it along a curve @xmath1 .the first return map is discontinuous at @xmath1 .the _ geometric lorenz flow _ has been introduced in @xcite : a lorenz flow with an extra condition that the return map preserves a one - dimensional foliation in the section , and contracts distances between points in the leafs of this foliation at a geometric rate .since the return maps is contracting in the leafs , its dynamics is asymptotically one - dimensional , and can be understood in terms of a map acting on the space of leafs ( an interval ) .this interval map has a discontinuity at the point of the interval corresponding to @xmath1 , and is commonly called the _ lorenz map_. we will start by defining what is known as the standard lorenz family .our work is a continuation of the study started in @xcite , and we will , therefore , make a conscientious effort to use the notation of @xcite so that it would be easier for the reader to compare the approach of this paper with that of @xcite .[ slf ] let @xmath2 $ ] , @xmath3 $ ] , @xmath4 and @xmath5 .the standard lorenz family @xmath6 is the family of maps @xmath7 \setminus \ { c \ } \mapsto [ 0,1]$ ] defined as @xmath8q(x)=\ { [ cols= " < , < " , ] a lorenz map @xmath9 is called _ renormalizable _ if there exist @xmath10 and @xmath11 , @xmath12 , such that the first return map @xmath13 , @xmath14 , of @xmath15 $ ] is affinely conjugate to a nontrivial lorenz map .choose @xmath16 such that it is maximal .the rescaled first return map of such @xmath17 is called the renormalization of @xmath9 and denoted @xmath18 $ ]. we will denote @xmath19,\ ] ] while the first return map will be denoted @xmath20 $ ] and referred to as the prerenormalization .if @xmath9 is renormalizable , then there exist minimal positive integers @xmath21 and @xmath22 such that @xmath23(x)=\left\{f^{n+1}(x ) , \quad x \in l , \atop f^{m+1}(x ) , \quad x \in r , \right.\ ] ] then , explicitly , @xmath24=a^{-1 } \circ \cp[f ] \circ a,\ ] ] where @xmath25 is the affine orientation preserving rescaling of @xmath26 $ ] onto @xmath16 .we will also use the notation @xmath27 for the renormalization of @xmath9 .the intervals @xmath28 , @xmath29 , are pairwise disjoint , and disjoint from @xmath16 . so are the intervals , @xmath30 , @xmath31 .since these intervals do not contain @xmath32 , we can associate a finite sequence of @xmath33 and @xmath34 to each of these two sequences of intervals : @xmath35 which will be called the type of renormalization . the subset of lorenz maps as in the definition @xmath36 which are renormalizable of type @xmath37 is referred to as the domain of renormalization @xmath38 ( cf .@xcite ) .let @xmath39 if @xmath40 $ ] is @xmath41-renormalizable for all @xmath42 , then @xmath9 is called infinitely renormalizable of combinatorial type @xmath43 .the set of @xmath44-renormalizable maps will be denoted @xmath45 , the set of maps @xmath9 such that @xmath40 $ ] is @xmath41-renormalizable will be called @xmath46 , @xmath47 , with @xmath21 finite or infinite . if @xmath43 is such that @xmath48 , @xmath49 , for some @xmath50 , we say that @xmath43 is of _ bounded type_. we would like to draw the attention of the reader to the position of the indices in our notation : @xmath51 is a pair of two words , while @xmath52 is an integer @xmath33 or @xmath34 in a single word ( cf .@xmath53 ) . the combinatorics @xmath54 will be called _monotone_. the set of all monotone combinatorial types will be denoted @xmath55 , while @xmath56 will denote all lorenz maps which are @xmath44-renormalizable with @xmath57 .given an integer @xmath58 , the subset of @xmath55 of all @xmath44 s such that the length of words in @xmath44 satisfies @xmath59 and @xmath60 , will be denoted @xmath61 . given two integers @xmath62 , @xmath63 will denote the subset of @xmath55 of all @xmath44 s such that the length of words in @xmath44 satisfies @xmath64 and @xmath65 . given a subset @xmath66, @xmath67 will denote all lorenz maps which are @xmath44-renormalizable with @xmath68 .we will also use the notation @xmath69 .the main results of our paper are the following proposition and theorems .[ mainprop1 ] ( _ a priori bounds _ ) . for every @xmath70 there exist an integer @xmath71 , and , for every integer @xmath72 , a subset @xmath73 , relatively compact in @xmath74 , such that @xmath75 \subset \ck$ ] . at this pointwe were able to prove _ a priori _ bounds only for @xmath70 .the somewhat technical reasons for that will become clear in the proof of the invariance of bounds on the critical point in proposition @xmath76 .proposition @xmath77 is used to obtain the existence of the periodic points of renormalization : [ mainthm1 ] ( _ renormalization periodic points _ ) . for every @xmath70 and every @xmath78 , where @xmath79 is as in in the main proposition @xmath77, the renormalization operator @xmath80 has a periodic point in @xmath81 of type @xmath43 .the proof of the next main theorem @xmath82 follows verbatim that of a similar result in @xcite , after on establishes _ a priori bounds_. we , however , chose to state this as separate main result since the existence of a cantor attractor for the dynamics merits a special emphasis . for completeness , the proof will be included in the appendix . [ mainthm2 ] ( _ cantor attractors _ ) .let @xmath70 , and suppose that @xmath83 , where @xmath79 is as in in the main proportion @xmath77 , is of bounded type .consider @xmath84 , and let @xmath85 be the closure of the orbits of the critical values .then , * @xmath85 is a cantor set of a hausdorff dimension strictly inside @xmath86 ; * @xmath85 is uniquely ergodic ; * the complement of the basin of attraction of @xmath85 in @xmath26 $ ] has zero lebesgue measure .the study of renormalizable lorenz maps was initiated by tresser et al .( see e.g. @xcite ) . a more recent work of martens and de melo @xcite produced a series of important results , specifically about the domains of renormalization and the structure of the parameter plane for two - dimensional lorenz families . the work @xcite presented a computer assisted proof of existence of a renormalization fixed point for the renormalization operator of type @xmath87 .the renormalization operator of this particular type has been later shown to have a fixed point in the class of maps analytic on a neighbourhood of the unit interval using only complex analytic techniques in @xcite . in a more generalsetting , issues of existence of renormalization periodic points and hyperbolicity have been addressed in @xcite , where it is proved that the limit set of renormalization , restricted to monotone combinatorics with the return time of one branch being large and much larger than the return time for the other branch , is a cantor set , and that each point in the limit set has a two - dimensional unstable manifold .specifically , @xcite proves equivalents of our main proposition @xmath77 and main theorem @xmath88 for monotone combinatorial types with the following return times : @xmath89 \le |\omega^-|-1 \le [ 2 \rho-1 ] , \quad n_- \le |\omega^+|-1 \le n_+,\ ] ] where @xmath90 is sufficiently large , and @xmath91 depends on the choice of @xmath90 . in comparison , we prove the _ a priori _ bounds for a different class of combinatorial types .we are able to avoid the disparity of return times evident in @xmath92 , as well as boundedness of return times from above .nevertheless , we could not avoid a condition of largeness of return times , nor can we cover the case @xmath93. we would like to emphasise , however , that the lower bounds on the return times for which our results are valid can be expressed in terms of explicit but very cumbersome functions of @xmath94 .a careful computation of these bounds will result in definite ( and , likely , not too large ) values of @xmath79 .however , we have not performed these estimates in the present paper .we will start by quoting the koebe principle which is of a fundamental importance in real dynamics ( see , ex .@xcite ) . we will say that an interval @xmath95 is a @xmath96-scaled neighbourhood of @xmath97 , if both components of @xmath98 have length at least @xmath99 .[ kp ] let @xmath100 be intervals , and @xmath101 be a @xmath102-diffeomorphism with @xmath103 .if @xmath104 contains a @xmath96-scaled neighbourhood of @xmath105 , then @xmath106 let @xmath107 be the set of @xmath108-continuously differentiable maps from @xmath25 to @xmath109 .we denote @xmath110 the subset of orientation preserving homeomorphisms whose inverse lie in @xmath107 .we will use the notation @xmath111 and @xmath112 whenever @xmath113 $ ] .the _ nonlinearity _ operator @xmath114 is defined as @xmath115 while @xmath116 is the nonlinearity of @xmath117 at point @xmath118 .given @xmath119 , the quantity @xmath120=\max_{x , y \in a } \ln \left ( { d \phi(y ) \over d \phi(x ) } \right)\ ] ] is called the _ distortion _ of @xmath117 .notice , that @xmath121 the following lemma results from a straightforward computation .the nonlinearity operator @xmath114 is a bijection . in the case@xmath113 $ ] , the inverse is defined as @xmath122 one can turn @xmath123 into a banach space using the nonlinearity operator . specifically , for @xmath117 , @xmath124 in @xmath123 and @xmath125 , the linear structure and the norm are defined via @xmath126 finally , we give a list of useful bounds on derivatives and distortion in @xmath123 in terms on the nonlinearity ( see @xcite or @xcite for the proofs ) . if @xmath127 then , for all @xmath128 , @xmath129 @xmath130 @xmath131 we will introduce two subsets of lorenz maps , defined via conditions on their distortion and critical points . given a real constants@xmath132 , we set @xmath133 \le \pi , { \rm dist}[\phi ] \le \pi \right\}.\ ] ] given real constants @xmath132 , @xmath134 , set @xmath135 \right\}.\ ] ] the reason for the introduction of these sets is the following compactness result .given @xmath132 and @xmath136 , the set @xmath137 is relatively compact in @xmath74 .recall that @xmath138 is isomorphic to @xmath26 ^ 2 \times ( 0,1 ) \times \cd^1 \times \cd^1 $ ] . since @xmath32 is bounded away from @xmath33 and @xmath34 by a constant , it is , therefore , contained in a compact subset of @xmath86 .consider the set @xmath139\le \pi , { \rm dist}[\psi]\le\pi \right\}.\ ] ] any sequence from @xmath140 is equicontinous since @xmath141 , and , clearly , uniformly bounded , therefore by the arzel - ascoli theorem .we will quote a lemma from @xcite ( lemma @xmath142 ) which gives the formulae for the factors of a renormalization of a lorenz map in @xmath56 .let @xmath143 be an interval and @xmath144 be an orientation preserving diffeomorphism .we denote the affine transformation that takes @xmath26 $ ] onto @xmath143 as @xmath145 .define the _ zoom operator _ : @xmath146 [ renorm_parts ] if @xmath147 is renormalizable of monotone combinatorics , then @xmath148=(\tilde{u},\tilde{v},\tilde{c},\tilde{\phi},\tilde{\psi})\ ] ] is given by @xmath149 @xmath150 where @xmath151 , @xmath152 .in this section we will obtain bounds on the critical points , critical values and lengths of the central subintervals @xmath153 and @xmath154 for lorens maps with monotone combinatorics whose diffeomorphic coefficients have bounded distortion . denote @xmath155 , \quad f_1(r)=r_1 , \quad f_0^{\circ k } \circ f_1(r)=r_{k+1}.\end{aligned}\ ] ] we will mention the following simple lemma ( cf .@xcite for a proof ) .[ psiphi_ext_lemma ] suppose @xmath156 , then @xmath157 and @xmath158 are diffeomorphisms on @xmath159 and @xmath160. we will continue with a sequence of lemmas which will prepare us for a construction of _ a priori _ bounds construction of a relatively compact set invariant under renormalization .first of all , we will need simple bounds on the difference of @xmath161 and @xmath162 at two points of the domain .[ anbn ] suppose that @xmath163 \le \pi$ ] , @xmath164 \le \pi$ ] , then @xmath165 for any @xmath166 in @xmath167 , and @xmath168 for any @xmath166 in @xmath169 $ ] .notice , that the average derivative of @xmath117 on @xmath170 is @xmath171 , therefore , the derivative @xmath172 at any point in @xmath170 is bounded as @xmath173 similarly , for @xmath174 .@xmath175 therefore , we get for @xmath166 in @xmath167 @xmath176 the lower bound is obtained as follows : @xmath177 bounds on the difference of @xmath162 can be obtained in a similar way . for the sake of brevity , let us introduce the following notation : @xmath178 since @xmath179 , we have that @xmath180 , and , therefore , for monotone combinatorics @xmath181 .similarly , @xmath182 .the next lemma uses this fact , and provides a lower bound on the length of the intervals @xmath183 $ ] and @xmath184 $ ] , which is also a lower bound on the length of the intervals @xmath185 $ ] and @xmath186 $ ] .[ delt ] let @xmath187 for some @xmath188 and @xmath189 with @xmath190 , @xmath191 .then @xmath192 we will first demonstrate that @xmath193 for all @xmath194 . to prove @xmath195 we use the following expressions for the inverse branches of a lorenz map : @xmath196)| \over |\phi^{-1}([0,c_1 ^ -])| } \right)^{1 \over \rho}= c - c \left({u-\phi^{-1}(x ) \over u } \right)^{1 \over \rho } , \\ \label{f1m1 } f_1^{-1}(x)&=&c+\mu \left(1-{|\psi^{-1}([x,1])| \over |\psi^{-1}([c_1^+,1])| } \right)^{1 \over \rho}= c+\mu \left(1-{1-\psi^{-1}(x ) \over v } \right)^{1\over \rho } , \end{aligned}\ ] ] and start with @xmath197 for @xmath194 , and use induction on this inequality to obtain @xmath198 according to lemma @xmath199 : @xmath200 on the other hand , @xmath201 , and according to @xmath195 , @xmath202 therefore , @xmath203 and @xmath204 which results in the required bound @xmath205 .the bound on @xmath206 is obtained in a similar way .lower bounds on the differences @xmath207 and @xmath208 can be used to bound @xmath209 and @xmath210 from below .[ lem2 ] let @xmath187 for some @xmath188 and @xmath189 with @xmath190 , @xmath191 .then , @xmath211 to get the lower bound on @xmath212 we notice that the derivatives of the inverse branches of @xmath213 ( formulae @xmath214 and @xmath215 with @xmath216 ) are increasing functions , while the derivatives of @xmath117 and @xmath124 are bounded as in @xmath217 and @xmath218 .this can be used to get a straightforward bound @xmath219 for all @xmath220 .therefore , @xmath221 so @xmath222 and @xmath223 the lower bound on @xmath224 is obtained in a similar way .we will now turn our attention to the bounds on @xmath153 and @xmath154 .[ lem3 ] let @xmath187 where @xmath225 and @xmath189 with @xmath190 , @xmath191 .then there exist a constant @xmath226 , such that @xmath227 and @xmath228 _ 1 ) upper bounds_. denote @xmath229 and @xmath230 ( notice , @xmath231 and @xmath232 ) , and , as before , @xmath233 .suppose , point @xmath234 is in the interval @xmath235 , and denote points in the orbit of @xmath234 as @xmath236 : @xmath237 .then , according to @xmath195 , @xmath238 and one gets for all @xmath239 @xmath240 and @xmath241 we can now see that @xmath242 for all @xmath243 . therefore ,@xmath244 notice , that for monotone combinatorics all intervals @xmath245 , @xmath246 , are contained in the interval @xmath247 , while the intervals @xmath248 , @xmath249 , are all contained in @xmath250 .therefore , @xmath251 where we have used that @xmath252 in the last inequality .we can now use the fact that @xmath253 for all @xmath254 as in the hypothesis of the lemma , to simplify the above expression . @xmath255 and the upper bound @xmath256 from the claim follows .the bound @xmath257 on @xmath154 is obtained in a similar way . to derive the upper bound @xmath258, we return to @xmath259 , and notice that for monotone combinatorics all intervals @xmath245 , @xmath260 , are contained in the interval @xmath261 , while the intervals @xmath248 , @xmath262 , are all contained in @xmath263 . therefore , @xmath264 while on the other hand , according to lemma @xmath199 , @xmath265 which together with @xmath266 results in @xmath267 and the second upper bound @xmath258 follows .the bound @xmath268 is obtained in a similar way ._ 2 ) lower bounds_. we will use the fact that @xmath269 , or @xmath270 then , according to the previous lemma , @xmath271 we will now obtain an estimate on @xmath272 .to that end , first notice , that @xmath273 for all @xmath274 , therefore , using the lower bound on @xmath224 from lemma @xmath275 , @xmath276 where @xmath226 is some immaterial constant of order @xmath34 .finally , @xmath277 becomes @xmath278 which results in the required lower bound for @xmath153 .the lower bound for @xmath154 is obtained in a similar way .recall that by lemma @xmath279 , the diffeomorphic coefficients of the renormalized map are @xmath280 the next proposition establishes the conditions for the invariance of the distortion of the coefficients under renormalization .[ inv_distortion ] ( invariance of distortion ) . for every @xmath0 and every @xmath281 ,there exist @xmath282 , such that if @xmath283 where @xmath284 with @xmath285 and @xmath286 , then @xmath287 \le \pi , \quad { \rm and } \quad { \rm dist}[\tilde{\psi } ] \le \pi.\ ] ] we consider the exponential of the distortion of @xmath288 on @xmath26 $ ] . for any @xmath289 $ ] , @xmath290 recall , that @xmath15 $ ] , and that , by lemma @xmath291 , @xmath292 and @xmath293 are well - defined diffeomorphisms at least on @xmath294 and @xmath295 , respectively . by koebe principle @xmath296 @xmath297 where @xmath298 , and @xmath299 similarly , for @xmath300 , @xmath301 where @xmath302 notice , that @xmath303 is a decreasing function of @xmath118 , therefore , @xmath304 .similarly , @xmath305 .therefore , @xmath306,{\rm dist}[\tilde{\psi } ] \ } & \le & \max\left\{\left(1+\tau_2 \over \tau_2 \right)^2,\left(1+\zeta_2 \over \zeta_2 \right)^2\right\ } \\\label{b1 } & = & \max\left\{\left({q - c_1^+ \over p - c_1^+ } \right)^2,\left({c_1 ^ --p \over c_1 ^ --q } \right)^2\right\ } .\end{aligned}\ ] ] below we will demonstrate that @xmath307 is less than @xmath308 for sufficiently large @xmath21 and @xmath22 .recall , that @xmath309 from lemma @xmath310 serves as a lower bound on @xmath311 , while @xmath312 is a lower bound on @xmath313 .then , using that @xmath314 , together with the upper bounds on @xmath153 and @xmath154 from lemma @xmath315 , we get @xmath316 notice , that the function @xmath317 assumes its maximum at @xmath318 , therefore , @xmath319 and similarly for @xmath320 .now , let @xmath321 , then @xmath322 where we have also used that the minimum if @xmath323 for @xmath324 $ ] , is @xmath325 .one can now see that if @xmath326 sufficiently large , then the small factor @xmath327 dominates other terms , and for every @xmath254 as in the hypothesis , there exists a sufficiently large @xmath326 , such that @xmath328 is smaller than @xmath329 whenever the distortion of the coefficients of @xmath9 is smaller than @xmath254 . in a similar way@xmath330 is less than @xmath331 for sufficiently large @xmath21 and @xmath22 .recall the definition of the subset @xmath61 from the introduction : this is the subset of @xmath55 ( monotone types ) of all @xmath44 s such that the length of words in @xmath44 satisfies @xmath285 and @xmath286 . also , recall that according to lemma @xmath279 , the critical point of a renormalized lorenz map is given by @xmath332 ( invariance of the bounds on the critical point ) .[ inv - crpoint ] let @xmath254 satisfy @xmath281 , and let @xmath70 .then there exist @xmath333 and , for every @xmath72 , an @xmath334 such that if @xmath335 then the critical point @xmath336 of the renormalization satisfies @xmath337.\ ] ] our immediate goal is to show , that for @xmath9 as in the hypothesis of the proposition , there exists a positive @xmath338 , such that @xmath336 lies in @xmath339 $ ] whenever @xmath32 does .we will start with the lower bound on @xmath336 . according to @xmath340 , for @xmath336 to be larger or equal to some @xmath341it is sufficient that @xmath342 the maximum of the ratio of the lengths of @xmath154 and @xmath153 can be estimated using bounds from lemma @xmath315 : @xmath343 we will identify the behaviour of the right hand side of the above inequality as @xmath344 . for that , we plug the definition @xmath345 of @xmath312 and the upper bound @xmath257 on @xmath346 in to @xmath347 , and isolate the powers of @xmath348 : @xmath349 where @xmath25 , @xmath109 , @xmath350 , @xmath351 and @xmath352 are some functions that depend on @xmath21 , @xmath22 , @xmath94 and @xmath254 , but do not depend on @xmath348 .since @xmath21 and @xmath22 are bounded from below and above by @xmath79 and @xmath353 , these functions are also bounded , independent of @xmath348 .therefore , for @xmath354 to hold whenever @xmath348 is small , its is sufficient that @xmath355 since @xmath348 strictly less than @xmath34 , for sufficiently large ( but bounded from above and below ) @xmath21 , @xmath22 the small exponential dominates the large factor in front of it , and , there exists a number @xmath58 , and , for each @xmath72 , an @xmath341 such that the inequality holds for @xmath356 and all @xmath357 .we now turn to the case @xmath358 . herewe will use the upper bound @xmath268 on @xmath346 : @xmath359 and we isolate all powers of @xmath32 and @xmath209 : @xmath360 since @xmath361 , for all @xmath0 the power of @xmath209 is bounded from below by the expression @xmath362 which is positive for all @xmath0 , and we can bound @xmath209 raised to the power by from above by @xmath34 . then the condition @xmath363 is implied by @xmath364 and it is sufficient for all @xmath365 that @xmath366 as @xmath22 grows , the power of @xmath367 in the above expression converges to @xmath368 for all @xmath70 , there exists @xmath333 , such that for all @xmath369 this expression is positive , and we get that the invariance condition is satisfied by a sufficiently small @xmath367 . to summarise , denote the maximum of the upper bounds @xmath347 and @xmath370 by @xmath371 ( we suppress the dependence of @xmath353 on @xmath94 , @xmath254 , @xmath372 , @xmath21 and @xmath22 in our notation ) ,then we have shown that there exists @xmath341 such that @xmath373 for all @xmath374 , and that @xmath373 for all @xmath375 .since @xmath376 is clearly a continuous function of @xmath118 , it achieves a minimum @xmath377 on any interval @xmath378 $ ] .we can now choose @xmath379 to be the lower bound on @xmath32 .existence of @xmath380 such that @xmath381 is proved in a similar way by considering the maximum of the ratio of the lengths of @xmath153 and @xmath154 .finally , take @xmath382 one can now see that the reason for a somewhat restrictive condition @xmath70 is the positivity of the exponent in @xmath383 .one might hope that with more work , for example , a better upper bound on @xmath209 , which would add to smallness in @xmath383 , one can relax this constraint .the following results is an immediate corollary of propositions @xmath384 and @xmath76 .( a priori bounds ) .[ inv - crpoint ] let @xmath254 satisfy @xmath281 .then , for every @xmath70 there exists a natural @xmath333 , and , for every natural @xmath72 , a @xmath385 , such that @xmath386 \subset \ck^\pi_\vareps$ ] .we consider a restriction @xmath387 of the renormalization operator to some @xmath388 where @xmath79 is as in proposition @xmath76 . in this sectionwe will demonstrate that @xmath387 has a fixed point .we will generally follow the approach of @xcite ( and we will make a conscientious attempt to keep the notation in line with that work ) .one important difference with the case considered in @xcite , however , is that we are looking at a different class of return times .this will introduce some extra difficulties , especially evident in the proof of lemma @xmath389 , somewhat more involved than its analogue from @xcite .we will start by quoting several previously established results .a branch @xmath143 of @xmath390 is _ full _ if @xmath390 maps @xmath143 onto the domain of @xmath9 .@xmath143 is _ trivial _ if @xmath390 fixes both endpoints of @xmath143 .we will now quote several facts about lorenz maps , established in @xcite .a slice in the parameter plane is any set of the form @xmath391 ^ 2 \times \{c\ } \times \{\phi\ } \times \{\psi\},\ ] ] where @xmath32 , @xmath117 and @xmath124 are fixed .we will use the simplified notation @xmath392 .a slice @xmath393 induces a family of lorenz maps @xmath394 any family induced by a slice is _ full _ , that is it contains maps of all possible combinatorics .specifically ( see @xcite for details ) , [ theorema](_theorem a from @xcite _ ) .let @xmath395 be a family induced by a slice .then this family intersects @xmath396 for every @xmath43 ( finite or infinite ) such that @xmath397 .[ lemma41](_lemma 4.1 from @xcite _ ) .assume that @xmath9 is renormalizable .let @xmath398 be the branch of @xmath399 and @xmath400 be that of @xmath401. then @xmath402 let @xmath254 , @xmath338 and @xmath403 be as in the previous section .consider the set @xmath404 the boundary of @xmath405 consists of three parts : @xmath406 iff at least one of the following holds : * the left and the right branches of @xmath18 $ ] are full or trivial ; * @xmath163=\pi$ ] or @xmath164=\pi$ ] ; * @xmath407 , or @xmath408 .consider the boundary of @xmath409 . if either branch of @xmath410 $ ] is full or trivial, then there exists an perturbation of @xmath9 , however small , such that @xmath9 is no longer renormalizable .hence c1 holds on @xmath411 .if @xmath412 does not satisfy c1 then , according to lemma @xmath413 , all small perturbations of it will be still renormalizable .conditions c2 and c3 are part of the boundary of @xmath403 . by proposition@xmath414 these boundaries intersect @xmath415 , and hence c2 and c3 are also the boundary conditions for @xmath405 .fix @xmath416 , and let @xmath417 ^ 2 \times\{c_0\ } \times \{\rm i d \ }\times \{\rm id\}$ ] .recall , that the linear structure on the space @xmath418 is defined via the nonlinearity operator : @xmath419 introduce the _ deformation retract _ onto @xmath393 as @xmath420 let @xmath421 we will strengthen the conditions on the set @xmath405 and consider a smaller set @xmath422 ) \ge \delta\}.\ ] ] the boundary of @xmath423 is given by conditions @xmath424-@xmath425 together with * @xmath426)=\delta\}$ ] .[ iff_lemma ] the exists a choice of @xmath427 in @xmath428 and @xmath429 , such that @xmath430 has a fixed point in @xmath431 iff @xmath432 has a fixed point in @xmath433 for some @xmath434 $ ] . the direct statement is obvious since @xmath435 .the converse is also obvious when @xmath436 , and we , therefore , consider @xmath437 .assume that @xmath438 with the coefficients @xmath439 is such that @xmath440 for some @xmath441 $ ] , and assume that @xmath430 has no fixed point on @xmath433 .we will demonstrate that this is impossible .choose @xmath427 close to @xmath442 : @xmath443 for some small @xmath444 .by proposition @xmath76 , @xmath445 } ) \in [ \vareps,1-\vareps]$ ] whenever @xmath446 is .together with the condition @xmath447 ) \ge \delta$ ] this implies that @xmath448 } ) \in [ \delta,1-\vareps].\ ] ] since @xmath437 , by formula @xmath428 @xmath449)$ ] is strictly in the interior of @xmath450 $ ] for all @xmath441 $ ] . therefore ,neither c3 nor c4 can hold for @xmath451 $ ] for @xmath441 $ ] .the distortion of the coefficients of @xmath18 $ ] is not greater than @xmath254 by proposition @xmath384 . for @xmath441 $ ] distortion of the diffeomorphic parts @xmath452 of @xmath453 $ ]is strictly smaller than that of @xmath454 ( diffeomorphic coefficients for @xmath18 $ ] ) .this can be seen from the following computation : @xmath455 } \over \int_0 ^ 1 { \exp { [ \int_0^r ( 1-t ) n_{\tilde{\phi}}(s ) + t n_{{\rmid}}(s ) d s ] } d r } } \over { \exp{[\int_0^x ( 1-t ) n_{\tilde{\phi}}(s ) + t n_{{\rm id}}(s ) d s ] } \over \int_0 ^ 1 { \exp { [ \int_0^r ( 1-t ) n_{\tilde{\phi}}(s ) + t n_{{\rm id}}(s ) d s ] } d r } } } = { { \exp{[\int_0^y ( 1-t ) n_{\tilde{\phi}}(s ) d s ] } } \over { \exp{[\int_0^x ( 1-t ) n_{\tilde{\phi}}(s ) d s ] } } } = \left(d \tilde{\phi}(y ) \over d \tilde{\phi}(x ) \right)^{1-t } < e^\pi.\ ] ] similarly for @xmath456 .therefore , we have that c2 does not hold for @xmath451 $ ] for @xmath441 $ ] .the only possibility is that , if @xmath451 \in \partial \cy_\delta$ ] then it belongs to the part of the boundary described by c1 .suppose that either branch of @xmath18 $ ] is full ; for definitiveness , suppose @xmath457)=1 $ ] .since @xmath117 fixes both end points of the unit interval , this implies that @xmath458)=1 $ ] , and since the deformation retract does not change the value of @xmath459 , @xmath460)=1 $ ] .since @xmath461 fixes @xmath34 as well , we get that @xmath462)=1 $ ] , and therefore , the corresponding branch of @xmath453 $ ] is full as well .this shows that @xmath9 can not be fixed by @xmath432 since a renormalizable map can not have a full branch .therefore , one of the branches of @xmath18 $ ] must be trivial . before we proceed with the last case of trivial branches, we will derive an upper bound on @xmath463 and a lower bound on @xmath464 .recall , that @xmath465 where the linear structure is given by @xmath466 .then , on one hand , @xmath467 and @xmath468 on the other hand , @xmath469 and @xmath470 we can now take a linear combination of @xmath471 and @xmath472 as an upper bound on @xmath463 .a particularly convenient choice is @xmath473 notice , that the maximum of the function @xmath474 is achieved at @xmath475 . in a similar way ,@xmath476 suppose , the left branch is trivial : @xmath447 ) \ge c^-_1(\cr[f])$ ] .recall , that for a map renormalizable with monotone combinatorics , @xmath477 , and according to lemmas @xmath315 and @xmath310 the differences @xmath478 and @xmath479 , where @xmath226 and @xmath480 depend on @xmath94 , @xmath254 , @xmath338 , @xmath21 and @xmath22 , but do not depend on the particular form of the map . also , for large @xmath21 and @xmath22 , @xmath226 and @xmath480 become independent of @xmath21 and @xmath22 : @xmath481 which follows from the expressions @xmath205 for @xmath309 and @xmath345 for @xmath312 .next , suppose @xmath444 is small : @xmath482 .then , on one hand , @xmath483)-c(\cr_t[f ] ) \le ( c_1 ^ -(\cr[f]))^{1-t } u(\cr[f])^t \left ( t e^{\pi ( 1-t ) } + ( 1-t ) e^{\pi t}\right ) -c(\cr[f])- t ( c_0-c(\cr[f])).\ ] ] recall , that by proposition @xmath384 , @xmath254 can be chosen small if one considers large @xmath21 and @xmath22 . since @xmath457)-c(\cr[f])<0 $ ] , the expression @xmath484))^{1-t } u(\cr[f])^t \left ( t e^{\pi ( 1-t ) } + ( 1-t ) e^{\pi t}\right ) -c(\cr[f ] ) < k-\nu\ ] ] if @xmath254 is small .then since @xmath485)$ ] is larger than @xmath486 , we have that @xmath484))^{1-t } u(\cr[f])^t \left ( t e^{\pi ( 1-t ) } + ( 1-t ) e^{\pi t}\right ) -c(\cr[f])- t ( c_0-c(\cr[f ] ) ) < k,\ ] ] and hence @xmath451 $ ] is not renormalizable with the monotone combinatorics @xmath44 . now , suppose that the right branch is trivial .then @xmath487)-c_1^+(\cr_t[f ] ) \le c(\cr[f ] ) & + & t(c_0-c(\cr[f ] ) ) \\ \nonumber & - & ( c_1^+(\cr[f]))^{1-t } ( 1-v(\cr[f]))^t \left ( t e^{-\pi ( 1-t ) } + ( 1-t ) e^{-\pi t}\right).\end{aligned}\ ] ] since @xmath447 ) - c_1^+(\cr[f])<0 $ ] , we have that for a sufficiently small @xmath254 , @xmath488)- ( c_1^+(\cr[f]))^{1-t } ( 1-v(\cr[f]))^t \left ( t e^{-\pi ( 1-t ) } + ( 1-t ) e^{-\pi t}\right ) < { j \over 2},\ ] ] while @xmath488 ) \!- \ ! ( c_1^+(\cr[f]))^{1-t } ( 1\!-\ ! v(\cr[f]))^t \left ( t e^{-\pi ( 1-t ) } \!+\!(1\!-\!t ) e^{-\pi t}\right ) \!+\ ! t(c_0-c(\cr[f]))\le { j \over 2 } \!+\!(1\!-\!\vareps\!-\!\nu\!-\!\delta).\ ] ] therefore , the map @xmath453 $ ] is not renormalizable with the monotone combinatorics @xmath44 for @xmath434 $ ] , if we chose @xmath489 so that @xmath490 we now notice , that according to lemmas @xmath315 and @xmath310 @xmath491 if @xmath489 is small , then the above expression demonstrates that @xmath492 , and the inequality @xmath493 is not satisfied . on the other hand ,if @xmath489 is close to @xmath494 , then the exponential in @xmath495 becomes large and dominates others terms , and @xmath493 is easily satisfied .therefore , there exists @xmath496 , not necessarily very close to @xmath497 , such that @xmath493 holds for all @xmath498 .we conclude that @xmath451 \notin \partial \cy_\delta$ ] which is a contradiction with the assumption in the beginning of the proof .according to the theorem b in @xcite the intersection of @xmath393 with @xmath415 contains a connected component @xmath143 of the interior , called a _ full island _ , such that the family @xmath499 $ ] is full .[ ext_lemma ] any extension of @xmath500 to @xmath423 has a fixed point .assume that @xmath501 has no fixed point in @xmath433 ( otherwise the theorem is trivial ) .let @xmath417 ^ 2 \times \{c_0\ } \times \{{\rm id}\ } \times \{{\rm id}\}$ ] , where @xmath427 is as in the previous lemma .this set contains a full island @xmath143 with @xmath502 .pick any @xmath503 such that @xmath504 .define the displacement map @xmath505 by @xmath506 which is well - defined since @xmath154 does not have fixed points on @xmath502 .the degree of @xmath507 is non - zero since @xmath143 is full .therefore , @xmath154 has a fixed point in @xmath143 ( otherwise @xmath507 would extend to all of @xmath143 , and would have a degree zero ) .to finish the proof of the existence of the fixed points we will require the following theorem from @xcite : [ theoremgd ] let @xmath508 where @xmath509 is closed and @xmath510 is a normal topological space . if @xmath511 is homotopic to a map @xmath512 with the property that every extension of @xmath513 to @xmath509 has a fixed point in @xmath509 , and if the homotopy @xmath514 has no fixed point on @xmath515 for every @xmath434 $ ] , then @xmath9 has a fixed point in @xmath509 .@xmath387 has a fixed point .@xmath501 either has a fixed point in @xmath433 , or otherwise by lemma @xmath516 any of extensions of @xmath500 to @xmath423 has a fixed point . in the second casewe can apply theorem @xmath517 and lemma @xmath518 , to immediately obtain the required result .now we can finish the proof of the main theorem @xmath88 .suppose that @xmath79 is as in propositions @xmath384 and @xmath76 .pick a sequence @xmath519 .one can use @xmath520 in place of @xmath387 in the previous proposition to demonstrate that @xmath520 has a fixed point , which , hence , is a periodic point of @xmath430 of combinatorial type @xmath43 .to present a proof of main theorem @xmath82 we will need to introduce the concept of transfer maps .an interval @xmath16 is called a nice interval of @xmath9 , if @xmath16 is open , the critical point of @xmath9 is in @xmath16 , and the orbit of the boundary of @xmath16 is disjoint from @xmath16 .fix @xmath9 and a nice interval @xmath16 .the transfer map to @xmath16 induced by @xmath9 , @xmath521 is defined as @xmath522 , where @xmath523 , is the transfer time to @xmath16 , that is the smallest non - negative integer @xmath21 such that @xmath524 .[ prop37](proposition 3.7 in @xcite ) assume that @xmath9 has no periodic attractors and that @xmath525 .let @xmath526 be the transfer map of @xmath9 to a nice interval @xmath16 .then the complement if the domain of @xmath526 is a compact , @xmath9-invariant and hyperbolic set .let @xmath527 and @xmath528 \setminus u$ ] . since @xmath529 is open , @xmath530 is closed , and , being bounded , is compact . by definition @xmath531 .@xmath530 is the set of points @xmath118 such that @xmath532 for all @xmath533 .since @xmath103 , @xmath9 does not have non - hyperbolic periodic points ( cf @xcite , theorem 1.3 ) , and , by assumption , @xmath9 has no periodic attracting orbits , so @xmath530 is hyperbolic ( cf @xcite , lemma iii.2.1 ) . since a compact , invariant , hyperbolic set for @xmath534 functions has zero lebesgue measure ( cf .@xcite , theorem iii.2.6 ) , we have @xmath26 \setminus { \rm dom } \ t$ ] has zero lebesgue measure . the last result that we will require forthe proof of unique ergodicity is the following theorem due gambaudo and martens ( cf @xcite ) .[ gmtheorem ] if @xmath9 is infinitely renormalizable ( of any combinatorial type ) with a cantor attractor @xmath85 , then @xmath85 supports one or two ergodic invariant probability measures . if the combinatorial type of @xmath9 is bounded , then @xmath85 is uniquely ergodic .we can now present a proof of main theorem @xmath82 , which is identical to theorem @xmath535 from @xcite .let @xmath536 and @xmath537 denote the left and right half intervals for the @xmath21-th first return map , and let @xmath538 and @xmath539 be the return times for the corresponding intervals .set @xmath540 $ ] , and @xmath541 where @xmath542 stands for the closure of a set .components of @xmath543 are called intervals of generation @xmath21 and components of @xmath544 are called gaps of generation @xmath21 .let @xmath545 be intervals of generations @xmath546 and @xmath21 , respectively , and let @xmath547 be a gap of generation @xmath546 .take the @xmath74 closure of the set @xmath548\}$ ] . since @xmath548 \}$ ] is compact in @xmath74 , the infimum and supremum of @xmath549 and @xmath550 over @xmath143 , @xmath480 and @xmath551 of the corresponding generation are bounded away from @xmath33 and @xmath34otherwise , there would be an infinitely renormalizable map in @xmath74 with @xmath552 or @xmath553 ( @xmath554 or @xmath555 ) .this is impossible , since this would imply that , for that map , one ( or both ) of @xmath536 or @xmath537 is of zero length , which contradicts renormalizability .therefore , there exist constants @xmath556 and @xmath557 , such that @xmath558 next , @xmath559 , since the critical values are contained in the closure of @xmath560 for each @xmath21 . from @xmath561 , @xmath562 , therefore , the lengths of intervals of generation @xmath21 tend to zero , and @xmath563 .a standard argument demonstrates that @xmath85 is a cantor set if measure zero ( since @xmath557 ) , of hausdorff dimension in @xmath86 .next , we prove that almost all points are attracted to @xmath85 .let @xmath564 denote the transfer map to the @xmath21-th interval @xmath565 .by proposition @xmath566 the domain of @xmath564 has full lebesgue measure for every @xmath21 , and , therefore , a.e .point enters @xmath567 for every @xmath21 .finally , the unique ergodicity follows theorem @xmath568 .i would like to extend my gratitude to the families of hernan franco and mariana amadei , and pedro beltran and mariana franco , for their hospitality during the preparation of this paper .99 p. collet , p. coullet and c. tresser , _ scenarios under constraint _ , j. physique lett . *46*(4 ) ( 1985 ) , 143147 .gambaudo and m. martens , _ algebraic topology for minimal cantor sets _ , ann . henri poincar .* 7*(3 ) ( 2006 ) , 423446 .a. granas and j. dugundji , _ fixed point theory _ , springer monographs in mathematics ( 2003 ) , springer - verlag , new - york .d. gaidashev and b. winckler , _ existence of a lorenz renormalization fixed point of an arbitrary critical order _ , preprint .j. guckenheimer and r. f. williams , _ structural stability of the lorenz attractors _ , publ .ihes * 50 * ( 1979 ) , 5972 .j. h. hubbard , c. t. sparrow , _ the classification of topologically expansive lorenz maps _ , comm .pure appl . math . *xliii * ( 1990 ) , 431443 .w. de melo and s. van strien , _ one - dimensional dynamics _ , * 25 * ergbnisse der matehematikund ihrer grenzgebiete ( 3 ) ergod .theor . anddyn . sys .* 21*(3 ) ( 1993 ) , springer - verlag , berlin .m. martens and w. de melo , _ universal models for lorenz maps _ , ergod . theor . anddyn . sys .* 21*(3 ) ( 2001 ) , 883860 .m. martens and b. winckler , _ on the hyperbolicity of lorenz renormalization _ ,arxiv:1205.0829 [ math.ds ] , ( 2012 ) . m. misiurewicz , _ absolutely continuous measures for certain maps of an interval _ , inst .hautes tudes sci .publ . math . *53 * , ( 1981 ) , 1751 .e. n. lorenz , _ deterministic non - periodic flow _ , j. atmos .sci * 20 * ( 1963 ) , 130141 .r. f. williams , _ the structure of the lorenz attractors _ , publ . math .ihes * 50 * ( 1979 ) , 7379 .b. winckler , _ a renormalization fixed point for lorenz map _ , nonlinearity * 23*(6 ) ( 2010 ) , 12911303 .

<S> lorenz maps are maps of the unit interval with one critical point of order @xmath0 , and a discontinuity at that point . </S><S> they appear as return maps of sections of the geometric lorenz flow . </S><S> we construct real _ a priori _ bounds for renormalizable lorenz maps with long monotone combinatorics , and use these bounds to show existence of periodic points of renormalization , as well as existence of cantor attractors for dynamics of infinitely renormalizable lorenz maps . </S>

astro-ph0101287

compact groups ( cg ) of galaxies are relatively isolated systems composed of a small number of galaxies ( three or more ) forming configurations with very low median projected intergalactic separation , providing the highest projected galaxy number density systems in the sky .hickson ( 1982 ) has pioneered the use of specific , quantitative selection criteria to create cg catalogs .other catalogs now available include the southern cg catalog ( prandoni , iovino & macgillivray 1994 ) , and those produced from the cfa2 ( barton et al .1996 ) and las campanas ( allam & tucker 1998 ) redshift surveys .groups in hickson s catalog and their member galaxies have been observed at many wavelengths and the extensive available data pose a number of challenging questions .their high galaxy number density and relatively low velocity dispersion would imply , on theoretical grounds , short dynamical timescales and rapid evolution , making cgs the ideal sites for galaxy tidal interactions ( and , eventually , tidal disruption ) and mergers to occur .these would result in the appearance of more frequent dynamical peculiarities in cg galaxies , as well as enhanced star formation rates and nuclear activity compared with galaxies in less dense environments .enhanced galaxy merging activity would produce merger remnants and would lead to the formation of blue luminous ellipticals in a time scale of @xmath0 1 gyr .tidal disruption would cause , eventually , the cg disappearance in a few crossing times ( @xmath1 gyr ; barnes 1989 ) .however , observations of hickson compact groups ( hcgs ) do not support these predictions .a fraction as high as 43@xmath2 of galaxies in hcgs may be interacting ( mendes de oliveira & hickson 1994 ) , but no correlation has been found between the number of interacting galaxies in a given group and the group global parameters ( velocity dispersion , crossing times , x - ray properties ; mendes de oliveira & hickson 1994 ; pildis , bregman & schombert 1995 ) . also , a high fraction of spirals in hcgs have asymmetric and peculiar rotation curves , but these dynamical peculiarities do not correlate with the cg global properties , even if the fraction of galaxies with peculiar rotation curves is higher among those belonging to cgs with lower galaxy velocity dispersion inside the group , @xmath3 km s@xmath4 ( nishiura et al . 2000 ) .the @xmath5 , co and fir observations of the hcg galaxies have shown neither enhancement in their present - day star formation rates nor in their star formation histories , when compared with control samples of field galaxies ( see verdes - montenegro et al .1998 ; iglesias - pramo & vlchez 1999 and references therein ) .coziol et al .( 1997 ) , coziol , iovino & de carvalho ( 2000 ) and allam et al .( 1999 ) have found depressed star formation , relative to control samples , in cgs of the hickson , southern and las campanas catalogs , respectively , and , also , a density - morphology - activity relation in both hcgs and southern cgs .moreover , even if the fraction of galaxies in hcgs that are merger remnants is still controversial , observations suggest that it is not high ( only @xmath6 in zepf 1993 ; see also hickson 1997 and references therein ) . in hcgs ,the first ranked galaxy sample and the global population have the same fraction of e or s0 galaxies ( hickson 1982 ) .observations also suggest that there are not more blue ellipticals in hcgs than in other environments ( zepf & whitmore 1991 ) .hcgs have a relatively large spiral fraction , but it is anticorrelated with the galaxy velocity dispersion ( hickson , kindl & huchra 1988 ) and is lower in x - ray luminous cgs ( these two parameters are , on their turn , correlated , with x - ray detected cgs having @xmath7 km s@xmath4 , see mulchaey 2000 ) .these results and the hcg number density are in contradiction with the theoretical framework of rapid evolution , supported by the first n - body simulations on cg evolution ( barnes 1985 , 1989 ; mamon 1987 ) . to surmount these difficulties ,some authors have pointed out that a large fraction of hcgs are not real dense configurations , but chance superpositions of galaxies ( mamon 1986 ; diaferio , geller & ramella 1994 ; hernquist , katz & weinberg 1995 , but also see hickson & rood 1988 ) .but this scenario does not account for the existence of a diffuse x - ray emission from hot intergroup gas in 75@xmath2 hcgs ( ponman et al .this gas has been found to have almost the same temperature in hcgs where it has been detected ( @xmath80.9 kev , with a detection limit of @xmath80.3 kev ) .the projected gas density profiles are consistent with @xmath9^{(3\beta_{\rm gas}-1)/2}$ ] , with core radius @xmath10 and slope @xmath11 for hcgs ( see mulchaey et al . 1996 and references therein ) . assuming hydrostatic equilibrium ( @xmath12-model , see cavaliere 1973 ) , these observations imply a total group binding mass of @xmath13 m@xmath14 at @xmath15 kpc for groups of the hcg catalog ( see mulchaey et al .1996 , and references therein ) , implying that galaxies in these cgs are embedded in a common massive , concentrated dark matter halo .we present in this letter a new framework to understand x - ray emitting cgs of galaxies as stable systems against tidal disruption of their galaxies , in the quiescent phases of their evolution .the main hypothesis is that the common virialized massive concentrated dark matter halo , in which galaxies are embedded in cgs , stabilizes both the group and the individual galaxies , so they are able to remain in dynamical equilibrium during many group crossing times after halo formation .this possibility has been tested through a series of n - body simulations , whose results confirm this assumption .to study the stability of the galaxies in cgs and the group itself , we have built initial model galaxies and left them to orbit in the external halo potential .we have considered that the massive dark matter halos in which galaxies are embedded are consistent with the gas distribution ( @xmath12-model ) as explained above . a statistical analysis of the distribution of projected galaxy pair separations in accordant redshift quartets of hcgs by montoya et al .( 1996 ) shows that galaxy density profiles are consistent with those of the x - ray emitting gas ( see above ) , with slope parameter @xmath16 instead of @xmath17 . in this way ,the galaxy and halo mass distributions are determined by four parameters among the following : @xmath18 , @xmath19 , @xmath17 , @xmath20 ( given by x - ray data ) , slope of the galaxy distribution profile , @xmath16 , and galaxy velocity dispersion inside the group , @xmath21 .we recall that , solving the jeans equation , the @xmath12-model predicts that @xmath22 , and indeed this is the case for hcgs ( montoya et al .1996 ) . to illustrate the stabilizing role of massive halos , results for two different halo massesare reported : cg1 group model , with @xmath23 m@xmath14 , and cg2 group model , with @xmath24 m@xmath14 at @xmath25 kpc .the core radius of the group is @xmath26 kpc ( mulchaey et al . 1996 ) .the remaining parameters of the halos ( table [ tbl1 ] ) are also consistent with those of hcgs , taking @xmath27 km s@xmath4 mpc@xmath4 ( hickson et al .1992 ) . the initial positions and velocities of the galaxies in the halo , @xmath28 and @xmath29 , @xmath30n@xmath31 ( n@xmath31 is the number of galaxies in the group ; as very often accordant hcgs are quartets , we have taken n@xmath32=4 ) have been assigned in the following way : ( a ) the positions are monte - carlo realizations of the average hcg galaxy number density profile found by montoya et al .( 1996 ) , and ( b ) the velocities are assigned randomly with the condition that @xmath33 .the @xmath28 , @xmath29 and the orbital angular momentum , @xmath34 , are given in table [ tbl2 ] .two sets of galaxy initial positions and velocities are reported for the massive cg1 group model ; for one of them ( cg1 m ) , they have been prepared in such a way that they lead to a binary merger event of two galaxies .galaxies have been assumed to be self - gravitating configurations in quasi - equilibrium with the environment , that is , with the tidal field of the halo . as the simplest choice for the initial galaxy model, we assume that they are spheres with an isotropic velocity dispersion tensor . in this case, the tidal field determines the limiting or tidal radius , @xmath35 , of the configuration ( king 1962 ; gmez - flechoso & domnguez - tenreiro 2001 , hereafter gd01 ) , leading to the so - called tidally - limited ( @xmath36-limited ) king spheres , based on the king - michie velocity distribution function ( king 1966 ; michie 1963 ) .these @xmath37-limited king spheres are characterized by two free parameters among the total galaxy mass , @xmath38 , the 1-dimensional internal velocity dispersion , @xmath39 , the galaxy core radius , @xmath40 , and the central potential , @xmath41 ( see gd01 for details ) . note that the widely used standard king spheres , whose limiting radius is left free ( and not , as due , fixed by the tidal field ) , have instead three free parameters . in gd01we describe a method to build @xmath37-limited king spheres with prefixed values of @xmath38 and @xmath40 .the initial configurations for our cg galaxies are monte - carlo realizations of the @xmath37-limited king spheres , with n@xmath42=10000 particles , @xmath43 m@xmath14 and @xmath44 kpc . in table[ tbl2 ] we give the tidal radius , @xmath45 , and the @xmath39 and @xmath41 parameters corresponding to each galaxy model of the cgs reported in this letter .the values of these parameters are within their observational range .these cg models have been left to evolve during a time interval of 10@xmath46 years ( @xmath47 group crossing times ) .evolution has been followed by integrating the equation of motion for each constituent particle in the combined potential of the other particles and the halo by means of a vectorized version of the treecode ( hernquist 1987 ) .as we are interested in galaxy stability against tidal disruption , the halo potential has been taken to be smooth , so dynamical friction effects have not been considered . however , no important dynamical friction effects are expected ( see @xmath48[summary ] ) . for the massive type cg1 group ,both the individual galaxies and the group itself are stable against tidal stripping , therefore very low tidally induced evolution can be expected in any of them . by galaxy stabilitywe mean that : ( a ) the individual galaxies do not lose appreciably mass , as the fraction of mass stripped by tidal forces is low ( @xmath49 , depending on the galaxy initial position ) , ( b ) the mass density profiles of the individual galaxies do not change , and , ( c ) the velocity distribution function of each galaxy does not evolve significantly , and , therefore , their internal velocity dispersion , @xmath50 , is almost constant ; only a mild velocity anisotropy is developed in each galaxy .the cg does not change either , that is : ( a ) galaxies do not end up at the center of the configuration ; on the contrary , the values of the galaxy pericentric and apocentric distances , as well as the median intergalactic separation , remain stable ( see figure 1a ) , and , ( b ) the galaxy velocity dispersion of the group , @xmath21 , does not change on average ( figure 1b ) .the same qualitative results have been found for the cg1 m group , where the merger of the a and c galaxies soon after the begining of the simulation does not affect either the dynamical properties of the other galaxies in the group , nor the overall group evolution ( see figure 1 ) .all these results are mostly determined by the choice of the initial galaxy model configurations with an average density consistent with the halo density at galaxy pericenter , that is , they are tidal quasi - equilibrium solutions in the tidal field , and they are roughly independent on the galaxy or halo density profiles ( see gd01 for details ) .as @xmath18 decreases , binary interactions become more important relative to the overall halo potential . as a consequence ,galaxies orbiting inside this halo become more likely to be tidally stripped and lose mass , and they tend to fall towards the halo center , where their remnants eventually end up , merging with each other and causing the group tidal disruption .an example of such a process is provided by the evolution of the low halo mass cg2 group model , where half of the mass inside 100 kpc is in the galaxies .the galaxies of the cg2 group lose @xmath51 of their mass during the first 4 gyrs of the simulation .their remnants merge and fall to the halo center .figure 1 illustrates the group disappearance as a consequence of these mass loss .taking more massive galaxies in this last experiment would result in a faster group disappearance , because the determining point here is the importance of the galaxy - galaxy interactions relative to the galaxy - halo interactions .according to current scenarios of hierarchical structure formation , dark matter halos form through violent merger events from previously formed lower mass halos .these halos host baryonic clumps ( either normal galaxies , dwarfs , and possibly hi clouds ) that survive the halo merging and virialization , and , after this process is completed , they orbit inside the newly formed halo ( navarro , frenk & white 1995 , hereafter nfw ; tissera & domnguez - tenreiro 1998 ) . in the present letter , we show that very little mass is stripped from typical galaxies orbiting inside the potential well of a quiescent , massive cg - like halo , provided that the halo mass is similar to that inferred from observations of cgs with diffuse x - ray emission and that the initial galaxy models are self - gravitating configurations in quasi - equilibrium within the tidal field of the halo . when these two conditions are met , the rate of tidal stripping is roughly independent of the galaxy or halo density profile , as shown by gd01 , where a cuspy halo density profile as those obtained by nfw has been adoptedso , profiles do not need to have low concentrations , as demanded by athanassoula , makino & bosma ( 1997 ) .these authors used standard ( as opposed to @xmath36-limited ) king spheres as initial model galaxies , that suffer from important mass loss unless that the cg halos they orbit are almost homogeneous .an important consequence of the tidal quasi - equilibrium in astronomical systems is that dynamical timescales for galaxy infall in cgs , and the ensuing cg disappearance , are not set by mass loss due to tidal stripping . among other processes causing galaxy infall to the center of the configurationthe most efficient is dynamical friction , whose timescales , @xmath52 , however , are difficult to predict .the well known chandrasekhar formula gives , for a @xmath53 m@xmath14 nfw halo and a @xmath54 m@xmath14 galaxy initially placed at @xmath0 40 - 90 kpc from the halo center , a value of @xmath55 3 - 6 gyrs ( klypin et al . 1999 ) .this situation corresponds roughly to the virialized cgs described in this paper .domnguez - tenreiro & gmez - flechoso ( 1998 , hereafter dg98 ) have shown that if the internal velocity dispersion of the orbiting galaxies is of the order of @xmath21 , then @xmath56 could easily be a factor of @xmath0 2 - 3 longer than the values above , and even much more longer , depending very strongly on the particular values of the gc halo and galaxy parameters . unfortunately , these analytical estimations of @xmath56 are difficult to test through numerical simulations of the evolution of cgs with live halos .the main pitfall is the correct implementation in the simulations of the ratio between the fluctuating forces ( causing dynamical friction ) and the smooth forces , @xmath57 , that must be low to accurately simulate the dynamical friction effects . and so , the proper sampling of a virialized cg - like halo would demand a too high number of particles for current computer facilities .if @xmath57 is unphysically high , it results into undesired numerical effects , leading to an overestimation of the fluctuating force intensity and , consequently , to a too low @xmath56 value .this is the so - called discreetness effect ( see hernquist & weinberg 1989 ; see also eq .( 14 ) in dg98 ) .the simulations of cg evolution by bode , cohn & lugger ( 1993 ) suggest that @xmath56 is roughly consistent with chandrasekhar s formula predictions ; however , these simulations could suffer from a discreetness problem , as they use a too low number of particles per galaxy and halo . in conclusion ,we see that even if the case for inefficient dynamical friction in cgs is not definitively closed , analytical estimations of @xmath56 suggest that , on theoretical grounds , the need of a rapid evolution scenario for virialized cgs becomes rather questionable . a scenario where virialized cgs are dynamically quiescent since the end of the violent relaxation phase ensuing the merger episode that formed their massive dark matter halos could explain a number of difficulties met by the rapid evolution scenario ( see @xmath48[intro ] ) .first , the lack of clear correlations between the rate of galaxy interactions and cg global parameters is naturally explained if the galaxies are embedded in a common massive halo that determines their trajectories .the galaxy - galaxy binary interactions only represent perturbations to the overall halo potential and they are uncorrelated with global cg parameters .as shown for the cg1 m group , if one such binary interaction occurs , the remaining cg galaxies are completely unaffected by this process .binary interactions do destroy cgs if they are important relative to the global field of forces caused by the halo , that is , if the common halo is not massive enough compared with the mass of individual galaxies ( the cg2 group ) .so , a range of cg halo masses could be responsible for the different degrees of dynamical evolutionary stages of cgs found by ribeiro et al .( 1998 ) . butthese could also result from different degrees of virialization after the merging event involving galaxy halos and leading to the common massive dark cg halo .second , the lack of enhanced merger activity _ inside quiescent _ cg halos found in our simulations suggests that cg early - type galaxies are unlikely to form in the quiet phases .it explains the anticorrelation found between the spiral fraction in cgs and @xmath21 or the x - ray luminosity .the cg early - type galaxies can be formed through the same merger trees responsible for cg halo formation in a proto - group environment .the hierarchical scenario predicts that the merger rates are higher in dense environments than in isolation ( evrard , silk & szalay 1990 ; tissera & domnguez - tenreiro 1998 ; see also vzquez - semadeni 1994 ) , where systems form earlier , and lead to more massive compact halos ( cgs would be more likely to appear within looser groups than isolated , see barton , de carvalho & geller 1998 ) and to more early - type galaxies within more massive ( or more x - ray luminous or with a higher @xmath21 , given the correlations among these quantities , see mulchaey 2000 ) cg halos .these pre - virialization violent mergers could also have triggered important gas inflows in the baryonic systems , giving rise , at that time , to starburst galaxies , that evolved into low luminosity agns in the post - merger phase , as the gas supply was gradually exhausted .this could be the origin of the density - morphology - activity relation found by coziol et al .( 1997 , 2000 ) , difficult to explain in the framework of the replenishment scenario ( governato et al . 1996 ) .finally , the lack of enhancement of star formation in the main body of cgs as compared to field galaxies can be seen as another consequence of the density - morphology - activity relation : starbursts would have been induced sometime in the past , and the faster evolving galaxy members , placed at the cg cores , are now observed in a quiet phase , as are early - type galaxies in galaxy clusters . it is a pleasure to thank lobservatoire de genve for its kind hospitality while this work was finished .magf was supported by the direccin general de enseanza superior ( dges , spain ) through a fellowship .it also supported in part this work , grants aen93 - 0673 , pb93 - 0252 and pb96 - 0029 .allam , s.s . , & tucker , d.l .2000 , an , 321 , 101 allam , s.s . , tucker , d.l . ,lin , h. , & hashimoto , y. 1999 , apj , 522 , l89 athanassoula , e. , makino , j. & bosma , a.1997 , mnras , 286 , 825 barnes , j.e .1985 , mnras , 215 , 517 ; 1989 , nature , 338 , 123 barton , e.j . , geller , m. , ramella , m. , marzke , r.o . , &da costa , l.n .1996 , aj , 112 , 871 barton , e.j ., de carvalho , r.r . , & geller , m.j .1998 , aj , 116 , 157 binney , j. , & tremaine , s. 1987 , galactic dynamics ( princeton : princeton univ . press ) coziol , r. , ribeiro , a.l.d . , de carvalho , r.r . , &capelato , h.v .1997 , apj , 493 , 563 coziol , r. , iovino , a. , & de carvalho , r.r .2000 , aj , 120 , 47 diaferio , a. , geller , m.j . , & ramella , m. 1994 , aj , 107 , 868 lcccccccccccccc @xmath39 ( km / s ) & 344 & 326 & 277 & 284 & & 351 & 336 & 297 & 279 & & 266 & 237 & 227 & 233 + @xmath45 ( kpc ) & 8.4 & 9.6 & 14.5 & 14.2 & & 7.6 & 8.9 & 12.0 & 14.6 & & 16.6 & 20.7 & 22.4 & 22.1 + @xmath60 & 7.3 & 7.5 & 8.2 & 8.1&&7.2 & 7.4 & 7.9 & 8.1 & & 8.3 & 8.7 & 8.9&8.8 + @xmath28 ( kpc ) & 32.7 & 42.6 & 72.6 & 88.1&&6.8&38.8&48.6&85.7&&29.7&38.7&45.8&52.5 + @xmath29 ( km / s)&497&465&234&495&&327&489&186&618&&181&79&253&107 + @xmath34 ( mpc km / s)&15.5&18.8&6.0&37.4&&1.2&19.0&7.4&27.8&&4.4&2.9&2.1&5.4 + [ tbl2 ]

<S> we report on results of n - body simulations aimed at testing the hypothesis that galaxies in x - ray emitting ( i.e. , virialized ) compact groups are not tidally stripped when they are embedded in a common , massive , quiescent dark matter halo . to disentangle the effects of interactions from spurious effects due to an incorrect choice of the initial galaxy model configurations , </S><S> these have been chosen to be tidally - limited king spheres , representing systems in quasi - equilibrium within the tidal field of the halo . </S><S> the potential of the halo has been assumed to be frozen and the braking due to dynamical friction neglected . </S><S> our results confirm the hypothesis of low rates of tidal stripping and suggest a scenario for virialized compact group evolution in their quiescent phases with only very moderate tidally induced galaxy evolution can be generally expected . </S><S> this implies the group stability , provided that the dynamical friction timescales in these systems are not much shorter than the hubble time . </S><S> we discuss briefly this possibility , in particular taking account of the similarity between the velocity dispersions of a typical virialized compact groups and the internal velocity dispersion of typical member galaxies . </S><S> a number of puzzling observational data on compact groups can be easily explained in this framework . </S><S> other observations would be better understood as the result of enhanced merging activity in the proto - group environment , leading to virialized compact group formation through mergers of lower mass halos , as predicted by hierarchical scenarios of structure formation . </S>

astro-ph0611403

the _ cobe _ @xmath7 image of the bulge @xcite and models of the projected 2@xmath8 light shows a bar - like structure that is also detected in star counts of red clump stars @xcite .the anomalously high optical depth of the bulge to microlensing @xcite can be explained only by assuming a bar whose major axis extends roughly toward the sun , thus raising the rate of star - star events @xcite .theoretical models of the bulge initially followed axisymmetric models @xcite but have graduated to self - consistent rapidly rotating bars @xcite , with the density and the potential strongly constrained by the observed microlensing rates in the bulge as well as gas kinematics .however , the phase space of the bar is relatively incompletely constrained by stellar kinematic data .study of the kinematics of the bulge is complicated by the large and variable foreground extinction , the presence of a contaminating disk population extending from the foreground well into the galactic center , and source confusion arising from the high density of stars .positional measurements from wide field schmidt plates are therefore impossible , where source confusion makes any astrometric exercise daunting .m giants , while faint in the traditional optical bandpasses due to their cool temperatures and tio bands , are bright in the @xmath9 band @xmath10 and are easy targets for spectroscopy , if positions are known .they further have the advantage of being ubiquitous throughout the bulge and are luminous enough to be studied even in fields with substantial extinction .finally , the short lifetimes of agb stars and luminous giants limit their numbers enough that source confusion is not an issue .@xcite was the first to measure the velocity dispersion of the bulge in baade s window , based on the m giant sample of @xcite that was first classified from low dispersion slitless spectra .multi - fiber spectroscopy of this sample @xcite subsequently harvested roughly 250 bulge giant velocities . despite numerous investigations of the dynamics of stars in the direction of baade s window @xcite and in other bulge fields @xcitethere has been , up to now , no large scale survey of the dynamics of the stellar population in the bulge .@xcite survey the planetary nebula population ; both the rarity of pne ( due to their brief lifetimes ) , the problem of disk or bulge membership , and the considerable distance uncertainty make pne a problematic population ; we believe a well selected sample of m giants is likely the best probe of the bulge / bar population .@xcite showed that the m giant luminosity function is consistent with an old population roughly the age of the halo and globular clusters .the first detailed abundance analysis of m giants @xcite finds an abundance range similar to that of the k giants @xcite . in principle, the most metal - poor component of the bulge population might not evolve through the m giant phase as the k giant abundance distribution is peaked at slightly sub - solar metallicities , but most of the stars are more metal - rich than 47 tuc ( at @xmath11 dex ) and are therefore candidates to reach the luminosities and effective temperatures characteristic of the m giants @xcite .the proper motion study of @xcite addresses a number of fields in the bulge ( avoiding high extinction ) and eventually , a second epoch of _2mass _ imaging would provide proper motions for large numbers of giants .however , the addition of radial velocities and ultimately metal abundances , is needed for a complete dynamical model and the m giant population provides the perfect sample of stars to target .the dynamical model for the bulge / bar has a number of important implications .large samples of uniform radial velocity data are still of great value in constraining the bar vs. axisymmetric models , and the nature of the orbit families supporting the bar .further , the interpretation of the microlensing events in the bulge depends on the use of an accurate dynamical model @xcite . the recent discovery of planetary transit host stars in the bulge @xcite gives an additional incentive to improve our knowledge of the bulge / bar model . with the availability of the two micron all - skysurvey _ 2mass _ @xcite we realized that the key ingredients of high precision positions and photometry would finally be available everywhere in the galactic bulge except near the plane of the galaxy .at this time , we present @xmath12 spectra and have obtained a radial velocity precision of @xmath13 for our most recent ( 2006 ) data . in the past ,optical radial velocity studies in the bulge have not emphasized high precision , because of the large velocity dispersion and the expectation that the short orbital periods would phase - wrap any cold structures out of existence in well under a gyr . yetour precision is sufficient to enable a search for more cold streams analagous to those associated with the sagittarious dwarf spheroidal galaxy ; some candidate cold features are seen and followup observations are underway ( reitzel et al .2007 in prep . ) . herewe report on the kinematics of stars along a band at @xmath1 , obtaining a rotation curve and velocity dispersion profile .the choice for optical spectroscopy is driven by the availability of multi - object wide field spectroscopy .ir spectroscopy , such as that of @xcite , can be used to great effect in the fields of highest reddening but obviously yields far fewer spectra . herewe present a brief sketch of the selection method and analysis ; full details will be given in howard et al.(2007 in prep ) .we use the survey method of @xcite as the model for our program .they observe that stars with @xmath14 have a lower velocity dispersion and are likely disk members .when the same field is examined in _2mass _ , the @xmath15 vs @xmath16 cmd shows a clearly defined red giant branch . in the scw90 field , we find that the magnitude limit of @xmath14 corresponds roughly to @xmath17 .further , there is break in the luminosity function at this @xmath15 magnitude @xcite . while a few bulge agb members may be present at brighter magnitudes, the bulk of such bright stars will be foreground contaminants .we adopt a range of @xmath18 as the basis for selecting the _ 2mass _ m giants ( figure 1 ) .a parallelogram - shaped selection region is adjusted in color and magnitude to encompass completely the red giant branch locus of old stars at the distance and reddening of the field as indicated by the reddening map of @xcite .we find that reddening varies greatly , ranging from @xmath19 . in fields with high extinction, differential reddening is an issue and we widen the selection region to accommodate the red giant branch as observed ( figure 1 ) .the indicated isochrones show that we admit stars with @xmath20[fe / h]@xmath21 a range that spans the entirety of the @xcite abundance distribution we use the hydra multi - fiber spectrograph at the cassegrain focus of the blanco 4 m telescope at cerro tololo .we optimize for spectroscopy in the red , taking advantage of the red colors of m giants and employ the kpgld grating , blazed at 8500 giving 0.45@xmath22 with 2-pixel on - chip binning yielding an effective resolution of 0.88 @xmath23 and a full spectra range of 1800 .we use the @xmath24 slit plate , giving us an increase in resolution ; the loss of light from the slit plate is inconsequential to our s / n because the stars are so bright .our useful spectral range in 2006 was 6891 to 8714 ; the first two lines of the @xmath25 infrared triplet are often well detected and contribute to our success in obtaining a correlation peak . however , there are also stars that are so red that the ca triplet is overwhelmed by tio absorption ; in these cases , the wide wavelength range is essential for the velocity measurement .our exposures are typically @xmath26 but longer exposures were needed when cirrus was present .each field has on average , 108 successfully exposed m giants and 20 fibers that are used to obtain a sky background spectrum .flat fielding , sky subtraction , throughput correction , scattered light subtraction , and wavelength calibration for each exposure were all accomplished using the iraf task dohydra .the spectra are binned to 34.3 @xmath27 and normalized .radial velocities are measured using the fxcor task in iraf ; this requires the spectra to be fourier filtered exclude features greater than 50 pixels or smaller than 3 pixels in extent .regions of the spectrum with obvious telluric features were excluded , leaving only 60% of the spectrum usable for cross correlation .our final cross correlation regions are 7000 - 7150 , 7300 - 7580 , 7700 - 8100 , and 8300 - 8600 .3 standard stars are all used in the cross correlation , with the final velocity for any given star being the average of the 3 derived velocities .the three standards return velocities that agree to within @xmath28 on average , with a standard deviation of @xmath29 in these differences .we report radial velocities for a total of 2294 m giants . herewe consider the fields spanning across @xmath1 .we compare our observed velocities and velocity dispersion with those predicted by the self consistent rotating bulge / bar model of @xcite .we now discuss the results from our study , beginning with the minor axis velocity dispersion ( figure 2 ) resulting from the sum of all fields on the minor axis at @xmath30 to @xmath31 as well as three examples of the fields that went into the summed distribution .the best fit gaussian to the data gives @xmath32 with @xmath33 .a number of peaks are present in this histogram ; these are prominent in the contributing histograms ( figure 2a - c ) . a comparison with the equivalent region extracted from the @xcite model is also shown ; the agreement is very good , but the velocities in the data appear to be more clumpy than they are in the model .more striking peaks @xmath34 are are found in some of our other fields , but because such features can occur in random draws , new observations are required for confirmation ( reitzel et al .2007 ) .figure 3a , b shows our rotation curve and velocity dispersion profile compared with that predicted by @xcite ; we give the data in table 1 .we do not confirm the solid body rotation claimed by a number of previous studies ; after reaching an amplitude of @xmath35 , the rotation curve flattens beyond @xmath36 .the velocity dispersion profile remains higher than @xmath37 even for the fields at @xmath38 .we compare our data with the planetary nebulae ( pne)from @xcite . in order to have a reasonable number of pne, we have accepted those in the range @xmath39 and we have binned the data . considering the less secure distances and assignment of population for the pne , the agreement is good , and settles the question of solid body rotation for the bulge . the @xcite model is a self - consistent rapidly rotating bar that is constrained to have no retrograde orbits . also plotted are the n - body bars of @xcite and @xcite , both of which are n - body bars formed from initially unstable disks .sellwood s bar starts from a rigid plummer sphere with a live kuzmin disk , with mass ratio 3:7 and no dark matter .fux s bar starts from an equilibrium of dark halo , power law nucleus , and an exponential disk ; in contrast to the zhao model that is fit to the data , a best - fit model is selected from a large number of n - body realizations . we adopt the bar angles and mass normalisations as suggested by @xcite that give an overall good match with the appearance of the cobe map .we project these models and calculate the velocity moments in each line of sight without distinguishing disk and bulge particles .neither model is satisfactory : fux s best model gives a remarkably good fit to the rotation curve but has too low @xmath40 .sellwood s bar appears to have a very high @xmath40 .it appears that our data is challenging to fit by both disk - instability formed bars as well as schwarzschild models with fixed potentials . in figure 4, we compare the @xcite model in longitude - velocity space with our radial velocities , both sampling the slice at @xmath41 ; we have smoothed the data in longitude by @xmath42 and @xmath43 in velocity .the addition of retrograde orbits may improve the agreement of the model and data ; we are working toward this end ( zhao et al .2007 in prep . ) .while our claim of slower rotation for the bulge appears to contradict earlier studies , we emphasize our agreement with the pne .it is also interesting that the kinematics of k giants in the @xmath44 field of @xcite agrees with our rotation curve ( 44.5 @xmath45 ) ; faster rotation ( and a lower velocity dispersion ) is observed in his @xmath46 but that field that might be disk dominated .@xcite used a small number of miras to sample velocities over a wide range in longitude , and found rapid solid body rotation .small numbers , or contamination by members of the disk , might be invoked to explain the more rapid rotation observed for sio masers @xcite .our data are clearly at odds with the widely held view that the bulge rotates as a solid body .the relatively high velocity dispersions of our fields is reassuring in the sense that the m giant selection criterion is yielding good kinematic probes of the bulge / bar population . extending this survey to a larger number of fields in the bulge offers the possibility of undertaking detailed tests of dynamical models , something that is not presently possible in distant galaxy populationsthis will give new insights into the structure and dynamics of the bulge , and into the formation of the milky way .rmr , dbr , and cdh acknowledge support from stsci grant go-10816 and nsf grant ast-0709731 .we thank the staff of ctio , especially knut olsen and angel guerra , and andreas koch and jan kleyna for helpful comments on drafts .we use data products from the two micron all sky survey , which is a joint project of the university of massachusetts and the infrared processing and analysis center / caltech , funded by nasa and the nsf . _facilities : _ .-9.98 & -3.99 & 17:36:31.47 & -39:30:42.9 & 93 & -60.8 & 7.5 & 72.1 & 5.3 + -9.01 & -4.00 & 17:39:10.83 & -38:41:41.7 & 101 & -33.4 & 7.5 & 74.9 & 5.3 + -7.98 & -4.00 & 17:41:56.69 & -37:49:13.2 & 101 & -46.3 & 7.5 & 75.7 & 5.3 + -6.98 & -3.99 & 17:44:33.22 & -36:57:55.3 & 103 & -61.3 & 8.5 & 85.9 & 6.0 + -6.00 & -3.99 & 17:47:03.54 & -36:07:46.5 & 105 & -42.8 & 8.5 & 87.0 & 6.0 + -3.99 & -3.98 & 17:52:02.74 & -34:24:11.0 & 108 & -44.6 & 9.3 & 96.4 & 6.6 + -3.01 & -3.99 & 17:54:26.79 & -33:33:26.6 & 111 & -45.7 & 10.5 & 111.0 & 7.5 + -1.99 & -3.98 & 17:56:50.87 & -32:40:49.5 & 108 & -25.3 & 10.1 & 104.8 & 7.1 + -0.02 & -3.99 & 18:01:28.27 & -30:58:18.6 & 110 & -14.7 & 10.8 & 112.8 & 7.6 + 1.04 & -3.95 & 18:03:43.51 & -30:01:49.9 & 73 & -1.1 & 12.9 & 110.2 & 9.1 + 2.00 & -4.00 & 18:06:02.64 & -29:13:17.4 & 109 & 29.0 & 11.0 & 115.2 & 7.8 + 3.99 & -3.98 & 18:10:19.36 & -27:28:32.3 & 106 & 29.0 & 10.2 & 105.2 & 7.2 + 6.01 & -3.99 & 18:14:40.05 & -25:42:06.9 & 108 & 49.3 & 8.5 & 88.3 & 6.0 + 6.99 & -3.96 & 18:16:34.61 & -24:49:41.9 & 104 & 29.8 & 9.2 & 94.2 & 6.5 + 7.99 & -3.99 & 18:18:44.88 & -23:57:51.7 & 107 & 44.0 & 8.5 & 88.2 & 6.0 + 9.01 & -3.98 & 18:20:47.51 & -23:03:31.1 & 104 & 20.1 & 7.2 & 73.9 & 5.1 + 10.00 & -3.99 & 18:22:50.13 & -22:11:12.6 & 106 & 34.4 & 8.2 & 84.0 & 5.8 +

<S> we are undertaking a large scale radial velocity survey of the galactic bulge which uses m giant stars selected from the _ </S><S> 2mass _ catalog as targets for the ctio 4 m hydra multi - object spectrograph . </S><S> the aim of this survey is to test dynamical models of the bulge and to quantify the importance , if any , of cold stellar streams in the bulge and its vicinity . </S><S> here we report on the kinematics of a strip of fields at @xmath0 and @xmath1 . </S><S> we construct a longitude - velocity plot for the bulge stars and the model data , and find that contrary to previous studies , the bulge does not rotate as a solid body . from @xmath2 </S><S> the rotation curve has a slope of roughly @xmath3 and flattens considerably at greater @xmath4 and reaches a maximum rotation of @xmath5 . </S><S> we compare our rotation curve and velocity dispersion profile to both the self - consistent model of @xcite and to n - body models ; neither fits both our observed rotation curve and velocity dispersion profile . </S><S> the high precision of our radial velocities @xmath6 yields an unexpected result : hints of cold kinematic features are seen in a number of the line of sight velocity distributions . </S>

cond-mat0103266

more than half a century ago kramers proposed a diffusion model for chemical reactions in terms of the theory of brownian motion in phase space @xcite . since then the model and several of its variants have been ubiquitous in many areas of physics , chemistry and biology for understanding the nature of activated processes in classical @xcite , quantum and semiclassical @xcite systems , in general .these have become the subject of several reviews @xcite and monograph @xcite in the recent past . in the majority of these treatments oneis essentially concerned with an equilibrium thermal bath at a finite temperature which stimulates the reaction co - ordinate to cross the activation energy barrier .the inherent noise of the medium is of internal origin .this implies that the dissipative force which the system experiences in course of its motion in the medium and the stochastic force acting on the system as a result of random impact from the constituents of the medium arise from a common mechanism . from a microscopic point of viewthe system - reservoir hamiltonian description @xcite developed over the decades suggests that the coupling of the system and the reservoir co - ordinates determines both the noise and the dissipative terms in the langevin equation describing the motion of the system .it is therefore not difficult to anticipate that these two entities get related through a fluctuation - dissipation relation @xcite ( these systems are sometimes classified as thermodynamically closed system in contrast to the systems driven by external noise in nonequilibrium statistical mechanics @xcite ) .however , when the reservoir is modulated by an external noise it is likely that this relation gets affected in a significant way . since the modulation of the reservoir crucially depend on its response function, one can further envisage a connection between the dissipation of the system and the response function of the reservoir due to the external noise from a microscopic point of view . in the present paperwe explore this connection in the context of activated rate processes when the reservoir is modulated by an external noise .specifically our object here is twofold : ( i ) to explore the role of reservoir response as a function of external noise on the system dynamics in contrast to direct driving of the system by external noise , ( ii ) to calculate the generalized kramers rate for the steady state of this nonequilibrium open system taking full care of thermodynamic consistency condition .both the internal and the external noises are gaussian , stationary and are characterized by arbitrary decaying correlation functions .while the internal noise of the reservoir is thermal , the external noise may be of thermal or non - thermal type .we consider the stochastic motion to be spatial diffusion limited and calculate the rate of escape in the intermediate to strong damping regime .it is worth to mention that the externally generated nonequilibrium fluctuations can bias the brownian motion of a particle in an anisotropic medium and may be used for designing molecular motors and pumps @xcite .we further mention that nonequilibrium , non - thermal systems have also been investigated phenomenologically by a number of workers in several other contexts , e.g. , for examining the role of color noise in stationary probabilities @xcite , properties of nonlinear systems @xcite , nature of crossover @xcite , rate of diffusion limited coagulation processes @xcite , effect of monochromatic noise @xcite , etc . while these treatments concern direct driving of the system by an external noise the present consideration is based on modulation of the bath. a number of different situations depicting the modulation of the bath by an external noise may be physically relevant . as ,for example , we consider a simple unimolecular conversion ( say , an isomerization reaction ) from @xmath0 .the reaction is carried out in a photochemically active solvent under the influence of external fluctuating light intensity . since the fluctuations in the light intensity result in the fluctuations in the polarization of the solvent molecules , the effective reaction field around the reactant system gets modified .provided the required stationarity of this nonequilibrium open system is maintained ( which is not difficult in view of the experiments performed in the studies of external noise - induced transitions in photochemical systems @xcite ) the dynamics of barrier crossing becomes amenable to the present theoretical analysis that follows .the remaining part of this paper is organized as follows : in sec .ii we discuss a system - reservoir model where the later is modulated by an external noise and establish an interesting connection between the dissipation of the system and the response function of the reservoir due to external noise .the stochastic motion in a linearized potential field is described in terms of a fokker - planck equation in sec .iii . based on the traditional flux over population method@xcite we derive in sec .iv the generalized expression for the kramers rate of escape from a metastable well . in sec .v we numerically analyze the model and the bath modulated dynamics for the full potential and verify the theoretical rate with numerical simulation .the paper is concluded in sec .we consider a classical particle of mass @xmath1 is linearly coupled to a heat bath of @xmath2 harmonic oscillators driven by an external noise .the total hamiltonian is given by @xcite @xmath3 in eq.([eq1 ] ) , @xmath4 and @xmath5 are the co - ordinate and momentum of the system particle ; @xmath6 are the variables associated with the @xmath7-th oscillator and @xmath8 and @xmath9 are the corresponding frequency and mass , respectively . @xmath10measures the interaction between the particle and the bath .@xmath11 is the potential energy of the particle .@xmath12 is assumed to be of the form @xmath13 the coupling function @xmath14 measures the strength of interaction and @xmath15 is the external noise which we assume to be stationary and gaussian with zero mean , i.e. , @xmath16 and is characterized by an arbitrary correlation function as follows ; @xmath17 here @xmath18 implies the averaging over the external noise .we then eliminate the bath degrees of freedom in the usual way @xcite to obtain the following generalized langevin equation @xmath19 [ while constructing eq.([eq4 ] ) we have set @xmath1 and @xmath9 equal to unity ] where @xmath20 @xmath21 is the internal forcing function generated through the coupling between the system and the heat bath and is given by @xmath22 \ ; \omega_i^2 \ ; \cos \omega_i t + v_i ( 0 ) \ ; \omega_i \ ; \sin \omega_i t \right \ } \ ; \ ; .\ ] ] in eq.([eq4 ] ) , @xmath23 is a fluctuating force term due to the external noise @xmath15 and is given by @xmath24 where @xmath25 the form of eq.([eq4 ] ) therefore suggests that the system is driven by two forcing functions @xmath21 and @xmath23 .@xmath21 depends on the initial conditions of the bath oscillators for a fixed choice of the initial condition of the system degrees of freedom . to define the statistical properties of @xmath21, we assume that the _ initial distribution _ is one in which the bath is equilibrated at @xmath26 in the _ presence _ of the system but in the _ absence _ of the external noise agency such that @xmath27 and @xmath28 .now , at @xmath29 , the external noise agency is switched on and the bath is modulated by @xmath15 .the system is governed by eq.([eq4 ] ) , where , apart from the internal noise @xmath21 , another fluctuating force @xmath23 appears , that depends on the external noise @xmath15 .therefore , one can define an effective noise @xmath30 whose correlation is given by @xmath31 along with @xmath32 , where @xmath33 means we have taken two averages independently .it should be emphasized that the above relation ( [ neq6 ] ) is not a fluctuation - dissipation relation due to the appearance of the external noise intensity .rather it serves as a thermodynamic consistency condition .let us now digress a little bit about @xmath23 .the statistical properties of @xmath23 are determined by the normal mode density of the bath frequencies , the coupling of the system with the bath , the coupling of the bath with the external noise and the external noise itself .eq.([eq8 ] ) is reminiscent of the familiar linear relation between the polarization and the external field where @xmath34 and @xmath35 play the role of the former and the later , respectively .@xmath36 can then be interpreted as a response function of the reservoir due to external noise @xmath15 .the very structure of @xmath23 suggests that this forcing function although originating from an external force is different from a direct driving force acting on the system .the distinction lies at the very nature of the bath characteristics ( rather than system characteristics ) as reflected in the relation ( [ eq8 ] ) and ( [ eq9 ] ) . with the coupling coefficients @xmath37 and @xmath38 , in the continuum limit @xcite @xmath39 and @xmath36reduce to the following forms @xmath40 and @xmath41 where @xmath42 and @xmath43 are constants and @xmath44 is the cutoff frequency of the oscillator .@xmath45 is the density of modes of the heat bath . from the above two relations , we obtain @xmath46 eq.([eq14 ] )is an important content of the present model .this expresses how the dissipative kernel @xmath39 depends on the response function @xmath36 of the medium due to external noise @xmath15 [ see eq.([eq8 ] ) ] .such a relation for the open system can be anticipated in view of the fact that both the dissipation and the response function crucially depend on the properties of the reservoir especially on its density of modes and its coupling to the system and the external noise source . in what follows weshall be concerned with the consequences of this relation in terms of the langevin description in the next section ( eq.([eq15 ] ) ) and numerical analysis of the full model potential in sec .we now consider the system to be a harmonically bound particle of unit mass and of frequency @xmath47 . then because of eq.([eq8 ] ) the langevin equation ( [ eq4 ] ) becomes @xmath48 the laplace transform of eq .( [ eq15 ] ) allows us to write a formal solution for the displacement of the form @xmath49 where we have made use of the relation ( [ eq14 ] ) explicitly .here @xmath50 with @xmath51 and @xmath52 being the initial position and initial velocity of the oscillator , respectively , which are nonrandom and @xmath53 \ ; \ ; .\ ] ] the kernel @xmath54 is the laplace inversion of @xmath55 where , @xmath56 , is the laplace transform of the friction kernel @xmath39 , and @xmath57 the time derivative of eq.([eq16 ] ) yields @xmath58 where @xmath59 @xmath60next we calculate the variances . from the formal solution of @xmath61 and @xmath62 ,the explicit expressions for the variances are obtained which are given below ; @xmath63 ^ 2 \rangle\rangle \nonumber \\ & = & 2 \int_0^t dt_1 \ ; h ( t_1 ) \int_0^{t_1 } dt_2 \ ; h ( t_2 ) \ ; \langle f(t_1 ) f(t_2 ) \rangle \nonumber \\ & & + 2 \left ( \frac{\kappa_0}{g_0 } \tau_c \omega_0 ^ 2 \right ) ^2 \int_0^t dt_1 \ ; h ( t_1 ) \int_0^{t_1 } dt_2 \ ; h ( t_2 ) \ ; \langle \epsilon ( t_1 ) \epsilon ( t_2 ) \rangle_e \nonumber \\ & & + 2 \left ( \frac{\kappa_0}{g_0 } \tau_c\right ) ^2 \int_0^t dt_1 \ ; h_2 ( t_1 ) \int_0^{t_1 } dt_2 \ ; h_2 ( t_2 ) \ ; \langle \epsilon ( t_1 ) \epsilon ( t_2 ) \rangle_e \nonumber \\ & & + 2 \left ( \frac{\kappa_0}{g_0 } \tau_c \right ) ^2 \omega_0 ^ 2 \ ; \int_0^t dt_1 \ ; h ( t_1 ) \int_0^{t_1 } dt_2 \ ; h_2 ( t_2 ) \ ; \langle \epsilon ( t_1 ) \epsilon ( t_2 ) \rangle_e \ ; \ ; , \end{aligned}\ ] ] @xmath64 ^ 2 \rangle\rangle \nonumber \\ & = & 2 \int_0^t dt_1 \ ; h_1 ( t_1 ) \int_0^{t_1 } dt_2 \ ; h_1 ( t_2 ) \ ; \langle f(t_1 ) f(t_2 ) \rangle \nonumber \\ & & + 2 \left ( \frac{\kappa_0}{g_0 } \tau_c \omega_0 ^ 2 \right ) ^2 \int_0^t dt_1 \ ; h_1 ( t_1 ) \int_0^{t_1 } dt_2 \ ; h_1 ( t_2 ) \ ; \langle \epsilon ( t_1 ) \epsilon ( t_2 ) \rangle_e \nonumber \\ & & + 2 \left ( \frac{\kappa_0}{g_0 } \tau_c\right ) ^2 \int_0^t dt_1 \ ; h_3 ( t_1 ) \int_0^{t_1 } dt_2 \ ; h_3 ( t_2 ) \ ; \langle \epsilon ( t_1 ) \epsilon ( t_2 ) \rangle_e \nonumber \\ & & + 2 \left ( \frac{\kappa_0}{g_0 } \tau_c \right ) ^2 \omega_0 ^ 2 \ ; \int_0^t dt_1 \ ; h_1 ( t_1 ) \int_0^{t_1 } dt_2 \ ; h_3 ( t_2 ) \ ; \langle \epsilon ( t_1 ) \epsilon ( t_2 ) \rangle_e \ ; \ ; , \end{aligned}\ ] ] and @xmath65 [ v ( t ) - \langle\langle v ( t ) \rangle\rangle ] \rangle\rangle \nonumber \\ & = & \frac{1}{2 } \dot{\sigma}_{xx}^2 ( t ) \end{aligned}\ ] ] where we have assumed that the noises @xmath21 and @xmath15 are symmetric with respect to the time argument and have made use the fact that @xmath21 and @xmath15 are uncorrelated . due to the gaussian property of the noises @xmath21 and @xmath15 and the linearity of the langevin equation ( [ eq15 ] ), we see that the joint probability density @xmath66 of the oscillator must be gaussian . the joint characteristic function associated with the densityis @xmath67 \right \ } \ ; \ ; .\ ] ] using the method of characteristic function @xcite and the above expression ( [ eq28 ] ) we find the general fokker - planck equation associated with the probability density function @xmath66 for the process ( [ eq15 ] ) ; @xmath68 where @xmath69 + h^2 ( t ) \ ; \ ; .\end{aligned}\ ] ] the functions @xmath70 and @xmath71 are defined by @xmath72 where the covariances are to be calculated for a particular given noise process . for the internal noise processesit had been shown earlier that for several models the various time dependent parameters @xmath73 , @xmath74 , etc .do exist asymptotically as @xmath75 .the above consideration shows that @xmath54 , @xmath76 , etc .do not depend on the nature of the noise but depend only on the relaxation @xmath74 .we now discuss the asymptotic properties of @xmath70 and @xmath71 , which in turn are dependent on the variances @xmath77 and @xmath78 , as @xmath75 since they play a significant role in our further analysis that follows . from eqs .( [ eq25 ] ) and ( [ eq26 ] ) , we may write @xmath79 where ` @xmath7 ' denotes the part corresponding to internal noise @xmath21 and ` @xmath80 ' corresponds to the external noise @xmath15 .since the average velocity of the oscillator is zero as @xmath75 we see from eq.([eq22 ] ) that @xmath81 and @xmath82 must be zero as @xmath75 . also from eq.([eq17 ] ) we observe that the function @xmath83 must decay to zero for long times . hence , from eq.([eq18 ] ) we see that the stationary value of the integral of @xmath54 is @xmath84 , i.e. , @xmath85 now , @xmath86 and @xmath87 of eqs.([eq25 ] ) and ( [ eq26 ] ) can be written in the form @xmath88 \label{eq30b}\end{aligned}\ ] ] and @xmath89 \ ; \ ; . \label{eq30c}\end{aligned}\ ] ] from the above two expressions [ eqs.([eq30b ] ) and ( [ eq30c ] ) ] we see that @xmath90 it is important to note that these stationary values are not related to the intensity and correlation time of the internal noise .we next consider the parts , @xmath91 and @xmath92 , due to the presence of the external noise .the laplace transform of eq.([eq16 ] ) yields the expression @xmath93 where @xmath94 } \right \ } x ( 0 ) + \frac{1}{s^2 + s \tilde{\gamma } ( s ) + \omega_0 ^ 2 } \ ; v ( 0 ) \nonumber \\ & = & \left \ { \frac{1}{s } - \omega_0 ^ 2 \frac { \tilde{h } ( s)}{s } \right \ } x ( 0 ) + \tilde{h } ( s ) v ( 0 ) \ ; \ ; . \label{eq30f}\end{aligned}\ ] ] from the above equation ( [ eq30e ] ) we can calculate the variance @xmath95 in the laplace - transformed space which can be identified as the laplace transform of eq.([eq25 ] ) .thus , for the part @xmath91 we observe that , @xmath96 contains terms like @xmath97 . since, we have assumed the stationarity of the noise @xmath15 , we conclude that if @xmath98 exists [ where @xmath99 ] , then the stationary value of @xmath91 exists and becomes a constant that depends on the correlation time and the strength of the noise .similar argument is also valid for @xmath92 . summarizing the above discussions we note that , \(i ) the internal noise - driven parts of @xmath77 and @xmath78 , i.e. , @xmath100 and @xmath101 approach the fixed values which are independent of the noise correlation and the intensity as @xmath75 , \(ii ) the external noise driven parts of variances also approach the constant values at the stationary ( @xmath75 ) limit which are dependent on the strength and the correlation time of the noise . hence we conclude , following the ref.(33 ) and our preceding discussions that even in presence of an external noise the above terms do exist asymptotically and we write the steady state fokker - planck equation for the asymptotic values of the parameters as , @xmath102 where , @xmath103 , @xmath104 , @xmath105 , @xmath106 , etc .are to be calculated from the general definition ( [ eq30 ] ) for the steady state . as an explicit example we consider the case of a @xmath107-correlated external noise and ornstein - uhlenbeck internal noise for which we provide the expressions for variances@xmath77 , @xmath78 and @xmath108 and the relaxation function @xmath54 given in the appendix - a. the general steady state solution of the above equation ( [ eq31 ] ) is @xmath109\ ] ] where @xmath110 and @xmath111 is the normalization constant .the solution ( [ eq32 ] ) can be verified by direct substitution .the distribution ( [ eq32 ] ) is not an equilibrium distribution .this stationary distribution for the nonequilibrium open system plays the role of an equilibrium distribution of the closed system which may , however , be recovered in the absence of external noise term .we now turn to the problem of decay of a metastable state . in kramersapproach @xcite , the particle coordinate @xmath4 corresponds to the reaction coordinate and its values at the minima of the potential @xmath11 denotes the reactant and product states . linearizing the motion around barrier top at @xmath112 the langevin equation ( [ eq4 ] )can be written down as @xmath113 where , @xmath114 and the barrier frequency @xmath115 is defined by @xmath116 correspondingly the motion of the particle is governed by the fokker - planck equation ( [ eq29 ] ) @xmath117 where , the suffix ` @xmath118 ' indicates that all the coefficients are to be calculated using the general definition ( [ eq30 ] ) for the barrier top region .it is apparent from eqs.([eq31 ] ) and ( [ eq36 ] ) that since the dynamics is non - markovian and the system is thermodynamically open one has to deal with the renormalized frequencies @xmath119 and @xmath120 near the bottom or top of the well , respectively .we make the ansatz that the nonequilibrium , steady state probability @xmath121 , generating a nonvanishing diffusion current @xmath122 , across the barrier is given by @xmath123 \ ; \xi ( x , v)\ ] ] where @xmath124 @xmath125 is the renormalized linear potential as @xmath126 with @xmath103 , @xmath127 .the unknown function @xmath128 obeys the natural boundary condition that for @xmath129 , @xmath128 vanishes .the ansatz of the form ( [ eq37 ] ) denoting the steady state distribution is motivated by the local analysis near the bottom and the top of the barrier in the kramers sense @xcite . for a stationary nonequilibrium system , on the other hand , the relative population of the two regions , in general , depends on the global properties of the potential leading to an additional factor in the rate expression .although because of the kramers type ansatz @xcitewhich is valid for the local analysis , such a consideration is outside the scope of the present treatment , we point out a distinctive feature in the ansatz ( [ eq37 ] ) compared to kramers ansatz . while in the latter case one considers a complete factorization of the equilibrium part ( boltzmann ) and the dynamical part , the ansatz ( [ eq37 ] ) incorporates the additional dynamical contribution through dissipation and strength of the noise into the exponential part .this modification of kramers ansatz ( by dynamics ) is due to nonequilibrium nature of the system .thus unlike kramers , the exponential factors in ( [ eq37 ] ) and in the stationary distribution ( [ eq32 ] ) which serves as a boundary condition are markedly different . before carrying out global analysis in the present section our aim hereis to understand the modification of the rate due to modulation of the bath driven by an external noise , within the perview of kramers type ansatz .the internal consistency of the treatment , however , can be checked by recovering the kramers result when the external noise is switched off . from equation ( [ eq36 ] ) , using ( [ eq37 ] )we obtain the equation for @xmath130 in the steady state in the neighborhood of @xmath131 , the equation @xmath132\frac{\partial \xi}{\partial v } + \phi_b ( \infty ) \frac{\partial^2 \xi}{\partial v^2 } + \psi_b ( \infty ) \frac{\partial^2 \xi}{\partial v \partial y } = 0 \ ; \ ; .\ ] ] after making use of the appropriate transformations and boundary conditions for reduced distribution functions@xcite we obtain the barrier crossing rate @xmath133 given by @xmath134\ ] ] where @xmath135 with @xmath136 here @xmath137 is the activation energy , @xmath138 . since the temperature due to internal thermal noise , the strength of the external noise and the damping constant are buried in the parameters @xmath139 , @xmath140 , @xmath141 , @xmath141 and @xmath142 the generalized expression look somewhat cumbersome .we point out that the subscripts ` @xmath143 ' and ` @xmath118 ' in @xmath144 and @xmath145 refer to the well or barrier top region , respectively .eq.([eq54 ] ) is one of the key results of this paper .we note here that @xmath146 in the exponential factor defines a new _ effective _ temperature characteristic of the steady state of the nonequilibrium open system and an _ effective _ transmission factor is contained in the prefactor controlling the barrier crossing dynamics .as expected both are the functions of the external noise strength @xmath144 and coupling of noise to the bath modes .the dependence of the rate on the parameters can be exposed explicitly once we consider the typical cases .we first explore the distinctive aspects of the external noise - driven - reservoir - modulated dynamics of the system ( in contact with the external noise driven bath ) in contrast to direct driving of the system by the external noise .this will help us to elucidate the special role of the reservoir response function in controlling the rate . to this endwe first return to our basic equation of the model , eq.([eq4 ] ) where @xmath147 as given by eq.([eq9 ] ) characterizes the response function .we solve this equation numerically using second order stochastic algorithm of fox @xcite for the full potential @xmath148 and plot the results of computation of the inverse of the mean first passage time as a function of the damping constant @xmath149 ( see eq.([eq12 ] ) ) in fig.(1 ) .the density of modes of the bath has been assumed to be of the standard lorentzian form , @xmath150 with vanishing internal correlation time @xmath151 .typically for the other scaled parameters we have used @xmath152 = 0.1 , the strength of the @xmath107-correlated external noise , @xmath144 = 0.1 .the mean first passage times have been calculated by averaging over 10000 trajectories .we first show the continuous curve in fig.(1 ) which represents the case of direct driving by the external @xmath107-correlated noise .what is immediately apparent is the loss of turnover of the rate as one allows the variation of the dissipation constant @xmath149 from low to high friction regime .this is in sharp contrast to what is shown by the dotted curve in fig.(1 ) which depicts the situation of bath modulation with the response function @xmath147 determined by eq.([eq13 ] ) .both the curves ( continuous and dotted ) representing the open system are compared to that for the standard kramers turnover , i.e. , when the dynamics is calculated in absence of the external noise ( dashed curve ) .as expected the rate in the driven system ( be it directly or through bath modulation ) is always higher that in the undriven one .fig.(1 ) also shows that although in the low damping region the direct driving causes a much higher rate , it is , however , the bath modulation which becomes more effective in inducing activated barrier crossing in the high friction regime .it is thus interesting to note that the barrier crossing dynamics of the system in contact with an external noise modulated bath captures the essential turnover features of the kramers dynamics of the closed system .we therefore realize that although open , the nature of the response function of the reservoir as well as the thermodynamic consistency condition ( [ neq6 ] ) , make the open system feel like a closed system . in fig.(2 )we show the variation of inverse of the mean first passage time as a function of the strength of the external noise @xmath144 , keeping all other parameter same as before but for @xmath149 = 2.0 .it is apparent that the barrier crossing is more facilitated by modulating the bath than driving the system directly for higher values of external noise strength .so far we have considered the full potential @xmath148 and vanishingly small correlation times for external and internal noises for numerical computation .we now turn to our basic theoretical result eq.([eq54 ] ) which is a generalization of kramers rate for bath modulated dynamics for intermediate to strong damping regime . to examine its validitywe calculate the rate ( [ eq54 ] ) as a function of the damping constant @xmath149 ( @xmath42 is related to both @xmath39 and @xmath36 in eqs.([eq12 ] ) and ( [ eq13 ] ) respectively ) for several values of external noise strength @xmath144 .the scaled barrier height @xmath137 and @xmath152 have been set to 0.25 and 0.1 , respectively .the results are plotted in fig.(3 ) ( continuous curve ) and compared to the rate , inverse of the mean first passage time ( dotted curve ) calculated numerically using full potential for the set of parameter values ( as given in the earlier subsection ) with eq.([eq4 ] ) . in figs.(4 ) and ( 5 ) we further compare our theoretical results ( continuous curve ) with numerical simulation ( dotted curve ) for the variation of rate as a function of external noise strength and its correlation time , respectively for several values of damping constant .it is apparent from figs.(3 - 5 ) that the theoretical and numerical results are in good agreement .based on a system - reservoir microscopic model where the reservoir is modulated by an external , stationary and gaussian noise with arbitrary decaying correlation function , we have numerically analyzed the model and generalized the kramers theory to calculate the steady state rate of escape from a metastable well .the main conclusions of this study are as follows ; \(i ) we have shown that since the reservoir is driven by the external noise and the dissipative properties of the system depend on the reservoir , a simple connection between the dissipation and the response function of the medium due to the external noise can be established .this connection is important for realizing an effective temperature of the reservoir characterizing the stationary state of the thermodynamically open system , as well as an _ effective transmission factor _ controlling the rate .both of these quantities depend on the strength and correlation time of the external noise .\(ii ) many of the earlier treatments of the rate concern direct ( phenomenological ) driving of the system and did not emphasize the question of thermodynamic consistency in the context of open systems .the present theory being microscopic the fluctuation - dissipation - like relation ( [ neq6 ] ) remains an inbuilt characteristic of the model itself as an essential thermodynamic consistency condition .\(iii ) based on numerical simulation of the full model potential we show that one can recover the turnover features of the kramers dynamics when the external noise modulates the reservoir rather than the system directly .this recovery , we believe , is an offshoot of the derived thermodynamic consistency condition ( [ neq6 ] ) .\(iv ) provided the long time limit of the moments for the stochastic processes pertaining to the external and internal noises characterized by arbitrary decaying correlation functions exist , the expression for generalized kramers rate of barrier crossing for the open system we derive here is fairly general . we have shown that it agrees reasonably well with numerical simulation using full potential for this model .the creation of a typical nonequilibrium open situation by modulating a bath with the help of an external noise is not an uncommon phenomenon in applications and industrial processing .the external agency generating noise does work on the bath by stirring , pumping , agitating , etc . , to which the system dissipates internally . in the present treatment we are concerned with a nonequilibrium steady state which signifies a constant throughput of energy in contrast to thermal equilibrium defined by an constant temperature .we believe that these considerations are likely to be important in other related issues in nonequilibrium open systems and may serve as a basis for studying processes occurring within irreversibly driven environments @xcite and for thermal ratchet problems @xcite .we thank e. pollak for his kind interest in this work and for pointing out the ref .( 35 ) to us .skb in indebted to council of scientific and industrial research ( csir ) , government of india for financial support .we consider a particular case as an example where the external noise @xmath15 is @xmath107-correlated and the internal noise is an ornstein - uhlenbeck process , i.e. , @xmath153 @xmath169^{1/2 } \ ; \ ; , \\\alpha & = & \tan^{-1 } \left ( \frac{\beta}{d-\delta_2 } \right ) \ ; \ ; { \rm and}\\ { \cal q } & = & -\frac{a^2b^2}{108 } + \frac{b^3}{27 } + \frac{a^3 c_0}{27 } - \frac{abc_0}{6 } + \frac{c_0 ^ 2}{4 } \ ; \ ; .\end{aligned}\ ] ] here we note that for a physically allowed solution @xmath165 , @xmath166 must be positive . since by eq.([eq19 ] ) @xmath54 depends on the memory kernel @xmath39 which is of decaying type and all the moments , in general , reach asymptotic constancy as shown in sec .iii , these quantities are positive ( which depends on the correlation time @xmath151 , the strength of the noise and other potential parameters ) which may be checked ( after some algebra ) by considering the limiting cases such as @xmath156 and @xmath170 large . where @xmath172 \nonumber \\ & & + k_bt \left \ { -\frac{c_1}{\delta_1 }e^{-\delta_1 t } \left [ 2 - 2 \omega_0 ^ 2 c_2 r - \frac{2 \omega_0 ^ 2 c_1}{\delta_1 } + e^{-\delta_1 t } \left ( \delta_1 c_1 + \frac{\omega_0 ^ 2 c_1}{\delta_1 } \right ) \right ] \right . \nonumber \\ & & - \frac{2 c_2 e^{-\delta_2 t } } { \delta_2 ^ 2 + \beta^2 } \left [ 1 - \omega_0 ^ 2 c_2 r + \frac{\omega_0 ^ 2 c_1}{\delta_1 } ( e^{-\delta_1 t } - 1 ) \right ] \left [ \delta_2 \sin ( \beta t + \alpha ) + \beta \cos( \beta t + \alpha ) \right ] \nonumber \\ & & - 2 c_1 c_2 e^ { - ( \delta_1+\delta_2 ) t } \sin ( \beta t + \alpha ) \nonumber \\ & & - \frac{\delta_2 \beta \omega_0 ^ 2 c_2 ^ 2 e^{-2\delta_2 t } } { ( \delta_2 ^ 2+\beta^2)^2 } \sin 2 ( \beta t + \alpha ) - \frac{\beta^2 \omega_0 ^ 2 c_2 ^ 2 e^{-2\delta_2 t } } { ( \delta_2 ^ 2+\beta^2)^2 } \nonumber \\ & & \left .+ \left [ \frac { \omega_0 ^ 2 ( 2\beta^2 - \delta_2 ^ 2 ) } { ( \delta_2 ^ 2+\beta^2)^2 } -1 \right ] c_2 ^ 2 e^{-2\delta_2 t } \sin^2 ( \beta t + \alpha ) \right \}\end{aligned}\ ] ] where , @xmath175 s are defined in eq.([eqa4]-[eqa5 ] ) .the explicit expression for @xmath108 can be derived from eq.([eq27 ] ) . in the limit @xmath75we calculate the stationary values of the variances .the variances @xmath180 , @xmath181 and @xmath182 yield @xmath183 and @xmath184 and other relevant quantities .

<S> we consider a system - reservoir model where the reservoir is modulated by an external noise . </S><S> both the internal noise of the reservoir and the external noise are stationary , gaussian and are characterized by arbitrary decaying correlation functions . </S><S> based on a relation between the dissipation of the system and the response function of the reservoir driven by external noise we numerically examine the model using a full bistable potential to show that one can recover the turn - over features of the usual kramers dynamics when the external noise modulates the reservoir rather than the system directly . we derive the generalized kramers rate for this nonequilibrium open system . </S><S> the theoretical results are verified by numerical simulation . </S>

cond-mat0312469

the screening of charged macromolecules in an electrolyte solution is a long standing problem which has prompted many attempts aiming at a theoretical explanation . in their pioneering work gouy @xcite and chapman @xciteused what is now referred to as poisson - boltzmann ( pb ) theory as the basis for a mean field treatment of the electrical double layer .this approach found its culmination about thirty years later in the famous dlvo theory of charged colloids @xcite .the major flaw of these mean field approaches is their neglect of correlations between the ions .the first attempt to work out such correlations for _ homogeneous _ electrolytes are due to debye and hckel @xcite , whose work remarkably ( and at first glance confusingly ) is also based on ( linearized ) poisson - boltzmann theory . in the _inhomogeneous _ case integral equation theories @xcite and recently field theories @xcite have become very popular in calculating correlation corrections to mean field double layers .however , in order to make progress and calculate physical quantities , approximations have to be made which , unfortunately , instead of clarifying the physics sometimes tend to obscure it .moreover , since in some of these methods , the free energy is not defined in a unique way , it becomes impossible to determine the specific role played by each source of correlations in the system .it would therefore be desirable to have a theoretical framework which retains the simplicity of the early attempts , but also accommodates correlation effects .this is the case for density functional theories .it is possible to rigorously rewrite the partition function of , say , a system of charged colloids , as a density functional @xcite , in which the contribution beyond mean field is seen to be expressible as an additive correlation correction to the free energy density , whose functional form is of course unknown and for which one has to make a reasonable ansatz .the spirit is very similar to the fundamental problem of integral equations , where one also has to make an educated guess ( namely , the closure relation ) , but in the functional case the ansatz involves a free energy density rather than a relation between two- and three - point functions .it thus relies on a different kind of intuition and thus permits complementary insight .one suggestion for such a functional correction has been made by nordholm @xcite .it relies on a debye - hckel treatment of the one component plasma ( ocp ) @xcite , in which the short - distance failure of linearization is cleverly overcome by postulating a correlation hole . since beyond a certain densitythe resulting ocp free energy density is a concave function of density , this favors the development of inhomogeneities . in the pure ocpthese are balanced by the homogeneously charged background .however , if one uses the ocp free energy density as a correlation correction to the mean field functional describing the double layer at a charged surface , one has all the charge opposite to the counterions located on that surface , rather than homogeneously distributed as a stabilizing background .the consequence is that the double layer becomes unstable and all ions collapse onto the surface , an effect which has been termed `` structuring catastrophe '' @xcite . to circumvent this instability without losing the physical transparency of a local functional, we recently proposed the debye - hckel - hole - cavity ( dhhc ) theory @xcite , in which we suggested a convex correlation functional .this was achieved by excluding the homogeneous background from a region of radius @xmath0 around the central ion during the debye charging process . for counterions with size we identified @xmath0 tentatively as the ion diameter .we then applied our theory to the screening of a charged rod by its counterions .comparisons of the ionic charge distribution obtained showed a very good agreement with the simulations for both monovalent and trivalent counterions .in this paper we test our theory for a different geometry : charged spherical colloids with point - like counterions . in general ,colloidal systems exhibit a rich phase behavior .the particles can agglomerate at high densities , generally an irreversible process , but they may also show a reversible liquid - vapor phase separation similar to the one present in simple molecular liquids . in order to prevent them from simply falling out of solution , one needs some kind of repulsion between the particles . introducing charged groups at the surface of the colloid is one way to do that . the large gain in entropy following the dissociation of a vast number of counterions into solutionstabilizes the system , because an aggregation of colloids into a small sub - volume would for reasons of global charge neutrality also require the counterions to occupy this small volume and thereby give up much entropy .of course , the final state of the system is always a balance between energy and entropy , and if electrostatic interactions are strong , they will ultimately overcome entropy and lead to aggregation of the colloids @xcite .the resulting phenomenon of `` like charge attraction '' has received much attention , but it is of course only mysterious if one forgets that the entire system is neutral .admittedly , confusion persists about whether such a phase separation could also happen within mean field theory .even though rigorous proofs exist that pb theory will not permit attraction between like charged macroions under reasonably general circumstances @xcite , and that in a cell model treatment the compressibility will be positive @xcite , it has been claimed that an expansion of the free energy of a charged colloidal suspension into zero- , one- , two- etc .body terms will contain configuration independent volume terms , which may drive a phase separation even though the pair terms are purely repulsive @xcite .since unfortunately all these derivations rely on a linearization of pb theory , which might render the findings as artifacts @xcite , the issue appears to be open yet .all these phenomena ultimately depend on the screening produced by the ionic cloud , which in turn depends on the geometry of the system . in this regard ,a charged spherical colloid differs from a charged rod in two fundamental ways : the electrostatic potential and the spatial extension . the logarithmic potential present in the case of charged cylinders leads to the phenomenon known as manning condensation @xcite .if the line charge density exceeds a critical threshold , a certain fraction will remain loosely associated with the rod , even at infinite dilution , and renormalize the rod charge .a quantitative pb treatment of this provides a unique criterion for defining the effective charge of the system , even at finite densities @xcite .the situation is different for charged spherical colloids , which lose all their counterions in the limit of infinite dilution ; thus , the colloidal charge does not get renormalized . still , on often talks about effective charges , which mimic the stronger condensation of nonlinear theory within a linearized treatment @xcite . that , however , is clearly not a physical but rather a formal renormalization , necessitated by the simplified linear treatment , and is thus a different story .another important difference between the spherical and the cylindrical symmetry lies in the spatial extend . if a charged rod is infinitely long ( as is usually assumed in theoretical treatments ) , the number of counterions at any given distance from the rod is always infinite .in contrast , for a charged spherical colloid the number of counterions at any distance is always finite , since of course there is no direction along which the colloid is infinite .hence , fluctuations of the radial charge density are more likely to be important in the spherical case .the systems we will consider here are strongly charged colloids with point - like ions of some specific valence and no added salt inside a spherical cell . since all the particles are limited to be within one cell , correlations between _ different macroions _ and between microions belonging to _ different cells _ are not present . in our treatmentwe will thus exclusively focus on questions regarding the description of a _ single double layer_. furthermore , for point - like ions the interpretation of our cutoff parameter , @xmath0 , can obviously no longer be the particle diameter .we will introduce an alternative prescription for @xmath0 , based again on local density considerations and keeping in mind that its entire purpose is to prevent the functional from becoming unstable .we also derive an approximated version of our correlation functional , namely , its zero temperature limit .it has the huge advantage that it can be calculated analytically , while still predicting ion profiles quite close to the full dhhc expression for a wide range of parameters .it also demonstrates the spirit of our stabilization correction very directly .finally , we compare our predictions for ion profiles with monte carlo simulations , in which we independently vary valence @xmath1 and plasma parameter @xmath2 , where @xmath3 is the density of surface charges and @xmath4 is the bjerrum length . it has been shown that beyond @xmath5 the force - distance curves between charged plates cease to be monotonic , and beyond @xmath6 attractions set in @xcite. these effects result from correlations between different double layers ( like , for instance , ion interlocking @xcite ) , which we can not account for , and it has in fact been shown that they can not be described within a local density functional theory with a convex correlation correction @xcite . however , for the description of a _ single _ double layer the regime of applicability of our theory is larger , even though it clearly must fail for too high coupling . the paper is organized as follows . in sec .ii the dhhc correlation functional is revisited and its zero temperature limit is introduced .it is then applied as a local correlation correction to the problem of screening of charged colloids in sec .the case of point - like ions is discussed in detail and the new expression for @xmath0 is proposed .technical details of the simulations are described in sec.iv .the results of the simulations , full theory and zero temperature limit are compared in sec .v , and we end with our conclusions sec.vi .the one component plasma consists of @xmath7 identical point - particles of valence @xmath1 and ( positive ) unit charge @xmath8 inside a volume @xmath9 with a uniform neutralizing background of charge density @xmath10 and dielectric constant @xmath11 . as a first approximationthe free energy of this system can be derived in the framework of the debye - hckel approach .then , the electrostatic potential @xmath12 created by some ion , fixed at the origin for instance , and all its surrounding ions satisfies the spherically symmetric poisson equation @xmath13 .the charge density has a contribution from the central ion , @xmath14 , a contribution from the surrounding ions which are distributed within mean field theory ! according to the boltzmann factor @xmath15 , and finally from the charged background . inserting this into the poisson equation and linearizing the exponential yields the linearized poisson - boltzmann equation @xmath16 where @xmath17 is an inverse screening length , @xmath18, @xmath19 is the bjerrum length , and @xmath20 is the inverse thermal energy . the solution of eqn .( [ eq : lpb ] ) is the well known expression @xmath21 .however , the problem with debye - hckel theory is that the condition for linearization is obviously not satisfied for small @xmath22 , where the potential is large . indeed, since all ions have the same sign of charge , this implies that the particle density becomes _ negative _ and finally diverges at the origin .this defect was overcome by the debye - hckel - hole theory @xcite , which artificially postulates a correlation hole of radius @xmath23 around the central ion into which no other ions are allowed to penetrate . in this casethe charge density is given by @xmath24 -\displaystyle\frac{\varepsilon \kappa^2}{4\pi } \psi(r ) & r > h \end{array } \right .\ .\ ] ] the hole size @xmath23 is fixed by excluding particles from a region where their coulomb energy is larger than @xmath25 , which gives @xmath26 . once the potential at the position of the central ion is known , the electrostatic contribution to the free energy density , @xmath27 , can be obtained by the debye charging process @xcite .the simple debye - hckel - hole analysis of the one - component plasma theory offers considerable insight into ionic systems and is in good agreement with monte - carlo simulations @xcite when fluctuations on the charge density are not relevant @xcite . in principle , one can attempt to include such fluctuations by applying the bulk density - functional theory in a local way .the basic idea is to obtain the density distribution via functional minimization of the free energy @xmath28 = f_{{\rm { pb}}}[n({\boldsymbol{r } } ) ] + \int\!{{\mbox{d}}}^3r \ ; f_{{\operatorname{corr}}}\big(n({\boldsymbol{r}})\big ) \ .\ ] ] the first part , the pb free energy @xmath29 = \int \ ! { { \mbox{d}}}^3r \ ; \big \ { k_{{\operatorname{b}}}t n({\boldsymbol{r } } ) \big [ \ln \big ( n({\boldsymbol{r } } ) v_{{\rm { p}}}\big ) -1 \big]\ , + \ , f_{{\operatorname{el}}}\big\ } \ , \ ] ] contains the entropy of the mobile ions , the interaction of the small ions with the macroion potential and the mean - field interaction between the counterions . here@xmath30 is the particle volume .the expression @xmath31 in eqn .( [ eq : f_ocp ] ) accounts for the correlation between the mobile ions .the ion distribution can be derived by minimization of eqn .( [ eq : f_ocp ] ) under the constraint of charge neutrality .unfortunately , this variational process does not lead to a well defined density profile if one uses @xmath27 as the correlation correction @xmath31 .the reason is that @xmath27 is a concave function beyond @xmath32 and asymptotically behaves as @xmath33 .since there is no stabilizing homogeneously charged background but rather a concentration of opposite charge on the macroion surface , this favors the development of a distribution in which all the ions sit on the surface of the macroion . the instabilities present in the dhh approach can be properly overcome by recognizing that the failure of this model is due to the too strong requirement of local charge neutrality imposed by the local density approximation : a local fluctuation leading to an increase of particle density implies a corresponding increase in background density .therefore , the fluctuation is not suppressed by an increase in repulsive coulomb interactions but quite on the contrary favored by its decrease . to circumvent the instabilities occurring at high densities , we proposed recently a simple solution in which one excludes the neutralizing background from a cavity of radius @xmath0 placed around the central ion ( for details of the derivation of the model see ref .@xcite ) . in this case, the charge density can be split in three different regions , namely @xmath34 where the hole size @xmath23 is chosen such as to yield the same screening ( i.e. , the same amount of charge within @xmath23 ) as the dhh theory , which results in @xmath35^{1/3 } \ , \ ] ] with @xmath36 . using this prescription for @xmath23, the free energy is obtained by debye - charging the fluid : @xmath37 where @xmath38 since the dhhc free energy is a convex function of density , @xmath39 can thus be used to account for correlations within a local density approximation . the fact that the integral in eqn .( [ eq : f_dhhc ] ) has to be solved numerically obstructs a direct view on how thermodynamic stability is actually restored .luckily , the crucial point can already be seen by focusing on the limit of zero temperature . in this case( [ eq : h_dhhc ] ) gives the expression @xmath41 for the correlation hole of the dhhc theory .this conveniently implies the potential to vanish outside @xmath23 . in other words ,the region @xmath42 contains the right amount of background charge to exactly neutralize the central ion , and it is appropriate to refer to this limit as `` complete screening '' .the potential in the two other regions then simplifies considerably : @xmath43 1 \ ; + \ ; { \widehat{n}}_{{\operatorname{b}}}\big(1+{\displaystyle}\frac{r^3}{2a^3}\big ) \ ; - \ ; { \displaystyle}\frac{3r}{2h}\big(1+{\widehat{n}}_{{\operatorname{b}}}\big ) & a \le r < h \end{array } \right . \ , \end{aligned}\] ] with the dimensionless scaled density @xmath44 given by @xmath45 .after the debye charging process one obtains the following closed expression for the excess free energy density : @xmath46 note that the limits @xmath47 and @xmath48 do _ not _ commute : for high densities , @xmath49 scales asymptotically like @xmath50 , i.e. , linear with density .however , in the limit @xmath47 eqn .( [ eq : f_dhhc0 ] ) becomes @xmath51 and this concave scaling with density prevents it from being used within a local density approximation .the zero temperature limit thus demonstrates in a clear way the key role played by the cavity of size @xmath0 , which excludes the uniform background from the vicinity of the central ion . before applying this strategy to various valences and ionic strengths , we need to specify the parameter @xmath0 .if the counterions have a diameter @xmath52 , no other charge should be found at a distance @xmath53 .therefore , in the spirit of the debye - hckel theory , we tentatively interpreted @xmath0 in ref .@xcite as the ion diameter .this choice has led to an excellent agreement with simulations when applied to rod - like polyelectrolytes @xcite with mono- , di- , and trivalent counterions ( and no added salt ) , but it is of course infeasible for point ions . in the following we suggest an alternative way to choose a value for @xmath0 which is independent of excluded volume arguments , and show that this choice yields a good description of our monte carlo results and trends .we already mentioned the crucial role played by @xmath0 in maintaining the free energy convex .we also have seen in our discussion of the zero - temperature case that this is achieved because @xmath0 in eqn .( [ eq : h_dhhc0 ] ) balances the length @xmath54 , which is basically the mean distance between ions .one could thus try to selfconsistently choose @xmath0 proportional to the local ion distance , but this would be unsuccessful : the balance would not work , since each density increase would shrink @xmath0 proportionally , and the collapse could not be stopped .one thus needs a length which is local and _ somehow _ related to the ion density but which does _ not _ change as the local ion density changes .this suggests to pick the average distance between ions _ as predicted by pb theory _ : @xmath55 .our density functional then quite naturally emerges as a next order correction to the mean field result .after these general considerations on the cutoff @xmath0 , let us continue with a practical remark .far away from the charged surface the ion density is always quite low , correlations are weakly developed , and the precise value of @xmath0 is immaterial .in fact , we only ever need a stabilizing cutoff close to the charged surface , where the ion density is largest .this suggests the following simplification : instead of using a cutoff function @xmath56 depending on the local pb density , we pick a constant @xmath0 from a worst - case scenario , namely , the value which it has at contact .this then finally yields the following prescription for @xmath0 : @xmath57 in fact , since the cutoff will become important in the regime of strong correlations , we could even replace the contact density @xmath58 by its limiting value @xmath59 , where @xmath3 is the density of surface charges @xcite .we then find @xmath60 where @xmath61 is the @xmath62 plasma coupling parameter @xcite .formula ( [ eq : a_lim ] ) nicely demonstrates that in this limit the cavity size , measured in the appropriate length scale @xmath63 ( see also the scaling discussion in the appendix ) , is simply another measure of the coupling strength .charged spherical colloids are common and well characterizable systems for studying many electrostatic phenomena in pure culture , and they can often serve as simplified models for more complicated systems like polyelectrolytes or proteins . solutions containing such charged structures are indeed complicated to describe due to the long - range nature of the coulombic interactions .however , as long as these long range forces are repulsive , the colloids will create large correlation holes ( `` cells '' ) around themselves which are void of other colloids . in a first approximation one can then decouple the macroion interactions and concentrate on what s going on within a single correlation hole an approach which is termed `` cell model '' @xcite .the cell picture is known to give a good approximation for many realistic systems , and most of the physics of the system is determined by the screening of the macroion by the microions inside a cell . as a test case for our theory we shall therefore consider a charged spherical colloid of radius @xmath64 containing @xmath65 charged groups , which are neutralized by point - like ions of valence @xmath1 .this macroion is embedded in the center of a spherical cell of radius @xmath66 , corresponding thus to a volume fraction @xmath67 of colloids .the thermodynamic behavior of the colloidal system is determined by the distribution of mobile ions around the macroion .this distribution is obtained by minimization of the free energy functional , eqn .( [ eq : f_ocp ] ) . for the colloidal system ,the interaction of the small ions with the macroion and the mean - field interaction between the counterions are given by @xmath68 where @xmath69 is the total electrostatic potential at position @xmath70 and @xmath71 is the potential due to the charged macroion alone .the inter - particle correlations are taken into account by employing @xmath72 .the minimization itself is accomplished by numerically solving the corresponding euler - lagrange - equation .special care had to be taken to obtain a sufficient accuracy of the rapidly varying density profiles close to the colloid surface .in the following we will concentrate on two observables .the first is the integrated fraction of ions within a radial distance @xmath22 from the colloid center , which is given by @xmath73 we will sometimes plot @xmath74 as a function of @xmath75 , which is the green function of the spherical laplacian .this will visually expand the region close to the colloid , but it also has practical advantages when estimating the amount of closely associated ions , see e.g.ref .@xcite .measuring all lengths in the full partition function of the cell model in units of @xmath18 reveals that the distribution function @xmath76 is invariant under a rescaling which keeps the number of counterions @xmath77 , the reduced colloid size @xmath78 , and the volume fraction @xmath67 constant ( see appendix ) .the same holds for pb theory , and it is also true for dhhc theory . in the latter casethis not only relies on the form of the dhhc free energy correction ( [ eq : f_dhhc ] ) , but also on our particular choice of @xmath0 .this invariance property is thus a further support for eqn .( [ eq : a2 ] ) .the second observable we look at is the osmotic pressure @xmath79 . for pb like free energy functionals with an additional density term like our @xmath31 it is given by @xcite @xmath80_{n = n(r ) } \ .\label{eq : pressure}\ ] ] for the pb case , @xmath81 , this reduces to the well known fact that the pressure is given by the boundary density @xcite .since this result actually holds _ rigorously _ for the full restricted primitive model @xcite , one could also argue that dhhc theory is an approximate way to calculate the boundary density , and then calculate the pressure from @xmath82 , i.e. , leave out the additional term @xmath83 .this would lead to a different result , reminding us that selfconsistency and consistency with other rigorous results can not generally be achieved .we will always use the internally consistent equation ( [ eq : pressure ] ) for our pressure calculations . inserting the dhhc expression ( [ eq : f_dhhc ] ) for @xmath31, we find @xmath84 \ , \end{aligned}\ ] ] where @xmath85 .in the zero temperature limit dhhc@xmath40 this simplifies considerably .a closed expression can easily be derived by combining eqns .( [ eq : f_dhhc0 ] ) and ( [ eq : pressure ] ) : @xmath86 where @xmath87 .observe that the contribution originating from the @xmath83 term is negative for all densities .the systems we study consist of a spherical macroion of radius @xmath64 and ( negative ) central charge @xmath88 .electroneutrality is ensured by the presence of @xmath77 point - like counterions of valence @xmath1 , confined inside an impermeable spherical cell of radius @xmath66 .this also fixes the colloid volume fraction to @xmath67 .no additional salt is added .the dielectric constant @xmath11 is assumed to be uniform throughout the system , such that no image forces @xcite occur .our choices for the system parameters can be found in table [ systems ] cccccddd system & @xmath65 & @xmath7 & @xmath1 & @xmath78 & & & + 1 & 100 & 100 & 1 & 2 & 2.5 & 20.77 & 0.23 + 2 & 120 & 120 & 1 & 5.477 & 1 & 0.4090 & 0.836 + 3 & 120 & 120 & 1 & 2.739 & 2 & 8.275 & 0.307 + 4 & 120 & 120 & 1 & 1.826 & 3 & 45.07 & 0.174 + 5 & 120 & 60 & 2 & 1.937 & 2 & 7.526 & 0.317 + 6 & 120 & 40 & 3 & 1.581 & 2 & 6.976 & 0.324 standard canonical mc simulations following the metropolis scheme @xcite were employed to sample the ion distributions .after an initial equilibration time of 200000 mc steps , where we attempted to move every ion once to a new position , we sampled the system for @xmath89 mc steps , producing 13002000 configurations for analysis .we will measure energies in units of @xmath25 and use the coupling length @xmath18 as our unit of length ( for monovalent ions under aqueous conditions and room temperature we would have @xmath90 , and the unit of concentration becomes @xmath91 ) . in the followingwe will present mc results for the integrated ion distribution , eqn .( [ eq : p ] ) , and for the pressure , eqns .( [ eq : p_dhhc ] ) and ( [ eq : p_dhhc_0 ] ) .for system 1 ( see tab .[ systems ] ) .the solid curve is the result of the mc simulation , while the dash - dotted curve is the prediction from pb theory .the inset shows the local density @xmath92 .the increase in the counterion condensation due to correlations is well captured by the dhhc theory ( dashed curve ) and its zero temperature limit dhhc@xmath40 ( dotted curve ) . the difference in @xmath92 between the latter two is invisible on the chosen scale , and only dhhc is shown . ] figure [ fig : z=100 ] shows the integrated charge distribution @xmath76 for system 1 from tab .[ systems ] .the solid curve is the result from the mc simulation , and it lies distinctly above the pb result ( dash - dotted curve ) , indicating a stronger condensation of ions due to correlations neglected in pb theory . most of this enhancement of ion localization close to the colloid is captured by dhhc theory ( dashed curve ) or its zero temperature limit dhhc@xmath40 .this is also evident from the local density @xmath92 , which relatively to pb is enhanced at close proximity to the colloid , while it drops below pb at the outer cell boundary . from what we have said in section [ sec : application ] this also indicates that the pressure will be lower , and this is indeed what we shall find ( see below ) .it should be noted that the 2d plasma parameter @xmath93 is already slightly beyond the point where attractions between two planes would arise @xcite .we should not expect dhhc theory to work for significantly higher plasma parameters , since it can not account for effects like attractions @xcite .however , we want to point out that here we only aim at properties of a _ single _ electrostatic double layer and not at phenomena arising from the interaction between two of them , and in fact the agreement seen in fig .[ fig : z=100 ] is very encouraging .it is also quite pleasing that the significantly simpler zero temperature limit dhhc@xmath40 from eqn .( [ eq : f_dhhc0 ] ) yields essentially the same result as the full dhhc theory . due to the scaling invariance of the partition function discussed in the appendix , a system with e.g. divalent ions and @xmath94 or trivalent ions and @xmath95 ( and properly rescaled bjerrum lengths@xmath96 ) shows exactly the same distribution function ( not shown ) . for systems 2 , 3 , and 4 from tab .[ systems ] .the line styles are the same as in figure [ fig : z=100 ] , the counterions are monovalent , and the value of the plasma - parameter @xmath97 is indicated . ] in figure [ fig : gamma_scan ] we show distribution functions @xmath76 for systems 24 from tab .[ systems ] .these have monovalent counterions and only differ in their value of the plasma parameter @xmath97 .clearly , a larger plasma - parameter leads to an increased condensation ( the curves are shifted up)an effect which naturally is already present in pb theory . however , apart from this , at a larger plasma parameter the influence of correlations becomes more important , and therefore the _ deviation _ between the pb prediction and the mc result increases for increasing @xmath97 , which is also clearly seen in fig .[ fig : gamma_scan ] . again, this effect is well captured by dhhc theory , which is always much closer to the mc data than to the pb result , even though its accuracy diminishes as @xmath97 becomes large . for systems 3 , 5 , and 6 from tab .[ systems ] .the line styles are the same as in figure [ fig : z=100 ] , the plasma parameter is @xmath98 , and the value of the counterion valence is indicated . for clarity , the pb curve is only shown for @xmath99 . ] in fig .[ fig : v_scan ] we show a `` complementary '' scan , in which we fixed the value of the plasma - parameter @xmath98 , but changed the counterion valence ( systems 3 , 5 , and 6 from tab .[ systems ] ) .maybe surprisingly , an _ increase _ in valence leads to a _ decrease _ in condensation if it happens at constant plasma - parameter and colloid charge . if we had changed @xmath1 from 1 to 2 and simultaneously replaced @xmath100 _ and _ @xmath101 , the plasma parameter would also have remained unchanged , but due to the scaling property of the partition function that would actually have been true for the whole distribution function .instead , we have reduced @xmath102 ( i.e. , a little less strongly ) , but have failed to increase @xmath65 .the net result is that condensation drops slightly .however , since the plasma parameter , which is the best indicator for the strength of correlations , remains the same , the _ deviation _ between pb theory and mc simulation are always about the same ( not shown in the figure ) . and as a consequence , the deviation between dhhc theory , which approximately accounts for correlations , and the mc simulation , which captures them all , is about the same in all three cases , and actually not very big .another strategy to check how successful our approach captures correlations is to compute the osmotic pressure . in real systemsthis pressure will depend on correlations between ions of _ different _ cells , something which neither our theory nor actually our simulation ( of a single colloid ! ) takes into account .so in the following by `` pressure '' we do not , strictly speaking , refer to the bulk pressure of a colloidal suspension at some given volume fraction , but only to the pressure exerted on the rigid wall at @xmath103 of our cell model . .] within the simulations , the pressure is given by the contact density at @xmath103 , which was obtained by fitting the mc density profile close to the cell boundary to a quadratic expression @xmath104 .an example for how the simulated densities compare to the pb approximation , our analytic dhhc approach , and its simpler zero temperature limit dhhc@xmath40 , can be found in fig .[ fig.c1 ] . [ cols="^,^,^,^,^,^,^,^ " , ] table [ pressure ] shows the predictions for the pressure in the case of point - like ions given by pb- , dhhc- , and dhhc@xmath40-theory , as well as by mc simulations . in the case of pb - theory andmc the pressure is simply the density at the outer boundary , while for the dhhc approach we employ eqn .( [ eq : p_dhhc ] ) and for dhhc@xmath40eqn .( [ eq : cspressure ] ) .these data , as well as fig .[ fig.c1 ] , demonstrate that as anticipated the simulated pressures lie below the pb prediction .this decrease in pressure is rather accurately captured by our functional @xmath105 and by its zero temperature limit , @xmath106 , with the mc result lying significantly below pb and ( in these cases ) below dhhc and above dhhc@xmath40 .the difference between the two correlation corrected approaches is consistent with the idea that entropic effects neglected in @xmath106 would push ions away from the macroions , or in other words , that the zero temperature limit implies stronger correlations than the regular dhhc theory and therefore yields even lower pressures .in this paper we showed how to apply our previously proposed local density functional approach based on a stable correlation correction to a spherical macroion confined in a spherical cell .one of the crucial parameters in this theory is the size @xmath0 of the exclusion cavity of the background charge density . for point - like ions, we suggest to associate the exclusion region with the mean distance between ions as predicted by pb theory , and for simplicity use the value present at colloidal contact . by going to the zero temperature limit we were able to derive an even simpler free energy functional @xmath107 , which is almost as good as the full dhhc theory , but much easier to handle .we also derived exact expressions for the osmotic pressure in this system .we successfully compared our predictions to simulations of the same model and compared the integrated counterion density and the osmotic pressure values for two complementary `` scans '' of the coupling strength , namely valence and plasma parameter .we demonstrated that our local density functional approach based on a stable correlation correction leads to a major improvement over the pb prediction .this work has been supported by the brazilian agency , cnpq ( conselho nacional de desenvolvimento cientfico e tecnolgico ) and the german science foundation ( dfg ) through sfb 625 , tr6 , ho-1108/11 - 1 and de@xmath108775/1 - 2 .mcb would like to thank prof .k. kremer for the hospitality during her stay in mainz .the canonical partition function @xmath109 of the colloid surrounded by its counterion is given by : @xmath110 where @xmath77 is the total number of counterions and the hamiltonian @xmath111 splits into kinetic and potential degrees of freedom . in the classical description employed herethe kinetic part @xmath112 will contribute the usual factor @xmath113 to the partition function , where @xmath114 is the thermal de broglie wavelength .the potential energy can be expressed as @xmath115 after rescaling all length by @xmath63 , i.e. introducing @xmath116 , the total partition function can be written as @xmath117 in this form it becomes evident that appropriately scaled thermal observables like the integrated charge density ( measured in units of @xmath118 ) or the pressure ( measured in units of @xmath119 ) are invariant under system changes which leave the number of counterions @xmath7 , the rescaled colloid size @xmath120 , and the volume fraction @xmath121 fixed .

<S> we study the counterion distribution around a spherical macroion and its osmotic pressure in the framework of the recently developed debye - hckel - hole - cavity ( dhhc ) theory . </S><S> this is a local density functional approach which incorporates correlations into poisson - boltzmann theory by adding a free energy correction based on the one component plasma . </S><S> we compare the predictions for ion distribution and osmotic pressure obtained by the full theory and by its zero temperature limit with monte carlo simulations . </S><S> they agree excellently for weakly developed correlations and give the correct trend for stronger ones . in all investigated cases </S><S> the dhhc theory and its computationally simpler zero temperature limit yield better results than the poisson - boltzmann theory . </S>

1604.04829

the bi - connectivity of graphs is essential for many real world applications ranging from power distribution systems , communication networks and many others .the reason for this is that such typologies provide a certain level of resistance to failures , and as a result more robust and reliable systems @xcite .these benefits are a direct consequence of the fact that such graphs have two disjoint paths connecting any two nodes in the graphs .this property is often called 2-connectivity of graphs . in literaturethere is differentiation between vertex and edge 2-connected graphs , where the two paths are vertex or edge disjoint , respectively . in practicethis means that the vertex / edge 2-connected graph stays connected if any single vertex / edge is removed from it .vertex 2-connectivity is a stronger property in the sense that every such graph is also 2-edge connected but the reverse is not necessarily true .the term bi - connected graph is used for 2-vertex connected graphs .testing if a graph is bi - connected can be done efficiently using the standard algorithm for finding articulation points of a graph based on depth first search in linear time @xcite .a similar approach has also been used for finding a 3-connected partitioning of a graph @xcite .other examples of algorithms for testing bi - connectivity of a graph exploit the fact that bi - connected graphs have an ear @xcite or chain @xcite decomposition .there are several types of mixed integer programs based on multi - commodity flow constraints that are used for exploring bi - connectivity , but such models generally have a large number of variables @xcite. there are alternative formulations containing a lower number of variables but having an exponential number of constraints @xcite . for many graph optimization problems , like the weighted vertex cover and the independent set , it is sufficient to solve the problem separately on each of the bi - connected components @xcite .the most commonly used method for partitioning graphs into bi - connected components is tarjan s algorithm , which accomplishes this task in linear time @xcite .there are several interesting variations of the original algorithm with similar computational times @xcite .the problem with these types of algorithms is that it is hard to modify them to a setting where the subgraphs need to satisfy some additional constraints .the proposed work is focused on developing a method that can partition a graph into bi - connected subgraphs with a maximal allowed size , but the general concept can be adapted to other interesting constraints . in current literaturethere are many practical problems which are modeled using the problem of partitioning graphs into connected subgraphs .some examples are applications in surveillance systems @xcite , data clustering @xcite and education @xcite .an interesting group of applications comes from the satisfactory graph partitioning problem , like finding communities within social or biological networks , defense alliances , artificial intelligence development , etc .@xcite . partitioning problems defined on supply / demand graphs have proven to be essential in modeling systems of interconnected microgrids @xcite .clustering in for many of them it would be reasonable to substitute the constraint of connectivity with bi - connectivity , producing the benefit of higher reliability of the system .this type of extension has frequently been applied to modeling real - world systems based on general graph problems including connectivity .some examples are power optimization in ad hoc wireless networks @xcite and facility layout problems @xcite .a common approach for solving such problems is to start from an approximate solution that is only connected and extend it with additional nodes to achieve bi - connectivity , like in the case of the problem of constructing a 2-connected virtual backbone in wireless networks @xcite . in case of graph problems containing connectivity constraints ,a standard approach is to grow a partial solution by adding neighboring nodes .one example is the method used for finding the minimal connected dominating set of a graph @xcite .the concept of growth has also been extended to problems where a graph is divided into connected components . in case of such problemsthe general approach is to grow several subgraphs in parallel with the constraint that no vertex can be added to more than one of them .the effectiveness of this type of method is well presented on the problem of partitioning supply / demand graphs into connected subgraphs @xcite .it is important to note that the concept of growing a solution gives a high level of flexibility of the method , in the sense of potential applications .another advantage of growth based algorithms is that they can easily be improved by their extension to metaheuristics like the ant colony optimization @xcite , the grasp algorithm @xcite and the variable neighborhood search @xcite . in this paperwe introduce a new problem of finding the maximal bi - connected partitioning of a graph with a size constraint ( mbcpg - sc ) . for the newly defined problemwe show np - hardness .because of this a growth based algorithm is developed for mbcpg - sc for finding approximate solutions .more precisely , it is solved using a heuristic procedure that exploits the fact that each bi - connected graph has an ear decomposition .the algorithm is based on a breadth first search ( bfs ) that also tracks additional properties of the nodes in the bfs tree .the additional information makes it possible to have an efficient way to `` grow '' a bi - connected subgraph by expanding it with suitable ears .the concept of growing a bi - connected graph using an ear decomposition is also used in case of constructing a fault - tolerant connected set cover problem @xcite . in this algorithm the best open ears for extending the current solution are found using the idea of shortest cycles in the original graph .a similar approach is used for the minimum 2-connected r - hop dominating set problem @xcite .the proposed method consists in growing several subgraphs in parallel , with some auxiliary corrections applied to the corresponding bfs trees . to improve the performance of the basic algorithm several methods of randomization are developed .the quality of found solutions is further enhanced using a local search procedure .the practical objective of the proposed graph problem and the corresponding solution method is its application to real word problems in the field of smartgrids , more precisely on the underlying wireless networks .such systems are well presented using unit disc graphs .the proposed graph problem is closely related to the clustering scheme for hierarchical control of wireless networks @xcite .because of this , the focus of the numerical experiments is on unit disc graphs . to be able to evaluate the method , an extensive effort has been dedicated to developing an algorithm for generating problem instances with known optimal solutions .our computational results show that the proposed method is able to find optimal solutions for small graphs . in case of large graphs ( 10.000 nodes )the method manages to find solutions within a few percent of error , in reasonable time .the paper is organized as follows . in the second sectionwe give the definition for mbcpg - sc and a proof of np- hardness .in the following section we present the method for growing a bi - connected subgraph .the third section gives details of the algorithm for parallel growth of several subgraphs and randomization .the next section describes the proposed local search mechanism . in section 5 ,we provide details of the data generation mechanism for problem instances with known optimal solutions .moreover , we discuss results of our computational experiments and provide some conclusions and ideas for future research .the problem is defined on a graph @xmath0 , where @xmath1 is the set of nodes and @xmath2 is the set of edges .we also define a set @xmath3 , whose elements will be called root nodes .the aim of the problem is to divide the graph @xmath4 into a set of subgraphs @xmath5 , where @xmath6 , satisfying the constraints given in the following text .the notation @xmath7 will be used for the set of nodes that induces subgraph @xmath8 .each of the @xmath7 must contain only one distinct root node @xmath9 .a node @xmath10 can be an element of at most one @xmath7 .the number of nodes in each of the subgraphs @xmath11 is less or equal to some constant @xmath12 .the last constraint is that each of the subgraphs @xmath8 is bi - connected .the goal of the problem is to find @xmath13 containing the maximal number of nodes in all the subgraphs .more formally , we wish to maximize the sum : @xmath14 + where each of the subgraphs @xmath7 satisfies @xmath15 in the definition we use notation @xmath16 to indicate the number of nodes in a graph .an illustration of a problem instance and solution for the mbcpg - sc is given in figure [ fig : problemcolor ] .it is important to note that the mbcpg - sc does not produce a partitioning in the strict sense , since some nodes may not be included in any @xmath7 . .the circle nodes represent root nodes .different shades of gray are used for different subgraphs.,scaledwidth=100.0% ] the mbcpg - sc is a hard optimization problem in the sense that it is np hard . we give a proof by reduction to a restriction of the problem of maximal partition of supply / demand graphs ( mpgsd ) @xcite .it has been shown that the mpgsd is np - hard in the case of a graph containing only one supply node and having a star structure @xcite . for a more comprehensive presentationwe first give the definition of the mpgsd for star graphs , which is a simplified version of the original problem .it is defined for an undirected star graph @xmath17 with a set of nodes @xmath1 and a set of edges @xmath2 .the center node of the graph @xmath18 is called a supply node and it has a corresponding supply value @xmath19 .all the other nodes @xmath20 are called demand nodes , and they have a corresponding demand value which is a positive integer @xmath21 .the aim is to find a set of nodes @xmath22 which satisfies the constraint that the supply @xmath19 must be greater or equal to the total demand of nodes in @xmath23 .the goal is to maximize the fulfillment of demands . in the following textwe prove that mbcpg - sc is np hard by reducing the mpgsd for star graphs to it .let mpgsd be defined for a star graph @xmath24 having a central supply node @xmath18 with a supply value @xmath19 connected to @xmath25 demand nodes @xmath26 having demand values @xmath27 .let us convert this problem to mbcpg - sc in the following way .the parameter @xmath12 of mbcpg - sc will be set to @xmath28 and there will be only one root node @xmath29 .let us convert the star graph @xmath24 to @xmath30 used in mbcpg - sc .first , we include the root node @xmath29 in @xmath31 .the node @xmath29 is connected to two nodes @xmath18 ( corresponding to the supply node in @xmath24 ) and node @xmath32 which will be used for some extra edges .let us remember that the edge set @xmath33 of @xmath24 consists of edges @xmath34 .for each demand node @xmath26 we will add a path @xmath35 .next , we add an additional edge @xmath36 for each demand node @xmath26 .an illustration of this conversion is given in figure [ fig : convertnp ] .represents the root node , @xmath18 the supply node from mpgsd and @xmath32 the auxiliary node .dashed lines are used for the auxiliary edges .the size limit mbcpg - sc is @xmath37.,scaledwidth=100.0% ] let us make a few observations on the solution of a mbcpg - sc on the convert graph @xmath24 .first , any solution must contain nodes @xmath32 , @xmath18 since there must be at least two disjoint paths from @xmath29 to any other node .second if any node @xmath38 is included in the solution @xmath39 all the nodes @xmath40 must be included in it .let us assume the opposite that node @xmath41 is not included in the solution and that the nodes @xmath42 for @xmath43 are .in such a case the removal of node @xmath32 from @xmath39 results in all the nodes @xmath44 where @xmath45 becoming disconnected from the graph induced by @xmath39 . in the same way , the removal of node @xmath18 from @xmath39 results in all the nodes @xmath44 , where @xmath46 , becoming disconnected from the graph induced by @xmath39 . as a consequence any @xmath47 is connected to @xmath29 by two disjoint paths ( containing only elements in @xmath39 ) @xmath48 , @xmath49 , @xmath50 , @xmath51 and by path @xmath52,@xmath53 , @xmath50 , @xmath54 . note that nodes @xmath29 , @xmath18 , @xmath55 , ... , @xmath56 , @xmath32 form a cycle and as a consequence two disjoint paths exist among any 2 of them .we will prove that mpgsd on @xmath24 can be reduced to mbcpg - sc on @xmath57 by showing that any solution @xmath58 of mpgsd has a corresponding solution @xmath39 of the mbcpg - sc such that @xmath59 and and vice versa .let us first prove the direction mpgsd to mbcpg - sc , by constructing the solution @xmath39 from @xmath58 .@xmath39 will consist of nodes @xmath29 , @xmath32 , @xmath18 and all nodes @xmath60 were @xmath61 and @xmath62 .first , by construction we have @xmath63 .the only additional case that needs to be considered to prove bi - connectivity of @xmath39 is for two nodes @xmath64 @xmath65 such that @xmath66 . to be more precise we need to prove that there are two disjoint paths connecting them . from the previous observationwe have that each of these two nodes is connected to nodes @xmath32 , @xmath18 by two disjoint paths .so , nodes @xmath64 @xmath65 are connected with disjoint paths @xmath67 and @xmath68 .the construction of the solution @xmath58 of mpgsd from a solution of @xmath39 of the mbcpg - sc is trivial and is based on the construction of graph @xmath57 and the fact , shown in the previously made observation , that any solution @xmath39 consists of @xmath29 , @xmath32 , @xmath18 and all nodes @xmath69 where @xmath70 for some nodes @xmath71 .the main motivation for defining the mbcpg - sc is its application to systems of interconnected microgrids @xcite .it has been shown that optimization of self - adequacy of individual microgrids in such systems can be well modeled using the mpgsd .the proposed problem can also be understood as a partitioning of a power supply network in which all supply nodes ( elements of @xmath72 ) have a value @xmath12 and demand nodes ( elements of @xmath73 ) have value 1 .the problem with using the mpgsd for this type of systems is that it does not address the problem of failure resistance . in mbcpg - scwe exploit the fact that by strengthening the constraint of connectivity to bi - connectivity , compared to mpgsd , we are able to have a model that produces more robust subsystems . another potential application of the proposed problem is on the hierarchical clustering for wireless networks . in the work of @xcite , a network is divided into clusters that can be hierarchically controlled .the basic properties of a cluster are that it has a single control node , the number of nodes it can contain is limited , and the corresponding subgraph needs to be connected .it is natural to extend this formulation with an additional constraint of bi - connectivity of subgraphs to enhance reliability .since the proposed algorithm is based on the property that a bi - connected graph has an open ear decomposition we start with its definition .an open ear decomposition of a graph @xmath4 is defined as a series of paths @xmath74 called ears .the term path is used for an ordered sequence ( @xmath75 , @xmath76 , , @xmath77 ) such that all edges @xmath78 . in the following textwe will use the notation @xmath79 for the set of nodes in @xmath80 .the notation @xmath81 is used for a sequence of nodes in ear @xmath82 .all the ears @xmath83 , @xmath84 in the decomposition satisfy @xmath85 if @xmath86 .the exception is the first ear @xmath87 which is a cycle @xmath88 . for @xmath89, we have that for each of the two terminating nodes @xmath90 exists @xmath91 such that @xmath92 , @xmath93 where it is not necessarily @xmath94 .we wish to point out that @xmath95 , so each @xmath96 is an open ear . except for such terminating nodes ,there is no @xmath97 if @xmath98 .finally , for each @xmath99 there exists at least one @xmath100 such that @xmath101 .if such a decomposition exists for graph @xmath4 then @xmath4 is bi - connected .an illustration of an open ear decomposition of a graph is given in figure [ fig : openeardecomposition ] .in the proposed algorithm the idea is to grow a bi - connected subgraph @xmath23 , starting from an initial cycle and extending it with adequate ears .more formally , we will be iteratively generating a sequence of subgraphs @xmath102 , where @xmath103 and @xmath104 is an open ear for @xmath105 .it is evident that if a subgraph is generated in this way it will always have an open ear decomposition .although it is possible to develop such an algorithm using a depth first search ( dfs ) the use of bfs is more suitable since it gives us more control of the size @xmath106 of the ear that will be added . in the subsequent text we will be using the following notation . * @xmath107 the set of adjacent / neighboring nodes to node @xmath108 in graph @xmath4 . * @xmath109/@xmath110 are the set of nodes / edges in the bfs tree .* @xmath111 is the set of children of node @xmath112 in the bfs tree .* @xmath113 is the parent node of @xmath112 in the bfs tree .* @xmath114 is the set of all descendants of node @xmath112 in the bfs tree . *@xmath115 $ ] is defined for nodes @xmath116 .it represents the path @xmath117 , @xmath118 , @xmath119 , @xmath50 , @xmath120 , @xmath121 connecting @xmath108 , @xmath122 such that all @xmath123 .brackets `` @xmath124''/``@xmath125 '' are used to indicate if @xmath108 , @xmath122 are included / excluded in the path .the notation @xmath126 $ ] is used for the corresponding set of nodes .* @xmath127 , is node @xmath122 which is the first ancestor of @xmath108 , in the bfs tree , such that @xmath128 . in case @xmath129, then @xmath130 .* @xmath131 , is node @xmath122 which is the first ancestor of @xmath108 , in the bfs tree , such that @xmath132 . in case @xmath133, then @xmath130 .* @xmath134 is the length of the path ( number of nodes ) , @xmath135 .note , that it does not include node @xmath127 . in case @xmath136, we have @xmath137 .it is well known that bfs can be used to find cycles which we exploit in the proposed method .let us assume that we start the bfs from some initial node @xmath29 . as we expand the bfs tree , or in other words , new nodes are visited , the first time we encounter a back - edge @xmath138 an initial cycle @xmath23 is found .more precisely , @xmath23 is acquired by connecting three segments : path from @xmath29 to @xmath108 , the back edge @xmath139 and the path from @xmath122 to @xmath29 . in the proposed notation @xmath140 , p_b[v , r))$ ] and it has all its nodes in the bfs tree . in a similar way we can find new open ears .let us assume the bfs tree is further expanded and a new back - edge @xmath141 has been found , and that at least one of the nodes @xmath142 is not in @xmath23 .it is obvious that if the following is satisfied , @xmath143 then the sequence @xmath144 , \bar{p}_b[s , root(t)])\ ] ] will produce a new open ear connected to @xmath23 . as a consequence @xmath145will also be a bi - connected subgraph .we will use the notation @xmath146 for the set of nodes corresponding to @xmath147 .the same procedure can be used to further expand @xmath23 with new open ears .the second constraint that exists in the mbcpg - sc is that @xmath148 . while growing @xmath23 , it can easily be maintained if we only allow adding ear @xmath146 if the following equation is satisfied latexmath:[\[\label{conlength } explore node @xmath108 having the lowest value of @xmath134 the length @xmath150 of the newly found open ear will be among the shorter ones ( except in rare cases ) .this is due to the method of construction in which newly found ears will always have one terminating node equivalent to the currently visited node in the bfs .if @xmath151 has the largest value of all nodes in the bfs tree which are not in @xmath23 then the maximal difference between @xmath150 and the shortest ear will be @xmath151 . in case of adapting the bfs in this way some additional work will be necessary to update the values of @xmath134 and @xmath152 as new ears are added to @xmath23 .it is important to note that after such updates it is possible that several back edges @xmath153 that have been previously tested may now satisfy the constraints given in eqs . , .the algorithm for growing a bi - connected subgraph based of the idea presented in the previous section will start the bfs from some root node @xmath29 .as the first ear in the decomposition @xmath154 differs from the rest as it is a cycle , some special initialization needs to be done for @xmath29 and its neighbors @xmath155 .for all these nodes @xmath108 we will initially set @xmath156 . in the adaptation of bfs for growing a bi - connected graphthe distance in the bfs tree will have a different meaning . instead of following the distance of a node @xmath108 from the root node @xmath29 we will track the distance from @xmath108 to the already generated bi - connected subgraph @xmath157 .at the initial step we will consider @xmath158 .let us assume that we have found two nodes @xmath108 , @xmath122 that satisfy the constraints given in eqs ., , then the first ear @xmath154 can be constructed as @xmath159 as previously stated in case a new ear @xmath79 is added to @xmath23 the values of @xmath127 , @xmath134 will need to be updated for some elements of the bfs tree .it will be necessary to update these values for all @xmath160 in the following way .@xmath161 @xmath162 it is important to note that the proposed correction for functions @xmath134 will produce approximations to the exact distance @xmath163 . for the function@xmath164 we have that @xmath165 since it is possible to have an alternative path to @xmath23 using some back edges which is shorter .a consequence of this is that the constraint defined in eq . will never give false positives if function @xmath164 is used instead of @xmath166 . by using this approximate approach itis possible to have a simpler and less computationally expensive implementation . in the standard bfsthere is no change in the distance for visited nodes and no node is re - visited . in the proposed adaptation of bfssuch changes can occur and some revisits are necessary .it is possible to use a heap or similar structure instead of a queue to always test the node @xmath108 with the lowest value @xmath163 . in practicethis is not necessary especially since we are only using an approximation to @xmath134 . on the other handthe need for retesting some nodes further increases the complexity of implementation .both of these issues are addressed simultaneously using the following approach .first , nodes will be re - added to the queue as a new ear is added to @xmath23 and their re - evaluation is needed .since it is possible for the same node to be added multiple times to the queue due to the addition of multiple ears , an additional value will be used to track if an evaluation is needed .the algorithm for growing a bi - connected subgraph is better understood by observing algorithm [ alg : grow ] .the proposed algorithm starts with a standard bfs initialization of the distance , parents and descendants for all the nodes with the additional property of the need for evaluation .initially all the values of @xmath167 will be set to @xmath168 .an auxiliary structure is used to store all the properties of individual nodes , which can be accessed and updated using the node i d .next , we initialize the root @xmath29 and all its neighbors @xmath155 as previously described and all nodes in @xmath155 are added to the queue @xmath169 . the main loop is executed for each node @xmath170 in @xmath169 until @xmath171 . for each such nodewe first check if an evaluation is needed and if so all its neighbors @xmath172 are evaluated . for each @xmath173we check if @xmath174 is a back - edge . in caseit is not we add @xmath108 to the @xmath169 as in the bfs , and we set @xmath175 . in case @xmath108 is a back - edge we check if @xmath176 is an open ear connected to @xmath23 .if this is true the subgraph @xmath23 is extended with @xmath177 and necessary updates are performed using procedure @xmath178 .after all the elements of @xmath172 are visited the evaluation of node @xmath170 is complete and we set @xmath179 .the update procedure for a newly found open ear , @xmath178 , is used to change the state of @xmath169 and nodes based on the @xmath177 .the details of the procedure are given in algorithm [ alg : updateaddear ] .in it , for all @xmath133 the distance is set to @xmath180 .each node @xmath108 now becomes a root of a new potential ear , so we set @xmath181 .for each node @xmath133 we wish to update the branch of the bfs tree whose root is @xmath108 .this is done using a recursive procedure @xmath182 . in itwe go through all the bfs descendants @xmath122 of @xmath108 that are not already in @xmath23 and set the @xmath183 and @xmath184 . by doing so the node properties are set to the values that correspond to the new state of @xmath23 . for each such node re - eavaluationis needed to check if new open ears have been created so we set @xmath185 .the addition to the queue @xmath169 is done after the recursive call @xmath186 for updating the descendants .this order is important since nodes will be added to the queue in reverse order of their distance and as a consequence shorter potential ears are checked first . by doing so@xmath23 is more gradually grown .when solving the mbcpg - sc multiple subgraphs @xmath187 should be grown together .the idea of the algorithm is to randomize this process .the randomization is done on two levels .first , the growth of individual subgraphs @xmath8 should be randomized .this can simply be done by adding an additional parameter @xmath188 $ ] and a corresponding random variable @xmath189 $ ] which will be used to decide if a valid open ear is added to @xmath23 or not .it is important to note that by not adding an open ear @xmath190 at iteration @xmath191 does not necessarily exclude the nodes inside @xmath190 from the corresponding subgraph @xmath157 .this is due to the fact that they can be a part of some ear @xmath192 that will be added to @xmath23 at a later iteration .it is evident that the growth of subgraphs @xmath193 is interdependent since a node @xmath108 can only be an element of a unique @xmath7 . on the other hand ,the growth of some @xmath7 will also effect the direction of expansion of the bfs tree of neighboring subgraphs . because of this the second type of randomization should effect the speed and order in which all the @xmath7 will be grown .the proposed method can be better understood through the pseudocode given in algorithm [ alg : generatesolution ] . in the proposed algorithmseparate bfs trees and corresponding auxiliary structures exist for each of the @xmath7 .the auxiliary structure for tracking node properties is extended with a property @xmath194 for nodes , which is used to indicate if a node has been added to some of the other subgraphs .initially this value is set to @xmath168 for all nodes in all subgraphs .the first step , is setting @xmath195 for all the @xmath196 where @xmath98 . herewe use the notation @xmath197 for node @xmath198 in the auxiliary structure of subgraph @xmath199 .next , the @xmath200 is initialized for each subgraph @xmath8 using the corresponding @xmath201 as described in the previous section .the main loop is repeated until no subgraph can be further expanded . at each iteration a random subgraph @xmath8 is selected for expansion of a maximal allowed size @xmath202 .@xmath202 is a random variable from some range @xmath203 $ ] .the goal of the inner loop is to extend @xmath7 with at least @xmath202 nodes , or until it is not possible to extend it . in this loopan adaptation of the procedure presented in the previous section is used in the form of function @xmath204 which only adds a single ear @xmath79 to @xmath7 .this function does not consider nodes having @xmath205 .after an ear @xmath79 is added to @xmath7 , the bfs tree and corresponding structures need to be updated for all @xmath196 where @xmath98 using procedure @xmath206 .the update procedure needs to perform several tasks .first , all the nodes @xmath133 need to be made unavailable for the growth of @xmath196 , where @xmath207 , which can be done by setting @xmath208 in the corresponding auxiliary structures .secondly , all the nodes in @xmath209 which have been cutoff from the bfs tree ( for some @xmath196 , @xmath98 ) , by the removal of @xmath108 , need to be re - initialized so they can potentially be re - added to the bfs tree .this is done by setting @xmath210 and @xmath185 .although there is a potential that a node @xmath209 may be reached by continuing the growth of the bfs tree , there is no guarantee for this .the question is which nodes need to be re - evaluated to make this possible .it is obvious that if a back - edge @xmath211 existed for the bfs tree that node @xmath212 can be reconnected to @xmath122 . on the other handif such a back edge existed and @xmath213 then the corresponding open ear would have been already added to @xmath23 .the only potentially disregarded back edges are of the type @xmath214 . because of this the nodes in @xmath215 should be re - evaluated if @xmath216 , since they can establish a connection with @xmath122 .the details of the update procedure for deleting a node are given in algorithm [ alg : deletenode ] .the algorithm in the previous section gives us a method for generating a single solution for mbcpg - sc .a simple way of finding higher quality solutions is to perform multiple runs with different random seeds and selecting the best one .the problem with this approach is that no experience is gained from previously generated solutions and as a consequence a very high number of them needs to be generated to get good quality ones .an alternative approach is to develop a local search procedure which improves sections of already generated solutions .the basic idea of the proposed local search is to regrow only a subsection @xmath217 of the previously best found solution . ideally , we wish to do this only for subsections for which an improvement is possible .the growth procedure presented in the previous section can easily be adapted to such a setting .let us define the set on non - located nodes as @xmath218 it is obvious that @xmath219 should contain at least one @xmath8 such that @xmath220 .the other requirement is that @xmath219 has a potential to expand to some @xmath221 . for simplicitylet us assume that there is only one non - located node @xmath221 , and there is only one @xmath222 such that @xmath223 .since @xmath8 must be bi - connected there must be a @xmath224 having only nodes in @xmath219 .we can define the set of neighboring nodes of a subgraph in the following way @xmath225 it is evident that for an appropriate @xmath226 to exist , @xmath108 must satisfy @xmath227 for some @xmath228 . using this logic , we can specify @xmath219 containing @xmath229 subgraphs based on this necessary condition .select one @xmath230 having @xmath223 and @xmath231 such that it @xmath232 .select @xmath233 random subgraphs from @xmath234 .we will call this method for selecting the elements of @xmath219 growth random ( @xmath235 ) .this method for specifying @xmath219 satisfies the necessary condition but further constraints can be added to make the local search more efficient .let us observe a graph @xmath236 induced by @xmath13 in the following way .each node in @xmath237 corresponds to a subgraph @xmath230 .an edge @xmath238 is in @xmath33 if there exist nodes @xmath239 , @xmath240 such that @xmath241 . if we are re - growing only two subgraphs @xmath242 there is a potential that nodes may be exchanged between them if @xmath243 , and more generally @xmath244 will influence each other s growth .transitively , the influence will extend to all subgraphs in a connected subgraph @xmath245 . and @xmath57 are @xmath246 and @xmath247 , respectively ., scaledwidth=85.0% ] an exchange can occur between subgraphs @xmath242 in the same setting even if @xmath248 through nodes in @xmath249 .let us assume that for some @xmath239 , @xmath240 there is a connecting @xmath250 which only contains nodes in @xmath249 .in such a case there is a chance that node @xmath122 can be added to @xmath7 using this path .we give an illustration in figure [ fig : nonlocpath ] .let us define @xmath251 as a set of edges , where @xmath252 only if a connecting path exists containing only non - located nodes .let us now define a new graph @xmath253 using the following edge set @xmath254 now , we can say that for the same @xmath219 an exchange between subgraphs @xmath8 , @xmath199 can only occur if and only if @xmath255 and more generally @xmath244 will influence each other s growth .transitively the same applies to any connected subgraph @xmath256 .a consequence of this is that regrowing should only be considered for a connected ( in @xmath57 ) subsections @xmath219 , since each isolated section can be treated separately . using graph @xmath57we can define a method for generating a set @xmath257 suitable for regrowth in the following way .first select an initial random node @xmath258 such that @xmath259 .note as a reminder that @xmath260 represents a subgraph in @xmath13 .iteratively , we add a random node @xmath261 such that it is connected ( in @xmath57 ) to at least one node @xmath262 .when @xmath263 , we check if at least one node @xmath262 satisfies @xmath264 .if this is true @xmath219 is accepted as the set that will be regrown . in casethis is not true a new set @xmath219 is generated in the same way .if after @xmath265 attempts no such @xmath219 is generated @xmath229 is increased by one .this process is repeated until @xmath219 satisfying the necessary constraints is generated .we will call this method for selecting the elements of @xmath219 growth neighbor ( @xmath266 ) .the method based on the local search for the mbcpg - ls is given in algorithm [ alg : localsearch ] .the methods starts with generating an initial solution using procedure @xmath267 which is set as the @xmath268 .for this solution we generate the graph @xmath57 as described in the previous text . in the main loop we first select a random subgraph @xmath8 such that @xmath223 and a random number @xmath269 $ ] which indicates how many subgraphs will be regrown .next , we generate a random subgraph @xmath256 based on the node @xmath7 containing @xmath229 nodes using one of the two methods @xmath235 or @xmath266 .a new solution @xmath270 is acquired by regrowing @xmath219 using procedure @xmath271 .this procedure is the same as @xmath267 with the following changes .first , for all nodes @xmath272 we set @xmath208 .next , only subgraphs in @xmath219 are grown and considered in the update procedure .the rest of the subgraphs are left unchanged .the next step in the main loop consists in testing if @xmath273 . in casethis is true the best found solution is updated and the graph @xmath57 is recalculated .note that we have allowed the update of the best solution even if it has the same quality as @xmath13 to diversify the search .the main loop is repeated until a maximal number of iterations or no further improvement can be achieved .in this section we present the results of the computational experiments used to evaluate the performance of the proposed method .we have compared the basic randomized growth algorithm with its extension using the local search procedure with and without exploiting the subgraph neighborhoods .all the algorithms have been implemented in c # using microsoft visual studio 2015 .the source code and the execution files have been made available at @xcite .the calculations have been done on a machine with intel(r ) core(tm ) i7 - 2630 qm cpu 2.00 ghz , 4 gb of ddr3 - 1333 ram , running on microsoft windows 7 home premium 64-bit . in the evaluation of the proposed algorithmour focus is on unit disc graphs due to there close relation with wireless networks . as stated above the proposed problemcan be used to find clustering schemes for hierarchical control of such systems with enhanced fault tolerance . due to the fact that this is a newly defined problemthere are no standard benchmark instances available for comparison it was necessary to also generate problem instances with known optimal solutions . in the generationprocedure our goal is to generate graphs that are `` as random as possible '' . to be more precise we attempt to have the node positions close to the uniform probability distribution and to have the number of edges that are adjacent to each node relatively balanced . to achieve this we use the following procedure .let us say that we are generating a problem instance having @xmath25 root nodes and @xmath12 is the maximal allowed number of nodes in a subgraph , having a solution in which @xmath13 contains all the nodes . for problem instances of this form we generate unit disc graphs inside of a box @xmath274 with dimensions @xmath275 , @xmath275 .the value for the length for establishing edges is @xmath276 .@xmath277 is an additional parameter used to vary the density of the graph used in specifying different data sets .the first step is generating @xmath25 random bi - connected unit disc graphs having @xmath12 nodes .as our aim is to have the nodes of graph @xmath4 spread over the entire area of the box @xmath274 having an area equal to @xmath275 .we want to have each of the subgraphs @xmath278 covering an area of close to @xmath279 . to achieve this for each such graphwe first define a box with dimensions @xmath280 , @xmath281 where @xmath72 is randomly selected from the interval @xmath282 used to vary the shape of @xmath278 . for this boxa random bi - connected graph @xmath278 is generated .this is done by repeatedly generating @xmath283 random points and checking if the resulting unit dist graph contains a bi - connected subgraph @xmath278 containing @xmath12 nodes . to increase the probability of having nodes of the graph covering an area close to @xmath279 ,four nodes have been forced to random positions at each edge of the box . to achieve thisa high number of graphs needs to be generated , especially in the case of a high value of @xmath12 .this is the reason for the additional points , specified using parameter @xmath284 , in the box , since it greatly reduces the number of graphs that are generated . out of all the nodes in @xmath278one would be randomly selected as the @xmath201 .the next step in generating a problem instance with a known solution is combining graphs @xmath278 .let us remember that by the method of construction for each subgraph @xmath278 contains @xmath12 nodes having some values of coordinates @xmath285 , as a consequence we can easily translate @xmath278 by a vector @xmath286 by changing all node positions to @xmath287 .the basic idea of combining the graphs @xmath278 is to randomly position ( translate ) them in a way that all the nodes fit in a box having dimensions @xmath275 , @xmath275 .as previously stated our goal is to have a relatively balanced number of edges for all the nodes .secondly , we want to distribute the nodes over the entire box @xmath274 .this is achieved through the following iterative procedure .initially , the graph @xmath4 is set to @xmath288 . at each iteration a new graph @xmath278 is set at a random position and added to @xmath4 , but some constraints must be satisfied .first , all the points in @xmath289 must fit into a box having dimensions @xmath275,@xmath275 .the number of new edges @xmath290 connecting @xmath4 and @xmath278 , has to satisfy @xmath291 . to achieve this multiple random positions have to be tested for each subgraph @xmath278 .the parameter @xmath292 is used to bound the allowed number of new edges which is related to the number of edges @xmath293 in the graph as @xmath294 .empirical tests have shown that values @xmath295 produce satisfactory results .for each graph @xmath278 a @xmath296 positions are tested and the one satisfying the constraints and having the lowest number of @xmath297 would be chosen for adding to @xmath4 .the last step is randomly enumerating all the nodes in @xmath4 . to have an extensive evaluation of the proposed algorithm , we have generated a wide range of problem instances for different numbers of root nodes @xmath25 ( 5 - 100 ) and maximal allowed number of nodes in a subgraph @xmath12 ( 5 - 100 ) .for each pair @xmath25 , @xmath12 40 problem instances have been generated using the algorithm presented in the previous subsection with different values for the seed of the random number generator .we have generated two problem sets varying in the level of connectivity by setting @xmath298 .for each of the problem instances a single run of the growth based algorithms is performed .this is done for the two local search methods based on @xmath235 and @xmath266 for generating the set of subgraphs @xmath219 which will be regrown .the same set of parameters for specifying the algorithm is used for all the problem sizes .the value of the parameter for accepting an ear is @xmath299 in the growth algorithm .the parameter specifying the maximal expansion of a subgraph at one step is @xmath300 .the value @xmath301 was used for the the upper bound on the value @xmath229 which is used for generating the size @xmath302 of the set of subgraphs that will be regrown .these values have been chosen empirically after a wide range of values have been tested in computational experiments . in all of the computational experimentsthe algorithm would execute until 10.000 iterations ( generated solutions ) have been reached or more than 2000 iterations have been completed without an improvement to the best found solution .the results of the computational experiments focus on the quality of found solutions and the computational cost which can be seen in tables [ table : alpha2 ] , [ table : alpha15 ] .the evaluation of solution quality is done using the average normalized error of the found solutions compared to the known optimal ones , for each of the used methods .more precisely , for each of the 40 test instances , for each pair @xmath303 , the normalized error is calculated by @xmath304 , and we show the average values . to have a better comprehension of the performance we have also included the standard deviation , maximal errors and the number of found optimal solutions ( hits ) .the computational cost is analyzed through the average execution time needed to find the best solution . to have a better understanding of these execution times ,we have also included the average number of iterations ( number of generated solutions ) to find the best solution and the corresponding standard deviation in tables [ table : alpha2 ] , [ table : alpha15 ] .430ptx*10c roots x m & & & & & + & @xmath72 & @xmath265 & @xmath72 & @xmath265 & @xmath72 & @xmath265 & @xmath72 & @xmath265 & @xmath72 & @xmath265 + 5 x 5 & 0.30(1.38 ) & 0.30(1.38 ) & 8.00 & 8.00 & 38 & 38 & 122(47 ) & 63(41 ) & 12 & 8 + 5 x 10 & 0.15(0.69 ) & 0.00(0.00 ) & 4.00 & 0.00 & 38 & 40 & 357(224 ) & 269(228 ) & 40 & 31 + 5 x 25 & 0.78(1.32 ) & 0.96(1.61 ) & 5.60 & 7.20 & 24 & 23 & 884(435 ) & 644(170 ) & 236 & 185 + 5 x 50 & 1.12(1.46 ) & 0.90(1.41 ) & 8.00 & 8.00 & 11 & 14 & 1428(1160 ) & 1455(364 ) & 851 & 940 + 5 x 100 & 1.40(1.39 ) & 1.38(1.64 ) & 5.00 & 6.00 & 5 & 7 & 1707(851 ) & 1738(215 ) & 2472 & 2547 + 10 x 5 & 0.50(1.40 ) & 0.25(0.80 ) & 6.00 & 4.00 & 35 & 36 & 393(903 ) & 359(507 ) & 34 & 33 + 10 x 10 & 0.43(0.80 ) & 0.28(0.63 ) & 4.00 & 3.00 & 28 & 32 & 1034(928 ) & 848(775 ) & 162 & 162 + 10 x 25 & 2.78(2.03 ) & 2.30(2.26 ) & 9.60 & 9.60 & 1 & 5 & 1810(1031 ) & 1569(594 ) & 748 & 680 + 10 x 50 & 2.04(1.34 ) & 1.51(1.19 ) & 6.80 & 5.00 & 0 & 1 & 2775(441 ) & 2946(1389 ) & 2497 & 2730 + 10 x 100 & 1.92(1.30 ) & 1.70(1.35 ) & 6.30 & 6.30 & 0 & 0 & 3470(555 ) & 3479(1997 ) & 7883 & 7870 + 25 x 5 & 3.36(3.13 ) & 1.98(2.71 ) & 9.60 & 8.80 & 12 & 21 & 2113(3057 ) & 1635(686 ) & 246 & 207 + 25 x 10 & 2.19(1.42 ) & 1.41(1.18 ) & 5.20 & 4.00 & 2 & 7 & 2268(2228 ) & 1731(890 ) & 431 & 399 + 25 x 25 & 3.35(1.31 ) & 2.04(0.87 ) & 6.88 & 4.16 & 0 & 0 & 4475(3739 ) & 4494(1885 ) & 2221 & 2248 + 25 x 50 & 2.91(0.98 ) & 2.41(0.95 ) & 4.88 & 4.56 & 0 & 0 & 6225(2062 ) & 4480(1170 ) & 6842 & 4985 + 25 x 100 & 2.59(0.93 ) & 2.32(1.04 ) & 6.00 & 5.16 & 0 & 0 & 7220(2221 ) & 6105(1092 ) & 22812 & 18533 + 50 x 5 & 6.47(2.08 ) & 3.08(1.97 ) & 10.40 & 8.00 & 0 & 2 & 3669(3037 ) & 4020(1764 ) & 746 & 728 + 50 x 10 & 2.93(1.16 ) & 1.37(0.82 ) & 6.00 & 3.20 & 0 & 1 & 5385(2962 ) & 3548(248 ) & 1681 & 1118 + 50 x 25 & 3.88(0.86 ) & 2.50(0.74 ) & 5.68 & 4.16 & 0 & 0 & 7779(1800 ) & 7322(1683 ) & 5911 & 4801 + 50 x 50 & 3.91(0.76 ) & 2.79(0.70 ) & 5.68 & 4.40 & 0 & 0 & 8822(813 ) & 7966(1921 ) & 15347 & 12788 + 50 x 100 & 3.51(0.79 ) & 2.47(0.59 ) & 5.26 & 4.40 & 0 & 0 & 9295(228 ) & 8995(916 ) & 42597 & 38596 + 100 x 5 & 8.87(1.53 ) & 4.87(1.19 ) & 13.00 & 7.40 & 0 & 0 & 5333(2432 ) & 5829(832 ) & 2681 & 2256 + 100 x 10 & 4.16(0.75 ) & 1.92(0.74 ) & 6.00 & 3.40 & 0 & 0 & 8100(1171 ) & 6944(2128 ) & 6452 & 3875 + 100 x 25 & 5.51(0.88 ) & 2.99(0.47 ) & 8.00 & 4.04 & 0 & 0 & 9449(306 ) & 9127(475 ) & 17044 & 11720 + 100 x 50 & 5.44(0.70 ) & 3.26(0.44 ) & 6.94 & 4.82 & 0 & 0 & 9710(240 ) & 9640(212 ) & 39697 & 29134 + 100 x 100 & 5.00(0.82 ) & 3.15(0.49 ) & 6.84 & 4.31 & 0 & 0 & 9844(168 ) & 9731(255 ) & 100213 & 78981 + 430ptx*10c roots x m & & & & & + & @xmath72 & @xmath265 & @xmath72 & @xmath265 & @xmath72 & @xmath265 & @xmath72 & @xmath265 & @xmath72 & @xmath265 + 5 x 5 & 0.10(0.62 ) & 0.10(0.62 ) & 4.00 & 4.00 & 39 & 39 & 158(150 ) & 153(124 ) & 10 & 10 + 5 x 10 & 0.00(0.00 ) & 0.10(0.44 ) & 0.00 & 2.00 & 40 & 38 & 254(147 ) & 211(180 ) & 27 & 28 + 5 x 25 & 0.38(0.69 ) & 0.18(0.42 ) & 2.40 & 1.60 & 29 & 33 & 511(273 ) & 526(480 ) & 137 & 153 + 5 x 50 & 0.05(0.13 ) & 0.06(0.17 ) & 0.40 & 0.80 & 35 & 35 & 911(252 ) & 542(456 ) & 536 & 341 + 5 x 100 & 0.17(0.31 ) & 0.16(0.43 ) & 1.60 & 2.60 & 23 & 26 & 1305(1029 ) & 875(502 ) & 1772 & 1243 + 10 x 5 & 0.75(1.53 ) & 1.00(2.10 ) & 6.00 & 8.00 & 31 & 31 & 624(434 ) & 349(185 ) & 49 & 35 + 10 x 10 & 0.65(1.15 ) & 0.33(0.75 ) & 5.00 & 4.00 & 26 & 31 & 807(457 ) & 578(464 ) & 126 & 121 + 10 x 25 & 0.70(1.01 ) & 0.49(0.82 ) & 4.40 & 3.60 & 18 & 24 & 1548(820 ) & 1269(324 ) & 659 & 595 + 10 x 50 & 0.39(0.55 ) & 0.38(0.40 ) & 2.80 & 1.60 & 15 & 13 & 2429(1540 ) & 1914(1255 ) & 2421 & 2043 + 10 x 100 & 0.50(0.75 ) & 0.38(0.60 ) & 3.00 & 2.80 & 6 & 13 & 2859(1494 ) & 2431(1869 ) & 7150 & 6134 + 25 x 5 & 3.48(1.47 ) & 2.18(1.51 ) & 6.40 & 6.40 & 0 & 6 & 1581(450 ) & 1336(838 ) & 208 & 182 + 25 x 10 & 1.17(1.01 ) & 0.76(0.79 ) & 4.40 & 2.80 & 3 & 13 & 2256(1616 ) & 1664(714 ) & 516 & 496 + 25 x 25 & 1.18(0.74 ) & 0.80(0.69 ) & 3.52 & 2.88 & 0 & 4 & 4392(2236 ) & 3231(730 ) & 2556 & 1970 + 25 x 50 & 0.98(0.64 ) & 0.76(0.79 ) & 3.12 & 4.00 & 0 & 1 & 5522(1851 ) & 4478(1968 ) & 7367 & 5983 + 25 x 100 & 1.22(0.76 ) & 1.02(0.80 ) & 3.52 & 3.44 & 0 & 0 & 6704(2977 ) & 4824(3691 ) & 22793 & 15582 + 50x 5 & 4.54(1.21 ) & 2.74(0.82 ) & 8.40 & 4.80 & 0 & 0 & 2805(825 ) & 3111(2346 ) & 652 & 644 + 50 x 10 & 1.63(0.78 ) & 0.85(0.49 ) & 3.80 & 2.00 & 0 & 1 & 4818(1732 ) & 3410(1566 ) & 1746 & 1518 + 50 x 25 & 1.72(0.66 ) & 0.90(0.37 ) & 3.36 & 1.60 & 0 & 0 & 7604(1662 ) & 5735(1714 ) & 7030 & 4764 + 50 x 50 & 1.61(0.45 ) & 0.97(0.36 ) & 2.76 & 1.92 & 0 & 0 & 8573(1729 ) & 7110(2077 ) & 18569 & 13504 + 50 x 100 & 1.50(0.60 ) & 1.01(0.44 ) & 2.96 & 2.72 & 0 & 0 & 9178(705 ) & 8132(1189 ) & 52250 & 39889 + 100 x 5 & 5.14(1.03 ) & 3.32(0.74 ) & 7.60 & 5.60 & 0 & 0 & 5442(1610 ) & 4417(976 ) & 2876 & 1843 + 100 x 10 & 2.36(0.57 ) & 1.03(0.45 ) & 3.50 & 1.80 & 0 & 0 & 8590(1161 ) & 6651(1594 ) & 6909 & 4301 + 100 x 25 & 2.52(0.58 ) & 1.19(0.42 ) & 4.24 & 2.16 & 0 & 0 & 9675(146 ) & 8769(3070 ) & 20390 & 12047 + 100 x 50 & 2.51(0.59 ) & 1.23(0.34 ) & 4.18 & 2.28 & 0 & 0 & 9641(78 ) & 9319(387 ) & 46193 & 30645 + 100 x 100 & 2.36(0.47 ) & 1.26(0.33 ) & 3.61 & 1.95 & 0 & 0 & 9708(245 ) & 9702(808 ) & 129312 & 87792 + as it can be seen in tables [ table : alpha2 ] , [ table : alpha15 ] the local search based on the use of subgraphs @xmath266 manages to significantly outperform @xmath235 . as expected , @xmath266 produces a higher level of improvement for problem instances having a higher number of roots , or in other words when the graph is divided into a greater number of subgraphs . in case of problems having 100 subgraphsthe average error is nearly halved . for problem instances with lower values of @xmath25this improvement is smaller but consistent . in the 50 problem sets , in only 5 @xmath235 produces slightly better resultsthis only occurred for problems having five subgraphs , in which the use of neighborhoods is not essential .overall both methods had a good performance , with an average error going up to @xmath305/@xmath306 and @xmath306/@xmath307 for problem instance generated using @xmath308 for @xmath235 and @xmath266 , respectively . in practice the problem instances having the maximal number of allowed nodes @xmath309 proved to be the hardest .the methods managed to find better solutions for graphs with higher edge densities ( @xmath310 ) .we believe that the main reason for this is that such problem instances are in practice easier to solve since there is a greater number of potential bi - connected subgraphs .the hardest instances to solve are the ones in which the maximal allowed size of a graph is the smallest and the number of subgraphs is the highest . for a notable number of test instances @xmath266 manages to find optimal solutions , in case of @xmath308 it is close to @xmath311 .it is important to note that optimal solutions are only found for graphs having up to 500 nodes .@xmath266 has a robust behavior in the sense that the maximal error for a problem set is @xmath312 but in the majority of the cases it is less than @xmath313 for @xmath308 .when we observe the computational speed of the proposed methods the additional cost for calculating the neighborhoods in @xmath266 is overall neglectable when compared to @xmath235 . in case of large problem instancesthe use of neighborhoods even decreases the calculation time .overall both methods prove to be computationally very efficient having taken 100/120 and 80/87 seconds to generate close to 10 000 solutions in case of graphs having 10 000 nodes and @xmath308 for @xmath235 and @xmath266 , respectively .it is important to note that for the largest instances in case of both methods the execution of the algorithm was terminated before the stagnation occurred ( no improvements of the best found solution ) .the scaling of the methods is good in the sense that the increase of average execution time from instances having 100 nodes to 10000 was around 10 000 times .the number of iterations performed for different problem instances for the same pair @xmath314 would highly vary which can be seen from the standard variation .we wish to point out that @xmath266 manages to find high quality solutions with a low number of generated solutions when compared to the solution space .in this paper we have introduced a new problem of finding the maximal number of nodes contained in a set of disjoint bi - connected subgraphs of a graph with the additional constraint on the maximal size of a subgraph .this type of problem can be potentially applied for many practical problems .one example is the partitioning of electrical grids into a system of interconnected microgrids with a high level of resistance to failure . for solving the mbcpg - sc ,a novel computationally efficient method for growing bi - connected subgraphs has been introduced .this method has been adapted to the setting of growing multiple graphs in parallel to generate solutions for mbcpg - sc .the quality of solutions generated in this way was further improved using a local search method exploiting neighboring relations between subgraphs .the proposed method managed to acquire approximate solutions having an average error of up to @xmath306 when compared to known optimal solutions .further , the method is highly computationally efficient in the sense that it manages to find such solutions within two minutes for graphs having 10.000 nodes . in the futurewe plan to extend the current research in several directions .first , we aim to explore the weighted version of the problem which would be more suitable for problems occurring in electrical distribution systems .once the weighted version is solved we may also relate the problem towards multi - depot vehicle rooting problems and the assignment of costumers to routes of different depots .secondly , we shall extend the method to metaheuristic approaches like ant colony optimization , grasp or variable neighborhood search which appear as good options .finally , we shall explore the potential of applying the proposed growth procedure for problems like the bi - connected dominating set and bi - connected vertex cover problem .banerjee s , khuller s ( 2001 ) a clustering scheme for hierarchical control in multi - hop wireless networks . in : infocom 2001 .twentieth annual joint conference of the ieee computer and communications societies . proceedings .ieee , ieee , vol 2 , pp 10281037 jovanovic r ( 2013 ) benchmark data sets for the problem of the maximal bi - connected partitioning of a graph with a size constraint .http://mail.ipb.ac.rs/~{}rakaj/home/mbcgpsc.htm[http://mail.ipb.ac.rs/~\{}rakaj/home/mbcgpsc.htm ]morgan m , grout v ( 2008 ) finding optimal solutions to backbone minimisation problems using mixed integer programming . in : seventh international network conference ( inc 2008 ) , plymouth , uk , july 8 - 10 , 2008. proceedings , pp 5363 popa a ( 2013 ) modelling the power supply network - hardness and approximation .in : chan th , lau l , trevisan l ( eds ) theory and applications of models of computation , lecture notes in computer science , vol 7876 , springer , berlin , pp 6271

<S> in this paper we propose a new problem of finding the maximal bi - connected partitioning of a graph with a size constraint ( mbcpg - sc ) . with the goal of finding approximate solutions for the mbcpg - sc , </S><S> a heuristic method is developed based on the open ear decomposition of graphs . </S><S> its essential part is an adaptation of the breadth first search which makes it possible to grow bi - connected subgraphs . </S><S> the proposed randomized algorithm consists of growing several subgraphs in parallel . </S><S> the quality of solutions generated in this way is further improved using a local search which exploits neighboring relations between the subgraphs . in order to evaluate the performance of the method , an algorithm for generating pseudo - random unit disc graphs with known optimal solutions </S><S> is created . </S><S> the conducted computational experiments show that the proposed method frequently manages to find optimal solutions and has an average error of only a few percent to known optimal solutions . </S><S> further , it manages to find high quality approximate solutions for graphs having up to 10.000 nodes in reasonable time . </S><S> example.eps gsave newpath 20 20 moveto 20 220 lineto 220 220 lineto 220 20 lineto closepath 2 setlinewidth gsave .4 setgray fill grestore stroke grestore </S>

1102.1247

a new technique called ` polarization ' has recently been introduced in @xcite to develop efficient channel coding schemes .the codes resulting from this technique , called polar codes , have several nice attributes : ( 1 ) they are linear codes generated by a low - complexity deterministic matrix ( 2 ) they can be analyzed mathematically and bounds on the error probability ( exponential in the square root of the block length ) can be _ proved _ ( 3 ) they have a low encoding and decoding complexity ( 4 ) they allow to reach the shannon capacity on any discrete memoryless channels ( dmc ) .these codes are indeed the first codes with low decoding complexity that are provably capacity achieving on any dmc .the key result in the development of polar code is the so - called ` polarization phenomenon ' , initially shown in the channel setting in @xcite .the same phenomenon admits a source setting formulation , as follows .[ thmari][@xcite ] let @xmath0 $ ] be i.i.d .bernoulli(@xmath1 ) , @xmath2 be a power of 2 , and @xmath3 , where @xmath4^{\otimes \log_2(n)}$ ] . then , for any @xmath5 , @xmath6 : h(y_j | y^{j-1 } ) \geq 1-{\varepsilon}\}| \stackrel{n \to \infty}{\longrightarrow } h(p ) , \label{polar}\end{aligned}\ ] ] where @xmath7 is the entropy of a bernoulli(@xmath1 ) distribution .note that implies that the proportion of components @xmath8 for which @xmath9 tends to 0 .hence most of the randomness has been extracted in about @xmath10 components having conditional entropy close to 1 and indexed by @xmath11 : h(y_j |y^{j-1 } ) \geq 1 -{\varepsilon}\ } \label{defr}\end{aligned}\ ] ] and besides @xmath12 fluctuating components , the remaining @xmath13 components have conditional entropy below @xmath14 .this theorem is extended in @xcite to @xmath0 $ ] being i.i.d . from an arbitrary distribution @xmath15 on @xmath16 , where @xmath17 is a prime , replacing @xmath7 by @xmath18 ( and using the logarithm in base @xmath17 ) .it is however mentioned that the theorem may fail when @xmath17 is not a prime but a power of a prime , with a counter - example provided for @xmath19 . in section [ galois ] of this paper ,we show a generalized version of the polarization phenomenon , i.e. , of theorem [ thmari ] , for powers of primes ( we show it explicitly for powers of 2 , but the same holds for arbitrary primes ) .also , the formulation of theorem [ thmari ] is slightly more general in @xcite , it includes an auxiliary random variable @xmath20 ( side - information ) , which is a random variable correlated with @xmath21 but not intended to be compressed , and which is introduced in the conditioning of each entropy term .although this formulation is mathematically close to theorem [ thmari ] , it is more suitable for an application to the slepian - wolf coding problem ( distributed data compression ) , by reducing the problem to single - user source coding problems .a direct approach for this problem using polar codes is left open for future work in @xcite ; we investigate this here in section [ sw ] .finally , we also generalize theorem [ thmari ] to a setting allowing dependencies within the source ( non i.i.d . setting . )this paper provides a unified treatment of the three problems mentioned above , namely , the compression of multiple correlated sources , non i.i.d . sources and non binary sources .the main result of this paper is theorem [ main ] , where a `` matrix polarization '' shows how not only randomness but also dependencies can be extracted using @xmath22 .some results presented in this paper can be viewed as counter - parts of the results in @xcite for a source rather than channel setting .reciprocally , some results presented here in the source setting can be extended to a channel setting ( such as channels with memory , or non - prime input alphabets ) .finally , connections with extractors in computer science and the matrix completion problem in machine learning are discussed in sections [ pexts ] and [ discussion ] .* @xmath23=\{1,2,\dots , n\}$ ] * for @xmath24 and @xmath25 $ ] , @xmath26=[x_i : i \in s]$ ] * for @xmath24 , @xmath27 $ ] * @xmath28 \cup [ 1-{\varepsilon},1+{\varepsilon } ] \cup \dots\cup [ m-{\varepsilon},m+{\varepsilon}]$ ] * @xmath29 * for a matrix @xmath30 , the matrix @xmath31 is obtained by taking @xmath32 kronecker products of @xmath30 with itself .a random variable @xmath33 over @xmath34 is @xmath14-uniform if @xmath35 , and it is @xmath14-deterministic if @xmath36 .we also say that @xmath33 is @xmath14-deterministic given @xmath37 if @xmath38 .[ main ] ( 1 ) let @xmath2 be a power of 2 and @xmath21 be an @xmath39 random matrix with i.i.d. columns of arbitrary distribution @xmath15 on @xmath40 .let @xmath3 where @xmath41^{\otimes \log_2(n)}$ ] .then , for any @xmath42 , there exist two disjoint subsets of indices @xmath43 \times [ n]$ ] with @xmath44 \times [ n ] \setminus ( r_{\varepsilon}\cup d_{\varepsilon})|=o(n)$ ] such that the subset of entries @xmath45 $ ] is @xmath14-uniform and @xmath46 $ ] is @xmath14-deterministic given @xmath47 $ ] .( hence @xmath48 , @xmath49 . )\(2 ) moreover , the computation of @xmath20 as well as the reconstruction of @xmath21 from the non - deterministic entries of @xmath20 can be done in @xmath50 , with an error probability of @xmath51 , @xmath52 , using the algorithm ` polar - matrix - dec ` .* remarks .* * the multiplication @xmath53 is over @xmath54 * the sets @xmath55 depend on the distribution @xmath15 ( and on the dimensions @xmath56 and @xmath2 ) , but not on the realization of @xmath20 .these sets can be accurately computed in linear time ( cf .section [ discussion ] ) . * to achieve an error probability of @xmath51 , one picks @xmath57 , for @xmath58 . the following lemma provides a characterization of the dependencies in the columns of @xmath20 , it is proved in section [ proofs1 ] .recall that @xmath59 denotes the @xmath8-th column of @xmath20 , @xmath60 the @xmath61-entry of @xmath20 , @xmath62=[y_j(i ) : i \in s]$ ] and @xmath63 $ ] .[ mainlemma ] for any @xmath42 , we have , @xmath64 : h(y_j[s]|y^{j-1 } ) \in \{0,1,\dots,|s|\ } \pm { \varepsilon } , \forall s \subseteq [ m ] \}| \\ \to 1 \end{aligned}\ ] ] this lemma implies the first part of theorem [ main ] , as shown in next section .the second part of the theorem is proved in section [ proofs2 ] , together with the following result , which further characterizes the dependency structure of @xmath20 .[ null ] for any @xmath42 and @xmath65 $ ] , let @xmath66 denote the binary matrix of maximal rank such that @xmath67 note that @xmath66 can have zero rank , i.e. , @xmath66 can be a matrix filled with zeros .we then have , @xmath68 moreover , the result still holds when @xmath69 , for @xmath58 .note that , if @xmath70 , @xmath71 is @xmath14-deterministic given @xmath72 , and if @xmath66 has rank @xmath73 , by freezing @xmath74 components in @xmath59 appropriately , say on @xmath75 , we have that @xmath71 can be reduced to a full rank matrix multiplication @xmath76 $ ] , and hence @xmath77 $ ] is @xmath14-deterministic given @xmath72 and @xmath78 $ ] . hence the number of bits to freeze , is exactly @xmath79 , and as stated in the lemma , this corresponds to the total entropy of y ( up to a @xmath12 ) .let @xmath42 and let @xmath82 be the set of indices @xmath83 $ ] for which @xmath84|y^{j-1 } ) \in \{0,1,\dots,|s|\ } \pm { \varepsilon}$ ] , for any @xmath85.$ ] from lemma [ mainlemma ] , @xmath86 .note that for @xmath87 , there exists a minimal set ( not necessarily unique ) @xmath88 such that @xmath89|y^{j-1 } ) \geq h(y_j |y^{j-1 } ) - { \varepsilon}\label{max}\end{aligned}\ ] ] which also implies @xmath89|y^{j-1 } ) \geq |t_j| - { \varepsilon}\label{max2},\end{aligned}\ ] ] and , by the chain rule and defining @xmath90 , @xmath91|y^{j-1 } y_j[s_j^c ] ) \leq { \varepsilon}. \label{corr}\end{aligned}\ ] ] ( note that if @xmath92 , we define @xmath93 so that @xmath94 $ ] . ) we then have @xmath95| ( \cup_{j \in e_n}y_j[s_j])^c ) \\& \leq \sum_{j \in e_n } h(y_j[s_j]| y^{j-1}y_j[s_j^c ] ) \leq { \varepsilon}n\end{aligned}\ ] ] and @xmath96 $ ] is @xmath14-deterministic given @xmath97)^c$ ] , so that @xmath98 .moreover , we have @xmath99 ) \geq \sum_{j \in e_n } h(y_j[t_j]|y^{j-1 } ] ) \notag \\ & \geq \sum_{j \in e_n } h(y_j |y^{j-1 } ) -{\varepsilon}n \notag \\ & \geq \sum_{j=1}^n h(y_j |y^{j-1 } ) -{\varepsilon}n - o(n)\notag \\ & = h(y ) -{\varepsilon}n - o(n ) , \label{same}\end{aligned}\ ] ] where the third inequality uses , and from , @xmath100 ) & \geq \sum_{j \in e_n } |t_j| -{\varepsilon}n.\end{aligned}\ ] ] since @xmath101 , we have @xmath102 and @xmath103 $ ] is @xmath104-uniform , so that @xmath105 . `polar - matrix - dec ` + inputs : @xmath106 \times [ n]$ ] , @xmath107 \in { \mathbb{f}}_2^{|d^c|}$ ] .+ output : @xmath108 .+ algorithm : + 0 .let @xmath109 ; + 1 . find the smallest @xmath8 such that @xmath110 is not empty ;compute @xmath111=\arg\max_{u \in { \mathbb{f}}_2^{|s_j| } } { \mathbb{p}}\ { y[s_j]= u | y^{j-1}=y^{j-1 } , y[s_j^c]=y[s_j^c ] \};\ ] ] 2 .update @xmath112 , @xmath113=y[m ] \cup \hat{y}[s_j]$ ] ; + 3 . if @xmath114 is empty output @xmath115 , otherwise go back to 1 .note that , using for the definition of @xmath116 ( and the corresponding @xmath81 ) , the realizations of @xmath72 and @xmath117 $ ] are known , and with high probability one guesses @xmath118 $ ] correctly in step 1 , because of .moreover , due to the kronecker structure of @xmath22 , and similarly to @xcite , step 1 . andthe entire algorithm require only @xmath50 computations .finally , from the proof of theorem [ main ] part ( 2 ) , it results that step 1 .can also be performed slightly differently , by finding sequentially the inputs @xmath119 $ ] for @xmath120 , reducing an optimization over all possible @xmath121 , where @xmath122 can be as large as @xmath56 , to only @xmath56 optimizations over @xmath54 ( which may be useful for large @xmath56 ) .we present now three direct applications of theorem [ main ] : * distributed data compression , i.e. , slepian - wolf coding * compression of sources on arbitrary finite fields * compression of non i.i.d .sources in @xcite , the two - user slepian - wolf coding problem is approached via polar codes by reducing the problem to single - user source coding problems .a direct approach is left open for future work ; we investigated this here , for arbitrary many users .consider @xmath56 binary sources which are correlated with an arbitrary distribution @xmath15 .we are interested in compressing an i.i.d .output of these sources .that is , let @xmath123 be i.i.d . under @xmath15 on @xmath40 ,i.e. , @xmath124 is an @xmath56 dimensional binary random vector and , for example , @xmath125,\dots , x_n[i]$ ] is the sources output for user @xmath126 .if we are encoding these sources together , a rate @xmath18 is sufficient ( and it is the lowest achievable rate ) . in @xcite , slepian and wolf showed that, even if the encoders are not able to cooperate , lossless compression can still be achieved at rate @xmath18 .we now present how to use theorem [ main ] to achieve this rate with a polar coding scheme ._ polar codes for distributed data compression : _ + 1 . for a given @xmath2 and @xmath14 ( which sets the error probability ) , since each user knows the joint distribution @xmath15 , each user can compute the `` chart '' of the deterministic indices , i.e. , the set @xmath127 \times [ n]$ ] and identify its own chart @xmath128 .each user computes @xmath129 and stores @xmath130 $ ] , so that the joint decoder is in possession of @xmath131 $ ] , and can run ` polar - dec - matrix ` with @xmath131 $ ] as input to get @xmath20 , with an error probability at most @xmath132 . since @xmath22 is invertible , indeed @xmath133 , one can then find @xmath134 . from theorem [ main ], we have the following result .[ swc][distributed polar compression ] for @xmath56 correlated sources of joint distribution @xmath15 , previously described scheme allows to perform lossless and distributed compression of the sources at sum - rate @xmath18 , with an error probability of @xmath135 , @xmath52 , and an encoding and decoding complexity of @xmath50 . notethat this result allows to achieve the sum - rate of the slepian - wolf region , i.e. , a rate belonging to the dominant face of the slepian - wolf achievable rate region , it does not say that any rate in that region can be reached with the proposed scheme . in @xcite ,the source polarization result is stated for sources that are i.i.d . and @xmath17-ary , where @xmath17 is prime .it is also mentioned that if @xmath17 is not prime , the theorem may fail .in particular , an example for @xmath19 is provided where the conclusion of theorem [ thmari ] does not hold .it is also mentioned that if additional randomness is introduced in the construction of the polar transformation ( leading no longer to a deterministic matrix @xmath22 ) , the result holds for arbitrary powers of primes .we show here that a generalized polarization phenomenon still holds for arbitrary powers of primes ( we formally show it for powers of 2 only but any prime would work ) even for the deterministic polar transform @xmath22 .[ galoisc][polarization for finite fields ] let @xmath0 $ ] be i.i.d . under @xmath15 on @xmath16 where @xmath136 , and let @xmath3 ( computed over @xmath16 ) .then , although @xmath20 may no polarize over @xmath137 , it polarizes over @xmath40 in the sense of theorem [ main ] , more precisely : define by @xmath138 a @xmath40 representation of @xmath137 , @xmath139 the distribution on @xmath40 induced by @xmath15 on @xmath137 , and set @xmath140 ( organized as an @xmath39 matrix ) .then the conclusions of theorem [ main ] hold for @xmath141 .note : this theorem still holds when @xmath17 is a power of any prime , by combining it with the result in @xcite for prime alphabets .the case where @xmath136 is particularly interesting for complexity considerations ( cf .section [ discussion ] ) ._ interpretation of corollary [ galoisc ] : _ when @xmath17 is a prime , @xmath142 , which means that @xmath59 is either roughly uniform and independent of the past or roughly a deterministic function of the past .however , for @xmath17 being a power of 2 ( or a power of a prime ) , we only get that @xmath143 , and previous conclusion can not be drawn , stressing indeed a different polarization phenomenon .however , corollary [ galoisc ] says that if we work with the vector representation of the elements in @xmath144 , we still have a ` polarization phenomenon ' in the sense of theorem [ main ] , i.e. , for almost all @xmath65 $ ] , a subset of the components of the @xmath145 are either roughly uniform and independent or deterministic functions of the past and the complementary components ._ compression of @xmath146-ary i.i.d .sources : _ for a given @xmath0 $ ] , compute @xmath3 and transform @xmath20 into @xmath141 based on the representation of @xmath137 by @xmath40 .organize @xmath141 to be an @xmath39 matrix .note one can equivalently map @xmath21 into @xmath147 and then take @xmath22 to get @xmath141 .this is due to the fact that the @xmath137 addition corresponds to the pointwise addition in @xmath40 .finally , store @xmath141 on @xmath148 , and run ` polar - matrix - dec ` to recover @xmath141 , hence @xmath20 and @xmath21 .let a binary source consist of i.i.d .blocks of length @xmath56 , each block having an arbitrary distribution @xmath15 .we can then compress the source as follows . from @xmath2 blocks@xmath123 each of length @xmath56 , i.e. , @xmath149 outputs of the source , create the matrix @xmath150 $ ] and apply the polar transform to get @xmath3 . then store the components of @xmath20 which belong to @xmath151 .to reconstruct @xmath21 , reconstruct @xmath20 from @xmath152 $ ] using ` polar - matrix - dec ` and find @xmath134 .if the source is not exactly block i.i.d .but is mixing , i.e. , if @xmath153 , for any @xmath154 , we can open windows of length @xmath155 between the blocks and store without compression these @xmath155 inter - block bits , which does not increase the compression rate .we are then left with a source formed by blocks which are ` almost ' i.i.d . anda similar procedure can be used . from theorem [ main ] , we have the following .[ memoryc]for a binary source consisting of i.i.d .blocks of length @xmath56 , each block having distribution @xmath15 , the polar coding scheme described previously allows to compress losslessly the source at rate @xmath18 , with an error probability of @xmath51 , @xmath52 , and an encoding and decoding complexity of @xmath50 .as discussed previously , a similar result holds for source which are mixing .we have discussed in this paper a procedure to extract randomness , i.e. , uniform bits , from non uniform bits .the applications we considered are in compression and coding , but there are also numerous applications of randomness extraction problems in computer science . in particular , there is a notion of `` extractor '' in theoretical computer science , which aims at extracting uniform bits from sources having much more general assumptions than the one considered here . phrased in our terminology , an extractor is roughly a map that extracts @xmath56 bits that are @xmath14-uniform from @xmath2 bits that have a total entropy at least @xmath32 , with the help of a seed of @xmath156 uniform bits . for more details and a survey on extractorssee for example @xcite .the notion of @xmath14-uniform , or @xmath14-close to uniform , used in computer science is usually measured by the @xmath157-norm , rather than the entropy as used in this paper .nevertheless , these two notions can be related and this is a minor distinction . also , the entropy used in the computer science literature is the min - entropy rather than the shannon - entropy , which is a stronger assumption , since the shannon - entropy is an upper bound to the min - entropy . on the other hand , the source for the extractor definitionis only assumed to have min - entropy @xmath32 , and no further assumptions are made on the distribution of @xmath123 , whereas in our setting , we consider sources that are at least ergodic and with a known distribution .one should also stress that we did not make use of any seed in our problems . in order to establish a more concrete connection between polar coding and formal extractors , we present here a result which takes into account one of the two caveat just mentioned : we only assume that the source has entropy at least @xmath32 , without requiring the exact knowledge of the distribution , but we keep an i.i.d . setting . using section [ memory ], one can generalize this result to a setting where the source is mixing , but in even then we do not make use of any seed .in particular , if one could use a seed , ideally of small size , e.g. , @xmath158 , to turn an arbitrary source of lower - bounded entropy , into a mixing source of comparable entropy , one could use the following result to construct real extractors ( work in progress ) .let @xmath159-@xmath160 be the matrix obtained by deleting the columns of @xmath22 that are not in @xmath161 , where @xmath162 is one of the two binary distribution having entropy @xmath163 ( and @xmath164 as defined in ) .note that @xmath165 benefits from the low encoding complexity of @xmath22 , namely @xmath50 .[ pext ] let @xmath2 be a power of two and @xmath0 $ ] be i.i.d .bernoulli such that @xmath166 ( where @xmath167 denotes the shannon or min - entropy ) . for any @xmath5 ,@xmath168 is @xmath14-uniform ( in the @xmath169 or entropy sense ) and @xmath170 this result is proved in section [ proofext ] , and using section [ memory ] it can be extended to a setting where the source is mixing .note that even in a mixing setting , the source entropy is @xmath171 , which is indeed a regime where good extractors are known @xcite .we have treated in this paper three problems , namely , compression of correlated sources , sources with memory and sources on finite fields , with a unified approach using a matrix polarization ( theorem [ main ] ) , and we provided polar coding schemes for each of these problems .the advantage of using polar coding schemes is that these schemes have low encoding and decoding complexity , and achieve the optimal performance ( shannon limit ) meanwhile affording mathematical guarantees on the performance , as described in corollaries [ swc ] , [ galoisc ] and [ memoryc ] .one can now also combine these different problems .namely , for multiple sources that are define on some finite fields , with some well - behaved correlations between and within themselves , one can , using the interleaving trick and the vector representation described respectively in sections [ memory ] and [ galois ] , organize the sources outputs in a matrix form so as to meet the hypotheses of theorem [ main ] , and hence have a polar compression scheme requiring the minimal compression rate .one can also translate the results in this paper to a channel setting , such as @xmath56-user multiple access channels ( already treated in @xcite ) , channels with memory or channels with non binary fields inputs , by using duality arguments .although the results in this paper are expected to hold when @xmath172 , one has to be careful with the complexity scaling when @xmath56 gets large . in that regard, an advantage of using finite fields of cardinality @xmath136 rather than modular fields of prime cardinality , is that some operations required in the polar decoding algorithm are convolution - like operations over the underlying field , and as the fft algorithm allows to reduce the computational cost of a convolution from @xmath173 to @xmath174 when @xmath17 is a power of 2 , one can benefit from this fact .we have assumed in this paper that the sets @xmath175 and @xmath176 can be computed , without discussing how .the first reason why we do not stress this aspect here , as in other papers in polar coding , is that these sets do not depend on the realization of the source(s ) .namely , if one is able to compute these sets once for several values of interest of @xmath14 and of the dimensions , one can then use the same sets for any outputs realizations .this is fundamentally different than the decoding algorithm which takes the source realization as an input .yet , it is still crucial to be able to compute these sets once , for the parameters of interests . in order to do so, there are at least two possible approaches .the first one is via simulations , and is discussed in @xcite : using the kronecker structure of @xmath22 , it is possible to run simulations and get accurate estimate of the conditional entropies @xmath177 , in particular ( from section [ proof1 ] ) of the sets @xmath175 and @xmath176 .another option is to use algorithms to approach the exact values of @xmath177 within a given precision , in linear time ; this has been proposed in particular in @xcite .it would also be interesting to have mathematical characterizations of these sets . at the moment , this is an open problem , even for the simplest settings ( single binary i.i.d .source , or in the channel setting , the binary erasure channel ) .finally , this work could also apply to the matrix completion setting .for example , if @xmath21 is an @xmath39 matrix where column @xmath178 contains the ratings of @xmath56 movies by user @xmath8 , we can use theorem [ main ] to show that by applying the matrix that are not in @xmath81 ] @xmath180 to @xmath21 , we are left with fewer entries ( the more correlations between the movie ratings the fewer entries ) that yet allow to recover the initial matrix . hence ,if we are given only a smaller set of appropriate entries ( and which sets can be characterized using section [ proof1 ] ) , we can reconstruct the initial matrix using ` polar - matrix - dec ` .for a random vector @xmath138 distributed over @xmath40 , define @xmath181 and @xmath182 , where @xmath183 is an i.i.d .copy of @xmath138 .let @xmath184 be i.i.d . binary random variables in @xmath185 with uniform probability distribution , andlet @xmath186=h(v^{b_1 \dots b_k}[s]| v^{c_1 \dots c_k } , \forall ( c_1 \dots c_k ) < ( b_1 \dots b_k))\end{aligned}\ ] ] for @xmath85 $ ] , where the order between @xmath187-sequences is the lexicographic order ( with @xmath188 ) . for @xmath197 , we have @xmath198 ) & = h(x_1 [ s ] x_2[s ] ] ) \notag \\ & = h(y_1 [ s ] y_2[s ] ) \notag \\ & = h(y_1 [ s ] ) + h(y_2 [ s ] | y_1[s ] ) \notag \\ & \geq h(y_1 [ s ] ) + h(y_2 [ s ] | y_1 ) \label{last } \ ] ] with equality in the if @xmath196 $ ] . for @xmath199 ,the same expansion holds including in the conditioning the appropriate `` past '' random variables . note that because @xmath191 $ ] is a martingale for @xmath196 $ ] , the sum - rate @xmath18 is conserved through the polarization process .now , using previous lemma and the fact that @xmath191 \in [ 0,|s|]$ ] for any @xmath200 , the martingale convergence theorem implies the following .[ invar ] for any @xmath42 , @xmath21 valued in @xmath201 , @xmath33 arbitrary , @xmath202 an i.i.d .copy of @xmath203 , @xmath85 $ ] , there exists @xmath204 such that @xmath205 | z ' ) - h(x ' [ s]| z , z',x[s]+x'[s ] ) \leq \delta$ ] implies @xmath206 | z')-h(x'[s \setminus i ] | z ' ) \in \{0,1 \ }\pm { \varepsilon}$ ] for any @xmath207 .we have @xmath208 | z ' ) - h(x ' [ s]| z , z',x[s]+x'[s ] ) \notag \\ & = i(x'[s ] ; x[s]+x'[s]| z , z ' ) \notag \\ & \geq i(x'[s ] ; x[i]+x'[i]| z , z ' ) \notag \\ & \geq i(x'[i ] ; x[i]+x'[i]| z , z ' , x'[s \setminus i ] ) \notag \\ & = h(x'[i]| z ' , x'[s \setminus i ] ) - h(x[i]+x'[i]| z , z ' , x'[s \setminus i ] ) .\label{squiz}\end{aligned}\ ] ] it is shown in @xcite that if @xmath209 are binary random variables and @xmath210 are arbitrary such that @xmath211 , for some conditional probability @xmath212 , then , for any @xmath213 , there exists @xmath214 such that @xmath215 implies @xmath216 . using this result, we can pick @xmath217 small enough to lower bound and show that @xmath218| z ' , x'[s \setminus i ] ) \in \{0,1\ } \pm { \varepsilon}$ ] . from the chain rule , we conclude that @xmath219| z ' ) \in \{0,1\ } \pm { \varepsilon}$ ] . in order to prove theorem [ main ] part ( 2 ) ,we basically need to show that part ( 1 ) still holds when taking @xmath14 scaling like @xmath221 for @xmath222 , as in @xcite .we did not find a direct way to show that when @xmath191 $ ] converges to @xmath223 , it must do it that fast ( the sub - martingale characterization is too week to apply results of @xcite directly ) .this is why we looked into lemma [ null ] . by developing a correspondence between previous results and analogue results dealing with linear forms of the @xmath224$ ] s , we are able to use the speed convergence results shown for the single - user setting and conclude .this approach was developed in @xcite for the multiple access channel , below is the counter - part for our source setting .[ tech1 ] for a random vector @xmath20 valued in @xmath40 , and an arbitrary random vector @xmath33 , if @xmath225| z ) \in \{0,1,\dots,|s|\ } \pm { \varepsilon}\ ] ] for any @xmath85 $ ] , we have @xmath226|z ) \in \{0,1\ } \pm \delta({\varepsilon}),\ ] ] with @xmath227. this lemma is proved in @xcite .using this result , we have that for @xmath228 , there exists a matrix @xmath66 of rank @xmath229 , such that @xmath230 this implies the first part of lemma [ null ] , and we now show how we can use this other characterization of the dependencies in @xmath20 to conclude a speed convergence result .we first need the following `` single - user '' result .we define the auxiliary family of random processes @xmath234 $ ] , for @xmath85 $ ] , by @xmath235=z(\sum_{i \in s}v^{b_1 \dots b_k}[i]| v^{c_1 \dots c_k } , \forall ( c_1 \dots c_k ) < ( b_1 \dots b_k))\end{aligned}\ ] ] where , for a binary uniform random variable @xmath30 and an arbitrary random variable @xmath236 , @xmath237 is the bhattacharyya parameter .note that @xmath238 ( this also follows from proposition 2 in @xcite . ) we then have , using the chain rule and source polarization inequalities on the bhattacharyya parameter , namely proposition 1 in @xcite , that @xmath239 \leq \zeta_{k}[s]^2 \text { if } b_{k+1}=1,\\ & \zeta_{k+1}[s ] \leq 2 \zeta_k[s ] \text { if } b_{k+1}=0 , \ ] ] and using theorem 3 of @xcite , we conclude that for any @xmath58 @xmath240 finally , we conclude using .we then use lemma [ tech1 ] and [ tech2 ] to conclude that @xmath233 : & h(y_j [ s]| y^{j-1 } ) \in \{0,1,\dots,|s|\ }\pm { \varepsilon } , \forall s \subseteq [ m ] , \notag \\ & \exists a_j \text { with } { \mathrm{rank}}(a_j)= \mathrm{int } ( m - h(y_j | y^{j-1})),\notag \\ & h(a_jy_j |y^{j-1 } )< { \varepsilon}_n \ } | \to 1 , \label{set}\end{aligned}\ ] ] which implies lemma [ null ] . to conclude the proof of theorem [ main ] part ( 2 ) , let @xmath221 and @xmath241 be the set defined through ( which , in view of previous results , is equivalent to the definition given in section [ proof1 ] ) .we then have for @xmath228 that the components @xmath116 to be decoded in @xmath59 are not correctly decoded with probability @xmath242 and the block error probability is bounded as @xmath243 so that taking @xmath222 large enough , we can reach a block error probability of @xmath244 for any @xmath52 . for @xmath245 , @xmath246 where @xmath247 and @xmath248 where @xmath249 is i.i.d .under @xmath162 . moreover , for any distribution @xmath1 on @xmath54 such that @xmath250 , there exists a distribution @xmath251 on @xmath54 such that @xmath252 , where @xmath253 denotes the circular convolution .equivalently , there exists @xmath254 independent of @xmath255 , such that @xmath256 .define @xmath257 , @xmath258 and @xmath259 , hence @xmath260 .we have @xmath261 where the last equality follows from the fact that @xmath262 is independent of @xmath37 since @xmath263 is independent of @xmath33 .therefore , for any @xmath263 i.i.d . such that @xmath264 and for any @xmath265 , we have @xmath266 and @xmath267 ) & \geq \sum_{j \in r_{\tilde{{\varepsilon}}}(p(k ) ) } h(y_j(p)|y^{j-1}(p ) ] ) \notag \\ & \geq |r_{\tilde{{\varepsilon}}}(p(k))| ( 1-\tilde{{\varepsilon } } ) .\notag \end{aligned}\ ] ] hence , defining by @xmath268 the distribution of @xmath269 $ ] and @xmath270 the uniform distribution on @xmath271 , we have @xmath272 using pinsker inequality and , we obtain @xmath273 finally , we have from theorem [ thmari ] latexmath:[\[\begin{aligned }

<S> the basic polarization phenomenon for i.i.d . </S><S> sources is extended to a framework allowing dependencies within and between multiple sources . </S><S> in particular , it is shown that taking the polar transform of a random matrix with i.i.d . </S><S> columns of arbitrary ( correlated ) distribution allows to extract the randomness and dependencies . </S><S> this result is the used to develop polar coding schemes ( having low complexity ) for : ( 1 ) distributed data compression , i.e. , slepian - wolf coding ( without decomposing the problem into single - user problems ) , ( 2 ) compression of sources with memory , ( 3 ) compression of sources on finite fields , extending the polarization phenomenon for alphabets of prime cardinality to powers of primes . </S>

1112.1584

the concept of physical layer network coding has attracted a lot of attention in recent times .the idea of physical layer network coding for the two - way relay channel was first introduced in [ 1 ] , where the multiple access interference occurring at the relay was exploited so that the communication between the end nodes can be done using a two stage protocol .information theoretic studies for the physical layer network coding scenario were reported in [ 2 ] , [ 3 ] .the design principles governing the choice of modulation schemes to be used at the nodes for uncoded transmission were studied in [ 4 ] .an extension for the case when the nodes use convolutional codes was done in [ 5 ] . a multi - level coding scheme for the two - way relaying was proposed in [ 6 ] .+ we consider the three - way wireless relaying scenario shown in fig .1 , where three - way data transfer takes place among the nodes a , b and c with the help of the relay r. it is assumed that the three nodes operate in half - duplex mode .the relaying protocol consists of two phases , _ multiple access _ ( ma )phase , consisting of one channel use during which a , b and c transmit to r ; and _ broadcast _ ( bc ) phase , in which r transmits to a , b and c in a single channel use .network coding is employed at r in such a way that a(/b / c ) can decode b s and cs(/a s and cs /a s and b s ) messages , given that a(/b / c ) knows its own message .+ for a two - way wireless relay channel , it was observed in [ 4 ] for 4-psk , that for uncoded transmission , the network coding map used at the relay needs to be changed adaptively according to the channel fade coefficient , in order to minimize the impact of multiple access interference . in other words ,the set of all possible channel realizations is quantized into a finite number of regions , with a specific network coding map giving the best performance in a particular region .it is shown in [ 7 ] for every m - psk constellation used by both the users , that every such network coding map that satisfies the _ exclusive law _ is representable as a latin square and this relationship can be used to get the network coding maps satisfying the exclusive law .a latin square of order @xmath0 is defined to be an @xmath1 array in which each cell contains a symbol from @xmath2 such that each symbol occurs at most once in each row and column [ 8 ] .+ while most of the research has been done for two - way relay channels , some work has been done for the relay channels with three or more user nodes as well .liu and arapostathis [ 9 ] , proposed a joint network coding and superposition coding for three user relay channels , and claim that the results can be easily extended to information relaying cases with more than three relay nodes . in this scheme ,two stage operations are required for encoding and decoding , and four channel uses are required for the information exchange , three channel uses for the ma phase , and one channel use for the bc phase , while for our scheme totally two channel uses suffice . in [ 9 ] ,the first three channel uses are utilized by each user node transmitting its packet to the relay node .then the relay makes two xor - ed packets and superimposes them together for broadcast .the relay makes two xor - ed packets , the packet from the node with the worse channel gain is xor - ed respectively with the other two packets .pischella and ruyet [ 10 ] also discuss the three - way wireless relaying scenario , and propose a method of information exchange among the users composed of alternate ma and bc phases .the physical layer network coding strategy for this relaying protocol is a lattice - based coding scheme combined with power control , so that the relay receives an integer linear combination of the symbols transmitted by the user nodes .this scheme also , however , consists of four channel uses .park and oh , in [ 11 ] , propose a network coding scheme for the three - way relay channels and present a ` latin square - like condition ' for the three - way network code .they also discuss schemes in order to improve these codes using cell swapping techniques . in this work , though latin cubes have been suggested as being equivalent to the map used by the relay , the number of channel uses the scheme uses is five , and the work does nt deal with the channel gains associated with the channels explicitly . in [ 12 ] , authors jeon et al . adopt an ` opportunistic scheduling technique ' for physical network coding where users in the ma as well as the bc phase are selected on the basis of instantaneous snr using a channel norm criterion and a minimum distance criterion and plot graphs to show that the proposed scheme outperforms systems without this selection .their approach , however , utilizes six channel uses . + for our physical layer network coding strategy , we use a mathematical structure called a latin cube , that has three dimensions out of which one is represented along the rows , one along the columns , and the third dimension is represented along ` files ' . in our case , we have a s transmitted symbol along the files , b s symbol along the rows , and c s symbol along the columns . for our purposes , we define latin cubes as follows : + a _ latin cube l of second order of side m _ on the symbols from the set @xmath3 is an @xmath4 array , in which each cell contains one symbol and each symbol occurs at most once in each row , column and file .+ the above definition , is given in [ 13 ] with @xmath5 ._ multiple access phase : _+ let @xmath6 denote the symmetric 4-psk constellation @xmath7 as shown in fig . 2 , used at a, b and c. assume that a(/b / c ) wants to send a 2-bit binary tuple to b and c(/a and c / a and b ) .let @xmath8 denote the mapping from bits to complex symbols used at a , b and c where @xmath9 .let @xmath10 denote the complex symbols transmitted by a , b and c respectively , where @xmath11 .it is assumed that the channel state information is not available at the transmitting nodes a , b and c during the ma phase .the received signal at r in the ma phase is given by @xmath12 @xmath13 where @xmath14 , @xmath15 and @xmath16 are the fading coefficients associated with the a - r , b - r and c - r link respectively .the additive noise @xmath17 is assumed to be @xmath18 , where @xmath18 denotes the circularly symmetric complex gaussian random variable with variance @xmath19 .+ let @xmath20 denote the effective constellations seen at the relay during the ma phase channel use , i.e. , @xmath21 let @xmath22 denote the minimum distance between the points in the constellation @xmath23 as given in ( [ dist ] ) , where @xmath24 . from ( [ dist ] ) ,it is clear that there exists values of @xmath25 , for which @xmath26 .let @xmath27 .the elements of @xmath28 are called singular fade states . for singular fade states , @xmath29 .+ a fade state @xmath25 is defined to be a _ singular fade state _ for the ma phase of three - way relaying , if the cardinality of the signal set @xmath20 is less than @xmath30 .let @xmath28 denote the set of all singular fade states for the three - way data transfer among a , b and c. + let @xmath31 denote the maximum likelihood ( ml ) estimate of @xmath32 at r based on the received complex number @xmath33 , i.e. , @xmath34 where , @xmath35\\ \nonumber \vspace{0.1 cm } & x= \left [ { \begin{array}{cc } \vspace{0.15 cm } x_{a } \\ \vspace{0.15 cm } x_{b } \\ x_{c } \\\end{array } } \right ] .\nonumber\end{aligned}\ ] ] _ broadcast ( bc ) phase : _ + the received signals at a , b and c during the bc phase are respectively given by , @xmath36 where @xmath37 is the complex number transmitted by r. the fading coefficients corresponding to the r - a , r - b and r - c links are given by @xmath38 @xmath39 and @xmath40 respectively and the additive noises @xmath41 @xmath42 and @xmath43 are @xmath44 . depending on the values of @xmath45 , @xmath46 and @xmath47 ,r chooses a many to one map @xmath48 where @xmath49 is a signal set of size between @xmath50 and @xmath30 used by r during _ bc _ phase .notice that the minimum required size for @xmath49 is 16 , since 4 bits about the other two users needs to be conveyed to each of a , b and c. + the elements in @xmath51 which are mapped to the same signal point in @xmath49 by the map @xmath52 are said to form a cluster .let @xmath53 denote the set of all such clusters .the formation of clusters is called clustering , denoted by @xmath54 .+ the cluster distance between a pair of clusters @xmath55 and @xmath56 is the minimum among all the distances calculated between the points @xmath57 and @xmath58 in the effective constellation seen at the relay node r , as given in ( [ cl1 ] ) above .+ the _ minimum cluster distance _ of the clustering @xmath54 is the minimum among all the cluster distances , as given in ( [ cl2 ] ) at the top of this page . + in order to ensure that a(/b / c ) is able to decode b s and cs(/a s and c s /a s and b s ) message , the clustering @xmath59 should satisfy the exclusive law , as given in ( [ mel1 ] ) , ( [ mel2 ] ) , ( [ mel3 ] ) above .+ the minimum cluster distance determines the performance during the ma phase of relaying .the performance during the bc phase is determined by the minimum distance of the signal set @xmath49 . for values of @xmath60 in the neighborhood of the singular fade states ,the value of @xmath61 is greatly reduced , a phenomenon referred to as _ distance shortening _ [ 4 ] . to avoid distance shortening , for each singular fade coefficient, a clustering needs to be chosen such that the minimum cluster distance at the singular fade state is non zero and is also maximized . +a clustering @xmath54 is said to remove singular fade state @xmath62 if @xmath63 . for a singular fade state @xmath64 ,let @xmath65 denote the clustering which removes the singular fade state @xmath60 ( if there are multiple clusterings which remove the same singular fade state @xmath60 , consider a clustering which maximizes the minimum cluster distance ) .let @xmath66 denote the set of all such clusterings .+ the minimum cluster distance of the clustering @xmath67 , when the fade state @xmath25 occurs in the ma phase , denoted by @xmath68 , is the minimum among all its cluster distances , as given in ( [ cl3 ] ) .+ for @xmath69 the clustering @xmath70 is chosen to be @xmath71 , which satisfies @xmath72 in [ 7 ] , such clusterings that remove singular fade states are obtained with the help of latin squares while concentrating only on the first minimum cluster distance .the clustering used by the relay is indicated to a , b and c using overhead bits .+ the contributions of this paper are as follows : * using our proposed scheme , exchange of information in the wireless three - way relaying scenario is made possible with totally two channel uses . *it is shown that if the three users a , b , c transmit points from the same 4-psk constellation , the requirement of satisfying the exclusive law is same as the clustering being represented by a latin cube of second order of side 4 . to the best of our knowledge ,this is the first work with only two channel uses for the three - way relaying scenario .* the singular fade states for the three user case are identified . *clusterings that removes these singular fade states are obtained , that result in the size of the constellation used by the relay node r in the bc phase to lie between 16 to 23 .* simulation results are provided to verify that the adaptive clustering as obtained in the paper indeed performs better than non - adaptive clustering .the remaining content is organized as follows : section ii demonstrates how a latin cube of second order and side 4 can be utilized to represent the network code for three user communication . in sectioniii the singular fade states are specified and in section iv , clusterings corresponding to removal of each singular fade state are obtained using latin cubes of second order .simulation results are shown in section v. section vi concludes the paper .the nodes a , b and c transmit symbols from the same constellation , viz . ,our aim is to find the map that relay node r should use in order to cluster the @xmath30 possibilities of @xmath73 such that the exclusive law ( given by ( [ mel1 ] ) , ( [ mel2 ] ) , ( [ mel3 ] ) ) is satisfied .the size of the constellation , or the number of clusters of the clustering that the relay utilizes has to be at least 16 , since each user needs the 4 bit information corresponding to the other two users . consider a @xmath74 array , whose 64 entries are indexed by @xmath73 , i.e. the three messages that a , b and c send in the ma phase .each file of this @xmath74 array , is indexed by a single value of @xmath75 . each row( column ) of each file is indexed by a value of @xmath76 ( @xmath77 ) , for a fixed value of @xmath78 .now , a repetition of a symbol in a file results in the failure of exclusive law given by ( [ mel1 ] ) . consider the @xmath79 array with its rows being the first(/second / third / forth ) rows of the @xmath80 array .each @xmath81 array so obtained , corresponds to a single value of @xmath82 .a repetition of a symbol in this array will result in the failure of exclusive law given by ( [ mel2 ] ) .similarly , consider the @xmath79 array with its columns being the first(/second / third / forth ) columns of the @xmath80 array .each @xmath79 array so obtained , corresponds to a single value of @xmath83 .a repetition of a symbol in this array will result in the failure of exclusive law given by ( [ mel3 ] ) .hence , if the exclusive law needs to be satisfied , then the cells of this array should be filled such that the @xmath80 array so obtained , is a latin cube of second order , for @xmath84 ( definition 1 ) .the clusters are obtained by putting together all the tuples @xmath85 such that the entry in the @xmath86-th slot is the same entry from @xmath87 . from above, we can say that all the relay clusterings that satisfy the mutually exclusive law forms latin cubes of second order of side 4 for @xmath84 , when the end nodes use 4-psk constellations .hence , now onwards , we consider the network code used by the relay node in the bc phase to be a @xmath74 array with files(/rows / columns ) being indexed by the constellation point used by a(/b / c ) , symbols from the set @xmath88 ( fig .the cells of the array will be filled with elements of @xmath87 in such a way , that the resulting array is a latin cube of second order of side 4 and @xmath84 .any arbitrary but unique symbol from @xmath87 denotes a unique cluster of a particular clustering .we earlier stated in section ii , that a clustering @xmath54 is said to remove singular fade state @xmath62 if @xmath63 .alternatively , removing singular fade states for a three - way relay channel can also be defined as follows : + a clustering @xmath54 is said to _ remove the singular fade state _ @xmath89 , if any two possibilities of the messages sent by the users @xmath90 that satisfy + @xmath91 are placed together in the same cluster by the clustering .+ a set @xmath92 consisting of all the possibilities of @xmath93 that must be placed in the same cluster of the clustering used at relay node r in the bc phase in order to remove the singular fade state @xmath60 is referred to as a _ singularity removal constraint _ for the fade state @xmath60 for three - way relaying scenario .+ let @xmath94 be the fade coefficient in the ma phase .the work in [ 4 ] and [ 7 ] shows that for the two - way relaying scenario , the @xmath50 possible pairs of symbols from 4-psk constellation sent by the two users in the ma phase , can be clustered into a clustering dependent on a singular fade coefficient , of size 4 or 5 in a manner so as to remove this singular fade coefficient . in the case of three users , at the end of ma phase , relay receives a complex number , given by ( [ yr ] ) . instead of r transmitting a point from the @xmath30 point constellation resulting from all the possibilities of @xmath73 ,the relay r can choose to group these possibilities into clusters represented by a smaller constellation .we describe one such clustering in the following .+ let @xmath95 denote a singularity removal constraint corresponding to the singular fade state @xmath94 and let @xmath96 .then , @xmath97\right\rangle ^{\bot}\end{aligned}\]]where for a @xmath98 non - zero vector @xmath99 over @xmath100 , @xmath101 clearly , @xmath102 is a two - dimensional vector space of @xmath103 .these values of @xmath104 result in only finitely many possibilities for the right - hand side , since @xmath6 is finite .thus the singular fade states @xmath105 which are uncountably infinite , are points in a finite number of vector subspaces of @xmath103 .henceforth , we shall refer to these finite number of vector subspaces as the _ singular fade subspaces_. more precisely , there are three possibilities of singular fade subspaces for the three - way relaying as we explain individually in the following three cases .+ + _ case 1 : _ one of the following subcases arise : 1 .@xmath106 2 .@xmath107 3 .@xmath108 _ case 2 : _ one of the following subcases arise : 1 .@xmath109 2 .@xmath110 3 .@xmath111 _ case 3 : _ @xmath112 + + * _ case 1 : _ * without loss of generality, we discuss the third subcase of case 1 , i.e. , the case when @xmath108 .the singular fade subspace in this case is given by @xmath113\right\rangle ^{\bot}. $ ] the set of differences of the points of @xmath6 is given by , @xmath114 let @xmath115 and @xmath116 .then , @xmath117 thus , @xmath118 can take eight non - zero values . as a result , there are eight total possibilities for the vector @xmath119^t $ ] , where @xmath120 denotes the transpose of a vector @xmath99 .also , each one of @xmath121 , @xmath122 and @xmath123 can be obtained as scalar multiples of @xmath124 ( over @xmath100 ) .thus , + + @xmath125\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } \pm 1\pm j \\ \vspace{0.15 cm } 0 \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } \pm 2j \\ \vspace{0.15 cm } 0 \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } \pm 2 \\\vspace{0.15 cm } 0 \\ 0 \\ \end{array } } \right]\right\rangle.}\\$ ] so , @xmath126 only one singular fade subspace for the subcase , viz ., @xmath127\right\rangle ^{\bot}.\ ] ] similarly , for the other two subcases , there is one singular fade subspace , resulting in a total of 3 singular fade subspaces for the case .+ + * _ case 2 : _ * without loss of generality , we discuss the third subcase of case 2 , i.e. , the case when @xmath111 .the singular fade subspace in this case is given by @xmath128\right\rangle ^{\bot}.$ ] here @xmath129 can take eight non - zero values each .there are therefore , 64 total possibilities for the vector @xmath130^t $ ] . + for the case when @xmath131 , for a given vector @xmath132^t$ ] over @xmath133 , there are precisely 4 or 8 vectors ( including @xmath99 ) over @xmath133 that generate the same vector space over @xmath100 as @xmath99 . as given in section ii of @xcite , for the 4-psk constellation @xmath6, the difference constellation @xmath134 is of the form , @xmath135 where @xmath136 and @xmath137 .therefore , we can write , @xmath138 = \left [ { \begin{array}{cc } \vspace{0.15 cm } 2 sin \frac { \pi k_{1}}{4 } e^{j \phi_{1 } } \\ \vspace{0.15 cm } 2 sin \frac { \pi k_{2}}{4 } e^{j \phi_{2 } } \\ 0 \\ \end{array } } \right]\ ] ] where @xmath139 if @xmath140 is odd and @xmath141 if @xmath140 is even .+ a vector @xmath142 over @xmath133 shall generate the same vector space over @xmath100 iff @xmath142 is a scalar multiple of @xmath99 , i.e. for some complex number @xmath143 , @xmath144 = r e^{j\theta } \left [ { \begin{array}{cc } \vspace{0.15 cm } 2 sin \frac { \pi k_{3}}{4 } e^{j \phi_{3 } } \\\vspace{0.15 cm } 2 sin \frac { \pi k_{4}}{4 } e^{j \phi_{4 } } \\ 0 \\ \end{array } } \right]\ ] ] where for @xmath145 @xmath139 if @xmath140 is odd and @xmath141 if @xmath140 is even .+ then , @xmath146 and @xmath147 dividing ( [ firstcomp ] ) by ( [ secondcomp ] ) and taking modulus of both sides , we get @xmath148 as shown in @xcite , this is possible only if @xmath149 and @xmath150 . also , from ( [ firstcomp ] ) and ( [ secondcomp ] ) we have @xmath151 from ( [ rel ] ) and ( [ div ] ), we have @xmath152 note that here , the lhs is fixed .it therefore suffices to compute that for the fixed value of the lhs , the number of values that rhs takes .it can be verified , that for a fixed value of @xmath153 , there are precisely four pair of values of @xmath154 and @xmath155 that result in the same value of @xmath156 .we now look at the following possibilities : + _ case 1 : _ @xmath157 .then , @xmath158 , i.e. , there are exactly two possibilities for @xmath159 and @xmath160 , viz . , @xmath161 and @xmath162 . with two pairs of values for @xmath163 and @xmath164 and four pairs of values for @xmath154 and @xmath155, we have a total of eight set of values that @xmath142 can take .hence , in this case , the vector space generated by @xmath99 can be generated by exactly eight other vectors over @xmath133 ._ case 2 : _ @xmath165 .then , @xmath166 , i.e. , there is precisely one possibility for @xmath159 and @xmath160 , viz . , @xmath149 and @xmath167 . with only one possible set of values for @xmath163 and @xmath164 and four pairs of values for @xmath154 and @xmath155, we have a total of four set of values that @xmath142 can take .hence , in this case , the vector space generated by @xmath99 can be generated by exactly four other vectors over @xmath133 . in this casewe end up with 12 singular fade subspaces given by the null spaces of the space given on the next page in figure [ fig : sfscase2 ] . + @xmath168\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } -1-j \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } 1-j \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } -1+j \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2j \\ \vspace{0.15 cm } 2j \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2j \\ \vspace{0.15 cm } -2j \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2 \\ \vspace{0.15 cm } 2 \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2 \\ \vspace{0.15 cm } -2 \\ 0 \\ \end{array } } \right]\right\rangle $ ] + @xmath169\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } 1+j \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } 1-j \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } -1+j \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2j \\ \vspace{0.15 cm } -2j \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2j \\ \vspace{0.15 cm } 2j \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2 \\ \vspace{0.15 cm } -2 \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2 \\ \vspace{0.15 cm } 2 \\ 0 \\ \end{array } } \right]\right\rangle $ ] + @xmath170\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } -1+j \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } 1+j \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } -1-j \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2j \\ \vspace{0.15 cm } 2 \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2j \\ \vspace{0.15 cm } -2 \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2 \\ \vspace{0.15 cm } 2j \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2 \\ \vspace{0.15 cm } -2j \\ 0 \\ \end{array } } \right]\right\rangle $ ] + @xmath171\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } 1-j \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } -1-j \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } 1+j \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2j \\ \vspace{0.15 cm } -2 \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2j \\ \vspace{0.15 cm } 2 \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2 \\ \vspace{0.15 cm } 2j \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2 \\ \vspace{0.15 cm } -2j \\ 0 \\ \end{array } } \right]\right\rangle $ ] + @xmath172\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } -2j \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } 2 \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } -2 \\ 0 \\ \end{array } } \right]\right\rangle $ ] @xmath173\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } 2j \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } -2 \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } 2 \\ 0 \\ \end{array } } \right]\right\rangle $ ] + @xmath174\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } -2 \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } 2j \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } -2j \\ 0 \\ \end{array } } \right]\right\rangle $ ] @xmath175\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } 2 \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } -2j \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } 2j \\ 0 \\ \end{array } } \right]\right\rangle $ ] + @xmath176\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2j \\ \vspace{0.15 cm } -1-j \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2 \\ \vspace{0.15 cm } -1+j \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2 \\ \vspace{0.15 cm } 1-j \\ 0 \\ \end{array } } \right]\right\rangle $ ] @xmath177\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2j \\ \vspace{0.15 cm } -1-j \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2 \\ \vspace{0.15 cm } -1+j \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2 \\ \vspace{0.15 cm } 1-j \\ 0 \\ \end{array } } \right]\right\rangle $ ] + @xmath178\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2 \\ \vspace{0.15 cm } -1-j \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2j \\ \vspace{0.15 cm } -1+j \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2j \\ \vspace{0.15 cm } 1-j \\ 0 \\ \end{array } } \right]\right\rangle $ ] @xmath179\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2 \\ \vspace{0.15 cm } -1-j \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2j \\ \vspace{0.15 cm } -1+j \\ 0 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2j \\ \vspace{0.15 cm } 1-j \\ 0 \\ \end{array } } \right]\right\rangle .$ ] [ fig : sfscase2 ] so , @xmath126 12 singular fade subspaces for the subcase each being the null space of the above 12 spaces .similarly , for each of the other two subcases , there are 12 singular fade subspaces , resulting in a total of 36 singular fade subspaces for the case .+ + * _ case 3 : _ * in this case , @xmath112 .the singular fade subspace in this case is given by @xmath180\right\rangle ^{\bot}.\ ] ] here each of @xmath181 can take eight non - zero values .there are therefore , 512 total possibilities for the vector @xmath182^t $ ] . in this casewe end up with 112 singular fade subspaces .we now explain this . + for the case when @xmath183 , for a given vector @xmath184^t$ ] over @xmath133 , there are precisely 4 or 8 vectors ( including @xmath99 ) over @xmath133 that generate the same vector space over @xmath100 as @xmath99 . as mentioned in the proof of lemma 1 , from @xcite , @xmath185 where @xmath136 and @xmath137 .therefore , we can write , @xmath186 = \left [ { \begin{array}{cc } \vspace{0.15 cm } 2 sin \frac { \pi k_{1}}{4 } e^{j \phi_{1 } } \\ \vspace{0.15 cm } 2 sin \frac { \pi k_{2}}{4 } e^{j \phi_{2 } } \\ 2 sin \frac { \pi k_{3}}{4 } e^{j \phi_{3 } } \\\end{array } } \right]\ ] ] where @xmath139 if @xmath140 is odd and @xmath141 if @xmath140 is even .+ a vector @xmath142 over @xmath133 shall generate the same vector space over @xmath100 iff @xmath142 is a scalar multiple of @xmath99 , i.e. for some complex number @xmath143 , @xmath187 = r e^{j\theta } \left [ { \begin{array}{cc } \vspace{0.15 cm } 2 sin \frac { \pi k_{4}}{4 } e^{j \phi_{4 } } \\\vspace{0.15 cm } 2 sin \frac { \pi k_{5}}{4 } e^{j \phi_{5 } } \\ 2 sin \frac { \pi k_{6}}{4 } e^{j \phi_{6 } } \\ \end{array } } \right]\ ] ] where for @xmath188 @xmath139 if @xmath140 is odd and @xmath141 if @xmath140 is even .+ then , @xmath189 @xmath190 and , @xmath191 dividing ( [ firstcomp2 ] ) by ( [ secondcomp2 ] ) and taking modulus of both sides , we get @xmath192 as shown in @xcite , this is possible only if @xmath167 and @xmath193 . similarly , dividing ( [ firstcomp2 ] ) by ( [ thirdcomp2 ] ) and taking modulus of both sides ,we get @xmath194 as shown in @xcite , this is possible only if @xmath167 and @xmath195 . also , from ( [ firstcomp2 ] ) and ( [ secondcomp2 ] ) we have @xmath196 from ( [ rel2 ] ) and ( [ div2 ] ) , we have @xmath197 similarly , @xmath198 in ( [ div3 ] ) and ( [ div4 ] ) , the lhs is fixed. it therefore suffices to compute the number of values that rhs takes for fixed lhs in the two equations .it can be verified , that for a fixed value of @xmath199 , there are precisely four set of values of @xmath200 that result in the same value of @xmath153 and @xmath201 .we now look at the following possibilities : + _ case 1 : _ @xmath202 .then , @xmath203 , i.e. , there are exactly two possibilities for @xmath164 , @xmath204 and @xmath205 , viz . , @xmath206 and @xmath207 . with two sets of values for @xmath164 , @xmath204 and @xmath205 and four sets of values for @xmath208 , @xmath209 and @xmath210, we have a total of eight set of values that @xmath142 can take .hence , in this case , the vector space generated by @xmath99 can be generated by exactly eight other vectors over @xmath133 ._ case 2 : _ atleast one of @xmath165 , @xmath211 , or @xmath212 .then , @xmath213 , @xmath214 , or @xmath215 , so that , using ( [ rel2 ] ) and ( [ rel3 ] ) , there is precisely one possibility for @xmath216 , viz . , @xmath217 , @xmath218 and @xmath219 . with only one possible set of values for @xmath220 and @xmath205 and four sets of values for @xmath221 and @xmath222, we have a total of four set of values that @xmath142 can take .hence , in this case , the vector space generated by @xmath99 can be generated by exactly four other vectors over @xmath133 .+ since all of @xmath181 are non - zero , we can say that @xmath223 as a result , we have the following three subcases : 1 .one of @xmath224 2 .two of @xmath224 3 .all of @xmath224 we deal with each one of the subcases one - by - one .+ + _ subcase 1 : _ one of @xmath224 . without loss of generality, we assume that @xmath225 and @xmath226 . the singular fade subspace for the case is given by ( [ scase3 ] ) .there are 64 possibilities for the vector @xmath227^{t}$ ] . but some of the possibilities may generate the same vector space over @xmath100 .there are precisely 4 vectors of length 3 over @xmath228 , the @xmath229 scalar multiples of the vector . as a result , the case @xmath225 and @xmath226 leads to 16 singular fade subspaces as shown in figure [ fig : sfssubcase1 ] . the same holds for the case when @xmath230 and @xmath231 , or when @xmath232 and @xmath233 .this subcase therefore results in 48 singular fade subspaces . + @xmath234\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } -2j \\ -2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } -2 \\ -2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } 2 \\ 2 \\ \end{array } } \right]\right\rangle $ ] @xmath235\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } -2j \\ 2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } -2 \\ 2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } 2 \\ -2 \\ \end{array } } \right]\right\rangle $ ] + @xmath236\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } -2j \\ -2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } -2 \\ 2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } 2 \\ -2j \\ \end{array } } \right]\right\rangle $ ] @xmath237\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } -2j \\ 2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } -2 \\ -2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } 2 \\ 2j \\ \end{array } } \right]\right\rangle $ ] + @xmath238\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } 2j \\ -2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } 2 \\ -2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } -2 \\ 2 \\ \end{array } } \right]\right\rangle $ ] @xmath239\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm }2j \\ 2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } 2 \\ 2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } -2 \\ -2 \\ \end{array } } \right]\right\rangle $ ] + @xmath240\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } 2j \\ -2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } 2 \\ 2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } -2 \\ -2j \\ \end{array } } \right]\right\rangle $ ] @xmath241\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } 2j \\ 2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } 2 \\ -2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } -2 \\2j \\ \end{array } } \right]\right\rangle .$ ] + @xmath242\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } -2 \\ -2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } 2j \\ -2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } -2j \\ 2 \\ \end{array } } \right]\right\rangle $ ] @xmath243\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } -2 \\ 2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } 2j \\ 2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } -2j \\ -2 \\ \end{array } } \right]\right\rangle $ ] + @xmath244\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } -2 \\ -2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } 2j \\ 2j \\ \end{array } }\right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } -2j \\ -2j \\ \end{array } } \right]\right\rangle $ ] @xmath245\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } -2 \\ 2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } 2j \\ -2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } -2j \\ 2j \\ \end{array } }\right]\right\rangle $ ] + @xmath246\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } 2 \\ -2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } -2j \\ -2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } 2j \\ 2 \\ \end{array } } \right]\right\rangle $ ] @xmath247\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } 2 \\ -2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } -2j \\ -2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } 2j \\ 2 \\ \end{array } } \right]\right\rangle $ ] + @xmath248\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } 2 \\ -2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } -2j \\ 2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } 2j \\ -2j \\ \end{array } } \right]\right\rangle $ ] @xmath249\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } 2 \\ 2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } -2j \\ -2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } 2j \\ 2j \\ \end{array } } \right]\right\rangle .$ ] [ fig : sfssubcase1 ] @xmath250\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } -1-j \\ -2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } -1+j \\ -2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } 1-j \\ 2 \\ \end{array } } \right]\right\rangle $ ] @xmath251\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } -1-j \\ 2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } -1+j \\ 2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15cm } 1-j \\ -2 \\ \end{array } } \right]\right\rangle $ ] + @xmath252\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } -1-j \\ -2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } -1+j \\ 2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } 1-j \\ -2j \\ \end{array } } \right]\right\rangle $ ] @xmath253\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } -1-j \\ 2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } -1+j \\ -2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } 1-j \\ 2j \\ \end{array } } \right]\right\rangle $ ] + @xmath254\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } 1+j \\ -2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } 1-j \\ -2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } -1+j \\ 2 \\ \end{array } } \right]\right\rangle $ ] @xmath255\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } 1+j \\ 2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } 1-j \\ 2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } -1+j \\ -2 \\ \end{array } } \right]\right\rangle $ ] + @xmath256\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } 1+j \\ -2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } 1-j \\ 2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } -1+j \\ -2j \\ \end{array } } \right]\right\rangle $ ] @xmath257\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } 1+j \\ 2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } 1-j \\ -2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } -1+j \\ 2j \\ \end{array } } \right]\right\rangle .$ ] + @xmath258\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } -1+j \\ -2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } 1+j \\ -2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } -1-j\\ 2 \\ \end{array } } \right]\right\rangle $ ] @xmath259\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } -1+j \\ 2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } 1+j \\ 2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } -1-j\\ -2 \\ \end{array } } \right]\right\rangle $ ] + @xmath260\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } -1+j \\ -2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm }1+j \\ 2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } -1-j \\ -2j \\ \end{array } } \right]\right\rangle $ ] @xmath261\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } -1+j \\ 2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } 1+j \\ -2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } -1-j \\ 2j \\ \end{array } }\right]\right\rangle $ ] + @xmath262\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } 1-j \\ -2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } -1-j \\ -2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } 1+j \\ 2 \\ \end{array } } \right]\right\rangle $ ] @xmath263\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } 1-j \\ 2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } -1-j \\ 2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } 1+j \\ -2 \\ \end{array } } \right]\right\rangle $ ] + @xmath264\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } 1-j \\ -2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } -1-j \\ 2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } 1+j \\ -2j \\ \end{array } } \right]\right\rangle $ ] @xmath265\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } 1-j \\ 2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } -1-j \\ -2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } 1+j \\ 2j \\ \end{array } } \right]\right\rangle .$ ] [ fig : sfssubcase2 ] @xmath266\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } -1-j \\ -1-j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } -1+j \\ -1+j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } 1-j \\ 1-j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2j \\ \vspace{0.15 cm } 2j \\ 2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2j \\ \vspace{0.15 cm } -2j \\ -2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2 \\ \vspace{0.15 cm } 2 \\ 2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2 \\ \vspace{0.15 cm } -2 \\ -2 \\ \end{array } } \right]\right\rangle $ ] + @xmath267\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } -1-j \\ 1+j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } -1+j \\ 1-j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } 1-j \\ -1+j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2j \\ \vspace{0.15 cm } 2j \\ -2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2j \\ \vspace{0.15 cm } -2j \\ 2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2 \\ \vspace{0.15 cm } 2 \\ -2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2 \\ \vspace{0.15 cm } -2 \\ 2 \\ \end{array } } \right]\right\rangle $ ] + @xmath268\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } -1-j \\ -1+j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } -1+j \\ 1+j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } 1-j \\ -1-j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2j \\ \vspace{0.15 cm } 2j \\ 2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2j \\ \vspace{0.15 cm } -2j \\ -2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2 \\ \vspace{0.15 cm } 2 \\ -2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2 \\ \vspace{0.15 cm } -2 \\ 2j \\ \end{array } } \right]\right\rangle $ ] + @xmath269\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } -1-j \\ 1-j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } -1+j \\ -1-j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } 1-j \\ 1+j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2j \\ \vspace{0.15 cm } 2j \\ -2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2j \\ \vspace{0.15 cm } -2j \\ 2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2 \\ \vspace{0.15 cm } 2 \\ 2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2 \\ \vspace{0.15 cm } -2 \\ -2j \\ \end{array } } \right]\right\rangle $ ] + @xmath270\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } 1+j \\ -1+j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } 1-j \\ -1+j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } -1+j \\ 1-j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2j \\ \vspace{0.15 cm } -2j \\ 2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2j \\ \vspace{0.15 cm } 2j \\ -2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2 \\ \vspace{0.15 cm } -2 \\ 2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2 \\ \vspace{0.15 cm } 2 \\ -2 \\ \end{array } } \right]\right\rangle $ ] + @xmath271\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } 1+j \\ 1+j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } 1-j \\ 1-j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } -1+j \\ -1+j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2j \\ \vspace{0.15 cm } -2j \\ -2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2j \\ \vspace{0.15 cm } 2j \\ 2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2 \\ \vspace{0.15 cm } -2 \\ -2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2 \\ \vspace{0.15 cm } 2 \\ 2 \\ \end{array } } \right]\right\rangle $ ] + @xmath272\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } 1+j \\ 1-j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } 1-j \\ -1-j \\ \end{array } } \right]\right\rangle = \left\langle \left[ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } -1+j \\ 1+j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2j \\ \vspace{0.15 cm } -2j \\ -2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2j \\ \vspace{0.15 cm } 2j \\ 2 \\ \end{array } } \right]\right\rangle= \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2 \\ \vspace{0.15 cm } -2 \\ 2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2 \\ \vspace{0.15 cm } 2 \\ -2j \\ \end{array } } \right]\right\rangle $ ] + @xmath273\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } 1+j \\ -1+j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } 1-j \\ 1+j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } -1+j \\ -1-j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2j \\ \vspace{0.15 cm } -2j \\ 2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2j \\ \vspace{0.15 cm } 2j \\ -2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2 \\ \vspace{0.15 cm } -2 \\ -2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2 \\ \vspace{0.15 cm } 2 \\ 2j \\ \end{array } } \right]\right\rangle $ ] + @xmath274\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } -1+j \\ -1-j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } 1+j \\ -1+j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } -1-j \\ 1-j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2j \\ \vspace{0.15 cm } 2 \\ 2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2j \\ \vspace{0.15 cm } -2 \\ -2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2 \\ \vspace{0.15 cm } -2j \\ 2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2 \\ \vspace{0.15 cm } -2 \\ 2j \\ \end{array } } \right]\right\rangle $ ] + @xmath275\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } -1+j \\ 1+j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } 1+j \\ 1-j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } -1-j \\ -1+j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2j \\ \vspace{0.15 cm } 2 \\ -2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2j \\ \vspace{0.15 cm } -2 \\ 2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2 \\ \vspace{0.15 cm } -2j \\ -2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2 \\ \vspace{0.15 cm } 2j \\ 2 \\ \end{array } } \right]\right\rangle $ ] + @xmath276\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } -1+j \\ -1+j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } 1+j \\ 1+j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } -1-j \\ -1-j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2j \\ \vspace{0.15 cm } 2 \\ 2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2j \\ \vspace{0.15 cm } -2 \\ -2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2 \\ \vspace{0.15 cm } -2j \\ -2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2 \\ \vspace{0.15 cm } 2j \\ 2j \\ \end{array } } \right]\right\rangle $ ] + @xmath277\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } -1+j \\ 1-j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } 1+j \\ -1-j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } -1-j \\ 1+j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2j \\ \vspace{0.15 cm } 2 \\ -2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2j \\ \vspace{0.15 cm } -2 \\ 2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2 \\ \vspace{0.15 cm } -2j \\ 2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2 \\ \vspace{0.15 cm } 2j \\ -2j \\ \end{array } } \right]\right\rangle $ ] + @xmath278\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } 1-j \\ -1-j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } -1-j \\ -1+j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } 1+j \\ 1-j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2j \\ \vspace{0.15 cm } -2 \\ 2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2j \\ \vspace{0.15 cm } 2 \\ -2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2 \\ \vspace{0.15 cm } 2j \\ 2 \\ \end{array } }\right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2 \\ \vspace{0.15 cm } -2j \\ -2 \\ \end{array } } \right]\right\rangle $ ] + @xmath279\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } 1-j \\ 1+j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } -1-j \\ 1-j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } 1+j \\ -1+j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2j \\ \vspace{0.15 cm } -2 \\ -2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2j \\ \vspace{0.15 cm } 2 \\ 2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2 \\ \vspace{0.15 cm } 2j \\ -2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2 \\ \vspace{0.15 cm } -2j \\ 2 \\ \end{array } } \right]\right\rangle $ ] + @xmath280\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } 1-j \\ -1+j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } -1-j \\ 1+j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } 1+j \\ -1-j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2j \\ \vspace{0.15 cm } -2 \\ 2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2j \\ \vspace{0.15 cm } 2 \\ -2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2 \\ \vspace{0.15 cm } 2j \\ -2j \\ \end{array } } \right]\right\rangle = \left\langle \left[ { \begin{array}{cc } \vspace{0.15 cm } -2 \\ \vspace{0.15 cm } -2j \\ 2j \\ \end{array } } \right]\right\rangle $ ] + @xmath281\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } 1-j \\ 1-j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } -1-j \\ -1-j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } 1+j \\ 1+j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2j \\ \vspace{0.15 cm } -2 \\ -2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2j \\ \vspace{0.15 cm } 2 \\ 2 \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2 \\ \vspace{0.15 cm } 2j \\ 2j \\ \end{array } } \right]\right\rangle = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2 \\ \vspace{0.15 cm } -2j\\ -2j \\ \end{array } } \right]\right\rangle $ ] . [fig : sfssubcase3 ] _ subcase 2 : _ two of @xmath224 . without loss of generality , we assume that @xmath282 and @xmath283. the singular fade subspace for the case is given by ( [ scase3 ] ) and again there are 64 possibilities for the vector @xmath284^{t}$ ] . for a given @xmath285 ,possibilities of other 3 length vectors over @xmath286 that generate the same vector space over @xmath100 are the @xmath229 scalar multiples of @xmath285 . as a result , the case @xmath225 and @xmath226 also leads to 16 singular fade subspaces as shown for this case in figure ( [ fig : sfssubcase2 ] ) .the same holds for the case when @xmath287 and @xmath288 , or when @xmath289 and @xmath290 resulting in therefore 48 singular fade subspaces .+ + _ subcase 3 : _ all of @xmath224 .there are 64 possibilities for the vector @xmath291^{t}$ ] over @xmath292 . for a given such vector , possibilities of other 3 length vectors over @xmath286 that generate the same vector space over @xmath100are the @xmath293 scalar multiples of the vector .the case @xmath294 leads to a total of 16 singular fade subspaces as shown in figure ( [ fig : sfssubcase3 ] ) .+ the three cases result in a total of 3 + 36 + 48 + 48 + 16=151 singular fade subspaces .we now discuss how these singular fade subspaces can be removed using latin cubes of second order .in the previous section , we classify the set of singular fade subspaces into three cases .we now cluster the possibilities of @xmath295 into a clustering using latin cubes .this clustering is represented by a constellation given by @xmath296 , which is utilized by the relay node r in the bc phase .the objective is to minimize the size of this constellation used by r. + the clustering to be used at r , first constrains the possibilities of @xmath295 received at the ma phase , with the objective of removing the singular fade subspaces , and fills the entries of a @xmath297 array representing the map to be used at the relay using these constraints , and then completes the partially empty array obtained to form a latin cube of second order .the mapping to be used at r can be obtained from the complete latin cube keeping in mind the equivalence between the relay map with the latin cube of second order as shown in section iii .+ in order to to obtain the constraints on the @xmath74 array representing the map at the relay node r during bc phase for a singular fade state , we utilize the vectors of differences , viz . , @xmath298 $ ] contributing to that particular singular fade state . during ma phase for the three - way relaying scenario ,nodes a , b and c transmit to the relay r. as shown in the previous section , there is a total of 151 singular fade subspaces .let @xmath299 denote a point in one of the 151 singular fade subspaces .then , there exist @xmath300 that satisfy @xmath301 . in order to remove the singular fade state @xmath299 ,the pair @xmath302 must be kept in the same cluster in the clustering .for instance , if @xmath303\right\rangle ^{\bot},\ ] ] then the pair @xmath302 must be kept in the same cluster in the clustering , i.e. , the entry corresponding to @xmath304 in the @xmath74 array must be the same as the entry corresponding to @xmath305 .similarly , every other such pair in @xmath306 contributing to the same singular fade subspace must be kept in the same cluster .apart from all such pairs in @xmath306 being kept in the same cluster of the clustering , in order to remove this particular fade state , there are no other constraints . upon filling up the @xmath74 array with these constraints ,the entire latin cube does not get filled up , but can be completed as we show later in this section .it is important to note that , this clustering can not be utilized to remove the singular fade subspaces of case 1 and case 2 of the previous section , as shown in the following lemma .+ the clustering map used at the relay node r can not remove a singular fade state corresponding to the case 1 and case 2 of the previous section and simultaneously satisfy the mutually exclusive law .let @xmath307\right\rangle ^{\bot } $ ] be a singular fade state for case 1 , and without loss of generality , assume that @xmath308 .then , in order to remove @xmath296 , @xmath93 and @xmath309 must be kept in the same cluster .this is impossible , since this clearly violates the mutually exclusive law as if the pair is placed in the same cluster , the users b and c will not be able to distinguish between the messages @xmath78 and @xmath310 sent by the user a. + let @xmath296 be a singular fade state for case 2 , and without loss of generality , assume that @xmath311 . then , in order to remove @xmath296 , @xmath93 and @xmath312 must be kept in the same cluster .this also clearly violates the mutually exclusive law since if the pair is placed in the same cluster , the user c will not be able to distinguish between the messages @xmath313 and messages @xmath314 sent by the users a and b. the singular fade subspaces given in case 1 ans case 2 of the previous section , whose harmful effects can not be removed by a proper choice of the clustering are referred to as the _ non - removable singular fade subspaces_. the rest of the singular fade subspaces , given in case 3 of the previous section , are referred as the _ removable singular fade subspaces_. we now illustrate the removal of a case 3 singular fade state with the help of examples .+ let the singular fade subspace be @xmath315\right\rangle ^{\bot } = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } -2j \\ 2j \\ \end{array } } \right]\right\rangle ^{\bot }= \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } -2 \\ 2 \\ \end{array } } \right]\right\rangle ^{\bot } = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } 2 \\ -2 \\ \end{array } } \right]\right\rangle ^{\bot}\ ] ] consider the first vector @xmath316^t$ ] . here, @xmath124 can be obtained as a difference of @xmath317 and @xmath318 or as a difference of @xmath319 and @xmath320 ; @xmath321 can be obtained as a difference of @xmath322 and @xmath323 ; and @xmath324 can be obtained as a difference of @xmath325 and @xmath326 .thus , the entries corresponding to @xmath327 and @xmath328 must be the same and the entries corresponding to @xmath329 and @xmath330 must be the same in the @xmath74 array representing the clustering , i.e. , entries @xmath331 and @xmath332 must be the same and entries @xmath333 and @xmath334 must be the same . + the second vector @xmath335^t$ ] where , @xmath336 can be obtained as a difference of @xmath337 and @xmath338 or as a difference of @xmath339 and @xmath340 ; @xmath324 can be obtained as a difference of @xmath341 and @xmath342 ; and @xmath321 can be obtained as a difference of @xmath343 and @xmath344 .thus , the entries corresponding to @xmath330 and @xmath329 must be the same and the entries corresponding to @xmath328 and @xmath327 must be the same in the @xmath74 array representing the clustering , i.e. , entries @xmath334 and @xmath333 must be the same and entries @xmath332 and @xmath331 must be the same . similarly , the constraints resulting from the third and forth vector ca be obtained .the constrained array is shown in figure 8 .+ we choose to fill the constrained array to form a latin cube of second order as shown in adjoining algorithm 1 , which simply fills the empty cells of the partially filled @xmath74 array with @xmath345 in increasing order of @xmath346 keeping in mind that the resulting array must be a latin cube of second order .the top - most and the left - most empty cell in the earliest file is filled at every iteration .the completed latin cube is shown in figure 9 .+ [ cols="^,^,^,^,^",options="header " , ] consider another example for which the singular fade subspace is given by @xmath347\right\rangle ^{\bot } = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1-j \\ \vspace{0.15 cm } -1-j \\ 1+j \\ \end{array } } \right]\right\rangle ^{\bot } = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -1+j \\ \vspace{0.15 cm } -1+j \\ 1-j \\ \end{array } } \right]\right\rangle ^{\bot } = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 1-j \\ \vspace{0.15 cm } 1-j \\ -1+j \\ \end{array } } \right]\right\rangle ^{\bot}\ ] ] @xmath348\right\rangle ^{\bot } = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2j \\ \vspace{0.15 cm } -2j \\ 2j \\ \end{array } } \right]\right\rangle ^{\bot } = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } -2 \\ \vspace{0.15 cm } -2 \\ 2 \\ \end{array } } \right]\right\rangle ^{\bot } = \left\langle \left [ { \begin{array}{cc } \vspace{0.15 cm } 2 \\ \vspace{0.15 cm } 2 \\ -2 \\ \end{array } } \right]\right\rangle ^{\bot}.\ ] ] the first vector is @xmath349 $ ] . here , @xmath124 can be obtained as a difference of @xmath317 and @xmath318 or as a difference of @xmath319 and @xmath320 ; @xmath336 can be obtained as a difference of @xmath350 and @xmath326 or as a difference of @xmath325 and @xmath351 .thus , the entries corresponding to @xmath352 , @xmath353,@xmath354,@xmath355 , @xmath356,@xmath357,@xmath358 , @xmath359 must be the same in the @xmath74 array representing the clustering , i.e. , entries @xmath360 , @xmath361 , @xmath362 , @xmath363 , @xmath364 , @xmath365 , @xmath366 , @xmath367 must be the same .similarly the other constraints can be obtained .the constrained array is shown in figure [ fig : ex2fig1 ] , and the clustering is as shown in figure [ fig : ex2fig2 ] .+ for each of the 112 possibilities of singular fade subspaces of _ case 3 _ of the previous section , a clustering of size between 16 to 23 can be achieved by first constraining the array representing the relay map in order to remove the singular fade state and then completing the constrained array using the provided algorithm .the proposed scheme is based on the removal of singular fade states for the three - way relaying sccenario .a minimum cluster distance greater than zero is ensured for all the fade states , excluding a subset of singular fade states , which are shown to be non - removable .it is attempted to ensure that in the given scenario , the number of clusters in the clustering , which is the same as the size of the signal set used during the bc phase is minimized .simulation results presented in this section identify some cases where the proposed scheme outperforms the naive approach that uses the same map for all fade states and vice verse all the simulation results shown in this section are for the case when the end nodes use 4-psk signal set .the simulation results for the end to end ber as a function of snr are presented in this section for different fading scenarios .+ consider the case when @xmath368 and @xmath369 are distributed according to rician distribution and channel variances equal to 0 db .the snr vs ber curve for this case , for a frame length of 256 bits .the plots for the cases with a rician factors of 5 db , 10 db , 15 db and 20 db are as shown in fig .13 , fig . 14 , fig . 15 and16 respectively .the figures show the snr vs bit - error - rate curves for the following schemes : adaptive clustering given in the paper , and non - adaptive clustering . for non - adaptive clustering, the relay node uses the same map given by figure [ fig : xor ] for all the channel realisations , we refer to this map as the non - adaptive map .it can be seen from fig .[ fig : plot_bc_rician ] that the schemes based on the adaptive clustering relaying perform better than the schemes based on non - adaptive clustering at low snr , since adaptive clustering removes 112 singular fade states .+ it can be seen from the simulation results presented that the non - adaptive network coding performs better than the proposed scheme above a certain snr when there is a dominant line of sight component , as in the case of rician fading scenario .the reason for this is as follows : the end to end snr vs ber as well as the throughput performance depend on the performance during the ma phase as well as the bc phase .as the line of sight component becomes more and more predominant , the performance during the bc phase gets better and better , but the effect of multiple access interference which occurs during the ma phase remains the same .hence , for the cases when line of sight component is predominant , the performance degradation due to the ma interference predominates over the degradation occurring during the bc phase .the case of non - adaptive network coding , the relay utilizes a constellation of least possible size , whereas in adaptive network coding , the relay attempts at removing singular fade states , thereby optimizing the performance during ma phase to the fullest extent possible , at the cost of degraded performance during the bc phase .hence , the non - adaptive network coding performs better than the proposed scheme at high snr . +our paper deals with the three - way wireless relaying scenario , assuming that the three nodes operate in half - duplex mode and that they transmit points from the same 4-psk constellation .it is shown that it is possible for information exchange to take place using just two channels uses , unlike the other work done for the case , to the best of our knowledge .the relay node clusters the @xmath30 possible transmitted tuples @xmath295 into various clusters such that _ the exclusive law _ is satisfied .this necessary requirement of satisfying the exclusive law is shown to be the same as the clustering being represented by a latin cube of second order . using the proposed schemes , not only is the exchange of information between the three nodes made possible using three channel uses , the size of the resulting constellation used by the relay node r in the bc phase is reduced from @xmath30 to lie between 16 to 23 .note that we do not claim that the size of the clustering utilizing modified clustering is the best that can be achieved , since our method of filling the latin cube of second order of side 4 may not be the most optimal process of doing so , and it might be possible to fill the array with less than 23 symbols .this work was supported partly by the drdo - iisc program on advanced research in mathematical engineering through a research grant as well as the inae chair professorship grant to b. s. rajan .+ t. koike - akino , p. popovski and v. tarokh , `` optimized constellation for two - way wireless relaying with physical network coding '' , ieee journal on selected areas in comm . ,vol.27 , pp . 773787 , june 2009 .mylene pischella and didier le ruyet , `` lattice based coding scheme for mimo bi - directional relaying with three nodes '' , 22nd ieee personal indoor mobile radio communications , toronto , canada , pp . 14591463 , sept . 2011 .youngil jeon , young - tae kim , moonseo park , inkyu lee , `` opportunistic scheduling for three - way relay systems with physical layer network coding '' , vehicular technology conference ( vtc spring ) 2011 , budapest , hungary , ieee 73rd , pp .15 , may 2011 .vijayvaradharaj t muralidharan , vishnu namboodiri and b. sundar rajan , `` channel quantization for physical - layer network coded two - way relaying '' , available online at arxiv:1109.6101v1 [ cs.it ] , 28 sept .

<S> the design of modulation schemes for the physical layer network - coded three - way wireless relaying scenario is considered . </S><S> the protocol employs two phases : multiple access ( ma ) phase and broadcast ( bc ) phase with each phase utilizing one channel use . for the two - way relaying scenario , </S><S> it was observed by koike - akino et al . </S><S> @xcite , that adaptively changing the network coding map used at the relay according to the channel conditions greatly reduces the impact of multiple access interference which occurs at the relay during the ma phase and all these network coding maps should satisfy a requirement called _ exclusive law_. this paper does the equivalent for the three - way relaying scenario . </S><S> we show that when the three users transmit points from the same 4-psk constellation , every such network coding map that satisfies the exclusive law can be represented by a latin cube of second order . </S><S> the network code map used by the relay for the bc phase is explicitly obtained and is aimed at reducing the effect of interference at the ma stage . </S>

astro-ph9905093

there is a bewildering array of agn classes .surprisingly , it s only in the last few years that we ve realized that much of this diversity is simply the result of viewing an axisymmetric central structure from different angles .synchrotron emission from the relativistic jets is much brighter when viewed along the jet direction ( the central engine s axis ) , and a torus of dust obscures a low - latitude view of the center .while the real physics is buried in the center , along with clues to the enormous range of agn luminosities , the mysterious cause of strong relationships among x - ray continuum and emission lines and the apparent radio - loud radio - quiet dichotomy , ` orientation ' unification schemes are not simply geometrical complications , but lead us to new ways of probing the central engine and its interaction with the surrounding space .this overview will emphasize qsos , the highest luminosity agn .first , and most importantly , we discuss the unification of radio - loud agn ( 2 ) .this is of prime importance because the axis of the central engine can be defined by the innermost radio jets .the strongest arguments for unification ( axisymmetry ) of properties at ir through x - ray wavebands can be made when observational properties can be referred to this axis , even for radio - weak ( radio - quiet ) agn . in 3 , we present an idealized picture that unifies a wealth of ir through x - ray properties , relating these to the radio axis . arguments in support of the general dusty torus picture are provided by a few specific examples ( 4 ) and by some statistical arguments ( 5 ) . an overview and speculationsare given in 6 .additional references are given in 7 .more detailed reviews are given by antonucci ( 1993 ) and , for radio - loud agn , by urry & padovani ( 1995 ) .the first recognized clues to unification came from the structure of radio sources because they can be spatially resolved , even on parsec scales for nearby agn .extragalactic radio sources often show double lobes separated by tens to hundreds of kpc , straddling a weak compact ` core ' coinciding with an optical galaxy or qso ( fig .1 ) . the emission mechanism is synchrotron radiation .the lobes have steep radio spectra ( f@xmath0 , @xmath1 ) so these lobe - dominated ( ld ) sources are most often selected by surveys at low radio frequencies ; the cores generally have flat radio spectra , the superposition of several peaked spectra from optically thick sub - components . in high - frequency radio surveys many compact , flat - spectrum sources have been discovered , often unresolved on kpc scales ( the core - dominated or cd sources ) , usually identified optically with qsos , occasionally with their almost lineless counterparts , the bl lac objects .the first clue to relativistically - beamed synchrotron emission came from the intensely bright , highly variable radio cores of these bl lac objects and the optically violently variable ( ovv ) qsos , with light travel sizes of light - hours implying impossibly high brightness temperatures exceeding the compton scattering limit .the second clue came from very long baseline interferometry ( vlbi ) that probed the compact radio sources on scales of milliarcseconds .blobs were ejected , apparently at superluminal speeds in the plane of the sky .as blobs travel towards the observer at relativistic speeds , @xmath20.7c , the emitted signal has less far to travel , that is , the blobs tend to ` catch up ' with their forward - emitted radiation ; thus , for a distant observer looking close to the direction of ejection , events appear to happen closer together in time , and the blob appears to travel at speeds up to 10c or more . under these conditions , the synchrotron radiation will be forward - boosted into an angle @xmath3 rad , , and the doppler factor @xmath4 . ] and with amplification or doppler boosting up to a factor @xmath5 ( n = 2 3 ) , simultaneously explaining the brilliant compact sources .the most general unification scheme for radio sources states that all cd sources are ld sources viewed close to the jet axis ( within 1520 of the line of sight . )this is illustrated in fig .1 . for a random distribution of orientationangles one expects to see @xmath6 ld sources for every cd source .this is more - or - less consistent with the statistics of radio sources , taking into account the frequency and flux - density limits of radio surveys ( urry & padovani 1995 ) .this is nice , but somewhat unexpected in the sense that it is not obvious why the electrons producing the doppler - boosted emission should have same @xmath7 as the flow pattern determined from apparent superluminal motion : first , the boosted radiation is dominated by those electrons that happen to be traveling in the observer s direction , so a spread in direction of the electron stream could result in an actual beam width much broader than @xmath8 ( lind & blandford 1985 ) .second , the pattern speed is probably an apparent speed , representing the phase velocity of shock fronts in the jet .several observations suggest that the pattern speed is in fact close to the speed of the doppler - boosting electrons , so we re in luck .the optically - thin emission from the lobes is isotropic , so what this means is that we have a measure of orientation of the central engine ( or inner jet ) , the core dominance , defined by : r = observed core flux - density / unbeamed lobe flux - density converted to rest - frame frequencies , often chosen as 5 ghz .this unified scheme predicted the following : + @xmath9 the jets should have very different brightnesses .the approaching jet should be brighter than the receding jet , by @xmath10^{n+\alpha}$ ] .in fact the fainter jet is rather faint even when it s thought that the jets are close to the sky plane .these jet asymmetries are observed out to several kpc ( in powerful 3cr sources , at least ) .the lobes are of similar brightnesses to each other because by the time the jet has reached them it has become much less relativistic , and environmental factors can dominate ( bridle et al . 1994 ; dennett - thorpe et al . 1997 ; wardle & aaron 1997 ) .+ @xmath9 the brighter jet is on the near side of the nucleus .this is beautifully demonstrated by radio faraday depolarization which is almost always greater for lobe emission on the side of the fainter ( invisible ) jet .recent observations have shown that the depolarization occurs in a foreground screen associated with the radio source , rather than being intrinsic to the lobes , so overcoming an earlier objection to this argument ( e.g. , leahy et al .1997 ; morganti et al .. + @xmath9 slight curvature of jets , when foreshortened as expected in cd sources , should appear to be much larger than in ld sources .this is observed ( hough & readhead 1989 ; cohen & unwin 1982 . )+ @xmath9 core - dominated sources , whose apparent brightness often exceeds the compton limit , should be seen as strong emitters of x - rays and @xmath7-rays , a prediction strikingly confirmed ( urry 1999 ) .+ @xmath9 apparent superluminal speeds should decrease with decreasing core dominance , and this is observed ( browne 1987 ; vermeulen & cohen 1994 ) . + @xmath9 the cd sources have relatively faint diffuse halos with structure , often double , and luminosity consistent with their being the extended doubles seen end - on . on this basis antonucci & ulvestad ( 1985 ) used the converse to argue that the radio source unified scheme must be basically correct : cd sources seen at larger angles to the jet must be seen as powerful steep spectrum radio sources dominated by extended , basically double - structured emission , and can account for most of the fraction of observed ld sources .+ @xmath9 there is an inverse relationship between r and projected linear size .barthel ( 1989 ) showed that the projected ( apparent ) radio linear sizes for 3crr qsos are smaller than for the lower - r frii radio galaxies , as predicted .a quantitative demonstration of this relationship has been thwarted by selection effects and probable linear size evolution ( gopal - krishna et al . 1996 ; neeser et al .1995 ) . while many radio sources show clear double - lobed structure ( e.g. , fig .1 ) , in real life structure can be more complex on both parsec and kiloparsec scales .about 20% of sources in low radio - frequency surveys consist of intrinsically compact double or single sources gigahertz peak spectrum ( gps ) and compact steep spectrum ( css ) sources ( urry & padovani 1995 ) .it is not yet clear whether unification works for these intrinsically compact sources . also , the ld sources come in two flavors , fri and frii ( fanaroff & riley 1974 ; owen et al .the lobes of the more luminous erg s@xmath11 hz@xmath11 , for h@xmath12 km s@xmath11 mpc@xmath11 . ] frii sources contain bright hotspots and are edge - brightened where they ram into the surrounding medium , whereas the less luminous fri lobes are relaxed and show neither of these features .the central regions of luminous agn are essentially unresolved at non - radio wavelengths , and their classification is based on optical spectra .optical spectra of type 1 are defined by their broad emission lines ( @xmath13 km / s fwhm ) , which originate in gas within 1 pc of a central supermassive black hole .they may also show narrow lines that arise from low density gas on scales of parsecs to kiloparsecs .type 1 spectra are accompanied by a broad optical - euv bump attributed to thermal emission with a range of temperatures , 10@xmath14 k to 10@xmath15 k , arising from an accretion disk that feeds the black hole ( shields 1978 ; malkan & sargeant 1983 ; mushotzsky 1997 ) .these observed properties define the qsos and their lower - luminosity cousins , the seyfert 1 nuclei .optical surveys for these objects usually depend on selection by blue or uv colors and strong broad emission lines . in radio - loud agnsit appears that type 1 spectra are found in frii radio sources , but almost never in the lower luminosity fr is ( falcke et al .1995 ) .optical spectra of type 2 are those seen to have narrow lines only , with a spectrum and velocity dispersion indistinguishable from the narrow lines of type 1 spectra . the optical continuum is dominated by starlight , and there is no convincing evidence for a optical - euv bump associated with narrow line spectra . when seen in the abscence of obscuration ,the thermal type 1 and type 2 emission lines and optical - euv bump are essentially unpolarized . in core - dominant radio - loud agns a steep - spectrum , polarized , smooth spectral energy distribution of an ir - optical - uv synchrotron component can contribute ( the blazars ) , in some cases overwhelming the emission lines and accretion continuum of a type 1 spectrum .the spectra of bl lac objects often appear featureless , and are always dominated by this synchrotron continuum , and may have no , or intrinsically only very weak broad lines .narrow emission and absorption lines may be present , as for any agn . fig .2 illustrates how agn can appear drastically different at different orientations .it shows ( not to scale ) a cross section through the dusty torus , whose shadow is shown in gray .a radio jet axis is shown aligned with the axis of the dusty torus .the broad lines arise from dense gas of the broad - line region ( blr ) and narrow emission lines from less - dense nlr gas .these are shown as filled and open circles in fig .2 . refer to fig . 2 in the remainder of this section .one can see the blr and accretion continuum directly from within the cone whose edge is defined by the shadow of the torus . within 15 to 25 of the jet axis the qsos are cd with r@xmath16 , and the strong variable synchrotron emission is revealed by a steep polarized infrared - uv continuum that may swamp the unpolarized blue bump and emission lines .these qsos are therefore often seen as optically violent variable ( ovv ) or highly - polarized ( variable ) qsos ( hpqs ) .beamed compton - scattered radiation from the jet contributes to and flattens the x - ray spectrum . at larger angles to the axis the qsos are ld ( r@xmath17 ) , and the beamed synchrotron emission is no longer detectable against the optical - euv bump . the broad h@xmath18 line and broad base of the civ@xmath191549 emission lines are broader than in cd qsos , perhaps indicating a disk - like configuration for the inner blr . at even larger inclinations of the jet axis , perhaps 40 to 50 , r decreases further , with increasing projected lobe separations ( barthel 1989 ) .broad lines may be very broad , up to 10,000 or even 20,000 km s@xmath11 in doppler width ( fwhm ) , and are usually reddened and weaker relative to the nlr emission ( grandi & osterbrock 1978 ; hill , goodrich & depoy 1996 ) .some of these reddened qsos and broad - lined radio galaxies ( blrg ) show weak , polarized broad - line and continuum spectra evidence for a scattered type i spectrum that is combined with the reddened view through the thinner regions of the torus . at even larger inclinations emission fromthe nlr completely dominates the spectrum ( fig .3 ) . these are the narrow - lined radio galaxies ( nlrg ) with their type 2 spectra .direct light from the blr and optical - euv continuum may be completely obscured , and even the higher - ionization nlr emission ( [ oiii]@xmath195007 ) may be partially obscured by the dusty torus .starlight completely dominates the near ir - optical continuum .the inner regions may be seen indirectly however , as a faint polarized spectrum that has been scattered from within the opening angle of the torus the ` ionization cone ' or ` scattering cone ' .even the scattered light may be absorbed as it grazes the torus or passes through the host galaxy .these ionization cones are sometimes resolved by hst and ground - based imaging .the `` 3@xmath20 m bump '' arises from warm dust near evaporation temperature ( @xmath21 k ) .this equilibrium temperature determines the distance of the inner torus from the central heat source , @xmath22 pc for luminous qsos of bolometric luminosity @xmath23 ergs s@xmath11 .this is present in most , if not all , rl qsos ( rlqs ) .does the above unified scheme also hold for the fri agn , which are less luminous ?most fri agn are ld sources and appear optically as radio galaxies .no fri radio galaxy had shown anything but weak narrow line emission until recently , when lara et al .( 1998 ) discovered an fri nucleus with broad emission lines .the bl lacs , which are always cd , are probably the ld fri radio galaxies seen at small angles to the radio jet .consistent with this , is their lack of , or only very weak , broad emission lines .the extended radio fuzz around bl lac cores is consistent in luminosity and structure , and probably in size as well , with the extended radio emission of double - lobed fri radio galaxies .the evidence for relativistic jets in bl lac objects has already been mentioned : ( i ) from their high apparent brightness temperature and rapid variability .their radio - ir - optical - uv continua are smoothly connected , linearly polarized and highly variable , as expected for a beamed synchrotron origin .the bl lacs are so far the only agn detected at tev ( @xmath7-ray ) photon energies attributable to beamed radiation from the relativistic jet , and their lower redshifts ( smaller distances ) than most core - dominant rlqs .( ii ) the jets are also seen in vlbi images . while the extended lobes of fri radio galaxies are relaxed , with no edge brightening or hotspots , their inner radio jets are indistinguishable from those of frii sources , and apparent superluminal motion has been measured ( giovannini et al .1998 ) .radio - quiet ( rq ) agn include both the rq qsos ( rqqs ) , which are of course optically luminous , and the seyfert galaxies , which are less luminous .their radio emission is weak , though detectable . for nearby seyfert 1s and 2s ,ionization - scattering cones can be related to their radio jet axes , and their unification by orientation is quite well established both statistically , and for many individual seyfert 2s where a buried seyfert 1 has been revealed in polarized light ( 4 & 5 ) .this does not preclude the existence of a small fraction of genuine seyfert 2s where the blr is intrinsically weak , perhaps temporarily ( e.g. , n4151 , penston & perez 1984 ) . does this unified scheme hold for radio - quiet qsos ?kukula et al . (1998 ) argue that the radio - quiet qsos form the high radio and optical luminosity extension of the seyfert 1 luminosity functions , which is suggestive that it does . while it has been shown that there is patchy covering by absorbing and scattering dust in radio - quiet qsos ( 6 ) , the dust geometry has not been clearly related to the axis of the central engine .also , no large , unbiased survey has been made for qso 2s whatever they may be expected to look like .the qso radio luminosity function appears bimodal ( hooper et al . 1995 ) so a distinction has been made between radio - loud qsos ( rlq ) and radio - quiet qsos ( rqq ) .a somewhat arbitrary definition often used to discriminate rqq from rlq is that rqq have a rest - frame flux - density ratio f@xmath24(5 ghz)/f@xmath24(b - band ) @xmath25 .results from deep radio surveys ( _ first _ , white et al .1999 preprint ) will show whether rqq and rlq are really separate populations ; the lower the survey frequency the better , to avoid the complications of beamed radio emission .recent surveys of the radio properties of radio - quiet qsos ( e.g. , falcke , wilson , & ho 1997 ; kukula et al .1998 ; blundell & beasley 1998a ; kellermann et al .1994 , 1998 ) show strong evidence for jet - producing central engines in many .falcke et al .( 1996a , b ) make a strong case that a few qsos with radio flux densities intermediate between radio quiet and radio loud are naturally explained as doppler - boosted radio - quiet qsos , the radio - quiet equivalent of cd rlqs . for these they derived similar @xmath7(doppler boosting ) .this is supported by the discovery of superluminal motion in at least one radio - quiet qso ( blundell & beasley 1998b ) .these new results are important for the interpretation of optical properties in terms of an axisymmetric central engine .however , it must be appreciated that only the least radio quiet of the radio - quiet qsos can be investigated for jet structure and superluminal motion .an important subclass of qsos are the broad absorption line qsos ( bal qsos ) .these objects show broad absorption features , blueshifted from a few thousand km s@xmath11 up to 45,000 km s@xmath11 or even more , from the systemic redshift .the observed radial terminal velocity of these bals is strongly anticorrelated with radio power ( weymann 1997 ) .the soft x - rays are completely absorbed in these objects ( green & mathur 1996 ) .these absorption features provide evidence for high - ionization and dusty , low - ionization nuclear outflows ( see 4 below ) .three main results have led to the idea that all rqqs have bal outflows : + ( i ) in individual bal qsos , from the limits on scattered light in the deep troughs , one can deduce that the absorbing gas covers @xmath2610% of the central continuum .+ ( ii ) the bal qsos and the non - bal qsos have very similar uv emission line spectra ( weymann et al . 1991 ) .+ ( iii ) bals are observed in @xmath2610% of all rqqs .+ can this ` orientation unification scheme ' be related to the above scheme for radio - loud qsos ? unfortunately ,most radio - quiet bal qsos have only weak , unresolved radio emission , too weak for present observations at milliarcsecond ( vlbi ) resolutions .so no jet axis can be determined .nevertheless , the radio structure of radio - loud bal qsos may soon provide important clues to orientation and radio - loud radio - quiet differences. there may be other clues to axisymmetry and inclination , however .the bal qsos were the first objects as a class to show significant optical linear polarization not counting the highly variable blazars whose high polarization arises from beamed synchrotron - emitting jets .spectropolarimetry of several bal qsos shows evidence for reddened direct views of the nucleus and multiple scattered light paths .unlike most other ` buried ' type 1 objects , the broad emission line polarization is generally low compared with the continuum , suggesting scatterers mixed with or at distances less than the blr . in principle , the polarization position angle , and modeling the degree of polarization , may lead to further clues to the radio - quiet qsos inner structure ( see the discussion of iras 07598 + 6508 4.7 , fig .serious doubt about the above simple orientation unification of bal qsos and non - bal qsos is suggested by the probable association of strong feii emission and weak [ oiii]@xmath195007 ( nlr ) emission with the existence of bals , that is , of ` principal component 1 ' with the existence of bals ( boroson & meyers 1992 ; turnshek et al .several investigations suggest a range in covering factor for bal material , so the truth may lie in orientation , together with a range in covering factor .by now there are many examples , so we choose a few for illustration , and give references to some others .ngc1068 is the prototypical seyfert 2 galaxy and the prototype of type 1 nuclei buried within a dusty torus .the following situation is similar to that illustrated in fig .3 . spectropolarimetry revealed a faint polarized type 1 spectrum ( miller & antonucci 1983 ) . in total lightthe scattered type 1 spectrum is dominated by an essentially unpolarized type 2 spectrum , nearly 100 times stronger .the type 1 spectrum shows wavelength - independent polarization from optical to uv wavelengths ( @xmath2616% ) , indicating electron scattering , and the polarization is perpendicular to the weak radio jet ( fig .4(a ) & ( c ) ) .direct light from the center ( ir to x - rays ) is obscured ( a@xmath27 ) by the edge - on torus , whose outer extensions have been imaged in the near infrared ( on a scale of tens of pc ) , in thermal emission from ionized clouds on vlba scales of @xmath261 pc , and detected in h@xmath28o and oh maser emission .even the unresolved radio nucleus is partially absorbed .the ratio of scattered optical continuum to x - ray continuum is like that for unobscured type 1 nuclei , indicating electron scattering of the x - ray spectrum as well .high equivalent width fe - k@xmath29 near 6.4 kev from the unobscured scattering region was predicted and confirmed . at optical and uv wavelengths , a one - sided ` cone ' of high - ionization and scattering gas , formed by the shadow of the dusty torus , subtends an opening angle of about 40 projected on the sky plane ( pogge 1988 ) .a bicone is detected in the near infrared ( packham et al .1997 ) . as in many other such obscured seyfert 1s ,the ionizing photon flux seen by the emission - line gas is significantly greater than can be accounted for by our line - of - sight view of the nucleus ( wilson 1996 ) . in fact , this was one of the first clues that the central continuum was anisotropic .in addition to the electron scattered view of the center of ngc 1068 , miller , goodrich , & mathews ( 1991 ) discovered a second fainter view in scattered light from a dust patch @xmath30ne of the nucleus . here , a polarized seyfert 1 flux spectrum rising to short wavelengths indicated rayleigh - type scattering from dust grains .the dust - scattered h@xmath18 profile from the second position shows a ` narrow - lined seyfert 1 ' spectrum , suggesting that the electron - scattered profile from the first position is broadened by thermal motion of the hot electrons .while ngc4258 is only a very low - level agn , it has special importance because of its closeness ( @xmath267 mpc ) allowing good linear resolution , but especially because it contains a thin edge - on warped molecular disk , enabling the exciting discovery of water vapor maser emission .the radial velocity curve of the maser lines can be mapped at vlbi resolution , and shows a very accurate r@xmath31 dependence to within 0.13 pc of the center ( miyoshi et al .this led to an estimate of @xmath32 m@xmath33 for the central mass , and the best evidence yet for a supermassive black hole in an active galaxy .spectropolarimetry revealed a faint agn - like blue continuum and emission lines polarized at 510% .the emission lines were broad for a low luminosity nucleus ( @xmath261000 km s@xmath11 ) .polarization is perpendicular to the disk axis .these observations suggest that the molecular disk or torus obscures an agn central engine , albeit a low - power one .high - resolution imaging in the near ir actually reveals an unresolved nucleus , suggesting a@xmath3417@xmath35 ( chary & becklin 1997 ) .a few additional maser sources with similar properties have since been discovered ( e.g. , trotter et al . 1998 ) .this iras - discovered agn is identified with a very luminous cd galaxy ( [email protected] ) in a rich cooling - flow cluster .it is important because its nucleus is of qso luminosity ( @xmath36 l@xmath33 ) and excitation , with a powerful type 2 spectrum a qso2 .a scattered qso type 1 spectrum is seen in polarized flux ( p@xmath2620% ) .the radio structure is more like fri than frii , which makes this object of importance for both fri and frii unified schemes .it is one of the few ` hidden ' qsos to show both opposed scattering - ionization cones ( opening angle @xmath37 ) .these giant cones extend @xmath30kpc from the nucleus .this is a ld radio - loud qso with high , wavelength - independent degree of polarization , indicating an electron - scattered type 1 spectrum and a completely obscured direct view of the nucleus .unpolarized narrow lines of high equivalent width suggest a direct view of the nlr and an obscured continuum .the wavelength - independent polarization parallel to the radio jets indicates scattering in an optically - thin edge - on disk ( fig .4(b ) and ( d ) ) .while oi287 is of high radio power , the extended radio features appear more like bridges than jets ; in this respect it is more like iras09104 + 4109 , which has been suggested to have fri radio structure .this z=1.2 agn is special because it is the most luminous , most lobe - dominant qso in the 3crr catalog .its optical - uv continuum is the reddest known and it shows a highly polarized type 1 spectrum , with degree of polarization increasing to @xmath2610% in the uv .this is interpreted in terms of a highly reddened direct spectrum , combined with a reddened scattered spectrum .like oi287 , 3cr68.1 shows strong associated absorption , which can be understood as the result of light from the center passing through absorbing clouds associated with the thinner regions of the torus ( fig .this is the archetype of highly - polarized radio - quiet qsos .it was the first qso to be discovered in the mid infrared ( 12 100 ) by iras . in the optical - uv it shows a strong , but reddened , type 1 spectrum ( z = 0.10 ) , with extremely weak nlr emission .its continuum and broad lines are increasingly polarized towards the uv , up to 10% , and its polarization ( * e*-vector ) is aligned with the major axis of a galaxy disk visible in the near ir .the explanation is in terms of a combination of a direct spectrum partially obscured by a dusty torus , and a much less - reddened ( polarized ) scattered spectrum .x - ray variability suggests a direct view to the center , but surprisingly , in view of the optical reddening , iras 13349 + 2438 is strong in soft x - rays ( @xmath38kev @xmath39kev ) with no sign of the expected cold absorption .this was one of the first examples of x - ray ` warm absorption ' .the dust appears to be mixed with warm ionized gas that produces absorption only at higher photon energies , e.g. , ovii and oviii edges at @xmath260.7 kev .another iras - discovered radio - quiet qso , iras07598 + 6508 , at z=0.15 , was found first to have high polarization , leading to the discovery of extremely strong blue - shifted broad absorption lines ( bals ) in the uv .spectropolarimetry showed a polarized continuum and unpolarized broad emission lines , unlike the qso 2s and seyfert 2s investigated so far .this was the first object for which an increased degree of polarization was seen in an absorption line indicating _ decreased _ dilution of a polarized continuum by an unpolarized absorbed continuum component .because the broad emission lines are apparently not scattered , but part of the continuum is , the scatterers must lie nearer the qso continuum source than the blr , and be distinct from the absorption region .this is an important constraint on scattering geometry . like other bal qsosit is not detected in x - rays , suggesting complete absorption by neutral gas with n@xmath40 cm@xmath31 .iras07598 + 6508 is also a ` super feii ' emitter .dusty tori can be recognized by their absorbing effects in two ways : first , by their reddening and obscuration of light at high inclinations , and second , by dimming direct nuclear light , allowing the detection of a faint , scattered , polarized spectrum .the alignment of the polarization of a scattered type 1 spectrum with the radio jet is the strongest evidence for a nuclear dusty torus .perpendicular or parallel alignment of the scattering polarization was recognized for a number of seyfert and radio galaxies by antonucci ( 1982 , 1983 ) , and interpreted as arising in the scattering geometries illustrated in fig .4 ( see 4 ) . the polarimetric evidence has been presented above by giving a few of many examples . in nearby low luminosity agn , where high spatial resolution is possible , flattened dusty regions or their outer 100pc scale extensions can be recognized by direct imaging , either in absorption or infrared emission ( see fig . 5 , and ngc1068 above ) .usually the disk axis is well aligned with the radio jets .there is much evidence for dust in the nlr , and between the nlr and blr , from differential reddening and differential polarization of the broad and narrow lines .evidence for flattened dusty disks of high optical depth in the nuclear regions of low luminosity agn was noted by de zotti and gaskell ( 1985 ) .we note three nice statistical studies relating increasing extinction to increasing lobe dominance : + ( i ) hill et al .( 1996 ) sought to detect the obscured broad line region via observations of pa@xmath29 in the near infrared where reddening effects are much smaller than the optical .they observed a complete sample of 3cr frii agns , that is , agn with radio luminosities in the qso class .they found increasing reddening with decreasing core - dominance r ( increasing inclination ) , consistent with a dusty torus model like that shown in fig.2 .+ ( ii ) baker and collaborators ( 1995 , 1997 ) have investigated a complete sample of low - frequency selected rl qsos , showing the steepening of the optical continuum and reddening of the balmer decrements with decreasing r. + ( iii ) baker ( 1997 ) and earlier , he s , fosbury & barthel ( 1994 ) have shown increasing narrow - line ratio [ oii]@xmath193727/[oiii]@xmath195007 with decreasing r , indicating the preferential obscuration of the higher - ionization [ oiii ] emission that is expected to be produced closer to the nucleus .alignment of the central engine axis with the host galaxy is important for understanding unified schemes too , and for completeness we note an investigation by schmitt et al .( 1997 ) and an earlier one by ulvestad and wilson ( 1984 ) , showing that the axis of small - scale radio structure in seyfert galaxies appeared to avoid alignment with the host galaxy minor axis .a more recent analysis suggests that this is the result of observational selection ( nagar & wilson 1999 ) .they find that observations are consistent with a uniform distribution of radio jet direction relative to the galaxy axis .additional important evidence for the existence of dusty tori in all or most luminous agn is the presence of the 3@xmath42 bump in essentially all radio - loud qsos and many radio - quiet qsos , where the redshift is low enough to allow detection ( 3.1 ) .models suggest that the near - infrared optical depths should depend on inclination ( ward 1995 ) .gas and dust within the opening of the torus , and therefore illuminated by the central qso , gives rise to the scattered broad line spectrum , as well as most of the nlr emission .the structure of these ` cones ' is clearest for nearby seyfert 2 galaxies , and more than a dozen are now known with quite clearly defined ` edges ' .however , the emitting gas and dust is obviously clumpy , so the edges are often ill - defined .the first ionization cones were investigated by pogge ( 1989 ) with arcsecond resolution .several examples of bicones are seen ( fig .6 ) , but the cone facing away from the observer is often obscured .opening angles range from 70 to 100 .the ionizing continuum deduced from the ionization state and density of the emitting gas is often brighter than the observed continuum .unified schemes suggest that the dusty torus or its extensions blocks our view of the continuum .the dusty torus should be heated by this continuum , and for a given torus geometry one can predict the amount of re - radiated infrared emission .storchi - bergmann et al . ( 1992 ) confirm such predictions for 8 out of a small sample of 9 seyfert 2s .a recent survey of early - type seyfert galaxies ( nagar et al .1999 ) and earlier work show that the axes of ionized emission are typically aligned within a few degrees of the inner radio jets . while some nlr emission is clearly excited by jet - related shocks ( e.g. , capetti et al . 1996; falcke et al . 1998 ) , much of the ionized gas is not directly associated with the radio emission ; there is ionized gas beyond the jet and the jets are rather more highly collimated than the ionized gas .therefore , wilson ( 1996 ) concludes , the radio plasma and ionizing photons are collimated by the same , or coplanar , disks or tori .another test is provided by a comparison of the optical and uv spectra of a sample of seyfert 2 galaxies .kinney et al .( 1991 ) estimate the ionizing photons available from the observed uv continuum , concluding that in most cases there are insufficient to produce the observed emission line fluxes .interestingly , the uv continua , while weak , have slopes indistinguishable from those of seyfert 1 galaxies , consistent with the seyfert 2 uv continuum being an electron - scattered seyfert 1 continuum . comparing a much larger sample of seyfert 1 and seyfert 2 agn , mulchaey et al .( 1994 ) are able to show that , while the infrared continua , [ oiii]@xmath195007 nlr strengths , and the hard x - ray are similar in both classes , consistent with isotropic emission , the uv and soft x - ray ( 0.2 - 4 kev ) continua of seyfert 2s are underluminous , consistent with line - of - sight absorption .in addition , they find that while the line emission and continuum fluxes are correlated in seyfert 1s , consistent with the uv photons powering the lines emission , there is no such correlation for the seyfert 2s , again consistent with our not seeing the ionizing continuum directly in seyfert 2s .there are similar results from calculations of the energy budget for a few higher luminosity radio - quiet qsos [ e.g. , iras 09104 + 4109 ( 4.7 ) ; iras 20460 + 1935 ( frogel et al . 1989 ) ] .one can also check the opening angles derived by the above direct imaging results with those derived by comparing relative numbers of type 2 and type 1 agns , assuming that most of the type 2 agns contain buried type 1 nuclei .opening angles derived from comparisons of seyfert 1 and seyfert 2 galaxies lie in the right ballpark about 60 to 80 ( e.g. , osterbrock & shaw 1988 ) . for rl agns , the comparison is easier because it is possible ( with a lot of hard work ) to select an orientation - independent sample by isotropic low - frequency radio emission . to avoid optical bias ,the sample must be completely optically identified .this is how barthel ( 1989 ) , using the completely identified 3crr sample ( 178 mhz laing , riley & longair 1983 ) , derived the value @xmath4345 used in 3.1 .willott , rawlings & blundell ( this volume ) use the much larger 7c sample selected at 151 mhz , and derive an opening angle of @xmath26120 for the most luminous qsos . see their paper for a discussion of this difference .qso 2s are to ` normal ' qso 1s , as seyfert 2s are to seyfert 1s .it has sometimes been hotly claimed that qso 2s do nt exist , with the suggestion that high luminosity agn have only small covering by a dusty torus .investigations of individual qsos ( e.g. , 4 ) lead to a wide range in dust - covering fraction , and this raises the question of just how many luminous qsos are we missing in samples selected by their optical - uv emission ?the only way to address this question is to observe a complete sample of agn selected by a property that is independent of orientation .selection by low - frequency radio emission is the best method .the 3crr sample selected at 178 mhz is excellent , and selection by frii structure identifies those of qso luminosities ( urry & padovani 1995 ; hill et al .this was the sample that led barthel ( 1989 ) to propose that frii radio galaxies were buried frii qsos .thus , some nlrg and 3cr68.1 ( 4.5 ) are qso 2s , and the blrg and reddened qsos ( smith & spinrad 1980 ) are of intermediate inclination .see also willott et al .( this volume ) .recent studies of reddening in rlqs are relevant .webster et al . (1995 ) have found a significant number of red qsos among a completely - identified sample of parkes flat radio - spectrum sources . while some fraction of these might be expected to have a steep ( and therefore red ) ir - uv synchrotron component ( serjeant & rawlings 1996 , see 7 ) a significant fraction are probably dust reddened ( francis et al .carilli et al .( 1998 ) have found 4 of 5 red flat radio - spectrum qsos with significant columns of neutral hydrogen either the result of intervening absorbers or , in two cases , intrinsic to the qso .this is interesting because , according to unified schemes , these flat - spectrum sources are cd qsos and are therefore expected to be the least dust - reddened , so absorption ( and reddening ) is likely to be rather more for lobe - dominant sources where the line - of - sight passes closer to the plane of the dusty torus .however the above studies apply only to rlqs .# 1#2#3#4#5#6#7 to#2 ' '' '' the rqqs far outnumber the rlqs , so it is of interest to understand how unified schemes work for rqqs .the hard x - ray region ( @xmath44kev , wilkes 1999 , this volume ) or the mid - infrared ( e.g. , rush , malkan & spinoglio 1993 ) should be good wavelengths to select a sample because absorption by cold gas or dust , while not negligible , is much less .the x - ray data are not yet available . in the infrared oneis overwhelmed by the huge numbers of galaxies , whose hii regions , excited by young stars , emit thermally in the infrared .a practical solution adopted by low et al .( 1988 ) and others is to select , in addition , by warm infrared colors .neugebauer et al .( 1986 ) had shown that warm infrared colors were an essentially unique signature of qsos and seyferts already discovered by optical techniques , so to find the same kinds of qsos , independent of whether or not their optical - uv emission is covered by dust , low et al .selected iras type 1 spectra with f@xmath45m)/f@xmath46 m ) @xmath47 .using the same criteria , wills & hines ( 1997 ) defined a complete sample of extragalactic objects with l(ir)@xmath48l(h@xmath12 km s@xmath11 mpc@xmath11 ) , regardless of optical classification .six objects ( 1/3 of the sample ) had type 2 spectra with type 1 spectra revealed in polarized light , and are therefore qso 2s . the spectral energy distribution of one of these is shown in fig.7 , where its steep optical uv spectrum , soft x - ray absorption , and recovery at harder x - ray energies , is illustrated .about 1/3 showed reddened type 1 spectra in total and polarized light , providing evidence for a significant scattered - light contribution to a partially obscured direct view of the center .the remainder were less reddened , but only 3 of the sample had been found by conventional optical - uv selection techniques .when corrected for reddening the ratio of ir to optical flux is no different from typical optically - selected pg qsos , so spectroscopically and with respect to their spectral energy distributions , these iras - selected luminous agn appear to harbor ` normal ' qsos .two thirds of the sample agn are either qso 2s , objects of qso luminosity dominated by type 2 spectra but with buried qso 1 nuclei , or qsos without strong nlr emission , but with a reddened line - of - sight to the center , and a significant scattered type 1 contribution reddened qso 1s .these would not have been discovered by conventional selection techniques .it is likely that even more - obscured luminous agns were missed by the iras color - selection , indicating that space densities of qsos have been underestimated by a large factor .the present description of unified schemes has intentionally been simplified .many aspects are the subject of active research .there is now a huge literature relevant to the subject certainly over 1000 papers .therefore i have selected some topics at the expense of others ; i apologize for important papers not included .unification of the qso majority , the most radio quiet , remains difficult without an indicator of inclination .one possibility is to search for ionization or scattering cones , but these may be small and faint .one might speculate whether beamed x - ray emission could be used as an orientation indicator for radio - quiet qsos .the basis for this is the discovery of radio - weak bl lac objects whose beamed radiation appears to be in the euv to soft x - ray region ( stocke et al .1985 ; padovani & giommi 1995 ) .the minimum ingredients for a unification scheme are orientation and patchy dust , but there is good evidence for a richer and more interesting unified scheme based on an axisymmetric central engine surrounded by an obscuring dusty torus . some tantalizing links are suggested between the torus parameters and other agn relationships .covering by dense feii - rich gas , and outflowing bal gas ( 3.3 ) , appear to be part of the ` principal component ' relationships among x - ray continuum , broad line profiles , and the inverse relationship between feii strength and [ oiii]@xmath195007 ( wills et al ., aoki et al . , this volume ) .radio loudness appears to be another parameter linked to these principal component relationships and to properties of bal outflows .links between strong feii emission and dust abound in astrophysical situations , so , is there a relationship between covering by dense line - emitting gas and the dusty torus ? it has been suggested that the torus opening angle increases with luminosity .the x - ray slope and broad - line equivalent widths depend on luminosity ( korista et al . ,this volume ) .broad line widths may also increase with luminosity .we have yet to learn whether or how all these relationships may be linked .unification of diverse classes of agn by changing aspect angle and extinction by dust is a reality , as foretold by rowan - robinson ( 1977 ) .while most astronomers would like to sweep it under the rug , the complicated subject of dust must be confronted ( e.g. , masci 1998 ) . not only does external dust absorb , scatter and emit thermal radiation, it affects the energy balance and chemistry of absorption and emission line regions .most surveys for agn are biased against dusty objects .this will change with new surveys in the radio , infrared and hard x - ray regimes .the lesson for cosmology is that the space density of qsos has been underestimated , with important implications for their evolution and contribution to the cosmic background . with increasing evidence for supermassive black holes in many galaxies , even those with little activity ,it is clear from the space density and luminosity functions of galaxies and qsos that qsos are a phenomenon common to perhaps all massive galaxies , even if short - lived and intermittent .therefore , what we learn about active galaxies and qsos is relevant to the evolution of all galaxies , and to cosmology .here i expand upon the references given in the above sections , with brief remarks .* radio - loud agn unification : * blandford & rees 1978 ( the classic paper on ideas behind present unified schemes ) vermeulen & cohen 1994 ( superluminal motion ) ghisellini et al . 1993 ( statistics relating radio , optical & x - ray observations for several classes of agn ) kellermann & owen 1988 ( general review of extragalactic radio sources at a basic level ) orr & browne 1982 ( rl qso cd vs. ld statistics ) hughes , p. a. 1991 ( good theory and observation text ) browne & perley 1986 ( halos of cd sources ) odea , c. 1998 ( gigahertz peaked spectrum sources - a review ) urry ( 1999 ) ( the fri frii break ) . *unification of frii agns : * tadhunter , dickson & shaw 1996 , cimatti et al .1998 ( polarimetry of blrg and reddened qsos ) sanders et al .1989 , barvainis 1987 ( 3 @xmath20 m bump ) shastri et al . 1993 ( x - ray spectrum and core - dominance ) wills & browne 1986 , vestergaard 1998 ( broad - line width vs. core - dominance ) .* seyfert galaxy unification : * tran 1995 ( spectropolarimetry ) , veilleux , goodrich & hill 1997 ( near ir spectroscopy ) .heisler , lumsden & bailey 1997 ( even scattered light obscured at highest inclinations ) .* radio - quiet qsos : * arav , shlosman & weymann ( eds ) 1997 , korista et al .1993 , glenn et al .1994 ( bal qsos , multiple light paths ) .* ngc1068 : * antonucci & miller 1985 , bailey et al .1988 , young et al .1995 ( spectropolarimetry ) roy et al .1998 ( absorbed radio nucleus ) bock et al . 1998( ir imaging ) capetti et al .1995 ( polarization imaging ) greenhill , l. 1996 ( vlbi h@xmath28o maser ) matt et al .1997 ( hard x - ray spectrum ) gallimore , baum , & odea 1995 ( pc - scale thermal emission from the torus ) . * the dusty torus : * lawrence & elvis 1982 ( perhaps the first paper to suggest that type 2 spectra were type 1 spectra with absorbed broad line emission and absorbed soft x - rays ) , ward 1995 ( the oxford torus workshop ) .i thank jack baldwin , mike brotherton and derek wills for many helpful comments .i am very grateful for the work of the staffs of the nasa astrophysics data system , the eprint astrophysics archives , and the nasa extragalactic database , without whom this review would have been impossible .b.j.w . is supported by nasa through ltsa grant number nag5 - 3431 and grant number go-06781 from the space telescope science institute , which is operated by the association of universities for research in astronomy , inc ., under nasa contract nas5 - 26555 .antonucci , r. r. j. 1982 nature , 299 , 605 antonucci , r. r. j. 1983 nature , 303 , 158 antonucci , r. r. j. 1993 , ara&a , 31 , 473 antonucci , r. r. j. , & miller , j. s. 1985 , , 297 , 476 antonucci , r. r. j. , kinney , a. l. , & hurt , t. 1993 , , 414 , 506 antonucci , r. r. j. , hurt , t. , & miller , j. s. 1994 , , 430 , 210 antonucci , r. r. j. , & ulvestad , j. s. 1985 , , 294 , 158 arav , n. , shlosman , i. , & weymann , r. a. ( eds . ) 1997 , mass ejection from agn ( san francisco : asp ) bailey , j. a. , axon , d. j. , hough , j. h. , ward , m. j. , mclean , i. , heathcote , s. r. 1988 , , 234 , 899 baker , j. c. 1997 , , 286 , 23 baker , j. c. , & hunstead , r. w. 1995 , , 452 , 95 baker , j. c. , hunstead , r. w. , & brinkmann , w. 1995 , , 277 , 553 barthel , p. d. 1989 , , 336 , 606 barvainis , r. 1987 , , 320 , 537 becker , r. h. , white , r. l. , helfand , d. j. 1995 , , 450 , 559 beichman , c. a. , soifer , b. t. , helou , g. , chester , t. j. , neugebauer , g. , gillett , f. c. , & low , f. j. 1986 , , 308 , 1 blandford , r. & rees , m. 1978 , in pittsburgh conference on bl lac objects , ed .a. m. wolfe ( pittsburgh : university of pittsburgh ) , 328 blundell , k. m. , & beasley , a. j. 1998a , , 299 , 165 blundell , k. m. , & beasley , a. j. 1998b , baas , meeting no .193 , 110.04 .bock , j. j. , marsh , k. a. , ressler , m. e. , werner , m. w. 1998 , , 504 , 5 boroson , t. a. , & meyers , k. a. 1992 , , 397 , 442 brandt , w. n. , fabian , a. c. , takahashi , k. , fujimoto , r. , yamashita , a. , inoue , h. , ogasaka , y. 1997a , , 290 , 617 brandt , w. n. , mathur , s. , reynolds , c. s. , & elvis , m. 1997b , , 292 , 407 bridle , a. h. , hough , d. h. , lonsdale , c. j. , burns , j. o. , laing , r. a. 1994 , , 108 , 766 brotherton , m. s. , wills , b. j. , dey , a. , van breugel , w. , & antonucci , r. r. j. 1998 , , 501 , 110 browne , i. w. a. 1987 , in superluminal radio sources , eds . j. a. zensus , & t. j. pearson ( cambridge : cup ) , 129 browne , i. w. a. , & perley , r. a. 1986 , , 222 , 149 burke , b. f. , & graham - smith , f. 1997 , an introduction to radio astronomy ( cambridge : cup ) capetti , a. , macchetto , f. , axon , d. j. , sparks , w. b. , & boksenberg , a. 1995 , , 452 , l87 capetti , a. , axon , d. j. , macchetto , f. , sparks , w. b. , boksenberg , a. 1996 , , 469 , 554 carilli , c. l. , menten , k. m. , reid , m. j. , rupen , m. p. , & yun , m. s. 1998 , , 494 , 175 chary , r. & becklin , e. e. 1997 , 485 , 75 cimatti , a. , di serego alighieri , s. , vernet , j. , cohen , m. , & fosbury , r. a. e. 1998 , , 499 , 21 cohen , m. h. & unwin , s. c. 1982 , proc .iau symposium no .97 , `` extragalactic radio sources '' , eds .d. s. heeschen & c. m. wade ( reidel : dordrecht ) , 345 corbett , e. a. , robinson , a. , axon , d. j. , hough , j. h. , jeffries , r. d. , thurston , m. r. , young , s. 1996 , , 281 , 737 crawford , c. s. , & vanderreist , c. 1996 , , 283 , 1003 de zotti , g. , & gaskell , c. m. 1985 , , 147 , 1 dennett - thorpe , j. , bridle , a. h. , scheuer , p. a. g. , laing , r. a. , leahy , j. p. 1997, , 289 , 753 falcke , h. , gopal - krishna , & biermann , p. l. 1995 , , 298 , 395 falcke , h. , wilson , a.s ., & ho , l.c .1997 , proc .relativistic jets in agns , 13 falcke , h. , patnaik , a. r. , sherwood , w. 1996a , , 473 , 13 falcke , h. , sherwood , w. , patnaik , a. r. 1996b , , 471 , 106 falcke , h. , wilson , a. s. , simpson , c. 1998 , , 502 , 199 fanaroff b. l. , & riley , j. m. 1974 , , 167 , 31p ferrarese , l. , ford , h. c. , & jaffe , w. 1996 , , 470 , 444 francis , p.j . , webster , r. l. , masci , f. j. , drinkwater , m. j. , peterson , b. a. 1997 , in emission lines in active galaxies : new methods & techniques , ed b. m. paterson , f .- z .cheng , & a. s. wilson ( san francisco : asp ) , 130 frogel , j. a. , gillett , f. c. , terndrup , d. m. , & vader , j. p. 1989, , 343 , 672 fugmann , w. 1988 , , 205 , 86 gallimore , j. f. , baum , s. a. , & odea , c. p. 1995 , nature , 388 , 852 ghisellini , g. , padovani , p. , celotti , a. , maraschi , l. 1993 , , 407 , 65 giovannini , g. , arbizzani , e. , feretti , l. , venturi , t. , cotton , w. d. , lara , l. , & taylor , g. b. 1998 , iau colloquium 164 : radio emission from galactic & extragalactic compact sources , asp conference ser .144 , eds .zensus , j.a ., taylor , g.b . , & wrobel , j.m .( asp : san francisco ) , 85 glenn , j. , schmidt , g. d. , & foltz , c. b. 1994 , , 434 , 47 gopal - krishna , kulkarni , v. k. , wiita , p. j. 1996 , , 463 , 1 grandi , s. a. & osterbrock , d. e. 1978 , , 220 , 783 green , p.j . , & mathur , s. 1996 , , 462 , 637 greenhill , l. 1996 , , 472 , 21 heisler , c. a. , lumsden , s. l. , & bailey , j. a. 1997 , nature , 385 , 700 he s , r. , fosbury , r. a. e. , & barthel , p. d. 1994 , in the first stromlo symposium : the physics of active galaxies , asp conference ser .g. v. bicknell , m. a. dopita , & p. j. quinn ( asp : san francisco ) , 367 hill , g. j. , goodrich , r. w. , & depoy , d. l. 1996 , , 462 , 163 hines , d. c. , schmidt , g. , wills , b. j. , smith , p. s. & sowinski , g. 1999 , , 512 , 000 hines , d. c. , & wills , b. j. 1993 , , 415 , 82 hines , d. c. , & wills , b. j. 1995 , , 448 , 69 hooper , e. j. , impey , c. d. , foltz , c. b. , hewett , p. c. 1995 , , 445 , 62 hough , d. h. , & readhead , a. c. s. 1989 , , 98 , 1208 hughes , p. a. ( ed . ) 1991 , beams & jets in astrophysics ( cambridge : cup ) impey , c. d. , lawrence , c. r. , & tapia , s. 1991 , , 375 , 461 jaffe , w. , ford , h. c. , ferrarese , l. , van den bosch , f. , & oconnell , r. w. 1993 , nature , 364 , 213 1996 , apj , 460 , 214 jones , d.l . & wehrle , a. e. 1998 , iau colloquium 164 : radio emission from galactic & extragalactic compact sources , asp conference ser .144 , eds .zensus , j.a . ,taylor , g.b . , & wrobel , j.m .( asp : san francisco ) , 63 kellermann , k. i. , & owen , f. n. 1988 , in galactic & extragalactic radio astronomy , second edition , eds . k. i. kellermann , & g. l. verschuur ( new york : springer - verlag ) , 563 kellermann , k. i. , sramek , r. a. , schmidt , m. , green , r. f. , shaffer , d. b. 1994 , , 108 , 1163 kellermann , k. i. , vermeulen , r. c. , zensus , j. a. , & cohen , m. h. 1998 , , 115 , 1295 kinney , a. l. , antonucci , r. r. j. , ward , m. j. , wilson , a. s. , whittle , m. 1991 , , 377 , 100 kleinmann , s. g. , hamilton , d. , keel , w. c. , wynn - williams , c. g. , eales , s. a. , becklin , e. e. , kuntz , k. d. 1988 , , 328 , 161 korista , k. t. , voit , g. m. , morris , s. l. , & weymann , r. j. 1993 , , 88,357 krolik , j. h. 1999 , active galactic nuclei ( princeton : princeton university press ) kukula , m. j. , dunlop , j. s. , hughes , d. h. , & rawlings , s. 1998 , , 297 , 366 kwan , j. , cheng , f .- z . , fang , l .- z . , zheng , w. , ge , j. 1995 , , 440 , 628 laing , r.a . , riley , j.m . , & longair , m.s .1983 , , 204 , 151 lara , l. , mrquez , i , cotton , w. d. , feretti , l. , giovannini , g. , marcaide , j. m. , & venturi , t. 1998 , astro - ph/9812254 lawrence , a. , & elvis , m. 1982 , , 256 , 410 lawrence , a. et al .1988 , , 235 , 261 leahy , j. p. , et al .1997 , , 291 , 20 lind , k. r. , & blandford , r. d. 1985 , , 295 , 358 lipari , s. 1994 , , 436 , 102 longair , m. s. 1981 , high energy astrophysics ( cambridge : cup ) low , f. j.,cutri , r. m. , huchra , j. p. , kleinmann , s. g. 1988 , , 327 , 41 malkan , m. a. & sargent , w. l. w. 1983 , , 254 , 22 masci , f. j. 1998 , pasa , 15 , 299 matt , g. , et al .1997 , , 325 , l13 miller , j. s. , & antonucci , r. r. j. 1983 , , 271 , 7 miller , j. s. , & goodrich , r. w. 1988 , , 331 , 332 miller , j. s. , goodrich , r. w. , & mathews , w. g. 1991 , , 378 , 47 miyoshi , m. , moran , j. m. , herrnstein , j. r. , greenhill , l. j. , nakai , n. , diamond , p. j. , & inoue , m. 1995 , nature , 373 , 127 morganti , r. , parma , p. , capetti , a. , fanti , r. , de ruiter , h.r ., 1997 , , 326 , 919 mulchaey , j. s. , et al .1994 , , 436 , 586 mushotzky , r. f. 1997 , in mass ejection from agn , asp conference ser .128 , eds .n. arav , i. shlosman , & r. j. weymann ( asp : san francisco ) , 141 moran , j. m. , greenhill , l. j. , herrnstein , j. r. , diamond , p. j. , miyoshi , m. , nakai , n. , & inoue , m. 1995 , proc ., 92 , 11427 morganti , r. , parma , p. , capetti , a. , fanti , r. & de ruiter , h. r. 1997 , , 326 , 919 nagar , n. m. , & wilson , a. s. 1999 , , 516 , no . 1 nagar , n. m. , wilson , a. s. , mulchaey , j. s. , & gallimore , j. f. 1999 , , 120 , no . 2 neeser , m. j. , eales , s. a. , law - green , j. d. , leahy , j. p. , rawlings , s. 1995 , , 451 , 76 neugebauer , g. , soifer , b. t. , miley , g. k. , & clegg , p.e .1986 , , 308 , 815 odea , c. p. 1998 , pasp , 110 , 493 orr , m. j. l. , & browne , i. w. a. 1982 , , 200 , 1067 osterbrock , d. e. 1989 , astrophysics of gaseous nebulae & active galactic nuclei ( mill valley : university science books ) osterbrock , d. e. , & shaw , r. a. 1988 , , 327 , 89 owen , f. n. , ledlow , m. j. , & keel , w. c. 1996 , , 111 , 53 packham , c. , young , s. , hough , j. h. , axon , d. j. , bailey , j. a. 1997 , , 288 , 375 padovani , p. , & giommi , p. 1995, , 444 , 567 penston , m. v. , & perez , e. 1984 , , 211 , 33p peterson , b. m. 1997 , an introduction to active galactic nuclei ( cambridge : cup ) pogge , r. w. 1988 , , 328 , 519 pogge , r. w. 1989 , , 345 , 730 robson , i. active galactic nuclei ( chichester : wiley - praxis ) roy , a. l. , colbert , e. j. m. , wilson , a. s. , ulvestad , j. s. 1998 , , 504 , 147 rudy , r. j. & schmidt , g. d. 1988 , , 331 , 325 rush , b. , malkan , m. a. , & spinoglio , l. 1993 , , 89 , 1 sanders , d. b. , phinney , e. s. , neugebauer , g. , soifer , b. t. , matthews , k. 1989 , , 347 , 29 schmitt , h. r. , kinney , a. l. , storchi - bergmann , t. , & antonucci , r. r. 1997 , , 477 , 623 shastri , p. , wilkes , b. j. , elvis , m. , mcdowell , j. 1993 , , 410 , 29 shields , g. a. 1978 , nature , 272 , 706 smith , h. e. , & spinrad , h. 1980 , , 236 , 419 stickel , m , fried , j. w. , & khr , h. 1993 , , 98 , 393 stocke , j. t. , liebert , j. , schmidt , g , gioia , i. m. , maccacaro , t. , schild , r. e. , maccagni , d. , & arp , h. c. 1985 , , 298 , 619 storchi - bergmann , t. , mulchaey , j. s. , & wilson , a. s. 1992 , , 395 , 73 tadhunter , c. n. , dickson , r. c. & shaw , m. a. 1996 , 281 , 591 tran , h. d. 1995 , , 440 , 597 tran , h. d. , brotherton , m. s. , stanford , s. a. , van breugel , w. , dey , a. , stern , d. , & antonucci , r. r. j. 1999 , , in press trotter , a. s. , greenhill , l. j. , moran , j. m. , reid , m. j. , irwin , j. a. , & lo , k .- y .1998 , , 495 , 740 turnshek , d. a. , monier , e. m. , sirola , c. j. , & espey , b. r. 1997 , , 476 , 40 ulrich , m. h. , hackney , k. r. h. , hackney , r. l. , & kondo , y. 1984 , , 276 , 466 ulvestad , j. s. , & antonucci , r. r. j. 1988 , , 328 , 569 ulvestad , j. s. , & wilson , a. s. 1984 , , 285 , 439 urry , c. m. 1999 , in the bl lac phenomenon ( turku , finland ) urry , c. m. , & padovani , p. 1995 , pasp , 107 , 803 veilleux , s. , goodrich , r. w. , hill , g. j. 1997 , , 477 , 631 vermeulen , r. c. , & cohen , m. h. 1994 , , 430 , 467 vermeulen , r. c. , et al .1995 , , 452 , 15 vestergaard , m. 1998 , phd thesis , univ . of copenhagen ward , m. j. ( ed . ) 1994 , oxford astrophysics workshop on evidence for the torus wardle , j. f. c. , & aaron , s. e. 1997 , , 286 , 425 webster , r. l. , francis , p. j. , peterson , b. a. , drinkwater , m. j. , & masci , f. j. 1995 , nature , 375 , 469 weedman , d. w. 1986 , quasar astronomy ( cambridge : cup ) weymann , r. j. , morris , s. l. , foltz , c. b. , hewett , p. c. 1991 , , 373 , 23 weymann , r. j. 1997 , in mass ejection from agn , asp conference ser .128 , eds .n. arav , i. shlosman , & r. j. weymann ( asp : san francisco ) , 3 wilkes , b. j. , schmidt , g. d. , smith , p. s. , mathur , s. , & mcleod , k. k. 1995 , , 455 , l13 wills , b. j. 1991 , in variability of active galactic nuclei , eds . h. r. miller & p. j. wiita ( cambridge : cup ) , 87 wills , b. j. , & browne , i. w. a. 1996 , 302 , 56 wills , b. j. , & hines , d. c. , in mass ejection from agn , asp conference ser . 128 , eds .n. arav , i. shlosman , & r. j. weymann ( asp : san francisco ) , 99 wills , b. j. , wills , d. , breger , m. , antonucci , r. r. j. , & barvainis , r. 1992a , , 398 , 454 wills , b. j. , wills , d. , evans , n. j. , natta , a. , thompson , k. l. , & breger , m. 1992b , , 400 , 96 wilson , a. s. 1996 , vistas in astronomy , 40 , 63 wilson , a. s. , & tsvetanov , z. i. 1994 , , 107 , 1227 wilson , a. s. , braatz , j. a. , heckman , t. m. , krolik , j. h. , & miley , g. k. 1993 , , 419 , 61 white , r. l. et al .1999 , , in press young , s. , hough , j. h. , axon , d. j. , bailey , j. a. , ward , m. j. 1995 , , 272 , 513

<S> the differences among apparently diverse classes of agn are mainly the result of viewing the central engine at different orientations , because dust , which absorbs and scatters the light , partially covers the central source , and because synchrotron emission is highly beamed along the relativistic jet . also important are factors independent of orientation : the total power output , the unknown mover behind the eigenvector 1 relationships , and the radio - loudness . </S><S> these other factors may not be independent of the parameters of unified models , such as intrinsic jet physics , agn dust content , and torus thickness . </S><S> we outline the basic evidence for orientation unified schemes , and briefly discuss their importance for understanding the mechanisms of the central engine and its relation to the surrounding host galaxy and beyond . </S>

nlin0210067

fractal sets and measures appear naturally as invariant sets ( respectively measures ) of many nonlinear dynamical systems .periodic orbit theory @xcite provides an effective approach to computing useful properties such as averages , lyapunov exponents and dimensions , particularly when the fractal corresponds to a nonattracting set ( `` transient chaos '' ) so that direct simulation methods are harder to implement . in the case of axiom a dynamics , the convergence of periodic orbit ( or `` cycle '' ) expansions can be spectacular , see tab . 1 below .recent work has extended the theory to include chaotic systems perturbed by external noise , motivated by realism ( all physical systems are coupled to unknown degrees of freedom ) and smoothness ( delta functions are replaced by smooth distributions ) .while several approaches have been attempted , including feynman diagrams @xcite ( by analogy with quantum perturbation theory ) and smooth conjugations @xcite ( by analogy with classical perturbation theory ) , the most computationally effective method @xcite has been to represent the stochastic evolution ( fokker - planck ) operator by a matrix in a basis of polynomials about each periodic orbit , truncated to finite size using the fact that the elements decay exponentially , that is , the eigenfunctions are very smooth . the previous work @xcite computed the escape rate of a stochastically perturbed map to order 8 in the noise strength . in this paperthe same matrix method , extended to high ( 60 digit ) precision arithmetic , is used to compute the escape rate to order 64 in the noise strength . with this number of coefficients ,it is meaningful to consider the asymptotic form of the late terms , the subject of this paper .we will discover that to unlock the secrets of the noise expansion will require insights not only from the classical theory of asymptotic expansions , but also from more recent analytic calculations involving fractals .the relatively small number of numerical coefficients is compensated by their high precision , allowing a reliable fit to a functional form involving several parameters . comparing two interpretations of the series , borel summation and shafer ( generalised pad ) approximation to the `` exact '' function, we will find that exponentially small corrections will need to be considered by a future theory .at this point we note a few other relevant works .contour integration methods @xcite have obtained the asymptotic form of noise coefficients for fixed periodic orbit length , however this does not directly determine the noise expansion of the escape rate since the latter requires successively longer orbits for higher noise corrections .direct integration @xcite has shown numerically that the cumulant expansion on which the cycle expansions are based is valid for strong as well as weak noise .stochastically perturbed dynamical systems constitute a vast field , applying many methods other than periodic orbit theory .section 2 outlines the previous theory and methods needed to understand the results and their interpretation . for space reasons , readers interested in the full detailsare referred to the original works .section 3 gives the coefficients , the logic used to fit them to a particular functional form , and the borel summation or shafer approximation needed to assign a precise meaning to the asymptotic expansion .final discussion is given in section 4 .this subsection gives the background for asymptotic series .a very readable review of this subject and its applications is given by boyd @xcite .singular perturbations of integrals or differential equations , such as perturbative approaches to physical problems , frequently have power series expansions of the form @xmath0 where @xmath1 is often an integer or half integer , and @xmath2 ( or rather its inverse ) is called the _ singulant _ , and is related to the nearest critical point of an integrand .@xmath3 is called the _ modifying factor _ and tends to a constant as @xmath4 ; it contains all the slower varying functional behaviour . such series diverge for all @xmath5 , and a number of methods have been employed to make sense of them .the simplest ( albeit discontinuous ) is to truncate the sum at its smallest term .an alternative , which boyd states as usually the most computationally efficient , is to replace the series by its pad approximants , more specifically the shafer extension @xcite in which the function is written as the solution of a quadratic equation with polynomial coefficients ; the coefficients of the polynomials are found by a set of ( typically ill - conditioned ) linear equations .while pad methods give results on average as good as any other method to this order , an alternative , borel summation , offers the possibility of systematic exponentially accurate ( `` hyperasymptotic '' ) corrections .these methods ( for example berry and howls @xcite ) start from the borel summation method as formulated by dingle @xcite , which we follow .the latter retains the decreasing terms , then performs borel summation on the divergent `` tail '' .borel summation is an approach in which the factorial is replaced by its integral representation @xcite , and the sum and integral are interchanged . in the present context ,where the terms are all the same complex phase , this leads to a pole in the path of integration ( corresponding to a `` stokes line '' ) ; the constraint that the result must be real then indicates that the principal value of the integral should be taken .dingle also shows how to write the borel summed expression in terms of a few standard functions ( related to incomplete gamma functions @xcite ) which he calls _terminants _ ; we use his method based on the forward difference expansion .more details can be found in dingle s book @xcite .for the bulk of this paper , we will need only eq .( [ e : asym ] ) , in order to fit the parameters , and generalise it slightly .once the form for the coefficients has been established , the shafer and dingle methods will be applied , and compared with the `` exact '' numerical result ; the shafer method gives marginally more accurate results .since the coefficients are fitted , not known exactly , and our calculation does not attempt to identify exponentially small corrections ( although this should be possible in the future ) , we can not apply the more detailed hyperasymptotic methods of berry and howls @xcite and others ( refer to ref .@xcite for extensive references ) . nevertheless , we succeed in computing a function accurate to remarkably high values of the noise strength ( perturbation parameter ) .this subsection gives the background for analytic expansions pertaining to exactly selfsimilar fractals , giving additional clues to the value of the parameter @xmath1 in eq .( [ e : asym ] ) and generalisations of this equation that we might expect for the noise expansion in later sections . the first paper to introduce asymptotic methods applied to a fractal problem @xcite discussed julia sets , but is more technically involved than the exactly selfsimilar fractals in refs . @xcite .the discussion in this section is based on ref .@xcite .the middle - third cantor set consists of two copies of itself scaled down by a factor of three , thus it has dimension @xmath6 according to many definitions .the uniform measure on the set located between @xmath7 satisfies the relation @xmath8d\mu(x)\ ] ] for arbitrary smooth function @xmath9 , which together with the definitions of the electrostatic potential @xmath10 and the moments @xmath11 lead after several steps @xcite to the following expression for the potential near the edge of the fractal , @xmath12 where @xmath13 , @xmath14 and @xmath15 are known series given in terms of the @xmath16 , which are in term given as explicit rational numbers with a well understood asymptotic form .the @xmath13 have a large exponential decay rate for example @xmath17 , @xmath18 , @xmath19 .the oscillatory terms come from the @xmath20 solutions @xmath21 of the equation @xmath22 defining the dimension .such `` complex dimensions '' of fractals appear in a number of physical applications @xcite . even though the context is different from that of the noise corrections considered below in sec .[ s : osc ] , in particular the expansion for the potential has a finite radius of convergence while that of the noise is divergent , we may conjecture that there are a number of general principles applying to fractal expansions .in particular , the leading order of the expansion is not a single term but a sum of terms containing a variable ( here @xmath23 ) raised to powers @xmath24 for all integers @xmath25 , where @xmath1 and @xmath26 are determined by properties of the fractal : 1 .the parameter @xmath1 ( here @xmath27 ) is related to the dimension of the fractal .the parameter @xmath26 ( here @xmath28 ) is determined by the spatial scaling factor @xmath29 of the fractal .3 . due to the rapid decay of the oscillatory coefficients ,accurate results may be obtained by considering only one or two of these terms , ie @xmath30 .note that in the absence of the imaginary @xmath26 terms ( perhaps a `` nonfractal limit '' ) , the @xmath1 in this section corresponds to the @xmath1 in eq .( [ e : asym ] ) since the latter effectively has @xmath1 in the exponent : @xmath31 as @xmath4 .this subsection gives a background to the numerical coefficients discussed in the main part of the paper .the periodic orbit theory @xcite allows the computation of long time properties such as averages , lyapunov exponents , and correlation functions from periodic orbits .the required properties are related to the leading eigenvalue(s ) of an evolution operator ( defined below ) , which , in the most rapidly convergent formulation , are computed using determinants .the determinant is expressed as an expansion in traces , and the traces are expressed in terms of periodic orbits .consider the discrete time dynamics described by @xmath32 where @xmath33 is the noise strength and @xmath34 are instances of an uncorrelated random variable . in the present case@xmath35 is a real number , but higher dimensional generalisations are straightforward , at least in the deterministic case @xmath36 .continuous time dynamics can also be considered ; the stochastic version has been discussed in ref .the probability density @xmath37 evolves according to @xmath38 where @xmath39 as defined above is the ( discrete time fokker - planck ) evolution operator .the noise distribution @xmath40 is an arbitrary function of standard deviation @xmath33 ; it reduces to a dirac delta in the deterministic case @xmath36 .we compute the trace @xmath41 which is an @xmath42-dimensional integral . in the deterministic case, it reduces to a sum over periodic points @xmath35 satisfying @xmath43 of the relevant jacobian .the weak noise theory is effectively a stationary phase approximation in which the leading order behaviour is given by the deterministic limit , with corrections determined by higher derivatives of the map @xmath9 evaluated at the periodic points .exponentially small corrections are obtained by considering local extrema , which are given by the `` generalised periodic orbits '' of @xcite , that is , periodic orbits of the extended system @xmath44 ; these will be discussed later .the characteristic determinant is @xmath45 where we define the determinant in terms of its expansion in powers of the inverse eigenvalue @xmath46 ; this is consistent with our desire for the largest eigenvalue ( smallest @xmath46 ) .the eigenvalue itself can be obtained by truncating the above equation at some @xmath47 , which requires computing periodic orbits up to length @xmath42 , and solving numerically for the first zero .the leading eigenvalue @xmath48 has a direct interpretation : the quantity @xmath49 is the escape rate , that is , a uniform distribution of initial conditions leads to a number proportional to @xmath50 remaining at long times @xmath51 .other useful quantities can be obtained by small modifications of the method , for example weighting the evolution operator using the function for which an average is desired .we note that because the eigenvalue is not directly expressed as an integral , rather an expansion of integrals or the infinite dimensional limit of an integral , the dingle theory discussed in sec .[ s : asym ] does not strictly apply . in principle( although not in practice ) an infinite number of critical points are required to determine the eigenvalue .there have been two analytical approaches to evaluating the integral for the trace in the stochastic case , in particular feynman diagrams @xcite and smooth conjugations @xcite .these give the trace explicitly in terms of the derivatives of @xmath9 at the periodic points , but have been applied only up to order @xmath52 .a more numerical approach was used in ref .@xcite to obtain coefficients up to @xmath53 , and it is the latter approach which is used in this paper . the evolution operator is expressed in an explicit polynomial basis ( in contrast to the usual situation in periodic orbit theory , where all calculations are kept independent of the basis ) , defined in the vicinity of each periodic point .truncating the representation to a finite matrix is justified since the eigenfunction is very smooth , leading to exponential decay of the matrix elements .all quantities are expanded in powers of the small parameter @xmath33 , and the full trace is obtained as a sum over periodic points , to each order in @xmath33 .finally the leading eigenvalue is obtained as a formal power series @xmath54 where odd powers vanish by the symmetry of the gaussian noise distribution used .further details of the calculation are given in ref .in the previous paper @xcite the results were limited to @xmath53 since high precision is required ( the cumulant expansion ( [ e : cum ] ) involves many cancellations ) , and commercial high precision mathematical packages required too much memory and time .the results of this paper were achieved by code written in c , involving 60 digits precision , a maximum matrix size of 200 ( the largest matrices are required for the shortest orbits ) and periodic points up to @xmath55 .the matrix size for each orbit length and the maximum length were determined adaptively ; the precision was estimated conservatively using the results of shorter calculations . here , as in previous calculations @xcite, the noise is gaussian , and the map appearing in ( [ e : dyn ] ) is @xmath56\ ] ] ( 290,240 ) ( 100,20 ) ( 200,0)@xmath35 ( 60,120)@xmath9 this map is axiom a with complete binary symbolic dynamics , so the rate of convergence of the cycle expansion with orbit length is super - exponential , both for @xmath57 ( the deterministic case ) and the noise corrections. examples of this convergence ( noted in the previous studies ) are given in tab .[ t : conv ] .the results for orbit length @xmath55 , expressed as the logarithms of the ( always positive ) @xmath58 are given in tab .[ t : res ] ..[t : conv ] the noise coefficients of the eigenvalue , as defined in eq .( [ e : num ] ) , calculated using periodic orbits up to length @xmath42 . [ cols=">,>,>,>,>,>",options="header " , ] to conclude the analysis of the results , we apply the shafer ( generalised pad ) and dingle ( borel summation ) methods discussed briefly in sec . [ s : asym ] and compare the results with the exact eigenvalue @xmath59 computed using the discretized eigenfunction method of ref .@xcite which is accurate to about six digits .note that the full power of the borel summation applied to the full computed series up to order @xmath60 is not ( currently ) testable , since this is the smallest term when it ( and hence the expected errors ) is of order @xmath61 .instead , we will consider quite large values of @xmath33 , where the smallest term is very close to the beginning of the series .the only slight extension of the dingle approach discussed in sec .[ s : asym ] concerns the modifying factor @xmath3 .since the asymptotic series now has three components , with @xmath62 , it is not clear whether each component should have a modifying factor .the point of view taken here is that a single modifying factor @xmath3 is used for all three components .this choice is pragmatic ; while it is probably more natural to modify each series separately , there is no method of extracting this information from the numerical data .( 290,250 ) ( 30,0 ) ( 240,0)@xmath33 ( 0,130)@xmath59 the results of the borel summation are shown in fig .[ f : resum ] .the bunch of lower solid curves are the borel summed series , using five forward differences ( see sec .[ s : asym ] ) and from two to four terms of the series before truncation .note that the borel summed function is consistent for relatively large @xmath33 , despite the obvious approximations made in approximating the series by so few terms .the dashed line is the result of a shafer ( quadratic pad ) approximation to the first 13 nonzero coefficients , as discussed in sec .[ s : asym ] ; the exact number of coefficients fitted makes little difference .these two interpretations of the power series are quite consistent .the dotted line is the true eigenvalue , computed as in ref .it is extremely close to the borel summed series for @xmath63 , after which it is significantly higher .the difference may be modelled by a function which is exponentially small for @xmath64 .the most obvious candidate ( but one of many ) is @xmath65 , which is roughly the magnitude of the smallest term , and is also the first expected `` hyperasymptotic '' correction .the remaining solid curve in fig . [ f : resum ] gives one of the borel summed curves , plus this function with @xmath66 .the dot - dashed curve gives the equivalent result for the shafer approximant .the result is a fit valid for roughly @xmath67 , indicating that the exponential part of the correction has the right form .it is reasonable to identify this exponentially small correction as the contribution of the generalised periodic orbit in sec .[ s : fit ] .however , there is actually a fractal set of such generalised periodic orbits , starting from the critical point of the map and limiting to each orbit on the repeller . the orbit given in sec .[ s : fit ] is the most probable case , but other orbits have exponents which are arbitrarily close .the nontrivial task of summing these contributions is left to a future paper .incidentally , there exist rigorous results pertaining to borel summation of asymptotic series @xcite .the observation of hyperasymptotic corrections implies the presence of singularities in the complex noise domain , however it is difficult to understand what physical consequences this might have .periodic orbit theory of stochastic systems as presented in ref .@xcite has been used to compute the escape rate of a stochastically perturbed map to order 64 in the noise strength with sufficient precision to permit the theory of asymptotic expansions to be applied .similar to a previous calculation of an exactly selfsimilar fractal , complex exponents appear , and the parameters in the expansion were found by a combination of analytic arguments and curve fitting . although connections between this and the previous calculation were at a phenomenological level , the precise fit gives strong numerical evidence that the form of the expansion is correct to this order .finally the shafer approximation and borel summation were performed on the series and compared with the known escape rate function , giving evidence for hyperasymptotic corrections of the same order as predicted by the theory of asymptotic expansions and independently by nonleading stationary points of the action . in the future, the analytic connections proposed between the singulant and the probability of returning to the repeller from the critical point , and between the imaginary exponent and the spatial scaling factor of the repeller , should be verified by calculations on a variety of different systems .the remaining tentative connection , between the real exponent @xmath1 and the @xmath68 dimension of the repeller would then either be verified or contradicted .the latter possibility is the most intriguing , since in that case @xmath1 could define a new `` noise dimension '' .the author is grateful for helpful discussions with michael berry , predrag cvitanovi , giovanni gallavotti , jon keating , jonathan robbins and niels sndergaard .this research was supported by the nuffield foundation , grant nal/00353/g .99 p. cvitanovi , r. artuso , r. mainieri , g. tanner and g. vattay , _ classical and quantum chaos _ , www.nbi.dk/chaosbook/ , niels bohr institute ( copenhagen 2003 ) .p. cvitanovi , c. p. dettmann , r. mainieri and g. vattay , j. stat . phys . * 93 * , 981 ( 1998 ) .p. cvitanovi , c. p. dettmann , r. mainieri and g. vattay , nonlinearity * 12 * , 939 ( 1999 ) .p. cvitanovi , n. sndergaard , g. palla , g. vattay and c. p. dettmann , phys .e * 60 * , 3936 ( 1999 ) .g. palla , g. vattay , a. voros , n. sndergaard and c. p. dettmann , found .phys . * 31 * 641 ( 2001 ) . c. p. dettmann , phys .e * 59 * , 5231 ( 1999 ) .j. p. boyd , acta applic .math . * 56 * , 1 ( 1999 ) .r. e. shafer , siam j. numer .11 * , 447 ( 1974 ) .m. v. berry and c. j. howls , proc .london a , * 430 * 653 ( 1990 ) .r. b. dingle _asymptotic expansions : their derivation and interpretation _( academic press , london , 1973 ) . m. abramowitz and i. a. stegun _ handbook of mathematical functions _( national bureau of standards , 1964 ) .d. bessis , j. s. geronimo and p. moussa , j. stat . phys . * 34 * , 75 ( 1984 ) .c. p. dettmann and n. e. frankel , j. phys .a : math . gen .* 26 * , 1009 ( 1993 ) . c. p. dettmann and n. e. frankel , j. stat. phys . * 72 * , 241 ( 1993 ) .h. saleur and d. sornette , j. phys .i * 6 * , 327 ( 1996 ) .p. gaspard , j. stat .phys . * 106 * , 57 ( 2002 ) .g. h. hardy _ divergent series _( clarendon , oxford , 1949 ) .

<S> recent advances in the periodic orbit theory of stochastically perturbed systems have permitted a calculation of the escape rate of a noisy chaotic map to order 64 in the noise strength . </S><S> comparison with the usual asymptotic expansions obtained from integrals and with a previous calculation of the electrostatic potential of exactly selfsimilar fractal charge distributions , suggests a remarkably accurate form for the late terms in the expansion , with parameters determined independently from the fractal repeller and the critical point of the map . </S><S> two methods give a precise meaning to the asymptotic expansion , borel summation and shafer approximants . </S><S> these can then be compared to the escape rate as computed by alternative methods . </S><S> _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ keywords : asymptotic expansions , borel summation , pad approximation , cycle expansions , escape rates , fractals , maps , repellers , stochastic perturbations . </S><S> + pacs : 02.30.mv 05.40.ca 05.45.df _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ </S>

quant-ph0211065

there has been considerable interest in recent years for the fascinating properties of coherently prepared atomic media @xcite . among the most studied coherent effectsis the phenomenon of electromagnetically induced transparency ( eit ) @xcite .eit has generally been modelled and experimentally studied in three - level @xmath7 systems with two long lived lower ( ground ) states and a rapidly decaying upper ( excited ) state .the two electromagnetic fields ( pump and probe ) separately couple each of the two arms of the @xmath7 system . a distinctive feature of the eit resonance is its narrow linewidth which corresponds to the coherence decay rate of the ground state doublet .the occurrence of the eit resonance is directly linked to the existence of a dark state ( ds ) : i.e. a linear combination of the two ground states uncoupled to the excited state .eit is a consequence of the system being pumped into the ds .eit may also be observed in multi - level systems as those involving the zeeman substates of two atomic levels with angular momentum degeneracy , hereafter called degenerate two - level systems ( dtls ) . in such case too , the observation of eit is a direct consequence of the existence of a ds within the lower atomic level when @xmath8 ( @xmath9 and @xmath10 are the angular momenta of the ground and excited state respectively ) .the pump - probe spectroscopy of dtls with @xmath11 also presents resonances in the probe transmission when the raman resonance condition between ground state zeeman sublevels is fulfilled .however , in this case , the resonances correspond to an increase of the probe absorption and have consequently being designated as electromagnetically induced absorption ( eia ) @xcite . as in the case of eit resonances ,the eia linewidth is given by the coherence decay rate of the ground level .unlike eit , the eia resonances can not be associated to the existence of a ds in the ground state .also , there is no simple connection between the eia resonances and the existence of a ds in the _ excited _ state since such state has a lifetime limited by spontaneous emission and can not account for the narrow spectral features observed .an explanation for the eia resonance was provided by taichenachev et al @xcite analyzing a four level system in a @xmath12 configuration .they showed theoretically that in this system , which is the simplest to present eia , the enhanced absorption is due to the transfer , via spontaneous emission , of the coherence created by the exciting fields within the two upper levels .although the @xmath12 configuration does not correspond to the configurations actually explored in realistic dtls , the argument presented in @xcite can be extended to such systems @xcite .the purpose of this paper is to provide experimental evidence in support of the argument presented in @xcite by demonstrating that eia resonances are suppressed ( and even reversed ) if the coherence of the excited state is significantly destroyed by collisions before the occurrence of the spontaneous emission decay . a simple theoretical model in good agreement with the observations is also presented .a convenient experimental scheme for the observation of coherence resonances is the hanle type setup that uses a unique optical beam with linear polarization in near resonance with an atomic transition . in this scheme ,the two opposite circular polarization components of the light can be considered as the pump and probe fields .the light beam is sent through an atomic sample where the raman resonance condition is tuned , via the zeeman effect , by a magnetic field along the light propagation axis .the intensity of the transmitted light is monitored .hanle / eit / eia resonances on the @xmath13 lines of alkaline vapors were recently studied by several groups @xcite ( see inset in fig . [ isotope ] for an energy level scheme of the @xmath1rb @xmath13 line ) .dancheva et al @xcite were the first to observe a hanle / eia resonance on a @xmath14 transition of the @xmath0 line .they pointed out the fact that the eia resonance can be observed in spite of the transition being open ( radiative decay can occur from the excited state to either ground state hyperfine levels ) .the open character of the transition is responsible for the smallness of the hanle / eia resonance as compared with the hanle / eit resonances occurring when @xmath15 .the choice of the hanle experimental scheme for coherent spectroscopy is not only motivated by its simplicity .it also allows a special insight on the connection between coherence resonance and optical pumping .consider the situation depicted in fig .[ levelschemes ] .[ levelschemes]a ) and [ levelschemes]b ) correspond to a @xmath16 and @xmath17 transitions respectively . in both cases ,the quantization axis has been chosen parallel to the direction of the light propagation which coincides with the orientation of the magnetic field .the raman resonance condition is achieved for the two components of the optical field at zero magnetic field and destroyed for nonzero magnetic field . in fig.[levelschemes]a ) the optical field interact with a @xmath7 system for which eit is known to occur as a consequence of the existence of a ds formed by the anti - symmetric combination of the ground state zeeman sublevels @xmath18 and @xmath19 .the situation in fig .[ levelschemes]b ) corresponds to eia which results , according to @xcite , from the transfer of coherence from the upper to the lower level via spontaneous emission . figs .[ levelschemes]c ) and [ levelschemes]d ) represent the same physical situation than in a ) and b ) respectively with the quantization axis taken along the direction of the linear polarization of the light . in fig .[ levelschemes]c ) the transparency observed for zero magnetic field is the consequence of the optical pumping of the system into the @xmath18 and @xmath20 sublevels which are not coupled to the excited state . in fig .[ levelschemes]d ) the absorption increase for @xmath21 is a consequence of the redistribution of the population of the ground state zeeman sublevels ( alignment ) via optical pumping .the atomic population accumulates preferentially in the @xmath22 sublevel which has the largest coupling with the excited state . noticethat with this choice of the quantization axis , no coherence is built among the excited state zeeman sublevels and thus can not be transferred to the ground state . in the basis used in c ) and d ) the hanle / eit and hanle / eia resonances appear as a consequence of incoherent optical pumping .since the choice of the quantization axis is arbitrary , the two frames considered in the previous discussion can conveniently be considered for the analysis of the hanle / eia resonances under the effect of collisions affecting the excited state . in the scheme of fig .[ levelschemes]b ) which can be considered as an extension of the simple @xmath12 system analyzed by taichenachev et al @xcite , eia is a consequence of the spontaneous transfer of zeeman coherence from the excited to the ground state . in consequence ,the eia resonance should disappear if the excited state coherence is destroyed by collisions in a time comparable or shorter than the excited state lifetime . in the basis corresponding to fig .[ levelschemes]d ) eia can be prevented if the collisions are responsible for a significant equalization ( thermalization ) of the excited state sublevel populations before the occurrence of spontaneous emission ( fig .[ levelschemes]e ) .if the excited state is completely thermalized then the total spontaneous emission decay into a given ground state zeeman sublevel is the same for all ground state sublevels . in consequence , in the steady state , the population is preferentially accumulated in sublevels @xmath23 resulting in an increased transparency . to demonstrate the influence of the excited state coherence on eia we have observed the hanle / eit / eia resonances corresponding to the four transitions of the @xmath13 line of @xmath1rb both in a vapor cell containing only rb vapor ( negligible collisions ) and in a cell containing @xmath4 of ne as buffer gas .it is well known that room temperature collisions with light noble gas atoms ( he , ne , ar ) produce different effects on the @xmath24 ground state than in the @xmath25 or @xmath26 excited states of several alkaline vapors @xcite .the collisions of a noble gas atom with an alkaline atoms in its ground state has little effect on its electronic and nuclear spin . as a consequencethe alkaline atom can experience a very large number of collisions while preserving the ground state coherence .however , the atomic motion is affected by collisions becoming diffusive .this results in a longer interaction time before the atoms can leave the light beam or reach the cell walls .the increased interaction time allows the observation of coherence resonances with time - of - flight limited linewidth of only a few tenth of @xmath27 @xcite .the situation is rather different for collisions between noble gas atoms and alkaline atoms in the @xmath28 state@xcite . at room temperature and for the buffer gas density corresponding to the experiments described below, the collisions are sufficiently energetic and frequent to produce a considerable broadening of the homogeneous width of the optical transitions ( which remains nevertheless smaller than the doppler width and the @xmath28 hyperfine splitting ) . also under such conditionsvirtual transitions to the neighboring levels ( mainly @xmath29 ) occur @xcite resulting in the non preservation of the magnetic quantum number ( @xmath30 ) of the electronic angular momentum during collisions . as a consequence ,a significant thermalization of the excited state density matrix take place in a time shorter than the excited state lifetime .the cross section for excited state collisional @xmath30 mixing was measured by several authors @xcite .using the figures in @xcite and the ne density corresponding to the experiments the collisional decoherence rate of the excited state @xmath31 can be estimated as @xmath32 where @xmath33 is the excited state spontaneous emission decay rate .the experimental setup is shown in fig .[ setup ] . we have used two cylindrical vapor cells of the same dimensions ( diameter : @xmath34 , length : @xmath35 ) provided by the same manufacturer and filled with natural rubidium vapor .one of the cells contains @xmath4 of ne as buffer gas .the cell under consideration was placed ( at room temperature ) inside a magnetic shield formed by three coaxial cylindrical @xmath36-metal layers .after degaussing of the @xmath36-metal shield the total inhomogeneity of the residual magnetic field at the vapor cell was less than @xmath37 .a coaxial solenoid , internal to the magnetic shield was used to scan the magnetic field at the atomic sample .an extended cavity diode laser ( @xmath38 linewidth ) was used for the atomic excitation .the laser frequency was monitored and stabilized on specific hyperfine transitions with the help of an auxiliary saturated absorption setup .the total laser power at the sample was approximately @xmath39 .the light transmitted through the vapor cell was monitored with a photodiode ( @xmath40 bandwidth ) . in order to enhance sensitivitythe coherence resonances were detected through a parametric resonance technique @xcite .in addition to the slowly varying dc magnetic field , a small ac component of the magnetic field ( oriented along the light propagation axis ) was introduced with a secondary coil driven at frequency @xmath41 .a lock - in amplifier detected the phase and quadrature components of the photodiode current oscillating at frequency @xmath41 .two different values of @xmath41 were used in the measurements .for the cell without buffer gas we used @xmath42 and for the cell containing ne @xmath43 was used . in both cases @xmath41 was chosen to exceed the width of the corresponding hanle resonances . a typical recording of the lock - in output signal as a function of the solenoid dc currentis shown in fig .[ expbuffbroad ] .the central structure of the spectrum corresponds to @xmath21 .the spectrum sidebands occurring for @xmath44 ( @xmath45 is the ground state zeeman energy shift ) allow a precise calibration of the magnetic field inside the solenoid . in the cell containing the buffer gas where the resonances are well resolved ( fig .[ expbuffbroad ] ) the parametric resonance scheme used in the experiment allows for isotope selective spectroscopy . for this , the dc magnetic field is kept fixed at the value corresponding to the maximum of one of the ( lorentzian ) sidebands resonances observed on the quadrature signal and the quadrature output of the lock - in amplifier is monitored while the frequency of the laser is scanned on the @xmath13 line .since the resonance condition @xmath46 depends on the specific isotope through the gyromagnetic factor @xmath47 ( @xmath48 is the bohr magneton ) , only one isotope contributes to the observed spectrum .an example of such spectrum is shown in fig .[ isotope ] where only the @xmath13 transitions of @xmath1rb appear in spite of the cell being filled with natural rb ( @xmath49 of @xmath50rb ) .the gyromagnetic factors of the two hyperfine levels of the ground state of @xmath1rb differ in absolute value by less than @xmath51 .consequently all four hyperfine transitions can be observed in the same spectrum .notice that the hyperfine structure is well resolved in the spectrum of fig.[isotope ] .this is a clear indication that in spite of the rather strong collisional regime , the atomic level structure resulting from the hyperfine coupling is preserved and that the total angular momentum @xmath52 remains a good quantum number .the relative weight of the coherence resonances in the four @xmath13 transitions is appreciated in fig .[ isotope ] .the peak corresponding to the @xmath17 transition is smaller than that of the @xmath53 transition by a factor @xmath54 and is not visible on the scale of fig .[ isotope ] .it is worth mentioning that the small @xmath17 signal is nevertheless @xmath55 times larger than the residual signal due to the wings of the @xmath56 peak estimated from a gaussian fit. figs .[ expnobuffer ] and [ expbuffer ] show the observed signals with the laser in resonance with the four @xmath13 transitions in the cell without buffer gas and in the cell containing ne respectively using the same optical power .the beam diameter was @xmath57 and @xmath58 respectively .only the central resonance around @xmath21 for the in phase signal is shown in fig .[ expbuffer ] . without buffer gasthe observed resonances have a width which is determined by the time of flight through the optical beam .notice the sign reversal of the @xmath59 transition ( eia ) with respect to the three other resonances ( eit ) .in the cell with buffer gas ( fig .[ expbuffer ] ) , the observed resonances are much narrower as expected from the increase in interaction time due to the diffusive atomic motion . however , under the present experimental conditions the observed linewidth is limited by residual magnetic field inhomogeneities . a sign change is clearly observed for the @xmath59 transition with respect to the cell without ne demonstrating the quenching of the eia resonance and its reversal into eit .we present in this section a simple theoretical model containing the essential ingredients for the analysis of hanle / eit / eia coherence resonances in dtls .the model takes explicitly into account the zeeman degeneracy of the ground and excited states .it considers a unique electric dipole allowed atomic transition in a homogeneous sample of atoms at rest . with respect to the conditions of the experiment ,the model contains several simplifications .the effect of the atomic motion , the spatial distribution of the light field , the light propagation in the sample and the influence of neighboring transitions are not considered .also the effect of collisions is introduced in an quite simplified way through a single collisional relaxation rate as discussed below .we consider a two - level transition between a ground state @xmath47 of angular momentum @xmath60 and an excited state @xmath61 of angular momentum @xmath52 with energy separation @xmath62 .the atoms are illuminated by an optical field of amplitude @xmath63 and frequency @xmath64 linearly polarized along the unit vector @xmath65 and submitted to a magnetic field @xmath66 perpendicular to @xmath65 .the atoms are submitted to collisions with the buffer gas .it is assumed that the collisions result in dephasing of the atomic optical dipole and can cause real transitions between excited state zeeman sublevels . inthe frame rotating with the optical field and with the usual rotating wave approximation , the liouville equation for the density matrix @xmath67 of the system is : @xmath68 -\frac{\left ( \gamma + \gamma _ { coll}\right ) } { 2}\left\ { p_{e},\sigma \right\ } \nonumber \\ & & + b\gamma \sum_{q=-1,0,1}q_{ge}^{q}\sigma q_{eg}^{q}-\gamma \sigma + \gamma \sigma _ { 0}+\gamma _ { coll}\bar{\sigma } \label{liouville}\end{aligned}\ ] ] where @xmath69 , @xmath70 is the zeeman hamiltonian ( @xmath71 and @xmath72 are projectors on the ground and excited subspaces respectively , @xmath73 and @xmath74 are the gyromagnetic factors of the ground and excited states respectively , @xmath48 is the bohr magneton and @xmath75 the projection of the total angular momentum along the direction of the magnetic field ) .the atom field interaction is given by : @xmath76 @xmath77 is a dimensionless vectorial operator related to the electric dipole operator @xmath78 through : @xmath79 @xmath80 is the reduced rabi frequency of the optical field ( @xmath81 is the reduced matrix element of the electric dipole operator for the considered transition ) .@xmath82 are the spherical components of the operator @xmath83 .@xmath84 is a branching ratio coefficient ( @xmath85 ) .@xmath33 is the spontaneous emission decay rate . @xmath86 and @xmath31 are relaxation rates associated to transit time and excited state collisions respectively .@xmath87 corresponds to an isotropic density matrix for the ground level with unit total population and @xmath88 is an ( incoherent ) isotropic density matrix for the excited state with the same excited state population than @xmath67 . the first term on the r.h.s . of eq .[ liouville ] describes the hamiltonian evolution of the atom in the presence of the optical and magnetic field .the second term account for the relaxation of the excited state and the optical coherences .in addition to spontaneous emission we consider the relaxation due to collisions with the buffer gas atoms . for simplicity, it is assumed that the effect of buffer gas collisions on optical and zeeman coherences can be described by the single relaxation rate @xmath31 . the third term on the r.h.s . of eq .[ liouville ] describes the spontaneous emission transfer of population and coherence from the excited state to the ground state . for open transitions , the branching ratio coefficient @xmath84 accounts for the atomic loss due to radiative transitions to external levels ( @xmath89 corresponds to closed transitions ) . the fourth term on the r.h.s . of eq .[ liouville ] accounts , in the usual phenomenological way , for the finite interaction time .although this relaxation term concerns both the ground and the excited level , notice that @xmath90 is the only relaxation rate acting on the ground state .the escape of atoms from the interaction region at rate @xmath86 is compensated at steady state by the arrival of fresh atoms isotropically distributed in the ground state ( @xmath91 in eq .[ liouville ] ) . the last term on the r.h.s . of eq .[ liouville ] is an effective isotropical repumping term introduced to compensate the effect of the collisions ( included in the second term of the r.h.s . of eq .[ liouville ] ) on the total excited state population . the steady state solution of eq .[ liouville ] can be easily obtained numerically @xcite .the solution of eq .[ liouville ] for conditions corresponding to the parametric resonance scheme where the magnetic field along the light propagation axis is the sum of a dc and a small sine modulated component is presented in the appendix .the calculation of the light absorption signals in phase and in quadrature with respect to the magnetic field modulation are also derived in the appendix .the prediction of the model for the transitions of the @xmath13 line of @xmath92rb are presented in fig .[ teo ] . only the @xmath93 and @xmath59 transitions are shown .the parameters of the calculation were chosen to correspond approximately to the experimental conditions .the same light intensity is assumed in all figures .we used @xmath94 ( the two transitions shown have the same strength ) .the branching ratio was taken as @xmath95 and @xmath96 respectively ( evaluated through standard angular momentum calculus ) and the signal corresponding to each transition was weighted proportionally to the thermal occupation number of the lower level ( coefficients @xmath97 and @xmath98 respectively ) . the plots a - d in fig .[ teo ] correspond to the absence of collisions . in these plotsthe transit time relaxation rate was taken as @xmath99 . while plots a , b , e , f correspond to the direct output of our model for @xmath100 , plots c and d correspond to the signal averaged over @xmath101 in the range @xmath102 to include the contribution of different velocity classes to the signal ( @xmath103, @xmath104 is the wavevector and @xmath105 the velocity ) .when different from zero , the collisional rate was taken as @xmath106 a value estimated from the data in @xcite .the precise value of @xmath31 has small influence on the lineshape of the signal provided that it is taken sufficiently larger than @xmath33 .notice the sign reversal on the calculated signal corresponding to the @xmath59 transition when @xmath107 with respect to the case in which @xmath108in spite of the simplicity of the model , the main features of the experimental observations are well reproduced in the calculated spectra ( similar agreement is obtained for the two hyperfine transitions not shown ) .the variation in the amplitude of the signal between the two hyperfine transitions is essentially the consequence of the different values of the branching ratio @xmath84 . in the plots a and b of fig .[ teo ] a non resonant contribution is clearly visible giving rise to the observed slope in the quadrature signal .such non resonant contribution is negligible on the scale of plots e and f. we attribute this difference to the fact that the non resonant contribution originates in the linear response of the atomic sample while the resonances are due to its nonlinear response . in an open two level system ,the saturation intensity @xmath109 is of the order of @xmath110 . with the values of @xmath86 used in the two cases considered above, we have @xmath111 for the pure rb vapor ( plots a - d ) and @xmath112 for plots e and f. this explains the different relative weight of the linear and nonlinear contributions in the two cases . since the coherence resonance condition is essentially doppler - free , only when @xmath111 the integration over different detunings @xmath101 ( integration over velocity classes ) result in a significant change in the lineshape .after integration , the nonresonant background is cancelled ( assuming that the laser is centered on the doppler profile ) in agreement with the experimental observation .several simplifications were assumed in the theoretical model .the approximation consisting in considering an homogeneous sample of atoms at rest have already been addressed ; when necessary this simplification can be abandoned carrying on the integration over velocity classes as described above .another important approximation is the simplified treatment of the collisional process described through a unique scalar rate .also , it is assumed in the model that the atom remain confined to the excited state level during the collisional process .however , for the ne density used in the experiment , inelastic collisions have a non - negligible probability and a fraction of the excited atoms may be collisionaly transferred to a different excited hyperfine level or even to the @xmath113 manifold @xcite . in any case ,such inelastic processes are not expected to preserve the zeeman coherence of the excited levels .the overall consequence of the inelastic collision processes , followed by spontaneous emission , is to provide a second ( indirect ) path for the return of the atoms to the ground state . with the assumption that this additional decay channel is isotropic its contribution can be effectively included in the collisional rate @xmath114 .we have studied hanle / eit and hanle / eia resonances on the @xmath13 line of @xmath115rb vapor both in absence and in presence of a buffer gas .the coherence resonances were studied through a parametric resonance technique .we observed that the hanle / eia resonance , occurring for the @xmath17 transition , change sign in the presence of a buffer gas as a consequence of collisions in the excited state .a theoretical model was presented that allows the numerical calculation of the parametric resonance signals . in spite of its simplicity, the model reproduces the essential features of the experimental results on the assumption that the collision of the rb atoms with the buffer gas result in the isotropic decoherence of the excited state . the overall agreement between the experimental results andthe simple model clearly suggest that the preservation of the excited state coherence during the excited state lifetime and its transfer to the ground state are , as suggested in @xcite , key ingredients for the occurrence of eia .the authors are thankful to d. bloch for fruitful discussions .this work was supported by the uruguayan agencies : conicyt , csic and pedeciba and by ecos ( france ) .in this appendix we describe the calculation of the parametric resonance signals .we consider the evolution of the atom - optical field system under the influence of a longitudinal magnetic field of the form : @xmath116 we seek a solution for the density matrix on the form : @xmath117 where @xmath118 is constant and @xmath119 is a time dependent correction to first order in @xmath120 . substituting in eq .[ liouville ] , we have : @xmath121 -\frac{\left ( \gamma + \gamma _ { coll}\right ) } { 2}\left\ { p_{e},\sigma ^{0}\right\ } \nonumber \\ & & + b\gamma \sum_{q=-1,0,1}q_{ge}^{q}\sigma ^{0}q_{eg}^{q}-\gamma \sigma ^{0}+\gamma \sigma _ { 0}+\gamma _ { coll}\overline{\sigma ^{0 } } \label{orderzero}\end{aligned}\ ] ] @xmath122 + \frac{i}{\hbar } \left [ m_{z}b_{1},\sigma ^{0}\right ] \cos \left ( \delta t\right ) \nonumber \\ & & -\frac{\left ( \gamma + \gamma _ { coll}\right ) } { 2}\left\ { p_{e},\sigma ^{1}\right\ } + \gamma _ { coll}\overline{\sigma ^{1 } } \label{orderone } \\ & & + b\gamma \sum_{q=-1,0,1}q_{ge}^{q}\sigma ^{1}q_{eg}^{q}-\gamma \sigma ^{1 } \nonumber\end{aligned}\ ] ] if one identifies the matrix elements of @xmath123 and @xmath124 with the components of column vectors @xmath125 and @xmath126 , then eqs . [ orderzero ] and [ orderone ] can be rewritten as : @xmath127 @xmath128 where @xmath129 is a matrix with time independent coefficients and @xmath130 and @xmath131 are two column vectors associated to @xmath132 and @xmath133 $ ] respectively .after numerical evaluation of vectors @xmath138 and @xmath139 one can retrieve the corresponding density matrices @xmath140 and @xmath141 and the corresponding light absorption coefficients respectively in phase and in quadrature with the magnetic field modulation : @xmath142 \;\ \left ( i=\alpha , \ \beta \right ) \label{phaseandquad}\ ] ]

<S> the explanation presented in [ taichenachev et al , phys . </S><S> rev . </S><S> a * 61 * , 011802 ( 2000 ) ] according to which the electromagnetically induced absorption ( eia ) resonances observed in degenerate two level systems are due to coherence transfer from the excited to the ground state is experimentally tested in a hanle type experiment observing the parametric resonance on the @xmath0 line of @xmath1rb . </S><S> while eia occurs in the @xmath2 transition in a cell containing only @xmath3 vapor , collisions with a buffer gas ( @xmath4 of @xmath5 ) cause the sign reversal of this resonance as a consequence of collisional decoherence of the excited state . a theoretical model in good qualitative agreement with the experimental results </S><S> is presented . </S><S> # 1@xmath6#1 </S>

1212.2331

geometric function theory makes use of several metrics for subdomains of @xmath1 .it has turned out that sometimes hyperbolic metric or more generally , hyperbolic type metrics , are more natural than the euclidean metric , because of their better invariance or quasiinvariance properties under well - known classes of mappings such as mbius transformations , bilipschitz or quasiconformal maps . in the recent yearsmany authors have contributed to this field , see e.g. @xcite . on the other hand ,hyperbolic type metrics are sometimes difficult to estimate and it is desirable to find simple expressions to serve as comparison functions .two such expressions , the visual angular metric and the triangular ratio metric , were recently studied in @xcite . hereour goal is to continue the study the triangular ratio metric for proper subdomains of @xmath2 .in particular , we study the local convexity of balls in this metric for some simple domains .for some other metrics , results of this type were recently proved by r. kln in @xcite , in answer to a question posed in @xcite .we study also inclusion relations between balls in different metrics .some of the metrics we study are the triangular ratio metric , the quasihyperbolic metric and the @xmath0-metric .we consider the inclusion relations in general domains as well as in some specific examples as the punctured space and the half - space .these kind of results for hyperbolic type metrics have been studied in @xcite . for a domain @xmath3 , and @xmath4, we define the triangular ratio metric @xcite by @xmath5,\ ] ] the @xmath0-metric @xcite by @xmath6 where @xmath7 , and the quasihyperbolic metric @xcite by @xmath8 where the infimum is taken over all rectifiable curves @xmath9 joining @xmath10 and @xmath11 .martin and b.g .osgood proved in @xcite the following formula for the quasihyperbolic distance : @xmath12 where @xmath13 .for a metric space @xmath14 we define the metric ball for @xmath15 and @xmath16 by @xmath17 . in sections 3 - 5 we consider local convexity properties of balls in triangular ratio metric . in section 3we consider the punctured space @xmath18 , in section 4 the half - space @xmath19 and in section 5 the punctured half - space @xmath20 and polygons @xmath21 . in sections 6 and 7we study the inclusion of balls defined by the triangular ratio metric , the quasihyperbolic metric and the @xmath0-metric . in section 6we consider these metrics in general domains and in section 7 in the punctured space and in the half - space .our main results are the following two theorems .[ mainthm1 ] ( 1 ) let @xmath22 , @xmath23 , @xmath15 and @xmath24 .the metric ball @xmath25 is ( strictly ) convex if @xmath26 @xmath27 .\(2 ) let @xmath28 and @xmath29 .then @xmath30 and @xmath31 is thus strictly convex .\(3 ) let @xmath32 with @xmath33 and @xmath34 $ ] , where @xmath35 then @xmath31 is convex .\(4 ) let @xmath36 be a polygon and @xmath37 .then @xmath31 is convex for all @xmath38 $ ] .moreover , if @xmath39 is convex then @xmath31 is convex for all @xmath40 .[ mainthm2 ] ( 1 ) suppose that @xmath41 is a domain .for each @xmath42 and @xmath43 , we have @xmath44 \(2 ) suppose that @xmath41 is a domain . for each @xmath42 and @xmath43 , we have @xmath45 and the inclusion is sharp if there exists some point @xmath46 in @xmath47 such that @xmath48 and @xmath49 moreover , for each @xmath42 and @xmath50 , we have @xmath51 \(3 ) let @xmath52 . for each @xmath42 and @xmath43 , we have @xmath53 moreover , the inclusions are sharp .\(4 ) let @xmath52 .for each @xmath42 and @xmath43 , we have @xmath54 where @xmath55 , and @xmath56 . moreover , the inclusions are sharp .[ slopedef ] let us have a polar function @xmath57 .then the slope of the curve @xmath57 at the point @xmath58 is @xmath59 the supremum in the definition of @xmath60 is attained at a point @xmath61 that is on a maximal ellipsoid with focii at @xmath10 and @xmath11 . therefore it is clear that @xmath60 is monotone with respect to the domain , which means that if @xmath62 and @xmath63 are domains and @xmath64 then @xmath65 . by the definition and the monotonicity with respect to domainit is easy to prove that for all @xmath15 and @xmath40 @xmath66 we denote the hyperbolic distance by @xmath67 , where @xmath68 is either the unit ball @xmath69 or the upper half - space @xmath19 ( @xcite , @xcite ) . [ le61 ]let @xmath70 .+ ( 1 ) the function @xmath71 is strictly decreasing from @xmath72 onto @xmath73 .+ ( 2 ) the function @xmath74 is strictly decreasing from @xmath75 onto @xmath76 .+ ( 3 ) the function @xmath77 is strictly decreasing from @xmath75 onto @xmath78 .+ ( 4 ) the function @xmath79 is increasing from @xmath80 onto @xmath81 .\(1 ) let @xmath82 and @xmath83 .it is clear that @xmath84 . by differentiation ,@xmath85 which is strictly decreasing in @xmath72 .therefore , we get the monotonicity of @xmath86 .the limiting values are clear by lhpital s rule .\(2 ) by differentiation , @xmath87 which is negative .therefore , @xmath88 is strictly decreasing .the limiting values are clear .\(3 ) let @xmath89 and @xmath90 .it is clear that @xmath91 . by differentiation , @xmath92 by differentiation, we have @xmath93 then @xmath94 is strictly decreasing in @xmath75 . therefore , we get the monotonicity of @xmath95 .the limiting values are clear by lhpital s rule .\(4 ) by differentiation , @xmath96 therefore , @xmath97 is strictly increasing .the limiting values are clear .we consider the metric @xmath60 in the punctured space @xmath98 .we first compare @xmath60 with the hyperbolic metric @xmath99 and then study convexity of metric balls @xmath31 . by the definitionit is clear that for @xmath100 , we have @xmath101 this special case of the triangular ratio metric has been studied in @xcite .the inequality @xmath102 holds for all @xmath103 . without loss of generality, we may assume that @xmath104 .then for all @xmath105 , by lemma [ le61 ] ( 1 ) , we have @xmath106 equality holds if @xmath107 are collinear and @xmath108 .we notice that @xmath109 for @xmath110 .a natural problem is that is it true that @xmath111 holds for all subdomains @xmath41 ?we only get the following constant which is larger than @xmath112 but less than @xmath113 .but we note that very recently the inequality @xmath114 was proved in @xcite .suppose that @xmath41 is a domain .then for all @xmath115 we have @xmath116 for given@xmath115 , we may assume that @xmath117 . then if @xmath118 , obviously we have @xmath119 if @xmath120 , then @xmath121 because by a simple computation @xmath122 for @xmath123 $ ] .let us next consider convexity of metric balls .[ ekalemma ] for all @xmath124 and @xmath125 $ ] we have @xmath126 let us write the above inequality in the following form @xmath127 first we need to show that @xmath128 is strictly increasing . for thiswe need @xmath129 in order to show that @xmath130 we need to prove following inequality @xmath131 because @xmath132 it is easy to see that @xmath133 for all @xmath124 and @xmath125 $ ] .since @xmath134 is strictly increasing we see that @xmath135 and with a simple computation we get @xmath136 now we have shown that @xmath128 is strictly increasing . therefore @xmath137 and the assertion follows .[ nonconvextheorem ] let @xmath22 , @xmath23 , @xmath15 and @xmath138 $ ] .the metric ball @xmath25 is nonconvex if @xmath139 . by symmetryit is sufficient to consider only the case @xmath140 , @xmath141 and @xmath142 .let @xmath143 $ ] , @xmath144 and denote @xmath145 .now @xmath146 is equivalent to @xmath147 by the law of cosines we obtain @xmath148 which is equivalent to @xmath149 [ kuva1 ] in the proof of theorem [ nonconvextheorem].,title="fig:",width=377 ] therefore we get @xmath150 and by solving for @xmath151 we obtain @xmath152 in this proof we select @xmath153 next we solve the following inequality to check , which values of @xmath154 are interesting @xmath155 with a simple substitution @xmath156 we get @xmath157 it is clear that if @xmath158 , then @xmath159 .therefore it is enough to focus on angles @xmath160 .the slope of the tangent of @xmath161 with respect to @xmath154 according to the definition [ slopedef ] is @xmath162 where @xmath163 since @xmath164 is symmetric with respect to @xmath165-axis , we need to show that @xmath166 for some @xmath160 .it is clear that for all @xmath160 the inequality @xmath167 holds .now we need to show that for some @xmath160 the inequality @xmath168 holds . by a simple computation we see that @xmath169 and @xmath170 for all @xmath171 .therefore @xmath172 for some sufficiently small @xmath154 .now we have shown that @xmath166 for some @xmath160 and the assertion follows .[ punctspace ] let @xmath22 , @xmath23 , @xmath15 and @xmath24 .the metric ball @xmath25 is ( strictly ) convex if @xmath26 @xmath27 . by symmetryit is sufficient to consider only the case @xmath140 , @xmath141 and @xmath142 .let us use the following notation obtained in ( [ bothparts ] ) @xmath173 in the proof of theorem [ nonconvextheorem ] we showed that for @xmath174 the slope of the tangent of @xmath161 is @xmath175 and @xmath176 if @xmath158 . with thisit is easy to see that @xmath177 similarly , for @xmath178 we write @xmath179 and @xmath180 where @xmath181 if @xmath182 . to prove the theorem we need to show that @xmath183 and @xmath184 for all @xmath124 where @xmath125 $ ]firstly , we get @xmath185 where @xmath186 and @xmath187 . because @xmath188 it is easy to see that for all @xmath124 and @xmath125 $ ] the following inequality @xmath189 holds . to prove that @xmath183 we need to show that the inequality @xmath190 holds .it is easy to see that @xmath191 for all @xmath124 and @xmath125 $ ] . in lemma [ ekalemma ]we showed that @xmath192 therefore @xmath183 .for @xmath193 we get @xmath194 because @xmath195 we clearly see that @xmath196 for all @xmath124 and the assertion follows .[ kuva3 ] in @xmath197 where @xmath142 and @xmath198 and @xmath199.,title="fig:",width=453 ]we consider the triangular ratio metric in half - space @xmath200 .we compare first @xmath60 with @xmath99 and then prove that the metric balls @xmath31 are also euclidean and thus always convex . by the definition we obtain that for @xmath201 @xmath202 where @xmath203 . by figure [ angmetfig9 ]it is clear that the supremum in is attained at the point @xmath61 . in @xmath19 as in .[angmetfig9 ] , width=264 ] [ beardonprop ] then equality @xmath204 holds for all @xmath205 .the assertion follows from and ( * ? ? ?* ( v ) ) .[ beardonprop1 ] balls in the triangular ratio metric are euclidean balls . by (* ( 2.11 ) ) hyperbolic balls in @xmath19 are euclidean balls and by proposition [ beardonprop ] also balls in the triangular ratio metric are euclidean balls .[ halfplanelemma ] let @xmath28 and @xmath29.then @xmath206 by symmetry of the domain it suffices to consider only the case @xmath140 and we may assume that @xmath207 , @xmath208 .first we select points @xmath209 and @xmath210 such that @xmath211 where @xmath212 , @xmath213 and @xmath29 . by the definition of @xmath60 we get for @xmath214 @xmath215 which is equivalent for @xmath216 . in a similar way we obtain @xmath217 . with a simple computation we get @xmath218 and the assertion follows from proposition [ beardonprop1 ] .[ halfplanecor1 ] let @xmath28 and @xmath219 .then @xmath220 [ halfplanecor2 ] let @xmath28 and @xmath24 .then @xmath221 let @xmath222 and @xmath40. then @xmath223 by symmetry of the domain it sufficies to prove the result in the case @xmath140 for @xmath207 .by @xcite and theorem [ halfplanelemma ] @xmath224 which is equivalent to @xmath225 and the assertion follows .we consider the triangular ratio metric in the punctured half - space @xmath226 . by ,theorems [ punctspace ] and [ halfplanelemma ] it is clear that the following result holds . [ puncthalfplane ] let @xmath227 and @xmath38 $ ] .then @xmath31 is convex .however , the upper bound for the radius @xmath228 in lemma [ puncthalfplane ] is not sharp . to see this we can choose @xmath10 close to @xmath229 and far from @xmath230 .now @xmath31 is a euclidean ball even for @xmath231 $ ] .for example for @xmath232 it can be verified that @xmath31 is convex for @xmath233 $ ] , see figure [ phppict ] .[ kuva5 ] in @xmath234 where @xmath235 and @xmath236 , @xmath237 and @xmath238.[phppict],title="fig:",width=377 ] the disks @xmath31 for @xmath239 and large enough radius consists of two parts separated by curve @xmath240 .the lower part consists of @xmath31 with respect to @xmath241 and the upper part consists of @xmath31 with respect to @xmath242 .the following theorem uses this idea and gives upper bound for the radius of convexity , when the center point @xmath10 is close to @xmath243 .[ hupthm ] let @xmath32 with @xmath33 and @xmath34 $ ] , where @xmath35 then @xmath31 is convex . by simple computationwe obtain that @xmath244 and we show that @xmath245 is below the curve @xmath246 .let @xmath247 and @xmath248 be the euclidean radius and center of @xmath31 . by geometrywe obtain that @xmath249 and thus @xmath250 which implies the assertion .we consider next the triangular ratio metric in the angular domain @xmath251 the boundary @xmath252 consists of two half - lines , which we call sides of the domain .[ ellipse ] let @xmath253 , @xmath254 $ ] and @xmath255 be the line through the points 0 and @xmath256 .then for all @xmath257 the maximal ellipse in @xmath258 touches both sides of the angular domain . if @xmath259 , then @xmath260 and the assertion follows .let us denote the lines that contain @xmath252 by @xmath261 and @xmath262 .let @xmath263 and denote @xmath264 the reflection of @xmath11 with respect to line @xmath261 and similarily @xmath265 the reflection of @xmath11 with respect to line @xmath262 .we consider maximal ellipses with foci @xmath10 and @xmath11 in the half - planes defined by lines @xmath261 and @xmath262 .formula means geometrically that the maximal ellipse with foci @xmath10 and @xmath11 in half - plane defined by @xmath261 touches @xmath261 at the point @xmath266 $ ] .the same is true also for @xmath262 .note that the line containing @xmath267 and @xmath10 is the bisector of the lines through the origin and points @xmath264 and @xmath265 .now by geometry @xmath268 and thus the maximal ellipse in @xmath269 touches both sides @xmath261 and @xmath262 and the assertion follows .[ kuva6 ] .,title="fig:",width=264 ] [ alpha < pi ] let @xmath270 , @xmath254 $ ] , @xmath15 and @xmath40 .then @xmath271 where @xmath272 and @xmath273 are the half - planes with @xmath274 .moreover , @xmath275 is always convex and @xmath276 is smooth for @xmath277 , where @xmath278 . by we have @xmath279 and by theorem [ halfplanelemma ] , @xmath275 is convex as an intersection of two convex domains .let us denote the line through @xmath267 and @xmath256 by @xmath255 . by proposition [ ellipse ]it is clear that @xmath276 is a circle if it does not intersect with @xmath255 at two distinct points and if @xmath276 is not a circle then it is not smooth .if @xmath276 is a circle and only touches @xmath255 then @xmath280 . we obtain that @xmath281 and the assertion follows .[ alpha > pi ] let @xmath270 , @xmath282 and @xmath15 .then @xmath31 is convex for @xmath38 $ ] and the radius @xmath283 is sharp for @xmath284 .moreover , if @xmath285 then @xmath161 is smooth for @xmath286 the radius of convexity @xmath283 follows from and theorem [ punctspace ] .sharpness of the radius follows from the fact that if @xmath287 then @xmath288 . if @xmath285 then by the proof of lemma [ alpha < pi ] , @xmath161 is a circle if @xmath289 , where @xmath290 and @xmath291 . by combining the results in the angular domain we obtain the corresponding result in a polygon .[ polygonthm ] let @xmath36 be a polygon and @xmath37 .then @xmath31 is convex for all @xmath38 $ ] .moreover , if @xmath39 is convex then @xmath31 is convex for all @xmath40 .the assertion follows from and lemmas [ alpha < pi ] and [ alpha > pi ] .the assertion follows from theorems [ punctspace ] , [ halfplanelemma ] , [ hupthm ] and [ polygonthm ] .let @xmath3 be a convex domain and @xmath15 .is @xmath31 convex for all @xmath40 ?in this section and the following section we will consider the inclusion relations between metric balls in general domains and also some special domains . [ thm2 - 2]suppose that @xmath41 is a domain . for each @xmath42 and @xmath43 , we have @xmath292 we first show that @xmath293 for each @xmath294 we have @xmath295 when @xmath296 , and @xmath297 when @xmath298 because @xmath11 is arbitrary , we get @xmath299 for the second inclusion @xmath300 let @xmath301 and assume that @xmath302 .let @xmath303 be such that @xmath304 .then @xmath305 which implies that @xmath306 [ thm2 - 1 ] suppose that @xmath41 is a domain . for each @xmath42 and @xmath43, we have @xmath45 and the inclusion is sharp if there exists some points @xmath46 in @xmath47 such that @xmath48 and @xmath49 moreover , for each @xmath42 and @xmath50 , we have @xmath51 we first prove the first inclusion .for given @xmath42 and @xmath307 , we have @xmath308. then @xmath309 for the sharpness : if there exists some point @xmath46 in @xmath47 such that @xmath48 and @xmath310then let @xmath311 .hence @xmath312 which implies @xmath313 next , we consider the second inclusion . by* ( 3.9 ) ) , we have @xmath314 which together with the fact @xmath315 " show that @xmath316 hence , the second inclusion follows from theorem [ thm2 - 2 ] . by theorems [ thm2 - 2 ] and [ thm2 - 1 ] and ( * ? ? ? * ( 3.9 ) ) , we get the following corollary .suppose that @xmath41 is a domain .for each @xmath42 and @xmath43 , we have @xmath317 and for @xmath42 and @xmath50 , we have @xmath318 and the boundary of the @xmath60-metric disk in theorem [ thm3 - 1 ] visualized as the dashed curve . here @xmath319 and @xmath236 . [ fig1],scaledwidth=40.0%,scaledwidth=40.0% ]first we consider the punctured spaces and get what follows . andthe boundary of the @xmath60-metric disk in conjecture [ con3 - 1 ] visualized as the dashed curve . in the left picture@xmath142 , @xmath348 , in the right one , @xmath349 and @xmath348 .[ fig3],title="fig:",scaledwidth=50.0% ] and the boundary of the @xmath60-metric disk in conjecture [ con3 - 1 ] visualized as the dashed curve . in the left picture@xmath142 , @xmath348 , in the right one , @xmath349 and @xmath348 .[ fig3],title="fig:",scaledwidth=40.0% ] _ idea for the proof of conjecture [ con3 - 1 ] . _ the first part . for @xmath350 ,. then @xmath352 , and it suffices to prove @xmath353 that is @xmath354 and thus only need to prove the following : @xmath355@xmath356 where @xmath357 $ ] and @xmath358 $ ] .: _ metrics and quasiregular mappings .workshop on quasiconformal mappings and their applications , iit madras , dec 27 , 2005 - jan 1 , 2006 , ed . by s. ponnusamy , t. sugawa and m. vuorinen , _ quasiconformal mappings and their applications _ ,narosa publishing house , new delhi , india , 291325 , 2007 .

<S> we consider the triangular ratio metric and estimate the radius of convexity for balls in some special domains and prove the inclusion relations of metric balls defined by the triangular ratio metric , the quasihyperbolic metric and the @xmath0-metric . </S>

0911.1404

it is now a common practice for a democratic country to hold elections in order to choose an individual to hold formal office . in elections , there are indeed many factors that can influence the outcome of the result .it is noted that the outcome of the distribution of votes for individual candidates has taken interest from researchers in recent years[1 - 11 ] .however , hold regular elections by adopting different electoral systems . in fact , with different electoral systems , one would imagine that the vote distributions of candidates in elections would look very different . in the case of proportional elections held in brazil@xcite ,the distribution of votes among candidates for the whole country was shown to follow a power law @xmath0 with @xmath1 .on the other hand , in many european countries , the vote distributions turn out to follow a lognormal distribution@xcite .since different countries adopt different electoral systems , one would probably argue that these different distributions are indeed results of different electoral systems .figure 1 is an illustration of the vote distributions of elections from different countries@xcite. figures 1(a)-1(f ) are vote distributions from brazil , finland , usa , australia , canada and taiwan . the bracket next to the name of the country is the electoral system that the country is adopting . here , @xmath2 is the number of votes obtained by the candidate divided by the total number of votes in the election and @xmath3 is the probability density .it is interesting to note that vote distributions in fig .1(c)-(f ) display two peak behavior .this is significantly different from either power law or lognormal distributions as in fig .1(a ) and ( b ) .the apparent differences could probably be thought of related to the different electoral systems used in different countries .however , in the case of taiwan , the country has changed its electoral systems in 2008 but the essential features of the vote distributions in elections before and in that year is basically unchanged .therefore it is natural to ask if different vote distributions are in fact results coming from the same mechanism that governs elections instead of different electoral systems .vote distributions from ( a)brazil(proportional representation system , large voting district ) , ( b)finland(proportional representation system , large voting district ) , ( c)usa(majoritarian electoral system , small voting district ) , ( d)australia(majoritarian electoral system , small voting district ) , ( e)canada(majoritarian electoral system , small voting district ) and , ( f)taiwan(proportional representation system , single non - transferable vote and medium voting district ) . ] in this paper , we will show that although different countries adopt different electoral systems , the results of their vote distributions can indeed be explained by two competing effects , viz ., the majority effect and the media effect . we will demonstrate this by mimicking these two effects in a simple growth model and show that one can reproduce the distribution of votes in different countries which adopt different electoral systems .using models from physics to study topics in political science is common nowadays .for example , one can study political districting problem by using the well - known potts model in physics@xcite . to facilitate our discussion ,let us first define what we mean by majority and media effects . by majority effect ,we here mean that the decision of a person to vote for a particular candidate is mainly influenced by the portion of his neighbors and friends who want to vote for that particular candidate .the larger the portion is , the more likely that the person will vote for this particular candidate .hence we call it the majority effect .it is easy to imagine that most of the less known candidates in fact obtain their votes through this process . on the other hand ,a person can also obtain information of a particular candidate through other sources such as the mass media .for example , in the 2008 usa presidential election , most americans obtained their information about obama and mccain through mass media such as tv , internet , etc , much less from their friends and neighbors . in this way, we can say that people obtain their information about a certain candidate through some long range interactions .we categorize these long range interactions as the media effect .we believe that the two effects indeed compete against each other which results in different patterns of vote distributions as observed in different countries .we will investigate the two effects one by one and then combine the two effects in our growth model to understand how they compete against each other . let us first consider only the majority effect . to make things simple, we now put all voters on a regular lattice .let us first put them on a one - dimensional regular lattice ( ring if we impose periodic boundary condition ) .each of the voters will then take a site on this lattice . in this case , each voter will only be affected by his neighbors whom they will vote for . to begin with ,let us assume that there are @xmath4 voters ( lattice sites ) and @xmath5 candidates .the candidates will act as the seeds on the lattice and they are randomly distributed on the lattice . in the first time step , the candidates start to convince his neighbors ( neighboring sites ) to vote for him . in the next time step , his neighbors ( neighboring sites ) will convince their neighbors ( neighboring sites ) to vote for the same candidate .when an undecided voter has decided voters on both sides , his vote will be determined by randomly picking one of his neighbor s decision .the process will stop when all the voters have decided whom they will vote for .this process is indeed equivalent to a simple one - dimensional growth process with @xmath5 seeds to begin with .the distribution of @xmath5 random seeds ( candidates ) should follow an exponential distribution @xmath6 where @xmath2 is the distance between two neighboring seeds and @xmath7 , the average lattice distance between two neighboring seeds .we here normalize @xmath2 to be from 0 to 1 . at the end of the process , there will be a distribution of votes from all the seeds(candidates ) .the form of the distribution of votes in this simple case can be obtained analytically .the argument is as follows . in order for a candidate to get a portion of votes @xmath2, he has to have two nearest neighboring candidates to be of a distance @xmath8 apart .the probability to have two neighboring candidates of distance @xmath8 apart is @xmath9 .the seed can be anywhere between these two neighboring candidates .integrating out all possibilities will give a factor proportional to @xmath2 . the vote distribution in this simple caseis therefore equal to @xmath10 , where @xmath11 is the normalization factor .@xmath11 can be fixed by integrating over @xmath2 from 0 to 1/2 , since the largest separation between two sites with a seed in between should be less than 1 .notice that there are no free parameters in this simple one - dimensional case .one can easily extend this analysis to higher dimensional cases .for a certain number of seeds randomly distributed on a two - dimensional surface , their distribution takes the form @xmath12 , where @xmath13 is the distance measured from a specified seed and @xmath14 is the density of seeds in the two - dimensional plane . in a similar fashion ,the vote distribution on the two - dimensional square lattice would take a form @xmath15 , where @xmath16 are constants and @xmath2 is the portion of votes that a certain seed will get .figure 2 is an illustration of simulations of both the one- and two - dimensional lattices with only majority effect and with @xmath17 and @xmath18 .the curves from theoretical arguments are also included for comparison . in the two - dimensional case ,the theoretical fit gives @xmath19 and @xmath20 respectively .simulations of both the one- and two - dimensional lattices with only majority effect . here , @xmath17 and @xmath18 .] the distribution of votes for the majority effect changes significantly if one introduces an extra factor which takes the form @xmath21 where @xmath22 refers to the @xmath23 candidate , @xmath24 is the total number of voters who have decided to vote for the this candidate and @xmath25 is the total number of voters who have already decided to vote for one of the @xmath5 candidates .@xmath26 here is a free parameter .the introduction of the factor in eq.(1 ) has the following meaning .a decided voter would want to convince his neighbors and friends to vote for that particular candidate .however , the probability that he could succeed would also depend on how well the candidate has been doing so far .one can conceive that if the candidate had been doing very poorly , the probability that the decided voter could successfully convince his neighbors and friends to vote for the candidate would be low .on the other hand , if the candidate had been doing very well , it could have been much easier for the decided voter to convince his neighbors and friends to vote for the candidate . the case of @xmath27is the extreme case which corresponds to what we have discussed above .this is the case when a decided voter can always convince his neighbors to vote for the candidate that he decided to vote for , regardless of how the candidate has been doing so far .figure 3 is an illustration of how the distribution of votes will change as we change @xmath26 . as @xmath26 increases, the tails on both ends of the distribution will start to move up until it displays a power law like distribution when @xmath28 .vote distributions with only majority effect on a two - dimensional square lattice as @xmath26 in eq .( 1 ) changes . ] as mentioned above , a few of the candidates can obtain a large portion of votes largely by media effect in an election .we investigate this effect again on a lattice .we here again assume that there are @xmath5 candidates , each candidate can put @xmath29 random seeds on the lattice with a total number of @xmath4 sites .since the @xmath30 seeds are randomly put on the lattice , one would then expect the distribution to follow a gaussian distribution .figure 4 shows the result of the distribution of votes with @xmath31 candidates on a two dimensional square lattice with a total of @xmath32 sites .we have tried two different situations here . in figure 4(a ) , @xmath33 and the random seeds are all put on the lattice in the very beginning .we let the system evolve similar to the case of the majority effect as described above .one can see that the curve can be fitted well by a gaussian distribution with its peak around 0.25 as expected . in figure 4(b ) , we start with only one seed for each of the @xmath5 candidates . in each time step , we give a probability @xmath34 that one of the undecided sites will become the seed of one of the @xmath5 candidates .again , the curve can be described by a gaussian distribution with its peak around 0.25 .the results in both cases agree well with our expectation . from the above analysis, we therefore propose that the function that describes the various effects on vote distribution in elections would take the general form @xmath35 where @xmath36 and @xmath37 are all constants and with the vote distributions of the two effects studied in the above as particular cases .vote distributions with only media effect .the number of candidates @xmath38 and each has @xmath33 random seeds on a square lattice with 250,000 sites .( a ) all the random seeds are put on the lattice at the beginning and , ( b ) the number of seeds grow as time evolves . ]we now include both effects in our growth model and investigate how they compete against each other .we again use the two - dimensional square lattice as an example . to begin with, we have a number of candidates @xmath5 and a lattice with @xmath4 lattice sites . ofthe @xmath5 candidates , @xmath39 of them acquire media effect while the rest can only obtain their votes through majority effect .the rules that mimic the two effects have been introduced above so we will not repeat here .figure 5 are the distributions of votes with both effects included in the model .figures 5(a ) - ( f ) are the fits using both our simulations and theoretical fit of the vote distributions corresponding to figure 1(a)-(f ) . in each of the cases, we fit the real data using both simulations from our growth model and the theoretical function of eq .( 2 ) . for example , figure 5(a ) is a fit of the 2006 election in brazil .the solid triangles are from real data while the circles are the simulated result of our growth model with @xmath40 and @xmath41 .we use @xmath42 in the simulation .the solid line is a theoretical fit using our function above with @xmath43 and @xmath44 .table 1 lists the values of @xmath34 and @xmath26 that we use in order to fit the vote distributions in different countries .we should also mention here that in the simulation , we use the same number of voters ( sites ) corresponding to the number of voters in the election held by that country .therefore , even with the same set of @xmath34 and @xmath26 , the result of the simulation could be different .an example is the simulations of australia and taiwan .even with the same values of @xmath34 and @xmath26 , the simulated results look somewhat different .theoretical fits for other countries can be done in a similar way but a sum of two functions similar to eq .( 2 ) is needed for figure 5(c)-(f ) .the peaks on the right of the distributions in figure 5(c)-(f ) refer to the group of candidates who obtain their votes through media effect .one could therefore study the relative importance of the two effects in different electoral systems within our simple growth model which would have real applications in elections .@*9l & & brazil & finland & usa & australia & canada & taiwan + @xmath34 & & 0.0 & 0.0 & 1.0e-6 & 1.0e-5 & 1.0e-5 & 1.0e-5 + @xmath26 & & 0.65 & 0.6 & 0.35 & 0.6 & 0.55 & 0.6 +the fact that there are two different groups of candidates in an election might be understood in the language of networks .the candidates who only obtain votes through majority effect can be conceived as people who communicate with others through a regular network with nearest neighbor interactions and could only affect others through some kind of diffusion process which is rather slow . on the other hand , candidates who obtain votes through media effect are more likely connected to networks similar to small world or scale free networks in which case they can affect voters from distance away . in summary ,we have identified two dominant effects in elections and mimic them in a simple growth model .we put the growth model on a two - dimensional square lattice and test it against the available data from elections in various countries . by adjusting the two parameters @xmath34 and @xmath26 in the model , we are able to fit the vote distributions of elections in all the countries that we have studied .we also give a theoretical argument of the general form of the function it can take and fit it with data .other effects as well as lattices with higher dimensions and/or different structures can be used and their effects on vote distributions will be investigated in the future .10 r.n . costa filho ,almeida , j.s .andrade , jr ., and j.e .moreira 1999 phys . rev .. s. fortunato and c. castellano 2007 phys .bernardes , d. stauffer , and j. kertesz , 2002 , eur .j. b 25 , 123 .lyra , u.m.s .costa , r.n .costa filho and j.s .andrade jr .2003 europhys .costa filho , m.p .almeida , j.e .moreira and j.s .andrade jr .2003 physica a 322 698 .g. travieso and l. da fontoura costa 2006 phys .e 74 036112 .moreira , d.r .paula , r.n .costa filho , and j.s .andrade , jr .2006 phys .e 73 065101 .gonzalez , a.o .sousa and h.j .herrmann 2004 int .andresen , h.f .hansen , a. hansen , g.l .vasconcelos , and j.s .andrade , jr 2008 int .c 19 1647 .araripe and r.n .costa filho 2009 physica a 388 4167 . c. castellano , s. fortunato , and v. loreto 2009 rev .web sites : http://clerk.house.gov , + http://www.justicaeleitoral.gov.br , http://www.aec.gov.au , + http://www.elections.ca , http://www.cec.gov.tw , http://www.tilastokeskus.fi .chou and s.p .li 2006 physica a 369 799 .

<S> there are many factors that can influence the outcome of an election . </S><S> we here identify two dominant effects that can affect the votes obtained by a candidate , namely , the majority effect and the media effect . </S><S> we mimic these two effects in a simple growth model . </S><S> we put the model on a two - dimensional square lattice and test it against the available data from elections in various countries . by adjusting the two parameters in the model , we are able to fit the vote distributions in all the countries studied . </S>

1203.1114

in classical statistics , the so - called simpson s `` paradox '' , also referred to as the yule - simpson effect @xcite , occurs when the correlations observed in different groups are reversed when the groups are combined together .the typical examples come from social- or medical - science @xcite : suppose you are given two samples , @xmath0 and @xmath1 , on which the rates of success of two events , say , the success of two different therapies or of two ways of applying for a job , are given by @xmath2 and @xmath3 respectively , with @xmath4 .the rates of success express the degree of correlations between the events under consideration and the characteristic features of the samples @xmath0 and @xmath1 .then , it may occur that by splitting the initial samples in groups , say @xmath5 and @xmath6 , according to the the value of a certain _ lurking variable _( say , gender , age , geographical localization , .. ) the ordering of the rates of success , and thus of the correlations , is reversed , in formula @xmath7 and @xmath8 . as for example, a certain therapy may appear _ good for women _ and _ good for men _ , but _bad for people _ @xcite .actually , there is no mathematical paradox : the ys effect arises from a hidden correlation , and it may be explained in terms of the relative weights of the groups @xmath5 and @xmath6 , due to different sizes of the samples in the groups @xmath9 and @xmath10 @xcite .an example illustrating the point is reported in the appendix .on the other hand , the practical consequences of the effect are indeed counterintuitive for decision making , since the aggregated data and the partitioned ones are , in fact , suggesting opposite strategies @xcite .other examples arises in game theory , where it may be shown that two losing strategies may be randomly combined to form a winning one @xcite realizing , in fact , a variation of the ys effect . from the operational point of viewthere is an additional reason for the emergence of the ys effect , which may summarized by saying the two events are incompatible : once a therapy has been administered , or an application has been submitted , there is no way to determine the outcome of the other options for the same individual @xcite . in other words , there is no way to assign a definite meaning to the rate of success of the two events simultaneously . loosely speaking, this fact suggests that other incompatibilities @xcite , as those arising from the quantum mechanical description of measurements , may play a role for the occurrence of the ys effect . in this communication , we address this possibility in details and investigate the occurrence of the ys effect in systems subjected to the laws of quantum mechanics .in particular , we analyze the role of mixing and superposition of quantum states , and illustrate the occurrence of two kinds of ys effects in quantum measurements .the first , which we term _ quantum - classical _ ys effect ( qcys ) , may occur with quantum limited measurements and mixing of states , whereas the second , here referred to as _ quantum - quantum _ ys effect ( qqys ) , may take place when superpositions of quantum states are allowed .we describe both effects in some details , prove that they may occur independently , and present a class of states for which qqys effect occurs when , for the same values of the involved parameters , the qcys one does not .let us consider two quantum tests , i.e. two binary probability operator - valued measures ( povms ) @xmath11 and @xmath12 aimed at describing the occurrence of certain events @xmath0 and @xmath1 .given a quantum state @xmath13 , the expectation value @xmath14 returns the probability of the event @xmath15 on the state @xmath13 , and thus represents the correlations between the occurrence of the event and the preparation of the system .we assume that the system under investigation may be prepared in two possible states @xmath16 , @xmath17 and that the event @xmath0 is more likely to happen than the event @xmath1 for both preparations , i.e. @xmath18 we have the yule - simpson effect whenever it occurs that for some choice of the mixing parameters ( which play the role of relative weights due to different sample sizes ) the event @xmath1 become more probable for the system prepared in mixed state @xmath19 < q \equiv\hbox{tr}[\varrho_\beta\,\pi_{{\scriptscriptstyle b}}]\end{aligned}\ ] ] where @xmath20 = \cos^2\alpha\ , p_1 + \sin^2\alpha\ , p_2 $ ] and @xmath21 = \cos^2\beta\ , q_1 + \sin^2\beta\ , q_2 $ ] ( see table 1 for a summary ) . the condition @xmath22 may be written as @xmath23 and it is satisfied by @xmath24 if @xmath25 and @xmath26 @xcite , where @xmath27 , @xmath28 and @xmath29 .we refer to this form of the yule - simpson effect as to _ quantum - classical _simpson s paradox since it happens in quantum - limited measurements , however with the lurking variable coming from the classical mixing of two quantum states ..the quantum - classical yule - simpson effect [ cols="^,^,^,^,^,^,^",options="header " , ]in order to gain a quantitative insight into both effects let us start by focusing on bidimensional systems ( qubits ) . in this casewe may write the initial states as @xmath30 is a basis for the hilbert space , and the povms as @xmath31 , @xmath32 , where @xmath33 , and @xmath34 are pauli matrices . the constraint of positivity for the @xmath35 s is expressed by the conditions @xmath36 .projective measurements are those individuated by the relations @xmath37 . before going to the general case ,let us consider the specific example where @xmath0 and @xmath1 are mutually exclusive events .classically , there is no ys effect in this case , since no lurking variable may change the correlations .quantum mechanically , the two mutually exclusive events are described by projective measurements on two orthogonal states .for qubits they span the entire hilbert space , i.e. @xmath38 . as a consequence , we have @xmath39 and , in turn , @xmath40 if @xmath41 , i.e no qcys effect . on the other hand , by looking at eqs .( [ pq1 ] ) and ( [ pq2 ] ) , one sees that qqys effect takes place whenever we have @xmath42 for @xmath43 , e.g. upon choosing @xmath44 and @xmath45 , for @xmath46 , @xmath47 , @xmath48 , @xmath49 , @xmath50 , and @xmath51 .more generally , the occurrence of qcys and qqys may be investigated in terms of the four + eight + four parameters describing the initial states @xmath16 , @xmath17 , the povms @xmath35 , @xmath52 , and the lurking variables @xmath53 . in fig .[ f : d2 ] we report results for random states and measurements , generated to satisfy the constraints @xmath41 , @xmath17 , and for random lurking variables . in particular , we show the ratio @xmath54 as a function of the ratio @xmath55 .the square region @xmath56 corresponds to the occurrence of both the qc and qq ys effects , whereas the rectangular regions @xmath57 and @xmath58 correspond to cases when solely the qc effect , or the qq one , takes place , respectively .finally , the region @xmath59 corresponds to situations when we have no ys effects .on the left , we show the results for random states and measurements , and for random lurking variables @xmath60 $ ] , @xmath61 $ ] , whereas on the right we consider only situations for which @xmath62 . as a function of @xmath55 for @xmath63randomly generated sets of states , measurements , and lurking variables .right : the same plot for fixed values of the mixing parameters @xmath62.,title="fig : " ] as a function of @xmath55 for @xmath63 randomly generated sets of states , measurements , and lurking variables .right : the same plot for fixed values of the mixing parameters @xmath62.,title="fig : " ] as it is apparent from the left plot , the two effects may occur independently , i.e. we may have @xmath22 when @xmath64 , and viceversa . for random settings , i.e. random states , measurements and lurking variables ,the qqys effect is more likely to occur than the corresponding ccys effect . upon generating a sample of @xmath63 random settings, we found qcys effect without the qqys one in about @xmath65 of cases , just the reverse in about @xmath66 of cases , and both effects in about @xmath67 of cases .the right plot refers to random initial states and measurements , and to lurking variables with equal mixing parameters @xmath68 . in this case , as mentioned before , the qcys effect can not occur for any measurement . on the other hand , upon considering quantum superpositions, we have that the qqys effect occurs for a consistent fraction of the possible settings ( about @xmath69 ) .if we consider only projective measurements , the rates of occurrence of both the ys effects increases , but the overall picture is not qualitatively modified .similar results are also obtained by considering three - dimensional quantum systems ( qutrits ) .let us now assume that besides the complete mixture @xmath70 , and the coherent superposition @xmath71 , the system may be also prepared in partially coherent superpositions of the two initial states @xmath16 , i.e. we consider the general class of states @xmath72 , @xmath73 .the family @xmath74 continuously connects the coherent superposition to the complete mixture , and we now address the occurrence of the corresponding generalized qqys effect , which takes place , assuming again @xmath41 , @xmath17 , when we have @xmath75 , where @xmath76 = \lambda p + ( 1-\lambda ) p$ ] , @xmath77 = \lambda q + ( 1-\lambda ) q$ ] . using these expressions and introducing the threshold value @xmath78one easily see that the qqys effect persists in the range @xmath79 if @xmath80 , or @xmath81 if @xmath82 .is shown as a function of @xmath55 for @xmath63 randomly generated sets of states , measurements , and lurking variables . from left to right : @xmath83 respectively.,title="fig : " ] is shown as a function of @xmath55 for @xmath63 randomly generated sets of states , measurements , and lurking variables . from left to right : @xmath83 respectively.,title="fig : " ] is shown as a function of @xmath55 for @xmath63 randomly generated sets of states , measurements , and lurking variables . from left to right : @xmath83 respectively.,title="fig : " ] the transition from the qc to the qq effect is illustrated in fig . [f : lam ] , where we show the ratio @xmath84 as a function of the ratio @xmath55 for @xmath63 sets of randomly generated states , measurements and lurking variables ( as in fig .[ f : d2 ] ) , and for different values of @xmath85 .the yule - simpson is a paradigmatic paradoxical effect occurring in data aggregation in complex systems , and found applications in the physical description of , e.g. social processes @xcite , complex dynamics @xcite , and game theory @xcite . in this communication , we have illustrated the occurrence of the ys effect in quantum measurements , where the lurking variables are coming either from the mixing or the superposition of quantum states . by analyzing low dimensional systems , we have found that the two effects may occur independently , and that the quantum - quantum ys effect is more likely to occur than the corresponding quantum - classical one for the same set of states and measurements .we have also discussed an example , illustrating the occurence of the quantum - classical ys effect in a decision problem involving quantum measurements .one of the best known real life examples of simpson s paradox occurred when the university of california , berkeley was sued for bias against women who had applied for admission to graduate schools there @xcite .the admission figures for the fall of 1973 showed that men applying were more likely than women to be admitted , and the difference was so large that it was unlikely to be due to statistical fluctuations . on the other hand, when examining the individual departments , it appeared that no department was significantly biased against women .in fact , most departments had a small but statistically significant bias in favor of women @xcite .the explanation of the `` paradox '' is relatively simple : the women tended to apply to competitive departments , with low rates of admission even among qualified applicants ( such as in the english department ) , whereas men tended to apply to less - competitive departments with high rates of admission among the qualified applicants ( such as in engineering and chemistry ) , and overall this results in a smaller fraction of women admitted in total .99 e. h. simpson , j. roy .stat . soc .b * 13 * , 238 ( 1951 ) . c. r. blyth j. am .stat . ass . *67 * , 364 ( 1972 ) .p. j. bickel , e. a. hammel , j. w. oconnell , science * 187 * , 398 ( 1975 ) .d. m. messick , j. p. van de geer , psychol .bull . * 90 * , 582 ( 1981 ) .s. ramanana - rahary , m. zitt , r. rousseau , scientometrics * 79 * , 311 ( 2009 ) . j. s. chuang , o. rivoire , s. leibler , science * 323 * , 272 ( 2009 ) .s. g. baker , b. s. kramer , j women health gen - b * 10 * , 867 ( 2001 ) .s. lipovetsky , w. m. conklin , eur .172 * , 334 ( 2006 ) .m. a. hernan , d. clayton , n. keiding , int .j. epidem .* 40 * , 780 ( 2011 ) .j. aldrich , stat .sci . * 10 * , 364 ( 1995 ) .d. r. cox , w. wermuth , j. roy .b * 65 * , 937 ( 2003 ). j. m. r. parrondo , g. p. harmer , d. abbott , phys .lett . * 85 * , 5226 ( 2000 ) .j. buceta , k. lindenberg , j. m. r. parrondo , fluctuation noise lett . * 2 * , l21 ( 2002 ) .r. j. kay , n. f. johnson , phys .e * 67 * , 056128 ( 2003 ) .r. d. astumian , m. bier , phys .lett . * 72 * , 1766 ( 1994 ) ; r. d. astumian , am .* 73 * , 178 ( 2005 ) .s. p. gudder , _ quantum probability _ , ( academic press , boston , 1988 ) .j. e. cohen , am . stat . * 40 * , 32 ( 1986 ) .p. hadjicostas , lin .. appl . * 264 * , 475 ( 1997 ) .s. galam , j. stat . phys . * 61 * , 943 ( 1990 ) . j. almeida , d. peralta - salas , m. romera , physica d * 200 * , 124 ( 2005 ) .m. d. mcdonnell , d. abbot , proc .a * 465 * , 3309 ( 2009 ) .see more details and real figures at http://en.wikipedia.org/wiki/simpsons_paradox

<S> the so - called simpson s `` paradox '' , or yule - simpson ( ys ) effect , occurs in classical statistics when the correlations that are present among different sets of samples are reversed if the sets are combined together , thus ignoring one or more _ lurking _ variables . here </S><S> we illustrate the occurrence of two analogue effects in quantum measurements . </S><S> the first , which we term _ quantum - classical _ ys effect , may occur with quantum limited measurements and with lurking variables coming from the mixing of states , whereas the second , here referred to as _ quantum - quantum _ ys effect , may take place when coherent superpositions of quantum states are allowed . by analyzing quantum measurements on low dimensional systems ( qubits and qutrits ) , </S><S> we show that the two effects may occur independently , and that the quantum - quantum ys effect is more likely to occur than the corresponding quantum - classical one . </S><S> we also found that there exist classes of superposition states for which the quantum - classical ys effect can not occur for any measurement and , at the same time , the quantum - quantum ys effect takes place in a consistent fraction of the possible measurement settings . </S><S> the occurrence of the effect in the presence of partial coherence is discussed as well as its possible implications for quantum hypothesis testing . </S>

1207.4282

in solid state materials , the energy - momentum dispersion relation @xmath0 is traditionally derived by associating between particles and wave packets and then solving the wave equation @xcite , @xcite . in this work ,we demonstrate an alternative procedure suitable for isotropic materials . we propose a theoretical definition of the particle effective mass based on @xcite and @xcite .then , we use a known dependence of mass on energy , which can be obtained for example from cyclotron resonance measurements in semiconductors @xcite . finally , we integrate the mass expression over the allowed energy range in order to obtain the dispersion relation . as an example , we assume a linear approximation of mass as a function of energy , @xmath1 , which is often observed experimentally for relativistic particles and in solid state materials @xcite , @xcite , @xcite .finally , we demonstrate that the definition of the effective mass and the resulting dispersion relation are suitable for description of free relativistic particles , carriers in solid state materials , and massless particles such as carriers in graphene and photons .we begin with the basic definition of the particle effective mass as presented in @xcite and @xcite where we identified particles with wave packets .this one - dimensional approach can be used for any isotropic energy - momentum relation @xmath0 @xcite .simple parabolic energy band model in semiconductors . ] in fig .[ fig : energy - parab ] , we show an example of a simple parabolic energy band model typically used in direct - gap semiconductors . this model is usually obtained from a solution of the wave equation and results in two energy bands : one band for particles with positive energy , similar to electrons in semiconductors , and another for anti - particles with negative energy , similar to light holes .assume that the dispersion relation @xmath0 achieves a minimum with the energy @xmath2 for the particle momentum @xmath3 , also assume that @xmath0 is symmetrical with respect to @xmath4 and @xmath3 .the kinetic energy for parabolic electrons and holes can be defined respectively as @xmath5 since the electron is at rest when @xmath3 , we can define @xmath6 as the electron rest energy and the electron mass at the energy minimum , @xmath7 , as the electron rest mass . clearly for electrons@xmath8 while for holes @xmath9 .usually @xcite , @xcite , the one - dimensional group velocity of the wave packet is defined as , @xmath10 relationship between the momentum and group velocity for parabolic electrons ( solid line ) and holes ( dashed line ) in semiconductors . ] in fig .[ fig : group - parab ] , we show a relationship between particle group velocity and momentum for parabolic electrons and holes obtained from ( [ eq : kinetic_energy ] ) and ( [ eq : group ] ) . based on the semi - classical definition of the particle momentum ,the effective mass appears as a proportionality factor between the particle momentum and the group velocity of the corresponding wave packet . however , there appears a complication with the sign of the effective mass for holes since the momentum is in the opposite direction from the group velocity as can be seen in fig .[ fig : group - parab ] . for electrons we obtain as expected , @xmath11 , while the hole momentum is a negative function of the hole group velocity , @xmath12 .this can be explained by the fact that the impact of moving holes is actually due to the electrons moving in the opposite direction @xcite .there is compelling experimental evidence that the effective mass should have a positive value for both electrons and holes .this can be deduced from the fact that electrons and holes rotate in opposite directions under the influence of the magnetic field .the implication is that the ratio of the electric charge to the effective mass has the opposite sign for electrons and holes .electrons have a negative charge and holes behave like electrically positive charges , as confirmed for example by their electrical attraction to electrons .consequently , both electrons and holes should have positive effective mass .therefore , we propose using the absolute values of the particle momentum and group velocity in order to define the effective mass for both particles and anti - particles with the help of ( [ eq : group ] ) , @xmath13 note that the effective mass defined by ( [ eq : mass ] ) is generally energy dependent and by definition always positive for both electrons and holes . in parabolic materials , the effective mass is a constant , @xmath14 , independent of energy as can be seen from ( [ eq : kinetic_energy ] ) and ( [ eq : mass ] ) and demonstrated in fig .[ fig : mass - parab ] .since there are no particles for @xmath15 , the mass is not defined for that energy range .effective mass as a function of energy for parabolic energy bands . ]therefore based on the semi - classical definition of the particle momentum , we can define the effective mass ( [ eq : mass ] ) using the absolute values of the particle momentum and group velocity which gives positive values of the effective mass for both particles and anti - particles .one of the main disadvantages of the parabolic energy model is the lack of velocity saturation at high particle energies , which is observed experimentally for both relativistic particles and semiconductors .assume that in the vicinity of @xmath6 , the effective mass can be approximated as a linear function of energy as observed from cyclotron resonance measurements in semiconductors @xcite , @xcite and from @xmath16 for relativistic particles .therefore , we assume the following linear approximation for the effective mass , @xmath17 where @xmath18 is a constant velocity that can be identified as the particle saturation velocity , as we show later . at the bottom of the electron energy band , @xmath6 and @xmath7 , and we obtain from ( [ eq : mass - linear - e0 ] ) ,@xmath19 a similar result is obtained for anti - particles from symmetry . assuming that there are no particles present for energies @xmath20 , we can define the forbidden energy gap as @xmath21 which demonstrates dependence between the particle bandgap , rest mass , and saturation velocity .effective mass as a function of energy for particles and anti - particles . ]effective mass as a function of energy for zero rest mass particles . ]then , we obtain for the particle effective mass from ( [ eq : mass - linear - e0 ] ) and ( [ eq : band - gap ] ) , @xmath22 in fig .[ fig : mass - energy ] , we show mass - energy dependence for particles and anti - particles based on ( [ eq : mass - linear ] ) . for particles with zero rest mass , we assume that @xmath23 and consequently from ( [ eq : band - gap ] ) @xmath24 .however , @xmath25 , and we still can define the effective mass from ( [ eq : mass - linear ] ) as , @xmath26 it appears from ( [ eq : mass - linear - zero ] ) that for zero rest mass particles , the dependence of mass on energy is similar to regular particles only with an assumption that @xmath14 and @xmath27 are much smaller than the particle energy , @xmath28 or equivalently @xmath29 .the resulting @xmath1 is presented in fig .[ fig : mass - energy - zero ] .therefore , we can model the effective mass of regular particles , anti - particles , and zero rest mass particles using the same linear effective mass approximation ( [ eq : mass - linear ] ) .assume that we can experimentally determine the dependence of the effective mass on energy in ( [ eq : mass ] ) , for example using cyclotron resonance in semiconductors @xcite .then , we can obtain the dispersion relation @xmath0 by integrating ( [ eq : mass ] ) over allowed energy values , which for one - dimensional isotropic case leads to @xmath30 substituting the linear mass ( [ eq : mass - linear ] ) into ( [ eq : momentum - energy ] ) results in @xmath31 integrating equation ( [ eq : momentum - integral ] ) we obtain for particles @xmath32 an identical expression results for anti - particles .close to the bottom of the band , where @xmath33 , we can approximate for particle energy from ( [ eq : momentum - general ] ) @xmath34 thus , a parabolic dispersion relation is preserved at the bottom of the energy band .two - band non - parabolic energy model resulting from the linear effective mass . ] assuming that for high energy particles , @xmath35 or equivalently @xmath29 , we can approximate from ( [ eq : momentum - general ] ) for high energies @xmath36 for the group velocity of high energy particles we obtain from ( [ eq : group ] ) and ( [ eq : momentum - highe ] ) @xmath37 therefore , the particle velocity approaches a maximum saturation velocity @xmath18 for high energies .the velocity saturation is a direct consequence of the linear mass - energy dependence in ( [ eq : mass - linear ] ) .the theoretical velocity saturation is important because it is observed experimentally for high energy particles .we can calculate @xmath18 from ( [ eq : band - gap ] ) @xmath38 the resulting two - band energy model obtained from ( [ eq : momentum - general ] ) and ( [ eq : sat - velocity ] ) is presented in fig .[ fig : energy - momentum ] .it appears that the main parameters of the model @xmath27 , @xmath14 , and @xmath18 can be easily measured experimentally .two - band energy model for particles with zero rest mass . ]the dispersion relation for zero rest mass particles results from ( [ eq : mass ] ) , ( [ eq : mass - linear - zero ] ) , and ( [ eq : momentum - general ] ) by assuming @xmath35 leading to , @xmath39 this in - turn leads to the group velocity of zero rest mass particles from ( [ eq : momentum - energy - zero ] ) @xmath40 as expected , zero rest mass particles always propagate with maximum saturation velocity and have linear dependence of the particle momentum on energy similar to high energy massive particles .it appears from fig .[ fig : energy - momentum - zero ] that the energy band model for zero rest mass particles ( [ eq : momentum - energy - zero ] ) can be considered an extension of the regular two - band model ( [ eq : momentum - general ] ) for high energy particles .thus , we were able to demonstrate a simple derivation of the particle dispersion relation based on the definition of the effective mass and linear dependence of mass on energy .we can apply the general dispersion relation ( [ eq : momentum - general ] ) to free relativistic particles and carriers in solid state materials . for relativistic particles ,the saturation velocity is equal to the speed of light @xmath41 and @xmath42. then , from ( [ eq : mass - linear ] ) we obtain the usual relativistic mass - energy dependence , @xmath43 applying the general dispersion relation ( [ eq : momentum - general ] ) to free relativistic particles ( [ eq : mass - einstein ] ) , we obtain as expected : @xmath44 note that ( [ eq : momentum - relativistic ] ) was derived from the linear mass assumption and not from the traditional relativistic concepts .next , we apply ( [ eq : momentum - general ] ) to electrons in solid state materials where the energy reference is usually chosen at the bottom of the conduction band .the energy @xmath45 is then identified as the electron kinetic energy . under these assumptions ,@xmath7 for @xmath4 , which leads from ( [ eq : mass - linear ] ) to @xmath46 using ( [ eq : momentum - general ] ) , we derive the energy band model for electrons in solid state materials , @xmath47 the above approximation ( [ eq : momentum - semi ] ) is similar to the simplified two - band kane model @xcite , @xcite .this model is usually derived by a complex solution of the wave equation .it is verified experimentally and commonly used in non - parabolic semiconductors such as hgcdte and insb @xcite , @xcite .therefore , the resulting dispersion relations for free relativistic particles ( [ eq : momentum - relativistic ] ) and carriers in solid state materials ( [ eq : momentum - semi ] ) are equivalent since both are obtained from the same general relationship ( [ eq : momentum - general ] ) by a suitable choice of the saturation velocity and energy reference .next , we apply the effective mass model to zero rest mass particles .carriers in graphene can be described as zero rest mass fermions with the saturation velocity equal to the fermi velocity @xmath48 @xcite .then , for carriers in graphene we obtain from ( [ eq : momentum - energy - zero ] ) @xmath49 which is confirmed experimentally from cyclotron resonance measurements @xcite . while the carrier effective mass in graphene is observed experimentally , the concept of the effective mass is not commonly applied to photons .nevertheless , by using @xmath50 and @xmath51 we can define the photon effective mass ( [ eq : mass ] ) , @xmath52 thus , we demonstrate that a simple energy band model derived using the effective mass definition and linear dependence of mass on energy is the same for relativistic particles , carriers in semiconductors , and zero rest mass particles .in this work , we extend the one - dimensional effective mass definition to particles , anti - particles , and zero rest mass particles by using the absolute values of the particle group velocity and momentum .we demonstrated that we can derive an energy band model based on the effective mass definition and using experimentally observed linear dependence of mass on energy .the resulting energy bands are non - parabolic and are similar to the electron and light hole bands in non - parabolic semiconductors .the energy non - parabolicity is a direct consequence of the linear dependence of the effective mass on energy and leads to carrier velocity saturation at high energies .the model for zero rest mass particles can be considered a high energy extension of the regular energy model .we demonstrated that the resulting expressions for free relativistic particles and carriers in semiconductors are equivalent and only depend on the choice of the reference energy .we applied the dispersion relation to zero rest mass particles such as carriers in graphene and photons . in conclusion, it seems that the proposed energy band model derived from the effective mass is compatible with the traditional dispersion relations and can be useful for description of relativistic particles and carriers in solid state materials .

<S> in this work , we demonstrate an alternative method of deriving an isotropic energy band model using a one - dimensional definition of the effective mass and experimentally observed dependence of mass on energy . we extend the effective mass definition to anti - particles and particles with zero rest mass . </S><S> we assume an often observed linear dependence of mass on energy and derive a generalized non - parabolic energy - momentum relation . </S><S> the resulting non - parabolicity leads to velocity saturation at high particle energies . </S><S> we apply the energy band model to free relativistic particles and carriers in solid state materials and obtain commonly used dispersion relations and experimentally confirmed effective masses . </S><S> we apply the model to zero rest mass particles in graphene and propose using the effective mass for photons . </S><S> therefore , it appears that the new energy band model based on the effective mass can be applied to relativistic particles and carriers in solid state materials . </S>

math0607619

in recent years many works on functional analysis have been obtained in order to extend the well - known results of the classical theory of normed linear spaces to the framework of asymmetric normed linear spaces and quasi - normed cones . in particular , the dual of an asymmetric normed linear space has been constructed and studied in @xcite . in the same reference an asymmetric version of the celebrated alouglu theorem has been proved ( see also @xcite ) .several appropiate generalizations of the structure of the dual of an asymmetric normed linear space can be found in @xcite and @xcite .banach type theorems in the frame of quasi - normed spaces have been given in @xcite and @xcite . in @xcite and @xcite ,the completion of asymmetric normed linear spaces and quasi - normed cones have been explored .an asymmetric version of the riesz theorem for finite dimension linear spaces can be found in @xcite .it seems interesting to point out that quasi - normed cones and other related ` nonsymmetric ' structures from topological algebra and functional analysis , have been successfully applied , in the last few years , to several problems in theoretical computer science , approximation theory and physics , respectively ( see 11 and 12 of @xcite , and also @xcite ) .the purpose of this paper is to show that it is possible to generate in a natural way a quotient quasi - normed cone from a subcone of a given quasi - normed cone .actually , we analyse when such quotient cones are bicomplete. we also construct and study the dual cone of a quasi - normed cone and we prove that it can be identified as the dual of a quotient cone .this is done with the help of an appropiate notion of a ` polar ' cone . 1 throughout this paper, @xmath9 @xmath10 and @xmath11 will denote the set of nonnegative real numbers , the set of natural numbers and the set of nonnegative integer numbers , respectively .recall that a _ monoid _ is a semigroup @xmath12 with neutral element @xmath13 .according to @xcite , a _ cone _ ( on @xmath14 is a triple @xmath15 such that @xmath12 is an abelian monoid , and @xmath16 is a mapping from @xmath17 to @xmath18 such that for all @xmath19 and @xmath20 : 1 .@xmath21 ; 2 .@xmath22 ; 3 .@xmath23 ; 4 .@xmath24 ; 5 .@xmath25 a cone @xmath15 is called _ cancellative _ if for all @xmath26 , @xmath27 implies @xmath28 .obviously , every linear space @xmath15 can be considered as a cancellative cone when we restrict the operation @xmath16 to @xmath29 let us recall that a _ linear mapping _ from a cone @xmath15 to a cone @xmath30 is a mapping @xmath31 such that @xmath32 .a _ subcone _ of a cone @xmath15 is a cone @xmath33 such that @xmath3 is a subset of @xmath18 and @xmath34 and @xmath35 are the restriction of @xmath36 and @xmath37 to @xmath1 respectively .a _ quasi - norm _ on a cone @xmath15 is a function @xmath38 such that for all @xmath19 and @xmath39 : 1 . @xmath40 if and only if there is @xmath41 and @xmath42 2 .@xmath43 3 . @xmath44 if the quasi - norm @xmath45 satisfies : ( i@xmath46 ) @xmath47 if and only if @xmath48 then @xmath45 is called a _ norm _ on the cone @xmath49 a quasi - norm defined on a linear space is called an _ asymmetric norm _ in @xcite , @xcite and @xcite .our main references for quasi - pseudo - metric spaces are @xcite and @xcite .let us recall that a _ quasi - pseudo - metric _ on a set @xmath18 is a nonnegative real - valued function @xmath50 on @xmath51 such that for all @xmath52 : ( i ) @xmath53 ( ii ) @xmath54 in our context , by a _ quasi - metric on _we mean a quasi - pseudo - metric @xmath50 on @xmath18 that satisfies the following condition : @xmath55 if and only if @xmath56 we will also consider _ extended quasi-(pseudo-)metrics_. they satisfy the above three axioms , except that we allow @xmath57 each extended quasi - pseudo - metric @xmath50 on a set @xmath18 induces a topology @xmath58 on@xmath18 which has as a base the family of open @xmath50-balls @xmath59 @xmath60 where @xmath61 for all @xmath62 and @xmath63 a ( @xmath64 _ extended ) quasi-(pseudo-)metric space _ is a pair @xmath65 such that @xmath18 is a set and @xmath50 is a ( @xmath64 extended ) quasi-(pseudo-)metric on @xmath66 similarly to @xcite , an extended quasi - metric @xmath50 on a cone @xmath15 is said to be _ invariant _ if for each @xmath67 and @xmath68 @xmath69 and @xmath70 .if @xmath50 is a ( @xmath64 extended ) quasi - metric on a set @xmath18 , then the function @xmath71 defined on @xmath51 by @xmath72 is a ( @xmath64 extended ) metric on @xmath66 an extended quasi - metric @xmath50 on a set @xmath18 is said to be _ bicomplete _ if @xmath71 is a complete extended metric on @xmath66 in @xcite it was shown that it is possible to generate in a natural way extended quasi - metrics from quasi - norms on cancellative cones , extending to the well - known result that establishes that each norm on a linear space @xmath18 induces a metric on @xmath66 in particular it was proved the following result : [ ep]@xmath73 let @xmath74 be a quasi - norm on a cancellative cone @xmath15 .then the function @xmath75 defined on @xmath51 by @xmath76 is an invariant extended quasi - metric on @xmath66 furthermore for each @xmath77 @xmath78 and each @xmath79 @xmath80 and the translations are @xmath81-open .the ( extended ) sorgenfrey topology is obtained as a particular case of the above construction .the following well - known example will be useful later on . for each @xmath82 , let @xmath83 then @xmath84 is clearly a quasi - norm on @xmath85 whose induced extended quasi - metric is the so - called ( extended ) upper quasi - metric @xmath86 on @xmath87 according to @xcite , a quasi - normed cone is a pair @xmath0 where @xmath18 is a cancellative cone and @xmath74 is a quasi - norm on @xmath66we can find several techniques to construct new normed linear spaces from a given one in the literature ( see , for example , @xcite and @xcite ) . in particular one of themconsists of deriving quotient linear spaces from a given linear space .let us briefly recall this construction .let @xmath18 be a linear space and let @xmath3 be a linear subspace of @xmath66 for each @xmath77 let @xmath88=x+y.$ ] the family of such subsets is called the quotient of @xmath18 by @xmath3 and it is denoted by @xmath89 the usual set - addition defines a sum on @xmath90 by @xmath91 moreover , the product by scalars is defined in @xmath90 by @xmath92 under these operations @xmath90 forms a linear space .furthemore , if @xmath93 is a norm defined on @xmath18 and @xmath3 is closed then @xmath94 admits a normed linear structure , with @xmath95\|_{e / y}=\inf\{\|x+y\|\hbox{\rm : } \break y\in y\}$ ] .next we introduce a general method for generating quotient spaces from a quasi - normed cone @xmath4 which preserves the quasi - normed cone structure , obtaining as a particular case of our construction the above classical technique .let @xmath18 be a cancellative cone and let @xmath3 be a subcone of @xmath18 .denote by @xmath96 the set @xmath97 clearly @xmath96 is a nonempty set because of @xmath98 .it is easy to see that @xmath96 admits a subcone structure , in fact it is a linear space .the following relation will allow us to construct a quasi - normed cone with quotient space structure . a pair of elements @xmath99 are @xmath100-related if there exists @xmath101 such that @xmath102 in this case, we write @xmath103 this relation is obviously an equivalence relation , i.e. , reflexive , symmetric and transitive .we will denote by @xmath88 $ ] the equivalence class of @xmath104 with respect to @xmath100 .thus , the classes are given by @xmath88=\{x+y\hbox{\rm : } \y\in g_{y}\}$ ] and the family of them by @xmath105\hbox{\rm : } \x\in x\}.$ ] @xmath73 the set @xmath90 endowed with the usual sum and product of equivalence classes @xmath88+[z]:=[x+z]$ ] and @xmath106:=[\lambda \cdot x]$ ] where @xmath99 and @xmath107 is a cancellative cone .let us show that the sum of equivalence classes is well - defined .indeed , if @xmath108 $ ] and @xmath109 $ ] there are elements @xmath110 such that @xmath111 and @xmath112 then @xmath113 whence @xmath114=[x+z].$ ] the natural product is also well - defined because @xmath115 so that @xmath116=[\lambda \cdot v].$ ] direct calculations show that @xmath117 is a cone with neutral element @xmath118=g_{y}$ ] . finally , we show that @xmath90 has the cancellation law .let @xmath52 such that @xmath88+[y]=[x]+[z].$ ] then @xmath119=[x+z].$ ] it follows that there exists @xmath101 with @xmath120 since @xmath18 is cancellative , @xmath121 so that @xmath122.$ ] consequently @xmath123=[z].$]@xmath124 @xmath73 let @xmath18 be a cancellative cone and @xmath3 be a subcone . then @xmath90 is called _ the quotient cone of _ @xmath18 _ by _@xmath125 [ inverse ] note that if @xmath88+[y]=[0]$ ] for any @xmath126 then @xmath119=[0]$ ] and , as a consequence there exists @xmath101 such that @xmath127 hence @xmath128 .therefore @xmath129 and @xmath130.$ ] in case @xmath18 is a linear space and @xmath3 is a linear subspace of @xmath18 we obtain as a particular case of our construction the classical technique for generating quotient linear spaces from linear spaces mentioned above , since @xmath118=y$ ] and @xmath88=x+y.$ ] let @xmath131 be endowed with the usual operations .let @xmath3 be the subcone of @xmath132 given by @xmath133 .obviously the subcone @xmath134 satisfies that @xmath135 and @xmath136 thus the ( cancellative ) quotient cone @xmath137\hbox{\rm : } \ ( x , y)\in \bbb{r}^{2}\},$ ] where @xmath138=(x , y)+a=\{(z+x , y)\hbox{\rm : } \z\in \bbb{r}\}.$ ] in the sequel , if @xmath139 is a subset of a quasi - metric space @xmath140 we will denote by @xmath141 the closure of @xmath139 with respect to @xmath142 the following property , whose proof is well - known and we omit , will be useful later on .[ qmclausura]@xmath73 let @xmath65 be a quasi - metric space and let @xmath143 .then , @xmath144 if and only if @xmath145 where @xmath146 next we denote by @xmath147 the function defined as @xmath148 in proposition [ pinf ] , we construct a quasi - norm on @xmath90 from a quasi - normed cone @xmath0 . for this , it is necessary to assume that @xmath96 is a closed subcone of @xmath18 with respect to @xmath81 ( see example [ ejprenorma ] below ) .[ pinf]@xmath73 let @xmath0 be a quasi - normed cone and let @xmath3 be a subcone of @xmath18 such that @xmath96 is closed in @xmath149 then the pair @xmath150 is a quasi - normed cone , where the function @xmath151 is defined by @xmath152)=\inf \{p(x+y)\hbox{\rm : } \y\in g_{y}\}.\ ] ] it is clear that @xmath153)=\inf \{p(y)\hbox{\rm : } \y\in g_{y}\}=0.\ ] ] let @xmath62 such that @xmath41 and @xmath154)=\hat{p}([-x])=0.$ ] then , @xmath155 hence , given @xmath156 there exist @xmath157 and @xmath158 with 1 .@xmath159 and @xmath160 2 .@xmath161 3 .@xmath162)+\varepsilon.$ ] since @xmath163 we deduce that @xmath164)\leq p(-x+g_{1})=p(b)<d_{p}(x , g_{y})+\varepsilon\ ] ] and @xmath165)+\varepsilon.\ ] ] then @xmath166)\leq d_{p}(x , g_{y})\leq \hat{p}([-x]).$ ] therefore @xmath167 .similarly it was showed that @xmath154)\leq d_{p}(-x , g_{y})\leq \hat{p}([x]).$ ] since @xmath96 is closed in @xmath168 and by proposition [ qmclausura ] , @xmath169 .thus @xmath88=[0].$ ] next we prove that @xmath170 is homogeneous with respect to nonnegative real numbers .we distinguish two cases : @xmath171 then the homogeneousness is immediately obtained by definition of @xmath170 and the fact that @xmath90 is a cone .@xmath172 then @xmath173 ) & = \inf \{p(\lambda x+g)\hbox{\rm : } \ g\in g_{y}\}=\inf \left\{\lambda p \left(x+\frac{g}{\lambda } \right)\hbox{\rm : } \ g\in g_{y}\right\ } \\[.3pc ] & = \inf \{\lambda p(x+h)\hbox{\rm : } \g_{y}\}=\lambda \inf \{p(x+h)\hbox{\rm : } \ h\in g_{y}\}\\[.3pc ] & = \lambda \hat{p}([x]).\end{aligned}\ ] ] it remains to show the triangular inequality .let @xmath99 and @xmath174 then there exist @xmath175 such that @xmath176)+\varepsilon /2\quad \hbox{and}\quad p(z+h)<\hat{p}([z])+\varepsilon /2.\ ] ] it follows that @xmath177+[z ] ) & = \hat{p}([x+z])\leq p((x+z)+(g+h))\\[.3pc ] & \leq p(x+g)+p(z+h)<\hat{p}([x])+\hat{p}([z])+\varepsilon.\end{aligned}\ ] ] we conclude that @xmath154+[z])\leq \hat{p}([x])+\hat{p } ( [ z]).$ ] @xmath178 the following example shows that the condition that the subcone @xmath96 is closed in @xmath168 can not be omitted in the statement of the preceding theorem . [ ejprenorma]consider @xmath179 endowed with the usual operations @xmath36 and @xmath16 .let @xmath139 be the subcone of @xmath132 given by @xmath180 .thus , it is clear that @xmath181 on the other hand , the function @xmath182 is a quasi - norm on @xmath183 so that @xmath184 is a quasi - normed cone .furthermore , @xmath185 because of @xmath186 for all @xmath187 and , as a consequence , @xmath188 is not closed in @xmath189 finally , the function @xmath170 satisfies that @xmath190)=\hat{p}([(-2,3)])=0.$ ] whence @xmath170 is a prenorm on @xmath191 but is not a quasi - norm .let us recall that a ( @xmath64 extended ) quasi - metric space @xmath192 is said to be a bicompletion of the ( extended ) quasi - metric space @xmath65 if @xmath192 is a bicomplete ( extended ) quasi - metric space such that @xmath65 is isometric to a dense subspace of the ( extended ) metric space @xmath193 it is well - known that each ( extended ) quasi - metric space @xmath65 has an ( up to isometry ) unique bicompletion @xmath194 ( see @xcite ) . in @xciteit was introduced that the notion of bicomplete quasi - normed cone and the construction of the bicompleteness was given . following @xcite ,a quasi - normed cone @xmath0 is said to be _ bicomplete _ if the induced extended quasi - metric @xmath75 is bicomplete . in connection with bicompleteness of quasi - normed structures ,some results for quasi - normed monoids and asymmetric normed linear spaces may be found in @xcite and@xcite . in this sectionwe characterize the bicompleteness of the quasi - normed quotient cone in terms of the bicompleteness of the original quasi - normed cone .[ dom ] let @xmath0 be a quasi - normed cone .then @xmath195,[y])\leq d_{p}(x , y)$ ] for all @xmath196 we distinguish two cases : @xmath197 then @xmath195,[y])=+\infty,$ ] because otherwise we have that there exists @xmath198\in x / y$ ] such that @xmath123=[x]+[z].$ ] it follows that @xmath199 for any @xmath200 hence we obtain that @xmath201 a contradiction with the hypothesis .now we suppose that @xmath202 then , given @xmath156 there exists @xmath203 such that @xmath204 and @xmath205. therefore @xmath123=[x]+[c_{\varepsilon } ] $ ] with @xmath195,[y])\leq \hat{p } ( [ c_{\varepsilon } ] ) < p(c_{\varepsilon } ) < d_{p}(x , y)+\varepsilon.$]@xmath206 [ cancellative]note that if @xmath18 is a cancellative cone , then @xmath207 and @xmath208 imply @xmath209 for all @xmath210 the next theorem provides necessary and sufficient conditions for bicompleteness of the quasi - normed quotient cone .[ bicomplete ] let @xmath0 be a quasi - normed cone and let @xmath3 be a subcone of @xmath18 such that @xmath96 is closed in @xmath149 then @xmath0 is bicomplete if and only if @xmath211 and @xmath212 are bicomplete .first we assume that @xmath0 is bicomplete .let @xmath213)_{n\in \bbb{n}}$ ] be a cauchy sequence in the extended metric space @xmath214 then there exists a subsequence @xmath215)_{i\in \bbb{n}}$ ] such that , given @xmath156 there is @xmath216 with @xmath217,[x_{n_{i+1}}])<\varepsilon $ ] whenever @xmath218 it follows that there exist @xmath219,[b_{n_{i , i+1}}]\in x / y,$ ] such that @xmath220=[x_{n_{i+1}}]+[a_{n_{i , i+1 } } ] , \quad [ x_{n_{i+1}}]=[x_{n_{i}}]+[b_{n_{i+1,i}}]\ ] ] and @xmath221),\hat{p}([ba_{n_{i+1,i}}])\ } < \varepsilon $ ] for all @xmath218 by remark [ cancellative ] , @xmath219=-[b_{n_{i , i+1}}]$ ] and by remark [ inverse ] , @xmath222=[-a_{n_{i , i+1}}].$ ] consequently , there is @xmath223 which satisfies @xmath224 with @xmath225 whenever @xmath218 hence @xmath226 is a cauchy sequence in the extended metric space @xmath227 since @xmath0 is a bicomplete quasi - normed cone there exists @xmath62 such that @xmath228 thus , by lemma [ dom ] , @xmath229,[x_{n_{i}}])\leq \lim_{i\rightarrow + \infty } ( d_{p})^{s}(x , x_{n_{i}})=0.$ ] therefore @xmath213)_{n\in \bbb{n}}$ ] is a convergent sequence so that @xmath150 is a bicomplete quasi - normed cone .furthermore , if @xmath0 is bicomplete we deduce that all cauchy sequence in @xmath230 are convergent .since the convergence with respect to @xmath231 implies convergence with respect to @xmath232 and @xmath96 is closed in @xmath168 we obtain that @xmath211 isbicomplete . conversely , let @xmath233 be a cauchy sequence in @xmath227 given @xmath156 there exists @xmath234 such that @xmath235 whenever @xmath236 whence there exists @xmath237 such that @xmath238 with @xmath239 whenever @xmath240 by lemma [ dom ] we obtain that @xmath241,[x_{m}])\leq ( d_{p})^{s}(x_{n},x_{m})=\max\{p(a_{nm}),p(-a_{nm})\}<\varepsilon $ ] whenever @xmath236 it follows that @xmath213)_{n\in \bbb{n}}$ ] is a cauchy sequence in @xmath214 since @xmath150 is bicomplete , @xmath242,[x_{n}])=0 $ ] for any @xmath88\in x / y.$ ] hence there exist @xmath243 and @xmath244\in x / y$ ] such that @xmath88=[x_{n}]+[a_{n}],$ ] @xmath245=[x]+[-a_{n}]$ ] and @xmath246),\hat{p}([-a_{n}])\}<\varepsilon$ ] whenever @xmath247 put @xmath248 then there exists @xmath249 such that @xmath250 and @xmath251 whenever @xmath252 .thus we deduce that @xmath253 , @xmath254 and @xmath239 whenever @xmath255 whence we deduce that @xmath256=\hat{p}([a_{n}])<\varepsilon $ ] and @xmath257)+\varepsilon < 2\varepsilon $ ] for all @xmath258since @xmath259 & \leq \inf_{g\in g_{y}}d_{p}(w_{n}-a_{n},g_{y})+\inf_{g\in g_{y}}d_{p}(g_{y},w_{n}-a_{n})\\[.3pc ] & = d_{p}(w_{n}-a_{n},g_{y})+d_{p}(g_{y},w_{n}-a_{n})<3\varepsilon,\end{aligned}\ ] ] there exist @xmath260 and @xmath261 such that @xmath262 , @xmath263 and @xmath264 whenever @xmath258 thus @xmath265 and @xmath266 with @xmath267 for all @xmath268 so @xmath269 whenever @xmath270 it follows that @xmath271 is a cauchy sequence in the extended metric space @xmath272 hence @xmath273 for any @xmath101 because @xmath211 is bicomplete .therefore @xmath274 and @xmath275 this concludes the proof.@xmath124 as a consequence of theorem [ bicomplete ] we have the following well - known result for normed linear spaces ( see @xcite ) .@xmath73 let @xmath276 be a normed linear space and let @xmath3 be a closed subspace of @xmath66 then @xmath276 is a banach space if and only if @xmath277 and @xmath278 are banach spaces .throughout this section we give several results concerning continuous linear mappings , which will be useful in the next section . a mapping from a quasi - normed cone @xmath0 to a quasi - normed cone @xmath192 will be called _ continuous _ if it is continuous from @xmath279 to @xmath280 . in the sequel the set of all continuous linear mappings between quasi - normed cones @xmath0 and @xmath192 will be denoted by @xmath281 obviously @xmath7 is a cancellative cone for the usual point - wise operations . in particular , when the pair @xmath192 is the quasi - normed cone @xmath282 we will denote @xmath283 by @xmath284 and @xmath285 will be called the _ dual cone _ of @xmath66 note that @xmath286 if and only if it is a linear and upper semicontinuous real - valued function on @xmath287 some results concerning metrization and quasi - metrization for several weak @xmath288 topologies defined on the dual of a quasi - normed cone and a generalization of alouglu s theorem to quasi - normed cone context were obtained in @xcite . a version of theorem [ continuity ] and proposition [ qnormed ] for asymmetric normed linear spaces can be found in @xcite and @xcite .we omit the easy proof of the following lemma .[ homo ] let @xmath0 and @xmath192 be two quasi - normed cones and let @xmath289 be a linear mapping .then @xmath290 for all @xmath291 the next result characterizes the continuity of linear mappings defined between quasi - normed cones . in order to help the reader we include the complete proof of the below theoremalthough the equivalence ( 1)(3 ) has been proved in @xcite .[ continuity ] let @xmath0 and @xmath192 be two quasi - normed cones and let @xmath289 be a linear mapping .then the following statements are equivalent : 1 .@xmath292 is continuous .@xmath292 is bounded in @xmath293 for every @xmath63 3 .there exists @xmath294 such that @xmath295 for all @xmath296 first , we prove that @xmath297 implies @xmath298 given @xmath299 there exists @xmath300 such that @xmath301 whenever @xmath302 thus , @xmath303 for all @xmath304 .consider @xmath305 .next we show that @xmath292 is bounded in @xmath306 to this end we only consider that @xmath307 , otherwise there is nothing to prove . by lemma [ homo ] , @xmath308 it follows that @xmath309 for all @xmath310 and so @xmath292 is bounded .since @xmath292 is bounded in @xmath293 with @xmath300 and @xmath311 @xmath292 is bounded in @xmath312 next we show that @xmath313 implies @xmath314 .let @xmath315 satisfying @xmath316 for all @xmath62 such that @xmath317 let @xmath318 .now we show that @xmath319 we distinguish two cases .@xmath320 suppose that @xmath321 .then @xmath322 hence there exists @xmath323 such that @xmath324 it follows that @xmath325 with @xmath326 which contradicts @xmath313 .@xmath327 then @xmath328 furthermore , by @xmath329 @xmath330 immediatelly we deduce that @xmath331 therefore @xmath332 @xmath333 put @xmath334 then @xmath335 for all @xmath336 finally , we prove that @xmath314 implies @xmath337 indeed , let @xmath156 @xmath338 and @xmath19 such that @xmath339 then there exists @xmath340 with @xmath341 and @xmath342 thus @xmath343 this concludes the proof .@xmath344 [ product]let @xmath15 be a cancellative cone . then , @xmath40 if and only if @xmath345 for any @xmath172 suppose that @xmath345 for any @xmath172 since @xmath346 for all @xmath347 , @xmath348 whence @xmath349 for all @xmath350 and so @xmath351 conversely , consider that @xmath351 then , for every @xmath352 @xmath353 for all @xmath354 thus @xmath355 for all @xmath354 whence @xmath356 for all @xmath172@xmath124 observe that @xmath357 where @xmath358 and @xmath359 let @xmath360 be the real numbers endowed with the usual operations .it is clear that @xmath361 is a quasi - normed cone such that @xmath362 $ ] and @xmath363 so @xmath364 is not a closed set in @xmath365 [ null ] let @xmath0 and @xmath192 be two quasi - normed cones and let @xmath366 . if @xmath367 then @xmath368 on @xmath66 let @xmath62 .we distinguish two cases .@xmath369 then @xmath370 and @xmath371 so @xmath372 @xmath373 then @xmath374 and @xmath375 .consequently @xmath376 @xmath377 and by remark [ product ] , @xmath372@xmath344 [qnormed]@xmath73 let @xmath0 and @xmath192 be two quasi - normed cones . then @xmath378 is a quasi - normed cone , where @xmath379 obviously , the function @xmath380 is well - defined by theorem [ continuity ] .next , we only have to show that @xmath381 is a quasi - norm on @xmath66 indeed , @xmath382 let @xmath383 satisfy @xmath384 such that @xmath385 then @xmath386 hence , we deduce that @xmath387 for all @xmath388 and by lemma [ null ] , @xmath389 on the other hand , for each @xmath390 and @xmath366 we obtain @xmath391 & = \lambda \sup_{x\in \overline{b_{d_{p}}}(x,1)}q(f(x))=\lambda \|f\|_{p , q}.\end{aligned}\ ] ] finally , we show the triangle inequality .let @xmath392 then , given @xmath156 there exists @xmath393 such that @xmath394 and @xmath395 & \leq \|f\|_{p , q}+\|g\|_{p , q}+\varepsilon.\end{aligned}\ ] ] therefore @xmath396@xmath124 we omit the easy proof of the following useful results .[ normineq]@xmath73 let @xmath0 and @xmath192 be two quasi - normed cones and let @xmath383 .then @xmath397 @xmath398 for all @xmath399 [ three]@xmath73 let @xmath4 @xmath192 and @xmath400 be three quasi - normed cones . if @xmath383 and @xmath401 then @xmath402 and @xmath403 [ asynls][injective]@xmath73 let @xmath404 and @xmath405 be two asymmetric normed linear spaces and let @xmath406 be a linear mapping . then the following assertions are equivalent : 1 .@xmath407 is injective and @xmath408 is continuous .2 . there exists a positive constant @xmath409 such that @xmath410 the key to the proof of proposition [ injective ] is based on the fact that a linear mapping is injective if and only if @xmath411 . as a particular case of proposition [ injective ]we obtain the well - known result for the case of normed linear spaces ( see , for example @xcite ) .however , example [ ker ] shows that it is possible to construct a noninjective mapping on the quasi - normed cones which satisfies the condition ( 2 ) of the preceding proposition .this is due to the fact that there are functions whose @xmath412 is exactly the neutral element that are not injective , so that the above proposition does not hold for quasi - normed cones .[ isometry][ker ] motivated by the applications to the analysis of complexity of programs and algorithms given in @xcite , it is introduced and studied in @xcite that the so - called dual complexity space , which consists of the pair @xmath413 , where @xmath414 and @xmath415 is the quasi - metric on @xmath416 given by @xmath417 $ ] .several properties of @xmath415 are discussed in @xcite . in particular , observe that the topology induced by @xmath415 is not @xmath418 . on the other hand, @xmath419 is clearly a cancellative cone , with neutral element @xmath420 given by @xmath421 for all @xmath422 , where @xmath36 is the usual pointwise addition and @xmath16 is the operation defined by @xmath423 for all @xmath424 . let @xmath425 defined by @xmath426 .it is routine to see that @xmath74 is a quasi - norm on @xmath416 .then the induced extended quasi - metric @xmath427 on @xmath416 is given by @xmath428 several properties of the extended quasi - metric @xmath427 have been studied in @xcite and @xcite .let @xmath429 .it is routine to see that @xmath18 is a subcone of @xmath416 .define @xmath38 by @xmath430 .clearly @xmath45 is a quasi - norm on @xmath18 .let @xmath431 defined by @xmath432 and @xmath433 for all @xmath434 and @xmath435 .obviously @xmath436 is linear from @xmath437 to @xmath419 .moreover @xmath438 for all @xmath434 .so @xmath436 satisfies the condition @xmath313 of proposition [ injective ] and @xmath439 however , if @xmath440 satisfy @xmath441 and @xmath442 , we obtain @xmath443 , and thus @xmath436 is not injective . in order to obtain a quasi - normed cone version of proposition [ injective ] we introduce the next result .[ finjective]@xmath73 let @xmath0 and @xmath192 be two quasi - normed cones and let @xmath289 be a linear mapping .if @xmath292is injective then the following statements are equivalent : 1 .there exists @xmath294 such that @xmath444 for all @xmath445 .there exists @xmath446 such that @xmath447 for all @xmath399 first , we assume that the statement @xmath297 holds .then for every @xmath62 we have @xmath448 and @xmath449 where @xmath450 .now , assume that the statement @xmath313 holds and let @xmath451 then there exists @xmath62 such that @xmath452 and @xmath453 thus @xmath454 so that @xmath455 with @xmath456@xmath124 we finish the section extending some results related to open continuous linear mappings between normed linear spaces to our context .to this end , we introduce the @xmath74-_injectivity_. @xmath73 let @xmath0 be a quasi - normed cone and let @xmath3 be a nonempty subset of @xmath18 .we will say that @xmath31 is @xmath74-injective if @xmath457 implies @xmath458 whenever @xmath196 note that from the above definition one gathers that @xmath459 and that all injective mappings are @xmath74-injective@xmath460 furthermore , if @xmath0 is an asymmetric normed linear space the linearity joint with the @xmath74-injectivity implies the injectivity .however , in the next example it is shown that the equivalence is not true for quasi - normed cones .consider again , the quasi - normed cone @xmath461 of example [ isometry ] .it is easy to see that the mapping @xmath431 defined as in the mentioned example is linear and @xmath45-injective , but it is not injective .[ open ] let @xmath0 and @xmath192 be two quasi - normed cones and let @xmath289 be a @xmath74-injective linear mapping .then the following assertions are equivalent : 1 .@xmath292 is open and onto . 2 .there exists @xmath323 such that @xmath462 3 .there exists @xmath463 such that , given @xmath464 there is @xmath62 with @xmath452 and @xmath465 \(i ) clearly implies ( ii ) .next , suppose that ( ii ) holds and let @xmath466 .we distinguish two cases : @xmath467 by hypothesis there exists @xmath388 such that @xmath468 .suppose that @xmath469 then there exists @xmath300 with @xmath470 and @xmath471 since @xmath472 @xmath473 . on the other hand , by ( ii ) we obtain that there exists @xmath474 such that @xmath475 with @xmath476 however @xmath477 and @xmath478 which contradicts the fact that @xmath292 is @xmath74-injective@xmath460 therefore @xmath369 @xmath479 then , let @xmath480 by hypothesis , there exists @xmath481 such that @xmath482 put @xmath483 thus @xmath484 next we prove that ( iii ) implies ( i ) .let @xmath62 and let @xmath485 be a neighborhood of @xmath486 then , by proposition [ ep ] , there exists @xmath487 such that @xmath488 for all @xmath489 with @xmath490 .moreover , if @xmath466 satisfies @xmath491 we have that there exists @xmath492 such that @xmath493 and @xmath494 hence @xmath495 and @xmath496 is a neighborhood of @xmath497 this concludes the proof .@xmath344 @xmath73let @xmath0 and @xmath461 be two quasi - normed cones . if @xmath289 is an open , bijective , continuous linear mapping then it is an isomorphism .the condition of @xmath74-injectivity of theorem [ open ] can not be omitted as seen in the following example .consider the cancellative cones @xmath131 and @xmath498 endowed with the usual operations . define @xmath499 by @xmath500 it is clear that @xmath292 is linear and onto .consider the quasi - norms @xmath501 and @xmath502 defined by @xmath503 and @xmath83 obviously @xmath504 and @xmath505 and as a consequence @xmath292 is not @xmath74-injective .next we show that @xmath292 holds in ( ii ) in theorem [ open ] , but not in ( iii ) .indeed , we have that @xmath506=\overline{b_{d_{u } } } ( 0,1)\bbb{\subset } f(\overline{b_{d_{p}}}(0,1))=\bbb{r}$ ] and @xmath507 for all @xmath463 and @xmath508given a normed linear space @xmath276 and a linear subspace @xmath509 it is defined the _ polar of _ @xmath3 by @xmath510 where @xmath285 is the dual of @xmath66 it is well - known that the polar inherits the linear structure of @xmath285 and it is isometrically isomorphic to the dual of the quotient linear space @xmath90 whenever @xmath3 is closed ( see @xcite)@xmath460 next , we introduce a suitable notion of polar which allows us to extend the mentioned property to our context .define the mapping @xmath511 by @xmath512.$ ] obviously @xmath513 is linear and onto@xmath460 such a mapping will be called the _ quotient mapping _ _ of _ @xmath18 _ induced by _ @xmath125 next we show that the mapping @xmath514 is , in addition , continuous .@xmath73 let @xmath0 be a quasi - normed cone and let @xmath3 be a subcone of @xmath18 such that @xmath96 is closed in @xmath149 then the quotient mapping @xmath515 is continuous .let @xmath399 then @xmath516)= \inf_{y\in g_{y}}p(x+y)\leq \inf_{y\in g_{y}}p(x)+p(y)=p(x).$]@xmath124 [ phi]@xmath73 @xmath517 @xmath73 let @xmath0 be a quasi - normed cone .we will call _ polar _ of the subcone @xmath3 of @xmath18 to the set @xmath518 .note that as a particular case of our definition we have the polar of a normed linear space , because of @xmath519 whenever @xmath3 is a linear subspace of a normed linear space .let @xmath0 and @xmath192 be two quasi - normed cones .a linear mapping @xmath520 is an _ isometry _if @xmath521 for all @xmath399 if , in addition , @xmath407 is an isomorphism , i.e. , @xmath407 is a continuous linear bijective mapping with @xmath408 continuous , then @xmath407 is called an _isometric isomorphism_. observe that , contrary to the asymmetric linear case , there are isometries on normed cones which are not injective ( see example [ ker ] above ) .two quasi - normed cones are called _ isometrically isomorphic _ ( _ isomorphic _ ) if there exists an isometric isomorphism ( isomorphism ) between them . @xmath73let @xmath0 be a quasi - normed cone and let @xmath3 be a subcone of @xmath18 such that @xmath522 then @xmath523 is a subcone of @xmath285 isometrically isomorphic to @xmath524 we only show that @xmath523 is isometrically isomorphic to @xmath525 .define the mapping @xmath526 by @xmath527).$ ] it is clear that @xmath407 is a well - defined mapping because : 1 .@xmath528)=0 $ ] whenever @xmath529 since @xmath530 2 .@xmath531 is linear on @xmath3 for all @xmath532 and so is @xmath533 , 3 .@xmath531 is continuous for all @xmath534 , because applying proposition [ normineq ] for each @xmath62 we have @xmath535 from the definition of @xmath407 we immediately deduce that it is linear .on the other hand , we obtain from @xmath314 and corollary [ phi ] that @xmath536 so @xmath407 is continuous . since @xmath537)\leq p(x)$ ] for all @xmath77 @xmath538hence @xmath539 and @xmath407 is an isometry .next we show that @xmath407 is onto . for each @xmath540 define @xmath541 as @xmath542)=h(x).$ ]thus @xmath543 is a well - defined function . indeed , 1 .given @xmath544 $ ] such that @xmath545 for any @xmath101 we have @xmath546)=h(z)=h(x+g)=h(x)+h(g)=h(x)=f_{h}([x]).\ ] ] 2 .@xmath543 is linear , and so is @xmath547 .3 . since for each @xmath62 and @xmath101 , @xmath548))=u(h(x+g))\leq \|h\|_{p , u}p(x),\ ] ]we follow that @xmath548))\leq \|h\|_{p , u}\inf \{p(z)\hbox{\rm : } \ z\in [ x]\}=\|h\|_{p , u}\hat{p } ( [ x]),\ ] ] and @xmath549 it is easy to see that @xmath550 on the other hand , @xmath407 is injective .suppose that @xmath551 with @xmath552 then @xmath553 for all @xmath399 whence @xmath554)=f([x])$ ] for all @xmath77 and @xmath555 finally , we prove that @xmath556 is continuous .let @xmath540 then @xmath557 it follows that @xmath558))=u(f_{h}([x]))\leq \|h\|_{p , u}\hat{p}([x]).\ ] ] so @xmath559 and from theorem [ continuity ] , @xmath408 is continuous.@xmath124 in the classical context it is showed that , under open continuous linear mappings , every normed linear space is isomorphic to a distinguished quotient linear space ( see @xcite ) . to end this section we obtain an asymmetric version of such a result .let @xmath0 and @xmath192 be two quasi - normed cones and let @xmath520 be an onto linear mapping .define @xmath560 by @xmath561)=t(x).$ ] it is obvious that @xmath562 is well - defined , since if @xmath544 $ ] , then @xmath545 for any @xmath563 and @xmath564)=t(z)=t(x)+t(g)=t(x).$ ] since @xmath407 is linear and onto , so is @xmath562 .furthemore , if @xmath565 is the quotient mapping induced by @xmath566 it is a routine to check that @xmath567 [ ginject]@xmath73 let @xmath0 and @xmath192 be two quasi - normed cones. then , @xmath562 is injective if and only if @xmath568 implies @xmath569.$ ] first , suppose that @xmath562 is injective .then , if @xmath568 for any @xmath19 we follow that @xmath561)=\tilde{t}([y]).$ ] thus @xmath88=[y]$ ] and there exists @xmath570 such that @xmath571 so @xmath572 now assume that @xmath568 implies @xmath569.$ ] then , if @xmath561)= \tilde{t}([y])$ ] for any @xmath19 we have that @xmath569.$ ] consequently @xmath123=[x]$ ] and @xmath562 is injective.@xmath124 @xmath73 let @xmath0 and @xmath192 be two quasi - normed cones .a mapping @xmath520 is @xmath573-injective if @xmath568 implies @xmath569 $ ] .note that injectivity implies @xmath573-injectivity .the below examples show that contrary to the case of normed linear spaces @xcite , @xmath574 is not closed in general in @xmath575 consider @xmath131 with the usual operations @xmath36 and @xmath37 .let @xmath139 be the subcone of @xmath132 given by @xmath134 .thus , it is clear that @xmath181 on the other hand , the function @xmath182 is a quasi - norm on @xmath183 so that @xmath184 is a quasi - normed cone .furthermore , @xmath185 because of @xmath576 for all @xmath187 and , as a consequence , @xmath188 is not closed in @xmath577 define the linear mapping @xmath578 consequently @xmath579 @xmath580 and @xmath581 is not closed in @xmath168 .let @xmath582 be endowed with the usual operations of @xmath583 thus @xmath584 is a quasi - normed cone , where @xmath585 define the linear mapping @xmath586 by @xmath578 hence @xmath587 and @xmath588 suppose that the sequence @xmath589 is convergent to @xmath590 in @xmath591 then @xmath592 eventually .it follows that @xmath593 eventually .therefore @xmath594 and @xmath581 is closed in @xmath595 in the light of the preceding examples we assume in the following desired result that @xmath596 @xmath73 let @xmath0 and @xmath192 be two quasi - normed cones .let @xmath520 be a onto , continuous linear mapping such that @xmath574 is closed in @xmath149 then the following holds : 1 .@xmath562 is continuous .2 . @xmath597 3 .if in addition @xmath407 is @xmath74-injective , open and @xmath573-injective , then @xmath562 is an isomorphism .let @xmath88\in x/\ker f.$ ] by proposition [ normineq ] , @xmath598))=q(t(z))\leq \|t\|_{p , q}p(z)$ ] for all @xmath544.$ ] then @xmath599))\leq \|t\|_{p , q}\hat{p}([x]).$ ] thus @xmath562 is continuous and @xmath600 since @xmath601 we deduce that @xmath602 and so , by proposition [ three ] , @xmath603 by corollary [ phi ] , @xmath604 we conclude that @xmath605 thus we have proved ( 1 ) and ( 2 ) . item ( 3 ) remains .assume that @xmath407 is , in addition , open and @xmath74-injective .then , by theorem [ open ] , there exists @xmath463 such that every element @xmath466 is expressible as @xmath606 for any @xmath62 , holding @xmath465 since @xmath407 is @xmath573-injective , by proposition [ ginject ] , @xmath562 is injective and @xmath607.$ ] furthermore , @xmath154)\leq p(x)\leq mq(y)$ ] and , by proposition [ finjective ] , @xmath608 is continuous with @xmath609@xmath124the author acknowledges the support of the spanish ministry of science and technology , plan nacional i+d+i , and feder , grant bmf2003 - 02302 .knzi h p a , nonsymmetric distances and their associated topologies : about the origins of basic ideas in the area of asymmetric topology , in : handbook of the history of general topology ( eds ) c e aull and r lowen ( 2001 ) ( dordrecht : kluwer acad .vol . 3 , pp .

<S> given a normed cone @xmath0 and a subcone @xmath1 we construct and study the quotient normed cone @xmath2 generated by @xmath3 . </S><S> in particular we characterize the bicompleteness of @xmath2 in terms of the bicompleteness of @xmath4 and prove that the dual quotient cone @xmath5 can be identified as a distinguished subcone of the dual cone @xmath6 . </S><S> furthermore , some parts of the theory are presented in the general setting of the space @xmath7 of all continuous linear mappings from a normed cone @xmath0 to a normed cone @xmath8 extending several well - known results related to open continuous linear mappings between normed linear spaces . </S><S> [ theor]definition [ theor]lemma [ theor]remark [ theor]example [ theor]proposition [ theor]corollary = msam10 at 10pt </S>

1503.04823

neutrino astronomy in the mev range has been very successful .measurements of solar neutrinos confirmed many aspects of the nuclear fusion reactions that power the sun ; they also provided essential information about neutrino mass and mixing , especially the matter - induced effects .the detection of neutrinos from sn 1987a and the identification of its progenitor star together confirmed the prediction that type ii supernovae arise from the collapse of the core of a massive star into a proto - neutron star ; the extreme conditions allowed many novel tests of neutrino properties . however , there are unresolved questions about the sun and supernovae that can only be answered with improved sensitivity. a better measurement of @xmath2b neutrinos could improve knowledge of the solar core temperature , test the energy dependence of the electron - neutrino survival probability , and strengthen the signal of the day - night effect ( presently 3 @xmath3 @xcite ) @xcite .a first detection of the @xmath4 flux would provide new tests of the solar model and neutrino mixing .an eventual milky way supernova will allow high - statistics tests of the physical conditions attending neutron - star birth , flavor mixing in extreme conditions , and possibly black hole formation @xcite .an immediate goal is the first detection of the diffuse supernova neutrino background ( dsnb ) , which will provide new insights about supernova neutrino emission and the cosmic star formation history @xcite .discoveries could be made with existing experiments if detector backgrounds were reduced .we focus on super - kamiokande ( super - k ) , by far the largest low - energy neutrino detector , with 22.5 kton of pure water in its fiducial volume ( fv ) @xcite. great success in reducing backgrounds has already been achieved , but further gains have been stubborn . for the robustly detected solar - neutrino signal ,the signal / background ratio is only @xmath5 after standard cuts ; at forward angles relative to the sun , the ratio is @xmath6 @xcite . for the dsnb search ,the high background rate means that the analysis energy threshold is above the peak energy of the signal spectrum @xcite .decreasing the background rate by a factor @xmath7 would substantially advance solar neutrino studies and the dsnb search .is this possible without building a bigger , deeper detector ?yes . after standard cuts ,the dominant background in the super - k fv between 618 mev is beta decays of nuclear spallation products @xcite , which are short - lived isotopes produced from oxygen in association with cosmic - ray muons .( at lower energies , longer - lived isotopes produced through radon ingress and decay are dominant . )when a muon passes through super - k , a cut around the measured position of the muon track is made to reject the spallation decays that follow ; a difficulty is that some decay lifetimes are long ( up to 20 s ; see table i in ref .@xcite ) compared to the average time between muons ( @xmath8 s ) .more precisely , a likelihood method is used to test events based on time elapsed since the muon , distance from the track , and a variable related to muon energy loss .the empirical cut that super - k has developed for solar neutrino studies effectively removes @xmath0 of the backgrounds but introduces @xmath1 deadtime , making it hard to improve . in a previous paper @xcite, we performed the first theoretical calculation of the production and properties of the spallation decay backgrounds for water - based cherenkov detectors such as super - k .our predictions are in good agreement , within a factor of 2 , with super - k data on the energy spectrum and time profile for the sum of spallation decay isotopes , and could be improved by calibration and more careful comparison .( comparable accuracy is found in spallation studies for scintillator detectors @xcite . )we detailed the physical processes behind isotope production and ways to use this knowledge to improve cuts .an important point is that _nearly all isotopes are produced not by the muons themselves , but by the secondary particles associated with their energy - loss processes_. at the depth of super - k ( 2700 m water equivalent ) , where the average muon energy is 270 gev , the average energy loss for a vertical throughgoing muon is 11 gev , of which 7 gev is from continuous processes such as ionization and 4 gev from radiative processes such as delta - ray production and bremsstrahlung .fluctuations can make the radiative losses much larger .a recent super - k paper on the dsnb search @xcite showed that the cherenkov light yield associated with a muon varies along its track , exceeding that expected for a single muon and presenting a broad peak ( comparable in length to the height of the fv ) , and that subsequent spallation decays are correlated in position with this peak .the reasons for this variation , its properties , and its association with spallation decays went unexplained . however , it was found that these facts could be exploited to improve the rejection of spallation decays .using an effectively shorter section of the muon track , several times less than the height of the fv , a more aggressive cut was used while keeping the deadtime moderate .this allowed super - k to lower the analysis threshold for the dsnb search from 18 to 16 mev , with zero spallation events remaining . herewe provide the first explanation of the physics behind the super - k technique , as well as new insights to substantially improve its effectiveness . because the cherenkov intensity ( light emitted per unit length ) of a relativistic muon is constant, the extra light and its variation must be due to additional charged particles , and a natural explanation is that these are produced in showers .however , the variations shown by super - k ( fig . 2 in ref .@xcite ) and fig . 4.2 in ref. @xcite appear to be grossly inconsistent with this explanation , because the spatial extent is too large and the amplitude too small .nevertheless , we find that the excess light is indeed due to particles in showers ; that these showers are of short extent with high light intensity but appear long with low intensity due to cherenkov reconstruction issues ; that the correlation between the light profile peak and spallation production is because nearly all isotopes are made in showers ; and that these reconstructions can be improved .using our results , super - k could refine their new cut down to 6 mev to improve solar neutrino and dsnb studies .the framework for our calculations closely follows that of our previous paper , and details are given there @xcite .we use the particle transport code fluka @xcite ( version 2011.2b.6 ) for our calculations , which has been used extensively for simulating muon - induced backgrounds in underground detectors @xcite .we use the same physics choices for fluka , details of the super - k geometry setup , and the muon spectrum .our calculations are for single throughgoing muons traveling 32.2 m vertically down the center of the fv .we assume that the positions of muon tracks are always well determined by a combination of outer - detector and inner - detector information , aided by the long lever arm of the muon track . for nonvertical throughgoing muons with shorter path lengths or for muon bundles, our results could be adjusted appropriately .we discuss stopping muons separately . whereas our previous paper considered the average behavior of muons ( from one to the next , and along each track ) , we now follow the energy - loss variations of individual muons .we separately simulate how muons create daughter particles , how these daughters induce showers , and how these showers produce isotopes . with our new approach, we recover our previous results .all particles eventually produced following a muon are called secondaries ; those in the first generation are called daughters .the scope of this work is defined by a few choices .we focus on explaining and extending the results of ref .we explain just the main features of the super - k results ; improving the details would require further input from them .we do not yet attempt a full calculation of the reduction in backgrounds ; our estimates are enough to show the promise of new techniques . in our next paper, we will show why the super - k cherenkov reconstruction results appear to be inconsistent with showers and how they can be improved .the remainder of this paper is organized as follows . in sec .[ sec : showers ] , we focus on the physics of showers the energy spectra of their secondaries , their geometric properties , and the rates of showers as a function of their energy to highlight physics insights critical to understanding later results . in sec . [sec : correlation ] , we detail how isotopes are produced and how this explains the observed correlation between cherenkov light yield and spallation decays . finally , we conclude in sec .[ sec : conclusion ] .particle shower ( or cascade ) processes are central to this paper .showers can be produced by radiative energy losses of cosmic - ray muons in super - k , especially at high energies .the basic physics is that particles multiply in number through repeated interactions , with the particle energy decreasing in each generation .this continues until the average energy drops below a critical energy @xmath9 that depends on the type of shower and the medium . below this , charged particles mostly lose energy by ionization . for electromagnetic showers , the main secondary particles are electrons , positrons , and gamma rays .the dominant interactions in water are electrons and positrons producing gamma rays through bremsstrahlung with nuclei , and gamma rays pair - producing electrons and positrons , also with nuclei . for hadronic showers ,the main secondary particles are charged and neutral pions .hadron interactions with nucleons produce pions , and pion interactions with nucleons can change both of their charges .the basic processes in electromagnetic showers leave the target nuclei largely intact , but that is not true for hadronic showers , which brings additional complications .figure [ real_shower ] shows a typical shower .shower lengths are around a few meters ; shower widths are around tens of centimeters .most electrons and positrons in showers are forward , with @xmath10 0.9 .showers are defined most generally by the phase - space density of their secondary particles , i.e. , the joint number density in momentum and position , with time as a parameter . to express the cumulative effects of a shower , integrated over time , we use not the number density of secondary particles , which is only defined at a given instant , but rather some measure of their integrated effects . for a cherenkov detector ,it is useful to weight by path length ; for charged particles , this is proportional to the light produced ( and , especially for electrons , is nearly proportional to the energy deposited ) .different integrals of the phase - space density are convenient for different purposes .the path length profile in longitudinal position ( integrating over momenta and lateral positions ) is probably the most familiar , and it determines the observable muon light profile in super - k .the path length spectrum in energy ( integrating over positions and the momentum directions ) is not commonly shown , but it determines isotope production in super - k .we present these in the opposite order , covering path length spectra in sec .[ subsec : em ] and [ subsec : hadronic ] and longitudinal profiles in sec .[ subsec : geometry ] . to provide more detail on path length spectra , @xmath11 _ describes the sum of distances traveled by all particles of a given species at each energy_. this is obtained by integrating over the positions of the particles , and is called the volume - integrated fluence in fluka @xcite .this spectrum multiplied by the cross section as a function of energy is the integrand for calculating the interaction rate .the integrated path length above the cherenkov threshold determines the total cherenkov intensity and thus the number of photomultiplier tube hits .super - k , a water - based cherenkov detector , directly observes only relativistic charged particles .we focus on the light produced by showers induced by muons .the muons themselves produce cherenkov light at constant intensity along the muon track @xcite .super - k can not separate electrons from positrons or @xmath12 from @xmath13 , so , hereafter , electrons means the sum of electrons and positrons and pions means the sum of @xmath12 and @xmath13 , unless specified otherwise .charged particles below their cherenkov thresholds ( kinetic energy 0.257 mev for electrons and 70.1 mev for pions @xcite ) are not detectable .gamma rays and neutrons are not detectable directly , but only through their interactions .we do not discuss isotope production by showers in this section .however , it is helpful to keep in mind that the most important parent particles for background isotopes are neutrons and pions ; gamma rays make a small fraction of isotopes and electrons do not make isotopes @xcite .hence , even though neutrons , pions , and gamma rays contribute negligibly to cherenkov light production , we discuss their behavior in showers . in the remainder of this section ,we first study the physics of showers in water independent of primary muons .then we discuss how cosmic - ray muons make daughter particles and thus showers with a variety of energies in super - k . some important aspects of electromagnetic showers can be understood using simple principles . in a model proposed by heitler @xcite , it is assumed that bremsstrahlung and pair production have the same mean free path ( radiation length @xmath14 ) , that this is energy independent , and that all other interactions , including electron ionization , can be ignored .further , it is assumed that in each generation , particles travel the same fixed distance ( @xmath15 ) before they split into two particles , each with half the parent particle energy .figure [ em_shower ] illustrates this process .if the shower starts with one particle of energy @xmath16 , then after @xmath17 generations , there are @xmath18 secondary particles , each with energy @xmath19 the shower stops growing when the average particle energy is below the critical energy @xmath9 , which is set by the electron ionization energy loss in one radiation length @xcite .then , a shower reaches its maximum , where the number of particles is the greatest , after @xmath20 generations . because the particles are mostly forward due to being relativistic , the distance to the shower maximum is @xmath21 in water , @xmath22 cm and @xmath23 mev @xcite .electrons with energy @xmath9 lose all of their energy by ionization in one radiation length .after shower maximum , gamma rays and the electrons they scatter will travel somewhat further ( a few radiation lengths ) .for a 10 gev shower , the longitudinal extent of a shower would be @xmath24 2 m , far less than the height of super - k .the true shower extent is greater than this , but not much , and is discussed in sec .[ subsec : geometry ] . .] further properties of showers can be obtained analytically with more complex models @xcite .an example of the latter is the work by rossi and greisen @xcite , where they derived results by solving the boltzmann equations under certain assumptions . in their approximationa , which is only valid for high particle energies , asymptotic cross sections for bremsstrahlung and pair production are assumed and electron ionization energy loss is neglected . for the electron path length spectrum in an electron - initiated shower , they find @xmath25 for electron energies @xmath26 . for electrons with energy greater than @xmath27 ,the distance to their maximum is @xmath28 this is similar to the heitler result if eq .( [ showermax ] ) is ( inappropriately ) evaluated at @xmath9 . 0.1 , 1 , 10 , and 100 gev .the features seen at the injection energy arise because the showers have not yet reached an equilibrium mixture of @xmath29 , @xmath30 , and @xmath31 .the gamma - ray path length is shown only for @xmath32 gev ; the other cases are similar , except for having lower endpoints . the pion path length spectraare shown for @xmath33 1 , 10 , and 100 gev ( it is zero for 0.1 gev ) .all spectra are normalized by @xmath16 . ]contemporary work on showers is based on monte carlo simulation of all microscopic processes @xcite .the fluctuations ( distance , energy , etc . ) in every interaction are taken into account , instead of solving for the average behavior with the boltzmann equation .this enables the study of individual showers , as well as the variations among them .the simulation results are valid for the entire energy range , and the precision is excellent . in the following , we use theoretical insights to illustrate the physics behind our numerical results .figure [ shower_comparison ] shows particle path length spectra for electron - initiated showers .we inject electrons with fixed energies into the super - k fv , which is large enough to contain all secondary particles .we discuss fig .[ shower_comparison ] from high to low energy . asindividual showers develop , the average energy of the shower particles decreases . at the peak , which is somewhat below @xmath9 ,the particle number is at a maximum . at lower energies ,particles stop multiplying and the path length decreases due to particle ionization losses . _ the way these and other results are shownis designed to highlight key physics points_. as discussed , the numerator is the total path length traveled by a group of particles , and not just the number of particles .we divide by the injection energy @xmath16 to show when there is universality ( more energetic showers being just multiples of less energetic showers ) or deviations from that .because of the large range of energies , we use a @xmath34 scale on the @xmath35 axis ; also , this is especially appropriate for the showering phase , where particle energies change by factors , not shifts , between each generation . to calculate integrals of the curves , one should use @xmath36 as the integration variable . to match this choice of axis ,we take derivatives with respect to @xmath36 , which makes the height of the curve proportional to its importance in the integral ; note that @xmath37 ( see ref .@xcite for further discussion ) .all energies in logarithms are in gev units .a @xmath34 scale is often used on the @xmath38 axis .this is of no particular importance , except that one should judge the relative contributions to the integral by numerical , not visual , height .the spectra at high energies , during the shower phase , go as @xmath39 for both electrons and gamma rays .the differential cross sections for bremsstrahlung and pair production can be factorized to roughly depend only on the fractional energy of the outgoing particles @xcite .the path length spectra should be a function of @xmath40 , and a power law shows this scale invariance @xcite .the shower is extensive in ( proportional to ) @xmath16 , so the length must be proportional to @xmath16 .the result must also scale linearly with the radiation length @xmath14 .then , using simple dimensional analysis , we know the path length spectrum must scale as @xmath41 .this is consistent with the results of rossi and greisen @xcite .the slight difference between the gamma - ray and electron path lengths at high energies in fig .[ shower_comparison ] is due to electron ionization , which matters more as the energy decreases .the electron path length spectra at low energies , during the ionization phase , go as @xmath42 . to first order , ionization conserves particle number , but dissipates energy in the shower , so we might expect @xmath43 constant and @xmath44 .however , below the peak , there are many gamma rays from bremsstrahlung , as shown in fig .[ shower_comparison ] , and these inject energy to electrons from the medium through compton scattering . the competition between this and ionization produces the electron spectrum shown , including shifting the peak to an energy below @xmath9 . for an injection energy of 0.1 gev or lower, showers do not typically develop .electrons range out by ionization and do not produce or accelerate other particles .gamma rays undergo compton scattering and pair production , but they do not produce particles other than electrons .the hadronic particle content in electromagnetic showers is quite small on average , and the pion path lengths are a few orders of magnitude less than those for electrons .the shapes of the pion spectra reflect the large pion mass and the large energy required for pion production by photo - nuclear interactions .we discuss this in the next subsection .the electron path length spectra are nearly extensive in @xmath16 ( same for the gamma - ray path lengths ) .these lie on top of each other when we divide out this initial energy .in other words , particles in an electromagnetic showers quickly lose information about the initial energy , and such showers are self - similar except for total energy @xcite .( this is less true for the hadronic components of the showers . )consequently , the total path lengths are extensive in @xmath16 .because electron ionization is the dominant dissipative energy - loss process , the total path length of electrons in water is @xmath45 for electromagnetic showers of fixed energy , the total path length for electrons does not fluctuate much . for example , for a 10 gev electron initiated shower , the average total electron path is @xmath46 5500 cm , while the standard deviation is only @xmath46 200 cm .most of the fluctuations arise from the rare production of hadronic components , for which there is some energy loss without cherenkov light ( e.g. , neutrons , nonrelativistic protons ) .in addition , there is some contribution to the fluctuations because the electron ionization rate depends on energy .the cherenkov light intensity is proportional to the electron path length .figure [ shower_comparison ] shows that most of the cherenkov light comes from electrons near the critical energy @xcite .the electron path length differences near the endpoints for different injection energies contribute negligibly to the total path length . also , there is little electron path length accumulated below the cherenkov threshold .pion path lengths contribute negligibly because they are much shorter and pions have a higher cherenkov threshold . in sum ,_ the injection energy of an electromagnetic shower is accurately revealed by its total cherenkov light_. the visible energy of each shower is within a few percent of the true shower energy .for gamma - ray - initiated showers , the path length spectra of particles ( including the hadronic component ) are almost identical to those of electron - initiated showers , except near the endpoint , because showers quickly lose information about the initial particle @xcite .spallation isotopes are dominantly produced by particles that produce little ( pions ) or no ( neutrons , gamma rays ) cherenkov light themselves . however , these particles are accompanied by electrons through shower processes . in sec .[ sec : correlation ] , we detail how to exploit this connection and identify spallation products using cherenkov light .in hadronic interactions in the gev range and above , the dominant particles produced are pions , with roughly equal numbers of each charge .hadronic showers of @xmath12 , @xmath13 , and @xmath47 have much in common with electromagnetic showers of @xmath29 , @xmath30 , and @xmath31 , because both arise from particle multiplication processes and because the interaction lengths happen to be comparable .hadronic showers have a critical energy of about 1 gev , where the probabilities for pions to multiply or to lose energy by ionization are equal .the multiplicity of pions in hadronic showers increases with energy , being a few in the gev range and a few tens in the tev range @xcite . for further discussion ,see refs .@xcite , though note that their focus is on high energies and low densities . in the following ,we emphasize some differences between hadronic and electromagnetic showers . _although electromagnetic showers have , on average , only a small hadronic component , hadronic showers always have a dominant electromagnetic component_. charged pions interact , producing more pions and continuing the hadronic shower .however , neutral pions promptly decay to gamma rays , feeding an electromagnetic shower . with each new generation in the hadronic shower , roughly 1/3 of the remaining energy is transferred to the electromagnetic shower . in principle , a hadronic shower with enough interactions would transfer all of its energy to the electromagnetic shower ; in practice , the final hadronic fraction asymptotes at @xmath48 @xcite .the number of charged pions reaching low energies is larger than would be naively expected due to large pion multiplicities at high energy and fluctuations in the energy division in each interaction . 0.1 , 1 , 10 , and 100 gev .again , the features at the endpoints are injection effects .the small features in the electron line for @xmath33 0.1 gev arise due to gamma rays from @xmath47 and nuclear decays .all spectra are normalized by @xmath16 . ]figure [ shower_comparison_pi ] shows particle path length spectra for showers initiated by charged pions . for @xmath33 1 , 10 , and 100 gev ,the primary pions have enough energy to induce hadronic showers ; for the 0.1 gev case , there is no pion multiplication and we discuss it separately .the electron and pion spectra are not quite extensive in the injection energy .this can be seen from the fact that the curves shown in fig .[ shower_comparison_pi ] do not overlap . with increasing injection energy ,the fraction transferred to the electromagnetic shower increases . for @xmath33 1 , 10 , and 100gev pion - initiated showers , the fractional energy in electromagnetic showers is 31% , 49% , and 65% .the rest of the energy is dissipated by hadron and muon ionization energy loss , with a small fraction carried away by neutrinos . accordingly , as the injection energy increases , the pion curves fall and the electron curves rise .pion - initiated showers thus appear to be less energetic than electromagnetic showers with the same initial energy .the visible energy is proportional to the total particle path length above the cherenkov thresholds .the energy that goes into the electromagnetic component of the shower produces cherenkov light due to the @xmath46 500 cm / gev of relativistic electron path length .however , the energy that remains in the hadronic component of the shower is less efficient , with only @xmath46 100200 cm / gev of relativistic pion path length .the difference is because some energy is lost to neutral particles and because pions become nonrelativistic at a higher energy than electrons . in terms of light yield , pions are subdominant even in pion - initiated showers @xcite .the visible energies for @xmath33 1 , 10 and 100 gev pion showers are 0.57 , 6.3 , and 74 gev .the general features of the pion spectrum follow from the same principles that govern the electron spectrum : showering processes dominate at high energies , causing the increase in path length with decreasing energy , while ionization dominates at low energies , causing the decrease in path length with decreasing energy .the critical energy for hadronic showers is higher than that for electromagnetic showers , due to the large pion mass and other factors , and the behavior of the path length spectrum in the peak region is more complex .the peak near 0.4 gev corresponds the most probable pion production energy . at slightly lower energies , 0.10.3 gev ,some pions disappear through inelastic interactions of the form @xmath49 and @xmath50 with bound nucleons , with the residual energy and momentum absorbed by their nuclei .once charged pions become nonrelativistic , the ionization rate increases quickly and the path length accumulated is small and decreases more steeply than for electrons below the peak .when the pion injection energy is too low to create new pions , an electromagnetic shower can not typically develop .the pion path length spectrum is large , as all the energy remains with the pions , and this is the same for both @xmath13 and @xmath12 .for the @xmath33 0.1 gev case shown in fig .[ shower_comparison_pi ] , the total pion path length is 23 cm .although this curve is much higher than the others , its integral is only slightly larger , corresponding to 230 cm / gev , because nonrelativistic particles lose energy rapidly .rarely , a charged pion interacts with a nucleon and converts to a neutral pion , leading to some electromagnetic activity ( on average 11 cm of electron path length ) .low energy @xmath12 are especially efficient at making isotopes through atomic and then nuclear capture @xcite ; low energy @xmath13 do not efficiently make isotopes because they decay , not capture , once at rest . 1 , 10 and 100 gev . here d@xmath51 is the charged - particle path length in all directions accumulated in a step d@xmath52 10 cm along the initial direction .we separately shift the starting positions of the showers , each with one electron and height @xmath24 1/@xmath16 , so that the peaks line up at @xmath53 .all profiles are normalized by @xmath16 . ]the physical distributions of showers and how they compare to the size of the super - k detector are crucial for understanding why the new super - k cut technique @xcite gives such a big improvement .the longitudinal and lateral sizes of showers define the region around the muon track where isotopes are made .the exact profile and the deflection of shower particles determine the pattern of cherenkov light .we focus on electromagnetic showers in this section , because they are more common , because hadronic showers have a large electromagnetic shower component , and because hadronic showers are similar to electromagnetic showers in geometry ( slightly different , and discussed below ) .figure [ shower_size ] shows the average longitudinal shower profile for three different injection energies .we plot the electron path length per unit length along the initial direction , i.e. , the cherenkov intensity from the shower relative to that from a single particle .this is roughly the instantaneous number of charged particles in the shower times ( gev/@xmath16 ) .this is not exactly true due to nonforward motion and particles starting or stopping within bins ; in addition , these curves represent averages over many showers .the area under the curve is the total electron path length scaled by the injection energy , and is nearly the same for all energies .the showers extend 46 m for energies between 1100 gev .this length is much shorter than the height of the super - k fv , even for high - energy showers , which are rare .these average profiles show a rising phase , a peak , and a declining phase .the distance to the peak position of the shower is an important parameter .even though eq .( [ heitler ] ) and eq .( [ showermax ] ) were derived from simplified models , they are in good agreement with the full numerical results . in more detail ,the shape is consistent with standard formulas for the longitudinal profiles of showers , such as the greisen @xcite and gaisser - hillas profiles @xcite . the overall profile shape , especially the length asymmetry between the rising and falling parts of the shower , is important for our discussions of shower correlations with spallation backgrounds in super - k . compared to the naive heitler model , where all electrons stop in one radiation length after shower maximum , the tails of realistic showers are long .this arises from two types of fluctuations in showers : the distances particles travel before splitting obey an exponential distribution , and secondary particles do not always split the energy equally @xcite .these fluctuations give a distribution to the particle energies in the shower at a given depth , instead of all particles having the same energy at the same location .after the shower maximum , there are particles in the shower with energy higher than @xmath9 because they have interacted for fewer generations or because they have taken more energy from their parent particles .these higher - energy particles stay in the shower longer , creating the long tail .figure [ shower_example ] shows examples of longitudinal profiles of individual showers , as fluctuations will affect shower reconstruction .showers with primary energies of 1 gev look very different from one another and from the average profile . with increasing initial energy ,the relative fluctuations in shower profiles decrease .showers with 100 gev have little variation in widths , peak position , and shape .because the shower energy is proportional to the cherenkov light intensity , it is easy to measure the total energy in a shower ( up to the ambiguity of whether it is electromagnetic or hadronic ) . for high - energy showers , it might be possible to reconstruct them using the average profile as a template . for low - energy showers , which are the most common, it is not clear if template fits will be helpful , due to the large fluctuations .so far , we have simplified showers to be one dimensional and collinear .particles in showers do have lateral displacements .the most important reason is electron displacement due to multiple scattering during propagation @xcite .this is characterized by the molire radius , which is about 10 cm in water @xcite .this is very small compared to either the super - k muon track resolution or the distance between the spallation decay and the muon track .the effects of the lateral extent of showers are negligible , so we skip discussions of their average profile or fluctuations . however , though the lateral displacement of electrons is small on average , their angular deflections greatly affect how the shower appears in the detector .note from fig .[ real_shower ] that individual electron paths are short but that deviations away from the forward direction are common .we will discuss this in detail in our next paper .as noted , hadronic showers are similar to electromagnetic showers in geometry , but there are some differences . for 1 gev hadronic showers , the longitudinal extent is similar to that shown in fig .[ shower_size ] , but the shape is quite different . because this is so close to the hadronic critical energy , there are few generations , and we mostly see the average number of pions decrease according to an exponential set by the hadronic interaction length. this might provide a way to identify low - energy hadronic showers , which are especially important for isotope production .the longitudinal profiles for 10 and 100 gev hadronic showers are quite similar to those of electromagnetic showers . at all energies ,the fluctuations in the longitudinal profiles of individual hadronic showers around the average are greater than for electromagnetic shower of the same energy ; this might be used to distinguish hadronic showers on a statistical basis .a more promising means might be to use the fact that hadronic showers have larger lateral extent ( see fig . 4 of ref .@xcite ) .cosmic - ray muons abundantly produce daughter particles that initiate electromagnetic and hadronic showers .figure [ shower_spectrum ] shows the daughter particle production spectra obtained using the super - k muon spectrum .the frequencies are scaled by the muon rate in super - k , and are thus numbers per muon .the electron spectrum goes as @xmath54 .this comes mainly from delta - ray production collisions of muons with atomic electrons where the energy transfer is large .( far more frequently , these collisions transfer little energy , and are treated as continuous ionization . ) for a muon energy of 270 gev , the average at super - k , the maximum energy transfer to an electron is 260 gev @xcite .the differential cross section for delta - ray production scales as @xmath55 for electron energy transfers well below the maximum @xcite .this , plus the fact that we plot @xmath56 , largely explains the results shown .line is the sum of @xmath13 and @xmath12 ; the @xmath47 line is about half of the @xmath57 line and is not shown . ]the positron spectrum comes entirely from pair production , mostly through muon interactions with nuclei .the differential cross section does not have a simple power - law form .using an approximate formula @xcite , we find that the differential cross section can be approximated by a broken power law : @xmath58 at low energies and @xmath59 at high energies .the transition energy is around @xmath60 , which is about 2 gev for the muons in super - k .again , reasonable agreement is seen .electrons are also produced in pair production , and this component is the same as the positron spectrum .the gamma - ray spectrum is rather flat , which follows from the form of the bremsstrahlung differential cross section , which is @xmath61 @xcite . except at the highest energies , showers initiated by gamma rays are subdominant .the rate of hadronic showers is small because muons primarily lose energy by electromagnetic processes .the dominant hadrons made directly by muons are pions , with comparable numbers of each charge .relative to a mono - energetic muon spectrum , using the full super - k spectrum in fig .[ shower_spectrum ] ( as we do ) leads to only modest differences . at the highest energies , the differential cross sections fordelta - ray production and pair production quickly increase with muon energy @xcite .consequently , the electron and positron production are increased at high energies . for the other particles and energies ,the differences are less .the spectra of muon daughter particles , and hence the showers they induce , favor low energies .for electromagnetic showers , because of the dominant rate of delta - ray production , the total spectrum has a @xmath54 shape .the delta - ray spectrum does not stop at 0.1 gev but keeps rising at lower energies .these low - energy delta rays do not shower or make isotopes , but they do create an almost continuous light intensity on top of the flat light profile from the muon , with little variation between muons .the hadronic shower spectrum is relatively flat , with a wide peak near 0.4 gev .( the hadronic component in electromagnetic showers is of comparable , but smaller frequency . )though hadronic showers are rare , with rate below 1% of all showers , they are quite important for producing isotopes . to obtain the expected number of all showers per muon above a given energy ,we integrate the curves in fig .[ shower_spectrum ] ; above 0.1 , 1 , 10 , and 100 gev , we obtain 3.6 , 0.4 , 0.04 , and 0.003 . for each muon, there will be poisson fluctuations in the number of showers . in sec .[ subsec : individualshowers ] , we calculate the energy distributions of showers weighted by isotope and light production .in our previous paper @xcite , we showed that isotopes are typically not produced directly by muons , but rather by their low - energy secondaries .( an exception , discussed in sec .[ subsec : energyloss ] , is stopping @xmath62 . )the isotope yields follow from convolutions of secondary - particle path - length spectra with isotope - production cross sections .neutrons and pions are the most important secondaries for producing background isotopes those that decay with detectable signals in super - k .in contrast , gamma - ray secondaries primarily produce harmless isotopes those that are stable or decay invisibly .we focus on background isotopes . in this section , we show that most isotopes are produced in rare , individual showers . on one hand , this is not surprising , because isotope production increases with secondary particle path length , and showers produce many secondaries in a short distance . on the other hand, it has been assumed that isotopes are made continuously along the muon tracks .a consequence of our claim is that isotopes are produced at random but specific locations , coincident with showers , along muon tracks .this picture is different from one where we average over muons ( as in ref .@xcite ) , so that isotopes are produced nearly uniformly along the muon track . as we show , showers can identify and localize isotope production , because showers are detectable through their cherenkov signals .using position information for preceding showers , the cuts to reduce spallation decays need to be applied only to a short section of the muon track that effectively covers the shower .compared to most super - k analyses , where cuts are made along the whole muon track , this would allow decreased backgrounds without increased deadtime . with the same deadtime , cutting less volume allows a longer time cut , improving background rejection . a version of this technique was pioneered by super - k in a search for the diffuse supernova neutrino background @xcite , and it was shown to work to remove spallation backgrounds down to decay energies of 16 mev . our goal , besides giving the first explanation of why this technique works , is to show how to extend it down to 6 mev , where the spallation rate is much higher , and apply it to solar neutrino studies . in the remainder of this section ,we show how light and isotope production correlate with muon energy loss , how they causally depend on the initiating particle and energy of showers , and how well in principle these showers could be identified and localized .we calculate the distributions of products showers , light , and isotopes from individual muons .super - k could use these distributions , following their likelihood approach , to assess the probability that an observed signal is of a particular origin , e.g. , if a low - energy event is signal or background ( and , if so , which muon was likely the cause ) .there can be several independent showers along a muon track .when that is the case , detecting each shower and measuring its energy would require geometric reconstruction .it is easier to measure the total visible muon energy loss through the total cherenkov light intensity .the true muon energy loss is slightly larger than the apparent energy loss because of the reduced light yield of hadronic showers . increased muon energy loss results in greater path length in secondaries and , hence , more cherenkov light .most of the radiative energy loss goes into producing electromagnetic showers , and the subsequent electrons are contained in the detector .thus super - k can measure the energy loss ( but not the absolute energy ) of a throughgoing muon by the total light deposited .the radiative part can be obtained by subtracting the amount expected from a muon with the minimum energy loss ( greater than the minimum ionization rate because these muons are relativistic ) .even in the rare cases where there are hadronic energy losses , the total light is a reasonably faithful ( better than a factor of 2 ; see above ) measurement of the muon energy loss . increased muon energy loss results in more isotopes , also due to more secondaries .however , there is an important difference : _ while light production is common , isotope production is rare_. most background isotopes in water are produced by low- to medium - energy hadronic secondaries ( see fig . 7 of ref .@xcite ) , which are rarely produced and which are subdominant to electromagnetic secondaries . recall that hadronic showers always induce electromagnetic showers ( but not vice versa ) , and that the light from the latter is typically dominant .figure [ light_iso ] shows our calculation of how the production of background isotopes increases with total muon energy loss .we also show the super - k measurement , which is part of their likelihood function for spallation cuts , defined in terms of residual charge , @xmath63 , the number of detected photoelectrons in excess of that expected from a muon with the minimum energy loss .we made conversions between residual charge and energy loss for which we could find only an approximate factor ( 1000 photo - electrons @xmath46 130 mev @xcite ) .we assume that the super - k results are for the expected number of isotopes per muon and that they need to be corrected by a factor @xmath64 because only a fraction of isotopes are included by the cuts used to select spallation events ; refs .@xcite are not clear about either point .we obtain 0.1 by direct calculation , not _ad hoc _ adjustment ; this arises from two factors , each @xmath46 0.3 , for a time cut of @xmath65 0.1 s and an energy cut of @xmath66 7 mev .in addition , we assume that all muons are vertically throughgoing .nevertheless , our estimates should be reasonably accurate .the good agreement with the super - k measurement indicates that our simulation is correctly modeling muon energy loss and isotope production .this simple figure illustrates several important points that hint at the physics of isotope production in showers .first , the average production rate of background isotopes is small , even for large muon energy losses .( the yield of harmless isotopes is about ten times larger . )second , this function becomes nonzero only beyond about 7 gev , which is where muon radiative loss processes start @xcite .third , the curve rises faster than linearly for low values of muon energy loss .we separately checked individual isotopes , and found that they follow the same trend as the total shown in the figure .there are two possible shower frequency scenarios that could lead to fig .[ light_iso ] .a point common to both simply follows from poisson statistics , which we illustrate using an energy loss of 30 gev .because the number of background isotopes per muon is 0.1 on average , the number of isotopes produced is 1 for 1 muon and is 0 for 9 muons .however , fig .[ light_iso ] does not tell us the frequency of showers that make isotopes .small electromagnetic showers are more frequent and less efficient at making isotopes . if the isotopes were made by such showers , then the number of showers per muon would be @xmath24 1 , with a fraction @xmath24 0.1 of them making isotopes .hadronic showers or very energetic electromagnetic showers are less frequent and more efficient at making isotopes . if the isotopes were made in these showers , then the number of such showers per muon would be @xmath24 0.1 with a fraction @xmath24 1 of them making isotopes .distinguishing these scenarios is important .if isotope - producing showers were small in energy and common in position , then spallation cuts would have to be applied along the whole muon track ; in contrast , if these are big and rare , they could be localized to short regions along the muon track .the physics of isotope production by showers determines which shower energy range is most important .although isotope production rises with muon energy loss , the frequency of muon energy loss falls steeply ( see fig . 2 of ref .when the muon energy loss is large , strong cuts can be applied without increasing deadtime because the frequency of such events is low .for example , _ muon energy losses of 30 gev or more lead to @xmath46 60% of the isotopes in super - k , while being only 2% of all muons_. a simple cylinder cut along the muon track could thus eliminate a majority of isotopes with little deadtime . using a radius of 3 m and delay of 20 s for just the muons with large energy losses, super - k could cut @xmath46 58% of isotopes with only @xmath46 4% deadtime .( for comparison , a radius of 1 m and a delay of 20 s , applied to all muons , would cut @xmath46 80% of isotopes with @xmath46 20% deadtime , close to what super - k achieves with more sophisticated likelihood techniques . ) in ref .@xcite , super - k introduced a new cut on showering muons , " defined to be those with an energy loss @xmath67 gev ; for these , all data in the next 4 s from the whole detector are discarded .we estimate that this has substantially worse efficiency and deadtime than our proposed new cut .our investigations also demonstrate that no spallation cuts are necessary along the tracks of stopping muons .muons with low energy ( @xmath65 7 gev ) lose all their energy by ionization in the fv .because their energies are low , they do not typically lose energy by radiative processes .consequently , very few isotopes ( 0.4% of all isotopes ) are produced along their tracks . at the ends of their tracks ,however , negative muons can capture on oxygen , which can lead to nuclear breakup .thus , a separate cut for stopping muons where only events inside a sphere centered on the end of the muon track are rejected would be highly efficient with minimal deadtime ( super - k has such a cut for @xmath68n @xcite ) .figure [ isotope_muon_rate ] shows a histogram of muon energy loss for muons that make isotopes , weighted by the number of isotopes produced .the shape of the histogram is the frequency of muon energy loss in super - k ( fig . 2 of ref .@xcite ) multiplied with the yield of isotopes from muons in super - k ( fig . [ light_iso ] ) .we focus on a small energy range ( below 30 gev ) , assuming that high - energy - loss muons can be cut as suggested above .this figure shows that the most probable energy loss for isotope production is small . however , there is a long tail , extending to hundreds of gev .once the energy loss range is constrained to a reasonable range , the cut should be optimized for small energy losses .when we average over muons and along their tracks , as above , light and isotope production are correlated through the total muon energy loss . herewe break that energy loss into individual showers , and detail how light and isotopes are causally related to showers with different injection energies and initiating particles .these relationships determine the geometry of the spallation cuts .figure [ efficiency ] shows our results for the average yields of light and isotopes made by showers as a function of energy . to calculate how muon - induced showers produce light and isotopes , we obtain the number spectra of daughter particles produced directly by muons using fig .[ shower_spectrum ] , then multiply these number spectra with the yields of light and isotopes by showers with those energies .this approach accounts for nearly all the daughter particles from the radiative energy losses of muons ; we discuss the exceptions below .we define showers initiated by @xmath69 ( including a small contribution from kaons and other hadrons ) to be hadronic , and those initiated by @xmath70 , @xmath31 , or @xmath47 to be electromagnetic .to compare to experiment , we use visible energy , determined from the total cherenkov light ( proportional to the integrated path length above the cherenkov thresholds ) made by relativistic particles ( see sec . [ subsec : em ] , [ subsec : hadronic ] ) . at injection energies below 0.1 gev , the curves drop off because showers do not form ; at energies above @xmath71 gev , they drop off because such injection energies are rare .an immediate conclusion is that _ light production is strongly dominated by electromagnetic showers _ , which are by far the most common .another is that _ background isotope production is somewhat dominated by hadronic showers _, even though they are much more rare . , @xmath31 , and @xmath47 ; the `` hadr .'' curves include showers initiated by charged pions , kaons , and other hadrons .yields are per vertical throughgoing muon in super - k , taking into account the cosmic - ray muon spectrum . ]the light yield distributions depend on the physics of muon energy loss and of shower development . at lowest order ,the light yield @xmath72 follows @xmath73 , which can be obtained by multiplying fig .[ shower_spectrum ] by @xmath27 .electromagnetic showers in this energy range are primarily induced by delta rays from muons , and their frequency falls as @xmath74 shower per energy decade per muon traveling the length of the super - k fv ( 3220 cm ) .the light yield of an electromagnetic shower rises as @xmath75 . in combination ,the result is @xmath76 cm , almost independent of shower energy .( this continues to even lower energies , dropping slightly , due to low - energy delta rays . )that is , 5000 cm of light is equally likely to be from one 10 gev shower or ten 1 gev showers ; these cases can be distinguished by reconstruction of the light profile along the muon track .hadronic showers in this energy range are primarily induced by pions from muons ; the rate relative to delta - ray production is @xmath77 near 1 gev but increases steeply with injection energy .hadronic showers convert most of their energy to electromagnetic showers , which produce nearly all of the light , and this efficiency increases with injection energy .the light yield for hadronic showers as a class is therefore quite suppressed and is not as flat as for electromagnetic showers . at low energies, this variation is especially pronounced because of low pion production by muons .the total light yields ( integrated over energy ) provide an important check of our calculation .the average light yield per muon is @xmath78 cm , corresponding to a radiative energy loss of about 4 gev , or a total energy loss of about 11 gev , in good agreement with the average we found in ref .@xcite .the isotope yield distributions depend on similar physics , plus the interaction cross sections of secondaries with nuclei .although the frequency of hadronic showers is low , the neutrons and pions they produce are quite efficient at making background isotopes .( above a total muon energy loss of about 30 gev , this efficiency is so high that it becomes possible that 2 or more isotopes are produced in the same shower , which would allow their clear identification and localization as background events . ) em showers make isotopes mostly through the neutrons and pions they produce , but also directly through gamma rays .the shapes of the isotope distributions are similar to each other and to the light distributions , but there are some important differences .low - energy hadronic showers are especially efficient ( per injected energy ) at making isotopes , because they convert less of their energy to electromagnetic showers ; low - energy electromagnetic showers are especially inefficient because of the threshold energy needed to induce hadronic showers .the shape of the isotope production curve here is closely related to that in fig .[ isotope_muon_rate ] .here we consider the energy of individual showers , each of which contributes to the radiative energy loss ; the total energy loss in fig .[ isotope_muon_rate ] includes about 7 gev for ionization energy loss .also , here we use a @xmath34 axis , which stretches out small radiative losses , and a log derivative , which has the effect of multiplying the shape by a factor @xmath79 .these facts show why the total muon energy loss and isotope production are correlated but not causally connected .most of the detected muon energy loss comes from electromagnetic showers .in contrast , most isotopes are made by hadronic showers .both types of shower increase with muon energy loss .the correlation between energy loss and isotope production is not simply linear because of the steep rise of isotope production as a function of shower energy at low energies . even at the level of individual showers ,the production of light and isotope production are not completely causal .the isotope production per shower is typically low , which means the presence of a shower does not necessarily indicate the production of an isotope .however , when an isotope is produced , it is almost always preceded by light from a shower , and that is what makes the super - k background - reduction technique possible .how well the super - k technique works depends on the frequency of showers that make isotopes . _ the drop in isotope production in low energy showers shown in fig .[ efficiency ] is crucial ._ few isotopes are made by low - energy showers , which are common , or low - energy delta rays , which are near continuous . from fig .[ shower_spectrum ] , we calculate that the integrated rate of showers becomes @xmath46 1 per muon when the minimum daughter particle energy is @xmath46 0.4 gev .because almost all isotopes are made by higher - energy showers , this technique can work with minimal confusion about which shower to associate with an isotope .if low - energy showers had produced too large a fraction of isotopes , the associated showers would be too frequent along the muon track for this technique to be practical . , andthe neutron component is discussed in the text .`` total '' means the sum of all processes . ]how well the super - k technique works also depends on the fraction of background isotopes produced in showers .figure [ efficiency_cumulative ] shows the fraction of isotopes contained in showers above a given energy .the curves are integrations of the isotope yield curves in fig .[ efficiency ] , now using true shower energy .the hadronic and electromagnetic components shown in fig .[ efficiency_cumulative ] are the same as in fig .[ efficiency ] .the neutron component , not shown in fig . [ efficiency ] because it produces so little light , is special . above a few hundred mev ,neutron secondaries act as part of the hadronic component . at lower energies, they can induce `` neutronic '' showers , where neutrons collide with nuclei , ejecting neutrons ( and protons ) , continuing the process , producing isotopes but very little light ; this accounts for only a few percent of isotopes ._ figure [ efficiency_cumulative ] shows that nearly all isotopes are made in showers induced by muon daughter particles _ , which is also crucial for this technique .( we exclude isotopes made directly by primary muons , which make 3% of all isotopes , mostly through processes that then produce identifiable showers . ) above 0.01 gev , we recover 96% of the isotopes that are not directly produced by primary muons . within the precision of our calculations , this agrees well with the isotope yield in ref .@xcite , where we did not separate the processes leading to isotope production .this supports our claim that nearly all isotopes are made in showers .in future work , we will show that nearly all of the showers in fig .[ efficiency_cumulative ] are identifiable .the results above show that isotopes are almost always produced in showers , and that these showers are detectable by their light . the probability of isotope production increases with shower energy , though it is small at the most important energies .these facts agree with the usual super - k spallation likelihood function , for which isotope production increases with the total muon energy loss .if this energy loss can be localized to a shower , it will allow the cut to be applied to a shorter section of muon track .the success of the super - k cut technique depends on the fraction of isotopes produced in identifiable showers .figure [ seperation_distribution ] shows the distribution of separation distances between the peak of the muon light profile and isotope production point .we first describe our calculation in detail , and then compare to the super - k result . for each individual muon, the peak of the muon light profile is taken to be the point of the maximum charged particle path length along this muon track .we define the @xmath80 coordinate to increase along the muon track , beginning at the top of the detector , and the separation distance to be the @xmath80 position of the shower minus that of the isotope . for calculating the maximum light position, we use a binning of 50 cm , comparable to the position resolution in super - k at low energies ; other reasonable choices give similar results .when more than one isotope is produced , we compute the separation distances for each .because of how the distribution is defined and would be used in a likelihood approach , there is no conceptual problem with having more than one isotope produced by one muon . practically speaking, the most common such scenario should be two isotopes produced in a rare , high - energy shower .the separation distribution has a large peak and small tails ._ the peak comes from the case where the isotope is produced in the largest shower along the muon track ._ the isotope production profile generally follows the shower longitudinal profile .the peak in fig .[ seperation_distribution ] is thus centered at zero .the full width of the peak of @xmath81 m at half - maximum follows from that of the longitudinal shower profiles ; it extends further to the left because showers are longer after the peak than before . because the peak is quite sharp, it can define a new spallation likelihood function with stronger cuts over a shorter section of muon track , as empirically discovered in ref .the tails , which can be barely seen at separations of tens of meters , arise from cases where the isotopes and showers are uncorrelated .as the distributions of showers and isotopes are nearly flat along the muon track , the tails have a well - defined shape a symmetric triangle peaked at zero separation . for our calculation, we find that the area in this triangle is 13% of the total . to improve background rejection , it is important to understand the reasons for this uncorrelated component .we find that @xmath82 is due to isotope production accompanied by very little light ; the parent particles are high - energy muons or low - energy neutrons made by them , in a ratio of about 1 to 2 .the largest portion , @xmath48 , is due to cases where the isotope is produced in a visible shower , but where there is a larger shower elsewhere on the muon track ; we determine this by examining isotope - shower pairs with large separations . as a check, we find that this portion increases if we increase the height of the simulated detector .a key issue for reducing the uncorrelated component will thus be improving the identification of multiple independent showers along the muon track .the uncorrelated component is as small as it is , even with this simple approach , because the expected number of showers per muon is small .figure [ seperation_distribution ] also shows the super - k result from ref . @xcite .although it is similar , there are some important differences .the most important is that the area in the tails is @xmath83 instead of 13% .this excess is due to cases where the isotope is produced in a shower that would have been visible in our simulation but was not visible in the super - k analysis , at least after the smearing effects of imperfect cherenkov reconstruction .we can approximately recover the super - k fraction of @xmath83 if we assume that showers below @xmath84 gev ( muon energy losses below @xmath85 gev ) can not be reconstructed .in addition , the peak and tails are not symmetric , which we think is due to problems with shower reconstruction , as discussed in our next paper .our estimates about the super - k results are crude , as their analysis has low statistics and large bin widths ( this could be improved by their using spallation decay energies lower than 16 mev ) ; the functions used to fit their data seem nonideal ; the noted asymmetries cause uncertainties ; and there is the possibility of differences in the selection of single - throughgoing muons in the super - k analysis and in our simulations .there are two major steps super - k can take to strengthen the correlation between showers and isotopes .first , they could attempt to reconstruct showers of lower energy . a @xmath86 gev shower more than doubles the light from a muon track , and we expect that much smaller showers could be identified .if they can do this down to very low energies , their measured result should match what we obtained in our simulated data , and they could reduce @xmath83 to @xmath87 .second , they could attempt to recognize multiple showers per muon , defining cut regions around each . because showers are relatively rare , it would probably be enough to reconstruct up to two showers .if this were successful , they could reduce @xmath87 down to @xmath82 . in future work, we will show that it should be possible for super - k to improve their reconstruction technique well enough to match our results in fig .[ seperation_distribution ] , and then even further , i.e. , reducing the tails of the distribution function with new methods .this will allow significantly better background rejection .low - energy neutrino detectors could continue to provide invaluable information about the sun , supernovae , and neutrino properties .prominent goals include the @xmath4 solar flux , the dsnb flux , and the solar day - night mixing effect .super - k is large enough , but progress depends on reducing detector backgrounds . in the energy range618 mev , the dominant background is from the beta decays of unstable nuclei produced by cosmic - ray muons and their secondaries .super - k has strong cuts to reduce these backgrounds , but the residual rates are large .we are undertaking a multipart project to provide tools to significantly reduce these spallation backgrounds in super - k .our project , based on a foundation of careful simulation and theoretical insights , is the most extensive such effort undertaken for any detector . with modest adjustments, our results will be useful for other water - based detectors , e.g. , watchman @xcite and hyper - kamiokande @xcite .since these detectors are likely to be shallower than super - k , spallation backgrounds will be even more severe .more generally , our results will provide valuable insights about backgrounds in other underground detectors for neutrinos , dark matter , and other rare processes such as neutrinoless double beta decay . in our previous paper @xcite, we presented the first theoretical calculation of the spallation background yields in super - k .we focused on the steady - state background rates , averaged over muons and along their tracks .we found that almost all isotopes are produced by secondary particles , and not the primary muons themselves .our predictions for the spallation decay backgrounds agree with super - k aggregate data to within a factor of 2 , which is very good and could be improved .our results provide new information about components , correlations , and production mechanisms that can be used to develop cuts that are more powerful than those based on empirical studies .our next steps were inspired by a recent super - k dsnb analysis @xcite , where the cherenkov light profiles associated with individual muons were measured .these were found to vary along the muon tracks , showing peaks , with the positions of the peaks correlated with the sites of isotope production .a new cut was developed using this correlation , and was shown to be effective for improving the dsnb search .however , the cause for the variation in the light profile and its correlation with isotope production remained mysteries .this new cut has not yet been used for solar neutrino analysis .it seems very promising for reducing backgrounds without increasing deadtime . in the present paper ,we consider how isotope production varies between muons and along their tracks .we break the process of muons producing isotopes into muons producing energetic daughter particles , these daughter particles inducing electromagnetic and hadronic showers , and these showers producing isotopes .we provide details about each step and combine them in the end .our calculations here break our previous calculations @xcite into more steps , but agree in overall approach and results .our fundamental result is that showers are the key to explaining the correlation between muon light profiles and spallation backgrounds , as well as their total yields of spallation products in super - k .showers produce electrons , which make cherenkov light , and neutrons and pions , which make background isotopes . in fig .[ efficiency ] , we show how showers of different types and energies contribute to the production of light and isotopes . because of the high rate of electromagnetic showers , and the high efficiency of hadronic showers for making isotopes , electromagnetic showers strongly dominate light production and hadronic showers somewhat dominate isotope production .isotopes are nearly always proceeded by showers , though only a small fraction of showers produce isotopes . with these results , we reproduce super - k results on muon energy loss ( fig . [ light_iso ] ) , isotope production ( fig . [ efficiency_cumulative ] ) , and their correlations ( fig .[ seperation_distribution ] ) .we are the first to show that the background isotopes in super - k are dominantly made in discrete , identifiable showers .( it has long been known that isotope production is associated with muons with high radiative energy loss , e.g. , refs .@xcite and much subsequent work , but it had not been shown that these showers are rare enough and energetic enough to be identifiable , and that they account for the production of nearly all isotopes . ) though this paper focuses on super - k , our results have much more general applicability . the calculations and insights of this paper and of ref .@xcite can be used to define new cuts that should be very effective for solar and dsnb analyses .some could be implemented easily ( the muon energy loss and stopping muon cuts in sec .[ subsec : energyloss ] ) ; others improve the technique of ref .@xcite ( the efficiency of the technique depends on how well super - k reconstructs the muon light profile , sec . [ subsec : tagshower ] ) ; and others need new development ( our forthcoming papers ) .we will soon demonstrate new ways to better identify showers . as mentioned above ,the super - k reconstructed light profiles are inconsistent with what we expect from showers . in our next paper, we identify the reason for this inconsistency and will demonstrate better ways to reconstruct muon cherenkov light profiles . in the super - k reconstruction equation , which solves for the emission position of each individualphotomultiplier hit , there are two possible solutions for the light from deflected electrons ; we will show how to select the better solution , and that doing so improves the resolution .in addition , we will show how to isolate shower light from muon light , which also helps significantly .in subsequent papers , we will discuss new signals that can identify showers with even higher efficiency , followed by quantitative studies of the effects of new cuts on background rates and the implications for solar and supernova neutrino analyses .s.w.l . and j.f.b .were supported by nsf grants phy-1101216 and phy-1404311 to j.f.b .we are grateful to patrick allison , kirk bays , mauricio bustamante , anton empl , ciriyam jayaprakash , paolo lipari , masayuki nakahata , kenny ng , eric speckhard , michael smy , yasuo takeuchi , and mark vagins for helpful discussions .69ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ] + 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1103/physrevd.69.073006 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.112.091805 [ * * , ( ) ] link:\doibase 10.1016/j.physrep.2007.12.004 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.83.195 [ * * , ( ) ] link:\doibase 10.1155/2013/351926 [ * * , ( ) ] link:\doibase 10.1146/annurev - astro-081811 - 125539 [ * * , ( ) ] link:\doibase 10.1146/annurev.nucl.012809.104524 [ * * , ( ) ] link:\doibase 10.1146/annurev - nucl-102711 - 094901 [ * * , ( ) ] link:\doibase 10.1146/annurev - nucl-102711 - 095006 [ * * , ( ) ] link:\doibase 10.1088/0004 - 637x/778/2/164 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.85.245 [ * * , ( ) ] link:\doibase 10.1146/annurev.nucl.010909.083331 [ * * , ( ) ] link:\doibase 10.1016/s0168 - 9002(03)00425-x [ * * , ( ) ] link:\doibase 10.1016/j.nima.2013.11.081 [ * * , ( ) ] link:\doibase10.1103/physrevd.73.112001 [ * * , ( ) ] link:\doibase 10.1103/physrevd.78.032002 [ * * , ( ) ] link:\doibase 10.1103/physrevd.83.052010 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.90.061101 [ * * , ( ) ] link:\doibase 10.1103/physrevd.85.052007 [ * * , ( ) ] link:\doibase 10.1016/j.astropartphys.2014.05.004 [ * * , ( ) ] _ _ , http://www-sk.icrr.u-tokyo.ac.jp/~koshio/dthesis.html[ph.d . thesis ] , ( ) link:\doibase 10.1103/physrevlett.90.171302 [ * * , ( ) ] link:\doibase 10.1103/physrevc.89.045801 [ * * , ( ) ] link:\doibase 10.1103/physrevc.81.025807 [ * * , ( ) ] link:\doibase 10.1088/1475 - 7516/2013/08/049 [ * * , ( ) ] _ _ , http://www-sk.icrr.u-tokyo.ac.jp/sk/pub/kirk%20ryan%20bays_thesis%20feb%202012.pdf[ph.d . thesis ] , ( ) @noop `` , '' link:\doibase 10.1063/1.2720455 [ * * , ( ) ] link:\doibase 10.1103/physrevd.64.013012 [ * * , ( ) ] link:\doibase 10.1016/s0168 - 9002(03)00983 - 5 [ * * , ( ) ] link:\doibase 10.1103/physrevc.71.055805 [ * * , ( ) ] link:\doibase 10.1103/physrevc.72.025807 10.1103/physrevc.73.049906 , 10.1103/physrevc.73.049906 , 10.1103/physrevc.72.025807 [ * * , ( ) ] link:\doibase 10.1103/physrevd.73.053004 [ * * , ( ) ] @noop link:\doibase 10.1088/1475 - 7516/2014/08/064 [ * * , ( ) ] link:\doibase 10.1088/0031 - 9155/39/6/011 [ * * , ( ) ] @noop _ _ , ed .( , ) @noop _ _ , ed .( , ) link:\doibase 10.1103/physrevd.86.010001 [ * * , ( ) ] link:\doibase 10.1103/physrev.51.220 [ * * , ( ) ] link:\doibase 10.1098/rspa.1937.0082 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.13.240 [ * * , ( ) ] @noop _ _ ( , ) link:\doibase 10.1146/annurev.ns.10.120160.000431 [ * * , ( ) ] link:\doibase 10.1016/j.nuclphysbps.2009.09.060 [ * * , ( ) ] link:\doibase 10.1103/physrevd.79.063001 [ * * , ( ) ] link:\doibase 10.1088/0305 - 4616/8/10/016 [ * * , ( ) ] link:\doibase 10.1103/physrevd.45.362 [ * * , ( ) ] link:\doibase 10.1103/physrevd.65.103002 [ * * , ( ) ] link:\doibase 10.1016/j.nima.2005.03.169 [ * * , ( ) ] @noop `` , '' , link:\doibase 10.1088/0954 - 3899/30/2/009 [ * * , ( ) ] link:\doibase 10.1016/j.astropartphys.2004.09.003 [ * * , ( ) ] link:\doibase 10.1016/j.astropartphys.2005.09.002 [ * * , ( ) ] link:\doibase 10.1146/annurev.nucl.012809.104544 [ * * , ( ) ] link:\doibase 10.1016/j.astropartphys.2014.03.010 [ * * , ( ) ] link:\doibase 10.1146/annurev.nucl.58.110707.171154 [ * * , ( ) ] link:\doibase 10.1103/physrevd.88.055005 [ * * , ( ) ] link:\doibase 10.1146/annurev.ns.23.120173.002143 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1016/j.astropartphys.2012.01.006 [ * * , ( ) ] link:\doibase 10.1016/0168 - 9002(91)90396 - 8 [ * * , ( ) ] _ _ , http://ir.uiowa.edu/etd/472[ph.d . thesis ] , ( ) _ _ , http://www-sk.icrr.u-tokyo.ac.jp/sk/pub/[ph.d .thesis ] , ( ) link:\doibase 10.1016/s0168 - 9002(00)00900 - 1 [ * * , ( ) ] @noop ( ) , @noop _ _ , @noop ph.d .thesis , ( ) link:\doibase 10.1103/physrevlett.63.16 [ * * , ( ) ]

<S> crucial questions about solar and supernova neutrinos remain unanswered . </S><S> super - kamiokande has the exposure needed for progress , but detector backgrounds are a limiting factor . </S><S> a leading component is the beta decays of isotopes produced by cosmic - ray muons and their secondaries , which initiate nuclear spallation reactions . </S><S> cuts of events after and surrounding muon tracks reduce this spallation decay background by @xmath0 ( at a cost of @xmath1 deadtime ) , but its rate at 618 mev is still dominant . a better way to cut this background was suggested in a super - kamiokande paper [ bays _ et al . _ , </S><S> phys . </S><S> rev . </S><S> d * 85 * , 052007 ( 2012 ) ] on a search for the diffuse supernova neutrino background . </S><S> they found that spallation decays above 16 mev were preceded near the same location by a peak in the apparent cherenkov light profile from the muon ; a more aggressive cut was applied to a limited section of the muon track , leading to decreased background without increased deadtime . </S><S> we put their empirical discovery on a firm theoretical foundation . </S><S> we show that almost all spallation decay isotopes are produced by muon - induced showers and that these showers are rare enough and energetic enough to be identifiable . </S><S> this is the first such demonstration for any detector . </S><S> we detail how the physics of showers explains the peak in the muon cherenkov light profile and other super - k observations . </S><S> our results provide a physical basis for practical improvements in background rejection that will benefit multiple studies . for solar neutrinos , </S><S> in particular , it should be possible to dramatically reduce backgrounds at energies as low as 6 mev . </S>

1501.03504

blazars are extragalactic radio loud sources whose jet is pointing toward us ( for recent reviews , see e.g. bttcher 2007 ; ghisellini 2011 ; dermer 2014 ) . to be more precise, we may define blazars as the sources whose jet axis is at an angle @xmath1 with respect to the line of sight , where @xmath2 is the bulk lorentz factor .this implies that for each observed blazars , there are other @xmath3 intrinsically identical sources , but pointing in other directions .blazars are classically divided in two subclasses : flat spectrum radio quasars ( fsrqs ) and bl lacs .the usual divide between the two subclasses is based on the equivalent width ( ew ) of the optical broad emission lines : bl lacs have ew@xmath45 ( rest frame , urry & padovani 1995 ; see ghisellini et al .2011 and sbarrato et al .2012 for an alternative definition ) .blazars are strong @xmath0ray emitters as a class . after the detection of 3c 273 by the _ cos b _ satellite ( swanenburg et al . 1978 ) , the _ compton gamma ray observatory _ discovered nearly 100 blazars ( nandikotkur et al .now , in the era of the _ fermi _ satellite , nearly 1,000 blazars ( 2lac ; ackermann et al .2011 ) have been detected in the 0.1100 gev energy range .this number is sufficiently large to start population studies ( e.g. ajello et al . 2012 ; 2014 ) , to derive the average properties of these sources , and to compare them to other radio loud but not necessarily so extreme sources and to the general population of radio quiet quasars . to this aim, we can benefit from the study of shaw et al .( 2012 ) concerning fsrqs detected by _fermi _ and systematically observed spectroscopically . to these sources , we added the few bl lacs in shaw et al .( 2013 ) that do show broad emission lines .together , they form an homogeneous sample of blazars detected in @xmath0rays . due to the uniformall sky exposure of the _ fermi_/lat instrument , these blazars form a @xmath0ray flux limited sample of blazars with broad emission lines .we first compare their optical properties with the radio loud and the radio quiet quasars of the compilation made by shen et al .( 2011 ) , comprising about 105,000 quasars of the sloan digital sky survey ( sdss ) .then we collect multiwavelength data for all blazars , and apply a simple one zone leptonic model to find the physical parameters of the sources . with respect to our earlier studies of blazars ,there are four important additions : i ) the knowledge , for all sources , of the luminosity of the broad emission lines ; ii ) the knowledge of the black hole mass derived through the virial method ( these two information are given in the study of shaw et al .2012 ) ; iii ) the high frequency radio data given by the _ wmap _ and _ planck _ satellite observations ; iv ) the far ir data given by the _ wise _ satellite , that detects almost all sources . these data at low frequency allow to determine the synchrotron peak of powerful blazars much better than previously .we make use of the known broad emission line luminosity to derive the luminosity emitted by the accretion disc . by assuming a standard shakura & sunyaev ( 1973 ) disc , whose emitted spectrum depends only on the black hole mass and accretion rate ( or equivalently , on the emitted luminosity ) , we are able to infer the black hole mass for the majority of fsrqs , and compare this value with the mass derived with the virial method ( peterson & wandel 2000 ; mclure & dunlop 2004 , peterson 2014 ) .two important parameters of the model ( black hole mass and disc luminosity ) are therefore determined by the observations of the emission line luminosity and width and by disc fitting .when the spectral energy distribution ( sed ) are sufficiently sampled , ( i.e. such that the two spectral peaks of the non thermal emission are robustly determined ) the parameters of the model well constrained , although some extra assumptions are needed to univocally specify them .since blazars are rapidly variable , and often with large amplitude , one should use simultaneous multiwavelength data , which in our case is impossible , given the very large number of sources .therefore the results of each individual source can be affected by the non simultaneity of data . on the other handwe argue that , in a statistical sense , the derived distributions of the parameters are reliable .furthermore , consider that the accretion disc should vary with timescale longer than the non thermal jet continuum , and with smaller amplitudes .results on the black hole mass and the accretion luminosity should therefore be valid also for individual sources .the paper is organized as follows : in2 we present the samples of _ fermi_/lat detected blazars and the comparison sample of radio loud and radio quiet quasars . in 3 we discuss the general observational properties of the blazars of our sample in comparison with radio loud and radio quiet quasars ( namely the redshift and broad emission line strength distributions ) .we also describe how we derive the black hole masses using the disc fitting method , and compare our results with the ones derived through the virial method . in 4 we describe the blazar model we use to derive the physical parameters of the jet emitting region , and discuss how these parameters can be univocally determined by knowing the disc luminosity and the black hole mass , and in 5 we present the corresponding results . in 6we draw our conclusions .the results concerning the jet power and its relation to the accretion luminosity have been already discussed in ghisellini et al .( 2014 ) . in this work ,we adopt a flat cosmology with @xmath5 km s@xmath6 mpc@xmath6 and @xmath7 .we selected our sources from the fsrqs of shaw et al .( 2012 ) and the bl lacs of shaw et al . (we compare them with the radio quiet and radio loud quasars of shen et al .( 2011 ) . tab .[ numbers ] shows a break down of the samples .all our _ fermi_/lat fsrqs come from the sample of shaw et al .( 2012 , hereafter s12 ) .the sample includes 229 objects , present in the 1lac sample of abdo et al .( 2010 ) , that have been spectroscopically observed by shen et al .( 2012 ; 165 sources ) or by the sdss data release 7 ( dr7 , shen et al .2011 , 64 objects ) . of these, we have studied the 191 objects with enough multiwavelength information necessary to apply our model .the s12 sample of fsrqs does not include several bright and famous blazars with historical spectroscopic classifications in the literature .shaw et al . ( 2013 , hereafter s13 ) studied a very large sample of bl lac objects present on the second _ fermi _ catalog of agn ( 2lac , ackermann et al .2011 ) . in the original 2lac sample , there are 410 bl lacs , 357 fsrqs and 28 agn of other known types , and 326 agn of unknown type .s13 themselves were able to classify some of the several unknown type " agns present in the 2lac catalog , increasing the number of bl lacs to 475 , and decreasing the number of the sources of unknown type to 215 . by spectroscopically observing a large number of bl lacs , and by adding bl lacs of already known redshift, they assembled a sample of 209 bl lac with the redshift spectroscopically measured .in addition , they could constrain the redshift of other 241 bl lacs ( finding a lower limit on the redshift , see s13 for details and their tab . 1 for a break down on the number of known redshift for source type ) . by visually inspecting all sed of the bl lacs with redshifts , we selected the 26 objects with a clear presence of broad emission lines .although they are classified as bl lacs according to the classical definition ( rest frame equivalent width of the emission line less than 5 ) , they should be rather considered to belong to the tail , at low accretion luminosities , of fsrqs , and for this reason we include them in our sample .shen et al . ( 2011 , heafter s11 ) studied a large number of quasars selected from the sdss ( schneider et al .2010 ) according to the following criteria : i ) the ( rest frame ) fwhm of the broad lines greater than 1,000 km s@xmath6 , and ii ) the absolute magnitude is brighter than @xmath8=22.0 the quasars selected in this way are 105,783 .we then selected the region of the sky covered by the first survey . in this regionthere are 89,783 quasars that have been observed , but not detected by the first ( with 1 mjy flux limit at 1.4 ghz ) . for brevity , we call these sources radio quiet ". moreover , there are 9,393 radio detected quasars .of these , there are 8,257 radio loud sources , i.e. objects with a radio loudness @xmath9 .the radio loudness is defined as in s11 , i.e. the ratio of the rest frame 5 ghz flux and the 2500 flux , where the 5 ghz flux is extrapolated from the observed 1.4 ghz flux ( assuming a power law slope @xmath10 ) .

<S> we study the broad emission line blazars detected in the @xmath0ray band by the large area telescope onboard the _ fermi _ satellite and with the optical spectrum studied by shaw et al . </S><S> ( 2012 , 2013 ) . </S><S> the observed broad line strength provides a measure of the ionizing luminosity of the accretion disk , while the @xmath0luminosity is a proxy for the bolometric non thermal beamed jet emission . </S><S> the resulting sample , composed by 217 blazars , is the best suited to study the connection between accretion and jet properties . </S><S> we compare the broad emission line properties of these blazars with those of radio quiet and radio loud quasars present in the sloan digital sky survey , to asses differences and similarities of the disc luminosity and the virial black hole mass . for most sources </S><S> , we could derive the black hole mass by reproducing the ir </S><S> optical uv data with a standard accretion disc spectrum , and we compared the black hole masses derived with the two methods . </S><S> the distributions of the masses estimated in the two ways agree satisfactorily . </S><S> we then apply a simple , one zone , leptonic model to all the 217 objects of our sample . </S><S> the knowledge of the black hole mass and disc luminosity helps to constrain the jet parameters . on average </S><S> they are similar to what found by previous studies of smaller samples of sources . </S><S> [ firstpage ] galaxies : bl lacertae objects : general galaxies : quasars : general radiation mechanisms : non - thermal </S>

1504.01297

in 1948 , casimir @xcite calculated that quantum fluctuations of electromagnetic fields cause attraction between two parallel metal plates at zero temperature .this can be interpreted as a physical manifestation of the zero - point energy predicted by quantum field theory @xcite . as the plates are brought near each other ,the allowed modes between them are limited . because each electromagnetic mode has an energy of @xmath3 at @xmath4 k, the total energy of the configuration depends on the plate separation .this energy landscape results in an attractive force between the plates .while casimir s description holds for ideal metal plates , a more general expression was derived by lifshitz , dzyaloshinskii , and pitaevskii @xcite .this formulation is used for comparison with experiments and is valid for both real metals and dielectrics . in this theory, the force between two uncharged surfaces can be derived according to an analytical formula ( often called the lifshitz or casimir - lifshitz equation ) that relates the zero - point energy to the dielectric functions of the interacting surfaces and of the medium in which they are immersed . in 1972 ,parsegian and weiss @xcite derived an expression for the nonretarded ( where the speed of light is taken to be infinite ) interaction energy between two dielectrically anisotropic plates immersed in a third anisotropic material .barash @xcite analyzed a similar problem including retardation effects and found an equation for the helmholtz free energy per unit area . in the nonretarded limit ,the results of parsegian , weiss , and barash are in agreement . in the expressions derived by parsegian , weiss , and barash ,the casimir energy between two parallel , birefringent plates depends on both their separation and their orientation . between two parallel , positive birefringent slabs with in - plane optical anisotropy ,the lowest energy state is when the two optical axes are aligned .this results in a torque that rotates the plates to this configuration .while the casimir - lifshitz force is the subject of much discussion and has been verified in a number of experiments @xcite , there have been no published experimental attempts to measure the torque between anisotropic materials .in addition to the early work of parsegian , weiss and barash , recent theoretical work has been performed including a derivation of a more simplified equation of the torque between two plates in one dimension @xcite , the torque between two dielectric slabs with different directions of conductivity @xcite , and numerical calculations based on real materials @xcite .most discussions of potential casimir torque measurements involve either levitating microdisks @xcite or a torsion pendulum @xcite . in the first method, microscale birefringent disks would be levitated over a birefringent substrate with casimir - lifshitz repulsion or an electrostatic force .the casimir torque would then cause the freely rotating disks to rotate so that their optical axes would align with that of the substrate . in the proposed torsion pendulum experiment, a macroscopic birefringent crystal would be attached to a quartz torsion pendulum and brought near another birefringent crystal .the casimir torque would affect the period of natural oscillations of the torsion pendulum , allowing for its detection . herewe propose a method that is in analogy to a static torsion pendulum with a thick liquid crystal layer as the twisted bulk .as the uniformly aligned liquid crystal is brought near a birefringent crystal , the casimir torque aligns the liquid crystal molecules with the solid crystal s optic axis , which in turn causes a twist through the bulk of the liquid crystal .a similar experiment was proposed by smith and ninham @xcite in 1973 but , to our knowledge , was never carried out . herewe provide a calculation including retardation effects of the expected results of a similar geometry , as well as detailed experimental considerations . instead of using the liquid crystal in the isotropic phase as a spacer layer ( which is experimentally unfeasible ), a thin layer of sio@xmath0 separates the liquid crystal from the solid crystal .varying the sio@xmath0 thickness is equivalent to changing the distance between parallel plates .if both the liquid crystal and the birefringent crystal have positive uniaxial anisotropy , then the casimir torque causes the liquid crystal molecules to twist towards the extraordinary axis of the solid crystal .because the liquid crystal is anchored at the glass interface , the director ( i.e. the locally preferred molecular orientation ) is twisted through the bulk by the casimir torque at the boundary at the opposite interface .this geometry is depicted in fig . [fig : setup ] .this experimental design is similar to methods for measuring the azimuthal anchoring energy of liquid crystals on treated substrates @xcite .the mechanism of liquid crystal alignment induced by a rubbed polymer layer itself is the subject of much study but is thought to include physical grooves , aligned polymer chains , and the van der waals torque for surfaces in contact @xcite .the last of these is equivalent to the short - ranged casimir torque of our proposed study , though distinguishing the van der waals effect from other alignment effects is difficult .however , previous experiments have suggested the anisotropic van der waals effect as the mechanism of liquid crystal alignment at a surface .schadt and schmitt @xcite used linearly photopolymerized layers to align liquid crystals with a van der waals interaction .lu @xcite also provided evidence that the van der waals interaction is an important component of the liquid crystal alignment at treated polymer layers .however , these experiments do not isolate the van der waals torque from other surface effects , because the liquid crystal is in contact with the substrate .our proposed experiment would demonstrate this effect over a distance of tens of nanometers ( and in doing so measure the long - range retardation effects of the casimir torque ) . finally , we can relate measured data to casimir torques calculated from the dispersive properties of the materials .smith and ninham @xcite considered the nonretarded case of this system and predicted measurable distortions of the liquid crystal director . herewe carry out the full retarded calculation of the casimir torque by considering its effect on a thin boundary layer .that is , in comparison to the liquid crystal bulk with thickness @xmath5 m , most of the casimir torque is felt by a thin layer of thickness @xmath6 nm .also , because the total liquid crystal twist through the bulk is less than @xmath7 , the liquid crystal in the region of @xmath8 is nearly uniformly aligned .therefore , we approximate the casimir torque on this layer from the uniaxial crystal to be the same as that experienced by a uniformly aligned and semi - infinite liquid crystal slab . these approximations are treated with more detail in sec .[ sec : boundary ] .most materials undergo a phase transition from solid to liquid at a melting point .liquid crystals exhibit an intermediate phase in which there is some local ordering due to the molecular structure .nematic liquid crystals are the simplest class .they can be thought of as long molecules ( fig .[ fig : molecule ] ) that tend to align the long axis with their neighbors .the local direction of molecular orientation is written as a vector of unit length called the director @xmath9 . because the molecules have different dispersive properties along the different axes , one can change the properties of the liquid crystal bulk by manipulating the director .drawing of the 5cb molecule , which is a nematic liquid crystal at room temperature .the extraordinary axis is along the molecular axis .because 5cb has positive birefringence , @xmath10.,scaledwidth=50.0% ] we carry out calculations using the physical properties of 4-cyano-4@xmath1-pentylbiphenyl ( 5cb ) . in bulk , this is a positive uniaxial material so the dielectric function is higher along the long molecular axis .to derive the interaction between two infinite slabs , barash wrote the helmhotz free energy between two birefringent materials at finite temperature as an infinite sum over matsubara frequencies @xmath11 @xcite:@xmath12 where @xmath13 represents the distance between the two slabs and @xmath14 is the angular separation of their extraordinary axes .we have performed calculations for room temperature based on experimental considerations ; however , the zero - temperature calculation for the casimir torque will provide the correct result to within @xmath15@xmath16 for the separations presented in this work ( @xmath17 nm ) .the zeros of @xmath18 as a function of real frequency @xmath19 indicate allowed surface modes between the two infinite slabs .it can also be expressed in terms of the reflection matrices of the two materials , as in refs .@xcite . in our regime of interest , the energy has @xmath20 dependence to an excellent approximation , as in fig .[ fig : angle_dependence ] .the casimir torque is then calculated torque between batio@xmath21 and uniform 5cb bulk separated by 10 nm of sio@xmath0 as a function of angle @xmath14 between the extraordinary axes of the birefringent materials .the line is a fit to a @xmath22 dependence .the difference between the fit and the full calculation is at most 2% , and is less than 0.1% for @xmath23.,scaledwidth=50.0% ] @xmath24 where @xmath25 is a negative for the materials considered here ( which have positive birefringence ) .the dielectric functions of the two materials and of the intervening medium are evaluated at imaginary frequencies in eq .( [ eq : energy ] ) .we use the ninham - parsegian oscillator model to describe the dispersion of the solid crystals @xcite : @xmath26 for the 5cb liquid crystal , we use the dispersive properties at @xmath27 k calculated by kornilovitch @xcite using data from wu __ @xcite . there, the index of refraction is fit with a three - oscillator model , so the dielectric function is : @xmath28 for the birefringent materials , there are separate functions describing the ordinary and extraordinary axes .the model data used for our calculations are summarized in table [ oscillator ] ..model parameters for dielectric functions of relevant materials .oscillator data for batio@xmath21 , caco@xmath21 , tio@xmath0 , and sio@xmath0 are from ref.@xcite . [ cols="^,^,^,^,^,^,^,^",options="header " , ] we calculate the casimir torque between an infinite half - space of 5cb liquid crystal and an infinite half - space of several birefringent crystals using eqs .( [ eq : energy ] ) and ( [ eq : torque ] ) and the parameter data in table [ oscillator ] .the results for @xmath29 are shown in fig .[ fig : maxtorques ] .note that torques per unit area on the order of 10 nn / m are found for separations of @xmath30 nm . casimir torque per unit area between a half - slab of aligned 5cb and various birefringent crystals , when separated by a sio@xmath0 layer of thickness _d _ with a relative angle of @xmath7 between the extraordinary axes.,scaledwidth=50.0% ]the casimir torque causing a director twist at one boundary competes with the restoring torque from the twisted liquid crystal .the latter is modeled using the frank free energy density @xcite . in our geometry , the director is always aligned in the @xmath31-@xmath32 plane so its orientation can be written in cartesian coordinates as @xmath33 .there is no bend or splay of the liquid crystal , so only the twist term contributes to the distortion energy .the twist contribution is given by : @xmath34 where @xmath35 pn is the twist elastic constant of the 5cb@xcite .substituting our expression for @xmath9 , the frank free energy density is @xmath36 and the frank free energy per unit area is : @xmath37 where @xmath38 is the thickness of the liquid crystal layer . in our geometry ( as in fig .[ fig : setup ] ) , the extraordinary axis of the solid crystal is along the @xmath31-axis , so the top boundary at @xmath39 ( where the liquid crystal is in contact with glass ) is treated to induce alignment along @xmath40 .a torque applied at @xmath41 twists the director to @xmath42 . using calculus of variations ,the lowest energy configuration is given by a linear twist , @xmath43 , where @xmath44 .the elastic energy of the bulk per unit area is then : @xmath45 if the director at @xmath41 is twisted to @xmath46 , there is an energy penalty and associated restoring torque at that boundary .the restoring torque of the elastic bulk is given by : @xmath47 @xmath48 is the torque that must be applied at the boundary @xmath41 to twist the director to @xmath42 .the torque applied at the boundary twists the director until the torque balance equation is satisfied : @xmath49 .if the casimir torque is in the approximate form @xmath50 , where @xmath14 is the angle between the two extraordinary axes of the birefringent materials , then the torque is approximately @xmath51 ( which has the same form as the planar rapini - papoular approximation @xcite ) , and the torque balance equation yields : @xmath52 to predict the director twist for our proposed experiment , we calculate @xmath25 using eqs .( [ eq : energy ] ) and ( [ eq : torque ] ) and then numerically solve eq .( [ eq : balance ] ) to find the twist caused by the casimir torque , @xmath53 . in the proposed experiment , @xmath53 ( or @xmath46 ) will be measured to obtain @xmath25 via @xmath54the liquid crystal can be treated as an anisotropic bulk material for the calculation because the twisting is small throughout the thickness that experiences the casimir interaction .the spacing between the liquid crystal and solid birefringent crystal is on the order of @xmath5510 nm , and the liquid crystal layer is about 100 @xmath56 m thick . following parsegian @xcite , the penetration depth of the casimir interaction is on the order of the material separation .so , the most important region of the liquid crystal is the 10 nm in contact with the sio@xmath0 ( or conservatively , the 100 nm ) .this 100 nm is the 0.1% of the liquid crystal nearest the birefringent crystal .because the liquid crystal will be twisted a maximum of @xmath57 radians throughout the bulk , the liquid crystal director will vary by a maximum of @xmath580.05@xmath2 in the relevant region for the casimir torque , which has no appreciable effect on the torque s magnitude .this approximation can also be justified by considering the reflection matrix of the liquid crystal stack .the casimir energy of the system is a function of the reflection matrices of the two materials at the matsubara frequencies .our method assumes that , at the matsubara frequencies , the reflection matrix of the slowly twisted liquid crystal is nearly the same as that of an untwisted , bulk liquid crystal with the same alignment at the boundary .the first matsubara frequency at room temperature , @xmath59 thz , corresponds to a wavelength of @xmath60 m. the higher frequences correspond to shorter wavelengths , so this first term has the longest penetration depth .we calculated the reflection matrices at this frequency between the twisted and untwisted stacks using the berreman @xmath61 matrix method , and found them to be numerically identical to four significant figures @xcite .the casimir interaction energy is largely unaffected by the slow twist of the liquid crystal throughout the bulk .hence , to several significant figures , the torque experienced by the liquid crystal layer is felt entirely at the nearest boundary and is only a function of the director orientation at that boundary .common methods for fabricating single liquid crystal cells have been previously reported in refs . @xcite and can be used for this experiment .a rubbed alignment layer of polyvinyl alcohol ( pva ) can be used to cause the liquid crystal molecules to align along the rubbed direction at the surface .the birefringent crystal with a thin , isotropic sio@xmath0 layer ( with thickness @xmath62 nm ) is sandwiched with the pva - treated glass with a spacing of @xmath63 @xmath56 m ( this value can be measured optically ) .the liquid crystal is then introduced into the cell via capillary action .the filling process may induce some alignment along the direction of liquid crystal flow ; however , baking the sample above the lc clearing temperature ( @xmath64 c for 5cb ) and allowing it to cool slowly will eliminate this effect .as the liquid crystal cools to room temperature , the director settles into the lowest energy state described in sec .[ sec : balance ] .the magnitude of the casimir torque effect can then be measured by observing the twist of the liquid crystal director .the final director twist @xmath46 can be measured optically .this method is similar to a technique for measuring azimuthal surface anchoring strengths of liquid crystals @xcite . when linearly polarized light is incident on an adiabatically twisted nematic liquid crystal stack ( in which the pitch of the twist is much larger than the wavelength of light ) , the polarization state is rotated to follow the liquid crystal director .this is known as the adiabatic approximation for twisted nematics and is the principle behind twisted nematic liquid crystal displays @xcite . in this experiment ,white light polarized at @xmath7 shines onto the stack as in fig .[ fig : setup ] .the jones vector of this light is @xmath65 where @xmath66 is the intensity .the light polarization follows the director of the twisted nematic and is incident on the transparent sio@xmath0 layer with polarization @xmath46 .its jones vector is now @xmath67 . in a typical measurement of an anchoring force, the liquid crystal is sandwiched between two glass slides that do not interfere with the polarization state of the light , as in ref . @xcite .then , the polarization state @xmath46 can be measured with a second polarizer .caculated twist of a 100 m layer of 5cb caused by a casimir torque induced by various birefringent crystals at distance @xmath13 from one end of the liquid crystal stack .the incident light is polarized at @xmath7 to the ordinary axis at the top of the 5cb stack , but the director is twisted by the casimir torque , which causes the light polarization to rotate @xmath53 towards the extraordinary axis at @xmath68.,scaledwidth=50.0% ] to calculate the expected results from an experiment , we consider a liquid crystal layer thickness of 100 @xmath56 m and calculate the lowest energy state of the system using eq .( [ eq : balance ] ) .figure [ fig : angle_results ] shows the expected results for this case .the liquid crystal bulk is predicted to twist by over @xmath64 when the stack is separated by 5 nm from batio@xmath21 , and a twist of several degrees is expected for separations of @xmath69 nm ( well into the casimir regime ) .when near a birefringent material that has negative birefringence over a large frequency range ( such as lithium niobate ) , the liquid crystal would twist towards the ordinary axis instead of the extraordinary axis . this would provide further confirmation that dispersion effects are causing the director to twist .we have proposed an experiment for measuring a casimir torque between a birefringent crystal and a liquid crystal separated by an isotropic spacer layer .we provide complete calculations of the expected results for several materials at a range of separations and include details for a proposed experiment .this experimental design avoids many of the difficulties involved with a torsion pendulum or levitating microdisks .40ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ] + 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty http://scholar.google.com/scholar?hl=en&btng=search&q=intitle:on+the+attraction+between+two+perfectly+conducting+plates#0 [ * * ( ) ] link:\doibase 10.1119/1.17618 [ ] ( ) link : [ * * , ( ) ] http://www.turpion.org/php/full/infoft.phtml?journal_id=pu&paper_id=3330 [ * * , ( ) ] http://www.tandfonline.com/doi/abs/10.1080/00218467208072197 [ ( ) ] @noop * * , ( ) \doibase http://dx.doi.org/null [ * * , ( ) ] link:\doibase 10.1038/nature07610 [ * * , ( ) ] link:\doibase 10.1016/j.aop.2005.03.007 [ ( ] ) , * * , ( 2005 ) link:\doibase 10.1126/science.1057984 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.88.041804 [ * * , ( ) ] , link:\doibase 10.1103/physrevlett.81.4549 [ * * , ( ) ] link:\doibase 10.1103/physreva.52.2569 [ * * , ( ) ] link:\doibase 10.1103/physrevd.63.121701 [ * * , ( ) ] link:\doibase 10.1103/physreva.71.042102 [ * * , ( ) ] link:\doibase 10.1088/1367 - 2630/8/10/244 [ * * , ( ) ] link:\doibase 10.1002/pssb.201147150 [ * * , ( ) ] http://www.sciencedirect.com/science/article/pii/003189147390267x [ * * , ( ) ] link:\doibase 10.1103/physreve.68.026601 [ * * , ( ) ] link:\doibase 10.1063/1.121390 [ * * , ( ) ] link:\doibase 10.1109/jdt.2005.852493 [ * * , ( ) ] http://www.springerlink.com/index/r5n0936q71h65p43.pdf [ * * , ( ) ] link:\doibase 10.1039/c1sc00696g[c]hem .* * , ( ) http://iopscience.iop.org/1347-4065/31/7r/2155 [ * * , ( ) ] link:\doibase 10.1143/jjap.43.8156 [ ] * * , ( ) link:\doibase 10.1088/0305 - 4470/39/21/s81 [ * * , ( ) ] , link:\doibase 10.1364/oe.21.014943 [ * * , ( ) ] @noop _ _ ( , , ) link:\doibase 10.1088/0953 - 8984/25/3/035102 [ * * , ( ) ] http://opticalengineering.spiedigitallibrary.org/article.aspx?articleid=1072612 [ * * , ( ) ] http://www.sciencedirect.com/science/article/pii/s0001868697000031 [ * * , ( ) ] http://pubs.rsc.org/en/content/articlepdf/1958/df/df9582500019 [ * * , ( ) ] http://iopscience.iop.org/1347-4065/26/12r/1959 [ * * , ( ) ] http://jphyscol.journaldephysique.org/articles/jphyscol/abs/1969/04/jphyscol196930c413/jphyscol196930c413.html [ * * , ( ) ] http://books.google.com/books?hl=en&lr=&id=k4f7uhjveooc&oi=fnd&pg=pp1&dq=van+der+waals+forces:+a+handbook+for+biologists,+chemists,+engineers,+and+physicists&ots=hsdfqcowax&sig=f89nt3v_1vqtpggfjc5ein7zhgo[__ ] ( , , ) link:\doibase 10.1364/josa.62.000502 [ ] * * , ( ) @noop * * , ( ) http://pubs.acs.org/doi/abs/10.1021/ed081p854 [ * * ( ) ] link:\doibase 10.1063/1.350965 [ * * , ( ) ] link:\doibase 10.1103/physreve.72.061705 [ * * , ( ) ]

<S> we present a calculation of the casimir torque acting on a liquid crystal near a birefringent crystal . in this system , </S><S> a liquid crystal bulk is uniformly aligned at one surface and is twisted at the other surface by a birefringent crystal , e.g. barium titanate . </S><S> the liquid crystal is separated from the solid crystal by an isotropic , transparent material such as sio@xmath0 . by varying the thickness of the deposited layer </S><S> , we can observe the effect of retardation on the torque ( which differentiates it from the close - range van der waals torque ) . </S><S> we find that a barium titanate slab would cause 5cb ( 4-cyano-4@xmath1-pentylbiphenyl ) liquid crystal to rotate by 10@xmath2 through its bulk when separated by 35 nm of sio@xmath0 . </S><S> the optical technique for measuring this twist is also outlined . </S>

0706.3139

generalized parton distributions ( gpds ) are a versatile tool to quantify important aspects of hadron structure in qcd .they contain unique information on the spatial distribution of partons @xcite and on the orbital angular momentum they carry in the proton @xcite .the theoretically cleanest process where gpds can be studied is deeply virtual compton scattering ( similar to inclusive dis , which plays a dominant role in constraining the usual parton densities ) .hard exclusive meson production is harder to describe quantitatively , but it provides opportunities to obtain important complementary constraints . in particular , vector meson production is more directly sensitive to the gluon distributions , which enter the compton amplitude only at next - to - leading ( nlo ) order in @xmath1 .together with a wealth of high - quality data @xcite , this warrants efforts to bring meson production under theoretical control as much as possible . in the present contribution@xcite we investigate exclusive @xmath2 production ( @xmath3 ) using collinear factorization , which is applicable in the limit of large photon virtuality @xmath4 at fixed bjorken variable @xmath0 and fixed invariant momentum transfer @xmath5 to the proton @xcite . in practical terms , this means that the description is restricted to sufficiently large @xmath4 but can be used for both small and large @xmath0 , thus providing a common framework for analyzing both collider and fixed - target data .the process amplitude can then be expressed in terms of gpds for the proton , the @xmath6 distribution amplitude for the @xmath2 , and hard - scattering kernels .the kernels are known to nlo , i.e. to order @xmath7 @xcite .the requirement of `` sufficiently large '' @xmath4 is demanding for meson production .contributions that are formally suppressed by powers of @xmath8 can not be calculated in a completely systematic way , but the estimates @xcite agree that for @xmath4 of several @xmath9 the effect of the transverse quark momentum inside the meson can not be neglected in the hard - scattering subprocess , as it is done in the collinear approximation .this effect can be incorporated in the modified hard - scattering picture @xcite , in color dipole models @xcite , or in the mrt approach @xcite .unfortunately , the calculation of @xmath1 corrections remains not only a technical but even a conceptual challenge in these approaches , so that the perturbative stability of their results can not be investigated at present .one strategy in this situation is to study the nlo corrections in the collinear factorization framework , identifying kinematical regions where they are moderate or small .there one may use formulations incorporating power corrections from transverse quark momentum with greater confidence .this is the aim of the present contribution . in the followingwe show results for the convolution of the unpolarized quark and gluon gpds @xmath10 and @xmath11 with the corresponding hard - scattering kernels and the asymptotic form of the @xmath2 distribution amplitude .we model the gpds using a standard ansatz based on double distributions @xcite , with the cteq6 m distributions as input .unless indicated explicitly , we take @xmath12 and set the factorization and renormalization scales equal , @xmath13 .imaginary part of the convolution integral for the sum of gluon and quark singlet distributions as a function of the renormalization and factorization scale @xmath14.,title="fig : " ] imaginary part of the convolution integral for the sum of gluon and quark singlet distributions as a function of the renormalization and factorization scale @xmath14.,title="fig : " ] r0.5 in a wide kinematical range at small @xmath0 , we find huge nlo corrections which have opposite sign to the born term and almost cancel it .this is shown for @xmath15 in fig .[ fig : small - x ] , where there are indications for an onset of perturbative stability at @xmath16 , but not yet at @xmath17 . taking@xmath18 one finds no stability even at @xmath19 , whereas for @xmath20 the corrections are of tolerable size already at @xmath17 .figure [ fig : zeus ] shows that in kinematics relevant for hera measurements , nlo corrections have a huge effect on the cross section and moreover lead to a flat energy behavior in conflict with experiment . due to the strong cancellations between lo and nlo terms ,the dependence on factorization and renormalization scale is not reduced when going to nlo . as already observed in @xcite the large size of nlo corrections at small @xmath0can be traced back to bfkl - type logarithms appearing first at nlo for vector meson production .such logarithms are present in many processes ( including dis ) but have a rather large numerical prefactor in the present case .it is to be hoped that all - order resummation of these logarithms in the hard - scattering kernel will give perturbative stability at small @xmath0 . in the @xmath0 range relevant for experiments at compass , hermes , and jlab, we generally find corrections which are sizable but not huge .an exception is the real part of the gluon and quark singlet amplitudes , where corrections become large for decreasing @xmath0 , as is seen in the left panel of fig .[ fig : large - x ] .real and imaginary part of the convolution integral for the sum of gluon and quark singlet distributions for @xmath21.,title="fig : " ] real and imaginary part of the convolution integral for the sum of gluon and quark singlet distributions for @xmath21.,title="fig : " ] in the quark nonsinglet sector there are large terms in the nlo kernel due to gluon self - energy corrections .the blm procedure for setting the renormalization scale aims at resumming these to all orders in @xmath1 .applied to the process at hand , one finds however that this requires @xmath22 to be typically an order of magnitude smaller than @xmath23 @xcite .this is outside the validity of the perturbative calculation for most practically relevant @xmath23 .numerically we find that for @xmath24 the nlo corrections become unstable for several convolution integrals , as shown for examples in fig .[ fig : scale ] .renormalization scale dependence of the real part of the convolution integrals for the sum of gluon and quark singlet distributions ( left ) and for for the difference of @xmath25 and @xmath26 quark distributions ( right).,title="fig : " ] renormalization scale dependence of the real part of the convolution integrals for the sum of gluon and quark singlet distributions ( left ) and for for the difference of @xmath25 and @xmath26 quark distributions ( right).,title="fig : " ] we have therefore omitted this region when estimating the scale setting error in fig . [fig : fixed - tar ] , where we show the cross section in typical fixed - target kinematics .we see that nlo corrections are quite large for @xmath27 , whereas for @xmath28 and @xmath29 they become moderate .cross section for @xmath30 with a longitudinal photon .bands correspond to the range @xmath31 in the left and to @xmath32 in the right plot , and solid lines to @xmath33 in both cases ., title="fig : " ] cross section for @xmath30 with a longitudinal photon .bands correspond to the range @xmath31 in the left and to @xmath32 in the right plot , and solid lines to @xmath33 in both cases ., title="fig : " ]we gratefully acknowledge discussions with l. favart , h. fischer , d. yu . ivanov , a. rostomyan and a. schfer .this work is supported by the helmholtz association , contract number vh - ng-004 .m. burkardt , int .j. mod .a * 18 * ( 2003 ) 173 [ hep - ph/0207047 ] .x. d. ji , phys .* 78 * ( 1997 ) 610 [ hep - ph/9603249 ] .a. levy , these proceedings , arxiv:0706.1867 [ hep - ex ] . slides : ` http://indico.cern.ch/contributiondisplay.py?contribid=83&sessionid=12&confid=9499 ` j. c. collins , l. frankfurt and m. strikman , phys .d * 56 * ( 1997 ) 2982 [ hep - ph/9611433 ] .ivanov , l. szymanowski and g. krasnikov , jetp lett .* 80 * ( 2004 ) 226 [ hep - ph/0407207 ] .s. v. goloskokov and p. kroll , eur .j. c * 50 * ( 2007 ) 829 [ hep - ph/0611290 ] ; + p. kroll , these proceedings , arxiv:0706.2587 [ hep - ph ]. m. vanderhaeghen , p. a. m. guichon and m. guidal , phys .d * 60 * ( 1999 ) 094017 [ hep - ph/9905372 ] .l. frankfurt , w. koepf and m. strikman , phys .d * 54 * ( 1996 ) 3194 [ hep - ph/9509311 ] .a. d. martin , m. g. ryskin and t. teubner , phys .d * 55 * ( 1997 ) 4329 [ hep - ph/9609448 ] ; + t. teubner , these proceedings .i. v. musatov and a. v. radyushkin , phys . rev .* d61 * ( 2000 ) 074027 [ hep - ph/9905376 ] .zeus collaboration , paper 594 submitted to eps 2001 , budapest , hungary , + ` http://www-zeus.desy.de/physics/phch/conf/eps01_paper.html ` a. v. belitsky and d. mller , phys .b * 513 * ( 2001 ) 349 [ hep - ph/0105046 ] .i. v. anikin _ et al ._ , eur .j. c * 42 * ( 2005 ) 163 [ hep - ph/0411408 ] ; + s. j. brodsky and f. j. llanes - estrada , eur .j. c * 46 * ( 2006 ) 751 [ hep - ph/0512247 ] .

<S> we report on numerical studies of the nlo corrections to exclusive meson electroproduction , both in collider and fixed - target kinematics . </S><S> corrections are found to be huge at small @xmath0 and sizeable at intermediate or large @xmath0 . </S><S> desy 07 - 093 </S>

cond-mat0012250

as a start , let us write the 3d time - independent schrdinger equation in @xmath0 and @xmath1 ( fermi wavelength and fermi energy ) units : @xmath2 \phi(x , y , z ) = 0 \label{eq1}\ ] ] a general waveguide geometry can be formed by discretization in _ constant section slices _ ( where the confining potential @xmath3 only depends on @xmath4 ) . for a quantum waveguide of _ constant section _ , the solution of the time - independent schrdinger equation is separable as : @xmath5 , being the separated equations : @xmath6 @xmath7 \psi = 0 \label{eq4}\ ] ] where the longitudinal wavevector @xmath8 is the arbitrary separation constant .the general solution for the propagating part of the wave function is : @xmath9 , where @xmath10 and @xmath11 are arbitrary constants to be determined . in @xmath12we have a generic quantum well determined by the confinement potential @xmath13 .we will assume that this well has some `` hard wall '' boundary conditions at a given limit .the discrete set of eigenvalues @xmath14 ( being @xmath15 ) can be found analytically or numerically .the known solutions of equation [ eq4 ] form a complete basis of transversal eigenfunctions @xmath16 inside the hard - wall limit , so one can write the complete solution for the uniform section waveguide as an expansion in this basis : @xmath17 where the quantity in between brackets can be seen as the correspondent complex coefficients of the expansion .this leads to a vectorial notation in which the coordinates of @xmath18 in @xmath19 are given by @xmath20 , where the @xmath21-th elements of vectors @xmath22 and @xmath23 are @xmath24 and @xmath25 .the basis @xmath19 is completely characterized by the confining potential @xmath13 .mode - matching techniques can be used to obtain the scattering matrix ( sm ) for the simplest non uniform system : a single junction .a single junction discontinuity is shown in fig .[ fig1 ] , the general solution and derivatives in the two uniform waveguide sections are given , in matrix notation , by : @xmath26 @xmath27 where diagonal matrix @xmath28 is given by @xmath29 ( analogous definition for @xmath30 ) .continuity condition for the wave function and its normal derivative at the junction leads to these following two matrix equations : @xmath31 @xmath32 here , matrix @xmath33 characterizes the intermode scattering or mode coupling due to the step discontinuity and is defined by overlap integrals .the _ change of basis matrix _ to change from basis @xmath34 to basis @xmath19 , @xmath35 , can only be defined if the eigenvectors of the basis @xmath34 can be expanded in terms of basis @xmath19 ; this implies that the definition domain for @xmath19 should include the one for @xmath34 i.e. , in geometrical terms , the hard - wall closed line of @xmath34 is completely inside the one for @xmath19 ( see fig . [ fig1 ] ) .the element @xmath36 of the change of basis matrix @xmath37 is given by : @xmath38 where @xmath39 is the area in the plane @xmath12 enclosed by the hard - wall limit correspondent to @xmath19 . from matrix equations [ eq10 ] and [ eq11 ] the generalized sm defined as : @xmath40can be readily obtained and is given by : @xmath41 @xmath42 @xmath43 @xmath44 where : @xmath45 @xmath46 @xmath47 ^{-1 } \label{eq15c}\ ] ] @xmath48 the analysis of a single junction similar to that shown in fig .[ fig1 ] with the only difference that now @xmath34 includes @xmath19 , is completely analogous , equivalent equations and sms are given in the next .following two matrix equations are equivalent to eq .[ eq10 ] and [ eq11 ] : @xmath49 @xmath50 the same definition ( eq .[ eq12 ] ) is valid for the _ change of basis matrix _ to change from @xmath19 to @xmath34 . from matrix equations [ eq16 ] and [ eq17 ] the scattering matrix , defined as stated in eq .[ eq13 ] , is given by : @xmath51 @xmath52 @xmath53 @xmath54 where same definitions [ eq15a]-[eq15d ] are valid if one swaps @xmath19 and @xmath34 indexes .the sm of a composite waveguide structure may be obtained by combining sms of the corresponding junctions .the procedure to obtain the sm of a composite waveguide structure is described in the following .let us assume two consecutive junctions joined by a uniform waveguide section as shown in fig .individual sm for both single junctions have been obtained as described above and are defined by : @xmath55 @xmath56 we will describe now formulae to obtain the total sm of the system defined by : @xmath57 since the middle connection is a uniform waveguide we can write : @xmath58 where @xmath59 is a diagonal matrix with elements given by : @xmath60 . with some matrix algebra from matrix equations [ eq20 ] and [ eq21 ] , using [ eq23 ] ,one can obtain the wanted sm of the whole composite system : @xmath61 @xmath62 s_1^b \label{eq25b}\ ] ] @xmath63 @xmath64 where the diagonal identity matrix has been noted as `` 1 '' and @xmath65 is : @xmath66 ^{-1 } \label{eq26}\ ] ] this same solution can be expressed in other equivalent ways , nevertheless , we found that this is the best formula to reduce computational time since it requires minimum number of matrix inversions : just one . for a greater number of junctions system ,one starts similarly calculating each single - junction sm .then , the iterative use of this combining recipe leads to the total sm . in the case of cylindrical symmetry ,overlapping integrals get easier since can be reduced to one dimension .let us detail the expressions for the simple case of plane hard - wall confinement .the symmetry assures that will be no matching of modes belonging to different acimutal quantum numbers @xmath67@xcite , so we obtain a 2d equivalent problem for each @xmath67 . the number of 2d problems that one needs to solve increases with tubes width .let us consider a _cylindrical uniform _ waveguide section .electrons are confined with hard - walls at distance @xmath11 of axis @xmath68 , the confining potential @xmath69 is only function of the distance to this axis .thus , eq . [ eq1 ] can be separated in cylindrical coordinates , so that the wavefunction can be written as : @xmath70 where : @xmath71 being @xmath72 and @xmath73 to be specified below . due to the cylindrical symmetry, the problem can be considered separately for each value of @xmath67 ; _ therefore , in the following , we will focus in the problem for a fixed @xmath67_. functions @xmath72 are the solutions of the separated radial equation : @xmath74 + \\ \ , \ , \, + \left\ { 4 \pi^2 [ e - v(r ) ] - [ k_{ml}^2 + \frac{m^2}{r^2 } ] \right\ } \xi_{ml}^r(r ) = 0 \\ \nonumber \mbox{with the boundary condition : } \xi_{ml}^r(r ) = 0 \end{array } \right .\label{eqradial}\ ] ] let us consider the basis of these radial eigenfunctions defined between @xmath75 and @xmath11 : @xmath76 .then , the wavefunction @xmath77 can be expressed as an expansion in the basis @xmath78 with some coefficients that depend on @xmath79 and @xmath80 , written as @xmath81 above . in matrix notationthis is : @xmath82 where : @xmath83 @xmath84 being @xmath85 the @xmath80 component of the wavevector for the subband @xmath86 .therefore given the ( infinite ) set of complex numbers @xmath87 and given the basis @xmath78 , we have completely described the wavefunction @xmath77 . for any other radius @xmath88the solution is , obviously , analogous .this formulation of the problem gives the correspondent sm for each @xmath67 quantum number .matrix equations are the same than in previous sections if we specify the change of basis matrix @xmath33 and the transverse wavevector diagonal matrix @xmath89 .so that for a single junction joining tubes of radius @xmath19 and @xmath34 , being @xmath34 lower than @xmath19 ( similar to that represented in fig . [fig1 ] ) we have : @xmath90 @xmath91 ^ 2 = 4 \pi^2 e - [ q_{ml}(r)]^2\ ] ] being @xmath92 given by the transversal quantized levels .the calculation of the relevant integral , for the _ simple case of hard - wall boundary conditions ( @xmath93 constant inside the tubes ) _ , yields : @xmath94 } { [ \alpha_{mh}/b]^2 - [ \alpha_{mj}/a]^2 } \end{array}\ ] ] in this case , radial wavefunctions are given by : @xmath95 being @xmath96 the zero number @xmath97 of @xmath98 , the bessel function of integer order @xmath67 , and @xmath85 the @xmath80 component of the wavevector given by : @xmath99 - [ \alpha_{ml}/r]^2 } $ ] . within this framework , one can easily evaluate the conductance of a two terminal system on the basis of landauer s scattering approach @xcite , which is given by the formula : @xmath100 where the sum runs to all propagant modes in the input lead ( propagant modes are defined as those with a correspondent real @xmath8 , while the rest -those with imaginary longitudinal wavevector @xmath8- are called evanescent ) . here, @xmath101 is the universal spinless quantum of conductance with the value @xmath102 .the transmission probability for incidence through mode @xmath103 can be easily obtained and is given by : @xmath104 where @xmath105 is the @xmath106 element of the @xmath107 matrix given in equation [ eq25c ] , the sum runs to all propagant modes of the final lead , @xmath108 and @xmath109 are the @xmath21 and @xmath103 longitudinal wavevectors of the final and initial tubes respectively .one crucial point in the correct calculation of the sm of a quantum waveguide junction is the right truncation of two infinite series of modes representing wave functions at both sides of the junction . in the analysis of junctions ,different converged results may be obtained for different ratios of the number of modes retained on each side .this convergence problem is usually referred to as relative convergence phenomenon @xcite .theoretical studies have found that the correct converged result is obtained if the ratio of modes on each side is taken the same as the ratio of the corresponding waveguide widths @xcite .further calculations give another optimum value for this ratio which is 1.5 times the ratio of waveguide widths @xcite . in our analysiswe have found that one can generalize this optimal ratio criterium to an optimal `` procedure '' which gives different ratios in different cases .our approach is based on observations of the coupling between different modes at waveguide junctions . for simplicity ,let us consider the 2d axial symmetric hard - wall problem of a single junction like the one shown in the inset of fig .[ fig4 ] . assuming incidence from left to right and for a fixed incident mode @xmath103 ,the plot of the coupling integral @xmath110 versus `` wide '' transversal modes , @xmath21 , is `` peaked '' at @xmath111 such that its transverse wave vector satisfies @xmath112 .thick dots in fig .[ fig4 ] represent values of this overlaping integral , the peak can be clearly seen at @xmath112 ( in this simple case @xmath113 and @xmath114 ) . in a `` mean field approximation ''@xcite true overlaps are approximated by a uniform coupling to all modes within one level spacing , represented by the dashed line in fig .this approximation gives excellent total transmission results when the difference in widths is big @xcite .we can conclude that exists a given band of modes of the wide side effectively coupled to the @xmath103-th mode of the narrow side .one needs to take into account all modes in this continuous interval to assure a correct representation of the `` narrow '' mode @xmath103 in the `` wide '' basis of transversal eigenfunctions .the truncation recipe that we have found is based on this observation . though we have shown here a simple case , this mode matching scheme holds for different confining potentials and presumably for all of them ( in particular we have checked some axial symmetric potentials in 2d and 3d : hard - wall , soft wall and harmonic .also some non symmetric 2d and 3d potentials have been checked : hard - wall and soft - wall ) consequently , for one single junction similar to that in fig .[ fig1 ] , the truncation procedure is as follows .given transversal levels sorted in crescent energy : ( i ) take one arbitrary number of modes @xmath115 in the narrow side , greater or equal than the number of propagant ones .this implies that @xmath115 can be written as @xmath116 , where @xmath117 is the number of propagant modes and @xmath118 is the number of evanescent modes in the narrow tube .( ii ) take all modes of the wide side which are effectively connected to the ones chosen in the narrow side .this implies that we should use every wide - mode whose energy lies bellow @xmath119 given by @xmath120 .in other words , @xmath119 is the energy correspondent to the @xmath121 level of the narrow side . for the truncation in the case of a greater system , in which the number of junctions is larger, the procedure is the same but choosing the narrowest of the whole system ( i.e. , the one that has the sparser transversal energy levels ) as the tube to define the threshold transversal energy @xmath119 .this @xmath119 value is then used to fix the number of modes retained at the rest of the uniform waveguide sections of the system . using this procedure ,the number of evanescent modes taken in the narrowest tube @xmath118 is the only free parameter , it can be tuned for proper convergence of the results . in fig .[ fig5 ] we can see the convergence of the conductance versus @xmath118 ( note that the number of evanescent modes taken into account in the wide side could be quite larger than @xmath118 ) .the plot has been done for the 3d `` hard - wall '' problem and is quite representative of the general behaviour for different confining potentials .the problem converges for @xmath118 equal four with an error of @xmath122 .the crucial point is that this way of making the truncation ensures good coupling of every mode at the narrow side , and the appropriate description of these modes is the main influence in conductance values .it is worth to notice here that matrices involved are rectangular in general .algebraical operations done assure the appropriate use of them : no numerical inversion of non - squared matrices has been done .one good property of this particular way of treating matrices involved is that flux is conserved with any truncation chosen ; this implies that @xmath123 even for @xmath124 ( where @xmath125 is the total transmission and @xmath126 is the total reflection for one particular propagant mode @xmath103 in the initial lead ) . by the use of the matrix algebra proposed here, numerical unitarity and stability are achieved with ease .the sm formalism used , in which information from both ends of the system is taken in each step , avoids numerical roundoff errors accumulation in particular when combined with a suitable change of transversal basis as a function of the local width of the system .the truncation procedure explained above is the key for the achievement of these characteristics : it produces a self filtering of numerically unstable components without further data manipulation , hence the unitarity ( to compare with recursion - transfer - matrix methods see for instance ref .for several years now and in various experimental ways , the formation of ultrathin wires between two macroscopic pieces of metal has been routinely achieved and characterized @xcite .however , direct information about the atomic structure of these wires has not been available until recently when various experiments have achieved atomic resolution tem images on gold nanowires @xcite .one of the most interesting features seen is the almost ideal perfect geometry , as well as surprising helical structures@xcite . to the date , theoreticalwork has been focused on the study of the stability and electronic structure for the thinnest monoatomic chains @xcite ; while for the helical wires works have been focused esentially in the prediction of the very structure for various metals @xcite , and the influence of helicity in the ideal ballistic conductance @xcite . in the second part of this work-by making use of the calculation method we have just described above- we approach this problem .we will point out some remarkable effects in the current and conductance patterns due to the ideal _ sharp _ geometries observed , these effects are valid in the presence or not of helicity .for a study on the effects of global _ smooth _ shapes in the conductance of atomic - scale 3d contacts see ref .@xcite . a narrow - wide connection is eventually formed when an unsupported gold nanowire is done by experimental means@xcite .this can be modelled by a simple single junction connection in a 3d wire similar to that sketched in fig .[ fig8b ] .previous theoretical work on 2d nanowires have pointed out the existence of a vortex structure , for the quantum current density , as a result of a narrow - wide connection .interestingly , a change of pattern from laminar to `` turbulent '' flux is obtained for ballistic 2d free - electron calculations @xcite . in the analogous simplest case for 3d wires we observe a similar behaviour and transition , the onset of vortex structure formation is intimately linked to the income of the second transversal level in the wider part . in fig .[ fig8b ] and [ fig9 ] we can see the charge and the current density for this 3d narrow - wide structure .the width of the narrow wire allows just a single propagating channel while the wide part allows two channels , this is the minimum required to exhibit the vortex structure that is remarked in fig .[ fig9 ] . periodic charge and current density structures are formed along the final wider tube , this can be readily explained in terms of constructive and destructive interference patterns for the only two channels occupied in the wide lead .there are points with charge density zero regularly distributed at both sides of the charge density maximums in fig .[ fig8b ] , these points are vortex centers in the quantum current map of fig .note that , due to the axial symmetry , these vortex points define rings in real space , and the flux sourrounding them torus surfaces .when the wide lead gets much wider , many transversal levels come up and the charge density structure gets irregular as can be seen in fig .[ fig9b ] . the same explanation remains valid this time : the pattern results from the interference of many -in this case- longitudinal wavelenghts , the period of these patterns increases with the number of levels involved , eventually it gets seemingly non periodic as the case seen in fig . [ fig9b ] where no regular pattern is observed. these results could be important for the understanding -and therefore control- of coherent waves by tuning geometrical parameters .in particular the transition from a regular one dimensional lattice of maximums to a disordered pattern by changing the width of the outgoing waveguide could be amenable of experimental verification ( may be with electromagnetic waves rather than electrons ) .a single junction , as the one seen in previous section , can be regarded as a scattering center in general terms .realistic wires have atomic structure and therefore many possible scattering centers for the conduction electrons passing through .the inclusion of increasing number of scattering centers gives rise -as we will see- to a richer phenomenology . in this subsectionwe will focus in the case of just two scattering centers : a wide - narrow - wide ( wnw ) junction .two ( strong ) scattering centers suffice to predict negative differential resistance in point contact experiments .the ballistic conductance of a wnw structure , as the one in the inset of fig .[ fig5 ] , presents a dense resonant behaviour @xcite as a function of the narrow area and lenght as can be seen in fig .[ fig10 ] .being this a very simplistic model , it is a fair representation of what can be expected when we have these main ingredients : @xmath127 a couple of strong scattering centers at a given distance , @xmath128 , and @xmath129 transversal confinement ( characterized by the area @xmath19 ) . in these cases we can expect a `` wavy landscape '' for the conductance as a function of ( @xmath128 , @xmath19 ) , though may be not the same than the one in fig .[ fig10 ] .clearly , however , the variation of any of these relevant parameters can eventually give any derivative .this diversity of results is what -we will see- is possible for the case of monoatomic contacts . within this very simple wnw representation of atomic size contacts, we can easily locate the zone in fig .[ fig10 ] that reasonably belongs to monoatomic ( or very few atoms ) contacts for typical metals .this zone is what we have represented in fig .[ fig11 ] , on it we have labeled three different points belonging to 3 different possible particular geometries or ideal contacts .arrows indicate the evolution due to an eventual stretching process of these few - atoms contacts . before breaking , seems reasonable to think that the effective lenght will increase and effective area will slightly decrease @xcite , this is what the arrows indicate .now , as can be seen , as a function of the starting spot , the subsequent evolution of the conductance can be very different .if we are initially located in point 1 , the evolution will be mostly quantized ( values very near @xmath101 ) and essentially horizontal ; the probability of this event to happen is quite high due to the great number of points in the plane ( @xmath128 , @xmath19 ) belonging to this type .however , there is a good chance of falling in a place like the one labeled 3 , in this case -as the arrow indicates- we expect to observe a decreasing of @xmath130 before breaking the contact .finally , there is a little but non negligible chance for our starting point to be like the labeled 2 ; in this case , since we are located in the minimum of the `` valley '' , further stretching will make an increase in the conductance as indicated .the behaviour for case 2 is the more surprising since it predicts increasing of conductance while stretching the wire .should this model be valid , it could be expected to be found for given geometries and/or metals .in fact there is compelling experimental evidence of negative differential resistance for the case of aluminum@xcite , and theoretical explanation in terms of atomistic density functional calculations @xcite for the case of relatively wide cross - section wires .our present free - electron result provides a helpful simple view of good part of the relevant physics in these systems , a complementary piece of the whole picture , especially relevant for the few - atoms - contact case .long physical wires have their thinnest expression in what we can call monoatomic ( in width ) long wires .the existence of -relatively long- monoatomic gold wires has been experimentally reported @xcite and theoretically studied @xcite very recently . in a chain like this ,each atom is a system on its own that holds localized electronic states in addition to those extended states responsible for conduction .very simple modelling of this can be done by the repitition of a wnw structure .as we will see , this periodic repetition gives rise to a long chain of 3d `` resonance boxes '' . with the selection of proper dimensions for the geometrical parameters of these boxes, localized states can be contained inside .the correspondent energy levels of this states can be an open door for _ resonant tunneling _ phenomena in these 3d structures .this is proved here for a sufficiently long periodic wire . for a clear exhibition of resonant tunneling phenomena in 3d structureslet us start with the standard wnw problem as shown in fig .[ fig12 ] .the evaluation of @xmath130 versus incident energy yields the expected quantized steps with resonance oscilations overimposed .there is a threshold energy at @xmath131 marking the onset of the first propagating level in the narrow tube , which is seen as the first jump in conductance .now , if we mantain the same geometry but we add -in the central narrow tube- a wider box as shown in the inset , we can clearly appreciate the appearance of a resonant tunneling peak quite before this threshold energy .this is completely analogous to the obtained in 1d and 2d models .it can be clearly prooved that perfect transparence is achieved at the resonant tunneling energy .not only the value of transmission gets to 1 at a given e , but the periodic repetition of this structure does not filter down conductivity but simply increases the number of peaks . there is a new peak for each 3d quantum resonant box we add : a splitting effect between degenerate localized levels is proved as can be clearly seen in fig .[ fig13 ] . in this figure, a zoom of the resonant peak is seen for the cases of 1 , 2 and 28 resonant boxes .perfect transmission peaks are not the only consequence of interference , total reflection is also obtained as can be seen in fig .[ fig14 ] .a `` band like '' structure appears as a consequence of the periodicity imposed in the chain of 28 boxes .there are intervals of energy where transmission is nearly one and intervals where it is practically zero , this is in connection with conduction bands and energy gaps of simple one dimensional band theories @xcite .the complete inhibition of transmission is not obtained in pure 1d problems while for 2d and 3d problems total reflection is naturally achieved@xcite .we have developed a formulation of the scattering matrix method to solve general three dimensional waveguide problems .the truncation method , based on the way the mode coupling is done , provides quick convergence of results with one single parameter to be tuned : the number of evanescent modes in the narrowmost part of the system .the method has been already successfully used for the calculation in optical waveguides @xcite and electronic ballistic contacts@xcite .examples of convergence tests and calculations have been also shown .in particular , we have studied with some detail the problem of quantum conductance in metallic nanowires .various simple models of crescent number of junctions have been studied . for narrow - wide connections charge density and quantum current patternshave been shown , interestingly vortex rings are obtained for this simple geometry in analogy to previous 2d calculations . this vortex structure starts forming a regular pattern and becomes apparently disordered with increasing tubes width .the calculation for wide - narrow - wide connections gives rise to strong resonant behaviour of the conductance as a function of width and length of the contact .due to the very recent appearance of experiments in which impressively straight nanowires have been imaged , seems pertinent a closer look at this idealized geometry model . therefore , interpretation of wnw results , leads to the prediction of several behaviours of the conductance in the last plateaus of an experimental breaking proccess ; in particular negative differential resistance can be expected .further addition of junctions to form periodic wnw structures gives rise to resonant tunneling effects in what we can think of as a simple model for a long periodic nanowire .opossite to 1d models and analogous to 2d , not only perfect transmission is obtained for the resonant tunneling energy but perfect reflection appears for some finite intervals of energy .the authors are grateful to m. nieto - vesperinas , a. garca - martn , e. tosatti , j. j. kohanoff , m. okamoto , t. uda , k. takayanagi and k. terakura for enlightening discussions .one of the authors ( jat ) acknowledges funding support from : european union through tmr grant erbfmbict972563 and _ human capital and mobility program _ ( icarus ii project at cineca , bologna ) , and from japaneese nedo .jjs s work has been supported by the _ comunidad autnoma de madrid _ and the dgicyt through grants 07t/0024/1998 and no .pb98 - 0464 .computational resources have been availed from _ dep .de fsica de la materia condensada _ at uam , _ abdus salam international center for theoretical physics _ ( ictp ) in trieste , and jrcat in tsukuba , we are deeply thankful to all of them .n. agrat , j.g .rodrigo and s. vieira , phys .b * 47 * , 12345 ( 1993 ) ; z. gai_ , _ ibid ._ * 58 * , 2185 ( 1998 ) ; b. ludoph and j.m .van ruitenbeek , _ ibid . _* 61 * , 2273 ( 2000 ) ; c. sirvent _ et al ._ , _ ibid ._ * 53 * 16086 ( 1996 ) ; h. yasuda and a. sakai , _ ibid . _* 56 * , 1069 ( 1997 ) ; h. van den brom and j.m .van ruitenbeek , phys .. lett . * 82 * , 1526 ( 1999 ) ; j. l. costa - krmer , n. garca and h. olin , _ ibid . _ * 78 * , 4990 ( 1997 ) ; r.j.p .keijsers , o. i. shklyarevskii and h. van kempen , _ ibid ._ * 77 * , 3411 ( 1996 ) ; g. rubio , n. agrat and s. vieira , _ ibid . _ * 76 * , 2302 ( 1996 ) ; b. ludoph __ , _ ibid ._ * 82 * 1530 ( 1999 ) ; l. olesen __ , _ ibid . _* 72 * , 2251 ( 1994 ) ; j.i ._ , _ ibid ._ * 71 * , 1852 ( 1993 ) ; science * 267 * , 1793 ( 1993 ) ; a. yazdani , d.m . eigler and n.d .lang , _ ibid . _* 272 * , 1921 ( 1996 ) ; c. zhou _ et al .. lett . * 67 * , 1160 ( 1995 ) .k. hansen et .al , _ ibid . _ * 77 * , 708 ( 2000 ) ; rev .inst . * 71 * , 1793 ( 2000 ) ; j.l .costa - krmer , n. garca , p. garca - mochales and p. a. serena , surf .sci . * 342 * , l1144 ( 1995 ) ; y. kawahito , h. kasai , h. nakanishi and a. okiji , j. appl .phys . * 85 * , 947 ( 1999 ) ; j.m .krans et al . , nature ( london ) * 375 * , 767 ( 1995 ) ; r.n .barnett and u. landman , nature ( london ) * 387 * , 788 ( 1997 ) ; e.n .bogachek , a.g .scherbakov and u. landman , phys .b * 56 * , 1065 ( 1997 ) ; m. brandbyge _et al . _ , _ ibid . _ * 52 * , 8499 ( 1995 ) ; * 55 * , 2637 ( 1997 ) ; * 56 * , 14956 ( 1997 ) ; * 57 * , r15088 ( 1998 ) ; a.m. bratkovsky , a.p .sutton and t.n .todorov , _ ibid ._ * 52 * , 5036 ( 1995 ) ; h. j. choi and jisoon ihm , _ ibid . _* 59 * , 2267 ( 1999 ) ; a. levy yeyati , a. martn - rodero and f. flores , _ ibid . _ * 56 * , 10369 ( 1997 ) ; h. mehrez and s. ciraci , _ ibid ._ * 56 * 12632 , ( 1997 ) ; h. mehrez , s. ciraci , a. buldum and i.p .batra , _ ibid ._ * 55 * r1981 , ( 1997 ) ; j.l .et al . _ , _ ibid . _* 56 * , r4351 ( 1997 ) ; f. kassubek , c. a. stafford and h. grabert , _ ibid . _ * 59 * , 7560 ( 1999 ) ; j.m .van ruitenbeek , m. h. devoret , d. esteve and c. urbina , _ ibid . _* 56 * , 12566 ( 1997 ) ; e. tekman and s. ciraci , _ ibid ._ * 39 * , 8772 ( 1989 ) ; * 43 * , 7145 ( 1991 ) ; n.d .lang , _ ibid . _* 36 * , 8173 ( 1987 ) ; * 52 * , 5335 ( 1995 ) ; * 55 * , 4113 ( 1997 ) ; phys .lett . * 79 * , 1357 ( 1997 ) ; j.c .cuevas , a. levy yeyati and a. martn - rodero , _ ibid . _* 80 * , 1066 ( 1998 ) ; j. c. cuevas _ et .al _ , _ ibid . _ * 81 * , 2990 ( 1998 ) ; a. nakamura __ , _ ibid ._ * 82 * , 1538 ( 1999 ) ; t.n .todorov and a.p .sutton , _ ibid . _ * 70 * , 2138 ( 1993 ) ; u. landman __ , _ ibid ._ * 77 * , 1362 ( 1996 ) ; science * 248 * , 454 ( 1990 ) ; d.p.e .smith , _ ibid . _ * 269 * , 371 ( 1995 ) ; n. kobayashi , m. brandbyge and m. tsukada , jpn .. phys . * 38 * , 336 ( 1999 ) ; c.c . wan _. lett . * 71 * , 419 ( 1997 ) ; c. yannouleas and u. landman , j. phys .b * 101 * , 5780 ( 1997 ) ; j.a .et al . _ ,sci . * 426 * , l441 ( 1999 ) ; d. snchez - portal _ et al ._ , phys rev. lett . * 80 * , 3775 ( 1998 ) ; m. okamoto and k. takayanagi , phys . rev .b * 60 * , 7808 ( 1999 ) ; h. hakkinen , r.n .barnett and u. landman , j. phys .b * 103 * , 8814 ( 1999 ) .torres and j.j .senz in _ the ultimate limits of fabrication and measurement _ , edited by j.k .gimzewski and m. e. welland , nato asi series e : applied sciences , * 292 * , 129 ( kluwer , dordrecht 1995 ) .

<S> an easy to implement and powerful method for the solution of 3d scattering problems that can be well described by helmholtz equation is presented . </S><S> the matrix algebra used provides excellent stability versus the number of junctions as well as great computational speed . </S><S> the matrix truncation method yields an easy single - parameter convergence procedure . </S><S> subsequently , some aspects of the electronic transport through metal nanowires are studied by the use of landauer s scattering approach to the conductance . </S><S> we predict the existence of current vortex - rings patterns due to sharp enough narrow - wide connections in atomic size point contacts . </S><S> longitudinal resonances between scattering centers provide a simple physical picture for the understanding of negative differential resistance in ideal monoatomic contacts . </S><S> relatively long nanowires with high geometrical perfection -like those recently observed by transmission electron microscopy- are modelled exhibiting resonant tunnelling and total reflection at given incident energy intervals . </S><S> + 2 future daily - life technology will be greatly based on miniaturization of both electronic components and optical devices . </S><S> state - of - the - art research has already prototyped tools using simultaneously both electrons and photons confinement . </S><S> for instance , in artificially created electromagnetic micro - cavities interaction light - matter can best be described by new concepts like those of `` cavity polaritons''@xcite , devices in which the light emission process can be extra - ordinarily fast . </S><S> also , quantum well lasers utilizing two - dimensional electron gas and optical modulators with quantum confinement effect are now practically used . </S><S> the ultimate integration of any opto - electronic device has to overcome the theoretical and practical problem of the waveguiding at tiny sizes for both electrons and photons@xcite . </S><S> consequently , it is not surprising the huge research effort devoted , in the last decade , to these waveguides themselves : _ metallic wires at the nanometer scale _ and _ optical waveguides at the micrometer scale_. a common characteristic lays on the basis of these waveguide systems : elastic scattering processes are dominant for transport . </S><S> metallic atomic - size wires and micro - optical guides present lateral dimensions of the order of the wavelength associated to the particles passing through . </S><S> helmholtz equation governs elastic flux of waves through a general dispersive system ( note here that the independent free - electron stationary schrdinger equation is equivalent ) . </S><S> therefore , the ability to calculate solutions for this equation is of great importance . </S><S> as mentioned above , the experimental and theoretical study of metallic nanowires , namely their electrical and structural properties , has been focus of intense research in the last decade ( see e.g articles in ref . </S><S> @xcite as well as the various experimental works of ref . </S><S> @xcite , and the theoretical works in @xcite ) . </S><S> however , only very recently , the combination of scanning tunneling microscope ( stm ) with transmission electron microscopy ( tem ) has achieved atomic - resolution images of metallic nanowires@xcite . in particular , ultrathin gold wires have been observed exhibiting a striking geometrical regularity in some cases . here </S><S> we intend to look at some possible effects due to the ideal geometry of these nanowires . due to the simple -therefore general- phenomenology studied , </S><S> the same physics is expected to be found by more sophisticated approaches . </S><S> the paper is organized as follows : first we present a generalized scattering matrix method for the solution of helmholtz ( or schrdinger ) equation in section [ secsm ] . though the formalism developed can be applied to general scalar wave propagation problems and even scattering of light in some cases , we use it later to address the problem of conduction through atomic - size contacts : in section [ secwires ] we focus on the study of electronic conduction through metallic nanowires with several models . </S><S> finally , in section [ secconcl ] we summarize our results . </S>

1106.3744

in the canonical formulation of general relativity , the hamiltonian is a sum of constraints . in particular , within the asthekar framework @xcite , the hamiltonian is a sum of three constraints : @xmath0 = \frac{1}{2\kappa } \int_{\sigma}d^3x \left ( n^i c_i+n^a c_a+n c\right ) \approx 0 , \nonumber\ ] ] where @xmath1 , @xmath2 are lagrange multipliers , @xmath3 is called the gauss constraint , @xmath4 is the diffeomorphism constraint , and @xmath5 is the hamiltonian constraint .the sign `` @xmath6 '' means equality on the surface of constraints ( _ i.e. _ weak equality ) .one can also define the corresponding smeared constraints as follows : @xmath7 = \frac{1}{2\kappa } \int_{\sigma}d^3x\ n^i c_i , \\\mathcal{c}_2 & = & d[n^a ] = \frac{1}{2\kappa } \int_{\sigma}d^3x\ n^a c_a , \\\mathcal{c}_3 & = & s[n ] = \frac{1}{2\kappa } \int_{\sigma}d^3x\ n c,\end{aligned}\ ] ] that is such that @xmath8 = g[n^i ] + d[n^a]+s[n]$ ] .the hamiltonian is a total constraint which is vanishing for all multiplier functions @xmath2 . because @xmath9 \approx 0 $ ] at all times , the time derivative of the hamiltonian constraint is also weakly vanishing, @xmath10 \approx 0 $ ] . the hamilton equation @xmath11\}$ ] therefore leads to @xmath12 , h_{\rm g } [ m^i , m^a , m]\right\ } \approx 0 , \label{hh}\ ] ] which , when explicitly written , means : @xmath13 + d[n^a]+s[n ] , g[m^i ] + d[m^a]+s[m ] \right\ } \approx 0 .\nonumber\ ] ] due to the linearity of the poisson bracket , one can straightforwardly find that the condition ( [ hh ] ) is fulfilled if the smeared constraints belong to a first class algebra @xmath14 in ( [ algebra ] ) , the @xmath15 are structure functions which , in general , depend on the phase space ( ashtekar ) variables @xmath16 .the algebra of constraints is fulfilled at the classical level due to general covariance . to prevent the system from escaping the surface of constraints , leading to an unphysical behavior, the algebra must also be closed at the quantum level . in addition , it was pointed out in @xcite that the algebra of quantum constraints should be strongly closed ( _ off shell _ closure ) .this means that the relation ( [ algebra ] ) should hold in the whole kinematical phase space , and not only on the surface of constraints ( _ on shell _ closure ) .this should remain true after promoting the constraints to quantum operators .loop quantum gravity ( lqg ) @xcite is a promising approach to quantize gravity , based on a canonical formalism parametrized by ashtekar variables .the methods of lqg applied to cosmological models are known as loop quantum cosmology ( lqc ) @xcite . in lqc , quantum gravity effects are introduced by holonomies of ashtekar connection .this replacement is necessary because connection operators do not exist in lqg .rewriting classical constraints in terms of holonomies leads to two types of quantum corrections : the so - called inverse - volume and holonomy corrections .because the constraints are quantum - modified , the corresponding poisson algebra might not be closed : @xmath17 here , @xmath18 stands for the anomaly term which can appear due to the quantum modifications . for consistency ( closure of algebra ) , @xmath18 is required to vanish .the condition @xmath19 implies some restrictions on the form of the quantum corrections . in this paper, we will study this requirement to find a consistent way for introducing quantum holonomy corrections to the vector perturbations .the question of the construction of an anomaly - free algebra of constraints is especially interesting to address in inhomogeneous lqc .perturbations around the cosmological background are indeed responsible for structure formation in the universe .this gives a chance to link quantum gravity effects with astronomical observations . in the particular case of the flat flrw background, the ashtekar variables can be decomposed as follows @xmath20 where @xmath21 and @xmath22 parametrize the background phase space , and @xmath23 is the so - called barbero - immirzi parameter .the issue of anomaly freedom for the algebra of cosmological perturbations was extensively studied for inverse - volume corrections .it was shown that this requirement can be fulfilled for first order perturbations .this was derived for scalar @xcite , vector @xcite and tensor perturbations @xcite .it is worth mentioning that , for the tensor perturbations , the anomaly - freedom is automatically satisfied .based on the anomaly - free scalar perturbations , predictions for the power spectrum of cosmological perturbations were also performed @xcite .this gave a chance to put constraints on some parameters of the model using observations of the cosmic microwave background radiation ( cmb ) @xcite .the aim of this article is to address the issue of anomaly freedom for the holonomy - corrected vector perturbations in lqc .it was shown in @xcite that these perturbations can be anomaly free up to the fourth order in the canonical variable @xmath21 .this , however , is not sufficient to perform the analysis of the propagation of vector modes through the cosmic bounce .vector perturbations with _ higher order holonomy corrections _ were also recently studied @xcite .it was shown there that , in this case , an anomaly - free formulation can be found for the gravitational sector . in this paper , we apply a different method , which is based on the introduction of counter - terms in the hamiltonian constraint .we show that the anomaly freedom conditions for vector modes with holonomy corrections can be fulfilled in this way .the method is similar to the one used by bojowald __ in the case of cosmological perturbations with inverse - volume corrections .as we will see , the counter - terms do not introduce any higher - order holonomy corrections .this way of fulfilling the anomaly freedom is therefore different from what was done in @xcite , where higher order terms are involved .moreover , in ref .@xcite , the issue of anomaly freedom was studied for the gravity sector only and the formulation suffers from ambiguities . in our study ,scalar matter is introduced .the presence of this matter term fixes the ambiguity associated with the holonomy correction .it should be underlined that without a vector matter source , one can not rigorously prove the anomaly cancellation .however , as will be shown in the next sections , our approach is meaningful as the equations derived are , as in @xcite , compatible with vector matter assuming @xmath24 but @xmath25 .holonomy corrections arise while regularizing classical constraints , when expressing the ashtekar connection in terms of holonomies . in particular , the regularization of the curvature of the ashtekar connection @xmath26 leads to the factor @xmath27 , which simplifies to @xmath28 in the classical limit @xmath29 .however , the ashtekar connection does not appear only because of @xmath26 : in the classical perturbed constraints , terms linear in @xmath21 are also involved . in principle , such terms should be holonomy - corrected .however , there is no direct expression for them , analogous to the regularization of the @xmath26 factor .nevertheless , one can naturally expect that @xmath21 factors are corrected by the replacement @xmath30 where @xmath31 is some unknown integer .it should be an integer because , when quantizing the theory , the @xmath32 factor is promoted to be the shift operator acting on the lattice states .if @xmath31 was not an integer , the action of the operator corresponding to @xmath33 would be defined in a different basis .another issue is related with the choice of @xmath34 , which corresponds to the so - called _lattice refinement_. models with a power - law parametrization @xmath35 were discussed in details in the literature . while , in general , @xmath36 $ ] , it was pointed out that the choice @xmath37 is favored @xcite .this particular choice is called the @xmath38scheme ( new quantization scheme ) .studies in this article are performed for the general power - law case @xmath35 . for the sake of simplicity, we use the notation @xmath39 : = \left\ { \begin{tabular}{ccc } $ \frac{\sin(n\bar{\mu } \gamma \bar{k})}{n\bar{\mu}\gamma}$ & for & $ n \in \mathbb{z}/\{0\}$ , \\ & & \\ $ \bar{k}$ & for & $ n=0 $ , \end{tabular } \right.\ ] ] for the holonomy correction function . the introduction of holonomy corrections is therefore performed by replacing @xmath40 $ ] .however , factors @xmath28 are simply replaced by @xmath41 ^ 2 $ ] , because they arise from the curvature of the ashtehar connection .for the linear terms , the integers are parameters to be fixed .vector modes within the canonical formulation were studied in @xcite .it was shown there that @xmath42 where @xmath43 and the divergence - free condition @xmath44 is fulfilled .the values of @xmath45 and @xmath46 depend on the gauge choice .however , due to the gauss constraint , only symmetric variables are invariant under internal rotations .this is the case for @xmath47 , which is consequently independent on the specific choice of @xmath45 and @xmath46 , and should be preferred .the perturbation of the shift vector is parametrized as @xmath48 .we consider the quantum holonomy - corrected hamiltonian constraint given by @xmath49 = \frac{1}{2 \kappa } \int_{\sigma } d^3x \left [ \bar{n}(c^{(0)}+c^{(2)})\right ] , \label{scalar}\ ] ] where @xmath50 \right)^2 , \\c^{(2 ) } & = & -\frac{1}{2\bar{p}^{3/2}}\left ( \mathbb{k}[1 ] \right)^2(1+\alpha_1 ) ( \delta e^c_j \delta e^d_k \delta^k_c \delta^j_d ) + \sqrt{\bar{p}}(\delta k^j_c \delta k^k_d \delta^c_k \delta^d_j ) \nonumber \\ & -&\frac{2}{\sqrt{\bar{p}}}\left(\mathbb{k}[v_1]\right)(1+\alpha_2)(\delta e^c_j \delta k^j_c ) . \end{aligned}\ ] ] holonomy corrections were introduced by replacing @xmath40 $ ] .two counter - term functions @xmath51 and @xmath52 , whose interest will be made clear later , were also added . in the classical limit @xmath53\rightarrow \bar{k}$ ] , and @xmath54 , with @xmath55 .we have assumed here that @xmath56 are functions of the background variables only and that @xmath57 is an integer to be fixed .the hamiltonian constraint ( [ scalar ] ) corresponds to the one investigated in @xcite while setting @xmath58 .however , as we will show , it is necessary to introduce these additional factors , which vanish in the classical limit .these factors can , of course , also be viewed as contributions from the two counter - terms @xmath59 \right)^2 ( \delta e^c_j \delta e^d_k \delta^k_c \delta^j_d ) , \\s_{c2 } & = & -\frac{\alpha_2}{2 \kappa } \int_{\sigma } d^3x \frac{2 \bar{n}}{\sqrt{\bar{p}}}\left(\mathbb{k}[v_1]\right ) ( \delta e^c_j \delta k^j_c ) \end{aligned}\ ] ] to the holonomy - corrected hamiltonian constraint . a similar method of counter - terms was successfully applied for perturbations with inverse - volume corrections . in that case , it was possible to fix the counter - terms so as to make the algebra anomaly free . in this article ,we follow the same path so as to find explicit expressions for @xmath51 and @xmath52 . for the sake of completeness, we also introduce holonomy corrections to the diffeomorphism constraint , as follows : @xmath60 = \frac{1}{\kappa } \int_{\sigma } d^3x \delta n^c \left [ -\bar{p } ( \partial_k \delta k^k_c ) - \left ( \mathbb{k}[v_2]\right)\delta^k_c ( \partial_d \delta e^d_k ) \right ] , \label{diff}\end{aligned}\ ] ] where @xmath61 is an unknown integer .it is worth emphasizing here that within lqg , the diffeomorphism constraint is fulfilled at the classical level while constructing the diffeomorphism invariant spin network states .if lqc was really derived from the full lqg theory , the classical form of the diffeomorphism constraint should therefore be used .however , at this early stage of the understanding of lqc , it might be safe to allow for some generalizations by introducing the holonomy correction also to the diffeomorphism constraint .this hypothesis was already studied in @xcite in the case of holonomy - corrected scalar perturbations .it was assumed there that the holonomy correction function was given by @xmath62 $ ] . in this work, we prefer to keep a more general expression @xmath63 $ ] with a free @xmath61 parameter. we will investigate whether this additional modification can help to fulfill the anomaly freedom conditions . in order to investigate the algebra of constraints ,the poisson bracket has to be defined .we start with the gravity sector for which the poisson bracket can be decomposed as follows : @xmath64 the algebra of constraints ( [ scalar ] ) and ( [ diff ] ) shall now be investigated . using the poisson bracket ( [ poisson ] ), we find : @xmath65 , s^q[n_1 ] \right\ } & = & 0 ,\\ \left\{d^q[n^a_1 ] , d^q[n^a_2 ] \right\ } & = & 0 , \\ \left\{s^q[n ] , d^q[n^a]\right\ } & = & \frac{\bar{n}}{\sqrt{\bar{p } } } \mathcal{b } d^q[n^a ] \nonumber \\ & + & \frac{\bar{n}}{\kappa\sqrt{\bar{p } } } \int_{\sigma } d^3x\delta n^c \delta^k_c ( \partial_d \delta e^d_k ) \delta e^d_k \mathcal{a } , \end{aligned}\ ] ] where @xmath66 + \mathbb{k}[v_2]- 2 \mathbb{k}[2]$ ] , and @xmath67 is the anomaly function which , for reasons that shall be made clear later , is decomposed in two parts @xmath68 , where @xmath69 , \\ \mathcal{a}_2 & = & 2\mathbb{k}[2 ] \bar{p } \frac{\partial \mathbb{k}[v_2]}{\partial \bar{p } } - \frac{1}{2 } ( \mathbb{k}[1])^2 \cos(v_2 \bar{\mu } \gamma \bar{k } ) - 2\mathbb{k}[1 ] \bar{p } \frac{\partial \mathbb{k}[1]}{\partial \bar{p}}\cos(v_2 \bar{\mu } \gamma \bar{k } ) \nonumber \\ & + & ( 1+\alpha_2)\mathbb{k}[v_1 ] \mathbb{k}[v_2 ] -\frac{1}{2 } \mathbb{k}[1]^2(1+\alpha_1 ) .\end{aligned}\ ] ] this decomposition was made such that , in the classical limit ( @xmath29 ) , both contributions to the anomaly vanish separately . using the relation @xmath70}{\partial \bar{p } } = \left ( \bar{k } \cos(n\bar{\mu } \gamma \bar{k } ) -\mathbb{k}[n]\right)\beta,\ ] ]the second contribution can be re - written as : @xmath71 \mathbb{k}[v_2 ] + ( 1+\alpha_2)\mathbb{k}[v_1]\mathbb{k}[v_2 ] + ( 2\beta -1/2 ) ( \mathbb{k}[1])^2\cos(v_2 \bar{\mu } \gamma \bar{k } ) \nonumber \\ & - & \frac{1}{2}(\mathbb{k}[1])^2(1+\alpha_1 ) .\label{a2rewritten}\end{aligned}\ ] ] the full anomaly term is given by : @xmath72\mathbb{k}[v_2]-\frac{1}{2}(\mathbb{k}[1])^2(1+\alpha_1 ) -2(1+\beta ) \mathbb{k}[2 ] \mathbb{k}[v_2]+\mathbb{k}[v_2]^2 \nonumber \\ & + & ( 2\beta -1/2 ) ( \mathbb{k}[1])^2\cos(v_2 \bar{\mu } \gamma \bar{k } ) .\label{totalanomaly}\end{aligned}\ ] ]the requirement of the anomaly freedom for the gravity sector reads as @xmath73 . under this condition ,the algebra of constraints becomes closed but deformed , in particular : @xmath74 , d^q[n^a]\right\ } = d^q\left[\frac{\bar{n}}{\sqrt{\bar{p } } } \mathcal{b } n^a \right].\ ] ] the structure of space - time is therefore also modified .this is illustrate in fig .[ fig ] where one can notice that the hamiltonian and diffeomorphism constraints generate gauge transformations in directions respectively normal and parallel to the hypersurface . in the classical limit , @xmath75 andboth the transformations commute at the perturbative level .let us start by analyzing the condition @xmath73 without any counter - term ( _ i.e. _ with @xmath76 ) .this case corresponds to the one studied in @xcite generalized by the contribution from the corrected diffeomorphism constraint .it was shown in that work that , if @xmath77 , the anomaly - freedom condition can be satisfied up to the @xmath78 order only . here , we investigate whether this might be improved by the additional correction made to the diffeomorphism constraint . by setting @xmath76 , the anomaly term given by ( [ totalanomaly ] ) can be expanded in powers of the canonical variable @xmath21 as follows : @xmath79 where we have defined @xmath80 and @xmath81 $ ] .clearly , in the classical limit @xmath82 , the anomaly tends to zero . requiring the anomaly cancellation up to the fourth order leads to the condition :@xmath83 it can be shown that the condition of anomaly cancellation up to orders higher than four can not be met . for @xmath84 ( @xmath38scheme ) , the above equation simplifies to the quadratic diophantine equation : @xmath85 this equation can be reduced to a pell - type equation and solved for an infinite number of pairs of integers @xmath86 .the first three solutions are @xmath87 and @xmath88 .the first one @xmath89 corresponds to the case studied in @xcite , where the diffeomorphism constraint was kept at its classical form .the value @xmath90 obtained in this case was also used to fix the ambiguity for the holonomy - corrected tensor perturbations @xcite .if the holonomy modified diffeomorphism constraint is used , the ambiguity can not be fixed anymore due to the infinite number of solutions to eq .( [ diophantine ] ) .as we have shown , the modification of the diffeomorphism constraint does not help satisfying the anomaly freedom conditions in the absence of counter - terms . in this case, the anomaly freedom can be fulfilled up to the fourth order in @xmath91 . in the semi - classical limit @xmath92 ,the anomaly cancellation up to the fourth order might be a good approximation .however , when approaching the bounce , where @xmath93 , contributions from higher order terms become significant and the effects of the anomaly can not be neglected anymore .studies of vector perturbations during the bounce phase can not be performed in such a setup . in order to study vector perturbations through the bounce , the anomaly cancellation at all ordersis required .this probably makes mandatory the introduction of counter - terms .let us consider the general case with non - vanishing counter - terms . in this case, the requirement @xmath94 can be translated into a relation between the two counter - terms @xmath51 and @xmath52 : @xmath95\mathbb{k}[v_2]}{\mathbb{k}[1]^2 } -4(1+\beta ) \frac{\mathbb{k}[2 ] \mathbb{k}[v_2]}{\mathbb{k}[1]^2}+ 2\frac{\mathbb{k}[v_2]^2}{\mathbb{k}[1]^2 } \nonumber \\ & + & ( 4\beta -1)\cos(v_2 \bar{\mu } \gamma \bar{k}).\end{aligned}\ ] ] with this choice for the @xmath51 function , the anomaly is removed .however a significant ambiguity remains .namely , the function @xmath52 together with parameters @xmath57 and @xmath61 remain undetermined .a particularly interesting case corresponds to the choice @xmath96 .this determines @xmath51 .of course , this also works the other way round : one can set @xmath97 and derive the correct expression for @xmath52 .therefore , two special cases , heuristically motivated , where one of the counter - terms is vanishing , are worth studying : @xmath98\mathbb{k}[v_2]}{\mathbb{k}[1]^2 } -4(1+\beta ) \frac{\mathbb{k}[2 ] \mathbb{k}[v_2]}{\mathbb{k}[1]^2}+ 2\frac{\mathbb{k}[v_2]^2}{\mathbb{k}[1]^2 } \nonumber \\ & + & ( 4\beta -1)\cos(v_2 \bar{\mu } \gamma \bar{k } ) , \\ \alpha_2 & = & 0,\end{aligned}\ ] ] and @xmath99)^2}{\mathbb{k}[v_1]\mathbb{k}[v_2 ] } + ( 1+\beta ) \frac{\mathbb{k}[2]}{\mathbb{k}[v_1]}-\frac{1}{2 } \frac{\mathbb{k}[v_2]}{\mathbb{k}[v_1 ] } \nonumber \\ & -&(\beta -1/4 ) \frac{(\mathbb{k}[1])^2\cos(v_2 \bar{\mu } \gamma \bar{k})}{\mathbb{k}[v_1]\mathbb{k}[v_2]}.\end{aligned}\ ] ] to conclude , at least one counter - term is necessary to fulfill the anomaly freedom conditions for the gravity sector .another possible way to fix the ambiguity in the choice of the @xmath51 and @xmath52 functions could be to set @xmath100 . with this restriction ,the anomaly cancellation is fulfilled by imposing @xmath101 as @xmath102 .as mentioned earlier , both @xmath103 and @xmath104 separately tend to zero in the classical limit , making this decomposition meaningful . in this case , the poisson bracket between the hamiltonian and diffeomorphism constraints is just @xmath105 , d^q[n^a]\right\}=0 $ ] . the conditions @xmath100 and @xmath101 can be translated into expressions for the @xmath51 and @xmath52 functions : @xmath106\mathbb{k}[v_2]}{\mathbb{k}[1]^2 } -2\frac{\mathbb{k}[v_2]^2}{\mathbb{k}[1]^2 } + ( 4\beta-1)\cos(v_2 \bar{\mu } \gamma \bar{k } ) , \label{alpha1b0 } \\\alpha_2 & = & -1+\frac{2 \mathbb{k}[2]-\mathbb{k}[v_2]}{\mathbb{k}[v_1]}. \label{alpha2b0}\end{aligned}\ ] ] the expressions for @xmath51 and @xmath52 are parametrized by the integers @xmath57 and @xmath61 only .however , the dependence upon @xmath57 vanishes when @xmath52 is used in the hamiltonian constraint .the derived expressions for @xmath51 and @xmath52 do contain @xmath53 $ ] functions in the denominators . in principle , @xmath51 and @xmath52 could therefore diverge for some values of @xmath21 .however , in the counter - terms @xmath107 and @xmath108 , @xmath51 is multiplied by @xmath41 ^ 2 $ ] and @xmath52 by @xmath109 $ ] .the subsequent cancellation prevents any physical divergence from occurring .we have shown that the gravity sector of the vector perturbations with holonomy corrections can be made anomaly free .we will now extend this result by introducing scalar matter .the matter hamiltonian does not depend on the ashtekar connection and is therefore not subject to holonomy corrections .furthermore , for vector perturbations , @xmath110 .the matter hamiltonian is perturbed up to the second order as follows : @xmath111 = \bar{h}_{\rm m } + \delta h_{\rm m } = \int_{\sigma } d^3x \bar{n } ( c_{\rm m}^{(0)}+c_{\rm m}^{(2)}),\ ] ] where @xmath112.\ ] ] the value of @xmath113 is given by @xmath114 where we have used the condition @xmath115 .the matter diffeomorphism constraint is given by : @xmath116 = \int_{\sigma } d^3x \delta n^a \bar{\pi}(\partial_a \delta \varphi).\ ] ] the total hamiltonian and diffeomorphism constraints are @xmath117 & = & s^q[n]+h_{\rm{m}}[n ] ,\\ d_{\rm{tot}}[n^a ] & = & d^q[n^a ] + d_{\rm{m}}[n^a ] .\end{aligned}\ ] ] the resulting poisson brackets are the following : @xmath118 , s_{\rm{tot}}[n_1 ] \right\ } = 0 , \\ \left\{d_{\rm{tot}}[n^a_1 ] , d_{\rm{tot}}[n^a_2 ] \right\ } = 0 , \\ \left\{s_{\rm{tot}}[n ] , d_{\rm{tot}}[n^a]\right\ } = \frac{\bar{n}}{\sqrt{\bar{p } } } \mathcal{b } d^q[n^a ] + \frac{\bar{n}}{\kappa\sqrt{\bar{p } } } \int_{\sigma } d^3x\delta n^c \delta^k_c ( \partial_d \delta e^d_k ) \delta e^d_k \mathcal{a } \nonumber \\ + [ \cos(v_2 \bar{\mu } \gamma \bar{k } ) -1 ] \frac{\sqrt{\bar{p}}}{2}\left(\frac{\bar{\pi}^2}{2\bar{p}^3}-v(\bar{\varphi})\right ) \int_{\sigma } d^3x \bar{n } \partial_c ( \delta n^a ) \delta^j_a\delta e^c_j \nonumber\\ + \frac{\bar{\pi}}{\bar{p}^{3/2 } } \int_{\sigma}d^3x \bar{n } ( \partial_a \delta n^a ) \delta\pi -\bar{p}^{3/2}v_{\varphi}(\bar{\varphi } ) \int_{\sigma}d^3x \bar{n } ( \partial_a \delta n^a ) \delta \varphi.\end{aligned}\ ] ] anomaly freedom requires @xmath100 , @xmath94 , @xmath77 ( classical diffeomorphism constraint ) , and also @xmath119 .the latter conditions @xmath119 are due to the fact that metric scalar perturbations are not considered .consistently , scalar field perturbations are vanishing too .in fact , one could set @xmath119 from the very beginning but , without assuming this , it can be shown that the condition @xmath119 in fact resulting from the anomaly freedom .the associated counter - terms are given by ( [ alpha1b0 ] ) and ( [ alpha2b0 ] ) with @xmath77 .two non - vanishing counter - terms are required in contrast to the gravity sector , where only one counter - term was sufficient to fulfill the anomaly freedom conditions .the integer @xmath57 remains undetermined but the dependence upon this parameter cancels out in the hamiltonian constraint .namely , applying the counter - terms ( [ alpha1b0 ] ) and ( [ alpha2b0 ] ) with @xmath77 , we find that the anomaly free hamiltonian constraint is given by : @xmath120 = \frac{1}{2 \kappa } \int_{\sigma } d^3x \left [ \bar{n}(c^{(0)}_{\rm{free } } + c^{(2)}_{\rm{free } } ) \right ] , \label{scalarfree}\ ] ] where @xmath121 \right)^2 , \\ c^{(2)}_{\rm{free } } & = & -\frac{1}{2\bar{p}^{3/2 } } \left [ 4(1-\beta)\mathbb{k}[2]\bar{k } -2\bar{k}^2 + ( 4\beta-1)\mathbb{k}[1]^2 \right ] ( \delta e^c_j \delta e^d_k \delta^k_c \delta^j_d ) \nonumber \\ & + & \sqrt{\bar{p}}(\delta k^j_c \delta k^k_d \delta^c_k \delta^d_j ) -\frac{2}{\sqrt{\bar{p}}}\left(2\mathbb{k}[2 ] -\bar{k}\right)(\delta e^c_j \delta k^j_c ) . \label{anomalyfreec2}\end{aligned}\ ] ] the gravitational diffeomorphism constraint holds its classical form(@xmath77 ) .this is in agreement with lqg expectations .interestingly , this can also be obtained here as a result of anomaly freedom .the obtained anomaly - free hamiltonian ( [ scalarfree ] ) is determined up to the choice of the @xmath34 functions .there are no other remaining ambiguities .the @xmath34 function appears in definition of the @xmath53 $ ] function .because of this , there is also explicit appearance of the factor @xmath122 in equation ( [ anomalyfreec2 ] ) .the choice @xmath123 is preferred by various considerations @xcite .recently , this value was shown to be required also by the conditions on the anomaly - free scalar perturbations with holonomy corrections @xcite . for this choice of the @xmath122 parameter ,the remaining freedom is a parameter of proportionality in relation @xmath124 .this parameter can be written as @xmath125 , so @xmath126 .the parameter @xmath127 has interpretation of physical area , around which the elementary holonomy is defined .it is expected that @xmath128 , where @xmath129 is the planck length . however , determination of the accurate value of @xmath127 is a subject to empirical verifications . it is worth noticing about the hamiltonian constraint ( [ scalarfree ] ) that the effective holonomy corrections , due to the counter - terms , are no longer _ almost periodic functions _ ,defined as follows @xcite @xmath130 in this expression , @xmath31 runs over a finite number of integers and @xmath131 .this does not lead to any problem at the classical level .however , difficulties may appear when going to the quantum theory on lattice states .this is because the quantum operator corresponding to @xmath21 does not exist in contrast to the @xmath53 $ ] functions , which are almost periodic functions .this problem does not exist if the gravitational sector , without any matter content , is considered alone .however , the diffeomorphism constraint then has to be holonomy corrected , as studied previously . in such a case ,the background terms in the anomaly - free gravitational hamiltonian are almost periodic functions .the loop quantization can therefore be directly performed .the coordinate transformation @xmath132 generates a tensor gauge transformation . in the case of vector modes, the coordinate transformation is parametrized by the shift vector @xmath133 , where @xmath134 , therefore , the resulting gauge transformation is generated by the diffeomorphism constraint @xmath135 \}$ ] .the corresponding transformations for the canonical variables are : @xmath136 \}=- \bar{p } \partial_i \xi^a \label{xie } , \\\delta_{\xi } ( \delta k^i_a ) & = & \{\delta k^i_a , d^q[\xi^a ] \}= \mathbb{k}[v_2 ] \partial_a \xi^i \label{xik}.\end{aligned}\ ] ] based on the equation of motion @xmath137 , and the definition ( [ deltae ] ) , one finds the expression of @xmath138. the dot means differentiation with respect to the conformal time since we have chosen @xmath139 . using equations ( [ xie ] ) and ( [ xik ] ) one finds : @xmath140 -\mathbb{k}[v_1](1+\alpha_2)-\mathbb{k}[v_2 ] ) \xi^a.\end{aligned}\ ] ] based on this, one can define a gauge invariant variable @xmath141 -\mathbb{k}[v_1](1+\alpha_2)-\mathbb{k}[v_2])}_{=-\mathcal{b } } f^a,\ ] ] such that @xmath142 .in this section we derive the equation of motion for the gauge - invariant variable found in the previous section . for the sake of completeness , we recall that the equations of motion for the background part are : @xmath143 ) , \\ \dot{\bar{k } } & = & - \frac{\bar{n}}{\sqrt{\bar{p}}}\left [ \frac{1}{2 } ( \mathbb{k}[1])^2+\bar{p}\frac{\partial}{\partial \bar{p } } ( \mathbb{k}[1])^2 \right ] \nonumber \\ & + & \frac{\kappa}{3v_0 } \left(\frac{\partial \bar{h}_{\rm{m}}}{\partial \bar{p } } \right ) , \end{aligned}\ ] ] where @xmath144 and @xmath139 . for a free scalar field , an analytical solution to these equations can be found @xcite : @xmath145 this solution represents a symmetric bounce .the diffeomorphism constraint @xmath146= 0 $ ] leads to the equation @xmath147\right)\delta^k_a ( \partial_d \delta e^d_k ) = \kappa \bar{\pi } \partial_a(\delta \varphi ) .\label{diffconstr}\ ] ] using the symmetrized variables @xmath148-\mathbb{k}[v_1](1+\alpha_2))\left ( { f_{a,}}^i+{f^i}_{,a}\right ) \right .\nonumber \\ & + & \left .\left ( { f_{a,}}^i+{f^i}_{,a } \right)\dot{}-\left ( { s_{a,}}^i+{s^i}_{,a } \right ) \right ] \nonumber \\ & = & -\frac{1}{2 } \left ( { \sigma_{a,}}^i+{\sigma^i}_{,a}\right)+\frac{1}{2 } \mathbb{k}[v_2]\left ( { f_{a,}}^i+{f^i}_{,a}\right ) , \label{deltak}\end{aligned}\ ] ] and @xmath149 equation ( [ diffconstr ] ) can be rewritten as @xmath150 because @xmath151 ( from the anomaly - free condition ) , the symmetric diffeomorphism constraint simplifies to the laplace equation @xmath152 . since , the spatial slice is flat ( @xmath153 ) there are no boundary conditions on @xmath154this restricts the possible solutions of the laplace equation to @xmath155 , where @xmath156 and @xmath157 are sets of constants .however , because @xmath154 is a perturbation ( there is no contribution from the zero mode ) , @xmath158 as required from the consistency of the perturbative expansion .this is also the reason why the first order perturbation of the hamiltonian is vanishing , @xmath159 .condition ( [ condition ] ) implies @xmath160 and @xmath161 , which leads to @xmath162 .this shows that our gauge invariant variable @xmath154 is identically equal to zero in absence of vector matter , in agreement with earlier studies @xcite .this can also be proved by expanding @xmath154 into fourier modes . in order to have non - vanishing ( physical ) vector modes@xmath154 , a source term in equation ( [ nablasigma ] ) therefore has to be present . with `` vector matter '', this reads as @xcite : @xmath163 where @xmath164 and @xmath165 are the energy density and pressure of the vector matter and @xmath166 is a matter perturbation vector .if @xmath167 then @xmath168 so physical vector perturbations are expected .however , it should be pointed out that proving that the formulation remains anomaly - free in presence of the vector matter remains an open issue .this could be checked,_e.g . _ , by introducing an electromagnetic field in the hamiltonian formulation @xcite .we leave this problem to be analyzed elsewhere . due to the gauss constraint, we introduce the symmetrized variable @xmath169 the equation of motion for this variable reads as : @xmath170+\mathcal{b } ) \mathfrak{s}^i_a + \mathcal{a } { f^{(i } } _ { , a)}=\kappa \bar{p } \delta t^{(i}_{a ) } , \label{vecteqfull}\ ] ] where @xmath171.\ ] ] for scalar matter @xmath172 .the same holds for tensor modes @xcite ( the reasons are the same because @xmath44 and @xmath173 ) . when imposing the anomaly freedom conditions @xmath94 and @xmath100 , equation ( [ vecteqfull ] ) simplifies to @xmath174)}_{= \frac{1}{\bar{p } } \frac{d \bar{p}}{d\eta } } \mathfrak{s}^i_a = 0 , \label{vectoreq}\ ] ] with fully determined coefficientsof course without vector mater , as discussed above , the variable @xmath175 is equal to zero and the equation ( [ vectoreq ] ) is trivially satisfied .however , the presence on a non - vanishing contribution from @xmath166 allows for non - trivial solutions of equation ( [ vectoreq ] ) .in such a case , equation ( [ vectoreq ] ) leads to : @xmath176 for a symmetric bounce driven by a free scalar field : @xmath177 the evolution is smooth through the bounce .the amplitude of the perturbations grows during the contraction and decreases in the expanding phase .the maximum amplitude is reached at the transition point ( bounce ) . moreover ,this evolution is independent on the length of the considered mode , as can be seen by performing a fourier transform of the function @xmath154 . because of this , there is significant difference with respect to tensor and scalar perturbations . for the scalar and tensor perturbations ,the evolution is different depending on whether the mode length is shorter or longer that the hubble horizon .in particular , on super - horizon scales , the amplitude of the scalar and tensor perturbations is _ frozen_. in contrast , for the vector modes there is no such effect .therefore , in an expanding universe , the amplitude of vector modes decreases with respect to the super - horizon tensor and scalar perturbations .the contribution from vector modes becomes negligible during the expansion phase .however , the situation reverses in the contracting phase , before the bounce .then , the amplitude of the vector perturbations grows with respect to the super - horizon tensor and scalar perturbations .therefore , on very large scales the vector perturbations can play an important role , _e.g. _ leading to the generation of large scale magnetic fields @xcite .this could lead to a new tool to explore physics of the ( very ) early universe .in this paper we have studied the issue of anomaly cancellation for vector modes with holonomy corrections in lqc .our strategy is based on the introduction of counter - terms in the holonomy - corrected hamiltonian constraint . in our study , we have also introduced possible holonomy corrections to the diffeomorphism constraint .we have shown , first , that the anomaly cancellation can not be achieved without counter - terms .holonomy corrections to the diffeomorphism constraint do not help significantly to fulfill the anomaly freedom conditions , that are anyway satisfied up to the fourth order in the canonical variable @xmath21 .then , we have studied the anomaly issue for the gravitational sector with two counter terms .we have shown that the conditions of anomaly freedom can be met with at least one non - vanishing counter - term .the resulting effective holonomy corrections are almost periodic functions only if the diffeomorphism constraint is holonomy corrected .subsequently , we have investigated the issue of anomaly cancellation when a matter scalar field is added . in this case , the closure conditions are more restrictive and fully determine the form of the resulting hamiltonian constraint .moreover , this requires that the diffeomorphism constraint holds its classical form , in agreement with lqg expectations .because of this , the effective holonomy corrections , which take into account contributions from the counter - terms , are no more almost periodic functions .we have found the gauge invariant variable and the corresponding equation of motion .the solution to this equation were also given .we have analyzed this solution for the symmetric bounce model to point out that the vector perturbations smoothly pass through the bounce , where their amplitude reaches its maximum but finite value .the work performed here for scalar matter should be extended to vector matter to fully address the considered issue . in @xcite, we address the related issue of anomaly freedom for scalar perturbations with holonomy corrections .this is most important from the observational viewpoint .tc and jm were supported from the astrophysics poland - france ( astro - pf ) .jm has been supported by polish ministry of science and higher education grant n n203 386437 and by foundation of polish science award start .99 a. ashtekar , phys . rev .* 57 * ( 1986 ) 2244. h. nicolai , k. peeters and m. zamaklar , class .* 22 * ( 2005 ) r193 [ arxiv : hep - th/0501114 ] .a. ashtekar and j. lewandowski , class .* 21 * ( 2004 ) r53 [ arxiv : gr - qc/0404018 ] ; + c. rovelli , _ quantum gravity _ , cambridge , cambridge university press , 2004 ; + c. rovelli , living rev . relativity , * 1 * , 1 ( 1998 ) ; + l. smolin , arxiv : hep - th/0408048v3 ; + t. thiemann , lect .notes phys . *631 * , 41 ( 2003 ) ; + a. perez , arxiv : gr - qc/0409061v3 ; + p. dona & s. speziale , arxiv:1007.0402v1 .m. bojowald , g. m. hossain , m. kagan and s. shankaranarayanan , phys . rev .d * 78 * ( 2008 ) 063547 [ arxiv:0806.3929 [ gr - qc ] ] .m. bojowald , g. m. hossain , m. kagan and s. shankaranarayanan , phys .d * 79 * ( 2009 ) 043505 [ erratum - ibid .d * 82 * ( 2010 ) 109903 ] [ arxiv:0811.1572 [ gr - qc ] ] .m. bojowald and g. m. hossain , class .* 24 * ( 2007 ) 4801 [ arxiv:0709.0872 [ gr - qc ] ] . m. bojowald and g. m. hossain , phys .d * 77 * ( 2008 ) 023508 [ arxiv:0709.2365 [ gr - qc ] ] .m. bojowald and g. calcagni , arxiv:1011.2779 [ gr - qc ] . m. bojowald , g. calcagni and s. tsujikawa , arxiv:1101.5391 [ astro-ph.co ] . y. li and j. y. zhu , class .. grav .* 28 * ( 2011 ) 045007 [ arxiv:1102.2720 [ gr - qc ] ] .a. ashtekar , e. wilson - ewing , phys .* d80 * ( 2009 ) 123532 .[ arxiv:0910.1278 [ gr - qc ] ] .w. nelson and m. sakellariadou , phys .d * 76 * ( 2007 ) 104003 [ arxiv:0707.0588 [ gr - qc ] ] .w. nelson , m. sakellariadou , phys .* d76 * ( 2007 ) 044015 .[ arxiv:0706.0179 [ gr - qc ] ] .a. corichi , p. singh , phys .* d78 * ( 2008 ) 024034 .[ arxiv:0805.0136 [ gr - qc ] ] .m. sakellariadou , j. phys .* 189 * ( 2009 ) 012035 [ arxiv:0810.5356 [ gr - qc ] ] .a. ashtekar , m. bojowald and j. lewandowski , adv .* 7 * ( 2003 ) 233 [ arxiv : gr - qc/0304074 ] . j. mielczarek , t. stachowiak and m. szydlowski , phys .d * 77 * ( 2008 ) 123506 [ arxiv:0801.0502 [ gr - qc ] ] .

<S> we investigate vector perturbations with holonomy corrections in the framework of loop quantum cosmology . </S><S> conditions to achieve anomaly freedom for these perturbations are found at all orders . </S><S> this requires the introduction of counter - terms in the hamiltonian constraint . </S><S> we also show that anomaly freedom requires the diffeomorphism constraint to hold its classical form when scalar matter is added although the issue of a vector matter source , required for full consistency , remains to be investigated.the gauge - invariant variable and the corresponding equation of motion are derived . </S><S> the propagation of vector modes through the bounce is finally discussed . </S>

0707.4021

high - precision measurements of hadrons and hadron jets have become increasingly important in experimental particle physics .such measurements are considered a crucial ingredient of experiments at a future high - energy linear electron - positron collider . historically , by farthe best performance in this respect has been delivered by _ compensating _ hadron calorimeters @xcite . in these instruments ,the response to the electromagnetic ( em ) and non - electromagnetic shower components is equalized by design , and therefore the detrimental effects of event - to - event fluctuations in the energy sharing between these components are eliminated .these effects include hadronic signal non - linearity , a poor hadronic energy resolution , especially at high energies where deviations from @xmath0 scaling become the dominant factor , and a non - gaussian response function . in recent years, an alternative technique has been developed : the dual readout method ( dream ) .dream calorimeters offer the same advantages as compensating ones , and are not subject to the disadvantages of the latter .these disadvantages derive from the fact that compensation can only be achieved in sampling calorimeters with a small sampling fraction ( 2.3% in lead / plastic - scintillator ) , plus the fact that compensation relies upon efficient detection of the neutrons produced in the shower development .these requirements limit the electromagnetic ( em ) energy resolution achievable with these instruments , while the excellent hadronic performance is only achieved with sufficiently large detector volumes and integration times .dream calorimeters are based on a simultaneous measurement of different types of signals which provide complementary information about details of the shower development .the first calorimeter of this type that we developed and tested ( in the context of a generic r&d project ) was based on a copper absorber structure , equipped with two types of active media .scintillating fibers measure the total energy deposited by the shower particles , while erenkov light is only produced by the charged , relativistic shower particles .since the latter are almost exclusively found in the em shower component ( dominated by @xmath1s produced in hadronic showers ) , a comparison of the two signals makes it possible to measure the energy fraction carried by this component , @xmath2 , event by event . as a result , the effects of fluctuations in this component , which are responsible for all traditional problems in non - compensating calorimeters , can be eliminated .this leads to an important improvement in the hadronic calorimeter performance .the performance characteristics of this detector are described elsewhere @xcite .once the effects of the dominant source of fluctuations , fluctuations in the em energy fraction @xmath2 , are eliminated , the performance characteristics are determined ( and limited ) by other types of fluctuations . in the described detector ,a prominent role was played by the small number of erenkov photoelectrons constituting the signals ( 8 p.e./gev ) .however , there is absolutely no reason why the dream principle would only work in fiber calorimeters , or even in sampling calorimeters for that matter .one could in principle even use a homogeneous ( fully sensitive ) detector , provided that the light signals can be separated into scintillation and erenkov components . in this paper, we describe the results of tests of this idea .a small electromagnetic calorimeter made of lead tungstate ( pbwo@xmath3 ) crystals was tested in conjunction with the dream calorimeter mentioned above , and exposed to high - energy particle beams at cern s super proton synchrotron . in section 2 , we describe the detectors and the experimental setup in which they were tested . in section 3 , we discuss the experimental data that were taken and the methods used to analyze these data . in section 4 , the experimental results are described and discussed . a summary and conclusions are presented in section 5 .the calorimeter system used in these experiments comprised two sections.the electromagnetic section ( ecal ) consisted of 19 lead tungstate ( pbwo@xmath3 ) crystals .each crystal was 18 cm long , with a cross section of @xmath4 @xmath5 .these crystals were arranged in a matrix , as shown in figure [ ecal ] . for the purpose of these tests , this ensemble of crystalswas considered one unit .the crystals were not optically isolated from each other , and the light produced by showering particles was read out by only two photomultiplier tubes ( pmt ) , one located at each end of the crystal matrix .the light was funneled into these tubes by means of cones made of aluminized mylar , a highly reflective material . for the hadronic section ( hcal ) of the calorimeter system, we used the original dream calorimeter @xcite .the basic element of this detector is an extruded copper rod , 2 meters long and 4@xmath6 4 mm@xmath7 in cross section .this rod is hollow , and the central cylinder has a diameter of 2.5 mm .seven optical fibers were inserted in this hole .three of these were plastic scintillating fibers , the other four fibers were undoped fibers , intended for detecting erenkov light .the instrumented volume had a length of 2.0 m ( 10 @xmath8 , 100 @xmath9 ) , an effective radius of 16.2 cm and a mass of 1030 kg .the fibers were grouped to form 19 hexagonal towers .the effective radius of each tower was 37.1 mm ( @xmath10 ) .a central tower ( # 1 ) was surrounded by two hexagonal rings , the inner ring ( 6 towers , numbered 2 - 7 ) and the outer ring ( 12 towers , numbered 8 - 19 ) .the towers were longitudinally unsegmented .the fibers sticking out at the rear end of this structure were separated into 38 bunches : 19 bunches of scintillating fibers and 19 bunches of erenkov fibers . in this way , the readout structure was established .each bunch was coupled through a 2 mm air gap to a pmt .more information about this detector is given elsewhere @xcite .the measurements described in this paper were performed in the h4 beam line of the super proton synchrotron at cern .the detectors were mounted on a platform that could move vertically and sideways with respect to the beam .the ecal was rotated into different positions ( see section 4 ) by hand .two small scintillation counters provided the signals that were used to trigger the data acquisition system .these trigger counters were 2.5 mm thick , and the area of overlap was 6@xmath66 @xmath5 .a coincidence between the logic signals from these counters provided the trigger .measurement of the time structure of the calorimeter signals formed a very important part of the tests described here . in order to limit distortion of this structure as much as possible , we used 15 mm thick air - core cables to transport the ecal signals to the counting room .such cables were also used for the signals from the trigger counters , and these were routed such as to minimize delays in the daq system in these cables . ] .the hcal signals were transported through rg-58 cables with ( for timing purposes ) appropriate lengths to the counting room .the ecal signals were split ( passively ) into 5 equal parts in the counting room .one part was sent to a charge adc , the other 4 signals were used for analysis of the time structure by means of a fadc .the latter unit measured the amplitude of the signals at a rate of 200 mhz . during a time interval of 80 ns, 16 measurements of the amplitude were thus obtained .the 4 signals from the splitter box were measured separately in 4 different channels of the fadc module .signals 2 , 3 and 4 were delayed by 1.25 ns , 2.50 ns and 3.75 ns with respect to signal 1 . by using 4 channels of the fadc module in this way , the time structure of the signals was thus effectively measured with a resolution of 1.25 ns ( 800 mhz ) . the quality of the information obtained in this way is illustrated in figure [ esignal ] , which shows the average time structure of the signals from 50 gev electron showers developing in the lead tungstate ecal .the trailing edge of these signals is well described by an exponential decay with a time constant of 9.7 ns .the charge measurements were performed with 12-bit lecroy 1182 adcs . these had a sensitivity of 50 fc / count and a conversion time of 16 @xmath11s .the adc gate width was 100 ns , and the calorimeter signals arrived @xmath12 ns after the start of the gate .the data acquisition system used vme electronics .a single vme crate hosted all the needed readout and control boards .the trigger logic was implemented through nim modules and the signals were sent to a vme i / o register , which also collected the spill and the global busy information .the vme crate was linked to a linux based computer through an sbs 620 optical vme - pci interface that allowed memory mapping of the vme resources via an open source driversanshiro / kinoko - e / vmedrv/ ] .the computer was equipped with a 2 ghz pentium-4 cpu , 1 gb of ram , and was running a cern slc 4.3 operating system .the data acquisition was based on a single - event polling mechanism and performed by a pair of independent programs that communicated through a first - in - first - out buffer , built on top of a 32 mb shared memory .only exclusive accesses were allowed and concurrent requests were synchronised with semaphores .the chosen scheme optimized the cpu utilization and increased the data taking efficiency by exploiting the bunch structure of the sps , where beam particles were provided to our experiment during a spill of 4.8 s , out of a total cycle time of 16.8 s. during the spill , the readout program collected data from the vme modules and stored them into the shared memory , with small access times . during the remainder of the sps cycle , a recorder program dumped the events to the disk .moreover , the buffer presence allowed low - priority monitoring programs to run ( off - spill ) in spy mode . with this scheme , we were able to reach a data acquisition rate as high as 2 khz , limited by the fadc readout time .the typical event size was @xmath13 kb . all calorimeter signals and the signals from the auxiliary detectors were monitored on - line .the two pmts reading out the two sides of the ecal were calibrated with 50 gev electrons . to this end ,the ecal was oriented such that the beam entered the detector perpendicular to the crystal axis , and the two pmts collecting the light generated by the showers were located in a plane perpendicular to the beam axis ( figure [ setup1 ] ) .= 5.5 cm the main purpose of this setup was to equalize the gain of the two pmts in a geometry where the relative contributions of scintillation and erenkov light to the signals were the same for both . given the radiation length of pbwo@xmath3 ( 8.9 mm ) , the ecal was only @xmath14 deep in this geometry , and therefore , a substantial fraction of the shower energy leaked out .the energy equivalent of the signals thus had to be established on the basis of ( egs4 ) monte carlo simulations .the 38 pmts reading out the 19 towers of the hcal were also all calibrated with 50 gev electrons .the showers generated by these particles were not completely contained in a single calorimeter tower .the ( average ) containment was found from egs4 monte carlo simulations . when the electrons entered a tower in its geometrical center , on average @xmath15 of the scintillation light and @xmath16 of the erenkov light was generated in that tower @xcite .the remaining fraction of the light was shared by the surrounding towers .the signals observed in the exposed tower thus corresponded to an energy deposit of 46.3 gev in the case of the scintillating fibers and of 46.8 gev for the erenkov fibers .this , together with the precisely measured values of the average signals from the exposed tower , formed the basis for determining the calibration constants , the relationship between the measured number of adc counts and the corresponding energy deposit .the main purpose of these tests was to see if the ecal signals could be split into their scintillation and erenkov components . in order to optimize the possibilities in this respect ,the ecal was oriented such that the angle between the crystal axis and the beam axis was equal to the erenkov angle , @xmath17 : @xmath18 since the refractive index of pbwo@xmath3 is 2.2 , @xmath19 . detailed measurements of the angular dependence of the erenkov / scintillation ratio @xcite revealed indeed a maximum value near @xmath20 , and therefore we chose this angle for our tests .the detector setup is shown in figure [ setup2 ] . in this geometry ,the effective depth of the ecal amounted to @xmath21 , and it contained on average @xmath22 of 50 gev electron showers .the remaining 10% of the shower energy was recorded in the hcal .= 5.5 cm we exposed the calorimeter system to 50 gev electrons and to 50 and 100 gev @xmath23 . for each run,100 000 events were collected .the time structure of both ecal signals was measured with 1.25 ns resolution . the integrated charge carried by these signals and by those from the 38 hcal channelswas digitized with 12 bit resolution .the beams turned out to be very clean .contamination in the electron beam was less than 10@xmath24 .the pion beams contained muons , at the few-% level .these muons could be easily recognized and removed from the event samples .we used the muon events for separate studies .in lead tungstate , erenkov light is emitted at an angle of @xmath25 by charged shower particles that are sufficiently relativistic .unlike the scintillation light , which is isotropically emitted , there is a clear directional preference in the erenkov light production of high - energy showers .even though much of the shower energy is deposited by isotropically distributed electrons , produced in compton scattering and photoelectric processes , the effects of that on the angular distribution of the emitted erenkov light are limited , since much of this takes place below the erenkov threshold ( @xmath26 mev/@xmath27 ) .quantitative information on this was obtained from the angular dependence of the response of quartz - fiber calorimeters ( which are exclusively detecting erenkov light ) to high - energy electron showers @xcite .the maximum response was obtained when the fibers were oriented at the erenkov angle with respect to the shower axis .this response was reduced by a factor 2 - 3 when the fibers were aligned with the beam axis , and by a factor 10 when read out at the upstream end .we are exploiting this feature by comparing the signals from the two ecal channels in the @xmath25 geometry ( figure [ setup2 ] ) .any erenkov light produced in the shower development will be preferentially observed in the downstream channel ( to be named channel @xmath28 in the following ) , whereas the upstream channel ( channel @xmath29 ) will see a much smaller fraction of it . since both pmts see the same amount of scintillation light , the _ forward / backward asymmetry _ ,@xmath30 , is a measure for the fraction of erenkov light contained in the signals .= 6.6 cm figure [ asym_e]a shows the measured asymmetry for the 50 gev electron signals from the ecal .indeed , the @xmath28 signals are significantly larger than the @xmath29 ones , on average by @xmath31 .the forward / backward asymmetry was measured to be @xmath32 .forward / backward asymmetry was also observed in the ecal signals from pions and muons .the muon asymmetry is shown in figure [ asym_e]b .the asymmetry is larger than for electron showers , the @xmath28 signals were measured to be , on average , larger by @xmath33 than the @xmath29 ones ( the forward / backward asymmetry was measured to be @xmath34 ) .however , the event - to - event fluctuations are considerably larger .the @xmath35 of the distribution for muons is larger by a factor 35 - 40 than the width of the distribution for electron showers .the latter phenomenon is due to the fact that the electron signals are considerably larger ( 45 gev vs 0.1 gev for a mip ) , so that poisson fluctuations in the number of photoelectrons account for at least half of this differenceb gave a @xmath36 of 0.60 .if ( poisson ) photoelectron statistics was the only contributing factor to the width , one should expect a sigma of 0.46 . ] .in addition , because of the landau distribution of the muon signals , these statistical fluctuations are non - poissonian for these particles .the larger average asymmetry for muons should be expected on the basis of the fact that erenkov light generated in em showers contains an isotropic component , mainly produced by compton electrons @xcite . in a separate paper , we have shown that measurements of em showers in thin pbwo@xmath3 crystals revealed that the forward / backward asymmetry decreases as the em shower develops . in the first @xmath37 ,an asymmetry of @xmath38 was measured , but this asymmetry decreased considerably when deeper regions of the shower were probed .the asymmetries measured here for complete em showers and for single particles ( muons ) are in complete agreement with these observations .= 9.5 cm after removing the ( 6% ) muons from the 50 gev hadronic event sample , the remaining pion events exhibited some interesting characteristics , which are a direct consequence of the large @xmath39 ratio of the pbwo@xmath3 crystal calorimeter value was measured to be 2.4 @xcite . ] .figures [ pions]a and [ pions]b show the total signal distributions in the hcal and ecal , respectively .the hcal distributions exhibits a structure that is the sum of two distinct substructures , as illustrated in figures [ pions]c and [ pions]d .figure [ pions]c shows the hcal distribution for pions that penetrated the ecal .these events populate the mip peak in the ecal distribution .the fact that these pions represent 40% of the total means that the ecal s thickness in this geometry corresponded to @xmath40 nuclear interaction lengths .the pion events were subdivided into samples depending on the total signal produced in the ecal .the underlying idea is that the relatively thin ecal will , for all practical purposes , only contain the remnants of the first generation of nuclear reactions in the hadronic shower development .if the first interaction took place in the first two crystals of the ecal , then @xmath1s produced in this first interaction typically deposited most of their energy in the ecal .therefore , one should expect a correlation between the total ecal signal and the fraction of erenkov light .figure [ asymmetry ] clearly exhibits such a correlation . by comparing the results shown in figures [ asym_e]a and [ asymmetry ], the average fraction of the ecal signal carried by em shower components may be determined .however , this determination is complicated by another effect shown in figure [ asymmetry ] , namely the landau tail of pions penetrating the ecal .not surprisingly , the events in the mip peak of the pion distribution exhibit the same asymmetry as the muons from our sample : @xmath41 , where the smaller error reflects the difference in event statistics . however , event samples with signals larger than the mip peak in the ecal do contain some fraction of penetrating pions , from the landau tail .because of the much larger asymmetry for such events , the overall asymmetry for these event samples is measurably affected .a similar effect may be expected for events in which the first pion interaction occurred deep inside the ecal .measurements with a single crystal indicated that em showers developing only over a few radiation lengths exhibit an asymmetry of @xmath38 , not much smaller than that for mips @xcite . as a result of these complications ,the measured asymmetry starts to reflect the em component of the hadron showers only for events in which at least 10 gev ( 20% of the total energy ) was deposited in the ecal . in order to check the latter statement ,we investigated the relationship between the erenkov components of the signals observed in the em and hadronic sections of our calorimeter system . in order to limit the contaminating effects of the penetrating pions , we limited this study to pions that deposited more than 10 gev in the ecal .the relative contribution of erenkov light to the signals was derived from the @xmath30 asymmetry in the ecal and from the @xmath42 signal ratio in the dream hadronic section .figure [ ehccorr ] shows that there is a clear correlation between these quantities . andsince the @xmath42 signal ratio is directly related to the em shower fraction , @xmath2 , we conclude from this that the asymmetry measured in the ecal is indeed an indicator for the fraction of the ecal signal carried by the em shower component .this can also be seen as follows .if we denote the ecal erenkov signal by @xmath43 and the scintillation signal by @xmath44 , as in the dream hadronic section , then the forward / backward asymmetry in the ecal signals can be written as @xmath45 since in practice @xmath46 , the asymmetry is approximately equal to 0.5 @xmath42 , and therefore the asymmetry is related in a similar way to @xmath2 as the @xmath42 signal ratio measured for the dream hadronic section . in reference@xcite , we derived the exact relationship between @xmath42 and @xmath2 for dream : @xmath47 where the factors 0.21 and 0.76 are the inverse values of the @xmath39 ratios of the cu / quartz and cu / plastic - scintillator sampling structures , respectively .a similar relationship can be derived between the forward / backward signal asymmetry and the em shower fraction in ecal .two modifications of equation [ eq2 ] are important : 1 .the @xmath39 value of ecal as a scintillation device is much larger than for the cu / plastic sampling structure in dream : 2.4 1.3 2 .since the scintillation and erenkov signals of the ecal are not independently calibrated , as in dream , an overall calibration factor is needed that relates the strengths of the @xmath43 and @xmath44 signals in ecal .this factor is such that the asymmetry is 0.044 for @xmath48 ( pure em showers , see figure [ asym_e ] ) .these considerations lead to the following relationship between the measured forward / backward asymmetry in the ecal signals ( with the detector oriented at the erenkov angle ) and the em fraction of the shower energy deposited in this calorimeter section : @xmath49 this relationship is graphically shown in figure [ fem_bf ] .the same figure also shows ( on the top horizontal axis ) the @xmath42 ratio in the fiber calorimeter section .if the experimental data points were located on this curve , we would conclude that the em shower fraction derived from the em and hadronic signal characteristics was _ exactly _ the same .however , since the crystal and the fiber sections of the calorimeter system probed different parts of the shower development , and especially since the crystal section only probed the first 0.5 nuclear interaction lengths , such a perfect one - to - one correspondence should not be expected .however , figure [ fem_bf ] does show a clear correlation between the em shower fractions derived from both detector segments .the practical merits of all this were studied by investigating the relationship between the asymmetry in the ecal signals and the _ total _ response of the calorimeter system ( ecal + hcal ) , for this event sample .we recall that in dream , there is a direct relationship between the @xmath42 signal ratio , the ratio of the total signals measured in the erenkov fibers and the scintillation fibers , and the em shower fraction , @xmath2 @xcite . by selecting events with a certain @xmath42 value ,the @xmath2 value is fixed .the total calorimeter signal for these events is different from that for events with a different @xmath42 , and thus @xmath2 value .= 9.5 cm this is illustrated in figure [ proof ] , which shows in diagram @xmath50 two signal distributions for event samples with very different @xmath42 values .the @xmath42 bins used for these samples are shown in figure [ proof]a .the events selected for this purpose all concern pions that penetrated the ecal without a nuclear interaction .therefore , almost all the energy carried by these pions was deposited in the fiber calorimeter section .these results illustrate that a larger em shower fraction ( selected by means of the @xmath42 signal ratio ) leads to a considerably larger total calorimeter signal , especially in the extremely non - compensating copper / quartz - fiber structure ( @xmath51 ) used for this purpose . in figure [ proof]d ,two signal distributions are shown for events that were selected on the basis of the forward / backward asymmetry measured in the ecal crystal section .the @xmath30 bins used for these two samples are shown in figure [ proof]c .only events in which the pions deposited at least 10 gev in the crystal were used for this purpose , and the total signal was calculated as the sum of the signal measured in pmt @xmath28 and the signal from the scintillating fibers in the hadronic section .these results exhibit , at least qualitatively , the same characteristics as those shown above for the penetrating pions , where the @xmath42 ratio provided information on @xmath2 : the larger the asymmetry measured in the ecal signal , the larger the relative contribution of erenkov light to the ecal signals , the larger the em shower fraction , and thus the larger the total calorimeter signal becomes . in figure [ dreamtrick ] ,the described phenomena are shown for the entire range of possible @xmath2 values .the total calorimeter signal for penetrating events ( mip in ecal ) is plotted as a function of the @xmath42 value in figure [ dreamtrick]a , and in figure [ dreamtrick]b as a function of the @xmath30 value for pions interacting in the ecal ( and depositing more than 10 gev in the crystals ) .a few remarks are in order .the fact that the effect shown in figure [ dreamtrick]a ( which we will call the _ reference effect _ ) is much larger than that in figure [ dreamtrick]b can be ascribed to two factors : 1 .the reference effect concerns exclusively erenkov light .we are looking at the total erenkov signal as a function of @xmath2 .had we chosen the signals from the scintillating fibers instead , the observed increase in the calorimeter response would have been only 15% , instead of the factor of two measured for the erenkov reponse @xcite .this is a consequence of the very large difference between the @xmath39 values ( 4.7 1.3 ) , which forms the very basis of the dream principle @xcite . for the em contentderived on the basis of the forward / backward asymmetry in the crystal signals , we do nt have the option to look exclusively at erenkov light .a very large fraction of the signal from the crystals consists of scintillation light , even if we optimize the event selection for em shower content , erenkov contributions .the ecal is only @xmath52 deep .even though much of the em component of hadron showers derives from @xmath1 production in the first nuclear interaction of the showering particle , the measurement of @xmath2 for the entire shower on the basis of the signals from the first @xmath52 has its limitations . in light of these considerations , the effect observed in figure [ dreamtrick]b is actually quite remarkable , although it should also be pointed out that , since the @xmath39 value of a homogeneous pbwo@xmath3 calorimeter is @xmath53 , a larger increase than the 15% mentioned above should be expected for the combination considered here .figure [ linearity ] summarizes the non - linearity characteristics for the different signals and signal combinations .we also want to point to another difference between the data sets shown in figures [ dreamtrick]a and [ dreamtrick]b . as explained in reference @xcite ,there is a linear relationship between @xmath2 and the total calorimeter signal .even though the parameters @xmath42 and@xmath30 are not exactly proportional to @xmath2 , it is remarkable that the signal dependence for @xmath42 is approximately linear , while it is not for @xmath30 .this may be explained from the fact that the latter curve extends beyond the physically meaningful region , @xmath54 , contrary to the @xmath42 curve .this region is limited to @xmath55 , and the points outside this region probe the tails of the distribution .this may explain the s- shaped " curve of figure [ dreamtrick]b . a similar deviation from linearityis actually also observed if points covering the unphysical region @xmath56 are included in figure [ dreamtrick]a .a second valuable tool for recognizing the contributions of erenkov light to the calorimeter signals is derived from the time structure of the events .this is illustrated in figure [ fbtime ] , which shows the average time structure of the 50 gev shower signals recorded with the same pmt ( b ) in the two geometries shown in figures [ setup1 ] and [ setup2 ] .the two distributions have been normalized on the basis of their trailing edge ( 24 - 80 ns after the start of the fadc digitization ) , and are indeed in great detail identical in that domain , both for the electron ( figure [ fbtime]c ) and for the pion ( figure [ fbtime]d ) signals .this part of the pulses is completely determined by the decay characteristics of the scintillation processes in the pbwo@xmath3 crystals ( see also figure [ esignal ] ) and should thus be independent of the detector orientation .however , there is a very significant difference in the _ leading _ edge of the pulses .the ones measured in the 63@xmath57 geometry ( figure [ setup2 ] ) exhibit a steeper rise than the ones from the calibration geometry ( figure [ setup1 ] ) . figures [ fbtime]a and b show the result of subtracting the latter pulse shape from the `` 63@xmath57 '' one : the pulses recorded in the 63@xmath57 geometry contain an additional `` prompt '' component of the type one would expect from erenkov light . in the case of the electron showers , this additional component represents @xmath58 of the total signal . for comparison , we recall that the forward / backward asymmetry measurements led us to conclude that the signals from pmt b contained , on average , 9% of erenkov light .the pion signals also exhibit a prompt component , which represents a somewhat smaller fraction of the total signal than in the case of the electrons .we have studied the possible application of differences in the time structure of the calorimeter signals to determine the erenkov component of the light produced in the crystal calorimeter , in a similar way as we used the measured measured forward / backward asymmetry ( section 4.1 ) .to this end , we used the time at which the calorimeter signal reached half of its amplitude value . the larger the relative contribution of erenkov light ,the earlier this threshold was reached .since the contribution of erenkov light to the signals from the downstream pmt ( b ) was larger than that from the upstream pmt ( a ) , we should thus expect the _ difference _ between the times at which the two signals reach 50% of their amplitude level ( @xmath59 ) , to be a measure for the fraction of erenkov light in the signals from the crystal calorimeter .figure [ tatb ] shows this difference , for 50 gev @xmath23 showers , as a function of the fraction of erenkov light , derived from the @xmath42 signal ratio measured in the fiber section of the calorimeter .the time difference @xmath59 increases indeed with this fraction . in order to check the significance of this result, we also repeated this analysis for the measurements done with the same particles ( 50 gev @xmath23 ) in the calibration geometry ( figure [ setup1 ] ) .the results , shown as squares in figure [ tatb ] , indicate no significant dependence on the erenkov fraction in the latter ( 90@xmath57 ) setup . as in the case of the forward / backward asymmetry ,only events depositing at least 10 gev in the crystal calorimeter section were considered in these analyses , and therefore the results from figures [ tatb ] and [ ehccorr ] may be directly compared to each other .we have also studied the correlation between this time difference and the total calorimeter signals .the results are shown in figure [ dreamtrick2]a , which shows the sum of the signals from ecal @xmath28 and the hcal scintillating fibers , as a function of @xmath59 , for 50 gev @xmath23 showers that deposited at least 10 gev in the ecal ( early starters ) .figure [ dreamtrick2]b shows these total signal distributions for two subsets of events , selected on the basis of the time difference @xmath59 .these results may be directly compared with those depicted in figures [ dreamtrick]b and [ proof]d , which concern a similar analysis on the basis of the forward / backward asymmetry .we conclude that the time structure of the crystal signals offers equally valuable opportunities for unraveling the crystal signals into their scintillation and erenkov components , and thus of an event - by - event measurement of @xmath2 as the directionality of the light production .we have measured the contribution of erenkov light to the signals from electrons , muons and hadrons in an electromagnetic calorimeter made of lead tungstate crystals . in the chosen geometry ,which was optimized for detecting this component , information about this contribution was obtained from the forward / backward asymmetry in the signals and from their time structure .for single particles traversing the calorimeter ( muons , pions ) , the erenkov contribution was measured to be @xmath33 .the measurements for showers indicated contributions at about half that level , since a substantial fraction of the signal is in that case typically caused by isotropically distributed shower particles .the pion measurements made it possible to determine the electromagnetic content of the energy deposited in the crystal calorimeter event by event .this information was found to correlate well with explicit measurements of this fraction in the dual - readout calorimeter that served as the hadronic section in these measurements .it could also be used in the same way to determine the electromagnetic shower fraction and thus improve the hadronic calorimeter performance .both the forward / backward asymmetry between the signals from the pmts reading out both ends of the crystal , and the differences in the time structure of these two signals could be used for this purpose .the studies reported in this paper were carried out with pbwo@xmath3 crystals made available by the phos group of the alice collaboration .we sincerely thank drs .mikhail ippolitov and hans muller for their help and generosity in this context .we thank cern for making particle beams of excellent quality available .this study was carried out with financial support of the united states department of energy , under contract de - fg02 - 95er40938 .

<S> results are presented of beam tests in which a small electromagnetic calorimeter consisting of lead tungstate crystals was exposed to 50 gev electrons and pions . </S><S> this calorimeter was backed up by the dream dual - readout calorimeter , which measures the scintillation and erenkov light produced in the shower development , using two different media . </S><S> the signals from the crystal calorimeter were analyzed in great detail in an attempt to determine the contributions from these two types of light to the signals , event by event . </S><S> this information makes it possible to eliminate the dominating source of fluctuations and thus achieve an important improvement in hadronic calorimeter performance . </S><S> calorimetry , erenkov light , lead tungstate crystals , optical fibers </S>

1506.08840

the comparison and analysis of geometric shapes plays an important role in medical imaging @xcite , biology @xcite as well as many other fields @xcite . spaces of geometric shapes are inherently nonlinear . to make standard methods of statistical analysis applicable , one can linearize the space locally around each shape .this can be achieved by introducing a riemannian structure , which is able to describe jointly the global nonlinearity of the space as well as its local linearity . over the past decade, riemannian shape analysis has become an active area of research in pure and applied mathematics . driven by applications ,a variety of spaces , equipped with different riemannian metrics , have been used to represent geometrical shapes and their attributes .ideally , the particular choice of metric should be dictated by the data at hand rather than by mathematical or numerical convenience .this leads to the task of developing efficient numerical methods for the statistical analysis of shapes for general and flexible classes of metrics .the topic of this paper are second order sobolev metrics on the space of regular , planar curves .this space and its quotients by translations , rotations , scalings , and reparametrizations are important and widely used spaces in shape analysis .second order sobolev metrics are mathematically well - behaved : the geodesic equation is globally well - posed , any two curves in the same connected component can be connected by a minimizing geodesic , the metric completion consists of all @xmath0-immersions , and the metric extends to a strong riemannian metric on the metric completion @xcite .we provide the first numerical implementation of the initial and boundary value problems for geodesics with respect to these metrics and apply them to medical data .our implementation allows us to compute karcher means , principal components , and clusters of curves in reasonable time .the parameters in the metric can be chosen freely because we do not rely on transforms which exist only for special choices of parameters @xcite .we are also able to factor out the action of the finite - dimensional translation and rotation groups .we illustrate the behaviour of the metrics in two medical applications . in sect .[ hela ] we use images of hela cell nuclei from @xcite and compute the mean shape of the nucleus and the principal modes of shape variation ; in sect .[ cardiac ] we use traces of cardiac image sequences from @xcite and compare the mean shape of patients with tetralogy of fallot with the mean shape of a control group .the paper is structured as follows : sect .[ math ] contains the definition of second order sobolev metrics and some relevant mathematical background .section [ disc ] describes our discretization of the geodesic equation and the riemannian energy functional .finally , in sect . [ app ] , we apply the metrics to medical imaging data .in this article we center our attention on the space of smooth , regular curves with values in @xmath1 , @xmath2 where @xmath3 stands for _ immersion_. as an open subset of the frchet space @xmath4 , it is a frchet manifold .its tangent space , @xmath5 , at any curve @xmath6 is the vector space @xmath4 .we denote the euclidean inner product on @xmath7 by @xmath8 .moreover , for any fixed curve @xmath6 , we denote differentiation and integration with respect to arc length by @xmath9 and @xmath10 respectively .[ def : sobolev_metric ] a second order sobolev metric on @xmath11 is given by @xmath12 where @xmath13 , @xmath14 , and @xmath15 are tangent vectors .the metric @xmath16 is invariant with respect to translations , rotations , and reparametrizations by diffeomorphisms of @xmath17 , but not with respect to scalings .it can be made scale - invariant by introducing weights that depend on the length @xmath18 of the curve @xmath6 .a scale - invariant metric is given by @xmath19 depending on the application , it can be desirable to factor out some of these isometry groups .for example , we consider the hela cell nuclei in sect .[ hela ] as curves modulo translations and rotations . for the traces of cardiac images in sect .[ cardiac ] , however , both the absolute position of the curve and its orientation have intrinsic meaning and should therefore not be factored out .the constants @xmath20 and @xmath21 in the definition of the metric determine the relative weight of the @xmath22- , @xmath23- , and @xmath0-parts . their choice is a non - trivial and important task , and it should , in each application , be informed by the data at hand .the influence of different parameter values on geodesics and karcher means can be seen in fig .[ fig : cat2cow_constants ] and fig .[ fig : karcher_constants ] . in the first rowis increased by a factor 10 in the second and a factor 100 in the third row.,title="fig:",scaledwidth=95.0% ] in the first row is increased by a factor 10 in the second and a factor 100 in the third row.,title="fig:",scaledwidth=95.0% ] in the first row is increased by a factor 10 in the second and a factor 100 in the third row.,title="fig:",scaledwidth=95.0% ] of the metric used in the first figure is increased by a factor 10 in the second and by a factor 100 in the third figure.,title="fig : " ] of the metric used in the first figure is increased by a factor 10 in the second and by a factor 100 in the third figure.,title="fig : " ] of the metric used in the first figure is increased by a factor 10 in the second and by a factor 100 in the third figure.,title="fig : " ] sobolev metrics have natural generalizations to manifold - valued curves , embedded surfaces , and more generally spaces of immersions of a compact manifold @xmath24 into a riemannian manifold @xmath25 ; see @xcite for details and @xcite for a general overview .deformations of curves can be seen as smooth paths @xmath26\to\on{imm}(s^1,\mathbb r^d)$ ] .their velocity is @xmath27 , the subscript @xmath28 denoting differentiation .the length of a path of curves @xmath6 is @xmath29 the distance between two curves in @xmath30 ( with respect to the metric @xmath16 ) is the infimum of the lengths of all paths connecting these curves .@xmath31 geodesics are locally distance - minimizing paths .geodesics can be described by a partial differential equation , called the geodesic equation .it is the first order condition for minima of the energy @xmath32 recently , some local and global existence results for geodesics of sobolev metrics were shown in @xcite .since they provide the theoretical underpinnings for the numerical methods presented in this paper , we summarize them here .let @xmath13 and @xmath14 .the geodesic equation of the metric @xmath16 , written in terms of the momentum @xmath33 , is given by @xmath34 given any initial condition @xmath35 , the solution of the geodesic equation exists for all time .if , however , @xmath36 and @xmath37 , then @xmath16 is a first order sobolev metric .its geodesic equation is locally , but not globally , well - posed .closely related are elastic metrics @xcite , which in the planar case are given by @xmath38 here @xmath39 are constants and @xmath40 denote the unit tangent and normal vectors to @xmath6 .two special cases deserve to be highlighted : for @xmath41 , @xmath42 @xcite and @xmath43 @xcite there exist nonlinear transforms , the square root velocity transform and the basic mapping , that greatly simplify numerics .both of these metrics have been applied to a variety of problems in shape analysis .we note that the elastic metric with @xmath43 corresponds to a first order sobolev metric as in def .[ def : sobolev_metric ] with @xmath44 and @xmath45 .as it has no @xmath22-part , it is a riemannian metric only on the space of curves modulo translations .we propose to represent paths of curves as tensor product b - splines @xmath46 of degree @xmath47 in time ( @xmath28 ) and @xmath48 in space ( @xmath49 ) , respectively , using @xmath50 and @xmath51 basis functions and controls @xmath52 .the spatial basis functions @xmath53 are defined on the uniform knot sequence @xmath54 , @xmath55 , and satisfy periodic boundary conditions .the time basis functions @xmath56 are defined using uniform knots on the interior of the interval @xmath57 $ ] and with full multiplicity on the boundary .full multiplicity of the boundary knots ensures that fixing the initial and final curves is equivalent to fixing the control points @xmath58 and @xmath59 . regarding smoothness , we have @xmath60)$ ] and @xmath61)$ ] .the proposed discretization scheme is independent of the special form of the metric , and can be applied equally to the scale - invariant versions as well as to the family of elastic metrics . to evaluate the energy on discretized paths , we use gaussian quadrature on each interval between subsequent knots .the resulting discretized energy @xmath62 is minimized over all discrete paths with fixed initial and final control points @xmath58 and @xmath59 .this is a finite - dimensional , nonlinear , unconstrained .all smooth paths violating this condition have infinite energy because of the completeness of the metric . by approximation ,spline paths violating this condition have very high energy and will , in practice , be avoided during optimization . ]optimization problem .we solved this problem using either matlab s ` fminunc ` function with gradients computed by finite differences , or using ampl @xcite and the ipopt solver @xcite with gradients computed by automatic differentiation . by classical approximation results@xcite , the discrete energy of discrete paths converges to the continuous energy of smooth paths as the number of control points tends to infinity .hence we expect to obtain close approximations of geodesics .establishing rigorous convergence results for our discretization will be the subject of future work .the karcher mean @xmath63 of a set @xmath64 of curves is the minimizer of @xmath65 it can be calculated by a gradient descent on @xmath66 . letting @xmath67 denote the riemannian logarithm , the gradient of @xmath68 with respect to @xmath16 is @xcite @xmath69 to calculate the karcher mean by gradient descent, one has to repeatedly solve the geodesic equation .to this aim , we use the time - discrete variational geodesic calculus @xcite . given three curves @xmath70 ,one defines the discrete energy @xmath71 a 3-tuple @xmath72 is a discrete geodesic if it is a minimizer of the discrete energy @xmath73 with fixed endpoints @xmath74 .the discrete exponential map is defined as follows :@xmath75 , if @xmath72 is a discrete geodesic , in other words , if @xmath76 . to find @xmath77, we differentiate the discrete energy @xmath73 with respect to @xmath78 and solve the resulting system of nonlinear equations , @xmath79 we use the solver ` fsolve ` in matlab to solve this system of equations .this procedure is repeated depending on the desired time - resolution .while the convergence results in @xcite do not apply to our setting , we found good experimental agreement with the solutions of the boundary value problem .in our first example we want to characterize nuclear shape variation in hela cells .we use fluorescence microscope images of hela cell nuclei ( 87 images in total ) .the acquisition of the cells is described in @xcite .a similar study on this dataset was performed using the lddmm framework in @xcite ; applying intrinsically defined sobolev metrics to the same problem will provide a complimentary point of view . to extract the boundary of the nucleus , we apply a thresholding method @xcite to obtain a binary image and then fit using least squares a spline with @xmath80 and @xmath81 to the longest 4-connected component of the thresholded image .then we reparametrize the boundary to approximately constant speed .the remaining degree of freedom is the starting point of the parametrization ; when computing minimizing geodesics , we minimize over this parameter as well .the shapes of cell nuclei are thus represented by curves modulo translations , rotations , and constant shifts of the parametrisation , i.e. the two curves @xmath6 and @xmath82 are considered equivalent .examples of the extracted curves are depicted in fig .[ hela_fig_1 ] . [ cols="^,^ " , ] the parameters @xmath83 , @xmath84 , and @xmath21 of the riemannian metric are chosen as follows : we compute the average @xmath22- , @xmath23- and @xmath0-contributions @xmath85 , @xmath86 , @xmath87 to the energy of linear paths between each pair of curves in the dataset .as the @xmath22- and @xmath0-contributions scale differently , we rescale all curves such that @xmath88. then we choose constants @xmath83 , @xmath84 , and @xmath21 such that @xmath89 resulting in an average length @xmath90 and @xmath91 , @xmath92 , and @xmath93 . ; the bold curve is the mean .one can see four different characteristic deformations of the cell : bending , stretching along the long axis , stretching along the short axis , and a combination of stretching with partial bending.,scaledwidth=75.0% ] the average shape of the nucleus can be captured by the karcher mean @xmath63 . to compute the karcher mean of the 87 nuclei , we use a conjugate gradient method on the riemannian manifold of curves , as implemented in the manopt library @xcite , to solve the minimization problem .we obtain convergence in @xmath94 steps ; the final value of the objective function is @xmath95 and the norm of the gradient is @xmath96 .the mean shape can be seen in fig .[ hela_fig_1 ] . having computed the mean @xmath63 , we represent each nuclear shape @xmath97 by the initial velocity @xmath98 of the minimal geodesic connecting @xmath99 and @xmath97 .we perform principal component analysis with respect to the inner product @xmath100 on the set of initial velocities @xmath101 .the first four eigenvalues are @xmath102 and @xmath103 , and they explain @xmath104 , and @xmath105 of the total variance .geodesics from the mean in the directions of the first four principal directions can be seen in fig .[ hela_fig_2 ] .[ hela_fig_5 ] shows the data projected to the subspace spanned by the first two eigenvectors .finally , we sample from a normal distribution with the same covariance matrix as the data and project the sampled velocities @xmath106 back to the space of curves using the exponential map @xmath107 ; some examples can be seen in fig .[ hela_fig_1 ] . in our second applicationwe study curves that are obtained from images of the cardiac cycle .more precisely , we consider a sequence of 30 cardiac images , taken at equispaced time points along the cardiac cycle . each image is projected to a barycentric subspace of dimension two , yielding a closed curve in the two - dimensional space of barycentriccoordinates . after normalizing the coordinates (3 ) we obtain a closed , plane curve with the curve parameter representing time to which we can apply the methods presented in sect . [ math ] .details regarding the acquisition and projection of the images can be found in @xcite ; barycentric subspaces on manifolds are described in @xcite .the data consists of 10 cardiac cycles of patients with tetralogy of fallot and 9 patients from a control group .each cardiac cycle is originally represented by three - dimensional homogeneous coordinates @xmath108 , sampled at 30 time points .we project the homogeneous coordinates onto the plane @xmath109 and choose a two - dimensional coordinate system for this plane .then we use spline interpolation with degree @xmath110 and @xmath111 control points to reconstruct the planar curves from the data points ; see fig .[ fig : splinefit ] . with @xmath111 control points is used.,title="fig : " ] with @xmath111 control pointsis used.,title="fig : " ] with @xmath111 control points is used.,title="fig : " ] the parameters @xmath83 , @xmath84 , and @xmath21 in the metric are chosen similarly to sect .[ hela ] ; however , the scale of the curves is not changed and we use equal weighting between the @xmath22- , @xmath23- and @xmath0-parts of the average energy for linear paths .this leads to parameters @xmath112 , @xmath113 , and @xmath114 . to seeif the metric structure derived from the sobolev metric enables us to distinguish between diseased patients and the control group , we compute all 171 pairwise distances between the 19 curves ; this takes about 15 minutes on a 2 ghz single core processor .multi - dimensional scaling of the distance matrix shows that the metric separates healthy and diseased patients quite well ( fig .[ distancematrix]a ) .indeed , a cluster analysis based on the distance matrix recovers exactly with exception of one outlier ( patient 4 ) the subgroups of healthy and diseased patients ( fig .[ distancematrix]b ) .the karcher means of the healthy and diseased subgroups as well as of the entire population are depicted in fig .[ cardiackarcher ] .the mean was computed using a gradient descent method as described in sect .[ karcher ] with a threshold of @xmath115 for the norm of the gradient .the average distance from the mean for the diseased group is @xmath116 with a variance of @xmath117 , and for the control group the distance is @xmath118 with a variance of @xmath119 . to investigate the variability of the observed data , we performe principal component analysis on the initial velocities of the minimal geodesics connecting curves to the respective means ( c.f .[ hela ] ) .[ cardiackarcher2 ] shows the initial velocities projected to the subspace spanned by the first two principal directions . within the healthy and sick subgroups ,less then @xmath120 of the principal components are needed to explain @xmath121 of the shape variation .if , in contrast , principal components are analyzed for the entire dataset based on the global karcher mean , then @xmath122 of the principal components are needed to explain @xmath121 of the shape variation .in this article we numerically solved the initial and boundary value problems for geodesics on the space of regular curves under second order sobolev metrics .we analyzed two medical datasets using our approach .in future work we plan to prove rigorous convergence results for our discretizations , further investigate the impact of the constants in the metric , and treat unparametrized curves .michael v. boland and robert f. murphy .a neural network classifier capable of recognizing the patterns of all major subcellular structures in fluorescence microscope images of hela cells ., 17(12):12131223 , 2001 .kristin mcleod , maxime sermesant , philipp beerbaum , and xavier pennec .patio - temporal tensor decomposition of a polyaffine motion model for a better analysis of pathological left ventricular dynamics ., 2015 .gustavo k. rohde , wei wang , tao peng , and robert f. murphy .deformation - based nonlinear dimension reduction : applications to nuclear morphometry . in _5th ieee int .symposium on biomedical imaging : from nano to macro _ ,pages 500503 , 2008 .catalina tobon - gomez , mathieu de craene , kristin mcleod , lennart tautz , wenzhe shi , anja hennemuth , adityo prakosa , hengui wang , gerald carr - white , sergio kapetanakis , albert lutz , vernon rasche , tobias schaeffter , constantin butakoff , oskar friman , tommaso mansi , maxime sermesant , xiahai zhuang , sbastien ourselin , hans otto peitgen , xavier pennec , reza razavi , daniel rueckert , alejandro f. frangi , and kawal rhode . ., 17(6):632648 , 2013 .

<S> in the recent years , riemannian shape analysis of curves and surfaces has found several applications in medical image analysis . in this paper </S><S> we present a numerical discretization of second order sobolev metrics on the space of regular curves in euclidean space . </S><S> this class of metrics has several desirable mathematical properties . </S><S> we propose numerical solutions for the initial and boundary value problems of finding geodesics . </S><S> these two methods are combined in a riemannian gradient - based optimization scheme to compute the karcher mean . </S><S> we apply this to a study of the shape variation in hela cell nuclei and cycles of cardiac deformations , by computing means and principal modes of variations . </S>

1112.0203

this paper deals with the existence of minimizers for an increasing functional of the first @xmath0 eigenvalues of the dirichlet laplacian , among ( quasi-)open sets in @xmath5 with unit measure .this kind of spectral minimization problems is well - studied and natural in many situations : for instance , one can be interested in minimizing a certain eigenvalue , or a linear combination of eigenvalues , or a product of eigenvalues , and so on .however , despite the big interest on the question , only very little is known in the general situation when the problem is stated in the whole @xmath1 .the reason is basically the lack of compactness for generic sequences of open sets ; in fact , given a sequence of sets of unit measure , it is not easy to understand whether or not it converges to a limit set in a suitable sense .the first existence result ( ( * ? ? ?* theorem 2.5 ) , see also ( * ? ? ?* thorme 4.7.6 ) ) , which is now classical , has been found by buttazzo and dal maso in the context of bounded sets .[ bdm ] let @xmath6 be a bounded open set , and let @xmath7 be a sequence of ( quasi-)open sets of unit measure .then , there exists a subsequence @xmath8 and a ( quasi-)open set @xmath9 of unit measure such that @xmath10 for every @xmath11 . as an immediate consequence , for every l.s.c .functional @xmath12 increasing in each variable , there exists @xmath13 minimizing the value of @xmath14 among the ( quasi-)open sets of unit volume contained in @xmath15 .thanks to a concentration - compactness argument ( see for instance ( * ? ? ?* theorem 5.3.1 ) ) , it is now well - known that the boundedness assumption ( @xmath16 for every @xmath17 ) in the theorem above can be replaced by the following slightly weaker assumption : for every @xmath18 there exists @xmath19 such that every set @xmath20 is contained in a cube of side @xmath21 up to a volume at most @xmath22 .the results described above make it simple to study spectral minimization problems inside an ambient space which is essentially bounded . on the other hand ,very little is known concerning general problems in @xmath1 , because minimizing sequences , in principle , could have a significant portion of volume moving at infinity .more precisely , the existence of a minimizer for @xmath23 and @xmath24 is clear , since it is well - known that a ball of unit volume minimizes @xmath23 , while two disjoint balls of half volume minimize @xmath24 : these two classical results are usually referred to as faber krahn inequality and krahn szeg inequality respectively . instead, the existence of a minimizer for @xmath25 has been proved only in 2000 by bucur and henrot ( see @xcite ) , but it is still presently not known which set is the minimizer ( and this is a major open problem ) . for other open problems and partial results see @xcite . in this paper , we show that every increasing functional of the first @xmath0 eigenvalues admits a minimizer , which is in turn a bounded set . in other words , we prove that the buttazzo dal maso theorem [ bdm ] is true in the whole @xmath1 , with no need of the _ a priori _ boundedness assumption .the strategy of the proof consists in taking a generic open set @xmath26 of unit volume , and showing that there exists a modified open set @xmath27 which is uniformly bounded and which has all the first @xmath0 eigenvalues lower than those of @xmath28 . roughly speaking ,the basic idea why this works is that , if a set of unit volume has huge diameter , then there must be some very thin sections . and in turn , this is against the smallness of the rayleigh quotients of the eigenfunctions , since by definition they vanish on the boundary . as an immediate consequence of the above boundedness claim , there will clearly exist a uniformly bounded minimizing sequence , and therefore theorem will follow by theorem [ bdm ] . during our construction , we will need the following result , simple but of indipendent interest .it is interesting to note that , in the particular case when @xmath29 , theorem is a consequence of the stronger result by ashbaugh and benguria @xcite , which states that not only the ratio @xmath30 is bounded , but also that the balls maximize this ratio .notice that one can not apply theorem to study this problem , since the functional @xmath31 is not increasing in the first variable .notice also that theorem is stated for sets of any volume ; in fact , since by trivial rescaling one finds that for every @xmath32 and every @xmath11 one has @xmath33 the ratio @xmath31 does not change by rescaling .the plan of the paper is the following . in section [ ssec2 ] below we give a brief list of the notation that will be used through the paper , and some preliminaries .then , in section [ sec2 ] we show theorem , while in section [ sec3 ] we show theorem ._ * _ note : _ * dorin bucur recently announced the article in preparation @xcite , where he gives a proof of the existence of a minimizer for the @xmath0-th eigenvalue of the dirichlet laplacian in @xmath1 ( which corresponds , in the language of our theorem , to the particular case when @xmath34 is the projection on the last variable ) . _ throughout this paper , the ambient space will be @xmath1 for some given @xmath35 .the generic point of @xmath1 will be denoted by @xmath36 , or sometimes as @xmath37 , while the generic open set will be @xmath38 .the eigenvalues of the dirichlet laplacian on @xmath28 will be denoted by @xmath39 , while @xmath40 will be a corresponding sequence of orthogonal eigenfunctions , always normalized to have unit @xmath41 norm . for any function @xmath42 and any set @xmath43, we will consider the rayleigh quotient @xmath44 hence , in particular @xmath45 for every @xmath11 .the following is a very useful characterization of the eigenvalues , which we will need later , and whose proof can be found for instance in @xcite .it is often referred to as _ min - max principle_. [ chareigen ] let @xmath38 be an open set . then for every @xmath46 one has @xmath47 where the minimum is taken among all the @xmath48dimensional subspaces @xmath49 of @xmath50 . in our result, we need to use quasi - open sets .however , the construction of this paper only deals with open sets , and actually no knowledge of quasi - open sets is needed ; therefore , we only recall here the very basic definition , addressing the interested reader for instance to @xcite for a complete tractation .let @xmath28 be an open set , and @xmath51 a compactly supported subset .the _ capacity of @xmath52 in @xmath28 _ is defined as @xmath53 let then @xmath54 be a bounded set , and let @xmath28 be an open set such that @xmath55 .the set @xmath13 is said _ quasi - open _if for every @xmath18 there exists an open set @xmath56 such that @xmath57 .this definition does _ not _ depend on the choice of @xmath28 .finally , a generic set @xmath54 is said _ quasi - open _ if the intersection of @xmath13 with any ball of @xmath1 is a quasi - open bounded set .it is well - known that the notion of eigenvalues can be extended to the realm of quasi - open sets . an important feature of the quasi - open sets is given by the theorem [ bdm ] by buttazzo and dal maso ; namely , that any bounded sequence of open ( or quasi - open ) sets converges up to a subsequence to a quasi - open set , in the sense that all the eigenvalues converge . through the paper ,@xmath58 will be a fixed integer , and @xmath59 will be a l.s.c . function , increasing in each variable . for the sake of brevity ,we will often write @xmath60 in place of @xmath61 ; hence , the goal of this paper can be rephrased by saying that we seek for a quasi - open set @xmath13 minimizing @xmath62 . by density , it is clear that the infimum of @xmath34 among the quasi - open sets equals the infimum among the open ( or even smooth ) sets .the letter @xmath63 will be always used to denote a big geometric constant , possibly increasing from line to line ; the constant @xmath63 will always depend _ only _ on @xmath2 and on @xmath0 ( sometimes , possibly also on some constant @xmath64 , which in turn will eventually be chosen only depending on @xmath2 and @xmath0 ) , thus _ not _ on the particular choice of @xmath34 , and _ not _ on the set @xmath28 . sometimes , we will label the constants in our results as @xmath65 for successive reference .for any @xmath66 , we will define @xmath67 notice that @xmath68 and @xmath69 are subsets of @xmath1 , while @xmath70 is a subset of @xmath71 .figure [ leftright ] shows an example of a generic set @xmath28 with @xmath72 and @xmath70 . on the other hand ,given @xmath73 and @xmath74 , we define the level @xmath75 and the width @xmath76 as @xmath77 observe that one surely has @xmath78 whenever @xmath79 , as well as @xmath80 if @xmath81 .finally , given any set @xmath26 , we define its @xmath82-dimensional projections for @xmath83 as @xmath84in this section we present the proof of theorem .as already anticipated in the introduction , our strategy basically consists in showing that to minimize @xmath34 it is enough to concentrate on uniformly bounded sets .more precisely , we will show the following result .[ allhere ] for every @xmath85 there exists a constant @xmath86 , such that the following holds .if @xmath38 is an open set of unit volume and with @xmath87 , there exists another open set @xmath27 , still of unit volume but contained in a cube of side @xmath21 , and with @xmath88 for every @xmath89 .let us immediately see how theorem follows from this proposition , then the rest of the section will be devoted to show the proposition .let us take a minimizing sequence of open sets @xmath90 for problem ( [ problema ] ) .fix a generic @xmath91 , and assume for a moment that @xmath92 , being @xmath93 the ball of unit volume and @xmath94 the constant of theorem .if it is so , then by theorem one has @xmath95 , thus for every @xmath89 it is @xmath96 , hence @xmath97 , being @xmath34 increasing in each variable .thanks to this observation , it is admissible to assume that @xmath98 for every @xmath17 . by proposition [ allhere ] , then , there exists another sequence @xmath99 , made by open sets of unit volume contained in a cube of side @xmath21 , with @xmath100 for every @xmath89 and every @xmath91 . again by the assumption that @xmath34 is increasing in each variable , we derive that also @xmath99 is a minimizing sequence for ( [ problema ] ). we can then apply theorem [ bdm ] to find a quasi - open set @xmath13 , still contained in the cube of side @xmath21 , and such that , up to extract a subsequence of @xmath99 , one has @xmath101 for every @xmath89 . by the lower semi - continuity of @xmath34, we derive that @xmath13 is a minimizer for @xmath34 , thus the proof is concluded .the rest of this section is devoted to show proposition [ allhere ] . for the ease of presentation, we divide the construction in three subsections . in the first one we obtain the boundedness of the `` tails '' ( lemma [ lemmatail ] ) , while in the second one we consider the internal part ( lemma [ lemmainterior ] ) .then , in the last subsection we put everything together to give the proof of proposition [ allhere ] .this subsection is devoted to show that , under the assumptions of proposition [ allhere ] , we can reduce to the case when the `` tails '' of @xmath28 are bounded .more precisely , we fix once for all a small positive number @xmath102 in such a way that @xmath103 being @xmath104 the ball of unit volume in @xmath1 .we aim to show the following result .[ lemmatail ] for every @xmath85 there exist @xmath105 and @xmath106 such that , for any open set @xmath38 of unit volume and with @xmath87 , there exists another open set @xmath107 , still of unit volume , such that @xmath88 for every @xmath89 , and that for every @xmath108 @xmath109 the claim of the lemma , roughly speaking , says that it is always possible to assume that the `` tail '' of @xmath28 , i.e. , the set @xmath110 of volume @xmath111 , has horizontal projection of length at most @xmath112 .more precisely , condition ( [ estimatetail ] ) says that one can modify @xmath28 in such a way that the tail is uniformly horizontally bounded , while condition ( [ estimatediam ] ) says that this modification does not excessively worsen the remaining part of the set @xmath28 , nor its extension in the @xmath113 non - horizontal directions . to prove the lemma , we start setting for brevity @xmath114 , and for every @xmath115 we define @xmath116 observe that @xmath117 moreover , we let as usual @xmath118 be an orthonormal set of eigenfunctions with unit @xmath41 norm and corresponding to the first @xmath0 eigenvalues of @xmath28 .we define then also , for every @xmath89 and every @xmath115 , @xmath119 which makes sense since every @xmath120 is smooth .it is convenient to give the further notation @xmath121 and in analogy with ( [ int1 ] ) we also set @xmath122 applying the faber krahn inequality in @xmath71 to the set @xmath70 , and using ( [ rescaling ] ) on @xmath71 , we know that @xmath123 calling @xmath124 the unit ball in @xmath71 . as a trivial consequence , we can estimate @xmath125 in terms of @xmath22 and @xmath126 : in fact , noticing that @xmath127 and writing @xmath128 , we have @xmath129 we can now present two estimates which assure that @xmath120 and @xmath130 can not be too big in @xmath131 . [ primastima ] under the assumptions of lemma [ lemmatail ] , for every @xmath132 and @xmath115 the following inequalities hold : @xmath133 for some @xmath134 .let us fix @xmath115 .consider the set @xmath135 obtained by the union of @xmath131 and its reflection with respect to the plane @xmath136 , and call @xmath137 the function obtained by reflecting @xmath120 .calling @xmath104 the unit ball in @xmath1 , we find then @xmath138 by the symmetry of @xmath139 , and using again ( [ rescaling ] ) .this estimate gives @xmath140 which in particular , being @xmath141 and recalling ( [ defmh ] ) , implies @xmath142 on the other hand , recalling that @xmath143 , by schwarz inequality and using ( [ muest ] ) we have @xmath144 it is now easy to obtain ( [ udu- ] ) combining ( [ eq1 ] ) and ( [ eq2 ] ) .in fact , by inserting the latter into the first , we find @xmath145 which by ( [ defmh ] ) again yields @xmath146 the left estimate in ( [ udu- ] ) is then obtained . to get the right one, one has then just to insert ( [ leftudu- ] ) into ( [ eq2 ] ) .let us go further into our construction , giving some definitions .for any @xmath115 and @xmath147 , we define the cylinder @xmath148 , shown in figure [ taglio ] , as @xmath149 where for any @xmath115 we set @xmath150 we let also @xmath151 , and we introduce @xmath152 as @xmath153 \bal\frac{x - t+\sigma}{\sigma}\,u_i(t , y)\eal & \hbox{if $ ( x , y)\in q(t)$}\ , .\end{array } \right.\ ] ] the fact that @xmath154 vanishes on @xmath155 is obvious ; moreover , @xmath156 on @xmath157 , while on @xmath148 one has @xmath158 a simple calculation allows us to estimate the integrals of @xmath154 and @xmath159 on @xmath148 .[ lemmatest ] for every @xmath115 and @xmath89 , one has @xmath160 for a suitable constant @xmath161 .thanks to ( [ estdut ] ) , and using also ( [ muest ] ) and ( [ defsigma ] ) , one obtains the first estimate in ( [ newtest ] ) since @xmath162 on the other hand , the second estimate in ( [ newtest ] ) follows , also again using ( [ muest ] ) and ( [ defsigma ] ) , by @xmath163 another simple but useful estimate concerns the rayleigh quotients of the functions @xmath154 on the sets @xmath164 and the integral of the products @xmath165 .[ noth ] for every @xmath115 and @xmath89 , one has @xmath166 moreover , for every @xmath167 , one has @xmath168 recalling that @xmath169 , making use of ( [ estray2 ] ) and ( [ newtest ] ) and arguing as in ( [ eq2 ] ) , we obtain @xmath170 hence ( [ est1a ] ) is proved . on the other hand , recall that @xmath120 and @xmath171 are orthogonal on @xmath172 both in @xmath41 and in @xmath173 sense by definition ,hence by using ( [ udu- ] ) and ( [ newtest ] ) we find @xmath174 in the very same way , concerning @xmath159 and @xmath175 , we have @xmath176 adding up the last two estimates yields ( [ est1b ] ) . in order to prove lemma [ lemmatail ], we need to compare the eigenvalues of @xmath28 and those of @xmath164 ; this can be done by means of the min - max principle , theorem [ chareigen ] , which relates the eigenvalues with the rayleigh quotients of @xmath173 functions .[ lemmanu ] there exist a small constant @xmath177 and a constant @xmath178 such that , if @xmath179 for every @xmath89 , then @xmath180 we aim to use the characterization given by theorem [ chareigen ] .to do so , for every @xmath181 we define @xmath49 as the linear subspace of @xmath182 spanned by the functions @xmath154 with @xmath183 .first of all , we give the + * claim a. * for every @xmath184 , the space @xmath49 has dimension @xmath185 .+ suppose that the claim is not true .then , there should exist some @xmath186 and some coefficients @xmath187 for @xmath188 with all @xmath189 and @xmath190 notice now that by ( [ estray2 ] ) we know that @xmath191 , hence also by ( [ est1b ] ) we deduce @xmath192 where the last inequality is true provided that @xmath193 is chosen small enough .the absurd shows the validity of claim a. we can now show ( [ est2new ] ) : to do so , pick a generic function @xmath194 , which can be written ( up to a rescaling ) as @xmath195 , where @xmath196 .we need to evaluate @xmath197 : we start by noticing that @xmath198 where the last inequality comes by ( [ est1b ] ) . if @xmath199 is small enough, then @xmath200 hence by the choice of @xmath187 and by ( [ estray2 ] ) the denominator in the last fraction of ( [ jts ] ) is bigger than @xmath201 ( in particular , it is strictly positive ) . as a consequence , recalling also that by ( [ est1a ] ) one has for every @xmath183 @xmath202 from ( [ jts ] ) we deduce @xmath203 ( keep in mind that the constant @xmath204 may increase from line to line ) .the validity of ( [ est2new ] ) is then now an immediate consequence of theorem [ chareigen ] and of claim a. we can now enter in the central part of our construction .basically , we aim to show that either @xmath28 already satisfies the requirements of lemma [ lemmatail ] , or some @xmath164 does it , up to a rescaling .to do so , we need another definition , namely , for every @xmath115 we define the rescaled set @xmath205 so that @xmath206 .we can now show the following result .[ threeconditions ] let @xmath28 be as in the assumptions of lemma [ lemmatail ] , and let @xmath115. there exists @xmath207 such that exactly one of the three following conditions hold : 1 .@xmath208 ; 2 .\(1 ) does not hold and @xmath209 ; 3 .\(1 ) and ( 2 ) do not hold and for every @xmath89 , one has @xmath210 .in particular , if condition ( 3 ) holds for @xmath211 and @xmath212 , then for every @xmath89 one has @xmath213 , being @xmath214 .if ( 1 ) holds true , there is of course nothing to prove .otherwise , it is possible to apply lemma [ lemmanu ] , hence we have @xmath215 for every @xmath89 . by ( [ rescaling ] ) andthe fact that @xmath206 , we know that @xmath216 by construction , @xmath217 hence the above estimates and ( [ putinto ] ) lead to @xmath218 this allows us to conclude .in fact , defining @xmath219 , if @xmath209 , then condition ( 2 ) holds true .otherwise , ( [ muchhere ] ) directly implies that @xmath220 .finally , assume that condition ( 3 ) holds and @xmath212 : in this case , ( [ muchhere ] ) directly implies @xmath221 where @xmath222 and the last inequality is true up to decrease @xmath223 ( notice that _ decreasing _ the value of the constant @xmath223 of lemma [ lemmanu ] does not change the value of @xmath224 ) .we are finally in position to give the proof of lemma [ lemmatail ] .let us start defining @xmath225 with the usual convention that , if condition ( 3 ) is false for every @xmath115 , then @xmath226 . we introduce now the following subsets of @xmath227 , @xmath228 and we further subdivide them as @xmath229 we aim to show that both @xmath13 and @xmath93 are uniformly bounded .concerning @xmath230 , observe that @xmath231 so that @xmath232 . concerning @xmath233 , in the same way and also recalling that @xmath234 for every @xmath235 , we have @xmath236 so that @xmath237 . summarizing , we have proved that @xmath238 let us then pass to the set @xmath239 . to deal with it , we need a further subdivision , namely , we write @xmath240 , where @xmath241 keeping in mind ( [ int1 ] ) , we know that @xmath242 is an increasing function , and that for a.e. @xmath243 one has @xmath244 .moreover , for every @xmath245 one has by construction that @xmath246 as a consequence , for every @xmath247 one has @xmath248 this readily implies @xmath249 which in turn gives @xmath250 finally , we deduce @xmath251 notice that basically our argument consisted in using the fact that in @xmath239 one has @xmath252 concerning @xmath253 , we can almost repeat the same argument : in fact , thanks to ( [ udu- ] ) , for every @xmath254 we have @xmath255 which is the perfect analogous of ( [ analof ] ) with @xmath256 and @xmath257 in place of @xmath22 and @xmath258 respectively . since as already observed @xmath259 , in analogy with ( [ defb1n ] ) we can define @xmath260 thus the very same argument which lead to ( [ estb1 ] ) now gives @xmath261 putting ( [ esta ] ) , ( [ estb1 ] ) and ( [ estb2 ] ) together , we find @xmath262 we will prove the validity of the lemma with the following choice of @xmath112 and @xmath263 , @xmath264 + 1}\,,\end{aligned}\ ] ] where @xmath265 and @xmath266 have been introduced in ( [ almostr_1 ] ) and in lemma [ threeconditions ] respectively . to obtain our proof , we will distinguish the possible cases for @xmath28 .if this case happens , then condition ( 3 ) of lemma [ threeconditions ] never holds true , i.e. , for every @xmath115 either condition ( 1 ) or ( 2 ) holds . recalling the definition of @xmath13 and @xmath93 and ( [ almostr_1 ] ) , we deduce that @xmath267 .therefore , the claim of lemma [ lemmatail ] is immediately obtained simply taking @xmath268 , since @xmath269 and @xmath270 . in this case , let us notice that it must be @xmath271 , hence @xmath272 and thus by ( [ almostr_1 ] ) @xmath273 .let us now pick some @xmath274 $ ] for which condition ( 3 ) holds , and define @xmath275 . by definition, @xmath276 has unit volume , and @xmath277 for every @xmath89 , being condition ( 3 ) true for @xmath278 .observe now that by definition for every @xmath108 one has @xmath279 , hence @xmath280 where we have used that @xmath281 .concerning the widths of @xmath276 and @xmath28 , we can start observing that @xmath282 moreover , since it is admissible to assume @xmath283 and then @xmath284 we have @xmath285 as a consequence , we evaluate @xmath286 thus we deduce , again recalling @xmath281 , that @xmath287 summarizing , we have found that @xmath288 as a consequence , the choice @xmath289 satisfies all the requirements of lemma [ lemmatail ] , except possibly condition ( [ estimatetail ] ) . to deal with this last condition ,we need to further subdivide this case . in this case, we can show that the choice @xmath289 actually works . as noticed above, we have only to prove the validity of ( [ estimatetail ] ) . to do so , we assume for simplicity that @xmath290 , which is clearly admissible by translation .hence , @xmath291 , and thus @xmath292 recalling that @xmath293 , we deduce that @xmath294 moreover , since @xmath295 , @xmath296 and this implies that @xmath297 .the inclusion ( [ finhe ] ) ensures then that @xmath298 , and then ( [ estimatetail ] ) holds true since @xmath299 .we have now to face the last possible case , namely , when @xmath300 is finite but @xmath301 . in this case , thanks to lemma [ threeconditions ] the estimates ( [ thistrue ] ) can be strengthened as @xmath302 concerning the validity of ( [ estimatetail ] ) , it does not follow by ( [ finhe ] ) because the assumption @xmath303 does not imply that @xmath297 .however , we can argue as follow : if ( [ estimatetail ] ) holds true for @xmath276 , then of course we are done by setting @xmath289 .otherwise , we apply the above construction to the set @xmath276 in place of @xmath28 : since @xmath276 does not satisfy ( [ estimatetail ] ) , then case i is impossible , thus we are in case ii and then by ( [ thistrue ] ) we find an open set @xmath304 of unit measure such that @xmath305 if @xmath276 is in case iia , then as before we are done with the choice of @xmath306 .otherwise , if we are in case iib , then ( [ thistrue2b ] ) becomes @xmath307 going on with the obvious iteration , if the proof has not been concluded after @xmath308 steps then we have found an open set @xmath309satisfying @xmath310 this is of course impossible if @xmath311 , being @xmath87 : as a consequence , our iteration must stop after less than @xmath312 steps , thus our thesis is concluded with our choice of @xmath263 .the goal of this subsection is to obtain a uniform bound also for the interior part of a set @xmath28 , in the sense of lemma [ lemmatail ] .most of the arguments of this case will be identical to those that we made for the tails in section [ sub1 ] , but some modifications are essential .in particular we give new definitions for @xmath313 @xmath126 , @xmath125 , @xmath164 and @xmath314 in order to mantain clear the analogy with what was done in section [ sub1 ] .the result that we are going to prove is the following .[ lemmainterior ] for every @xmath85 there exist @xmath315 and @xmath316 such that , for any open set @xmath38 of unit volume and with @xmath87 , and for any choice of @xmath317 , there exists another open set @xmath107 , still of unit volume , such that @xmath88 for every @xmath89 , and such that for every @xmath108 @xmath318 to start with , we give the analogous of the definitions ( [ int0 ] ) , ( [ int1 ] ) and ( [ int2 ] ) of section [ sub1 ] that we need now ; figure [ intern ] helps to visualize the new situation .more precisely , we set for brevity @xmath319 keep in mind that , since @xmath320 , then @xmath321 . for any @xmath322 , we define @xmath323 moreover , having fixed an orthonormal set @xmath324 of eigenfunctions with unit @xmath41 norm corresponding to the first @xmath0 eigenvalues of @xmath28 , for every @xmath89 and @xmath322 we define @xmath325 in analogy with ( [ defphi ] ) , we define again @xmath326 , and we set again @xmath327 our strategy to prove lemma [ lemmainterior ] is very similar to what we did to show lemma [ lemmatail ] ; in fact , basically the only difference is that to show the analogous of lemma [ primastima ] we can not rely on the symmetrization of @xmath328 .let us see how to overcome this difficulty .[ newass ] there exists a small constant @xmath177 such that , if @xmath28 and @xmath329 are as in the assumptions of lemma [ lemmainterior ] , and @xmath322 is such that @xmath330 , then for every @xmath89 one has @xmath331 consider the `` external cylinders '' @xmath332 where @xmath333 in perfect analogy with ( [ defcyl ] ) and ( [ defsigma ] ) . calling @xmath334 , we can extend ( [ utilde ] ) to obtain the following definition of @xmath335 , @xmath336 \bal\frac{x-(t_0-t-\sigma_1)}{\sigma_1}\,u_i(t_0-t , y)\eal\qquad & \hbox{if $ ( x , y)\in q_1$}\ , , \\[10pt ] \bal\frac{(t_0+t+\sigma_2)-x}{\sigma_2}\,u_i(t_0+t , y)\eal & \hbox{if $ ( x , y)\in q_2$}\ , .\end{array } \right.\ ] ] applying lemma [ lemmatest ] to the two cylinders @xmath337 and @xmath338 , and comparing the present definitions of @xmath22 and @xmath339 with those that we used in section [ sub1 ] , ( [ newtest ] ) gives us @xmath340 we can then obtain an estimate between @xmath341 and @xmath342 similar to ( [ eq1 ] ) , first noticing that @xmath343 and then deducing , recalling ( [ defmh ] ) and choosing @xmath223 small enough , @xmath344 observe that @xmath193 can be chosen so small that the last estimate implies @xmath345 we can also generalize ( [ eq2 ] ) ;in fact , since the very same argument used in ( [ muest ] ) again ensures that @xmath346 , we can obtain @xmath347 putting together ( [ wtebte ] ) and ( [ wtebte2 ] ) gives ( [ 2udu- ] ) .we need now to extend the result of lemma [ lemmanu ] to our new setting .to do so , going on in analogy with section [ sub1 ] , we give the following definition .let @xmath28 , @xmath329 and @xmath348 be as in the assumptions of lemma [ newass ] .consider the `` internal cylinders '' @xmath349 where @xmath333 and notice that by the assumption on @xmath350 and the fact that @xmath351 one has @xmath352 .the set @xmath164 is defined as @xmath353 see figure [ newomtt ] .notice that @xmath354 moreover , define again the rescaled set @xmath355 with this definition of the sets @xmath164 and with the obvious extension of ( [ utilde ] ) in order to define @xmath356 , we can now literally repeat the proofs of lemmas [ lemmatest ] , [ noth ] , [ lemmanu ] and [ threeconditions ] , the unique difference being the substitution of @xmath148 with @xmath357 , and of @xmath70 with @xmath358 .we obtain then the following result , which holds up to possibly decrease the constant @xmath193 of lemma [ newass ] .[ threebis ] let @xmath28 be as in the assumptions of lemma [ lemmainterior ] , and let @xmath359 .there exists @xmath360 such that exactly one of the three following conditions hold : 1 .@xmath208 ; 2 .\(1 ) does not hold and @xmath361 ; 3 .\(1 ) and ( 2 ) do not hold and for every @xmath89 , one has @xmath210 . in particular , if condition ( 3 ) holds for @xmath211 and @xmath362 , then for every @xmath89 one has @xmath213 , being @xmath214 .we can now conclude this section by presenting the proof of lemma [ lemmainterior ] , which will be a minor modification of the proof of lemma [ lemmatail ] .first of all , we want to show that it is admissible to assume @xmath363 in fact , suppose that it is not so , and let @xmath364 . then , @xmath28 is the disjoint union of @xmath365 and @xmath366 , and it does not intersect the whole strip @xmath367 .therefore , replacing @xmath28 with @xmath368 , that is , moving closer the two disjoint parts of @xmath28 , does not change any of the eigenvalues of @xmath28 and is clearly admissible for the proof of the lemma ; moreover , the property ( [ admissible ] ) of course holds true for this new set .hence , from now on we directly assume that ( [ admissible ] ) holds true for @xmath28 .define now @xmath300 analogously to ( [ defhatt ] ) by setting @xmath369 with the convention that , if condition ( 3 ) is false for every @xmath370 then @xmath371 .again we define @xmath13 and @xmath93 as @xmath372 the same argument of the proof of lemma [ lemmatail ] gives then @xmath373 and we are going to show the thesis with the choice @xmath374 + 1}\,,\end{aligned}\ ] ] being @xmath375 the constant of lemma [ threebis ] . we can again subdivide the possible cases for @xmath28 . in this case , by ( [ admissible ] ) one has that every @xmath376 belongs either to @xmath13 or to @xmath93 , thus by ( [ almostr_2 ] ) @xmath377 ; as a consequence , the choice @xmath378 satisfies the requirements of the lemma . indeed ,while the right condition of ( [ estimateinterior ] ) is obviously true , the left one follows just noticing that @xmath379 in this case , again by ( [ admissible ] ) we know that @xmath380 contains the whole segment @xmath227 , thus @xmath381 by ( [ almostr_2 ] ) . if we choose @xmath382 for which condition ( 3 ) holds and define @xmath275 , we know by construction that @xmath383 and @xmath277 for every @xmath89 . as in the proof of lemma [ lemmatail ] , the fact that @xmath384 ensures that @xmath385 for each @xmath108 . on the other hand , concerning the width of @xmath276 , recalling the definition of @xmath386 and observing that @xmath387 one finds @xmath388 summarizing , we have found that @xmath389 to conclude the inspection of the validity of ( [ estimateinterior ] ) , we will again need to consider separately two subcases . in this case , we can quickly observe that @xmath390 and , together with ( [ thistrueint ] ) , this concludes the proof of ( [ estimateinterior ] ) and of the lemma with the choice @xmath289 .let us conclude with this last case .by lemma [ threebis ] , in place of ( [ thistrueint ] ) we have then @xmath391 which still does not guarantee the validity of ( [ estimateinterior ] ) .however , we can apply the construction above to @xmath276 : if @xmath276 is in case i , then the choice @xmath289 concludes the proof ; otherwise , there exists an open set @xmath304 of unit measure satisfying @xmath392 if @xmath276 is in case iia then ( [ ifcaseiia ] ) gives @xmath393 so the choice @xmath306 concludes the proof .instead , if @xmath276 is in case iib then the first inequality of ( [ thisu_2 ] ) becomes @xmath394 for every @xmath89 .the obvious iteration ensures us that , if the proof has not been obtained after @xmath395 steps , then there must be some open set @xmath309satisfying @xmath396 since this is not possible for @xmath397 , the iteration must stop at some @xmath398 $ ] , and thus we conclude the proof thanks to the choice of @xmath399. we are finally in position to give the proof of proposition [ allhere ] , which is now a simple consequence of lemma [ lemmatail ] and lemma [ lemmainterior ] .let us pick a generic open set @xmath28 with @xmath87 .applying lemma [ lemmatail ] to @xmath28 , we find a set @xmath400 with @xmath401 for every @xmath402 .then , we apply lemma [ lemmainterior ] to @xmath400 with @xmath403 finding @xmath404 which satisfies @xmath405 iterating , for any @xmath406 such that @xmath407 we apply lemma [ lemmainterior ] to @xmath408 with @xmath409 finding @xmath410 such that @xmath411 possibly applying a last time lemma [ lemmainterior ] with @xmath412 , we have then found an open set @xmath413 satisfying @xmath414 - 1}-1}{\gamma_2 - 1}+ \gamma_2^{[1/\mh]-1 } r_1 , & & \frac{{\rm diam } \big(\pi_p(e)\big)}{{\rm diam } \big(\pi_p(\om)\big ) } \leq \gamma_2^{[1/\mh]-1}\gamma_1 .\end{aligned}\ ] ] calling @xmath415 the set obtained by reflecting @xmath413 with respect to the plane @xmath416 , the above estimates become @xmath417 - 1}-1}{\gamma_2 - 1}+ \gamma_2^{[1/\mh]-1 } r_1\ , , & & \frac{{\rm diam } \big(\pi_p(e')\big)}{{\rm diam } \big(\pi_p(\om)\big ) } \leq \gamma_2^{[1/\mh]-1}\gamma_1 \,,\end{aligned}\ ] ] so that applying once again lemma [ lemmatail ] to @xmath415 we find a set @xmath418 satisfying @xmath419 having set @xmath420 - 1}-1}{\gamma_2 - 1}+ \gamma_1\gamma_2^{[1/\mh]-1 } r_1\ , , & & \gamma_3 : = \gamma_2^{[1/\mh]-1}\gamma_1 ^ 2\,.\end{aligned}\ ] ] we can now repeat the whole construction , using as starting set @xmath418 in place of @xmath28 , and using the second coordinate in place of the first one .we will end up with a set @xmath421 with @xmath422 and such that for every @xmath423 it is @xmath424 .in particular , choosing @xmath425 we discover that @xmath426 .we have now to iterate also this argument : for any @xmath427 we repeat the above construction starting from @xmath428 and using the @xmath185-th coordinate in the whole procedure , obtaining a set @xmath429 which satisfies @xmath430 the thesis is then finally obtained by defining @xmath431 , being @xmath432 .this last section is devoted to the proof of theorem . for the ease of presentation, we will begin with a couple of technical lemmas , then passing to the proof of the theorem .it is important to observe that the core of our construction , namely lemma [ startingcase ] , is in fact an easy consequence of the well - known result by ashbaugh benguria in @xcite , which states that the ratio @xmath30 is bounded .however , we prefer to give a formal proof of this lemma to keep the present paper self - contained . [ smallrho ] for every @xmath433 $ ] and @xmath85 there exists @xmath434 such that the following holds .let @xmath435 with @xmath436 then for every cube @xmath437 with half - side @xmath438 one has @xmath439 suppose that the claim is not true .then there exists a sequence @xmath440 satisfying @xmath441 being @xmath442^n$ ] the cube of half - side @xmath443 centered at the origin . by definition ,this sequence is bounded in @xmath444 , hence up to a subsequence we have that @xmath445 weakly converges to some function @xmath435 .in particular , for any @xmath18 , @xmath445 strongly converges to @xmath446 in @xmath447 , so that , thanks to ( [ asspt ] ) , one has @xmath448 .since this is absurd , the claim follows .[ startingcase ] for every @xmath85 there exists @xmath449 such that , if @xmath28 is an open subset of @xmath1 with @xmath450 , then there are two disjoint open subsets @xmath451 of @xmath28 with @xmath452 for @xmath453 .we start applying lemma [ smallrho ] with @xmath64 and with @xmath454 , thus getting a positive number @xmath438 .let then @xmath38 be an open set with @xmath450 , and let @xmath455 be a first eigenfunction of @xmath28 with unit @xmath41 norm .extending @xmath446 by @xmath456 outside @xmath28 , we have then by definition @xmath457 let now @xmath458 be identified by @xmath459 we claim that it is possible to assume @xmath460 in fact ,if it is not so , this means that there is a vertical stripe of width @xmath461 out of which the squared @xmath41 norm of @xmath446 is less than @xmath462 ( by `` vertical '' we mean orthogonal to @xmath463 ) .if this happens for every direction @xmath464 , the intersection of the corresponding stripes is a square of half - side @xmath438 out of which the squared @xmath41 norm of @xmath446 is less than @xmath465 .since this is in contradiction with lemma [ smallrho ] , we obtain the validity of ( [ admass ] ) , up to a rotation .let us now call @xmath466 , define @xmath467 and @xmath468 , and let @xmath469 be defined as @xmath470 \bal\frac{t -x}{\rho}\eal\ , u(x , y ) \qquad & \hbox{for $ t - \rho\leq x \leq t$}\ , .\\ \end{array}\right.\ ] ] since @xmath455 , it is clear that @xmath471 . moreover ,writing @xmath472 , one has @xmath473 for every @xmath474 with @xmath475 . as a consequence ,minding ( [ firstprop ] ) one gets @xmath476 on the other hand , recalling ( [ admass ] ) and ( [ deft- ] ) it is @xmath477 putting together ( [ estimdu ] ) and ( [ estimu ] ) one immediately obtains @xmath478 finally , we can set @xmath479 : since we have shown that @xmath480 , and since by symmetry it is also @xmath481 , the thesis follows .let us start by setting @xmath487 , and then , applying lemma [ startingcase ] , we let recursively @xmath488 for every @xmath489 . finally , we define @xmath490 , where @xmath185 is the smallest natural number such that @xmath491 .we will show the claim of the theorem with such constant @xmath492 .to do so , we pick any open set @xmath28 with @xmath493 . applying lemma [ startingcase ] to @xmath28 with constant @xmath494, we find two disjoint open sets @xmath495 with @xmath496 for @xmath497 .then , we can apply lemma [ startingcase ] to @xmath498 and @xmath499 with constant @xmath500 , finding four disjoint subsets @xmath501 of @xmath28 , each of them with first eigenvalue smaller than @xmath502 .continuing with the obvious induction , we end up with @xmath503 disjoint open subsets of @xmath28 , say @xmath504 for @xmath505 , having @xmath506 for each @xmath507 . to get ( [ tofinish ] ) , for every @xmath509 let @xmath120 be a first eigenfunction of @xmath504 , again extended by @xmath456 on @xmath510 , so that @xmath511 and then @xmath512 .observe that the functions @xmath120 are mutually orthogonal ( both in the @xmath41 and in the @xmath513 sense ) by construction , since they are supported on disjoint sets .hence , the linear subspace @xmath514 of @xmath50 spanned by the functions @xmath120 for @xmath505 is @xmath503-dimensional .thanks to theorem [ chareigen ] , to prove ( [ tofinish ] ) it is enough to show that @xmath515 for every @xmath516 . andin fact , writing the generic function @xmath516 as @xmath517 , by the orthogonality of the different @xmath120 one has clearly @xmath518 as noticed before , this gives the validity of ( [ tofinish ] ) , hence the proof is concluded .first of all notice that , by density , it is admissible to consider only the case of the open sets .we apply lemma [ exthm ] with @xmath519 , so defining @xmath520 .we will prove theorem with such @xmath94 .let @xmath38 be an open set , and apply the rescaling formula ( [ rescaling ] ) choosing @xmath521 , thus getting @xmath522 .by lemma [ exthm ] , we derive @xmath523 , and then by ( [ rescaling ] ) again we find @xmath524 , thus the proof is concluded .

<S> in this paper we show that any increasing functional of the first @xmath0 eigenvalues of the dirichlet laplacian admits a ( quasi-)open minimizer among the subsets of @xmath1 of unit measure . </S><S> in particular , there exists such a minimizer which is bounded , where the bound depends on @xmath0 and @xmath2 , but not on the functional . in the meantime , we show that the ratio @xmath3 is uniformly bounded for sets @xmath4 . </S>

1607.05600

neutron stars are ideal testing grounds for studying strong gravitational effects as predicted by general relativity .the dragging of inertial frames is one such important relativistic effect in compact stars . as the surrounding spacetime of a rotating compact objectis dragged due to its rotation , the nearby spacetime is basically distorted due to this effect .general relativity predicts that the curvature of spacetime is produced not only by the distribution of mass - energy but also by its motion .similarly , in electromagnetism , the fields are produced not only by charge distributions but also by currents , i.e. , as the moving charge produces the magnetic field .considering this analogy , this effect is also called the _gravitomagnetic effect_. spacetime is distorted by its mass as well as by its rotation .if a small part of the matter starts to rotate , the inertial frame must be dragged along slightly .the small rotation of the earth also drags the local inertial frame along it .lense and thirring @xcite first discovered that the rotation of a massive body could distort the spacetime , making the spin of nearby test gyroscope precess @xcite .this does not happen in newtonian gravity for which the gravitational field of a body depends only on its mass , not on its rotation .this dragging frequency depends on the angular momentum of the rotating object as well as its compactness , hence its value is large for rapidly rotating neutron stars .thus the strong gravitational field in the vicinity of neutron stars may provide an opportunity to observe and identify frame dragging effects . as this effect decreases as the inverse cube of the distance from the source, the frame dragging rate is expected to be largest at the vicinity of the centre of a rotating neutron star and falls off with inverse cube law of distance toward the surface and the distance away from it in weak gravity regime . +lense - thirring ( lt ) precession is generally studied in the context of the accretion disc theory where the orbital plane precession frequency is calculated for a _ test particle _ orbiting in the equatorial plane relative to the asymptotic observer . in this article, we derive the lt precession frequency of a test gyroscope ( _ spinning test particle _ ) relative to the `` fixed stars '' ( copernican system ) which was first proposed in @xcite .he derived the lt precession frequency of a test gyroscope relative to the `` fixed stars '' for a rotating spacetime in the weak - field regime .a test gyro follows the gyroscope equation and not necessarily follows a geodesic path in the presence of gravitational fields .therefore , it is more general than the lt frequency of a test particle moving in a geodesic. it may be useful to mention here that chakraborty and majumdar derived @xcite the exact lt precession rate of a test gyro in kerr spacetime without invoking the weak field approximation and showed that the exact formulation reduced to the well - known lt frequency rate in weak gravity regime , derived in @xcite . in this article , we will deal with the exact lt precession rate of a _ spinning test particle _ due to the rotation of a pulsar . we should mention that based on the proposal by schiff , the gravity probe b experiment had been designed to measure the frame - dragging effect of a test gyroscope ( relative to the copernican system or `` fixed star '' hr8703 , also known as i m pegasi ) due to the rotation of the earth @xcite to very high accuracy .the former lageos satellite provided a low precision measurement of the lt precession of the orbit .however the prospect of measurement of lt frequency of a test gyroscope due to the rapidly rotating neutron stars is attractive , as the frame - dragging precession in neutron stars is larger than the precession due to oblateness .further , as the frame dragging effect is also found to occur inside rotating neutron stars and the magnitude of this effect depends on the density profile , it is an interesting prospect to try to constrain the equation of state exploiting this effect @xcite .the matter orbits in accretion disks can provide ways to probe the spacetime properties in the vicinity of neutron stars , including frame dragging effects .quasi - periodic oscillations ( qpos ) observed in x - ray bursts and diagnostics of x - ray spectra have been suggested as promising tools to investigate the motion of matter around neutron stars @xcite .recently , it has been predicted @xcite that frame - dragging can be generated by the electromagnetic fields . in this articlethey have basically calculated the _orbital plane precession frequency _ considering an _ orbiting test particle_. + using the slow - rotation approximation proposed in @xcite , the frame dragging effect inside a slowly rotating neutron star can be estimated . in this approximation , to second order in rotation frequencies , the structure of a star changes only by quadrupole terms and the equilibrium equations describing the structure reduce to a set of ordinary differential equations .the frame - dragging frequency in this scheme is only a function of the radial distance from the centre of the star . the precession frequency at the centreis then given by the frequency of rotation of the neutron star , and drops off monotonically towards the surface .in a recent investigation , the exact lt frequency of a test gyro was derived in @xcite .the frame dragging rate was found to be not only a function of the rotation frequency of the star but also of the other metric components . unlike in the slow rotation limit, it was found that the precession rate depends both on the radial distance ( @xmath0 ) and the colatitude ( @xmath1 ) of the gyroscope .thus , in rapidly rotating neutron stars , the precession rates are not found to be the same in the equatorial and polar directions . + it is well known that neutron stars are endowed with strong magnetic fields .the surface magnetic field value for normal pulsars , estimated from their spin down rates , points to values around @xmath2 g. there have been reports of discovery of high magnetic field neutron stars , both in isolated systems ( xdinss , rrats ) @xcite as well as in binaries ( sxts ) @xcite .the largest magnetic fields have been observed in anomalous x - ray pulsars ( axps ) and soft gamma - ray repeaters ( sgrs ) , commonly called magnetars .various observations , including direct detection of cyclotron features in the spectra @xcite confirm that these objects possess surface magnetic fields as large as @xmath3 g. however , magnetic fields in the interior of magnetars could be even larger . as it can not be directly measured ,the maximum interior magnetic field is usually estimated using the scalar virial theorem , which points towards a value of @xmath4 g. the presence of a strong magnetic field breaks of spherical symmetry of the star , resulting in a strong deformation of its shape from isotropy .the investigation of the effect of strong magnetic fields on the frame - dragging rate in magnetars is the aim of this paper .+ the scheme of the paper is as follows . in sec.[seceq ] of this paper , we define the basic equations to describe the exact frame dragging rate in rotating neutron stars in presence of an electromagnetic field .sec.[secnum ] explains the numerical scheme for solving the equations described in the previous section .the results obtained are discussed in sec.[secres ] and sec.[secnull ] is devoted to demonstrate a thought experiment to determine the magnetic field and the `` null region '' of a pulsar using lt precession .finally sec.[seccon ] closes the article with a summary and future outlook .to construct numerical models of rotating neutron stars endowed with a magnetic field within a framework of general relativity , we follow the approach of @xcite . with the assumption of a stationary , axisymmetric spacetime ,in which the matter part fulfils the circularity condition , the line element can be written as @xcite : @xmath5 where @xmath6 are functions of @xmath7 .the energy - momentum tensor in the presence of a magnetic field can be written as : @xmath8 here the first term represents the matter contribution ( perfect fluid ) , where @xmath9 is the energy density , @xmath10 is the pressure and @xmath11 is the fluid four velocity .the last term gives the electromagnetic field contribution to the energy - momentum tensor .it was shown in @xcite that the contributions of magnetization and the magnetic field dependence of the equation of state ( eos ) on the structure of strongly magnetised neutron stars are negligible , hence we disregard them in the calculations . + in the same way as in @xcite we consider here that the electromagnetic field originates from free currents , noted hereafter simply @xmath12 , which are _ a priori _ independent from the movements of inert mass ( with 4-velocity @xmath13 ) .this is a limiting assumption in our model , and one should in principle use a microscopic model to derive a distribution for the free currents , too .however , such a model would require a multi - fluid approach to model the movements of free protons and electrons , and we leave it for a future study .+ subject to the symmetries considered in our model , the definition of the electromagnetic field tensor @xmath14 determines the magnetic field configuration , either purely poloidal or purely toroidal .we choose a purely poloidal configuration .the electric and magnetic fields measured by the eulerian observer ( whose four - velocity is @xmath15 ) are then defined as @xmath16 and @xmath17 , with @xmath18 is the levi - civita tensor associated with the metric ( [ eq : metric ] ) .the non - zero components read ( see @xcite for detailed calculations ) : @xmath19 in presence of an electromagnetic field , the equilibrium configurations for rotating neutron stars are determined by solving the coupled maxwell - einstein equations . in the case of rigid rotation , this is equivalent to solving a first integral of the equation of fluid stationary motion @xcite : @xmath20 taking into account the lorentz force exerted by the electromagnetic field on the medium , via the electromagnetic term : @xmath21 where @xmath22 is an arbitrary current function relating the components of the electric current to the electromagnetic potential @xmath23 . in the above equation , @xmath24 is the fluid log - enthalpy , @xmath25 is the gravitational potential related to the metric component @xmath26 ( @xmath27 ) , while @xmath28 is the lorentz factor relating the eulerian observer to the fluid comoving observer . + we recapitulate here the basic equations for the exact frame dragging rate in rotating neutron stars ( see @xcite for a detailed derivation ) using the metric eq .( [ eq : metric ] ) . in the orthonormalcoordinate basis , the exact lt precession rate of a test gyro relative to the copernican system ( or `` fixed stars '' ) for the rotating neutron star with electromagnetic field can be written as : @xmath29\hat{\th}\right .\label{ltnp } \nonumber \\ & & \left.-[n^2(bn^{\phi}_{,\th}\sin\th+2n^{\phi } b_{,\th}\sin\th + 2n^{\phi } b\cos\th ) + ( n^{\phi})^2 r^2b^3n^{\phi}_{,\th}\sin^3\th-2n^{\phi } bnn_{,\th}\sin\th]\hat{r}\right]\nonumber \\\end{aligned}\ ] ] and the modulus of eq .( [ eq : omega_lt ] ) is @xmath30 ^ 2\right .\nonumber \\ & & \left.+[n^2(bn^{\phi}_{,\th}\sin\th+2n^{\phi } b_{,\th}\sin\th + 2n^{\phi } b\cos\th ) + ( n^{\phi})^2 r^2b^3n^{\phi}_{,\th}\sin^3\th-2n^{\phi } bnn_{,\th}\sin\th]^2\right]^\f{1}{2 } \nonumber \\\end{aligned}\ ] ] where ` @xmath31 ' and ` @xmath32 ' represent the derivatives with respect to @xmath0 and @xmath1 respectively .the einstein - maxwell equations described above are solved in the 3 + 1 formalism , where they reduce to poisson - like partial differential equations .these equations are then solved numerically using spectral scheme within the numerical library lorene @xcite .the input parameters for the model are an eos , the current function @xmath22 , the rotation frequency @xmath33 and the central log - enthalpy @xmath34 .the electromagnetic field originates from the free currents , which are controlled via the current function @xmath22 .varying the input current function @xmath22 allows us to control the intensity of the magnetic field @xmath35 or the magnetic moment @xmath36 . on the other hand , varying the central log - enthalpy permits variation of the mass of the neutron star ( gravitational mass @xmath37 or baryon mass @xmath38 ) . + ) for a non - rotating pulsar having a baryonic mass @xmath38 of 1.475 m@xmath39 . ] using the numerical scheme described above , we compute models of fully relativistic neutron stars for a given magnetic field configuration . we consider a poloidal magnetic field configuration .we choose as an example a particular pulsar psr j0737 - 3039 whose mass ( @xmath40 ) and spin ( @xmath41 s@xmath42 ) are well known .the exact frame - dragging rate for this pulsar was calculated for the non - magnetic case in @xcite .we now study the effect of magnetic field on its frame dragging rate . just as in the case of a rotating starit is not the rotation frequency but the moment of inertia which is a constant of motion , for strongly magnetised neutron stars the magnetic field is not constant of motion but rather the magnetic moment . to give a clearer correspondence between the polar magnetic field of the pulsar and its magnetic moment , we plot in fig .[ fig : bpmm ] the relation between the two for a given baryonic mass of 1.475 @xmath43 , which corresponds to a gravitational mass of 1.337 @xmath43 for the zero magnetic field case .we can see from the figure that the relation is almost linear , with the largest value @xmath44 for the magnetic moment corresponding to a magnetic field value of @xmath45 g. + to show the effect of strong magnetic field on the structure of a neutron star , we demonstrate an equilibrium neutron star configuration corresponding to a magnetic moment of @xmath46 a m@xmath47 . in fig .[ fig : xzcut ] , the magnetic field lines ( top panel ) and enthalpy isocontours ( bottom panel ) in the meridional plane for psr j0737 - 3039 , having a gravitational mass of 1.337 m@xmath39 and angular frequency of 276.8 s@xmath42 , for a magnetic moment of @xmath44 have been displayed .it is evident from the figure , that under the influence of the strong magnetic field , the stellar surface has been highly deformed from spherical symmetry .+ to estimate the deformation of the structure due to the magnetic field , we compared the values of the @xmath48 and the quadrupole moment of the star due to different constant magnetic moments .the values are displayed in table [ tab : deform ] .it is evident from the table that with increase in magnetic moment , the flattening increases i.e. , the ratio of polar radius with respect to the equatorial radius decreases from 1 ( for a spherical star ) . the equatorial circumferential radius as well as the quadrupole moment increase with increase in magnetic field . [ cols= " <, > , > , > " , ] [ tab : deform ]it was shown in @xcite that the behaviour of the lt - precession rate for different eoss is qualitatively similar . therefore in this study , we employ only the apr @xcite eos .the maximum masses as well as the corresponding radii of static neutron stars as well as those rotating at kepler frequency were displayed already in @xcite , hence we do not repeat this study here .our aim is to perform a study of the effect of magnetic field on the frame dragging rate @xmath49 for neutron stars rotating at sub - kepler frequencies .+ we construct equilibrium configurations of neutron stars for constant values of magnetic moment @xmath36 = @xmath50 . for a given magnetic moment , we calculate the value of the frame - dragging precession rate inside ( @xmath51 ) as well as outside the pulsar ( @xmath52 ) , where @xmath53 is the distance of the surface of the pulsar from its centre ( at the equator @xmath54 ; @xmath55 is the equatorial coordinate radius ) . + we should clarify that the lt frequency can be calculated using eq .( [ eq : omega_lt_mod ] ) both inside and outside of a pulsar .as all the parameters ( @xmath56 ) in eq .( [ eq : omega_lt_mod ] ) are functions of @xmath0 and @xmath1 , the equation is automatically modified by changing the values of @xmath0 and @xmath1 . inside the pulsar ( @xmath51 )all of the above mentioned parameters take the values corresponding to the solution of the einstein equation with the energy - momentum tensor mentioned in eq .( [ emtensor ] ) . outside the pulsar , ( @xmath57 ) eq .( [ emtensor ] ) reduces to the case with @xmath58 in absence of matter and eq.([eq : metric ] ) describes the electrovacuum solution of the einstein equation .thus , the frame - dragging formula ( eq . ( [ eq : omega_lt_mod ] ) ) also gets modified automatically with the corresponding values of @xmath56 and the lt precession rates ( for a given angle ) are continuous from the centre to the asymptotically large distances as well as on the surface of the pulsar .+ one must note that the effect of strong electromagnetic field is two fold : + ( i ) firstly , there is the effect of electromagnetic field directly on the matter and vacuum solutions , given by eq .( [ eq : omega_lt_mod ] ) .+ ( ii ) secondly , strong magnetic field causes a deformation of the stellar surface .thus it affects the radial distance at which the energy density falls to zero , and consequently the lt frequency also shifts at the radial distance at which the matter solution matches the electrovacuum solution . + in panel ( a ) and panel ( b ) of fig .[ fig : in ] , we display the frame dragging rate as a function of radial distance ( @xmath59 ) for different angular values from 0 ( along polar direction ) to 90 ( equatorial direction ) for two different constant magnetic moments .the results are qualitatively the same as obtained in @xcite : the frame dragging frequency has a smooth behaviour along the pole from the centre to the surface of the star but the behaviour is quite different along the equator , which is already been explained in @xcite and @xcite . along the equatorial directionthe frame dragging frequency goes through a `` dip '' or local minimum at a certain angle , labeled as critical angle " . in this case, the critical angle occurs around @xmath60 . for magnetic moments@xmath61 a m@xmath47 , the results are indistinguishable from the non - magnetic case .+ panels ( a ) to ( j ) of fig .[ diffmag ] reveal that for different values of magnetic moment , depending on the angular direction , the frame dragging rate differs inside the pulsar .interestingly , the frame dragging rates along the pole ( @xmath62 ) decrease with increasing the magnetic moments and the reverse effect is seen along the equator ( @xmath63 ) , i.e. , an increase in the frame dragging rates with increasing magnetic moments .therefore , there should exist a ` crossover angle ' where the effect of the magnetic field on the frame - dragging effect will be ` null ' .magnetic field of the pulsar does not affect the gravitomagnetic effect in this region , which we can call the `` null region '' . to find the null regionwe show the evolution of the frame - dragging effect for various magnetic moments ( starting from @xmath64 a.m@xmath47 ) from panel ( a ) to panel ( j ) ,i.e. , from the pole to the equator for every @xmath65 interval .the solid red line represents the zero magnetic moment and black dashed line represents the highest magnetic moment considered . in panel( a ) and panel ( j ) , the deviation between these two lines is maximum which means that frame - dragging rate is affected maximally by the magnetic moment of the pulsar in these two regions .the evolution from panel ( a ) to ( e ) shows that the distance between these two lines decreases if we proceed from the pole @xmath66 to an angle @xmath67 .deviation is minimum ( @xmath68 ) around @xmath69 ( panel ( e ) ) .this shows that magnetism has no effect on frame - dragging at this particular angle .if we proceed further from panel ( f ) to panel ( j ) , we see that the deviation between those two lines ( solid red and dashed black ) again increases but in this case it is in the reverse direction .thus the maximum effect of magnetic field is along the polar direction ( reducing the frame - dragging frequency ) and in the equatorial direction ( increasing the frame - dragging frequency ) , while for intermediate angles its effect goes through a minimum , having practically no influence on the frame - dragging frequency around @xmath70 .the frame dragging rates are plotted for radial distances extending outside the pulsar ( @xmath71 ) along polar ( panel a ) , @xmath72 ( panel b ) and equatorial directions ( panel c ) in fig .( [ out ] ) , respectively .this shows that the null region also exists outside the pulsar , at angle of about @xmath73 .this is expected as the effect of the magnetic field exists not only inside but also outside ( in principle @xmath74 ) the pulsar and the frame - dragging should therefore also extend upto @xmath74 .the change in the null angle inside and outside can be understood from the fact that , as the energy density vanishes at the surface of the pulsar , the frame - dragging frequency is no longer determined by the matter solution where @xmath75 is given by eq .( [ emtensor ] ) but by the electrovacuum solution with @xmath58 in @xmath75 .the curves for frame - dragging frequencies converge asymptotically at large distances ( @xmath76 ) .however , there is an interplay between the influence of the magnetic field on both the matter and vacuum solutions , as well as on the radial distance at which there is a change from the matter to vacuum solutions .+ comparing the energy density profiles along different angles for different magnetic moments , we find that for small values of magnetic moments ( e.g. @xmath77 a.m@xmath47 ) the energy density profiles remain the same along all angular directions ( see fig .[ fig : enprof_mm1e31 ] ) .this is expected as for small magnetic fields the stellar surface remains spherically symmetric .however for large magnetic moments ( e.g. @xmath78 a.m@xmath47 ) , the energy density profiles vary along different angles ( see fig . [fig : enprof_mm1e32 ] ) .this can be understood from the fact that the stellar surface is distorted from spherical symmetry due to the strong magnetic fields . for a poloidal configuration, there is a polar flattening and equatorial bulging , which means the energy density goes to zero at smaller radial distances along the polar direction and at larger radial distances along the equatorial direction , defining the distorted stellar surface .it is found that the energy density profile ( and hence the stellar surface ) is least affected along @xmath73 ( comparing figs .[ fig : enprof_mm1e31 ] and [ fig : enprof_mm1e32 ] ) . ) along polar direction ( red solid line ) , @xmath70 ( magenta dashed line ) , at @xmath73 ( green dotted line ) and equatorial direction ( brown dot - dashed line ) for a magnetic moment of @xmath77 a.m@xmath47.,width=192 ] ) along polar direction ( red solid line ) , @xmath70 ( magenta dashed line ) , at @xmath73 ( green dotted line ) and equatorial direction ( brown dot - dashed line ) for a magnetic moment of @xmath78 a.m@xmath47.,width=192 ] to determine the influence of the value of pulsar rotation on the position of the null region , we repeat the numerical calculations for @xmath33 = 0 , 100 , 1000 @xmath79 . as the rotation frequency increases to large values .the surface of the star is also deformed to a non - spherical shape .however , we found that when the rotation frequency is zero , the lt frequency vanishes , no matter what the value of the magnetic field .this implies that there is no contribution of free precession due to the oblateness caused by the magnetic field , and that the lt effect arises purely due to the rotation of the spacetime .the lt frequency is non - zero only in presence of rotation , and its magnitude is affected by the presence of the magnetic field .however , we have also observed that the value of the rotation frequency of the pulsar has no influence on the value of the angle at which the magnetic field ceases to affect the frame dragging rate . for the interior region , the null regionis found at @xmath80 while for the exterior region it is found at @xmath81 , as obtained for @xmath33 = 276.8 @xmath79 for the psr j0737 - 3039 . + here, we should emphasize that any kind of magnetic field is not directly responsible for the frame - dragging effect .this effect solely depends on the rotation of the spacetime ( there is also an exception where lt effect is not related to the rotating spacetime but some `` rotational sense '' was involved ( @xcite ) there ) .thus , if a magnetic field exists in a non - rotating spacetime , the spacetime will not be able to show the frame - dragging effect .if the spacetime starts to rotate , gravitomagnetism is induced in this spacetime and the magnetic field comes into play to show its effect on frame - dragging . in an analogy with electromagnetism , we can say that as a non - magnetic material is not affected by the magnetic field of any magnet , the non - rotating spacetime is also not affected by the magnetic field of that spacetime in the case of frame - dragging effect .it is evident form eq.([eq : metric ] ) that @xmath82 vanishes for a non - rotating spacetime ( @xmath83 ) and therefore the gravitomagnetic effect vanishes ( @xmath84 ) which is also evident from eq.([eq : omega_lt ] ) , though the magnetic field is non - zero ( see eq.([emtensor ] ) ) .the _ tussle between magnetic field and gravitomagnetic effect is only possible if they are both present in a particular spacetime_. if any one of them is absent , the tussle will also be absent .that is why it is meaningless to isolate the role of magnetic moment on lt effect from that due to rotation only , but the deviation of the lt effect due to the presence of the magnetic field can easily be deduced from the panels ( a)-(j ) of fig .[ diffmag ] for various magnetic moment values at various angles .in the previous section , we have discussed the lt effect of psr j0737 - 3039 and without making any realistic experiment we have predicted that the null region will be at @xmath80 ( inside the pulsar ) and @xmath81 ( outside the pulsar ) where magnetic field has no effect on gravitomagnetism , whatever magnetic field value psr j0737 - 3039 possesses .the crossover angle of the null region between @xmath70 to @xmath73 occurs on the surface of the pulsar at @xmath85 which is evident from panels ( e)-(f ) of fig .[ diffmag ] and panel ( b ) of fig .it is also evident from fig .[ fig : enprof_mm1e31 ] that the distance of the surface of the pulsar is @xmath86 at @xmath70 angle .+ in an axisymmetric spacetime, we can imagine that the null angle @xmath87 creates a hypothetical 3-d hollow ` cone ' inside the pulsar and @xmath88 outside , which is depicted in fig .[ fignl ] . therefore ,if a test gyro moves along the surface of the cone , lt frequency of the gyro will not be affected by the magnetic field of the pulsar whatever magnetic field it holds .if the gyro moves along any direction inside the cone , its lt frequency varies inversely with the magnetic moment and outside the cone the frequency will be directly proportional to the magnetic moment .+ now , using the concept of gravity probe b experiment we can demonstrate a thought experiment which can be performed by a test gyro .we can make it move towards the strong gravitational field of a pulsar along the pole or equator or whichever angle we want . for a particular angle outside a unknown pulsar , say , @xmath89 , we make the gyro move from _ infinity _ towards the pulsar and in principle we could measure the exact frame - dragging rate .we could then calculate the difference between this measured frame dragging rate and the theoretically calculated frame dragging rate for zero magnetic field .the difference in the frame dragging along this given angular direction will be determined by the magnetic field of the pulsar .thus , in principle we can calculate the intensity of the magnetic field of the pulsar in this way .now , we can move the gyro along different angles @xmath90 from @xmath91 to @xmath92 and the angle at which the measured frame dragging frequency is exactly equal to the calculated lt precession frequency for zero magnetic field , is the null angle ( @xmath93 ) outside the pulsar . as an example , it is @xmath73 in fig .[ fignl ] . in this way, we can determine the null angle for the effect of magnetic field on lt precession for the particular pulsar .after determining the @xmath94 we can move the gyro further toward the unknown pulsar .now , very close to the pulsar surface ( point a , b , c or d in fig .[ fignl ] ) we will see that the frame - dragging rate is starting to get affected by the magnetic field suddenly . now , we have to follow the similar procedure ( which we have followed outside the pulsar ) at the surface of the neutron star to find the null angle @xmath95 ( @xmath70 in fig .[ fignl ] ) inside the pulsar .azimuthal section of a _ theoretically predicted _ null region / surface of the pulsar psr j0737 - 3039 around the symmetry axis @xmath96 or the rotation axis of the pulsar .@xmath97 is the equatorial plane and solid black line represents the surface of the pulsar .. the boundary of the 3-d hypothetical hollow ` cone ' has been drawn along @xmath99 and @xmath100 around the symmetry axis . in principle , the cone is extended upto @xmath74 , which is shown by four arrows but outside i.e. , @xmath101 ( as @xmath102 ) , the null region appears at @xmath81 instead of @xmath70 which is evident from panels ( e)-(f ) of fig .[ diffmag ] and panel ( b ) of fig .if the gyro moves along the surface of this 3-d hypothetical ` cone ' , the effect of _ magnetic field _ of the pulsar will not be exerted on its frame - dragging frequency or gravitomagnetic effect . inside this cone, @xmath49 decreases with increasing magnetic moment and it shows completely reverse effect outside this cone ( this figure is not drawn in scale).,width=288 ]in this work , we extended the recent calculation @xcite for exact lt precession rates inside and outside neutron stars to include effects of electromagnetic field . this should be particularly interesting for the case of magnetars , which are strongly magnetised neutron stars , and for the observations of quasi - periodic oscillations ( qpos ) in magnetars .+ we employed a well established numerical scheme @xcite to obtain equilibrium models of fully relativistic strongly magnetised neutron stars .we computed the frame dragging rates inside and outside the stellar surface for such models .we obtained qualitatively similar peculiar features for the frame dragging rates along polar and equatorial directions as previously discussed for non - magnetic calculations @xcite , with the effect of the electromagnetic field evident on the frame dragging rate inside as well as outside the star .this could be potentially interesting for magnetars for constraining the magnetic field from a measurement of the lt frequency .as an example for psr j0737 - 3039 , we found that the maximum effect of magnetic field appears along the polar direction ( decreasing the frame - dragging frequency ) and in the equatorial direction ( increasing the frame - dragging frequency ) , while for intermediate angles its effect goes through a minimum , having practically no influence on the frame - dragging frequency around @xmath70 inside and @xmath73 outside the pulsar .we showed that this can be attributed to the twofold effect of a strong magnetic field , on the lt frequency as well as on the breaking of spherical symmetry of the neutron star .we also devised a thought - experiment to determine the magnetic field of the neutron star as well as the null angle using a measurement of its lt frequency .+ measurements of geodetic and frame dragging effects by dedicated missions such as lageos and gravity probe b @xcite have permitted us to impose constraints and better understand gravito - magnetic aspects of theories of gravity .there is a need to expand the sensitivity of future gravitational tests to neutron stars to understand phenomena such as qpos and lt precession .one of us ( cc ) gladly acknowledge kamakshya p. modak and prasanta char for their valuable suggestions .dc is grateful to jrome novak for his guidance regarding the numerical computation .

<S> rotating bodies in general relativity produce frame dragging ( or lense - thirring effect ) , also known as the _ gravitomagnetic effect _ in analogy with classical electromagnetism . in this work , </S><S> we study the effect of strong magnetic fields in neutron stars on the gravitomagnetic effect , which is produced as a result of its rotation . </S><S> we show that the magnetic field has a non - negligible impact on frame dragging . </S><S> the maximum effect of the magnetic field appears along the polar direction , where the lense - thirring frequency decreases with increase in magnetic field , and along the equatorial direction , where its magnitude increases . for intermediate angles , the effect of the magnetic field decreases , and goes through a minimum for a particular angular value at which magnetic field has no effect on gravitomagnetism . beyond </S><S> that particular angle gravitomagnetic effect increases with increasing magnetic field . </S><S> we try to identify this `` null region '' for the case of magnetars , both inside and outside , as a function of the magnetic field . </S><S> we demonstrate a thought experiment to find the magnetic field and the null region of a particular pulsar using the frame dragging effect . * </S><S> key words :* rotating neutron stars , frame dragging , magnetic field </S>

hep-lat0605012

in two seminal papers svetitsky and yaffe have tentatively linked the finite - temperature phase transitions in ` hot ' gauge theories to the simpler order - disorder phase transitions of spin models @xcite . in the general casetheir conjecture may be stated as follows : the effective theory describing finite - temperature yang - mills theory with gauge group @xmath3 in @xmath4 dimensions is a spin model in @xmath5 dimensions with a global symmetry group given by the centre @xmath6 of the gauge group .a somewhat stronger version of the conjecture can be formulated if the phase transitions in questions are of second order . in this caseyang - mills theory and spin model fall into the same universality class and the critical exponents coincide .this has been convincingly demonstrated for @xmath7 @xcite in @xmath8 .however , continuous phase transitions in hot gauge theories are not generic and hence the universality statement is almost empty at least in 3 + 1 dimensions @xcite . on the other hand , the more general version of the statement has already been used by svetitsky and yaffe to argue that the phase transition for @xmath9 yang - mills theory must be first order for @xmath8 as there is no @xmath0 rg fixed point in this case . since thenthis has been firmly established by a number of lattice calculations @xcite .the conjecture implies that the effective theories may be formulated as ` polyakov loop models ' @xcite . for @xmath3this means that the macroscopic dynamical variables have to reflect the complete gauge invariant information contained in the ( untraced ) polyakov loop , which in lattice notation is given by @xmath10 \equiv \prod_{t=1}^{n_{\tau } } u_{t,{\mbox{\scriptsize\boldmath$x$}};0 } \ ; .\ ] ] this is a temporal holonomy winding around the compact euclidean time direction of extent @xmath11 . for gauge groups with nontrivial centre the traced polyakov loop , @xmath12 with the trace being taken in the defining representation for the ease of later notation . ]serves as an order parameter for the deconfinement phase transition .the phase transition goes along with spontaneous breaking of the centre symmetry resulting from _ non - periodic _ gauge transformations under which @xmath13 the deconfined broken - symmetry phase at sufficiently large wilson coupling @xmath14 is characterised by @xmath15 ( see @xcite for a review ) .recently it has been found that gauge groups with trivial centre may also lead to a deconfinement transition depending on the _ size _ of the gauge group @xcite . at this pointthe choice of dynamical variables needs to be addressed . under_ periodic _ gauge transformations @xmath16 the holonomy ( [ eq : utrpoldef ] ) transforms as @xmath17 which leaves its eigenvalues and , as a consequence , its trace ( [ eq : trpoldef ] ) invariant .the eigenvalues are permuted arbitrarily by gauge transformations corresponding to weyl reflections @xcite .this invariance is taken into account by constructing symmetric polynomials in the @xmath18 eigenvalues . from unimodularity the product of the eigenvaluesis @xmath19 and there are only @xmath20 independent polynomials , for example the traces @xmath21 for @xmath22 .these in turn are in one - to - one correspondence with the characters of the @xmath23 fundamental representations ( see below ) .hence , for @xmath7 @xmath24 ( which is real ) is sufficient while @xmath9 requires @xmath24 and @xmath25 , the latter being a linear combination of @xmath26 and @xmath27 .only for @xmath28 traces of higher powers of @xmath29 are needed as independent dynamical variables @xcite .in two recent papers @xcite we have studied polyakov loop models on the lattice for the simplest non - abelian gauge group @xmath7 .the models have been derived using strong - coupling techniques at small wilson coupling @xmath30 and a newly developed inverse monte carlo ( imc ) method which works for arbitrary values of @xmath30 .the latter method allows for a mapping of yang - mills theory at a certain value of @xmath30 to any appropriately chosen polyakov loop model of the form @xmath31 where the summation is over nearest neighbours and group representations @xmath32 , @xmath33 .the group character @xmath34 is the trace of the polyakov loop in representation @xmath32 , @xmath35 \equiv { \text{tr}}_i \ , { \mathfrak{p}_{{\mbox{\footnotesize\boldmath$x$ } } } } \ ; .\ ] ] all characters @xmath34 are polynomials in the characters corresponding to the fundamental representations . for @xmath7the simplest model is of ising type , @xmath36 as first suggested in @xcite ( see also @xcite ) .the output of the imc routines are the @xmath30-dependent effective couplings @xmath37 . even without particular knowledge of theseone may study the models ( [ eq : polmodel ] ) in their own right as statistical field theories .the svetitsky - yaffe conjecture may then be utilised to deduce information about the yang - mills phase transition .for @xmath7 we have been able to show that the mean - field analysis of the effective models yields a surprisingly good agreement with the monte carlo analysis of both the model itself and the underlying yang - mills dynamics .it should be stressed at this point that the models ( [ eq : polmodel ] ) are not just simple scalar field theories with a linear target space as all characters ( [ eq : chars ] ) take values in a compact space parametrised by the @xmath38 fundamental characters .hence , in the lattice path integral each lattice site is endowed with the reduced haar measure on the gauge group rather than a lebesgue measure .this paper presents our first steps to generalise the results of @xcite to the gauge group @xmath9 .we label the characters by two integers @xmath39 and @xmath40 which count the numbers of fundamental and conjugate representations ( ` quarks ' and ` antiquarks ' in the @xmath9 flavour language ) required to construct the representation @xmath41 .equivalently , these integers characterise the horizontal extensions of @xmath9 young tableaux ( see fig . [fig : pq ] ) . character labels and @xmath9 young tableaux . ] under the @xmath0 centre transformations the characters transform according to the rule @xmath42 so that the most general centre - symmetric effective action with nearest - neighbour interaction may be written as @xmath43 = \sum_{\substack{{\langle}{\mbox{\scriptsize\boldmath$x$ } } { \mbox{\scriptsize\boldmath$y$}}{\rangle } , \ ,pq , \ , p'q ' \\p + p ' = q + q ' \ , \mathrm{mod } \ , 3 } } \lambda_{pq , p'q ' } \ , \big ( \chi_{pq } ( { \mbox{\boldmath$x$ } } ) \ , \chi_{p'q ' } ( { \mbox{\boldmath$y$ } } ) + \text{h.c . }\big ) \\ + \sum_{\substack{{\mbox{\scriptsize\boldmath$x$ } } , \ , pq \\\mathrm{mod } \ , 3 } } \lambda_{pq,00 } \ , \big ( \chi_{pq } ( { \mbox{\boldmath$x$ } } ) + \text{h.c .} \big ) \ ; .\end{gathered}\ ] ] this coincides with the ansatz suggested by dumitru et al.@xcite . the first sum in the effective action consists of hopping terms involving monomials of the form @xmath44 or @xmath45 ( and h.c . ) while the second sum is a ` potential ' term containing only powers @xmath46 ( and h.c . )localised at single sites .the remainder of the paper is organised as follows . in section [ sec: sce ] we confirm the ansatz ( [ eq : ssu3 ] ) by means of a strong coupling ( small-@xmath30 ) expansion for @xmath47 $ ] . for a restricted set of couplings ( and hence representations )we investigate the resulting @xmath0 models by minimising the classical action ( section [ sec : qca ] ) followed by an improved mean field analysis in section [ sec : mf ] . in agreement with the svetitsky - yaffe conjecturewe find a first order phase transition from the symmetric to a ferromagnetic phase .our improved mean field analysis already reveals an interesting phase structure with four different phases : two distinct ferromagnetic phases , one symmetric and one anti - ferromagnetic phase . besides the first order transitions we detect second order transitions from the symmetric to an anti - ferromagnetic and to a ferromagnetic phase .the continuous transition from the symmetric to the anti - ferromagnetic phase is to be expected since the models reduce to the @xmath48-state potts model for polyakov - loops having values in the centre of the gauge group .section [ sec : mcsim ] contains the results of our extensive monte carlo simulations performed on a linux cluster where we have implemented the powerful package jenlatt .similarly as for @xmath7 the mean field and numerical results are in surprisingly good agreement .this is presumably due to the existence of ( three ) tricritical points .depending on the order of the transition we localise the critical lines with either a metropolis , a multicanonical or a modified cluster algorithm .in addition we have checked that the critical exponents @xmath1 and @xmath2 at the second order transition to the anti - ferromagnetic phase agree with those of the @xmath48-state potts model in @xmath48 dimensions .finally , in section [ sec : so ] we wrap up with discussion and conclusions .in this section we briefly recapitulate the strong - coupling ( small-@xmath30 ) expansion for the @xmath9 wilson action at finite temperature @xcite .it is known that the leading order result ( @xmath49 ) stems from ladder diagrams that wind around the temporal lattice extension and corresponds to an ising type model analogous to ( [ eq : su2ising ] ) . by going beyond the leading order we will encounter higher group representations / characters andhence have an independent confirmation of the ansatz ( [ eq : ssu3 ] ) for the effective action .our starting point is the standard wilson action , @xmath50 where the summation over plaquettes contains both temporal and spatial links .the effective action @xmath51 $ ] is introduced as usual by inserting an appropriate ( group valued ) delta function in the path integral , @xmath52 } \\ & \equiv \int { \mathfrak{dp}}\ , e^{-{s_{\text{\tiny eff}}}[{\mathfrak{p } } ] } \ ; . \end{split}\ ] ] while it is not known how to perform the final integration analytically for the full wilson action one can straightforwardly integrate order by order in @xmath30 .thus we expand the boltzmann weight @xmath53 and integrate separately over temporal and spatial links , @xmath54 . adopting temporal gaugewe set all temporal links equal to unity apart from the links in the first timeslice which according to ( [ eq : utrpoldef ] ) may be identified with the polyakov loop . integrating out all spatial link variables we obtain the partition function @xmath55 \right ) \ ; , \end{split}\ ] ] where we have introduced the operators @xmath56 .the effective action may finally be written as @xmath57 where the coefficients @xmath58 are related to the operators @xmath59 via the linked - cluster theorem . in the remainder of this section we are going to determine the explicit form of the effective operators @xmath58 .we first rewrite the wilson - boltzmann weight ( [ eq : boltzmann1 ] ) as @xmath60 and expand the single - plaquette contribution in terms of @xmath9 characters , @xmath61 with @xmath62 ( see fig .[ fig : pq ] ) .all @xmath30 dependence now resides in the generalised fourier coefficients @xmath63 which accordingly may be further expanded , @xmath64 an explicit computation of the @xmath65 shows that these vanish whenever the representation labels become sufficiently large , namely if @xmath66 @xcite .this yields the important intermediate result that to any given order @xmath67 in the strong - coupling expansion only a _ finite _ number of characters contributes , @xmath68 the integrations over the spatial links are standard group integrals which can be found in the texts @xcite .the upshot is that only connected link arrangements ( ` polymers ' ) wrapping around the temporal extent of the lattice yield nonvanishing contributions . the leading term is a ladder diagram consisting of @xmath69 plaquetteseach of which contributes a factor of @xmath30 implying a total contribution of @xmath70 .the associated operator is explicitly found to be @xmath71 a typical operator of order @xmath72 is given by @xmath73 in terms of which the wilson - boltzmann weight ( [ eq : boltzmann2 ] ) becomes @xmath74 expanding this to next - to - leading order ( @xmath75 ) yields the simple expression @xmath76 note , however , that some care has to be taken in interpreting products such as @xmath77 which by also contain disconnected pieces . upon expanding the logarithm by means of the linked - cluster theorem we are led to keep only connected contributions of the form @xmath78 making use of the character reduction formula @xmath79 products of characters at the same site may be reduced to single characters . as a consequence ,the connected part of ( [ eq : s2n ] ) takes the explicit form @xmath80 for what follows it is useful to introduce the short - hand notation @xmath81 with the obvious term - by - term identifications as compared to ( [ eq : effactionlong ] ) .our conventions are such that all couplings in ( [ eq : effactionlong ] ) and ( [ eq : effactionshort ] ) are real functions of @xmath30 , the single leading one being @xmath82 ( as noted already in @xcite ) while the subleading ones are @xmath83 .it is straightforward to include higher - order terms the number of which increases rapidly . at order @xmath84 ,for instance , there are already 11 terms so that we refrain from going beyond next - to - leading order in @xmath30 . for later purposes it is useful to express the operators appearing in ( [ eq : effactionshort ] ) in terms of the fundamental loops @xmath85 and @xmath86 .octet and sextet characters ( @xmath87 and @xmath88 , respectively ) are eliminated via the standard reduction identities @xmath89 which are equivalent to the character relations ( recall that @xmath90 ) @xmath91 making use of the latter the different terms in ( [ eq : effactionshort ] ) become @xmath92 from these expressions it is obvious that each operator @xmath93 is manifestly real . under charge conjugation @xmath94 ,hence @xmath95 , whereupon all terms in @xmath96 are charge conjugation invariant as required @xcite .note that the orders included correspond to terms that are quadratic , cubic and quartic in @xmath85 and/or @xmath86 .higher powers will arise upon taking into account higher representations .thus , in this respect @xmath97 is somewhat singled out being of only cubic order .this will become important in a moment .it has already been pointed out by svetitsky and yaffe @xcite that effective actions with @xmath0 centre symmetry are closely related to the 3-state potts model which shows a first order phase transition from a symmetric to a ferromagnetic phase @xcite . to make the relation manifest we restrict the polyakov loop to the centre elements@xmath98 introduced in ( [ eq : centre ] ) .setting @xmath99 we find the general formula @xmath100 where @xmath101 denotes the dimension of the representation @xmath41 , @xmath102 applying this to the effective action in ( [ eq : effactionshort ] ) we find , up to an additive constant , @xmath103 = \lambda \sum _ { { \langle}{\mbox{\scriptsize\boldmath$x$ } } { \mbox{\scriptsize\boldmath$y$ } } { \rangle}}\cos \left ( \frac{2\pi}{3}(k_{{\mbox{\scriptsize\boldmath$x$ } } } - k_{{\mbox{\scriptsize\boldmath$y$ } } } ) \right ) \ ; , \quad k_{{\mbox{\scriptsize\boldmath$x$}}}\in\{0,1,2\}\ ; , \label{eq : pottsm}\ ] ] with effective coupling @xmath104 the action ( [ eq : pottsm ] ) is precisely the one of the 3-state potts model @xcite .we thus expect that our effective polyakov loop models will have a phase structure generalising the one of the 3-state potts model .the latter is known to have a ferromagnetic phase for large negative @xmath105 and an anti - ferromagnetic phase for large positive @xmath105 @xcite . as a preparation for the discussion of later sections it is hence useful to obtain some qualitative understanding of the phase structure of the polyakov loop models viewed as generalisations of the 3-state potts model . for the reason mentioned at the end of the previous section we choose as a minimal generalisation the following effective action , @xmath106 which in terms of the fundamental loops may be written explicitly as @xmath107 note that there are also quadratic contributions stemming from @xmath97 .the action ( [ eq:2cmodel ] ) is manifestly @xmath0 centre symmetric under @xmath108 .it is important to realise that ( [ eq:2cmodel ] ) differs from the standard lattice actions for scalar fields in several respects .first , the field @xmath24 is dimensionless , being the trace of a unitary matrix .this allows for the presence of cubic hopping terms connecting neighbouring sites .even more important is the fact that the target space of @xmath24 is compact . introducing the eigenvalues of @xmath109 via @xmath110 and writing @xmath111we find for the real and imaginary part of @xmath85 , @xmath112 the target space of @xmath85 may then be sketched in the complex @xmath85-plane ( see fig .[ fig : domain ] ) .the boundary corresponds to the points with @xmath113 , the singular ` corners ' being given by the three centre elements @xmath114 .target space of the polyakov loop @xmath85 in the complex @xmath85-plane .the corners represent the three centre elements .the intermediate points ( denoted anti - centre elements ) will also become relevant for the discussion of the phase structure . ]let us try to get some first rough idea of the phase structure associated with the two - coupling model ( [ eq:2cmodel ] ) in the @xmath115-@xmath116 plane by looking at the extrema of the classical action . if we vary the couplings these will trace out a certain ( possibly discontinuous ) trajectory in the target space given by the triangle of fig .[ fig : domain ] .as we argued earlier , for centre - valued polyakov loops the effective action ( [ eq:2cmodel ] ) reduces to the action of the potts model ( [ eq : pottsm ] ) with coupling @xmath117 .thus we expect a ferromagnetic phase ( f ) for large negative @xmath118 and an anti - ferromagnetic phase ( af ) for @xmath118 large and positive . in a region around the origin in the coupling plane entropydominates energy and we can not expect to actually obtain the correct phase - structure in this region by purely classical reasoning based on minimising the energy . qualitatively we expect a symmetric phase in a neighbourhood of the origin .this is represented schematically in fig .[ fig : clphases ] by the central rectangle . in order to study the ordered phases ( in particular af ) we divide the lattice in two sub - lattices ( denoted ` even ' and ` odd ' ) where the polyakov loop takes values @xmath119 and @xmath120 , respectively .two nearest neighbours belong to different sub - lattices .the absolute minima of the classical action @xmath121 will then be located at certain values @xmath122 and @xmath123 of the polyakov loop which are identified with its ` expectation values ' .we trust this reasoning as long as we are sufficiently far from the origin of the coupling plane i.e. from the disordered , entropy - dominated phase .any _ ferromagnetic _ ordering will be characterised by a minimum with @xmath124 while in an _ anti - ferromagnetic _ phase @xmath125 .quite interestingly we find two distinct ferromagnetic phases , one for which the polyakov loop is near a centre element or @xmath126 in the vicinity of @xmath127 and a different ferromagnetic phase with @xmath126 taking values near the intermediate points marked by triangles in fig .[ fig : domain ] .we call this an _ anti - centre _ phase ( ac ) .we expect a phase transition line separating the ferromagnetic and anti - ferromagnetic phases at vanishing potts - coupling @xmath118 .the resulting qualitative phase diagram is depicted in fig .[ fig : clphases ] . qualitative prediction of the phase diagram in the coupling constant plane for the effective action ( [ eq:2cmodel ] ) .the ferromagnetic ( f ) , anti - centre ( ac ) and anti - ferromagnetic ( af ) phases are obtained by looking for classical minima .the symmetric , disordered phase ( s ) is located where entropy is expected to dominate over energy . ] to discuss the ferromagnetic phases it suffices to minimize the action ( [ eq : smin ] ) with @xmath128 , in which case @xmath129 this can be done analytically . to localise the anti - ferromagnetic phase we have calculated the absolute minima of ( [ eq : smin ] ) on the target space depicted in fig .[ fig : domain ] numerically .the combined analytical and numerical results are summarised as follows .for negative @xmath116 we have a transition @xmath130 while for positive @xmath116 there is a richer structure , @xmath131 behaviour of the order parameter @xmath132 defined in ( [ eq : defprojector ] ) as a function of @xmath115 for fixed @xmath133 . for comparisonwe have added the result from the mean field analysis to be developed in the next section . ]the behaviour of a suitably projected order parameter @xmath132 ( the precise definition of which will only be needed later on ) for positive @xmath133 is shown in fig .[ fig : classicafer ] . upon inspection onenotes that for @xmath115 sufficiently negative the system starts out with the polyakov loop at a centre element .increasing @xmath115 beyond @xmath134 the order parameter drops monotonically until , at a critical coupling @xmath135 , there is a jump to the ac phase with @xmath126 near a anti - centre element .the jump of @xmath132 is due to a centre - transformation and does not imply that the polyakov loop itself jumps . indeed , @xmath136 changes smoothly and arrives at an anti - center element for @xmath137 .the system stays there until @xmath138 , where it jumps again , this time to the af phase .as expected , we see no symmetric phase in a purely classical analysis .actually , for @xmath139 there is just no symmetric phase .the next step of refinement to be presented in this section is a mean field ( mf ) analysis of the effective action ( [ eq:2cmodel ] ) .this will serve as a basis for a comparison with results from direct monte carlo simulations to be discussed later on .due to the peculiarities of the model as compared to standard scalar field theories the application of the mf approximation is not entirely straightforward . for the benefit of the readerwe will set the stage by giving a brief outline of the necessary modifications .for further details the reader is referred to our earlier paper @xcite . to keep the discussion sufficiently generalwe will first treat the effective action ( [ eq : effactionshort ] ) with five couplings focussing on simpler examples later on .we are interested in expectation values which are computed by evaluating integrals of the form @xmath140 } \int \mathcal{d}{\mathfrak{p}}\ , e^{-{s_{\text{\tiny eff}}}[{\mathfrak{p } } ] } \ , a[{\mathfrak{p } } ] \ ; , \\\mathcal{d } { \mathfrak{p}}&\equiv \prod_{{\mbox{\scriptsize\boldmath$x$ } } } d\mu({\mathfrak{p}}_{{\mbox{\scriptsize\boldmath$x$ } } } ) \ ; , \end{split}\ ] ] which apparently extend over the whole group manifold employing the haar measure @xmath141 .however , due to the gauge invariance of both action and measure the integrals can be reduced to the coset space of conjugacy classes which we ( somewhat symbolically ) denote by @xmath142 .hence we integrate with the reduced haar measure by replacing @xmath143 thus , is equivalent to @xmath144 } \int \mathcal{dp } \ , e^{-{s_{\text{\tiny eff}}}[\mathcal p ] } \ , a[\mathcal p ] \ ; , \\\mathcal{dp } & = \prod_{{\mbox{\scriptsize\boldmath$x$ } } } d\mu_{\rm red}(\mathcal p_{{\mbox{\scriptsize\boldmath$x$}}})\;. \end{split}\ ] ] the probability measure @xmath145 $ ] is characterized as the unique solution to the variational problem @xmath146 = -w[0 ] \;.\ ] ] the expectation value on the left - hand side has to be taken with respect to the integration measure @xmath147 $ ] with @xmath148 $ ] denoting the probability density of @xmath149 . from this point of viewthe mf approximation is nothing else but the restriction of the permissible densities to product form , @xmath150 \to p_{\mathrm{mf}}[\mathcal p ] \equiv \prod_{{\mbox{\scriptsize\boldmath$x$ } } } p_{{\mbox{\scriptsize\boldmath$x$ } } } ( \mathcal p_{{\mbox{\scriptsize\boldmath$x$}}})\ ; .\ ] ] expectation values can now be simply computed site by site via factorisation , @xmath151 our goal is to compute the effective potential as function of the mean characters @xmath152 . for that purposewe solve the variational problem ( [ eq : definfimum ] ) on the space of product measures with fixed expectation values of the characters .this is done by introducing appropriate lagrange multipliers @xmath153 . for ferromagnetic systemsone may assume that the weight functions @xmath154 at each site are identical , @xmath155 .this assumption corresponds to a translationally invariant ground state .according to the discussion of the previous section we expect anti - ferromagnetic phases and hence we must refine our choice for @xmath154 .we therefore introduce different weight functions on the even and odd sub - lattices , respectively , @xmath156 defining the sign of a lattice point as @xmath157 as a consequence , expectation values of characters will subsequently have two values depending on the sub - lattice where they are evaluated , @xmath158 the sources are taken constant as well when restricted to the even and odd sub - lattices , @xmath159 or @xmath160 , respectively . like the characters the sources are complex .the action couples only nearest - neighbour sites so that its expectation value entering ( [ eq : definfimum ] ) may be written as @xmath161 with @xmath162 denoting the lattice volume in @xmath5 spatial dimensions .the logarithm in ( [ eq : definfimum ] ) decomposes as @xmath163 it is convenient to drop the common volume factor @xmath164 and consider densities instead .the variation of finally yields the weight function for the even sub - lattice , @xmath165 and a completely analogous expression for the odd sub - lattice .here we have introduced @xmath166 as a short - hand for @xmath167 and the single - site partition function @xmath168 the sources @xmath169 are eliminated by inverting the relations @xmath170 to be evaluated separately on both sublattices .the schwinger function @xmath171 is defined as usual , @xmath172 introducing the legendre transform of according to @xmath173 the solution of is finally obtained as the mf potential ( density ) as a function of even and odd mean fields , @xmath174 from this expression one can easily derive relations between the sources @xmath175 and mean characters @xmath176 .for instance , by varying @xmath177 with respect to @xmath178 we obtain @xmath179 where we have used that the current is given as @xmath180 the first term in ( [ eq : distrsolution ] ) derives directly from the effective action ( [ eq : effactionshort ] ) and hence contains four couplings to which the potential coupling @xmath181 has to be added .a complete mf analysis of this system becomes very awkward . in what follows ,we therefore specialise to effective actions with only one or two couplings .the action ( [ eq : s10 ] ) defines what we call the ` minimal model ' with one coupling only , @xmath182 identifying @xmath183 and inserting ( [ eq : sgp ] ) the mf potential simplifies to @xmath184 it is useful to define the generic mf potential @xmath185 so that ( [ eq : simplesolution_2 ] ) can be rewritten as @xmath186 this expression clearly shows that for negative @xmath105 configurations with @xmath187 are favoured making the distinction between even and odd sub - lattices obsolete . the remaining unique expectation value, @xmath188 thus serves as an order parameter for the ferromagnetic phase transition . on the other hand ,if @xmath189 @xmath119 and @xmath120 will cease to be positively correlated and we expect an anti - ferromagnetic phase transition . in this case , a reasonable order parameter is given by @xcite @xmath190 let us also introduce the absolute values of the order parameters , henceforth denoted by @xmath191 if we assume for the moment that the occurring phase transitions are second order the corresponding mf critical couplings can be computed analytically as we are going to demonstrate next .the consistency condition reduces for the case at hand to @xmath192 as both @xmath119 and @xmath120 may be assumed small near the critical coupling @xmath193 the corresponding sources will also be small .thus , we may expand @xmath194 which again holds separately on each sublattice , @xmath195 and @xmath196 . plugging this into yields @xmath197 and hence the critical couplings @xmath198 for arbitrary order parameters we have solved the gap equations ( [ eq : gap ] ) numerically .the result is depicted in the following figure .the transition s - af is second order , the one to the phase f first order .thus , the anti - ferromagnetic transition must be at @xmath199 .the first order ferromagnetic transition is slightly above @xmath200 .mean field results for the minimal model with one coupling .the first order ferromagnetic transition s - f is at @xmath201 and the second order transition s - af at @xmath202 . ]our mf analysis thus confirms the qualitative results from the preceding section that already in the simplest model there is both a ferromagnetic and an anti - ferromagnetic transition .this is qualitatively consistent with the phase diagram fig .[ fig : clphases ] restricted to the horizontal axis . even for the simple model of the previous subsectionno explicit expression is known for @xmath203 as the @xmath9 group integrals can not be evaluated in closed form ( a fact well known from strong - coupling expansions , see e.g.@xcite ) .things naturally become worse if additional couplings are turned on . already for two couplings i.e. the action ( [ eq:2cmodel ] ) , the only way to proceed is by means of numerical methods . in order to obtain the mf version of the phase diagram fig .[ fig : clphases ] we have employed the following algorithm : 1 . at the extremal points of all sources @xmath204 occurring in can be eliminated in favour of the expectation values @xmath205 as in .since the character target space is compact it can be easily discretised defining measures @xmath206 and @xmath207 at each point . using these measuresexpectation values @xmath208 can now be computed which in general will differ from @xmath205 .hence , we first look for local minima of @xmath209 with norm @xmath210 .these minima , however , do not correspond to exact solutions yet but rather serve as the starting points of a recursion .2 . we now solve the equation @xmath211 for @xmath212 and @xmath213 by newton iteration using the local minima of the previous step as initial input .in this way we end up with multiple solution vectors @xmath205 each extremising .3 . the solution vector with minimal @xmath177 contains the desired ground state expectation values . with the help of this algorithmwe are able to compute the expectation values of @xmath85 and @xmath214 .we conclude this section with some remarks concerning our choice of the sources @xmath175 . allowing them to be complex yields an 8-dimensional parameter space for the observables of model ( [ eq:2cmodel ] ) with two characters , @xmath215 and @xmath216 . for this large parameter spaceit would actually be simpler to perform a high - precision monte carlo simulation than to find a good approximation for the global minima of .for this reason , we have chosen real sources as an input to our mf approximation .we expect this to be a very good approximation as long as the peak of the probability distribution for @xmath215 is concentrated near the real axis in fig .[ fig : domain ] . comparing with the classical analysiswe note that our real - source assumption is justified for all couplings which are located away from the boundary between the phases f and ac . the f - ac transition should be second order according to the analysis of section [ sec : qca ] while the mf approximation with real sources predicts a first order transition ( see fig .[ fig : classicafer ] ) .the contradiction will be finally resolved by monte carlo simulations showing that near the f - ac transition the peaks of the probability distribution for @xmath215 are off the real axis . apart from this fairly small region in parameter spacewe find a remarkable agreement between the mf results and the monte carlo results of the next section .this provides the ultimate justification for our simplifying choice of real sources in the mf approximation .the previous two sections have provided us with a good deal of information on the phase structure of the effective model in the most interesting regions of parameter space . based on thiswe have performed a large number of monte carlo simulations to quantitatively check the mf predictions and obtain a precise picture of the critical behaviour .as before , to avoid excessive complexity , we have concentrated on the action ( [ eq:2cmodel ] ) with two couplings @xmath115 and @xmath116 . at the beginning of section [ sec : mf ] we have already noted that both the action and the reduced haar measure depend only on the conjugacy class of the ( untraced ) polyakov loop .hence , one can choose the basic field variables either as the traces in the fundamental representations ( @xmath85 , @xmath86 ) and powers thereof or as a suitable parametrisation in terms of the eigenvalues as introduced in ( [ eq : pdiag ] ) .it turns out that for a numerical treatment the latter proves to be more appropriate and so we use ( [ eq : pdiag ] ) to represent the conjugacy class according to @xmath217 & \equiv \mathcal p_x ( \boldsymbol \phi ) = \text{diag } \big ( e^{i\phi_1 } , e^{i\phi_2 } , e^{-i(\phi_1+\phi_2 ) } \big ) \ ; , \\ \quad \boldsymbol\phi & \equiv ( \phi_{1},\phi_{2 } ) \ ; . \end{split}\ ] ] here , the following restrictions should be imposed such that the angular coordinates cover each class only once , @xmath218 these restrictions are somewhat awkward to implement in a simulation code .it is much more convenient to let @xmath219 take values in the full square @xmath220 which covers the fundamental domain given by ( [ eq : funddom ] ) six times . due to the residual gauge symmetry of the systemit is clear that expectation values will be unaffected by this over - counting . in the coordinates ( [ eq : defangulardecomp ] )the reduced haar measure becomes @xmath221 where the normalization is such that the measure integrates to unity over the square @xmath220 .all characters can be expressed in terms of @xmath222 and @xmath223 . using ( [ eq : redhaar ] ) the full measuremay be straightforwardly expressed in terms of the angular coordinates as @xmath224 } = \prod_x d\mu_{\mathrm{red } } ( \boldsymbol \phi_x ) e^{-{s_{\text{\tiny eff}}}[\boldsymbol \phi ] } \ ; .\ ] ] the focus of our numerical studies has been the phase diagram in the @xmath225 plane .there , we have scanned through the region @xmath226\times[-0.22\ldots 0.16]$ ] with a resolution of @xmath227 points , which were in total @xmath228 different monte carlo simulations . before we present our results a few words on our numerical techniques are in order . for the investigation of phase transitions , in particular their order ,histogram methods are widely used and accepted .this approach , however , requires large statistics and thus tends to consume a lot of computer time .in addition , we are interested in a fairly large range of coupling constants . for these reasons ,the updating algorithm for the polyakov loop models has to be fast and versatile .it turns out that the standard metropolis algorithm favourably matches both requirements if we aim at an accuracy of about @xmath229 . on the other hand , because of the highly nontrivial probability measure involved , a heat bath algorithm does not seem applicable or , in any case , would be too time consuming .in addition , its local nature should not yield any enhancement of statistics near a first order phase transition .we have thus refrained from implementing a heat bath update scheme but rather decided to optimise the metropolis algorithm as described in the following two paragraphs .when a system undergoes a first order phase transition , the histogram associated with the order parameter will generically display a multi - peak structure .depending on the total volume of the system the peaks can be very pronounced .in other words , the configuration space is decomposed into distinct sectors between which local algorithms can hardly mediate .one way to overcome the resultant failure in sampling the total configuration space is to make use of the multicanonical algorithm , see e.g.@xcite .the crucial improvement step consists in the replacement of the measure used in ( [ eq : haarphi ] ) according to @xmath230 the new improved measure on the right - hand side has a weight @xmath231 which depends on the ( modulus of the ) order parameter .one chooses the particular form @xmath232 where @xmath233 denotes the probability density of the order parameter .this choice leads to an enhancement of configurations that would otherwise be suppressed and thus allows for a much improved ergodic behaviour of the algorithm .histogram of @xmath234 with @xmath235 on a @xmath236-lattice sampled against the histogram for @xmath237 . ]the effect due to the altered measure is illustrated in fig .[ fig : multicanhist ] where two typical distributions are plotted . in the original distribution one clearly recognises two well - separated peaks .any local algorithm will fail to sample such a distribution properly once it is trapped at one of its peaks .the distribution actually used ` closes the gap ' enabling transitions between different peak regions during the simulation . in the end , of course , one has to correct for the change in the measure by reweighting with @xmath238 , @xmath239 where we have denoted expectation values taken with respect to the modified measure with a subscript ` mult ' .a slight problem with this approach , however , still has to be overcome .one actually needs right at the beginning what one set out to compute originally , namely the distribution @xmath240 .a practicable strategy is e.g. the following . from a small lattice volume ,say @xmath241 , where peaks are usually less pronounced , one obtains an approximate distribution function @xmath242 which is then used on a slightly bigger lattice , say of volume @xmath243 .this simple trick can be further refined if on the larger lattice one first computes @xmath244 using @xmath245 and subsequently repeats all measurements employing @xmath181 . in practice, this procedure is iterated several times to make larger and larger lattices available .going beyond a volume of @xmath246 requires additional knowledge of the scaling behaviour of @xmath233 .[ fig : multcanlog ] shows that , to a very good approximation , the scaling depends linearly on the volume , @xmath247 , @xmath248 plot of @xmath249 against volume @xmath250 for the minimal model with @xmath251 confirming the scaling behaviour .the coupling is very close to critical . ] in summary , the multicanonical algorithm yields substantial improvements compared to the standard metropolis algorithm and allows for very accurate simulations .lattices with volumes up to @xmath252 could thus be studied near the first order phase transition . however , as the implementation is very involved we applied it only to the minimal model ( @xmath253 ) . in principle , the generalisation to incorporating further couplings is straightforward , being merely a matter of having sufficient computing time available .the multicanonical algorithm is best suited for studying first order transitions .this is no longer true for second order transitions where it is outperformed by cluster algorithms at least if there is one available . for our purposes, none of the algorithms on the market could be immediately put to use .we therefore decided to modify the well - known wolff algorithm @xcite , an extension of the swendsen - wang @xcite algorithm originally proposed for discrete spin systems . before actually describing our modificationslet us briefly recall the main idea of @xcite .suppose the action ( and the measure ) are invariant under a certain global symmetry which acts on the fields according to @xmath254 typically , the symmetry operator @xmath255 will depend on some parameters which we collectively denote by @xmath256 ( continuous or discrete ) so that @xmath257 .it is important to note that the operator @xmath255 has to be idempotent , @xmath258 , in order to ensure detailed balance .the algorithm then works according to the following list . 1 .fix some parameter @xmath259 and hence some transformation @xmath260 . randomly choose a lattice point @xmath261 and apply the symmetry transformation @xmath262 which may be viewed as flipping the field variable @xmath263 .the point @xmath261 is checked and added to the cluster .[ it : flipnn ] repeat the following for all unchecked neighbours @xmath264 of @xmath261 : + let @xmath265 denote the contribution to the action from the link @xmath266 and compute @xmath267 which may be rewritten as @xmath268 since @xmath269 is already invariant under @xmath255 .the decision to add @xmath264 to the cluster is subject to an accept / reject step so that the probability to flip @xmath270 becomes @xmath271 if @xmath272 is flipped , check the point @xmath264 .3 . go back to step ( [ it : flipnn ] ) for all sites added to the cluster in the previous step .note that measures whether it is advantageous to flip @xmath270 once @xmath263 has been flipped . from the very construction of the algorithmit should be already clear that the clusters will increase with the correlation length of the system . in this way one suppresses the phenomenon of critical slowing down observed with local algorithms .this makes the cluster algorithm particularly suited for the study of second order phase transitions .the first step in adapting the cluster algorithm to our needs is to find suitable symmetries of the action .being a polyakov loop model the symmetry in question is the discrete @xmath0 symmetry , @xmath273 .in addition , the action has to be real which implies constraints on the way complex conjugation acts .from these symmetries one can construct three operators @xmath274 , @xmath275 , acting on the polyakov loop according to @xmath276 as required the @xmath274 square to unity .furthermore , it is easy to see that both the operators @xmath93 appearing in and the domain of the polyakov loop @xmath85 are left invariant by the action of @xmath274 .the latter is illustrated in fig .[ fig : reflections ] for a particular value of @xmath85 .illustration of the @xmath0 reflections @xmath274 used in our modified cluster algorithm . ] for the actual algorithm the transformations are not immediately applicable since the simulation is based on the angular variables @xmath277 introduced in ( [ eq : defangulardecomp ] ) .however , it is just a matter of a little algebra to show that the @xmath274 act on the angles @xmath277 via @xmath278 whereas in the original cluster algorithm flipping along randomly chosen lattice sites is sufficient to guarantee ergodicity this is no longer true in the present case .hence , we have to augment the update scheme by standard metropolis sweeps to make the algorithm ergodic . forone monte carlo step our cluster algorithm finally can be summarised as follows : 1 . choose a random number @xmath279 between @xmath280 and @xmath250 .2 . do @xmath279 standard metropolis sweeps at randomly drawn lattice points .3 . for a suitable fixed number @xmath281 repeat the steps for building a cluster as described above .do @xmath282 additional metropolis sweeps , again at randomly chosen lattice sites .from several test runs we have found that the number @xmath281 should be chosen such that total number of flipped sites after performing step ( 3 ) is approximately half the number of all lattice sites .thus , if @xmath283 denotes the typical size of a cluster , the following equation should hold ( at least approximately ) , @xmath284 one of the most interesting questions of course is the gain in performance compared to e.g. the standard metropolis algorithm .we have found that the autocorrelation time for the order parameter @xmath285 is independent of both the lattice extent ( at least in the range @xmath286 ) and the coupling constant ( if close to the second order phase transition ) implying a dynamical critical exponent of @xmath287 .moreover , for an optimal choice of @xmath281 the autocorrelation time is of order unity . as a result , our cluster algorithm outperforms the metropolis algorithm even for small lattices . on the largest lattices we have considered the cluster algorithm reduces autocorrelation times by two orders of magnitude as compared to metropolis updatingthis improvement comes at the cost of a slightly increased complexity , specifically a factor of 1.5 in computing time which clearly is negligible . on the other hand , in line with our expectations, no significant improvement has been found near the first order transition .we present the results of our monte carlo simulations in the same two steps as for the mf approximation .hence , we first report on the minimal model ( [ eq : sgp ] ) and switch on @xmath116 later on . with only @xmath115 different from zeroit is reasonably cheap to perform highly accurate measurements so that a precise quantitative comparison with the 3-state potts model is possible . for the model ( [ eq:2cmodel ] ) with two couplingswe determine the phase diagram in the @xmath115@xmath116 plane and compare with our expectations as laid out in the previous two sections .we conclude with a careful study of the nature of the phase transitions , in particular their continuity properties . to determine the ferromagnetic phase transition for @xmath288 we have used standard techniques which need no further explanation .it suffices to note that in order to study the first order ( s - f ) transition we performed @xmath289 sweeps on a @xmath290-lattice in the multicanonical ensemble for each value of the coupling constant .this led to highly accurate statistics until we approached the close vicinity of the critical coupling itself . in this regimeour statistics is restricted to @xmath291 independent samples due to large correlation times of the order of @xmath292 sweeps . as fig .[ fig : ferroptmc16 ] shows this is sufficient to demonstrate that the transition is first order . phase transition of fig .[ fig : ferroptmc16 ] when calculated within mf approximation . ]ferromagnetic phase transition computed with the multicanonical algorithm on a @xmath290-lattice .the expectation value of @xmath234 is plotted against its probability distribution given by the area shaded in grey .the latter clearly shows the correct discontinuous behaviour . note that , since we measure the modulus , statistical fluctuations manifest themselves in a ( small ) positive value of the order parameter even in the symmetric phase . ] in fig .[ fig : ferromf ] we compare our monte carlo result for @xmath293 with its mf approximation .the figure basically zooms into that part of fig .[ fig : meanscan ] where the first order ferromagnetic transition is located .again , the first order nature of the transition is corroborated .in addition to the coupling dependence of the order parameter we have determined both the critical coupling and the discontinuity @xmath294 at @xmath295 .the results are given in table [ tab : crit1 ] together with the mf prediction .again we find a surprisingly good _ quantitative _ agreement between simulations and mf approximation .moreover , if we consider the ratio of the two critical couplings , say @xmath296 , we are able to compare this with results of the 3-state potts model .the actual figures turn out to be fairly close , namely @xmath297 for the minimally coupled polyakov model and @xmath298 for the 3-state potts model @xcite ..[tab : crit1 ] critical couplings for the s - f and s - af phase transitions and jump @xmath294 . for the first order transition ( s - f ) there is excellent agreement between mf and monte carlo data .even the values for @xmath294 agree within 10% .for the second order transition ( s - af ) the critical couplings agree within approximately 20% .[ cols="<,^,^,^",options="header " , ] the uncertainty in @xmath299 quoted in table [ tab : crit1 ] is mainly due to the fact that the different cumulants do not precisely meet in a single intersection point ( cf .again fig .[ fig : bindercumulant ] ) . the error in the critical exponentsis estimated from a least square fit to the logarithm of ( [ eq : critexpdef ] ) and . to cross - check our results we have measured the expectation value of @xmath300 and its probability distribution @xmath301 in analogy with the ferromagnetic transition already discussed . as in the former case , the expectation value of the order parameter alone does not suffice to decide on the order of the transition .however , the probability distribution shown in fig .[ fig : aferrmcs ] is quite different from the one in fig .[ fig : ferroptmc16 ] .no discontinuous behaviour is observed now which provides further ( numerical ) evidence for a second order phase transition .equally important , the critical coupling obtained is compatible with the results presented in table [ tab : crit1 ] . for the sake of completeness fig .[ fig : aferrmf ] shows the mf prediction for the transition s - af . comparing with earlier results on the 3-state potts model @xcite we draw the important conclusion that the minimal polyakov loop model is in the same universality class .fundamental domain @xmath302 of the order parameter @xmath85 obtained by identifying @xmath0 copies according to the depicted arrows . ] having discussed the minimal model at length let us continue by switching on the second coupling @xmath116 in order to analyse the phase diagram in the coupling constant plane .this requires a suitably chosen ` indicator ' to distinguish the ( at least ) four phases ( s , f , ac and af ) we expect in accordance with our mf analysis of section [ sec : mf ] . while @xmath234 and @xmath303 clearly are order parameters for the minimal model they are numerically less suited for the model with two couplings .it turns out advantageous to construct a new observable denoted @xmath132 which may be obtained from @xmath234 by the following procedure .we first divide the domain of @xmath85 into six distinct parts as shown in fig .[ fig : rotatedobs ] .the light - shaded region represents the preferred locus of the polyakov loop in the ferromagnetic phase f , whereas the dark - shaded region corresponds to the anti - centre ferromagnetic phase ac . to eliminate the ( numerically superfluous ) @xmath0 symmetrythe first step in our projection is to identify the regions as indicated by the arrows in fig .[ fig : rotatedobs ] . in this way we end up with a fundamental domain @xmath302 for the @xmath0 symmetry centred along the real axis .every @xmath85 is mapped into @xmath302 by a centre transformation . to finally obtain a real observable we project the transformed @xmath85 onto the real axis .this projection results in the variable @xmath132 the sign of which clearly distinguishes between the two ferromagnetic phases .@xmath304 indicates the ac phase , @xmath305 the ferromagnetic phase f , while @xmath306 in the symmetric phase s. mathematically , the projection of @xmath85 to @xmath132 is given by @xmath307 to detect the af phase we simultaneously measure @xmath303 so that we can finally discriminate between all possible phases .our main results for the phase diagram are shown in fig .[ fig : phasediagmf ] obtained by the modified mf approximation explained earlier and fig .[ fig : phasediagmc8 ] which displays the monte carlo data and hence constitutes the most faithful representation of the phase structure .the monte carlo simulations were carried out on a @xmath308-lattice .the number of sweeps was chosen such as to reduce the jackknife error in the estimate of @xmath132 below @xmath309 .the independence of our samples was ensured by demanding that the autocorrelation time @xmath310 associated with the observable @xmath132 was less then one percent of the total number of sweeps . as a resultour simulations included at least @xmath311 sweeps far away from the critical regions and more than @xmath289 sweeps in their vicinity .it is reassuring to note that the qualitative phase diagram fig .[ fig : clphases ] predicted from energy - entropy arguments is quantitatively confirmed by both figs .[ fig : phasediagmf ] and [ fig : phasediagmc8 ] . comparing the latter in some detail it is once more remarkable how good the mf approximation works . within large regions of parameter spaceit agrees with the ` real ' data within 10% or less .interestingly , the observable @xmath132 also seems to be sensitive to the af phase ( see lower right part of figs . [fig : phasediagmf ] and [ fig : phasediagmc8 ] ) .however , any further discrimination between phases f and af by means of @xmath132 is impossible . to lift this degeneracyone clearly needs the af order parameter @xmath303 as an additional input .it remains to be discussed why @xmath132 is sensitive at all to af ordering .in what follows we will provide a heuristic answer in the context of the 3-state potts model . by definition ,all spins are ( anti)aligned in the ( anti)ferromagnetic ground state . in the @xmath312 symmetric ising model with only two spin states each possible ground stateis two - fold degenerate , independent of the particular ordering ( f or af ) .the counting of degeneracies , however , is totally different for systems with more spin states ( hence higher symmetry ) . in the afphase the additional freedom of choice between two ( or more ) states anti - aligned with a given one leads to an enormous degeneracy of the ground state _ in energy_. as a consequence , _ entropy _ will be the sole judge deciding what is to be observed in a measurement . for the @xmath0 polyakov loop modelboth mf approximation and the monte carlo simulations tell us that the most probable ground state corresponds to a preferred direction for the polyakov loop on one of the two sub - lattices and an equal distribution of the two remaining directions on the other sub - lattice .although we do not have an analytical justification for this statement the numerical evidence is compelling .based on the latter , the sensitivity of @xmath132 to af ordering can be explained by a net expectation value which for the case of the 3-state potts model is easily computed as @xmath313 phase boundaries and orders of transitions as obtained via monte carlo simulation on a @xmath308-lattice . we observe a mixture of both first and second order transitions depending on the particular values for the couplings .the symmetric phase is enclosed by ordered phases as already expected from the discussion of section [ sec : qca ] . for further details on the simulationsee the main text . ]phase diagram of fig .[ fig : phasestructure ] with oriented trajectories ( marked by arrows ) used for the histograms of figs .[ fig : histoone ] to [ fig : histoeight ] .the curves are directed from the symmetric to the broken phases intersecting the critical lines vertically .we have chosen a representative subset from a total of twenty such curves which were analysed to determine the order of the transitions . ]we conclude this discussion with an overview of the resulting phases , their boundaries and the order of the transitions in between presented in fig .[ fig : phasestructure ] .our reasoning here is based on additional measurements carried out along parametrised curves @xmath314 as depicted in fig .[ fig : histopositions ] .these simulations were exclusively focused on the order of the phase transition measuring the histograms of the observables @xmath85 and @xmath214 . as before all measurements were carried out on @xmath308-lattices each trajectory @xmath315 being sampled with twenty points . for every such point @xmath316 @xmath289monte carlo sweeps were performed . to improve the statistics of the histograms we made use of all previously discussed symmetries ( @xmath0 , complex conjugation and exchange of even and odd sub - lattice ) by binning @xmath85 or @xmath214 together with their centre images @xmath317 , @xmath318 , @xmath319 , @xmath320 and their complex conjugates .this amounts to using each measured value of @xmath85 six times and of @xmath214 even twelve times . in totalwe have recorded twenty such runs four of which are depicted in fig .[ fig : histopositions ] . to illuminate the order of the transitions corresponding to the four directed line crossings displayed in fig .[ fig : histopositions ] we present six ( out of twenty ) histograms in figs .[ fig : histoone ] to [ fig : histoeight ] which display the distribution of the observable @xmath85 respectively @xmath214 in the complex plane . the first figure in this series , fig .[ fig : histoone ] , corresponds to the ferromagnetic transition , s - f .the histograms displayed may be viewed as a ` movie sequence ' starting out in the symmetric phase with the distribution @xmath321 located at the origin .recalling the relation between probability distribution @xmath321 and the constraint effective potential @xcite , @xmath322 , this situation corresponds to a unique minimum of @xmath323 at @xmath324 . as the couplings change three further maxima which are @xmath0 copies of each other arise in addition to the one at the origin .hence , we observe coexistence of ordered and disordered phases .the new maxima are separated from the original unique maximum at the origin by a finite amount .the associated discontinuity together with coexistence clearly shows that the transition is first order .in contrast to this the situation depicted in fig . [fig : histotwo ] is quite different .first of all the distribution is much more delocalised as compared to fig .[ fig : histoone ] implying a much flatter effective potential .moreover , as soon as the maxima of the broken ( ordered ) phase emerge the maximum at the origin has dissolved .hence , there is no coexistence of different phases as in the previous case but rather a continuous appearance of new maxima , branching away from the origin towards the corners of the fundamental domain .these features characterise a second order phase transition .analogous behaviour can be observed in fig .[ fig : histothree ] with the maxima now moving in roughly the opposite directions indicating a transition to the ac phase . still different is the s - af transition of fig .[ fig : histoeight ] . by the same reasoning as given above this transition is of second order as well .however , the probability function emerging in the broken phase does not display the usual three - fold symmetry we have observed for the previous transitions . although the figures seemingly look rotationally invariant one actually finds a six - fold degeneracy rather than a continuous one . the symmetry enhancement by a factor of two has a simple explanation .note that fig .[ fig : histoeight ] shows the probability distribution of the af order parameter @xmath214 which is defined in terms of even and odd sub - lattices . interchanging the latter accounts for the additional factor of two .let us conclude this section with a short remark concerning figs .[ fig : histotwo ] and [ fig : histothree ] .we have seen that our numerical data seem to indicate the possibility of second order transitions from the disordered phase s to f or ac ordering .this is at variance with the folklore claiming the absence of a @xmath0 universality class in @xmath8 , see e.g. @xcite .it is not excluded that the observed second order characteristics near the tricritical point ( s - f - ac ) derive from artifacts due to our algorithms . at the moment , however , we can not make a definite statement about this issue .obviously , the puzzle deserves further investigation in the future .as outlined in the introduction the original motivation for polyakov loop models lies in their status as effective theories for finite - temperature gluodynamics .however , our investigation of the statistical mechanics involved should have made it clear that they are interesting in their own right . in support of this statementwe mention the 3-fold @xmath0 symmetry shared with the 3-state potts model , the nontrivial complex target space and the enormously rich phase structure resulting thereof .as we have seen , the latter can be understood qualitatively by simple energy - entropy considerations which predict a symmetric phase s close to the origin in coupling space ` surrounded ' by broken phases .these can exhibit ferromagnetic ( f ) , anti - ferromagnetic ( af ) or anti - centre ( ac ) ordering .this picture has been confirmed quantitatively both within a mean - field ( mf ) approximation and by extensive monte carlo simulations .both methods find a first order transition for s - f and a second order one for s - af . in order to capture the details of the phase diagram a total number of 8000 simulations ( corresponding to roughly 3000 hours of cpu time )was required .naturally , this has added further refinements to our classical and mf analysis like precise values for critical couplings and exponents .the agreement between mf and monte carlo result is much better than naively expected for a statistical model in @xmath8 .the s - f critical couplings agree to within 1% , the discontinuity @xmath325 within 10% and the s - af critical coupling within 20% .the precise agreement for the s - f transition suggests that the point where the s - f and s - ac critical lines merge is actually a _tricritical _ point where two first order transitions merge into a second order one .it is known that the upper critical dimension for a tricritical point is three @xcite so that in @xmath8 the mf approximation actually is exact apart from logarithmic corrections @xcite .this observation is corroborated by the fact that the polyakov loop model with two couplings is somewhat similar to the spin - one blume - capel @xcite model which apart from an ising term has an addition @xmath326 , @xmath327 , with 3 spin states .the model is known to have a tricritical point and is closely related to the 3-state potts model @xcite .the second order phase transition line is characterised by 3@xmath5 ising critical exponents .obviously , one should test the analogous exponents for our ac - f transition at the very left of our phase diagrams .one may also address the spatial localisation of the phases .on the 3@xmath5 lattice the coexisting phases should be separated by interfaces the tension of which can be measured ( see e.g.@xcite and references therein ) . in the end oneis of course interested in matching the effective couplings to the underlying microscopic theory , i.e. finite - temperature yang - mills .the resulting curve @xmath328 in the space of effective couplings should stay in the symmetric and ferromagentic phases .we have already solved the matching - problem for the ( simpler ) case of @xmath7 using inverse monte carlo methods based on schwinger - dyson equations . for @xmath9this is considerably more difficult ( at least technically ) due to the increased complexity of the haar measure .the problem is actually similar to one encountered in strong - coupling expansions where @xmath7 group integrals can be performed analytically which is no longer true for @xmath9 .nevertheless , we have been able to derive the relevant schwinger - dyson equations and hope to report on their applications in the near future .we thank guy buss for his contributions to the strong coupling expansion and georg bergner , wolfhard janke and erhard seiler for valuable hints .tk acknowledges support by the konrad - adenauer - stiftung e.v . ,cw by the studienstiftung des deutschen volkes and th by the plymouth particle theory group .

<S> we study effective lattice actions describing the polyakov loop dynamics originating from finite - temperature yang - mills theory . starting with a strong - coupling expansion </S><S> the effective action is obtained as a series of @xmath0-invariant operators involving higher and higher powers of the polyakov loop , each with its own coupling . </S><S> truncating to a subclass with two couplings we perform a detailed analysis of the statistical mechanics involved . to this end </S><S> we employ a modified mean field approximation and monte carlo simulations based on a novel cluster algorithm . </S><S> we find excellent agreement of both approaches concerning the phase structure of the theories . </S><S> the phase diagram exhibits both first and second order transitions between symmetric , ferromagnetic and anti - ferromagnetic phases with phase boundaries merging at three tricritical points . </S><S> the critical exponents @xmath1 and @xmath2 at the continuous transition between symmetric and anti - ferromagnetic phases are the same as for the 3-state potts model . </S>

1409.1775

the three - dimensional magnetic fields in the higher solar atmosphere provide crucial information toward understanding various solar activities , such as filament eruptions , flares , and coronal mass ejections ( cmes ) . to understand the physical mechanisms of these activities in the solar atmosphere ,an important step is to find out the underlying structure of the magnetic field above the related active regions .currently , due to the extremely low density and high temperature , the direct measurement of the magnetic field in the solar chromosphere and corona is not as sophisticated as observation in the photosphere @xcite . as an alternative to direct measurement of the three dimensional magnetic field in solar atmosphere ,numerical modeling are used to infer the field strength in the higher layers of the solar corona from the measured photospheric magnetic field .thus model assumption is called the force - free field assumption , as the lorentz force vanishes .this can be obtained by either vanishing electric currents or the currents are co - aligned with the magnetic field lines .it is generally assumed that the magnetic pressure in the corona is much higher than the plasma pressure ( small plasma @xmath0 ) and that therefore the magnetic field is nearly force - free ( for a critical view of this assumption see .force - free coronal magnetic fields are defined entirely by requiring that the field has no lorentz force and is divergence free ( the solenoidal condition ) : @xmath1 @xmath2 subject to the boundary condition @xmath3 where @xmath4 is the magnetic field and @xmath5 is measured vector field on the photosphere .equation ( [ one ] ) states that the currents are co - aligned with magnetic fields and equation ( [ two ] ) describes the absence of magnetic monopoles . using two equations as constraints equations, one can calculate the magnetic field density in a corona volume for photospheric measurements . in absence of coronal magnetic field measurements , nonlinear force - free ( nlff ) reconstruction techniques based on photospheric magnetic field measurements ( within their limitations ;see ) are to date one of the few means of approximating the coronal field structure with a near real - time temporal cadence given the spatial resolution provided by the measured field vector ( _ e.g. _ , ) . for more details on the success and future improvements of the reconstruction techniques , we direct the reader to @xcite . in the present study ,we use vector magnetic field data of hmi with two different field - of - views as an input to our spherical optimization code and compare the results .there has been relatively little consideration of the effect of the size of computational domain on nlfff modeling to date .therefore , we investigate whether the size of computational domain chosen for coronal magnetic field modeling significantly influences the resulting solutions to the model . in our experiment , we provide a wider field of view that can accommodate magnetic connections between the region of interest and both nearby plage and neighboring ars .we have selected a large region on the sun observed on 09 march 2012 at 20:55ut with four ars of which two of them are in northern hemisphere and the other two in the south . in the companion case , the region of interest with two ars ( ar11429 and ar11430 ) in the northern hemisphere is cropped more tightly , as if observed with an instrument with a limited field of view .we compare quantities like magnetic flux density , electric current density and free magnetic energy density of ars of interest obtained from the two different field - of - views . in the study ,the same spatial resolution is used for both cases .the _ helioseismic and magnetic imager _ ( hmi ) is part of the _ solar dynamics observatory _ ( sdo ) , which provides the first uninterrupted time series of space - based , full - disk , vector magnetic field observations of the sun with a 12-minute cadence @xcite .hmi consists of a refracting telescope , a polarization selector , an image stabilization system , a narrow band tunable filter and two 4096 pixel ccd cameras with mechanical shutters and control electronics .photospheric line - of - sight los and vector magnetograms are retrieved from filtergrams with a plate scale of 0.5 arc - second . from filtergramsaveraged over about ten minutes , stokes parameters are derived and inverted using the milne - eddington ( me ) inversion algorithm of .the transverse components of vector magnetic fields suffer from the so - called @xmath6 ambiguity .the inversion applied to each pixel can not resolve the inherent @xmath6 azimuth ambiguity in the transverse field direction .therefore , the @xmath6 ambiguity has been resolved by minimum energy method @xcite . as described in , in weak - field areas, the minimization may not return a good solution due to large noise .the noise level is @xmath7 10 g and @xmath7 100 g for the longitudinal and transverse magnetic field , respectively .therefore , in order to get a spatially smooth solution in weak - field areas , we divide the magnetic field into two regions , i.e. , strong - field region and weak - field region , which is defined to be where the field strength is below 200 g at the disk center , and 400 g on the limb .the values vary linearly with distance from the center to the limb . for more detailed descriptions of hmi data processing and production techniques , we direct the reader to .we solve the force - free equations ( 1)-(3 ) by using an optimization principle as proposed by wheatland , sturrock , and roumeliotis ( 2000 ) and extended by wiegelmann ( 2004 ) and wiegelmann and inhester ( 2010 ) in the form @xmath8@xmath9 @xmath10 @xmath11 where @xmath12 and @xmath13 measure how well the force - free equation [ equation ( [ one ] ) ] and divergence - free condition [ equation ( [ two ] ) ] are fulfilled , respectively . @xmath14 and @xmath15 are weighting functions , which are one in the region of interest and drop to zero in a 32 pixel boundary layer toward the lateral and top boundaries of the computational domain . in this work ,we implement the surface integral term , @xmath16 , in equation ( [ 4 ] ) to work with boundary data of different noise levels and qualities @xcite .this allows deviations between the model field @xmath4 and the input field , _ i.e. _ the observed @xmath17 surface field , so that the model field can be iterated closer to a force - free solution .@xmath18 is a space - dependent diagonal matrix whose elements @xmath19 are inversely proportional to the estimated squared measurement error of the respective field components .because the line - of - sight photospheric magnetic field is measured with much higher accuracy than the transverse field , we typically set the component @xmath20 to unity , while the transverse components of @xmath21 are typically small but positive . in regionswhere transverse field has not been measured or where the signal - to - noise ratio is very poor , we set @xmath22 . in order to control the speed with which the lower boundary condition is injected during the nlfff extrapolation, we have used the lagrangian multiplier of @xmath23 as suggested by .our optimization method uses vector field values @xmath17 over the entire lower boundary as boundary conditions at the photosphere .the inconsistency of the boundary data with the force - free assumption can lead to poor model solutions .therefore , we use a spherical preprocessing procedure to remove most of the net force and torque from hmi boundary data to be more consistent with nlfff modeling @xcite .there are no vector magnetic field measurements for the side and top boundaries of a localized domain .therefore , we have to make assumptions about these fields before performing a nlfff extrapolation .we assumed the lateral and upper boundaries of the computational domain as current - free . in order to initialize our nlfff code, we calculated potential field from sdo / hmi data set using preprocessed radial field components ( @xmath24 ) by spherical harmonic expansion method .therefore , the potential magnetic field is used with values of radial field matching the initial preprocessed lower boundary values , and this field provides also the initial field at all points in the volume except the lower boundary .the computational box is a wedge - shaped volume @xmath25 with six boundary surfaces ( four lateral side boundaries , top and photospheric boundaries ) .this box enables us to study the connectivities between ars and their surroundings for the large field - of - views .+ + + + + + + mm obtained from ( a ) the boundary of region of interest with smaller field of view and ( b ) the boundary of corresponding region of interest ( fov@xmath26 ) cropped from the region with larger field of view ( fov2).,title="fig:",width=461,height=56 ]the main purpose of this work is to investigate whether the size of computational domain chosen for coronal magnetic field modeling significantly influences the resulting solutions to the model . in our investigations , we provide two sdo / hmi vector magnetic field boundaries data from two different field of views to our spherical code .we extracted the two boundary data from full - disk hmi data observed on 09 march 2012 at at 20:55ut .the first data set accommodates the region of interest , which has two ars ( ar11429 and ar11430 ) located in the northern hemisphere ( see , figure [ fig1 ] ) .the photospheric surface boundary of the computational box over the region of interest is shown by the red rectangles in figure [ fig1 ] ( referred as fov1 , thereafter ) .this region is cropped more tightly , as if observed with an instrument with a limited field of view .the other boundary data has a wider field of view that can accommodate magnetic connections between the region of interest and neighboring ars ( referred as fov2 , thereafter ) .it has four ars of which two of them are in northern hemisphere and the other two in the south .the photospheric surface boundary of the larger computational box is shown by the yellow rectangles in figure [ fig1 ] .the region of interest , fov1 , was well isolated from the surrounding with the quiet sun region .however , it was magnetically connected to the ars in the southern hemisphere by trans - equatorial loops crossing the lateral boundary of fov1 in the south ( see , figures [ fig1 ] c and d ) . during modeling the 3-d magnetic field of fov1 , the photospheric magnetic field outside this region is ignored .this approach is the better justified the more the region of interest is isolated .often , however , active regions are not always completely isolated but magnetically connected with other active regions @xcite . in our previous work @xcite , we have studied the connectivity between two active regions one in the northern hemisphere and the other in the south .the study revealed that there were substantial number of fluxes shared between the ars and as the result there were trans - equatorial loops carrying electric current .therefore , those fluxes and electric currents crossing the lateral boundary of the region of interest might have effect on magnetic field solutions , electric current density and free magnetic energy .to calculate three - dimensional nlff magnetic field solutions , we minimize the functional @xmath27 given by equation ( [ 4 ] ) using preprocessed photospheric boundary data for the two cases . in the first case , the boundary data of fov1 ( region in the red rectangles of figure [ fig1 ] ) has been provided to our spherical optimization code as an input .we compute the solution on a number of different horizontal grid scales , using 3-levels multigrid ( _ e.g. _ @xmath28 , @xmath29 , and @xmath30 ) , with the results of the coarser resolution used to initialize the next , finer resolution .the solution from the coarser grids are interpolated onto finer grids as the initial state for the magnetic field in the computational domain of the next larger box .this gives a better starting equilibrium in the full resolution box , an improvement on an initial potential field .the region of interest ( fov1 ) corresponds to a wedge shape box with the field of view of @xmath31\times[\theta_{\rm{min}}=5.5^{\circ}:\theta_{\rm{max}}=40^{\circ}]\times[\phi_{\rm{min}}=160^{\circ}:\phi_{\rm{max}}=230^{\circ}]$ ] . in the second case, the boundary data of fov2 ( region in the yellow rectangles of figure [ fig1 ] ) has been provided to the code as an input .we compute the solution on a number of different horizontal grid scales , using the same 3-levels multigrid ( _ e.g. _ @xmath32 , @xmath33 , and @xmath34 ) .this region ( fov2 ) corresponds to a wedge shape box with the large field of view of @xmath35\times[\theta_{\rm{min}}=-30^{\circ}:\theta_{\rm{max}}=40^{\circ}]\times[\phi_{\rm{min}}= 160^{\circ}:\phi_{\rm{max}}=230^{\circ}]$ ] , twice the size of fov1 .once the 3d magnetic configuration is reconstructed for the two cases , we plot the surface vector magnetic field solutions of the region of interest from the two individual cases at the coronal height of @xmath36 mm ( see , figure [ fig2 ] ) . for the comparison of the two magnetic field solutions in the region of interest that are obtained from the two field - of - views ( fov1 and fov2 ), we cropped the volume of region of interest that corresponds to fov1 from fov2 ( referred as fov2@xmath37 , thereafter and where fov2@xmath37= fov2 @xmath38 fov1 , is an intersecting region of fov2 and region of interest , fov1 ) .figure [ fig2 ] a and b show the vector fields solution that are obtained from the boundary data of the region of interest from fov1 and fov2@xmath39 respectively .we also plot the difference in the magnetic field solutions of fov1 and fov2@xmath37 over the region of interest in figure [ fig2 ] c. the difference in magnetic field vector is higher in the area where the field lines are connected to outside the domain . in order to evaluate the three - dimensional nlfff of the region of interest from the two different solutions, we have calculated the following quantities like @xmath40 where @xmath41 .@xmath42 , @xmath43 , @xmath44 , and @xmath45 are lorentz - force , divergence of magnetic field , the average angle between magnetic field and electric current density , and vector correlation between the magnetic field from fov1 respectively .we normalize the magnetic field with the average photospheric field and length scale of solar radius .table [ table1 ] shows that the force - freeness ( @xmath42 ) and divergence ( @xmath43 ) conditions are best fulfilled for the magnetic field solutions of fov@xmath26 extracted from larger field - of - view .similarly , the magnetic field solution of fov1 has larger average angle between magnetic field and electric current density compared to that of fov@xmath26 , indicating that the magnetic field solution of fov1 are less force - free that that of fov@xmath26 .the vector correlations values between nlff magnetic fields and the corresponding potential fields from the two solutions indicate that magnetic field solution of fov1 is somewhat close to potential field than the solution of fov@xmath26 . in our previous works @xcite, we studied the connectivity between many neighboring ars which were found to share a significant amount of magnetic flux compared to their internal flux connecting one polarity to the other .in this study , we have used the same method to calculate the total ( `` shared '' ) flux , @xmath46 , for all field lines starting from where @xmath47 g in the region of interest , fov1 , and leaving the volume through the lateral boundary of fov1 towards the rest of the region of fov2 outside of fov@xmath26 .table [ table1 ] shows that there are more magnetic field lines crossing the lateral boundary of fov1 for the magnetic field solution of fov@xmath26 than those of fov1 .this indicates that there is large flux shared between the region of interest and its neighboring region , if we include all the connecting regions in the computational domain . in order to compare our 3-d field reconstructions with observation, we plot the selected field lines of the nlfff solutions for the two data sets and we overlay the field lines with corresponding aia composite ( 171 , 193 , 211 ) image ( see figure [ fig3 ] a and c ) . from those figures ,one can see that the field lines of nlfff model solutions obtained from the two field - of - views have significant difference . in figures [ fig3]b and d , we rotate the same selected field lines to the limb .the field lines reconstructed from the boundary data of fov1 deviate from observation than the one obtained from the boundary data of fov2 .especially for those field lines crossing the southern lateral boundary of fov1 , the deviation is more pronounced .there are spatial correspondence between the overall shape of the magnetic field lines of fov@xmath26 and the euv loops .those qualitative comparisons between nlfff model magnetic field lines and the observed euv loops of aia images indicate that the nlfff model provides a more consistent field for large field - of - views ..evaluation of the reconstruction quality for the 3-d magnetic field solutions of fov1 and fov@xmath26 . [ cols="^,^,^,^,^,^,^ " , ] once we have calculated the 3-d magnetic field solutions for the two field - of - views , we computed the vector current density @xmath48 as the curl of the field : @xmath49 where @xmath4 is nonlinear force - free 3d magnetic field solution .we use finite difference method to solve for the electric current density from curl of @xmath50 .we plot electric current density vector @xmath48 of the region of interest .figures [ fig4](a ) and ( b ) show the electric current densities on the layer at the corona height of @xmath36 mm from the two magnetic field solutions .there is substantial difference in electric current densities of the two solutions . in order to quantify the percentage share in the electric current , we first identified those field lines carrying electric currents and emanating from the region of interest , fov1 ( crossing its lateral boundary on the south ) and ending into the region in fov2 outside of fov1 .the ratio of total unsigned electric current density flux due to those electric current carrying field lines connecting those two regions to the total unsigned electric current density flux due to all field lines with current emanating from fov1 gives us the percentage share in the electric current between the two regions .for the case of electric current density calculated from the 3-d field of larger field - of - view , we found that @xmath51 of positive / negative polarity of the ars in fov@xmath26 ( region of interest ) in the northern hemisphere is connected to positive / negative polarity of ars in fov2 outside fov1 crossing the southern lateral boundary of fov1 .however , there is no electric current density crossing the southern lateral boundary of fov1 for the case of magnetic field solution obtained from fov1 boundary data .this might be due to the fact that we initialized the lateral boundaries of fov1 with the potential field .we estimate the free magnetic energy budget above the potential - field state , the difference between the extrapolated nlfff and the potential field with the same normal boundary conditions in the photosphere ( , @xcite ) .the free magnetic energy budget is a measure of the magnetic energy that can be stored in a magnetic configuration or can be released during an eruptive or reconnection event .we therefore estimate the upper limit to the free magnetic energy budget in excess to the potential field state , associated with coronal currents , by : @xmath52 where the magnetic energy @xmath53 is computed in the coronal volume @xmath25 as follows : @xmath54 our result for the estimation of the total free magnetic energy budget above the potential - field state throughout the computational volumes of the region of interest from fov1 and fov@xmath26 have @xmath55 and @xmath56 ( see , table [ table1 ] ) more energy than the corresponding potential - fields , respectively . in figure[ fig5 ] , we plot total surface free magnetic energy density of the two magnetic field solutions relative to the potential one at the corona height of @xmath36 mm . from those figures show that there is free magnetic energy difference in the area where loops are connected to outside the domain .in this study , we have investigated effects of the size of the domain chosen for coronal magnetic field modeling on resulting model solution .we applied spherical optimization procedure to vector magnetogram data of _ helioseismic and magnetic imager _ ( hmi ) onboard _ solar dynamics observatory _( sdo with four ars observed on 09 march 2012 at 20:55ut . in this study , we have used spherical multigrid technique where we computed the solution on a number of different horizontal grid scales , with the results of the coarser resolution used to initialize the next , finer resolution .the solution from the coarser grids were interpolated onto finer grids as the initial state for the magnetic field in the computational domain of the next larger box .this gives a better starting equilibrium in the full resolution box , an improvement on an initial potential field . in order to study the effects of the size of the domain over large area of the sun , we have selected a well isolated , but magnetically connected region of interest with two ars and located in the northern hemispherewe have used two boundary data with two different computational boxes , one exactly accommodating the region of interest and the other accommodating the region of interest and its southern surrounding region with two additional ars .we computed the 3-d magnetic field solutions from the two boundary data in the two computational boxes with different field - of - views .for the comparison with the solution from the smaller box over the selected region of interest , we extracted the 3-d magnetic field solution of the region of interest from the field solution of large field - of - view .we have compared the magnetic field lines of the two magnetic field solutions with the observed corresponding euv loops .the comparison shows that the field lines of the modeling fields computed within the larger field - of - view generally agree with the coronal loop images than the one computed within the smaller box . in the study, we found that there are substantial differences in magnetic fluxes , electric current densities and free magnetic energies at a selected coronal height from the two solutions .the difference is even more pronounced in the regions where there are connections to outside the domain .this is due to the fact that the fluxes and electric currents crossing the lateral boundary of the region of interest have effect on magnetic field solutions , electric current density and free magnetic energy .therefore , one has to consider the region of magnetic connectivity while modeling the magnetic field of the region of interest in spherical geometry .data are courtesy of nasa / sdo and the aia and hmi science teams . this research was supported by an appointment to the nasa postdoctoral program at the goddard space flight center ( gsfc ) , administered by oak ridge associated universities through a contract with nasa .

<S> the solar coronal magnetic field produces solar activity , including extremely energetic solar flares and coronal mass ejections ( cmes ) . </S><S> knowledge of the structure and evolution of the magnetic field of the solar corona is important for investigating and understanding the origins of space weather . </S><S> although the coronal field remains difficult to measure directly , there is considerable interest in accurate modeling of magnetic fields in and around sunspot regions on the sun using photospheric vector magnetograms as boundary data . in this work , </S><S> we investigate effects of the size of the domain chosen for coronal magnetic field modeling on resulting model solution . </S><S> we apply spherical optimization procedure to vector magnetogram data of _ helioseismic and magnetic imager _ ( hmi ) onboard _ solar dynamics observatory _ ( sdo with four active region observed on 09 march 2012 at 20:55ut . </S><S> the results imply that quantities like magnetic flux density , electric current density and free magnetic energy density of ars of interest are significantly different from the corresponding quantities obtained in the same region within the wider field of view . </S><S> the difference is even more pronounced in the regions where there are connections to outside the domain . </S>

1311.4377

the investigation of electron transport in nano - structures , such as quantum dots or carbon nanotubes , is of high actual interest .it leads to the observation of a multitude of mesoscopic phenomena , where the dual nature of quasi - particles is readily observed .its wave character is manifest in interference phenomena due to the phase coherence of charge carriers , whereas its granular character is best visible in the coulomb blockade effect @xcite . besides their fundamental character ,mesoscopic transport studies are also of high technological interest .it is sufficient to quote the development of carbon nanotube transistors @xcite , which can operate up to terahertz frequencies @xcite .several complications arise for the theoretical description of transport properties in nanodevices which are beyond solid state physics outlined in classical text - books .these include nonadiabatic effects and time dependent phenomena , as well as the wealth of properties induced by coulomb correlations . for a realistic description ,all these effects have to be treated simultaneously .this is a real challenge .nonadiabatic time dependent effects are usually treated within the framework of two - time keldysh green s functions @xcite , but the treatment of the coulomb interaction is indispensable in most cases .one can distinguish three levels of correlation effects : ( i ) weak coulomb correlation of either hartree ( electrostatic interaction ) or hartree - fock ( including spin exchange ) character , ( ii ) coulomb blockade , and ( iii ) kondo physics .the two crucial parameters that guide coulomb effects are the coupling between dot and electrodes , and the temperature . by decreasing these two parametersthe coulomb effects become more and more important , the kondo physics taking place at low temperature , for a low bias voltage . to simulate the terahertz response of carbon nanotube transistors ,the treatment of weak correlation effects ( in hartree approximation ) is already a standard procedure @xcite .however , a well developed time - dependent formalism in case of strong correlation and a fortiori when coulomb - induced collective phenomena set in , deserves to be improved , and is currently a very active research area . here, we present a formalism of non - adiabatic electron transport which is able to treat an arbitrary time - dependence of bias and gate voltages , as well as a time - dependence of the hybridization between leads and dot .the transient and steady - state properties can be evaluated , without any a priory assumption about adiabatic or sudden limits .our formalism includes the effects of coulomb correlations beyond weak - coupling treatment , and works for arbitrary on - site interaction @xmath0 .our approach is not restricted to the wide - band limit , although the presented numerical results rely on it .it treats correctly the uncorrelated ( @xmath1 ) and the atomic ( disconnected dot ) limits , but can not account for the kondo effect , albeit interplay between transient phenomena and kondo physics are very interesting issues @xcite , @xcite . as a consequence the temperatures considered here are higher than the kondo temperature @xmath2 .the formalism is based on a systematic use of two - time keldysh green s functions in the hubbard i approximation ( hia ) @xcite .the key and new point lies in the proposed way to circumvent the difficulties of the usual heavy computation when dealing with two time green s functions .our approach is easy to implement and constitutes a substantial computer - time saving method .this advantage is shared with the technique recently developed by croy and saalman @xcite .the approaches are different in detail but predict consistent results , as shown later . after having presented the general formalism in sec .2 , we show in some detail the main idea of our approximation in sec .first we apply the approach to a steady - state case ( sec.4 ) .this allows us to compare our correlation treatment to the hartree - fock ( hf ) approximation , and also to a more sophisticated correlation treatment , namely the noncrossing approximation ( nca ) @xcite .for the parameters under study , the nca with a renormalized hubbard bandwidth and hia in the non - equilibrium steady - state regime compare favorably , while hf turns to be unreliable @xcite . in the time - dependent situation ( sec .5 ) we investigate the case of a pulse modulation for the bias voltage , which enables to measure the charging time in the coulomb blockade regime , as well as the tunnel and displacement currents .we then investigate the case of a forced harmonic bias voltage , with transient and steady - state regimes , before turning to the case of a pure pumping experiment , where , along the lines developed by croy _et al . _@xcite , we address the question of adiabatic @xcite and non - adiabatic frontier .we finally close with our conclusions ( sec .we consider a system which is a single level interacting anderson quantum dot , coupled to two uncorrelated leads , acting as source and drain .the hamiltonian reads @xcite @xmath3 @xmath4 is a contact hamiltonian corresponding to electrons in leads , and takes the following form @xmath5 where @xmath6 denotes momentum index , @xmath7 stands for left and right leads and @xmath8 is the spin degree of freedom . here@xmath9 and @xmath10 are electron creation and annihilation operators for the @xmath11-lead state @xmath12 .@xmath13 is a coupling term @xmath14,\ ] ] where @xmath15 and @xmath16 are electron creation and annihilation operators at the dot for spin state @xmath17 .the central region hamiltonian @xmath18 includes coulomb repulsion term : @xmath19 the energy level in the dot @xmath20 , in the leads @xmath21 , as well as the hybridization coefficients @xmath22 , are all considered to be time dependent , and independent of each other .experimentally , that can be realized by applying different bias and gate voltages .we shall use the assumption that the time dependence of the hybridization parameters can be factorized as @xmath23 @xcite .current from left contact to the central region can be calculated as @xmath24 \rangle,\ ] ] where @xmath25 is the lead fermion number operator .similar problem was considered recently @xcite in the wide - band limit with a density matrix approach using a truncated equation of motion technique . in the present paperwe apply the time - dependent keldysh formalism @xcite .the expression for the current can be written in terms of central region keldysh green s functions ( see ref .@xcite ) @xmath26 \left .\right\}. \label{eq : lcur}\end{aligned}\ ] ] here @xmath27 is defined as @xmath28,\end{aligned}\ ] ] where @xmath29 is the density of states per lead and per spin , which we choose independent of @xmath11 .@xmath30 is the fermi distribution function of the left contact .finally @xmath31 and @xmath32 are the lesser and retarded keldysh green s functions of the central region @xmath33 even out of equilibrium , the green s functions are diagonal in spin , due to our choice for @xmath34 which conserves spin . in order to calculatethe time - dependent current one needs to calculate these green s functions first .we calculate the equation of motion for the retarded green s function @xmath35 .it leads to @xmath36 where two other green s functions appear : @xmath37 and @xmath38 .the equation of motion for @xmath39 is @xmath40 the formal solution of this equation can be written as @xmath41 where @xmath42 is the green s function for the uncoupled system @xmath43.\ ] ] substituting into the equation for @xmath35 , we get @xmath44 where @xmath45 is the hybridization self - energy @xmath46 equation is not a closed equation for @xmath35 because of the presence of so far unknown @xmath47 . in order to get a closed equation we need to make certain approximations regarding this last green s function . within the hf approximation we use the following factorization @xcite @xmath48 where @xmath49 . in this casewe can get an equation for @xmath35 @xmath50 the quantity @xmath51 can be determined from the lesser green s function as @xmath52 and therefore a self - consistent scheme is needed to solve eq .. in the hf approximation , the retarded dot green s function has the same form as in the case of non - correlated electrons , the only difference being @xmath53 .it means that the hartree - fock approximation reduces here to the hartree approximation , which is quite crude , as will be seen later .it is possible to get a better approximation for @xmath47 by considering the equation of motion for this function .first we write down the commutator @xmath54 $ ] and neglect all the terms without coinciding quantum numbers for at least two fermionic operators : @xmath55 = ( \varepsilon_0 ( t ) + u){\mathbf}d_{\sigma } { \mathbf}d_{\bar{\sigma}}^{\dag } { \mathbf}d_{\bar{\sigma } } \nonumber \\ & & + \sum_{k \alpha } \left ( v_{k \alpha}(t ) { \mathbf}c_{k \alpha \bar{\sigma}}^{\dag } { \mathbf}d_{\sigma } { \mathbf}d_{\bar{\sigma } } + v^*_{k \alpha}(t ) { \mathbf}d_{\bar{\sigma}}^{\dag } { \mathbf}c_{k \alpha \bar{\sigma } } { \mathbf}d_{\sigma } + v^*_{k\alpha}(t ) { \mathbf}c_{k \alpha \sigma } { \mathbf}d_{\bar{\sigma}}^{\dag } { \mathbf}d_{\bar{\sigma } } \right ) \nonumber \\ & & \qquad \approx ( \varepsilon_0 ( t ) + u ) { \mathbf}d_{\sigma } { \mathbf}d_{\bar{\sigma}}^{\dag } { \mathbf}d_{\bar{\sigma } } + \sum_{k \alpha } v^*_{k \alpha}(t ) { \mathbf}c_{k \alpha \sigma } { \mathbf}d_{\bar{\sigma}}^{\dag } { \mathbf}d_{\bar{\sigma}}. \label{eq : comm}\end{aligned}\ ] ] at this point we could follow the usual hia procedure ( henceforth called canonical ) , with the factorization @xmath56 , as used for example in ref . @xcite . this would close the system of equations .however this decoupling scheme leads to cumbersome numerical difficulties , while keeping this term untouched as shown later , enables to skirt them .the equation of motion for @xmath57 then takes the following form @xmath58 where appears a new green s function @xmath59 , for which , neglecting the same kind of terms as in , the equation of motion leads to @xmath60 eqs . and form a closed set of equations for @xmath47 and @xmath61 .it is useful to rewrite this system in integral form .to do so we define the following functions @xmath62 , \label{eq : g1}\\ g_{u}^r ( t , t ' ) & = & -i \operatorname{\theta}(t - t ' ) \exp\left [ -i \int_{t'}^t d t_1 \varepsilon_{0 } ( t_1 ) \right ] \nonumber \\ & \quad & \times \exp\left[-iu(t - t')\right],\label{eq : g2}\\ \tilde{g}_{0}^r ( t , t ' ) & = & ( 1 - n_{\bar{\sigma } } ( t ' ) ) g_{0}^r ( t , t'),\label{eq : g3}\\ \tilde{g}_{u}^r ( t , t ' ) & = & n_{\bar{\sigma } } ( t ' ) g_{u}^r ( t , t ' ) , \label{eq : g4}\end{aligned}\ ] ] and rewrite the equations and @xmath63 substituting eq . into and using for @xmath64 , we get an integral equation for @xmath47 only @xmath65 going back to the integral equation for the local green s function resulting from @xmath66 and using , we obtain @xmath67 next we will decompose the central region green s function into two components @xmath68 , where @xmath47 satisfies the equation and @xmath69 is a new unknown function . then , using the definition for @xmath70 we can write @xmath71 it is seen from that the last two terms here add up to form @xmath47 .thus we can get the integral equation for @xmath69 @xmath72 to summarize we have decomposed the diagonal central region green s function into two terms @xmath68 , where @xmath69 and @xmath47 satisfy equations and respectively .the terms @xmath73 and @xmath74 actually correspond to the lower and upper hubbard bands as will be detailed later . to get the lesser green s functions we will use the analytic continuation rules of the keldysh formalism .first , let us introduce the green s functions which are related to the hubbard bands in the following way @xmath75 from eqs .and we get the equations for the newly introduced green s functions @xmath76 these equations are actually of the type of standard dyson equations for the retarded green s function of the central region in the absence of electron interaction at the dot ( see ref .@xcite ) , the only difference being the change @xmath77 for the disconnected green s function in the second equation . we will introduce @xmath78 as @xmath79 for convenience .the total retarded green s function of the central region can then be expressed as @xmath80 the last three equations constitute the key results of our hia .it will be then straightforward to get the lesser green s functions and to do numerical calculations , thanks to the fact that @xmath81 and @xmath82 depend only on one time variable , namely @xmath83 .now we can get the expressions for the corresponding lesser green s functions which were derived in ref .@xcite using the langreth rules @xmath84 where the lesser self - energy @xmath85 is @xmath86 and the advanced green s function is @xmath87^*$ ] .the total lesser green s function of the central region reads @xmath88in the time - independent case the green s functions can be calculated with the use of fourier transform .let us calculate the self - energy first , and define useful quantities @xmath89.\end{aligned}\ ] ] the corresponding dyson equations for central region retarded ( advanced ) green s functions in hia are @xmath90 it is easy to obtain the explicit form for these green s functions @xmath91 from which we get the retarded ( advanced ) green s function of the central region : @xmath92 the spectral function @xmath93 $ ] is then @xmath94 ^ 2 + \left[\frac{\gamma(\omega)}{2 } \right]^2 } \nonumber \\ & & + \frac{n_{\bar\sigma } \gamma(\omega)}{\left[\omega - \varepsilon_0 - u - \lambda(\omega)\right]^2 + \left[\frac{\gamma(\omega)}{2 } \right]^2}. \label{eq : spectral}\end{aligned}\ ] ] one recovers the exact results in the atomic limit @xmath95 , and in the noninteracting @xmath96 one .it is straightforward to calculate the lesser green s function as well @xmath97 \nonumber \\& \times & \bigl ( \frac{1-n_{\bar{\sigma}}}{\left[\omega - \varepsilon_0 - \lambda(\omega)\right]^2 + \left[\frac{\gamma(\omega)}{2 } \right]^2 } \nonumber \\ & & + \frac{n_{\bar{\sigma}}}{\left[\omega - \varepsilon_0 - u - \lambda(\omega)\right]^2 + \left [ \frac{\gamma(\omega)}{2 } \right]^2 } \bigr)\end{aligned}\ ] ] in the time - independent case it is possible to perform the @xmath98 integration in eq . andto express the current as @xcite @xmath99 \right\}.\end{aligned}\ ] ] in terms of spectral function , this leads to @xmath100 , \label{eq : jlreq}\end{aligned}\ ] ] while the expression for dot occupancy for spin @xmath17 is @xmath101 where @xmath102 . here@xmath103 depends on @xmath104 , therefore is implicit. we can also write this equation in the following form @xmath105 where @xmath106 ^ 2 + \left[\frac{\gamma(\varepsilon)}{2 } \right]^2}.\nonumber \\ & & \end{aligned}\ ] ] equation , which is actually a system of two linear equations for @xmath107 and @xmath104 , can be solved explicitly to yield @xmath108 in the wide band limit , the real part of the retarded self - energy vanishes , while its imaginary part is constant @xmath109 we can now analyse the current - voltage characteristic , with the voltage defined as @xmath110 , and the symmetrized current @xmath111 calculated from .[ pic : volcur ] depicts the calculation results for the following parameters : @xmath112 .all energies are measured in @xmath113 units .the results for @xmath114 and those obtained in hartree - fock and noncrossing approximations are also shown for comparison ( this last one is discussed later ) ., in the hubbard i approximation , non interacting case , hartree - fock and noncrossing approximations ( see text ) . the differential conductance ( y axis on the right side ) for hiais also shown in gray . ]there are two jumps in @xmath115 which correspond to @xmath116 and @xmath117 .one might think that @xmath0 lowers the current , as seen in fig .[ pic : volcur ] .however , it is not a general result : the coulomb repulsion can even raise it , as observed in the upper part of fig . [pic : volcur2 ] and more obviously in the lower part . for the parameters used in these figures , a simple expression for the current , valid in the weak coupling limit ( @xmath118 ) @xmath119 associated with a weak - coupling expression for density @xmath120 , enables to evaluate the current plateau values observed in the three plots .the proximity between hia and weak - coupling predictions for bias corresponding to the middle of the plateaus is smaller than 5% . however the transition between two consecutive plateaus is too abrupt in the weak - coupling approach due to its delta - shaped spectral weight .@xmath121 increases with the number of conducting channels opened in the bias window , but not in proportion with this number .indeed the hubbard band spectral weight renormalizes each contribution , as explicitly stated in eq . .to further validate the foundations of our approach , that is hia out of equilibrium , to which the formalism reduces under steady - state conditions , we compare its predictions for the current with those evaluated within a more sophisticated approach , namely nca . in nca on - dot interaction is treated in a more reliable way than in hia and the spectral density displays a more elaborate structure : showing at low temperature the kondo resonance , and more generally wider hubbard bands ( typically four times wider ) .thus for a quantitative comparison , we choose nca parameters such as to obtain the same peak locations and bandwidths in both approaches @xcite . as can be seen in figs .[ pic : volcur ] and [ pic : volcur2 ] , we obtain an overall satisfying agreement between nca and hia .even for the considered temperature @xmath122 , where @xmath2 is the kondo temperature , some discrepancies can be observed , especially for @xmath123 , because of the incipient kondo resonance .however nca slightly overestimates the kondo resonance weight when this structure is close to @xmath124 @xcite . , but with @xmath125 ( top ) and @xmath126 ( bottom ) . ] , but with @xmath125 ( top ) and @xmath126 ( bottom ) . ]in the time - dependent case it is necessary to solve eqs . - , with @xmath127 given by eq . .it is possible to go further analytically in the case of the wide - band limit , that is neglecting the influence of bandstructure details : in that case @xmath128 is assumed to be independent of @xmath129 and the couplings to the leads become @xmath130 where @xmath131 are constant .this leads to @xmath132 , and the retarded self - energy becomes @xmath133 where @xmath134 .it is then possible to solve the equations for the retarded green s functions @xmath135 the expressions for @xmath136 were given previously in eqs . and .the total retarded green s function is then determined from eq . .it is worth noting that the analytical simple expression quoted in eq . which is very convenient for numerical evaluation ,is a direct consequence of the original manner used in this work to make the hubbard i approximation . with the canonical hia , we do not obtain such a handy result .the lesser green s function can be evaluated using the langreth analytic continuation rules , see eq . .to calculate the lesser green s functions it is useful to define , following ref @xcite @xmath137 using these functions it is possible to write the lesser green s function in a compact form @xmath138 |a_{l / r}^{0}(\varepsilon , t)|^2 \right . \nonumber \\ & & + \left .n_{\bar{\sigma}}(t ) |a_{l / r}^{u}(\varepsilon , t)|^2\right\}.\end{aligned}\ ] ] similarly to the stationary case we have equation for @xmath139 which can be solved explicitly to yield @xmath140 where @xmath141the current consists of two contributions @xmath142 , with @xmath143 , \\ \label{eq: jwb1 } j_{l / r}^{(2 ) } ( t ) & = & -\frac{e}{\hbar } \gamma^{l / r } u_{l / r } ( t ) \sum_{\sigma } \int \frac{d \varepsilon}{\pi } f_{l / r } ( \varepsilon ) \nonumber \\ & & \times \mathrm{im } \left\ { b_{l / r}^{0}(\varepsilon , t ) + b_{l / r}^{u}(\varepsilon , t)\right\}. \label{eq : jwb2}\end{aligned}\ ] ] quantities @xmath144 appear after the integration of @xmath145 over @xmath98 in the general expression for the current .they can be written as @xmath146 u_{l / r } ( t_1 ) \tilde{g}_{\sigma\sigma,0}^{r } ( t , t_1 ) \nonumber \\ & & \times e^{i \varepsilon ( t - t_1 ) } \exp\left ( -i\int_t^{t_1 } d t_2 \delta_{l / r } ( t_2 ) \right ) , \nonumber \\ & & \\ b_{l / r}^{u } ( \varepsilon , t ) & = & \int d t_1 n_{\bar{\sigma } } ( t_1 ) u_{l / r } ( t_1 ) \tilde{g}_{\sigma\sigma , u}^{r } ( t , t_1 ) \nonumber \\ & & \times e^{i \varepsilon ( t - t_1 ) } \exp\left ( -i\int_t^{t_1 } d t_2 \delta_{l / r } ( t_2 ) \right).\nonumber \\ & & \end{aligned}\ ] ] details of the numerical procedure used for calculation of current for arbitrary time dependences can be found in appendix a. in the case of a rectangular pulse shape modulation , we choose the following time dependences @xmath147 \delta_{l / r } , \label{eq : stepfirst}\\ \varepsilon_0 ( t ) & = & \varepsilon_0 + \left[\operatorname{\theta}(t ) - \operatorname{\theta}(t - s)\right ] \delta,\\ u_{l / r}(t ) & = & 1 , \end{aligned}\ ] ] it entails that @xmath148 figure [ pic : timedep1 ] depicts @xmath149 and @xmath150 for the following choice of parameters : @xmath151 , @xmath152 , for @xmath114 , and @xmath153 ( hf and hia ) , in @xmath113 unit for energy and @xmath154 unit for time . for finite @xmath0 in the hiathe current behaves similarly as in the case of no correlations , experiencing ringing ( pseudo - oscillations ) with the same period , but with reduced value when the transient regime fades .the hf approximation predicts a period which is roughly half as long . a higher current value for @xmath114 than for @xmath153 can be explained as follows : for @xmath153, only one channel corresponding to the energy transition @xmath155 lies in between @xmath156 and @xmath157 ( @xmath158 is outside the bias window ) while for @xmath114 both channels contribute to the current .the similarity between the non - interacting case and the hia , as well as the discrepancy between these and the hf result is also obvious in the dot occupancy versus time plot ( lower part of fig .[ pic : timedep1 ] ) . for @xmath114, @xmath159 always lies in the middle of the bias window , then by symmetry @xmath160 , such that the dot occupancy is not affected by the bias step : @xmath161 is therefore time - independent , and equals 1/2 .the dot occupancy in the hia is neither affected by the bias onset for the same reason , and the obtained constant value of 1/3 can be understood on time - independent grounds .this value attests the spectral weight transfer which takes place between the two hubbard bands : indeed the weight of the lower hubbard band is @xmath162 , furthermore this band lies symmetrically around the middle of the bias window , for which @xmath163 , such that one has @xmath164 , hence leading to the value @xmath165 .the occupancy in hf approach at @xmath166 is reduced compared to the non - interacting one due to the shift of the band towards higher energy .conversely and in an erroneous way , the hf result for @xmath107 is time - dependent and brings up an artificial characteristic time scale .quite generally charge conservation leads to @xmath167 where the first two terms are tunnel currents , while the last one is called displacement current .thus a time - independent dot occupancy is expected when , by symmetry , @xmath168 ., @xmath152 , for @xmath114 and @xmath153 ( hf and hia ) .the gray area depicts the pulse duration . ] , @xmath152 , for @xmath114 and @xmath153 ( hf and hia ) .the gray area depicts the pulse duration . ]+ when both bands lie in the bias window , as in fig .[ pic : timedep2 ] , the differences for current and occupancy between @xmath114 and hia for @xmath153 are attenuated . as previously noted ,the periods of pseudo - oscillations of @xmath149 are nearly the same between @xmath114 and hia , while the amplitudes are slightly different .now the density is affected by the bias setup , even for @xmath114 , due to an asymmetric distribution of spectral weight in the bias window . , but for @xmath169 . ] , but for @xmath169 . ]+ all these results , for one or two bands inside the bias window , display a transient regime which differs depending on whether the bias is turned on or off .the greatest qualitative difference between non - interacting and hia results occurs during the equilibrium restoration .we observe that the over - current values are quite close in hf and hia in fig .[ pic : timedep1 ] , but it may be fortuitous : it is not the case in fig .[ pic : timedep2 ] .the hf has an additional shortcoming , predicting a temporary sign reversal of the current immediately after the pulse end , a behavior absent in the hia and non interacting cases , which can be attributed in part to an overestimation of the dot occupancy in the steady state regime combined with an underestimation of @xmath107 in the equilibrium regime .finally we choose the voltages and local dot parameters in such a way as to visualize the coulomb blockade - conducting transition .this is shown in fig .[ pic : timedep3 ] where , at @xmath166 the dot leaves the insulating coulomb blockade region to enter the conducting one until @xmath170 .letting @xmath171 enables to access the charging time of the dot : for the present parameters , using an exponential modelization , we find @xmath172 in hia , this is about twice as long as the time predicted by hf , as seen in fig .[ pic : timedep3 ] . ,but for @xmath173 .] , but for @xmath173 . ]+ to explore further the discrepancies between hf and hia , we analyze the dependence on @xmath0 of the transferred charge or time - integrated symmetrized current . the interval between consecutive pulsesis assumed to be much larger than the length of the pulse , therefore we can treat consecutive pulses independently . since the length of the pulse @xmath174 is less than time duration of the transient regime ( see figs . [pic : timedep1]-[pic : timedep2 ] ) , this charge @xmath175 can illustrate properties of purely time - dependent phenomena , it is shown in fig .[ pic : chargeu ] . in hf and hia , for @xmath176 . ] for small @xmath0 , both hf and hia give the same result , as expected since they converge to the exact description for @xmath114 . in the hia ,increasing @xmath0 from 0 , the transferred charge @xmath175 decreases to settle at a constant value when the higher hubbard band leaves the bias window .the value @xmath177 which corresponds to @xmath178 , marks the upturn between @xmath114 and @xmath179 regimes .the smoothness of this decrease depends on @xmath113 .besides , in the hf approximation , the charge transferred @xmath175 keeps decreasing with the increase of @xmath0 .the hubbard i approximation is known to contain more physics than hf approximation , e.g. in stationary and equilibrium cases , and we can conclude that hf approximation is also insufficient to describe time - dependent transport in presence of coulomb repulsion @xcite .it is interesting to explore which parameters can influence the period of the `` ringing '' in the time - dependence of @xmath149 .when we change all the parameters @xmath180 , @xmath181 and @xmath182 , the time - dependence can be rather complicated ( see fig .[ pic : timedep2 ] ) , so we can focus on only one parameter .let us fix the values of @xmath181 and @xmath182 to zero and change only @xmath180 . figure [ pic : period ] depicts @xmath149 in case of a step - function characterized by @xmath183 , and with @xmath184 . in case of step voltage for different values of @xmath180 , for @xmath185 . ]the value of the steady - state current does not change with @xmath180 : indeed in all cases the conducting channels are the same . in the meantime , the period is strongly affected by @xmath180 : the product of @xmath180 and period @xmath186 approximately satisfies @xmath187 .this can be attributed to the presence of a phase multiplier of the form @xmath188 in the expressions which determine the time dependence of the current .this result is compatible with the observation reported in ref @xcite where the oscillations are ascribed to the electronic transitions between the lower dot state and the leads potentials .finally we can also study the temperature influence on @xmath149 .figure [ pic : temp ] depicts @xmath149 for different @xmath189 values . in case of step voltage for different temperatures , for @xmath190 . ]moderate temperature has little influence on `` ringing '' period , but can change significantly the amplitude of these oscillations .besides , very high temperature affects the steady - state value of the current , which is attained almost instantaneously .so , it is seen that complex time - dependence is restricted to low temperatures .it is interesting to explore the case when external voltages are periodic in time and how correlations , after the transient regimes , influence the forced ones . in case of harmonic modulation we choose in - phase voltages : @xmath191 .this kind of modulation was studied before for the non - interacting case in ref .@xcite and in ref .@xcite with an exponential modulation of the hybridization ; it was generalized to the interacting model in the hf approximation @xcite .such an harmonic time - dependence was also adressed in the kondo regime @xcite@xcite .figure [ pic : timedepharm ] depicts the time dependence of the current ( top ) and dot occupancy per spin ( bottom ) in case of harmonic modulation , for the following parameters : @xmath192 , @xmath193 ( pulsation measured in @xmath194 ) , for @xmath114 and @xmath153 ( hf and hia ) . , @xmath193 , for @xmath114 ( exact solution ) and @xmath153 . ] , @xmath193 , for @xmath114 ( exact solution ) and @xmath153 . ]+ in hia , the modulation amplitude for dot occupancy is very slight , but not strictly zero : it stays close to 1/3 , and never reaches 2/5 , which would be expected once per period , if the system were adiabatic .eventually , in this forced regime , the current becomes periodic in time , with the same period as the external perturbation .for the chosen parameters , a rise and fall regime sets in : the higher hubbard band alternatively reaches the border and leaves the varying bias window . however a higher frequency ( about three times larger ) also emerges .as also previously argued , the hf approximation is not reliable in the transient regime , not more in the forced regime , predicting a periodic current inversion , as well as a poor estimation of dot occupancy .as previously mentioned in the harmonic regime , current and densities do not follow adiabatic predictions , even at low frequency @xmath195 @xcite . the non - adiabatic behavior is exploited in charge pumping .the idea of producing a current in the absence of any bias voltage , called pure pumping , can also be addressed within the hia . a recent paper focussed on this issue in a formalismclose , but not equivalent , to the present work @xcite .these authors truncate the hierarchy of equation of motion at the same level than detailed above , and manage also to circumvent double - time green s functions evaluation , using an auxiliary - mode expansion .it thus seemed to us interesting to make a detailed comparison between these two resembling techniques . in the footsteps of these authors, we looked at the case of a gaussian pulse gate voltage , and calculated the left current .results are shown in fig .[ pic : gaussian - gate ] .varying the gaussian time - width @xmath196 allows to browse the cases of adiabatic and non adiabatic response .there is a qualitative agreement , and a very good quantitative consistency for rapid pulse ( low @xmath196 ) ; however some discrepancies in the adiabatic region are observed , the current being higher in our hia .the origin of this difference in not very clear for us but stems from different correlation treatment procedures ; indeed for @xmath114 our numerical procedure for arbitrary time dependence was checked by a detailed comparison with jauho s et al .results @xcite , we also agree with uncorrelated croy _s results . , in the hia for different values of @xmath196 , the other parameters are @xmath197 . ]a time - dependent formalism for a single level interacting quantum dot coupled to two leads has been developed within the hubbard i approximation , in a convenient and handy way for numerical evaluation .it enables us to consider general time dependences for hybridization as well as for gate or bias voltages . to validate the approach , the steady - state regime has been , in some parameter range , favorably compared with nca results .the formalism results in the appearance of two hubbard bands in the local spectral density . for those bands we introduce two green s functions , which offer the key advantage to get rid of double - time evaluation , without any a priori assumption about adiabatic or sudden limits .these bands undergo spectral weight transfers which influence the transport properties , in contrast with the rigid band frame .calculations show the influence of coulomb correlations on the current , which mainly consists in a change of amplitude .the influence on the time structure ( e.g. `` ringing '' period ) , appear to be mostly insignificant .comparisons between hubbard i and hartree - fock approximations show that the latter is insufficient to describe time - dependent transport .the presented method can also be extended beyond the wide band limit , indeed the key point consists in writing the dot green s function as a sum of two independent green s functions ; this acts upstream , before the assumption of wide band limit . to implement our formalism beyond this limitdeserves further study .we thank steffen schfer and oleh fedkevych for valuable discussions .v.v . appreciates hospitality of staff at im2np in marseilles , where part of this work was done and acknowledges support from the ministry of education and science of ukraine ( program `` 100 + 100 + 100 '' for studying and interning abroad ) .d.a . was supported by the program `` microscopical and phenomenological models of fundamental physical processes at micro and macro scales '' ( section of physics and astronomy of the nas of ukraine ) .here we describe the numerical procedure used for calculating the current and the dot occupancy for arbitrary time dependences in the wide - band limit for the hia .the time dependences are @xmath198 , @xmath199 , @xmath200 with the only condition being that all time - dependent perturbations start at @xmath201 . for the sake of simplicity we choose in this appendix @xmath202 ,i.e. for @xmath203 we have a stationary state .the time - dependent current and dot occupancy are calculated by integrating the @xmath204 and @xmath205 functions .the key point here is the numerical computation of these functions .we will split the integration , e.g. in the expression for @xmath206 into two parts : @xmath207 . then , after performing the integration in the first term we get @xmath208}{\varepsilon-\varepsilon_{0(u)}+\frac{i\gamma}{2 } } \nonumber \\ & & \times \exp\left[-i\int_0^{t } dt_1 \left(\delta(t_1)-\delta_{l(r)}(t_1 ) - \frac{i\gamma(t_1)}{2}\right ) \right ] \nonumber \\ & & - i \int_{0}^t d t_1 \exp\left[i \varepsilon ( t - t_1 ) \right ] u_{l / r}(t_1 ) \exp\left [ -i \int_{t_1}^t d t_2 \varepsilon_{0(u ) } ( t_2 ) \right ] \nonumber \\ & & \times \exp\left [ i \int_{t_1}^t d t_2 \delta_{l / r } ( t_2 ) \right ] \exp\left[- \int_{t_1}^t \frac{\gamma(t')}{2 } d t ' \right ] .\label{eq : a1wb}\end{aligned}\ ] ] expressions for @xmath144 are also expressed in this form , for instance we get @xmath209}{\varepsilon-\varepsilon_0+\frac{i\gamma}{2 } } \nonumber \\ & & \exp\left[-i\int_0^{t } dt_1 \left(\delta(t_1)-\delta_{l(r)}(t_1 ) - \frac{i\gamma(t_1)}{2}\right ) \right ] \nonumber \\ & & - i \int_{0}^t d t_1 \left(1-n_{\bar\sigma}(t_1)\right ) \exp\left[i \varepsilon ( t - t_1 ) \right ] u_{l / r}(t_1 ) \nonumber \\ & & \times \exp\left [ -i \int_{t_1}^t d t_2 \varepsilon_0 ( t_2 ) \right ] \exp\left [ i \int_{t_1}^t d t_2\delta_{l / r } ( t_2 ) \right ] \nonumber \\ & & \times \exp\left[- \int_{t_1}^t \frac{\gamma(t')}{2 } dt ' \right],\end{aligned}\ ] ] where @xmath210 is the stationary value prior to modulation .@xmath217 \nonumber \\ & & - i \int_{t}^{t+dt } d t_1\exp\left[i \varepsilon ( t+dt - t_1 ) \right ] u_{l / r}(t_1 ) \times \nonumber \\ & & \exp\left [ -i \int_{t_1}^{t+dt } d t_2 \varepsilon_{0(u ) } ( t_2 ) \right ] \times \exp\left [ i \int_{t_1}^{t+dt } d t_2 \delta_{l / r } ( t_2 ) \right ] \nonumber \\ & & \exp\left[- \int_{t_1}^{t+dt } \frac{\gamma(t')}{2 } d t ' \right].\end{aligned}\ ] ] because @xmath218 is small we can use the midpoint method to perform integration from @xmath83 to @xmath219 @xmath220 \nonumber \\ & & - i dt \exp\left[i \varepsilon \frac{dt}{2 } \right ] u_{l / r}(\tilde{t } ) \nonumber \\ & & \times \exp\left [ -i \frac{dt}{2 } \left ( \varepsilon_{0(u ) } ( \tilde{\tilde{t } } ) - \delta_{l / r } ( \tilde{\tilde{t } } ) - \frac{i\gamma(\tilde{\tilde{t}})}{2}\right)\right],\end{aligned}\ ] ] where @xmath221 and @xmath222 . for @xmath144it is possible to obtain similar approximation but @xmath223 also appears in the integration from @xmath83 to @xmath219 .we will use @xmath224 .indeed we can compute @xmath225 before computing @xmath226 because it is determined only by @xmath227 . in the end , we get , e.g. for @xmath228 @xmath229 \nonumber \\ & & - i dt \frac{2 - n_{\bar\sigma } ( t ) - n_{\bar\sigma } \left(t + dt\right)}{2 } \exp\left[i \varepsilon \frac{dt}{2 } \right ] u_{l / r}(\tilde{t } ) \nonumber \\ & & \times \exp\left [ -i \frac{dt}{2 } \left ( \varepsilon_0 ( \tilde{\tilde{t } } ) - \delta_{l / r } ( \tilde{\tilde{t } } ) - \frac{i\gamma(\tilde{\tilde{t}})}{2}\right)\right].\end{aligned}\ ] ]

<S> the time - dependent transport through a nano - scale device , consisting of a single spin - degenerate orbital with on - site coulomb interaction , coupled to two leads , is investigated . various gate and bias voltage time - dependences are considered . </S><S> the key and new point lies in the proposed way to avoid the difficulties of the usual heavy computation when dealing with two time green s functions within keldysh formalism . </S><S> the time - dependent retarded dot green s functions are evaluated , in an efficient manner within a non - canonical hubbard i approximation . </S><S> calculations of the time - dependent current are then presented in the wide - band limit for different parameter sets . </S><S> a comparison between the method and the hartree - fock approximation is performed as well . </S><S> it is shown that the later can not account reliably for dynamical aspects of transport phenomena . </S>

cond-mat0402227

recently , several papers have focused on the blow - up properties of a cluster of self - attracting particles @xcite .this concerns in particular the `` isothermal collapse '' @xcite of self - gravitating brownian particles and the `` chemotactic aggregation '' @xcite of bacterial populations in biology .these systems have a similar mathematical structure and they are described , in a first approximation , by the smoluchowski - poisson system @xcite .the smoluchowski equation is a particular fokker - planck equation in physical space involving a diffusion due to brownian motion and a drift . in the standard brownian theory developed by einstein ,the drift is due to an external potential @xcite .alternatively , we can consider the more complicated situation in which the potential is produced by the particles themselves . if we assume a newtonian type of interaction , we have to couple the smoluchowski equation to the poisson equation .it can be shown that the smoluchowski - poisson system decreases a lyapunov functional @xmath5 $ ] which can be interpreted as a boltzmann free energy @xcite .furthermore , the type of evolution depends on the value of a dimensionless parameter @xmath3 . for @xmath6 ,the smoluchowski - poisson system evolves to an equilibrium state which minimizes the free energy at fixed mass .it corresponds to an isothermal distribution of particles similar to isothermal stars and isothermal stellar systems @xcite . on the contrary , for @xmath7, there is no possible equilibrium state and the system blows up .this is the case for self - gravitating brownian particles below a critical temperature @xmath8 and for bacterial populations above a critical mass @xmath9 .it is found that the collapse is self - similar and that it develops a finite time singularity , _i.e. _ the central density becomes infinite in a finite time @xmath0 @xcite .then , a dirac peak is formed in the post - collapse regime @xcite .it is clear that the collapse time @xmath0 depends on the distance to the critical point @xmath4 and that it should diverge as @xmath10 .more precisely , by using heuristic arguments , it is argued in @xcite that @xmath11 .this scaling is consistent with numerical simulations of the smoluchowski - poisson system .however , the approach of @xcite is qualitative and does not provide the numerical value of @xmath12 .after some introductory material presented in section 2 , we develop a systematic procedure in section 3 , inspired from @xcite , which confirms the scaling law @xmath13 and leads to the explicit value of @xmath12 . in section 4 , we study the relaxation time @xmath14 to equilibrium below @xmath15 and show that it diverges like @xmath16 , where @xmath17 is given explicitly . in section 5 , we derive a systematic large time expansion of the density profile exactly at @xmath15 . in particular , we show that the central density approaches its equilibrium value according to @xmath18 , where @xmath19 is a universal constant which is explicitly computed .finally , in section 6 , we determine the pulsation period of bounded isothermal spheres described by the euler - jeans equations close to the critical point .in the overdamped regime , a system of self - gravitating brownian particles is described by the @xmath20 coupled stochastic equations @xmath21 where @xmath22 is the mobility ( @xmath23 is the friction coefficient ) , @xmath24 is the diffusion coefficient and @xmath25 is a white noise satisfying @xmath26 and @xmath27 , where @xmath28 refer to the coordinates of space and @xmath29 to the particles .we define the temperature @xmath30 through the einstein relation @xmath31 .the particles interact via the potential @xmath32 . in this paper, @xmath33 is the newtonian binary potential in @xmath34 dimensions . starting from the @xmath20-body fokker - planck equation and using a mean - field approximation @xcite, we can derive the nonlocal smoluchowski equation @xmath35 where the potential @xmath36 is produced by the density @xmath37 according to the poisson equation @xmath38 the self - gravitating brownian gas has a rigorous canonical thermodynamic structure @xcite .it can be seen therefore as the canonical counterpart of the usual @xmath20-stars problem governed by newton s equations and possessing a microcanonical structure . on the other hand , in a biological context , a model of chemotactic aggregation exhibiting blow - up phenomenais provided by the system of equations @xmath39 @xmath40 where @xmath41 is the concentration of the biological population ( amoebae ) , @xmath42 is the concentration of the substance secreted ( acrasin ) and @xmath43 measures the strength of the chemotactic drift . equations ( [ ge5 ] ) and ( [ ge6 ] ) can be viewed as a simplification of the keller - segel model @xcite .clearly , the two models are isomorphic provided that we make the identification @xmath44 , @xmath45 and @xmath46 .introducing the mass @xmath47 of the system and the radius @xmath48 of the domain , we can show that the problem depends on the single dimensionless parameter @xmath49 @xcite .therefore , a large value of @xmath3 corresponds to a small temperature @xmath50 or a large mass @xmath51 . in the following, we shall use the gravitational terminology but we stress that our results are equally valid for the biological problem due to the above analogy . by rescaling the physical parameters adequately ,it is possible to take @xmath52 without restriction .then , the equations of the problem are @xmath53 with proper boundary conditions in order to impose a vanishing particle flux on the surface of the confining sphere .if we restrict ourselves to spherically symmetric solutions , the boundary conditions are @xmath54 with this rescaling , the equations only depend on the temperature @xmath50 , or equivalently on the parameter @xmath55 . as mentioned previously ,a small temperature is equivalent ( through a rescaling ) to a large mass . integrating eq .( [ dim2 ] ) once , we can rewrite the smoluchowski - poisson system in the form of a single integrodifferential equation @xmath56 the smoluchowski - poisson system is also equivalent to a single differential equation @xmath57 for the quantity @xmath58 which represents the mass contained within the sphere of radius @xmath59 .the appropriate boundary conditions are @xmath60we know that the smoluchowski - poisson system blows up for @xmath61 , where @xmath8 is a critical temperature below which there is no equilibrium state @xcite .we shall place ourselves close to the critical point and expand the mass profile as @xmath62 where @xmath63 is the equilibrium profile at @xmath8 and @xmath64 is a small parameter defined as @xmath65 we also rescale the time according to @xmath66 if we substitute the expansion eq .( [ dint1 ] ) in eq .( [ sca ] ) and equal terms of same order , we get to order @xmath67 : @xmath68 to order @xmath69 : @xmath70 and to order @xmath71 : @xmath72 the boundary conditions are @xmath73 @xmath74 1.5 cm 1.5 cm at equilibrium , the condition of hydrostatic balance @xmath76 combined with the gauss theorem @xmath77 leads to the relation @xmath78 since @xmath79 , we obtain the differential equation @xmath80 introducing the function @xmath81 through the relation @xmath82 where @xmath83 is the central density , and using the normalized distance @xmath84 where @xmath85 , we can rewrite the relation eq .( [ dint9 ] ) in the form @xmath86 furthermore , according to eq .( [ dint10 ] ) , the function @xmath81 is solution of the emden equation @xmath87 @xmath88 the taylor expansion of the emden function near the origin is @xmath89 using these results , we can check that the emden equation ( [ emden ] ) is equivalent to eq .( [ dint4 ] ) , as it should . for a given temperature @xmath90, the emden equation has to be solved from @xmath91 to @xmath92 such that @xmath93 for the profile we are considering , at the critical temperature @xmath8 , we have to stop the integration at @xmath94 such that @xmath95 is minimum .the condition @xmath96'(\alpha_{c})=0 $ ] is equivalent to @xmath97 it is found numerically that @xmath98 and @xmath99 . since eq .( [ dint5 ] ) is linear , we look for solutions of the form @xmath101 the function @xmath100 satisfies the differential equation @xmath102 which is equivalent to @xmath103 where we have used the emden equation eq .( [ emden ] ) . equation ( [ dint19 ] ) is precisely the equation found at the critical point @xmath8 by analyzing the linear stability of isothermal spheres @xcite . indeed , considering a perturbation @xmath104 around a stationary solution @xmath105 of eq .( [ sca ] ) , we get @xmath106 if @xmath107 , the stationary solution is stable and if @xmath108 , the stationary solution is unstable . at the critical point @xmath8where @xmath109 ( marginal stability ) , we recover eq .( [ dint5 ] ) with @xmath110 .equation ( [ dint19 ] ) has to be solved with the boundary conditions @xmath111 we find that @xmath100 can be simply expressed in terms of @xmath112 as @xcite : @xmath113 the condition @xmath114 determines the critical temperature @xmath8 .( [ dint21 ] ) , this condition is equivalent to @xmath115 comparing this result with eq .( [ dint16 ] ) , we find that the criterion of marginal stability ( @xmath109 ) corresponds to the point @xmath94 in the series of equilibria @xmath116 where the temperature is minimum ( @xmath96'(\alpha_{c})=0 $ ] ) @xcite . using the previous results , eq .( [ dint6 ] ) can be rewritten @xmath118 we now multiply eq .( [ dint22 ] ) by a function @xmath119 and integrate between @xmath67 and @xmath120 .we choose the function @xmath43 ( see below ) so as to eliminate the terms where @xmath121 appears .we are thus left with @xmath122 where @xmath123 for future convenience , we note that @xmath124 by construction , the function @xmath43 satisfies the integral relation @xmath125 integrating by parts and using @xmath126 , we obtain @xmath127 thus , we impose that @xmath43 is a solution of the differential equation @xmath128 with the boundary condition @xmath129 it turns out that the solution of this equation can be expressed in terms of the emden function as @xmath130 with @xmath131 . the function @xmath119 is represented in fig . 3 along with the function @xmath100 . using eq .( [ dintconda ] ) , we readily check that @xmath132 .therefore , the condition eq .( [ dint26 ] ) is satisfied .accordingly , @xmath117 is determined by eq .( [ dint23 ] ) , where @xmath133 and @xmath134 have to be evaluated using eq .( [ dint25 ] ) .1.5 cm the solution of eq .( [ dint23 ] ) with @xmath135 as @xmath136 is @xmath137 returning to the original time variable , we get @xmath138 then , we find that the blow up occurs for @xmath139 with @xmath140 we find numerically that @xmath141 and @xmath142 .we thus get @xmath143 this asymptotic result is compared in fig . 4 with the exact law @xmath144 obtained by solving numerically the smoluchowski - poisson system . 1.2 cmwe now assume @xmath145 , so that an equilibrium state exists . we look for a time dependent solution for @xmath146 which converges exponentially to the equilibrium profile @xmath147 the purpose of this section is to compute the leading asymptotic form of the relaxation time @xmath148 as @xmath50 goes to @xmath149 . after inserting this @xmath150 into the equation of motion for @xmath146, we obtain @xmath151 where @xmath152 we have used @xmath59 as a coordinate ( instead of @xmath23 ) so as to remain within a fixed interval of space @xmath153 as we make the perturbative expansion in @xmath154 ( see below ) . when @xmath155 , we have @xmath156 , @xmath157 , and @xmath158 , where @xmath100 is given by eq .( [ dint21 ] ) .we now expand eq .( [ eqprop ] ) in power of @xmath159 , by introducing @xmath160 and @xmath161 such that @xmath162 and @xmath163 the relation between @xmath164 and @xmath165 will be given later .collecting terms proportional to @xmath165 , and setting @xmath166 , we obtain @xmath167 subject to the boundary condition @xmath168 .this eigenvalue problem selects a unique value for @xmath161 , hence for @xmath14 .we were not able to solve this problem analytically .however , we remark that this equation is similar to eq .( [ dint22 ] ) , without the term leading to @xmath134 .we can therefore use the same trick as in sec .[ sec_two ] . after multiplying eq .( [ eqmu ] ) by the function @xmath43 defined in the previous section , we end up with @xmath169 the direct numerical solution of eq .( [ eqmu ] ) leads of course to the same value .the function @xmath160 is plotted in fig .we now find a simple relation between @xmath164 and @xmath165 close to @xmath149 @xcite .using the condition @xmath170 and the fact that @xmath171 , we obtain @xmath172 which up to leading order in @xmath165 can be written in the form @xmath173 finally , we obtain the following diverging behavior for @xmath14 : @xmath174 numerically , we get @xmath175 with @xmath176 note that a heuristic argument leading to @xmath177 was given in @xcite , although the constant @xmath178 could not be calculated .strictly at @xmath179 , @xmath180 , and one expects a slow convergence to the equilibrium profile @xmath181 .let us write @xmath182 we implicitly assume that @xmath183 as @xmath184 .moreover , the presence of the minus signs in eq .( [ devt ] ) anticipates the fact that starting from a flat density the central density increases . in @xcite , it was argued based on a heuristic argument that @xmath185 .it is the purpose of this section to justify this statement as well as to give more quantitative estimates .let us postulate for the moment the natural choice @xmath186 which will be justified hereafter .inserting again this @xmath150 in the dynamical equation for @xmath146 , and collecting terms of order @xmath187 and @xmath188 , we immediately find that the first equation leads to @xmath189 where the constant has been set for later convenience , as multiplying @xmath190 and dividing @xmath191 by the same constant leaves the expansion for @xmath51 unchanged .setting @xmath192 , we find that @xmath193 satisfies @xmath194 with @xmath195 which must be a constant independent of time .thanks to the proper choice of the constant in eq .( [ m1mu ] ) , we find that eq .( [ eqlambda ] ) and eq .( [ eqmu ] ) are identical .the only physical solution with @xmath196 corresponds to the eigenvalue @xmath197 .finally , we find @xmath198 once the signs have been fixed in eq .( [ devt ] ) , we indeed find that the small time correction to the central density is necessarily negative , as @xmath190 , @xmath199 ( or @xmath193 ) and @xmath191 are found to be positive .therefore , the equilibrium profile @xmath63 is approached from below , which is natural since an excess of mass would provoke gravitational collapse .let us illustrate this result by calculating the correction to the central density .using eq .( [ m1mu ] ) and eq .( [ atmu ] ) , and the fact that @xmath200 and @xmath201 , we find that for large time the central density converges to @xmath202 from below in a universal manner @xmath203 with @xmath204 this result is illustrated quantitatively in fig .1.5 cm finally , we discuss the _ a priori _ unjustified choice @xmath205 .the case @xmath206 is clearly impossible as it amounts to take @xmath207 in eq .( [ eqlambda ] ) , leading to an equation which is never satisfied by @xmath208 , whatever the choice for @xmath187 .the opposite case @xmath209 leads to an equation similar to eq .( [ eqlambda ] ) but where the term proportional to @xmath210 is absent .for this equation , it is clear that if a solution @xmath211 exists for a specific @xmath212 , then @xmath213 is also solution associated to @xmath214 .this implies that the only solution satisfying the boundary condition @xmath196 is actually the null function associated to @xmath109 , which is not permitted for @xmath145 .hence @xmath205 is the only possibility leading to a physically sensible solution .in sec . [ sec_relax ] , we have determined the relaxation time of self - gravitating brownian particles described by the smoluchowski - poisson system close to the critical point @xmath8 .writing the perturbation as @xmath215 where @xmath216 , the eigenvalue equation for @xmath214 can be written in dimensional form as @xcite : @xmath217 where @xmath218 .this equation is the same as eq .( [ eqprop ] ) . using the result of sec .[ sec_relax ] , and returning to dimensional parameters , the inverse relaxation time is given by @xmath219 the sign @xmath220 corresponds to @xmath221 and leads to exponentially damped pertubations ( stable ) .the sign @xmath222 corresponds to @xmath223 and leads to exponentially growing pertubations ( unstable ) .we now consider the case of gaseous self - gravitating systems ( stars ) described by the euler - jeans equations : @xmath224 @xmath225 @xmath226 we use an isothermal equation of state @xmath227 and we assume that the system is confined within a region of size @xmath48 .these equations provide the usual starting point for the analysis of the gravitational stability of gaseous systems .the case of an infinite homogeneous medium was first investigated by jeans @xcite in his classical study .the case of a finite inhomogeneous medium was considered by chavanis @xcite . in that case , the equation of pulsation can be written in the form @xmath228 where , as before , @xmath215 . comparing with eq .( [ gs1 ] ) , we see that the results of sec .[ sec_relax ] remain valid provided that @xmath229 is replaced by @xmath230 .therefore , eq . ( [ gs2 ] ) translates into @xmath231 for @xmath223 , the eigenvalue @xmath232 can be positive implying instability . for @xmath221 , the eigenvalue @xmath233 is purely imaginary so that the time evolution of the perturbation has an oscillatory nature .close to the critical point , the pulsation period is given by @xmath234in this paper , we have obtained the asymptotic expressions of the blow - up time and relaxation time of self - gravitating brownian particles and biological populations close to the critical point of collapse in @xmath34 .an excellent agreement is obtained by solving numerically the smoluchowski - poisson system .this study confirms and improves the results obtained in @xcite on the basis of heuristic arguments. a possible extension of this work would be to consider the case of polytropic distributions arising in case of anomalous diffusion @xcite .more generally , we could consider the generalized smoluchowski equation ( 81 ) proposed in @xcite which is valid for an arbitrary equation of state .this equation can provide ( among other applications ) a generalized chemotactic model valid when the diffusion coefficient or the drift term depends on the density .these extensions will be considered in future works .one of us ( p.h.c ) acknowledges interesting discussions with n.i .kavallaris and d.e .tzanetis during the workshop on nonlocal elliptic and parabolic problems held in bedlewo .

<S> we determine an asymptotic expression of the blow - up time @xmath0 for self - gravitating brownian particles or bacterial populations ( chemotaxis ) close to the critical point . </S><S> we show that @xmath1 with @xmath2 , where @xmath3 represents the inverse temperature ( for brownian particles ) or the mass ( for bacterial colonies ) , and @xmath4 is the critical value of @xmath3 above which the system blows up . </S><S> this result is in perfect agreement with the numerical solution of the smoluchowski - poisson system . </S><S> we also determine the asymptotic expression of the relaxation time close but above the critical temperature and derive a large time asymptotic expansion for the density profile exactly at the critical point . </S><S> laboratoire de physique thorique ( umr 5152 du cnrs ) , universit paul sabatier , + 118 , route de narbonne , 31062 toulouse cedex 4 , france + e - mail : _ [email protected] , [email protected]_ </S>

1411.1915

in this paper we introduce a new mathematical value @xmath0 , called the spherical ratio of two points .such ratio is a generalization of the ratio of two segments originally introduced by hermann schwartz at the end of 19-th century , ( @xcite , pp .359 - 361 ) or ( @xcite , pp .168 - 169 ) .while studying complex analysis , he found that for a point inside a planar disk , the poisson kernel can be written as a ratio of two segments .for the notations in the formula below , look at figure [ geo_inte_schwarz ] .+ @xmath1 we shall see that a similar ratio can be defined for every two different points in a multidimensional space and can serve as a useful tool for integration of some functions over sphere .the technique of integration together with the results obtained here can be applied for spectral analysis in a hyperbolic space .this happens because the ratio of two segments represents an eigenfunction of the hyperbolic laplacian , while the averaging of @xmath0 over sphere represents a radial eigenfunction of the hyperbolic laplacian .therefore , all statements obtained in this paper can be restated in terms of radial eigenfunctions and used to obtain some results related , for example , to dirichlet eigenvalue problem in a disc of constant negative curvature .the main goal of this article is to describe the constraints for complex numbers @xmath2 and for a fixed point @xmath3 under which the following equivalence holds .@xmath4 where @xmath5 is the spherical ratio of two points that is defined below .[ spherical_ratio_definition ] let @xmath6 and assume that @xmath7 .let @xmath8 denotes the @xmath9-dimensional sphere of radius @xmath10 centered at the origin @xmath11 .then , let us introduce @xmath12 and @xmath13 as follows .if the line defined by @xmath14 and @xmath15 is tangent to @xmath16 , then we set @xmath17 .otherwise , @xmath12 be the point of @xmath16 such that @xmath18 and @xmath19 are collinear .note that @xmath20 in case @xmath21 .similarly , if the line through @xmath22 is tangent to @xmath23 , then we set @xmath24 .otherwise , @xmath13 be the point of @xmath23 such that @xmath25 and @xmath26 are collinear .note again that @xmath27 in the case @xmath21 .then , we can observe that @xmath28 and denote @xmath29 we call the ratio of these two segments @xmath30 and @xmath31 , as the spherical ratio of two points @xmath14 and @xmath15 , i.e. , @xmath32 where the last expression is exactly the 2-d poisson kernel for a planar disk . using the ratio of two segments @xmath30 and @xmath31 obtained above, we observe that @xmath33 remains constant as long as @xmath14 stays at a circle tangent to @xmath23 and centered at some point of the line @xmath34 .this follows from the lemma [ omega_constant - lemma ] , p. in appendix .then , the combination of lemma [ omega_constant - lemma ] and corollary [ limit - for - omega ] from p. shows that @xmath35 $ ] .[ basic - theorem ] let @xmath36 , @xmath37 , be a @xmath9-dimensional sphere of radius @xmath10 centered at the origin @xmath11 , @xmath38 such that @xmath39 and @xmath33 is defined in .until further notice we assume that @xmath10 and a point @xmath40 are fixed .then , the following statements hold .( a ) : : [ sphe - pro - direct - statement ] if @xmath41 and @xmath42 , then @xmath43 ( b ) : : [ sphe - pro - inverse - real - statement ] if @xmath44 are real , then @xmath45 ( c ) : : [ sphe - pro - liuville - statement ] for every @xmath46 there are infinitely many numbers @xmath47 such that @xmath48 ( d ) : : [ sphe - pro - inverse - complex - statement ] suppose that for the fixed point @xmath40 @xmath49 then @xmath50 statement ( c ) shows that the implication introduced in fails if we just let @xmath51 and @xmath52 be complex .however , a certain additional restriction formulated in statement ( d ) for the complex numbers @xmath44 allows us to obtain precisely the same implication as in .observe that statement ( b ) is a special case of statement ( d ) . indeed , if @xmath44 are real , then holds for every @xmath53 .[ imaginary - part - vanishing - corollary ] @xmath54 moreover , for every natural number @xmath55 @xmath56 and for every @xmath47 @xmath57 which is the taylor decomposition of @xmath58 at @xmath59 .[ sphe - pro - distance - basic - statement ] if @xmath60 and @xmath61 , then @xmath62 it is clear that formula can be obtained from statement ( a ) if we rewrite formula in terms of distance between @xmath14 and @xmath15 .in particular , for @xmath63 and for every @xmath40 , formula yields @xmath64 note that for @xmath65 the integrand on the left is the newtonian potential in @xmath66 and the integrand on the right is exactly the poisson kernel in @xmath66 . [ one - dimension - sphe - pro ] if @xmath67 , then for every @xmath68 and @xmath69,@xmath70 [ basic - th - integrable - case - general - item ] if @xmath71 , then @xmath72 in particular , if @xmath73 , @xmath74we are going to prove successively all of the statements ( a ) , ( b ) , ( c ) and ( d ) from the theorem . * case 1 ( @xmath75 ) . * for notation refer to figure [ change ] .let @xmath15 be an arbitrary point of @xmath76 . here , @xmath77 is defined as an intersection of the sphere @xmath76 and line @xmath78 .angle @xmath79 is defined as @xmath80 and @xmath81 is defined as @xmath82 .the circle @xmath83 appears on the cross - sectional plane defined by the points @xmath11 , @xmath14 and @xmath84 .let @xmath85 and @xmath86 and let the segments @xmath87 be perpendicular to @xmath88 .one can easily check that @xmath89 since @xmath90 and , by pythagorean theorem , @xmath91 . in the next stepwe are going to use the change of variables from @xmath79 to @xmath92 , introduced also by hermann schwarz in @xcite ( pp .359 - 361 ) .notice , first , that @xmath93 is similar to @xmath94 and therefore , @xmath95 it is not hard to see that @xmath96 therefore , combined with yields @xmath97 let @xmath98 and @xmath99 be orthogonal projections of @xmath15 and @xmath77 to line @xmath88 respectively .clearly , @xmath100 is similar to @xmath101 and thus @xmath102 using , , and , we have @xmath103 where @xmath104 is the area of @xmath105-dimensional unit sphere .the formula @xmath106 holds , since our integrand @xmath33 depends only on angle @xmath79 when @xmath14 is fixed .* case 2 ( @xmath108 ) .* let @xmath109 and @xmath110 .then , one can easily check that @xmath111 for every @xmath112 . clearly , the point @xmath113 is inside the sphere @xmath76 and then we can apply the result obtained in formula .this yields @xmath114 this completes the proof of statement ( a ) from p. .fix a real number @xmath51 .we have to show that the equation @xmath115 has only two solutions with respect to the variable @xmath116 ; namely , @xmath117 and @xmath118 .first note that , since @xmath119 , we have @xmath120 for every @xmath121 .therefore , the function @xmath122 is symmetric with respect to the point @xmath123 .the leibnitz rule of differentiation under the integral sign yields @xmath124 we can apply the same rule one more time , @xmath125 therefore , the graph of our function @xmath126 , plotted in the @xmath127-plane , is convex and symmetric with respect to the line @xmath123 .this means that @xmath123 is the minimum point for @xmath122 .therefore , for every @xmath128 , the equation @xmath129 has only two solutions @xmath117 and @xmath118 .this completes the proof of statement ( b ) from p .. let @xmath130 , where @xmath131 .then , the statement ( c ) can be obtained as a consequence of the following picard s great theorem .[ picard - great - theorem ] let @xmath132 be an isolated essential singularity of @xmath133 .then , in every neighborhood of @xmath134 , @xmath133 assumes every complex number as a value with at most one exception infinitely many times , see @xcite , ( p.240 ) .indeed , the direct computation shows that @xmath135 is an entire function , which means that @xmath136 is an isolated singularity of @xmath135 .note that the taylor decomposition presented in , p. shows that @xmath136 is the essential singularity for @xmath58 , since all even taylor coefficients are positive .it is clear also , we can not meet an exception mentioned in picard s theorem above , since we are looking for complex numbers @xmath116 , satisfying @xmath137 for some @xmath51 .this means that the value @xmath58 is already assumed by @xmath135 , and then , by picard s great theorem , @xmath135 must attain this value infinitely many times in every neighborhood of an isolated essentially singular point , i.e. , at every neighborhood of @xmath136 , in our case .this completes the proof of statement ( c ) .denote @xmath138 recall that @xmath10 and @xmath139 are fixed .then , @xmath140 where @xmath141 , @xmath142 $ ] and it is clear that @xmath143 for every possible @xmath139 . observe also that @xmath144 is increasing while @xmath142 $ ] .therefore , @xmath145 let @xmath146 be an arbitrary non - negative number such that @xmath147 the direct computations show that the second inequality in , in itself , implies @xmath148 therefore , combining and , we have @xmath149 or , equivalently , @xmath150\,.\ ] ] if @xmath151 and @xmath152 , then , clearly , implies @xmath153\,.\ ] ] hence , @xmath154\ ] ] and for every @xmath142 $ ] . fix a number @xmath155 satisfying or , equivalently , p. and introduce the set @xmath156 consider the function @xmath157 note now that statement ( e ) will be proven if we show that for any @xmath158 , the equation @xmath159 has only two solutions in @xmath160 , i.e. , @xmath161 and @xmath162 . to accomplish this goallet us split our argument into the following four steps .: : fix any @xmath158 and set the equation @xmath159 as the following system of two equations @xmath163 then , we establish a symmetry property for @xmath164 and @xmath165 .we shall see that @xmath164 is symmetric with respect to the two lines @xmath166 and @xmath167 , while @xmath165 is skew - symmetric with respect to the same lines .this step is described on p. .: : here we describe all solutions inside @xmath160 for the equation @xmath168 . because of the symmetry mentioned in step 1 it is enough to describe the set of solutions in the first quadrant @xmath169 $ ] .the description will be presented in proposition [ level - curve - description ] , page .: : here we shall study the behavior of @xmath165 along any of the @xmath170 curves from step 2 .proposition [ uniquiness - total - in - first - quadrant ] , p. shows that @xmath165 is a strictly increasing function along any of the level curves inside the first quadrant @xmath169 $ ] , which implies that the system has the unique solution in the first quadrant for any @xmath171 $ ] , i.e. , the solution is @xmath52 itself .this step is carried out on p. .: : here we study the behavior of @xmath165 in the whole strip @xmath160 .we shall see that for every @xmath171 $ ] there exist only two solutions of the system , i.e. , @xmath172 and @xmath173 .the same argument will show that for every @xmath174 there are at most two possible solutions of system , i.e. , @xmath175 and @xmath173 .this step is carried out on p. .now let us follow this plan described in the steps 1 - 4 above .[ step_curve_1 ] it is convenient to introduce the notation for the real and the imaginary parts of @xmath176 . since @xmath177 and @xmath139 are fixed , we can denote @xmath178 and similarly , @xmath179 fix some @xmath158 and rewrite system in the new notation .@xmath180 which is going to be solved .as we saw in statement ( a ) , @xmath181 for every @xmath182 and then , the direct computation yields @xmath183 which pictured on the diagram below , see figure [ re - symmetry ] .+ the symmetry relations presented in imply that @xmath184 is symmetric with respect to the two perpendicular lines @xmath166 and @xmath167 in the plane @xmath185 .similarly , @xmath186 is symmetric with respect to point @xmath187 and skew - symmetric with respect to the lines @xmath167 and @xmath166 .[ step_curve_2 ] because of the symmetry , without loss of generality , we may assume that @xmath188\}$ ] , which is the upper - right quadrant .now we are ready to solve the equation @xmath189 .first we need to describe all solutions for the real parts of this equation , i.e. , @xmath168 .let us define two quadrants @xmath190 , @xmath191 and the level set @xmath192 by @xmath193 , \\ & s(a , b)=\{(\xi,\zeta)\in\overline{q}\mid w(\xi,\zeta)=w(a , b)\}\ , .\end{split}\ ] ] it is clear that @xmath191 is the closure of @xmath190 . the following proposition describes all possible types of such level sets .[ level - curve - description ] 1 . for every fixed @xmath194 the set of solutions @xmath192 for the equation@xmath195 is a continuous curve @xmath196 .the function @xmath197 is @xmath198 and @xmath199 for every point @xmath200 such that @xmath201 .3 . no two of these curves have common points in @xmath191 .according to theorem 6.1 , ( @xcite , p. 128 ) , if a level curve starts at the corner @xmath202 , then it bisects the corner and if a level curve starts at the lower or at the left edge of @xmath191 , then it must be perpendicular to the edge .no two of these level curves have a common point in @xmath191 .first , we need to analyze the behavior of partial derivatives of @xmath203 in @xmath191 .[ partial - dw - d - xi - claim ] @xmath204\ , ; \\ & \frac{\partial w(\xi,\zeta)}{\partial\xi}=0\quad\text{if}\quad\xi= \frac{k}{2}\quad\text{and}\quad\zeta\in[0,p]\ , .\end{split}\ ] ] note that @xmath203 is @xmath198 function of both variables and , according to , @xmath203 is symmetric with respect to @xmath166 as a function of @xmath205 for every @xmath206 .this implies that @xmath207 note also that according to , p. , @xmath208 for every @xmath209 defined in .it is clear that and imply .this completes the proof of claim [ partial - dw - d - xi - claim ] .[ partial - vertical - claim ] @xmath210\ , ; \\ & \frac{\partial w(\xi,\zeta)}{\partial\zeta}=0\quad\text{if}\quad\zeta=0 \quad\text{and}\quad\xi\in[k/2 , \infty]\ , .\end{split}\ ] ] here we are going to use similar symmetry and second derivative arguments used in the proof of previous claim .recall that @xmath203 is @xmath198 function of both variables and , according to , @xmath203 is symmetric with respect to @xmath167 as a function of @xmath206 for every @xmath205 .this implies that @xmath211 note also that according to , @xmath212 for every @xmath209 defined in , p. .it is clear that and imply .this completes the proof of claim [ partial - vertical - claim ] . for the next stepwe need the following notation for the boundary parts of @xmath191 . with every @xmath194 we associate two curves @xmath213 and @xmath214 as shown on the figure [ special - boundary - curves ] below and defined as follows .@xmath215\}\cup \\ & \{(\xi,\zeta)\mid\xi\in[k/2,a]\,\,\text{and}\,\,\zeta=0\ } ; \end{split}\ ] ] @xmath216 note then , according to and , the value of @xmath203 is a continuous and strictly increasing function as the point @xmath217 runs along @xmath213 from @xmath218 to @xmath219 and @xmath220 on the other hand , @xmath221. therefore , there exists the unique point @xmath222 such that @xmath223 .note also that by , @xmath203 is continuous and strictly decreasing along the vertical segment connecting @xmath224 and @xmath225 .thus , @xmath226 . then , when the point @xmath185 runs along @xmath227 from @xmath225 to @xmath136 , the value of @xmath203 is continuously and strictly increasing to @xmath136 , according to and .therefore , there exists the unique point @xmath228 such that @xmath229 . clearly , @xmath230 .now we are ready to describe the level set @xmath192 for any fixed @xmath231 .[ level - set - boundary ] @xmath232\times[0,p]\,.\ ] ] using and we can observe that the value of @xmath233 is the strict maximum of @xmath203 for @xmath234\times[0,p]$ ] as well as the value of @xmath235 is the strict minimum for @xmath236\times[0,p]$ ] .this completes the proof of claim [ level - set - boundary ] .[ function - presentation - claim ] @xmath192 can be interpreted as the graph of a function @xmath197 , i.e. , @xmath237\}\,.\ ] ] fix any @xmath238 $ ]. then , for the vertical segment connecting two points @xmath239 and @xmath240 we can observe that @xmath241 since , according to , the value of @xmath203 is a strictly increasing function along every horizontal line in @xmath191 .note also that the value of @xmath242 , according to , is strictly decreasing for @xmath243 $ ] .hence , there exists the unique value of @xmath244 $ ] , such that @xmath245 , which completes the proof of claim [ function - presentation - claim ] . to complete the proof of proposition [ level - curve - description ] , from p., we need to study the functional properties of the function @xmath197 .first , let us observe that @xmath246 for every @xmath247 , since the value of @xmath203 , according to , is a strictly increasing function along the lower edge of @xmath191 .therefore , by the implicit function theorem , for every @xmath247 there exists some neighborhood @xmath248 , where @xmath197 is continuously differentiable @xmath249 times for any natural @xmath249 and then , according to and , we have @xmath250 a possible curve @xmath251 starting with the horizontal edge of @xmath191 is sketched on figure [ special - boundary - curves ] , page . notealso that @xmath197 is continuous in @xmath252 $ ] , since @xmath197 is monotone in @xmath253 and @xmath203 is continuous in @xmath191 .now the proof of the proposition [ level - curve - description ] from p. is complete .[ step_curve_3 ] the goal of this step is to obtain the following proposition .[ uniquiness - total - in - first - quadrant ] if @xmath254 , then the equation @xmath255 has the unique solution @xmath256 , i.e. , @xmath257 .it is clear that a solution @xmath185 may appear only on the level curve @xmath258 described above . for a point @xmath259 to be a solution we need @xmath260 .so , let us study the behavior of @xmath186 along @xmath261 . by the direct computation we can observe that @xmath262 which together with leads to @xmath263 ^ 2\right)\left[-\frac{\partial w(t , v(t))}{\partial v(t)}\right]>0\ ] ] for every @xmath247 , where the last inequality holds because of from claim [ partial - vertical - claim ] , p. .therefore , the value of @xmath264 is a strictly increasing function on @xmath265 $ ] and then , the function @xmath264 assumes each of its values for @xmath238 $ ] only once .hence , the equation @xmath266 must have only one solution in @xmath192 .this completes the proof of the proposition [ uniquiness - total - in - first - quadrant ] .[ step_curve_4 ] the next and the last stage in the proof of statement ( d ) is to analyze the behavior of @xmath267 in @xmath268\}.$ ] [ signum - behavior - of - i - propo ] @xmath269\}\cup\{(-\infty,0)\times(0,-p]\};\ ] ] @xmath270\}\cup\{(0,\infty)\times(0,-p]\}.\ ] ] the signum of @xmath271 behavior is sketched on figure [ signum - behavior - of - i ] below . + using the skew - symmetry of @xmath186 introduced in , p. , we can observe that @xmath272 on the other hand , we can observe that the condition @xmath273 $ ] yields @xmath274 for every value of @xmath0 , except @xmath275 and then , for every point @xmath209 , one of following relations depending on the signum of @xmath206 holds @xmath276_{\zeta=0}=0 ; \\ & \left.\frac{di(\xi,\zeta)}{d\xi}\right|_{\zeta>0}>0 \quad\text{or}\quad\left.\frac{di(\xi,\zeta)}{d\xi}\right|_{\zeta<0}<0 , \end{split}\ ] ] which to together with implies and .this completes the proof of proposition [ signum - behavior - of - i - propo ] .note also that @xmath186 , as well as @xmath203 are symmetric with respect to @xmath187 .therefore , the uniqueness obtained in proposition [ uniquiness - total - in - first - quadrant ] , p. together with the symmetry of @xmath277 and the skew - symmetry of @xmath271 with respect to the lines @xmath166 and @xmath167 implies that the equation @xmath255 has only two solutions @xmath257 and @xmath278 for every @xmath254 .it is clear that a similar symmetry argument shows that for every @xmath279 , the equation @xmath255 also must have only two solutions @xmath257 and @xmath278 .this completes the proof of statement ( d ) as well as completes the proof of one radius theorem [ basic - theorem ] , p .according to statement ( a ) , @xmath280 comparing the imaginary parts of this identity gives @xmath281 which implies that imaginary part is identically zero , and then , the proof of is complete .the repeated differentiation with respect to @xmath282 of the imaginary part leads directly to . to proof the taylor decomposition presented in ,notice that @xmath58 is entire function and then for every @xmath47 , @xmath283 the direct computation together with the first formula from , p. yields @xmath284 for @xmath285 and @xmath286 for @xmath287 .this completes the proof of the corollary [ imaginary - part - vanishing - corollary ] .note that the condition : @xmath69 is sufficient to find @xmath10 and @xmath288 such that @xmath289 and the equalities @xmath290 hold .note also that if we integrate a @xmath291-periodic function over the period from 0 to @xmath291 , we can replace @xmath292 by @xmath293 and vice versa without changing the result of the integration .thus , the function under the integral can be reduced to the poisson kernel , which yields @xmath294 where the fourth equality was obtained by .this completes the proof of corollary [ one - dimension - sphe - pro ] .we organize the proof as the chain of identities , which will be explained at each step .: : using the statement ( a ) from p. and comparing imaginary parts for both integrals with @xmath295 and @xmath296 , we have @xmath297 step 2 . : : let @xmath298 be the unit sphere centered at point @xmath14 .then , we can observe that @xmath299 where all notations are pictured on the figure [ reference - picture ] below that can be used for a reference .thus , the last integral in can be written as @xmath300 + + step 3 .: : recall that according to formula from p. and pythagorean theorem , we have @xmath301 note also that the integrand in depends only on angle @xmath302 pictured on figure [ reference - picture ] above . therefore , the integral in becomes @xmath303 step 4 .: : observe that the last integrand is symmetric with respect to @xmath304 . indeed , @xmath305 and @xmath306 where the symbol @xmath307 denotes composition .hence , the integral in can be written as @xmath308 step 5 .: : the substitution for @xmath309 presented in , allows to rewrite the integral in as follows @xmath310 step 6 . : : the last integral can be computed directly and then the integral in yields @xmath311 step 7 .: : note now that @xmath312 therefore , the direct computation in gives @xmath313 which completes the proof of , p. of the corollary . the partial case stated in, p. can be obtained by taking limit at both parts of as @xmath314 .this limit exists since the integrand in , p. or equivalently , in , p. converges uniformly as @xmath314 .this completes the proof of corollary [ basic - th - integrable - case - general - item ] , p. .the main goal of this section is to use the ratio of two segments @xmath30 and @xmath31 defined above , to analyze the behavior of @xmath33 as @xmath14 approaches @xmath15 .we show that @xmath35 $ ] and depends on the path chosen for @xmath14 to approach @xmath15 .[ omega_constant - lemma ] let @xmath315 be any point on the line @xmath34 such that @xmath316 and let @xmath317 be the sphere of radius @xmath318 centered at @xmath315 , where @xmath319 is the distance from @xmath315 to @xmath15 .then @xmath320 let @xmath322 , @xmath323 and @xmath324 be the intersections of line @xmath34 with the spheres @xmath23 and @xmath317 respectively .@xmath325 is defined by the definition [ spherical_ratio_definition ] .let @xmath326 .observe that @xmath327 since the segments @xmath328 and @xmath329 are diameters .therefore , @xmath330 and @xmath331 hence , the combination of , and yields @xmath332 therefore , for the chosen point @xmath315 formula is justified . a similar argument yields for all other positions of @xmath315 on the line @xmath34 .this completes the proof of lemma [ omega_constant - lemma ] .[ limit - for - omega ] let @xmath333 be the @xmath170 path through @xmath15 and let @xmath334. then @xmath335\,,\ ] ] where the value @xmath336 depends on the path @xmath333 chosen and can assume any value from the closed interval @xmath337 $ ] .note first , if @xmath338 , then @xmath339 for every @xmath340 and therefore , @xmath341 . if @xmath342 , then , by , @xmath343 which can be any number from @xmath344 depending on @xmath315 chosen on the line @xmath34 . in particular , if @xmath345 , then @xmath346 , which corresponds to the case when @xmath333 belongs to the tangent hyperplane to @xmath23 at point @xmath15 . in this case@xmath347 for every @xmath340 . finally , to get @xmath348, @xmath333 must be a @xmath170 curve non - tangential to @xmath23 at point @xmath15 .indeed , let @xmath349 be the angle between @xmath333 and line@xmath350 at point @xmath15 as it is pictured on figure [ omega - limit - picture ] below .schwarz , * gesammelte mathematische abhandlungen * , chelsea publishing company bronx , new york , 1972 , ( pp .359 - 361 ) .lars v. ahlfors , * complex analysis * , mcgraw - hill , inc . , 1966 , ( pp .168 - 169 ) .reinhold remmert , * classical topics in complex function theory * , springer - verlag new york , inc . , 1998 .s.artamoshin , * geometric interpretation of the 2-d poisson kernel and its applications * , http://arxiv.org/pdf/0912.0223v1.pdf , 2009 .

<S> at the end of 19-th century , 1874 , hermann schwartz found that for every point inside a planar disk , a two - dimensional poisson kernel can be written as a ratio of two segments , which he called as the geometric interpretation of that kernel @xcite . </S><S> about 90 years later , lars ahlfors , in his textbook , called this ratio as interesting @xcite . </S><S> we shall see here that every two different points , in a multidimensional space also define a similar ratio of two segments . </S><S> the main goal of this paper is to study some integral properties of the ratio as well as introduce and proof one radius theorem for spherical ratio of two points . in particular , we introduce a new integration technique involving the ratio of two segments and a non - trivial integral equivalence leading to an integral relationship between newtonian potential and poisson kernel in any multi - dimensional space . </S>

1304.2530

red clump ( rc ) stars are core helium - burning giants .it is a prominent feature in the colour - magnitude diagrams of open clusters as well as globular clusters .now , it is known that they are abundant in the solar neighbourhood .this sample of rc stars provides accurate absolute magnitude estimation due to their parallaxes which can be used in testing their suitability for a distance indicator .a mean absolute magnitude in any band with small scattering can work for the purpose of the researchers .many works has been carried out for optical and near infrared bands such as @xmath2 , @xmath9 and @xmath0 .their absolute magnitudes in the optical range lie from @xmath10 mag for those of spectral type g8 iii to @xmath11 mag for type k2 iii @xcite .the absolute magnitude of these stars in the @xmath12 band is @xmath13 mag with negligible dependence on metallicity @xcite , but with real dispersion .@xcite claimed an absolute magnitude for the rc stars rather close to the one just cited here , with some limitations however .based on 14 open clusters , they draw the conclusion that for clusters having @xmath14\leq0 $ ] dex and @xmath15 gyr , one can use @xmath16 mag .the dependence of the @xmath9-band magnitude on the rc stars was extensively studied in the past from an observational point of view . in most casesthe @xmath17 mean absolute magnitude is insensitive to age and metallicity @xcite. however , a modest variation in @xmath17 with colour and metallicity has been claimed in the literature @xcite .theoretical models from @xcite also show a dependence of metallicity and age in @xmath9 band .@xcite stated in their study based on the models of @xcite that @xmath18 is a complicated function of metallicity and age . in a recent work, @xcite used the two micron all sky survey ( 2mass ; * ? ? ? * ) infrared data for a sample of 24 open clusters to investigate how the @xmath0-band absolute magnitude of the red clump depends on age and metallicity .they showed that a constant value of @xmath19 mag is a reasonable assumption to use in distance determinations for clusters with metallicity between @xmath50.5 and + 0.4 dex and age between 0.31 and 7.94 gyr . following this work , @xcite claimed two absolute values in different bands for the rc stars , i.e. @xmath20 and @xmath21 mag , where the estimations were based on newly reduced the _ hipparcos _ catalogue @xcite . in a more recent work ,@xcite determined the mean @xmath18 absolute magnitude for rc stars in the solar neighbourhood to within 2 per cent ( @xmath22 mag ) and applied their results to the estimation of the distance of the large magellanic cloud .a mean value for the @xmath18 absolute magnitude with weak or negligible dependence on metallicity makes possible to use this population as a tracer of galactic structure and interstellar extinction , as several works have fully demonstrated in the last decade ( see for example * ? ? ?? * ; * ? ? ?* ; * ? ? ?* ; * ? ? ?* ; * ? ? ?* ; * ? ? ?* and references therein ) . in @xcitethe @xmath23 , @xmath24 , @xmath3 and @xmath25 absolute magnitudes of rc stars , identified by a set of constraints in the _ hipparcos _ catalogue , were calibrated in terms of colour with @xmath26 , @xmath27 and @xmath28 photometries . in the present paper, we will focus on a single absolute magnitude , @xmath3 , which is the most probable candidate for adopting as a distance indicator .our aim is to calibrate the @xmath3 absolute magnitude as a function of colour .thus , we expect more accurate absolute magnitudes relative to the procedure which adopts @xmath3 absolute magnitude as a constant value .the procedure is given in section 2 .the data and the relation between the colour and absolute magnitude are presented in sections 3 and 4 , respectively .the application of the procedure is devoted to section 5 , and finally a discussion is given in section 6 .the absolute magnitude of a star is a function of luminosity class , temperature or colour , age and metallicity . as it is assumed that the rc stars are at the same evolutionary stage , all the rc stars are of the same luminosity class .hence , we omit this parameter in the absolute magnitude estimation of the rc stars . many studies are based on the _ hipparcos _ catalogue @xcite which involves the solar neighbourhood stars . using the _ hipparcos _catalogue is a constraint both in metallicity and age for the sample stars .this is the reason that the researchers claim weak dependence of the @xmath3 absolute magnitude on age and metallicity .the advantage of the up - dated _ hipparcos _ catalogue @xcite is that the errors are smaller than the former edition .the data of open clusters are also age and metallicity constraint .for example the age and metallicity intervals for 24 open clusters in @xcite are @xmath29 gyr , and @xmath30\leq+0.4 $ ] dex , respectively .however , one can include the metal - poor globular clusters such as ngc 1261 with metallicity @xmath31=-1.35 $ ] dex into the sample used for @xmath3 absolute magnitude estimation of the rc stars . in this case , one expect metallicity- and age - dependent @xmath3 absolute magnitudes . despite age and metallicity limitations for the data used for the @xmath3 absolute magnitude estimation ,the range of the corresponding colour is large enough for considering in the estimation of the absolute magnitude in question .for example , the range of the @xmath4 colour of the data used in @xcite and @xcite is @xmath32 mag .then , one should consider the @xmath4 colour in the @xmath3 absolute magnitude estimation .that is , we expect more accurate @xmath3 absolute magnitudes for individual rc stars rather than a constant value for the whole sample .we used the data of @xcite for calibration of the @xmath3 absolute magnitude in terms of @xmath4 colour .there are 224 rc stars in the catalogue of @xcite .we provided the _hipparcos _ number , parallax , metallicity , @xmath0 magnitude , and @xmath3 absolute magnitude from the electronic version of the paper , and we added the @xmath2 magnitude to this catalogue to obtain the @xmath4 colours instead of @xmath1 ones .the @xmath2 magnitudes are taken from the _ hipparcos _ catalogue .the relative parallax errors lie in the interval @xmath33 and their median is 0.03 .the parallaxes were corrected by @xcite .the data are given in table 1 .the columns give : ( 1 ) current number , ( 2 ) _ hipparcos _ number , ( 3 ) the corrected parallax , ( 4 ) @xmath34 $ ] metallicity , ( 5 ) @xmath2 apparent magnitude , ( 6 ) @xmath0 apparent magnitude , ( 7 ) @xmath4 colour index and ( 8) @xmath3 absolute magnitude . as in @xcite , we assumed no foreground reddening . actually , the mean colour excess of 20 rc stars with _ hipparcos _ number between 671 and 7643 in the catalogue of @xcite , estimated by the following procedure is only @xmath35 mag .the @xmath36 colour excess of 20 rc stars have been evaluated in two steps .first , we used the maps of @xcite and evaluated a @xmath37 colour excess for each star .then , we reduced them using the following procedure of @xcite : @xmath38.\ ] ] here , @xmath39 and @xmath40 are the galactic latitude and distance to the star , respectively .@xmath41 is the scale height for the interstellar dust which is adopted as 125 pc @xcite . @xmath42 and @xmath43 are the total absorptions for the model and for the distance to the star , respectively .@xmath42 can be evaluated by means of the following equation : @xmath44 @xmath37 is the colour excess for the model taken from the @xcite .then , @xmath45 , i.e. the colour excess for the corresponding star at the distance @xmath40 , can be evaluated via the equation , @xmath46 the colour excess @xmath47 and the classical colour excess @xmath36 have the same meaning .the same case is valid for the total absorption @xmath48 and the classical absorption @xmath49 .the metallicities are given only for 100 rc stars .the diagram for @xmath3 absolute magnitude versus @xmath4 colour index is given in fig .most of the stars are concentrated in the region with @xmath50 and @xmath51 mag . however , there are about three dozen of stars beyond these limits .the extreme colours and absolute magnitudes belong to the rc stars with _ hipparcos _ numbers 3781 , 37901 , 38211 , 58697 , 63608 , 70306 , 72471 .the range of the metallicity of the sample is @xmath52<0.4 $ ] dex .however , 80 per cent of them lie in a smaller metallicity interval , i.e. @xmath53\leq + 0.15 $ ] dex ( fig .2 ) , indicating a thin disc sample .hence , we preferred to use the whole sample in the calibration of @xmath3 absolute magnitude to @xmath4 colour instead of separating it into different metallicity classes . the distribution of 80 per cent of the points in fig .1 is almost circular , while the complete figure gives the indication of a linear distribution .also , the large scattering and the inhomogeneous number density hinder the selection of fitting type of the @xmath3 absolute magnitude in terms of @xmath4 colour .however , we considered all the points and fitted @xmath3 to a linear equation in terms of @xmath4 as in the following ( fig .3 ) : @xmath54 we evaluated the @xmath3 absolute magnitude residuals , i.e. the difference between the absolute magnitude estimated by using eq .( 1 ) and the corresponding absolute magnitude in table 1 , and compared them with another set of absolute magnitude residuals of the same stars evaluated by adopting the value @xmath51.613 mag as the @xmath3 absolute magnitude for all rc stars .this value was claimed by @xcite and @xcite as the mean @xmath3 absolute magnitude for rc stars .( 4 ) is derived without considering the small foreground reddening in @xcite stated in section 3 , i.e. @xmath35 mag .if we transform this value to the colour excess @xmath55 by the equation @xmath56 @xcite , and use it in eq .( 4 ) we get an absolute magnitude fainter than 0.02 mag .the mean absolute magnitude residuals and the corresponding standard deviations for two sets are given in the third and fourth rows of table 2 .the mean of the absolute magnitude residuals evaluated by adopting the constant absolute magnitude value @xmath6 mag , @[email protected] , is 53 times larger than the one evaluated by the linear equation , @[email protected] , in our work .also , the standard deviation corresponding to the linear equation is smaller than the other one .however , the factor of the residuals in fig .4 is slightly higher than unity .we applied the procedure to the data in @xcite .the catalogue of @xcite involves 284 rc stars .we replaced the recent parallaxes and @xmath0 band magnitudes appeared in the _ hipparcos _ catalogue with the old ones .the relative parallax errors lie in the interval @xmath58 and their median is 0.02 .we used the following equation of @xcite to correct the observed _ hipparcos _ parallaxes @xcite : @xmath59,\ ] ] where @xmath60 and @xmath61 are the observed and corrected parallaxes , respectively , and @xmath62 denotes the error of the observed parallax . @xmath63apparent magnitudes are not given for two stars , _ hipparcos _ number : 33449 , 46952 .the sample is given in table 3 .the columns give ( 1 ) current number , ( 2 ) _ hipparcos _ number , ( 3 ) @xmath64 corrected parallax , ( 4 ) @xmath34 $ ] metallicity , ( 5 ) @xmath2 apparent magnitude , ( 6 ) @xmath0 apparent magnitude , ( 7 ) @xmath4 colour index , ( 8) @xmath3 absolute magnitude , and ( 9 ) @xmath65 parameter which indicates the quality of the @xmath27 magnitudes of a star .the quality of the rc stars are adopted from the _ hipparcos _ catalogue , and the @xmath3 absolute magnitudes are evaluated by the following equation : @xmath66 where @xmath67 is the corrected parallax of the star considered . as in @xcite , we assumed no foreground reddening ( see also section 3 ) .the total number of stars for which @xmath3 absolute magnitudes could be estimated is 282 .we plotted the @xmath3 absolute magnitudes versus @xmath4 colours for these stars in fig .the distribution of the diagram is rather different than the one given for the data in @xcite , i.e. there is a high condensation along a horizontal line and a large scattering beyond this formation .the range of the absolute magnitudes in fig .5 is much larger than the one in fig . 1 , @xmath68 mag .we evaluated the absolute magnitudes of 282 rc stars using eq .( 4 ) and compared them with the original ones in table 3 .the mean of the residuals and the corresponding standard deviation are @xmath69 and @xmath8 mag . as in section 4, we evaluated another set of statistics by adopting the value @xmath51.613 mag as the @xmath3 absolute magnitude for all rc stars , i.e. @xmath70 and @xmath71 mag ( table 2 , row 4 ) .although the standard deviation evaluated by using the linear eq .( 4 ) is smaller than the one evaluated for the constant absolute magnitude , the corresponding mean of the residuals is about 1.6 times larger than the one estimated via constant absolute magnitude .the distribution of the residuals estimated by means of two procedures are given in fig .we combined the ( @xmath0 , @xmath1 ) data in @xcite with the @xmath2 apparent magnitudes and trigonometric parallaxes taken from the _ hipparcos _ catalogue and used them to fit the @xmath3 absolute magnitude to a linear relation in terms of @xmath4 colour .we preferred the data in @xcite for absolute magnitude calibration in terms of colour due to the high precession of the observed magnitudes .@xcite used 0.75 m telescope at south african astronomical observatory and observed the brightest and nearest rc stars in the solar neighbourhood which provide accurate data .we evaluated the @xmath3 absolute magnitude residuals , i.e. the difference between the absolute magnitude estimated by using eq .( 4 ) and the corresponding absolute magnitude in table 1 , and compared them with another set of absolute magnitude residuals of the same stars evaluated by adopting the value @xmath51.613 mag as the @xmath3 absolute magnitude for all rc stars .this value was claimed by @xcite and @xcite as the mean @xmath3 absolute magnitude for rc stars .the mean of the absolute magnitude residuals evaluated by adopting the constant absolute magnitude value @xmath6 mag , @[email protected] mag , is 53 times larger than the one evaluated by the linear equation @[email protected] mag in our work .also , the standard deviation corresponding to the linear equation is smaller than the other one .this comparison shows that the @xmath3 absolute magnitudes estimated by a linear equation in terms of colour are more accurate than the constant absolute magnitudes .the result obtained from the application of the procedure to the data in @xcite is a bit different , however .although the standard deviation corresponding to the linear equation is smaller than the one evaluated for constant absolute magnitude , the mean of the absolute magnitude residuals estimated via linear equation is 1.6 times larger than the other one .differences between the statistics estimated for two sets of data originate from their trends of @xmath72 colour magnitude diagrams .the distribution of colour - magnitude diagram for the data in @xcite is almost diagonal , whereas the absolute magnitudes of rc stars in @xcite are condensed along a horizontal line , @xmath73 mag .as one can see in the table 3 , the 2mass data are not of best quality which probably affect the accuracy of the estimated absolute magnitudes .conclusion : it has been suggested that rc stars are standard candles and a mean value of @xmath6 mag based on the _ hipparcos _ data has been claimed for them in the literature . in this study , we showed that the absolute magnitude for the @xmath0 band is colour dependent .however , we need data of best quality and a large sample in order to obtain a standard linear calibration of @xmath3 absolute magnitude in terms of colour .this work has been supported in part by the scientific and technological research council ( tbitak ) 112t120 .we thank to dr .martin lopez - corredoira for his comments and suggestions .this research has made use of nasa s astrophysics data system and the simbad database , operated at cds , strasbourg , france .

<S> we combined the ( @xmath0 , @xmath1 ) data in @xcite with the @xmath2 apparent magnitudes and trigonometric parallaxes taken from the _ hipparcos _ catalogue and used them to fit the @xmath3 absolute magnitude to a linear polynomial in terms of @xmath4 colour . </S><S> the mean and standard deviation of the absolute magnitude residuals,@xmath50.001 and 0.195 mag , respectively , estimated for 224 red clump stars in @xcite are ( absolutely ) smaller than the corresponding ones estimated by the procedure which adopts a mean @xmath6 mag absolute magnitude for all red clump stars , @xmath50.053 and 0.218 mag , respectively . </S><S> the statistics estimated by applying the linear equation to the data of 282 red clump stars in @xcite are larger , @xmath7 and @xmath8 mag , which can be explained by a different absolute magnitude trend , i.e. condensation along a horizontal distribution . </S>

0909.4825

* the magnus kernel . *let @xmath4 be a compact , connected , oriented surface of genus @xmath2 with one boundary component .let @xmath5 denote the _ mapping class group _ of @xmath6 , which is the group of homotopy classes of orientation - preserving homeomorphisms of @xmath6 which fix @xmath7 pointwise .let @xmath8 denote the _ torelli group _, which is the subgroup of @xmath5 consisting of elements that act trivially on @xmath9 .@xmath5 acts on the fundamental group @xmath10 , inducing an action on the solvable quotient @xmath11 , where @xmath12 , @xmath13 $ ] and @xmath14 $ ] are the first three terms of the derived series of @xmath15 . in this paperwe consider the group @xmath16 in 1939 , magnus ( @xcite ; see also ( * ? ? ?* chapter 3 ) ) used the fox calculus to construct a representation @xmath17 now called the _ magnusrepresentation_. it follows from ( * ? ? ?* theorem 4.9 ) that the kernel of @xmath18 coincides with @xmath0 .this group is called the _ magnus kernel_. it was an open question for some time whether or not @xmath0 is nontrivial .this was settled in the affirmative by suzuki in @xcite .the first main result of this paper is that @xmath0 is in fact quite large .[ theorem : main ] for @xmath2 the group @xmath19 has infinite rank .as the abelianization of a finitely - generated group has finite rank , we deduce the following .[ corollary : main ] for @xmath2 the group @xmath0 has no finite generating set . the idea of our proof of theorem [ theorem : main ] is to define a kind of `` johnson - type '' homomorphism ( see @xcite ) : @xmath20 where @xmath21 $ ] and @xmath22 denotes the abelianization of @xmath23 .we then construct infinitely many linearly independent elements contained in the image . * the burau kernel .* let @xmath24 denote the braid group on @xmath25 strands .@xmath24 can be realized ( see section [ section : braid ] below ) as a subgroup of the automorphism group @xmath26 of the free group of rank @xmath25 .the _ burau representation _ is a homomorphism @xmath27).\ ] ] we define the _ burau kernel _ , denoted @xmath1 , to be the kernel of @xmath28 .let @xmath29 be the kernel of the homomorphism @xmath30 taking each fixed generator of @xmath31 to @xmath32 .it follows easily from @xcite that @xmath33)).\ ] ] while @xmath34 is faithful , it was a longstanding problem as to whether or not @xmath28 is faithful ( i.e. whether @xmath1 is nontrivial ) for @xmath35 .this was solved by moody @xcite , long paton @xcite , and bigelow @xcite in various cases , with the result that @xmath1 is nontrivial for @xmath36 ; the case of @xmath37 is still open .our next main result is that @xmath1 is in fact quite large for @xmath3 .[ theorem : main2 ] for @xmath3 the group @xmath38 has infinite rank ; in particular , @xmath1 has no finite generating set . to prove theorem [ theorem : main2 ] we construct , similarly to the proof of theorem [ theorem : main ] above , a homomorphism @xmath39 the elements which have been constructed in the kernel of the burau representation are geometrically elegant , but algebraically very complicated ;for example , the element of @xmath40 found by long paton can be described by a single diagram , but as a free group automorphism sends generators of @xmath41 to words of length up to @xmath42 .thus we need the assistance of a computer in order to calculate @xmath43 explicitly ( see section [ section : braid ] below for a full discussion ) . for the computations in this paper we use a simpler element @xmath44 for @xmath3 found by bigelow , which takes generators to words of length no more than 9841 .once we compute the form of @xmath45 , we then use an equivariance property of @xmath43 to prove that the image of @xmath43 has infinite rank , from which theorem [ theorem : main2 ] follows .we remark that , as problem 6.24 of @xcite , morita posed the problem of determining the kernel of the magnus and burau ( among other ) representations .theorem [ theorem : main ] and theorem [ theorem : main2 ] can be viewed as a partial answer to this problem .* acknowledgements .* we are grateful to william goldman , whose mathematica notebook freegroupautos.nb was very helpful in our computations of the expression in appendix a. we would also like to thank dan margalit and tam nguyen phan for careful comments on an earlier version of this paper .the following construction works whenever one considers a group of automorphisms of the universal @xmath46-step nilpotent quotient of a group @xmath23 acting trivially on its abelianization .johnson @xcite considered the case @xmath47 . with @xmath15equal to @xmath10 or @xmath31 as in the introduction , we take @xmath48 $ ] or @xmath49 respectively . in either case , let @xmath50 be the lower central series of @xmath23 , defined inductively by @xmath51 and @xmath52 $ ] .consider the exact sequence @xmath53 centralizing ( [ eq : basic1 ] ) gives @xmath54 since @xmath23 is free , taking ( [ eq : basic1 ] ) as a presentation for @xmath22 , hopf s formula gives that @xmath55 @xmath56 acts on @xmath15 , and thus on @xmath23 , and the isomorphism @xmath57 respects the action of @xmath56 on both sides .in particular , conjugation by @xmath15 descends to an action on @xmath22 by @xmath58 $ ] or by @xmath59 respectively . in the case@xmath21 $ ] , the fact that @xmath0 acts trivially on @xmath11 implies that @xmath0 acts trivially on @xmath60 and on @xmath61 .similarly , in the case @xmath62 , we have that @xmath1 acts trivially on @xmath22 and on @xmath61 .let @xmath63 ( resp .@xmath64 ) be given .for @xmath65 , pick any lift @xmath66 . since @xmath67 acts trivially on both the quotient and kernel of ( [ eq : key ] ) , we see that @xmath68 lies in the kernel @xmath69 , which we identify with @xmath61 via the isomorphism above . one checks , exactly as in @xcite , that @xmath70 defined by @xmath71 is a well - defined homomorphism ; in fact , the resulting map @xmath72 is @xmath73linear ( resp .@xmath74$]linear ) with respect to the conjugation action on @xmath22 .this is equivalent to the claim that @xmath75 which can be checked as follows .the difference between the left and right side is @xmath76 which is conjugate to @xmath77 $ ] .the condition on @xmath67 implies that @xmath78 , so @xmath79 and @xmath77\in g_3 $ ] as desired .one also checks , exactly as in @xcite , that in the case @xmath21 $ ] , defining the map @xmath80 by @xmath81 gives a well - defined homomorphism ; @xmath82 and , in the case @xmath62 , defining @xmath83 gives a well - defined homomorphism : @xmath84 the homomorphisms @xmath80 and @xmath43 are equivariant with respect to the natural @xmath56-actions on the source and target .let @xmath85 denote the @xmath46-sphere with @xmath86 open disks removed .a _ lantern _ in @xmath6 is an embedding @xmath87 .consider the two simple closed curves @xmath88 and @xmath89 and the three arcs @xmath90 and @xmath91 on @xmath85 given in figure . [ figure : curves ] and @xmath89 , and the arcs @xmath92.,title="fig:",width=396 ] + one directly computes the action of @xmath93 on @xmath94 , @xmath95 and @xmath91 , as follows ( see figure ) .let @xmath96 , @xmath97 , and @xmath98 be the loops which begin with @xmath94 , @xmath95 and @xmath91 , respectively , go clockwise around the appropriate boundary component of @xmath85 , then come back along the same arc @xmath99 .let @xmath100 be the inverses of @xmath101 in @xmath102 .then : @xmath103a_1\\\\ f(a_2)= zxzxa_2= [ z , x]a_2\\ \\ f(a_3)= zxzxyxzxzxyxa_3 = [ zxz , xyx]a_3 \end{array}\ ] ] let @xmath104 be an embedding of a lantern in @xmath6 with the property that each of the four boundary curves of @xmath104 are separating in @xmath6 . with elements of @xmath105 , we choose a basepoint on @xmath7 , and arcs from this basepoint to @xmath104 meeting @xmath104 in one point .since @xmath67 is the identity off of @xmath104 , any ambiguity in the choice of these paths to @xmath104 does not affect the computation . ] in this case we can observe that @xmath106 , as follows .note that the elements corresponding to @xmath101 all lie in @xmath107 .furthermore , @xmath105 has a basis where each element @xmath108 is either disjoint from @xmath104 , or else of the form @xmath109 , where @xmath110 is an arc intersecting @xmath104 in some @xmath99 and @xmath111 is a loop disjoint from @xmath104 . in the former casethe element @xmath112 fixes @xmath108 . in the latter case , assume for example that @xmath110 intersects @xmath104 in @xmath95 ; then we have @xmath113a\gamma a^{-1}[x , z]=[z , x]c[x , z]\ ] ] since @xmath114 , we have @xmath115\in \gamma^3 $ ] ; thus @xmath116 .the same is true for @xmath94 and @xmath91 , so we conclude that @xmath116 for all elements of a basis for @xmath15 , implying @xmath106 .suzuki gave a more illuminating proof that elements of this form lie in @xmath0 in @xcite .[ figure : arcs ] and @xmath117.,width=612 ] we are now ready to compute @xmath80 . for @xmath118 , we denote by @xmath119 the image of @xmath120\in g$ ] in @xmath22 under the abelianization map . [ prop : lantern ] let @xmath104 be a lantern embedded in @xmath6 so that each of the four boundary curves of @xmath104 are separating in @xmath6 .let @xmath121 and @xmath122 be loops intersecting @xmath104 in @xmath94 and @xmath95. then @xmath123\ ] ] note that the right hand side of ( [ eq : lantern5 ] ) is an element of @xmath61 , considered as a @xmath73-module , and @xmath124 are taken to be elements of @xmath125 .as in the computation above , we have @xmath126)=[f(a),f(b)]=[wa , vb]\ ] ] where @xmath127,a ] \\ \mbox{and}\ \ v=[[z , x],b].\ ] ] from the assumption on the embedding of @xmath104 we have @xmath128 , and thus @xmath129 .we will use the following commutator identities , which hold in any group ; we write @xmath130 for @xmath131 .@xmath132={}^w[a , b]\ [ w , b ] \qquad\qquad[a , vb]=[a , v]\ { } ^v[a , b]\ ] ] [ figure : surface ] and @xmath133 for @xmath134.,title="fig:",width=396 ] + we then find that @xmath135={}^w[a , v]\ { } ^{wv}[a , b]\ [ w , v]\ { } ^v[w , b]\ ] ] note that the second term lies in @xmath23 , the first and fourth in @xmath136 , and the third in @xmath137 .we want to compute @xmath138)[a , b]^{-1}$ ] as an element of @xmath69 .note that @xmath139\equiv 0\bmod g_3}$ ] , and that conjugating an element of @xmath23 by an element of @xmath136 is a trivial operation modulo @xmath137 .finally , since @xmath140,[w , b]]\in g_3 $ ] , we can move @xmath120 $ ] to the right to cancel @xmath120^{-1}$ ] .we thus obtain @xmath141)[a , b]^{-1}&={}^w[a , v]\ { } ^{wv}[a , b]\ [ w , v]\ { } ^v[w , b]\ [ a , b]^{-1}\\ & \equiv [ a , v][a , b][w , b][a , b]^{-1}\bmod g_3\\ & \equiv [ a , v][w , b]\bmod g_3. \end{aligned}\ ] ] recall that action of @xmath15 on @xmath15 by conjugation decends to a @xmath73 action on @xmath22 .recall from above the isomorphism @xmath142 . since the homology class of @xmath96 is trivial in @xmath125, we have @xmath143)=y\wedge z \\ \mbox{and}\ \\nu([z , x])=z\wedge x.\ ] ] it follows that @xmath144,a])=(1-a)y\wedge z\ ] ] and @xmath145,b])=(1-b)z\wedge x.\ ] ] we therefore have that @xmath146[w , b])=(a-1)v-(b-1)w=(a-1)(1-b)z\wedge x-(b-1)(1-a)y\wedge z.\ ] ] we conclude that @xmath147\ ] ] as desired . [ figure : fourcurves ] ; the subsurfaces cut off by these curves are shaded.,title="fig:",width=340 ] + the image of @xmath80 has infinite rank for @xmath148 .[ theorem : imagea ] let @xmath111 and @xmath133 be the curves depicted in figure .the figure depicts the case @xmath134 ; in general @xmath133 has @xmath149 twists around the upper right handle .( specifically , the curve @xmath133 is equal to @xmath150 , where @xmath151 is as in figure . )the regular neighborhood of @xmath152 is a lantern @xmath153 , and we fix an identification of @xmath153 with our reference lantern @xmath104 by specifying that @xmath111 and @xmath133 should correspond to @xmath154 and @xmath155 respectively .let @xmath156 be the element corresponding under this identification to the mapping class @xmath157 on @xmath104 ; it is easy to check using the lantern relation that @xmath158 is in fact @xmath159 $ ] .we will show that the images @xmath160 are linearly independent ( over @xmath161 ) .the boundary curves of @xmath153 are depicted in figure .[ figure : basis ] .,title="fig:",width=226 ] + with the basis @xmath162 for @xmath163 as illustrated in figure , we see that as curves @xmath96 , @xmath97 and @xmath98 can be represented by @xmath164 $ ] , @xmath165 $ ] , and @xmath166 $ ] respectively . as based loops , we actually have the conjugate @xmath167 $ ] , where @xmath168[b_2,a_2]a_2 $ ] .note that with this representative for @xmath98 , we have @xmath169[a_2,b_2][a_3,b_3]$ ] , the fourth boundary curve in figure .note that @xmath170 and @xmath171 intersect each @xmath153 in arcs corresponding to @xmath94 and @xmath95 .thus by proposition [ prop : lantern ] , we have that @xmath172\ ] ] denote this element of @xmath61 by @xmath173 .we now check that @xmath174 is linearly independent as follows .there is a standard embedding @xmath175 given by sending the class @xmath176 $ ] to @xmath177 , where @xmath178 is our basis for @xmath31 and where @xmath179 are the fox derivatives ( see e.g. @xcite for a detailed explanation of this embedding ) .the only property of this embedding that we will need is that the components that make up @xmath173 are mapped as follows by the embedding . herethe @xmath99 and @xmath180 make up a basis for @xmath181 .@xmath182 by expanding out @xmath173 , we see that @xmath183 is the only such element which contains the term @xmath184 with nonzero coefficient ; it follows that the @xmath173 are linearly independent , as desired .as the image of @xmath80 is abelian , theorem [ theorem : imagea ] immediately implies theorem [ theorem : main ] for @xmath148 . note that the proof of theorem [ theorem : imagea ] used in an essential way that @xmath148 .so in order to complete the proof of theorem [ theorem : main ] , we need another argument when @xmath185 .[ theorem : imageb ] @xmath186 has infinite rank .suzuki showed that the element @xmath187 $ ] is in @xmath188 for @xmath111 and @xmath189 as in figure ; in particular @xmath188 is nontrivial .let @xmath190 be a closed surface of genus 2 ; we denote by @xmath191 the torelli group of @xmath190 with respect to a marked point @xmath192 , and by @xmath193 the torelli group of the closed surface @xmath190 . by johnson @xcite, we have the exact sequence @xmath194 where the kernel is generated by a twist @xmath195 around the boundary @xmath196 .it is easy to check that the action of @xmath195 on @xmath197 is conjugation by @xmath198 ; since @xmath199 , we see that @xmath200 .[ figure : genustwo ] $ ] lies in @xmath188.,title="fig:",width=453 ] + it follows that @xmath201 restricts to an isomorphism between @xmath188 and a subgroup @xmath202 .again by johnson @xcite , we have the exact sequence @xmath203 where @xmath204 ; note that @xmath191 acts on @xmath205 , and the restriction to @xmath206 is just the action by conjugation .mess @xcite proved that @xmath193 is free of infinite rank .it is easy to see from figure that @xmath207 .we use the following well - known lemma .[ lemma ] any nontrivial infinite index normal subgroup of a surface group or free group is an infinite rank free group . if @xmath208 is nontrivial , then by lemma [ lemma ], @xmath188 surjects to the infinite rank free group @xmath209 , and we are done .suppose that @xmath210 .any @xmath211 acts trivially on @xmath11 ; thus @xmath212 acts trivially on @xmath213 .since the action of @xmath206 is by conjugation , this implies that @xmath212 lies in @xmath214 .thus @xmath215 has infinite index in @xmath206 , and so by lemma [ lemma ] , @xmath216 is an infinite rank free group . theorem [ theorem : main ] , and hence corollary [ corollary : main ] , follows immediately from theorems [ theorem : imagea ] and [ theorem : imageb ] .one can check by explicit computation that for suzuki s element @xmath217 above , @xmath218 .it would be interesting to know whether @xmath80 in fact vanishes on @xmath188 .the kernel @xmath29 of the map from @xmath219 to @xmath220 which sends each @xmath221 is normally generated by the elements @xmath222 .if we set @xmath223 for @xmath224 and @xmath225 , then @xmath226 gives a basis for @xmath29 as a free group .as above , the conjugation of @xmath29 by @xmath31 descends to a @xmath74 $ ] action on @xmath227 . with respect to this action we have @xmath228 , andthus @xmath227 is a free @xmath74$]module with basis @xmath229 .the braid group @xmath24 has generators @xmath230 ; the action of @xmath231 on @xmath31 sends @xmath232 , @xmath233 , and fixes the other generators .the action of @xmath24 on @xmath227 commutes with the @xmath74 $ ] action . [ figure : butterfly ] .,title="fig:",width=264 ] + the image of @xmath43 has infinite rank for @xmath3 .the element of the kernel found by bigelow in @xcite is the commutator of the half - twists along the arcs displayed in figure . in terms of the artin generators , this is @xmath234,\qquad\text{where } \psi_1= \sigma_4\sigma_5^{-1}\sigma_2^{-1}\sigma_1\quad\text{and } \psi_2=\sigma_4^{-1}\sigma_5 ^ 2\sigma_2\sigma_1^{-1}.\ ] ] in appendixa , we give the computation of @xmath235)=\phi(\phi_{b})(y_2)$ ] ; it has 262 terms .the only fact about @xmath88 that we will need is that its highest term of the form @xmath236 is @xmath237 , and its highest term of the form @xmath238 is @xmath239 ( these terms are set in boxes in the appendix ) .it is easy to check that @xmath240 by induction , for @xmath241 we have @xmath242 the action of @xmath243 on @xmath227 in terms of our basis is thus given by : @xmath244 now for @xmath245 set @xmath246 by the equivariance of @xmath43 , and since @xmath247 fixes @xmath248 , we have @xmath249 . from the action of @xmath243 on @xmath227 , we can see that the highest term in @xmath183 of the form @xmath236 will be @xmath250 .thus @xmath183 is not contained in the span of @xmath251 ; it follows that the @xmath173 are linearly independent over @xmath161 , and thus the image of @xmath43 has infinite rank .theorem [ theorem : main2 ] follows immediately .the following computation was made , with the method explained in section [ section : braid ] , with the help of _mathematica_. a _ mathematica _ notebook implementing these computations can be found at :

<S> consider the kernel @xmath0 of the magnus representation of the torelli group and the kernel @xmath1 of the burau representation of the braid group . </S><S> we prove that for @xmath2 and for @xmath3 the groups @xmath0 and @xmath1 have infinite rank first homology . as a consequence </S><S> we conclude that neither group has any finite generating set . </S><S> the method of proof in each case consists of producing a kind of `` johnson - type '' homomorphism to an infinite rank abelian group , and proving the image has infinite rank . for the case of @xmath1 </S><S> , we do this with the assistance of a computer calculation . </S>

hep-ph0310073

@xmath8 violation has been a puzzling phenomenon in the studies of elementary particle physics since the first observation of its effects in hadronic kaon decays almost four decades ago @xcite . in the standard model ( sm ) , @xmath8-violation is due entirely to the cabibbo - kobayashi - maskawa ( ckm ) mechanism @xcite , describing the mismatch between the unitary transformations relating the up and down type quark mass eigenstates to the corresponding weak eigenstates .the ckm matrix involves a single weak phase along with three mixing angles .the validity of the ckm picture is further strengthened by the fact that recent @xmath9 measurements from time - dependent @xmath8 asymmetries of decay modes involving the @xmath10 subprocess @xcite agree well with the range of the weak phase @xmath11 from many other constraints @xcite .however , it is still unknown whether there are any other sources that may give rise to @xmath8-violating effects .good places to search for deviations from the sm predictions are decay processes that are expected to be rare in the sm , which may reveal new physics through interference effects .in particular , discrepancies among the time - dependent @xmath8 asymmetries of different @xmath1 decay modes may show evidence for new physics @xcite .recently , an anomaly was reported in the time - dependent @xmath8 asymmetry measurement of the @xmath12 decay mode . within the framework of the sm, this process should also provide us with information on the weak phase @xmath11 , up to about @xmath13 theoretical uncertainty @xcite .however , the averaged value of @xmath6 reported by the babar and belle groups is @xcite ( individual measurements to be quoted in table [ tab : data ] ) s_k_s = -0.147 0.697 ( s=2.11 ) .this result is only about @xmath14 away from the corresponding quantity measured by the @xmath15 mode , @xmath16 @xcite .however , the scale factor @xmath17 suggests a discrepancy between the two experimental groups . before this discrepancyis settled , the difference between @xmath6 and @xmath18 suggests the possibility of new physics contributions . from the theoretical point of view , the @xmath19 decay is a loop - induced process involving @xmath20 penguin operators in the sm .therefore , it is susceptible to new physics contributions even if they are suppressed by a large mass parameter which characterizes the new physics scale .in addition to model - independent approaches @xcite , many studies have been made to explain the anomaly in supersymmetric and related models @xcite .such an effect can also be explained using models in which the bottom quark is mixed with heavy mirror fermions with masses of the order of the weak scale @xcite .it is the purpose of this work to show that a new physics effect of similar size can be obtained from some models with an extra @xmath5 boson .@xmath5 bosons are known to naturally exist in well - motivated extensions of the sm @xcite .the @xmath5 mass is constrained by direct searches at fermilab , weak neutral current data , and precision studies at lep and the slc @xcite , which give a model - dependent lower bound around 500 gev .the latter also severely limits the @xmath21-@xmath5 mixing angle @xmath22 a few @xmath23. a could be relevant to the nutev experiment @xcite and , if the couplings are not family universal @xcite , to the anomalous value of the forward - backward asymmetry @xmath24 @xcite .( earlier hints of a discrepancy in atomic parity violation have largely disappeared due to improved calculations of radiative corrections @xcite . )we therefore study the @xmath5 boson in the mass range of a few hundred gev to 1 tev , assuming no mixing between @xmath21 and @xmath5 .interesting phenomena arise when the @xmath5 couplings to physical fermion eigenstates are non - diagonal .this is possible if there exist additional exotic fermions that have different @xmath0 charges from the ordinary fermions , as found in @xmath25 models @xcite .however , in these models left - handed fermion mixings may induce undesirable flavor changing neutral currents ( fcnc ) mediated by the @xmath21 boson even in the absence of @xmath21-@xmath5 mixing or nonuniversal family couplings .one can avoid this consequence by confining the mixing to be between right - handed fermions and the exotic quarks @xcite . alternatively ,other models give family nonuniversal @xmath5 couplings as a result of different ways of constructing families in some string models @xcite .fcnc and possibly new @xmath8-violating phenomena will also occur in these models after fermion mixings are taken into account .these can occur for both left and right - handed fermions .although experiments on fcnc processes ( such as the mass difference between @xmath26 and @xmath27 and the @xmath28 decay ) have significantly constrained the @xmath5 couplings of the first and second generation quarks to be almost the same and diagonal , the couplings to the third generation are not well constrained .similar statements apply to the charged leptons .it has been shown in ref .@xcite that indeed the third generation fermions can have different @xmath5 couplings from the other two generations .we use all of the above - mentioned features to study the imprints of the @xmath5 boson on certain processes that involve @xmath29 transitions . in section[ sec : formalism ] , we present the model and framework to be studied . in section [ sec :phiks ] , we show the constraints on the model parameters from the current data of @xmath6 , @xmath30 and the branching ratio @xmath31 . in section [ sec : etaprks ] , we study a related process @xmath32 , also including its @xmath8 asymmetries and branching ratio .another way to constrain the @xmath33-@xmath34-@xmath5 coupling is the leptonic decay process @xmath35 .we study this process in section [ sec : mumu ] and show how sensitively run ii at fermilab can probe this coupling in the near future .we conclude in section [ sec : conclusions ] .in this paper , we concentrate on models in which the interactions between the @xmath5 boson and fermions are flavor nonuniversal for left - handed couplings and flavor diagonal for right - handed couplings .the analysis can be straightforwardly extended to general cases in which the right - handed couplings are also nonuniversal across generations .the basic formalism of flavor changing effects in the @xmath5 model with family nonuniversal and/or nondiagonal couplings has been laid out in ref .@xcite , to which we refer readers for detail .here we just briefly review the ingredients needed in this paper .we write the @xmath5 term of the neutral - current lagrangian in the gauge basis as [ eq : zpr ] l^z = -g j _ z^ , where @xmath36 is the gauge coupling associated with the @xmath0 group at the @xmath37 scale .we neglect its renormalization group ( rg ) running between @xmath37 and @xmath38 .the @xmath5 boson is assumed to have no mixing with the sm @xmath21 boson . and@xmath39 parameters defined in section [ sec : phiks - amp - br ] , but would not alter the discussion of the implications . ]the chiral current is j _ = _ i , j_ i^i _ ^i_j , where the sum extends over the flavors of fermion fields , the chirality projection operators are @xmath40 , the superscript @xmath41 refers to the gauge interaction eigenstates , and @xmath42 ( @xmath43 ) denote the left - handed ( right - handed ) chiral couplings . @xmath42 and @xmath43 are hermitian under the requirement of a real lagrangian .the fermion yukawa coupling matrices @xmath44 in the weak basis can be diagonalized as y_^d = v__r y _ v__l^using the bi - unitary matrices @xmath45 in @xmath46 , where @xmath47 and @xmath48 are the mass eigenstate fields .the usual ckm matrix is then given by v_ckm = v_u_l v_d_l^ .the chiral @xmath5 coupling matrices in the physical basis of down - type quarks thus read b^l_d & & v_d_l _d_l v_d_l^ , + b^r_d & & v_d_r _d_r v_d_r^ , where the @xmath49 are hermitian .we do not need the corresponding couplings for up - type quarks or charged leptons in our discussions . as long as @xmath50 is not proportional to the identity matrix, @xmath49 will have nonzero off - diagonal elements that induce fcnc interactions . to see this ,consider as an example the simplified @xmath51 matrix for the down - type quarks of the form _d_l = q_d ( ccc 1 & 0 & 0 + 0 & 1 & 0 + 0 & 0 & x ) , [ eldiag ] where both @xmath52 and @xmath34 quarks have the same @xmath5 charge @xmath53 and @xmath54 is the ratio of the @xmath5 charge of @xmath33 to @xmath53 .if we assume the mixing is among the down - type quarks only , @xmath55 and @xmath56 .any redefinition of the quark fields by pure phase shifts would have no effect on the resultant @xmath57 .all the off - diagonal couplings are proportional to the @xmath5 charge difference between the @xmath33 quark and the @xmath58 quarks , as expected . using the standard parametrization @xcite ,the explicit form of the off - diagonal @xmath5 coupling between @xmath33 and @xmath34 quarks , for example , is b^l_sb = ( 1-x ) q_d _ 13 _ 23 ( _ 12 _ 23 + _ 23 _ 12 _ 13 e^-i _ 13 ) , where @xmath59 are the mixing angles between the @xmath60th and the @xmath61th generations and @xmath62 is the @xmath8-violating weak phase . in this example, @xmath63 is proportional to the product @xmath64 of elements of the ckm matrix .more generally , one can always pick a basis for the weak eigenstates in which the @xmath50 matrices are diagonal and of the form ( [ eldiag ] ) , though with different @xmath53 and @xmath54 for the @xmath65 and @xmath66 .however , the yukawa matrices @xmath67 and @xmath68 will in general not be diagonal in that basis , so that @xmath69 and @xmath70 . in that case @xmath71will in general be nondiagonal and complex , with new mixing angles and @xmath8 violating phases not directly related to @xmath72 .basis , in which case @xmath50 would in general be off - diagonal and complex . ] instead of restricting ourselves to models with particular parameter choices in the couplings and mixings , we will take the effective theory point of view and constrain the effective couplings relevant to the decay modes of interest in the following analysis. however , to be more definite , we assume that the right - handed coupling matrix @xmath73 is flavor diagonal . if @xmath73 is nondiagonal , new operators involving different chirality structures will be induced in @xmath1 decays .within the sm , the @xmath74 decay proceeds through the loop - induced @xmath75 transition , which involves dominantly the qcd penguin but also some electroweak ( ew ) and chromomagnetic penguin contributions . to illustrate possible modifications due to the existence of an extra @xmath0 gauge boson, we will neglect the smaller contributions from weak annihilation diagram in the following analysis although they can play some role in enhancing the branching ratios @xcite .this two - body hadronic @xmath1 meson decay can be conveniently analyzed in the framework of the effective weak hamiltonian and factorization formalism @xcite .since the penguin diagrams receive dominant contributions from the top quark running in the loop , the effective hamiltonian relevant for the charmless @xmath76 decays can be written as _eff^sm = \ { v_ubv_us^ * - v_tbv_ts^ * } + . here o_1 = ( |u_b_)_v - a ( |s_u_)_v - a , o_2 = ( |u_b_)_v - a ( |s_u_)_v - a are tree - level color - favored and color - suppressed operators , o_3(5 ) = ( |s_b_)_v - a ( |s_s_)_v - a ( v+a ) , o_4(6 ) = ( |s_b_)_v - a ( |s_s_)_v - a ( v+a ) are the qcd penguin operators , o_7(9 ) = 32 e_s ( |s_b_)_v - a ( |s_s_)_v+a ( v - a ) , o_8(10 ) = 32 e_s ( |s_b_)_v - a ( |s_s_)_v+a ( v - a ) are the ew penguin operators ( @xmath77 is the electric charge of the strange quark ) , and o_g = m_b |s_^ t^a _ ( 1 + _ 5 ) b _g_^a is the chromomagnetic operator , where @xmath78 and @xmath79 refer to color indices .we mention in passing that the @xmath5 boson will also modify the @xmath80 effective hamiltonian relevant to @xmath81-@xmath82 mixing , but in an unnoticeable way .this is because the additional contribution is proportional to the square of the @xmath5 couplings between the first and third generations , @xmath83 , which is much more suppressed than the sm contribution . although the also contributes to @xmath84 transitions at the tree level and gains a color factor relative to the sm tree process , it is nevertheless suppressed by the @xmath85 couplings and the mass in comparison with @xmath86 and the @xmath87 mass @xcite .consequently , we do not study its effect on @xmath88 and @xmath9 in charmed modes .nevertheless , it can have significant effects on the @xmath89-@xmath90 system if the couplings @xmath85 are not too small , as we assume in the current analysis . in the generalized factorization approach @xcite , the @xmath91 decay amplitude is [ eq : sm - amp ] |a(|b_d ^0 )& = & - v_tbv_ts^ * x^(bk , ) , where [ eq : fact ] x^(bk , ) | ( |s_s_)_v - a | 0 | ( |s_b_)_v - a | |b = 2 f_m_f_1^bk(m_^2 ) ( ^ * p_b ) is a factorizable hadronic matrix element . the coefficients @xmath92 are given by [ eq : effwc ] a_2i-1 = c^eff_2i-1 + c^eff_2i , a_2i = c^eff_2i + c^eff_2i-1 , where @xmath93 are effective wilson coefficients that should be used when one replaces the one - loop hadronic matrix elements in the effective hamiltonian with the corresponding tree - level ones @xcite .nonfactorizable effects are encoded in the effective number of colors @xmath94 . throughout this paper , we take the naive choice @xmath95 for illustration . for the input parameters@xmath96 , @xmath97 ; the wolfenstein parameters @xcite @xmath98 , @xmath99 , @xmath100 and @xmath101 @xcite ; @xmath102 , @xmath103 gev ; and the running quark masses @xmath104 gev , @xmath105 gev , @xmath106 gev , @xmath107 mev , @xmath108 mev , and @xmath109 mev @xcite , the next - to - leading order ( nlo ) effective wilson coefficients @xcite for the @xmath76 weak hamiltonian at the scale @xmath110 gev within the sm are given in the second and third columns of table [ tab : wc ] .[ cols="<,<,<",options="header " , ] with fcnc , the @xmath5 boson contributes at tree level , and its contribution will interfere with the standard model contributions .in particular , the flavor - changing couplings of the @xmath5 with the left - handed fermions will contribute to the @xmath111 and @xmath112 operators for left ( right)-handed couplings at the flavor - conserving vertex , i.e. , @xmath113 receive new contributions from @xmath5 . on the other hand , the right - handed flavor changing couplings yield new operators with coefficients that contain another weak phase associated with @xmath114 .we will ignore these contributions in this paper .the effective hamiltonian of the @xmath75 transition mediated by the @xmath5 is _eff^z = - ( ) ^2 b^l_sb ( b^l_ss o_9 + b^r_ss o_7 ) + , where @xmath115 , and @xmath116 and @xmath117 refer to the left- and right - handed effective @xmath5 couplings of the quarks @xmath60 and @xmath61 at the weak scale , respectively .the diagonal elements are real due to the hermiticity of the effective hamiltonian , but the off - diagonal elements may contain weak phases . only one new weak phase associated with @xmath63can be introduced into the theory under our assumption of neglecting @xmath114 .we denote this by @xmath118 and write @xmath119 .as @xmath120 has the same operators @xmath111 and @xmath121 as in the sm effective hamiltonian , the strong phases from long - distance physics should be the same .since heavy degrees of freedom in the theory have already been integrated out at the scale of @xmath37 , the rg evolution of the wilson coefficients after including the new contributions from @xmath5 is exactly the same as in the sm .we obtain the branching ratio [ eq : brphik0 ] b^sm+z(|b ^0 ) ^sm(|b ^0 ) | 1 - e^i _l |^2 , where ^ll & & | ( ) ^2 | , + ^lr & & | ( ) ^2 | , and @xmath122 .the second and third terms in eq .( [ eq : brphik0 ] ) represent the @xmath5 contributions from left- and right - handed couplings with the @xmath123 in the final state , respectively .we have assumed for definiteness that @xmath124 and @xmath125 have the same sign , so that the @xmath126 and @xmath39 terms interfere constructively .the branching ratio predicted by our model depends on the absolute ratios @xmath126 and @xmath39 and the weak phase @xmath118 .we show the branching ratios as a function of @xmath118 in fig .[ fig : brphiks ] .generically , one expects a ratio @xmath127 and @xmath38 to be a few to around 10 times @xmath128 .we assume that the product @xmath129 is numerically about the same as @xmath130 , and take @xmath131 and @xmath132 as representative values for numerical analyses in this and the following sections .here with the bounds on the product of the effective @xmath34-@xmath33-@xmath5 coupling and the ratio @xmath133 obtained from certain semileptonic @xmath1 decays in ref .@xcite , although a specific @xmath25 model with leptophobic features is assumed in their study . ]it is straightforward to scale the results to other @xmath126 values . in units of @xmath134 versusthe weak phase @xmath118 associated with the effective @xmath5 coupling @xmath63 .the current experimental range at @xmath135cl is shown by the two horizontal dotted lines .the sm prediction is the thin horizontal line .the thick solid and dashed curves include both left - handed and right - handed couplings with @xmath136 and @xmath132 , respectively .the single - dot - dashed and double - dot - dashed curves involve only the left - handed couplings with @xmath131 and @xmath132 , respectively .[ fig : brphiks ] ] to quantify the effects of right - handed couplings , we consider @xmath137 and @xmath132 and show the corresponding curves in fig .[ fig : brphiks ] .the branching ratio curves are almost symmetric about @xmath138 , with the slight asymmetry set by the small strong phases in the wilson coefficients .this echos the fact that the contributing amplitudes in eq .( [ eq : brphik0 ] ) have the largest constructive interference when @xmath139 . to be consistent with the measured branching ratio of @xmath74 , our weak phase @xmath118 in the region @xmath140 is favored , with the exact range depending upon @xmath126 and @xmath39 in the model . for some parameter choices ,it can leave us a two - fold ambiguity , which can be resolved using further information to be discussed in the following subsection .the time - dependent @xmath8 asymmetry for @xmath19 is a_k_s(t ) & = & + & = & a_k_s ( m_b_dt ) + s_k_s ( m_b_dt ) , where the direct and the indirect @xmath8 asymmetry parameters are given respectively by a_k_s = , s_k_s = . the parameter @xmath141 is defined by [ eq : lambda_phiks ] _ k_s _ k_s ( ) _ b ( ) _ k , where @xmath142 is the @xmath8 eigenvalue of the @xmath143 state , and ( ) _ b = , ( ) _ k = are factors that account for the mixing effects in neutral @xmath1 and @xmath144 meson systems , respectively .\(a ) ( b ) asymmetries , @xmath6 and @xmath145 , versus @xmath118 .the current experimental ranges at @xmath146 level are shown by the thin horizontal solid and dotted lines , respectively .the sm predicts @xmath147 ( not shown ) .the thick solid and dashed curves are @xmath6 for @xmath131 and @xmath132 , respectively .the corresponding curves for @xmath30 are displayed using single - dotted and double - dotted dashes .plot ( a ) has @xmath148 ; plot ( b ) has @xmath149 and @xmath132 .[ fig : acpphiks],title="fig : " ] asymmetries , @xmath6 and @xmath145 , versus @xmath118 .the current experimental ranges at @xmath146 level are shown by the thin horizontal solid and dotted lines , respectively .the sm predicts @xmath147 ( not shown ) .the thick solid and dashed curves are @xmath6 for @xmath131 and @xmath132 , respectively .the corresponding curves for @xmath30 are displayed using single - dotted and double - dotted dashes .plot ( a ) has @xmath148 ; plot ( b ) has @xmath149 and @xmath132 .[ fig : acpphiks],title="fig : " ] we show our estimates of @xmath6 and @xmath30 as a function of the new weak phase @xmath118 in fig .[ fig : acpphiks ] . in fig .[ fig : acpphiks](a ) , we have choices @xmath131 and @xmath132 , but set @xmath148 . the sm prediction of @xmath6 and @xmath30 are @xmath150 and @xmath151 , respectively .we see that the measured @xmath30 does not give much constraint on the weak phase @xmath118 , except for the regions between @xmath152 when @xmath131 .the @xmath6 data can be readily fitted within @xmath146 for values of @xmath126 chosen here . in fig .[ fig : acpphiks](b ) , we turn on the right - handed couplings and set @xmath153 .we notice that the variation of @xmath30 is within the experimental @xmath146 limits for the most range of @xmath118 . as illustrated by the thick solid curve in fig .[ fig : acpphiks](b ) , there are four possible ranges of @xmath118 that can fit the averaged @xmath6 if both @xmath126 and @xmath39 are large enough . for smaller @xmath126 and @xmath39 ,however , only a region of negative @xmath118 is favored . in order to satisfy both cp asymmetry constraints, @xmath118 should have negative value in most cases .only @xmath136 can have some positive @xmath118 range . combining the constraints from @xmath154 at @xmath135cl and both @xmath6 and @xmath30 at @xmath146 level, we find the following allowed regions of @xmath118 . if we take @xmath155 , @xmath156 is favored .if we take @xmath157 , @xmath158 is favored .if we take @xmath131 and ignore @xmath39 , then @xmath159 is favored . for @xmath160 and @xmath148 ,@xmath161 is favored .the @xmath32 is another decay mode whose time - dependent @xmath8 asymmetry @xmath7 is expected to give us the sm @xmath9 .current data reported by babar and belle ( see table [ tab : data ] ) are both lower than the sm prediction , although consistent within @xmath162 . since this process also contains the @xmath112 and @xmath111 operators in the amplitude , we discuss the @xmath5 effects on its observables . the perturbative calculations of the @xmath163 branching ratios are significantly smaller than the observed values .this discrepancy can be explained by adding a singlet - penguin amplitude , where @xmath164 is produced through a flavor - singlet neutral current , to interfere constructively with the qcd penguin contributions @xcite .another analysis @xcite found that it is hard to obtain a sizeable flavor - singlet amplitude from perturbative calculations , but qcd penguin amplitudes can be enhanced by an asymmetric treatment of the @xmath165 component of the @xmath164 wavefunction . since this matter is still debatable , we will follow the usual effective hamiltonian approach @xcite and put the emphasis on what kind of effects the @xmath5 boson may provide .following the notation in ref .@xcite , the decay amplitude of @xmath166 can be written as [ eq : etaprk - sm - amp ] & & |a(|b^0 |k^0 ) + & & = i , where x_1 & = & i(m_b^2 - m_^2 ) f_k f_0^b(m_k^0 ^ 2 ) , + x_2 & = & i(m_b^2 - m_k^0 ^ 2 ) f _ f_0^bk(m_^2 ) , + x_3 & = & i(m_b^2 - m_k^0 ^ 2 ) y_ f_0^bk(m_^2 ) , + r_1 & = & , + r_2 & = & , and @xmath167 and @xmath168 are mixing parameters for the choice of the @xmath164 meson wavefunction to be @xmath169 .since the @xmath163 has two pseudoscalar mesons in the final state , the decay width is ( b_d k^0 ) = | a(b_d k^0 ) |^2 , where @xmath170 is defined in a similar way to eq .( [ eq : pc ] ) . with @xmath171 mev and @xmath172 mev @xcite , @xmath173 and @xmath174 @xcite, we have @xmath175 , which is much lower than the experimental average of @xmath176 ( see table [ tab : data ] ) . as in the case of @xmath177decays , our model makes extra contributions to @xmath111 and @xmath121 at the weak scale .the branching ratio is ^sm+z(|b |k^0 ) ^sm(|b |k^0 ) | 1 - e^i _we notice that the coefficient of @xmath126 and that of @xmath39 also tend to have constructive interference between themselves according to our assumption that @xmath124 and @xmath125 have the same sign .the magnitudes of these coefficients , however , are much smaller than those in eq .( [ eq : brphik0 ] ) .this is simply because the terms that receive contributions from the @xmath5 boson ( mostly @xmath178 ) have some cancellation between the @xmath179 and @xmath180 terms in eq .( [ eq : etaprk - sm - amp ] ) .these observations qualitatively tell us why the @xmath181 decays are not affected quite as much by the @xmath5 effects . in units of @xmath134 versus @xmath118 . the current experimental range at @xmath135cl is shown by the two horizontal dotted lines .the sm prediction is the thin horizontal line .the thick solid and dashed curves include both left - handed and right - handed couplings with @xmath136 and @xmath132 , respectively .the single - dot - dashed and double - dot - dashed curves involve only the left - handed couplings with @xmath131 and @xmath132 , respectively .[ fig : bretaprks ] ] we see in fig . [ fig : bretaprks ] that the @xmath5 boson can explain the gap between the observed branching ratio and the sm prediction only around @xmath182 even with large couplings in both @xmath126 and @xmath39 . as we will see , however , this region is not favored by the cp asymmetry constraints .therefore , we must attribute this anomaly to some other unknown source .\(a ) ( b ) asymmetries , @xmath7 and @xmath183 , versus @xmath118 .the current experimental ranges at @xmath146 level are shown by the thin horizontal solid and dotted lines , respectively .the sm predicts @xmath184 ( not shown ) .the thick solid and dash - dotted curves are @xmath7 for @xmath131 and @xmath132 , respectively . the corresponding curves for @xmath185 are displayed using single - dotted and double - dotted dashes .plot ( a ) has @xmath148 ; plot ( b ) has @xmath149 and @xmath132 .[ fig : acpetaprks],title="fig : " ] asymmetries , @xmath7 and @xmath183 , versus @xmath118 .the current experimental ranges at @xmath146 level are shown by the thin horizontal solid and dotted lines , respectively .the sm predicts @xmath184 ( not shown ) .the thick solid and dash - dotted curves are @xmath7 for @xmath131 and @xmath132 , respectively .the corresponding curves for @xmath185 are displayed using single - dotted and double - dotted dashes .plot ( a ) has @xmath148 ; plot ( b ) has @xmath149 and @xmath132 .[ fig : acpetaprks],title="fig : " ] the asymmetry curves for @xmath32 are shown in fig .[ fig : acpetaprks ] .we do not get useful constraints from current data on @xmath185 .the value of @xmath185 does not vary much from its sm prediction throughout the whole range of @xmath118 .the averaged value of @xmath7 can be explained at @xmath146 level by simultaneously taking large values of both left- and right - handed couplings ( the solid curve in fig .[ fig : acpetaprks](b ) ) . in this case , however , only negative @xmath118 around @xmath186 is favored from the @xmath7 constraint .other cases do not explain the @xmath7 anomaly though all of them favor negative value of @xmath118 to approach the @xmath187 limit .leaving @xmath135cl of branching ratio constraints , we have only @xmath188 case that can satisfy both @xmath163 and @xmath189 cp asymmetry with a two - fold range of @xmath118 , @xmath190 and @xmath191 . attributing the branching ratio of @xmath163 to some unknown effects , the latter is favored by the @xmath177 branching ratio .@xmath192 decays are good candidates to observe fcnc interactions beyond the sm . in the sm ,the branching ratio of such processes are proportional to @xmath193 and the corresponding ckm factors .therefore , @xmath89 decays in general have branching ratios larger than the corresponding @xmath81 decays by a factor of @xmath194 .although @xmath195 should have a still larger branching ratio , this mode is hard to detect at hadron colliders .the best mode for experimental study is thus @xmath196 .recently , studies in some minimal supersymmetric standard model ( mssm ) and minimal supergravity ( msugra ) models predict that the branching ratio of the rare decay @xmath35 can be large enough to be observable by tevatron run ii @xcite . using 113 pb@xmath197 of data from run ii ,the cdf collaboration placed an upper bound @xmath198 , while the d0 collaboration placed a bound @xmath199 based on 100 pb@xmath197 of data @xcite .the sm prediction of @xmath200 is more than two orders of magnitude smaller than these bounds .the sm contribution to @xmath35 is loop - suppressed .it is therefore possible for the decay to be dominated by @xmath5 physics . for an order of magnitude estimate, one can temporarily ignore the rg running effect at the @xmath33-@xmath34-@xmath5 vertex .the decay width of @xmath35 is given by ( b_s ^+ ^- ) & = & ( b_s ) f_b_s^2 m_^2 m_b_s |v_tb^ * v_ts|^2 + & & \{ | y ( ) + 2 ( ) ^2 |^2 + | 2 ( ) ^2 |^2 } , where @xmath201 is the effective @xmath202-@xmath202-@xmath5 coupling at the weak scale .one can find the definition of @xmath203 in the sm part in ref .@xcite ; its value is about @xmath204 here .no mixing between @xmath21 and @xmath5 is assumed here .using the central value of the averaged @xmath89 lifetime @xmath205 ps @xcite and @xmath206 mev , we obtain a sm branching ratio of @xmath207 .the current cdf run ii upper bound thus gives a constraint that \{ | 0.0028 + ( ) ^2 |^2 + | ( ) ^2 |^2 } ^1/2 0.043 . the weak phase @xmath118 associated with @xmath63 can in principle be extracted from the time - dependent @xmath8 asymmetry measurement in a fashion similar to the cases studied in the previous sections , but the relevant experimental information is not presently available .we note that the above bound has no direct relation with the couplings relevant to @xmath75 transitions .the lepton couplings to can be much smaller than the quark couplings , as is true in some ( quasi)leptophobic models .we are currently investigating the effects on @xmath208 decays , which have been measured to good precision recently and should provide a tighter bound @xcite .in this paper we have considered models with an extra @xmath5 in the mass range of a few hundred gev to around 1 tev . with a family nonuniversal structure in the @xmath5 couplings , flavor changing neutral currentsare induced via the fermion mixing , therefore producing interesting effects .currently , constraints on the @xmath5 coupling between the second and third generations are not restrictive . withnon - diagonal left - handed and diagonal right - handed @xmath5 couplings in the down - type quarks , we studied the impact of such @xmath5 models on rare @xmath1 meson decay processes that are sensitive to new physics . in the present analysis, we have assumed that the left- and right - chiral couplings @xmath124 and @xmath125 have the same sign , rendering constructive interference in the @xmath5 contributions .we do not include the right - handed flavor changing couplings , which will give rise to new operators not existent in the sm .involving these or choosing different values for the effective number of colors @xmath209 , for which the branching ratios change sensitively , would change the results . we have found that with the inclusion of the @xmath5 contributions , @xmath6 can be appreciably different from the sm prediction , while the branching ratio of @xmath74 and @xmath30 are still within the experimental ranges .we find that a sizeable weak phase associated with the @xmath63 coupling is favored in the ranges of @xmath210 , depending upon the @xmath126 and @xmath39 parameter choices .we have also studied the influence of the new @xmath5 on the @xmath211 decay .the @xmath185 data do not restrict the choice of @xmath118 .the @xmath7 constraint from the data can be satisfied if large couplings are taken .though the discrepancy between the observed branching ratio and the sm prediction can be explained with this @xmath5 effect , we can not explain both branching ratio and cp asymmetries constraints with a common weak angle .combining with the constraints from the @xmath74 decays , @xmath212 and @xmath185 , we find that a value of @xmath118 around @xmath191 is favored . we have observed that the @xmath8 asymmetries of the @xmath143 mode are more sensitive to the effects than the @xmath213 decay .this is because of a cancellation between different parts ( @xmath123 versus @xmath214 and @xmath215 ) in the @xmath164 wavefunction .finally , we have investigated @xmath35 decay , now being searched for at fermilab , in the same @xmath5 model . this process can be dominated by the @xmath5 contribution and the branching ratio can reach the expected sensitivity in run ii .c. would like to thank a. kagan , d. morrissey , j. rosner , c. wagner , and l. wolfenstein for useful discussions and comments .this work was supported in part by the united states department of energy under grants no .ey-76 - 02 - 3071 and no .de - fg02 - 95er40896 , and in part by the wisconsin alumini research foundation .j. h. christenson , j. w. cronin , v. l. fitch and r. turlay , phys .* 13 * ( 1964 ) 138 .n. cabibbo , phys .* 10 * ( 1963 ) 531 .m. kobayashi and t. maskawa , prog .* 49 * ( 1973 ) 652 .t. browder , presented at 21st international symposium on lepton and photon interactions at high energies ( lp 03 ) , batavia , illinois , 1116 august 2003 .y. grossman and m. p. worah , phys .b * 395 * , 241 ( 1997 ) [ arxiv : hep - ph/9612269 ] .r. fleischer , int .j. mod .a * 12 * , 2459 ( 1997 ) [ arxiv : hep - ph/9612446 ] .d. london and a. soni , phys .b * 407 * , 61 ( 1997 ) [ arxiv : hep - ph/9704277 ] .r. fleischer and t. mannel , phys .b * 511 * , 240 ( 2001 ) [ arxiv : hep - ph/0103121 ] .g. hiller , phys .d * 66 * , 071502 ( 2002 ) [ arxiv : hep - ph/0207356 ] .m. ciuchini and l. silvestrini , phys .lett . * 89 * , 231802 ( 2002 ) [ arxiv : hep - ph/0208087 ] . y. grossman , g. isidori and m. p. worah , phys . rev .d * 58 * , 057504 ( 1998 ) [ arxiv : hep - ph/9708305 ] .k. hagiwara _ et al ._ [ particle data group collaboration ] , phys . rev .d * 66 * , 010001 ( 2002 ) .c. w. chiang and j. l. rosner , phys . rev .d * 68 * , 014007 ( 2003 ) [ arxiv : hep - ph/0302094 ] . for recent model studies , see a. datta , phys .d * 66 * , 071702 ( 2002 ) [ arxiv : hep - ph/0208016 ] ; m. raidal , phys .lett . * 89 * , 231803 ( 2002 ) [ arxiv : hep - ph/0208091 ] ; b. dutta , c. s. kim and s. oh , phys .rev . lett . * 90 * , 011801 ( 2003 ) [ arxiv : hep - ph/0208226 ] ; s. khalil and e. kou , phys .d * 67 * , 055009 ( 2003 ) [ arxiv : hep - ph/0212023 ] ; g. l. kane , p. ko , h. b. wang , c. kolda , j. h. park and l. t. wang , arxiv : hep - ph/0212092 ; s. khalil and e. kou , arxiv : hep - ph/0303214 ; s. baek , phys .d * 67 * , 096004 ( 2003 ) [ arxiv : hep - ph/0301269 ] ; a. kundu and t. mitra , phys .d * 67 * , 116005 ( 2003 ) [ arxiv : hep - ph/0302123 ] ; r. arnowitt , b. dutta and b. hu , arxiv : hep - ph/0307152 . c. dariescu , m. a. dariescu , n. g. deshpande and d. k. ghosh , arxiv : hep - ph/0308305 .a. k. giri and r. mohanta , phys .d * 68 * , 014020 ( 2003 ) [ arxiv : hep - ph/0306041 ] .d. e. morrissey and c. e. wagner , arxiv : hep - ph/0308001 .for a recent discussion , see p. langacker , [ arxiv : hep - ph/0308033 ] .f. abe _ et al . _[ cdf collaboration ] , phys .lett . * 79 * , 2192 ( 1997 ) .j. erler and p. langacker , phys .b * 456 * , 68 ( 1999 ) [ arxiv : hep - ph/9903476 ] ; phys .rev . lett .* 84 * , 212 ( 2000 ) [ arxiv : hep - ph/9910315 ] , and references theirin .the lep electroweak working group and sld heavy flavour group , [ arxiv : hep - ex/0212036 ] .e. nardi , phys .d * 48 * , 1240 ( 1993 ) [ arxiv : hep - ph/9209223 ] ; j. bernabeu , e. nardi and d. tommasini , nucl .b * 409 * , 69 ( 1993 ) [ arxiv : hep - ph/9306251 ] ; y. nir and d. j. silverman , phys .d * 42 * , 1477 ( 1990 ) ; v. d. barger , m. s. berger and r. j. phillips , phys .d * 52 * , 1663 ( 1995 ) [ arxiv : hep - ph/9503204 ] ; m. b. popovic and e. h. simmons , phys .d * 62 * , 035002 ( 2000 ) [ arxiv : hep - ph/0001302 ] . k. s. babu , c. f. kolda and j. march - russell , phys .d * 54 * , 4635 ( 1996 ) [ arxiv : hep - ph/9603212 ] ; _ ibid . _ * 57 * , 6788 ( 1998 ) [ arxiv : hep - ph/9710441 ] .t. g. rizzo , phys .d * 59 * , 015020 ( 1999 ) [ arxiv : hep - ph/9806397 ] .k. leroux and d. london , phys .b * 526 * , 97 ( 2002 ) [ arxiv : hep - ph/0111246 ] .s. chaudhuri , s. w. chung , g. hockney and j. lykken , nucl .b * 456 * , 89 ( 1995 ) [ arxiv : hep - ph/9501361 ] .g. cleaver , m. cvetic , j. r. espinosa , l. l. everett , p. langacker and j. wang , phys .d * 59 * , 055005 ( 1999 ) [ arxiv : hep - ph/9807479 ] .l. l. chau and w. y. keung , phys .lett . * 53 * , 1802 ( 1984 ) .h. y. cheng and k. c. yang , phys .d * 64 * , 074004 ( 2001 ) [ arxiv : hep - ph/0012152 ] .g. buchalla , a. j. buras and m. e. lautenbacher , rev .* 68 * , 1125 ( 1996 ) [ arxiv : hep - ph/9512380 ] .a. ali and c. greub , phys .d * 57 * , 2996 ( 1998 ) [ arxiv : hep - ph/9707251 ] ; a. ali , g. kramer and c. d. lu , phys .d * 58 * , 094009 ( 1998 ) [ arxiv : hep - ph/9804363 ] ; y. h. chen , h. y. cheng , b. tseng and k. c. yang , phys .d * 60 * , 094014 ( 1999 ) [ arxiv : hep - ph/9903453 ] ; h. y. cheng and k. c. yang , phys .d * 62 * , 054029 ( 2000 ) [ arxiv : hep - ph/9910291 ] .d. atwood and g. hiller , arxiv : hep - ph/0307251 .l. wolfenstein , phys .lett . * 51 * , 1945 ( 1983 ) .m. battaglia _ et al ._ , arxiv : hep - ph/0304132 . z. j. xiao , c. s. li and k. t. chao , phys .d * 63 * , 074005 ( 2001 ) [ arxiv : hep - ph/0010326 ] .lep @xmath1 oscillation working group , presented at xxxviii rencontres de moriond , electroweak interactions and unified theories , les arcs , france , march 15 - 22 , 2003 .m. wirbel , b. stech and m. bauer , z. phys .c * 29 * , 637 ( 1985 ) ; m. bauer , b. stech and m. wirbel , z. phys .c * 34 * , 103 ( 1987 ) ; j. g. korner and g. a. schuler , z. phys .c * 38 * , 511 ( 1988 ) [ erratum - ibid .c * 41 * , 690 ( 1988 ) ] ; m. bauer and m. wirbel , z. phys .c * 42 * , 671 ( 1989 ) .babar collaboration , b. aubert _ et al ._ , slac report no .slac - pub-10086 , arxiv : hep - ex/0309025 .belle collaboration , k .- f .chen _ et al ._ , arxiv : hep - ex/0307014 ; cleo collaboration , r. a. briere _ et al ._ , phys .lett . * 86 * , 3718 ( 2001 ) [ arxiv : hep - ex/0101032 ] ; babar collaboration , b. aubert _ et al ._ , slac report no .slac - pub-9698 , hep - ex/0303046 .h. aihara , talk presented at fpcp2003 ( flavor physics and @xmath8 violation ) , paris , june 2003 .cleo collaboration , s. j. richichi _ et al ._ , phys .lett . * 85 * , 520 ( 2000 ) [ hep - ex/9912059 ] .h. j. lipkin , phys .b * 254 * , 247 ( 1991 ) ; _ ibid ._ b * 415 * , 186 ( 1997 ) [ arxiv : hep - ph/9710342 ] ._ b * 433 * ( 1998 ) 117 . c. w. chiang and j. l. rosner , phys .d * 65 * , 074035 ( 2002 ) [ erratum - ibid .d * 68 * , 039902 ( 2003 ) ] [ arxiv : hep - ph/0112285 ] ; c. w. chiang , m. gronau and j. l. rosner , arxiv : hep - ph/0306021 , to appear in phys .d ; c. w. chiang , m. gronau , z. luo , j. l. rosner and d. a. suprun , arxiv : hep - ph/0307395 , to appear in phys .d. m. beneke and m. neubert , nucl .b * 651 * , 225 ( 2003 ) [ arxiv : hep - ph/0210085 ] .k. s. babu and c. f. kolda , phys .lett . * 84 * , 228 ( 2000 ) [ arxiv : hep - ph/9909476 ] ; p. h. chankowski and l. slawianowska , phys .d * 63 * , 054012 ( 2001 ) [ arxiv : hep - ph/0008046 ] ; c. bobeth , t. ewerth , f. kruger and j. urban , phys .d * 64 * , 074014 ( 2001 ) [ arxiv : hep - ph/0104284 ] ; a. dedes , h. k. dreiner and u. nierste , phys .lett . * 87 * , 251804 ( 2001 ) [ arxiv : hep - ph/0108037 ] ; a. j. buras , p. h. chankowski , j. rosiek and l. slawianowska , phys .b * 546 * , 96 ( 2002 ) [ arxiv : hep - ph/0207241 ] ; j. k. mizukoshi , x. tata and y. wang , phys .d * 66 * , 115003 ( 2002 ) [ arxiv : hep - ph/0208078 ] ; a. j. buras , p. h. chankowski , j. rosiek and l. slawianowska , nucl .b * 659 * , 3 ( 2003 ) [ arxiv : hep - ph/0210145 ] .m. nakao , presented at 21st international symposium on lepton and photon interactions at high energies ( lp 03 ) , batavia , illinois , 1116 august 2003 .for a general discussion on the effects of non - standard @xmath21 couplings in @xmath216 transitions , see g. buchalla , g. hiller and g. isidori , phys .d * 63 * , 014015 ( 2001 ) [ arxiv : hep - ph/0006136 ] .

<S> we study the effects of an extra @xmath0 gauge boson with flavor changing couplings with fermion mass eigenstates on certain @xmath1 </S><S> meson decays that are sensitive to such new physics contributions . </S><S> in particular , we examine to what extent the current data on @xmath2 , @xmath3 and @xmath4 decays may be explained in such models , concentrating on the example in which the flavor changing couplings are left - chiral . </S><S> we find that within reasonable ranges of parameters , the @xmath5 contribution can readily account for the anomaly in @xmath6 but is not sufficient to explain large branching ratio of @xmath3 with the same parameter value . </S><S> @xmath6 and @xmath7 are seen to be the dominant observables that constrain the extra weak phase in the model . </S>

solv-int9903014

it is well - known that solutions for the painlev equations play a role of special functions in the nonlinear science @xcite .originally , painlev derived these equations in order to find new transcendental functions determined by second order ordinary differential equations possessing so - called the painlev property .recently , irreducibility of solutions of the painlev equations has been proved by umemura @xcite .umemura first gave a rigorous definition of classical functions : starting from the field of rational functions , if a function is obtained by finite numbers of iterations of the following permissible operations , * differentiation , * arithmetic calculations , * solving homogeneous linear ordinary differential equation , * substitution into abelian functions , then that function is called `` classical '' .umemura proved that solutions of the painlev equations are not classical in general in the above sense .however , it is known that they admit classical solutions for special values of parameters .one is the rational or algebraic solutions , and another is the transcendental classical solutions which are expressed by rational functions in various special functions and their derivatives . in this article , we discuss the painlev ii equation ( p@xmath0 ) , @xmath2 and the equation of no . 34 in gambier s classification@xcite , @xmath3 which we call the painlev xxxiv equation ( p@xmath4 ) . in equations ( [ p2 ] ) and ( [ p34 ] ) , we adopted different scale from their canonical forms @xcite , @xmath5 @xmath6 for the clarity of expression of solutions .the scale transformations , @xmath7 lead the canonical forms to our ones .we denote @xmath8 , p@xmath9 $ ] and @xmath10 , p@xmath11 $ ] , respectively , when it is necessary to show the value of parameter explicitly .these equations are related to each other by the miura transformation @xcite , @xmath12 and its complement , @xmath13 umemura also proved that p@xmath0 ( and thus p@xmath4 ) admits transcendental classical solutions which are expressed by the airy function when @xmath14 is an integer , rational solutions when @xmath14 is a half - integer , and that otherwise the solutions are non - classical .it is known that the painlev equations ( except for p@xmath15 ) admit the bcklund transformations(bt ) which form the affine weyl group @xcite .for example , bt of p@xmath0 is given by , @xmath16 and @xmath17 forms the affine weyl group of type @xmath18 . starting from a suitable `` seed '' solution, we obtain `` higher '' solutions by applying bt to it , which are expressed by rational functions in the seed solution and its derivatives . it is interesting and important to note two points on classical solutions for the painlev equations .the first is that , in general , the classical solutions are located on special points in the parameter space from the viewpoint of symmetry .namely , transcendental classical solutions are on the walls of the weyl chambers , and rational solutions on their barycenters .the second is that the classical solutions have additional structure .namely , it is known that some of the classical solutions admit such determinant formulas that they are expressed by log derivative of the ratio of some determinants .moreover , application of bt corresponds to increment of size of the determinant .in fact , both airy function type and rational solutions for p@xmath0 and p@xmath4 admit such determinant structure . then, does such determinant structure exist even for non - classical solutions , or are non - classical solutions so `` transcendental '' that do not admit even such structure ? herewe also note that such determinant structure are shown to be universal among the soliton equations due to the celebrated sato theory @xcite .the purpose of this article is to generalize the determinant formulas of the classical solutions of p@xmath0 and p@xmath4 , and show that such determinant structure is universal among the solutions of p@xmath0 and p@xmath4 .the key to this generalization is to use the technique of the darboux transformation together with hirota s bilinear formalism .this article is organized as follows . in section[ classical ] , we summarize the derivation of p@xmath4 and classical solutions of p@xmath0 and p@xmath4 .we illustrate our basic ideas in section [ darboux ] .our main results are presented in section [ result ] .the proof for our results are given in section [ proof ] .section [ conc ] is devoted to summary and discussions .it is well - known that p@xmath0 is derived as the similarity reduction from the modified kdv equation @xcite , @xmath19 in fact , putting @xmath20 equation ( [ mkdv ] ) is reduced to equation ( [ p2 ] ) .similarly , p@xmath4 ( [ p34 ] ) is derived as the similarity reduction from the kdv equation , @xmath21 by putting @xmath22 it is known that p@xmath0 ( [ p2 ] ) is bilinearized to @xcite @xmath23g\cdot f=0,\label{p2:bl2}\end{aligned}\ ] ] by the dependent variable transformation , @xmath24 where @xmath25 is an arbitrary function in @xmath26 , and @xmath27 is hirota s bilinear differential operator defined by @xmath28 the bilinear equations ( [ p2:bl1 ] ) and ( [ p2:bl2 ] ) are regarded as those for p@xmath0 , but it is also possible to derive p@xmath29 as follows . first we divide equations ( [ p2:bl1 ] ) and ( [ p2:bl2 ] ) by @xmath30 .then , putting @xmath31 and using the formulas @xcite , @xmath32\psi,\end{aligned}\ ] ] we obtain @xmath33 compatibility condition of the linear equations ( [ lin1:p34 ] ) and ( [ lin2:p34 ] ) gives p@xmath11 $ ] .let us discuss the determinant expressions for the classical solutions of p@xmath0 and p@xmath29 @xcite . in the following, the determinant with size zero should be regarded as @xmath34 .if we choose @xmath35 , then it is known that the bilinear equations ( [ p2:bl1 ] ) and ( [ p2:bl2 ] ) are satisfied by @xmath36 @xmath37 where @xmath38 and @xmath39 satisfies the airy equation , @xmath40 thus , the airy function type solutions of p@xmath41 $ ] and p@xmath42 $ ] are given by @xmath43 respectively .the rational solutions are obtained for the case of @xmath44 , and there are two expressions for them .one is the schur function type expression given by @xmath45 @xmath46 where @xmath47 are the polynomials in @xmath26 defined by @xmath48 another expression is given by hankel determinant as @xmath49 @xmath50 where @xmath51 , @xmath52 are the polynomials in @xmath26 , which are recursively defined by @xmath53 thus , the rational solutions for p@xmath54 $ ] and p@xmath55 $ ] are given by @xmath56 respectively .we note that it is possible to generalize the above result for negative @xmath14 by the reflection symmetry of p@xmath0 and p@xmath4 , @xmath57 and @xmath58 respectively .indeed , this symmetry is generated by @xmath59 in equations ( [ bt : s ] ) and ( [ bt : t ] ) .the technique of darboux transformation is well developed in the soliton theory @xcite .for example , in the case of kdv equation ( [ kdv ] ) , we start with its auxiliary linear problem , @xmath60 where @xmath61 is the spectral parameter .then , one can show that equations ( [ lin : kdv ] ) are covariant with respect to the darboux transformation @xmath62 $ ] , @xmath63 $ ] defined by @xmath64=\frac{wr(\psi_1,\psi_2,\cdots,\psi_n,\psi)}{wr(\psi_1,\psi_2,\cdots,\psi_n)},\ ] ] @xmath65=w+2\frac{\partial^2}{\partial x^2}\log wr(\psi_1,\psi_2,\cdots,\psi_n),\ ] ] where @xmath66 is a solution of the linear equations ( [ lin : kdv ] ) with @xmath67 , and @xmath68 is a wronskian with respect to the indicated functions .thus , choosing an appropriate seed solution of kdv equation as @xmath69 , one can construct series of exact solutions by using this method .for example , starting from the solution @xmath70 , then we obtain the wronskian expression of @xmath71-soliton solution @xcite .although most of the expression would be somewhat formal except for the soliton and rational type solutions or their variants , it is important that we can express quite wide class of solutions in terms of determinant .the above result is recovered by the following bilinear equations : @xmath72 @xmath73g\cdot f=0 , \label{g - bl2:kdv}\end{aligned}\ ] ] where @xmath74 , @xmath75 .we omit the detail , but we can directly prove that these bilinear equations holds as the identity of the determinants from the linear equations ( [ lin : kdv ] ) .indeed , we recover `` usual '' bilinear equations by putting @xmath70 , @xmath76 @xmath77 > from these bilinear equations , we recover the kdv equation ( [ kdv ] ) and the auxiliary linear problem ( [ lin : kdv ] ) by putting as @xmath78 now , keeping this correspondence between the darboux transformation and the bilinear formalism in mind , let us illustrate our strategy to generalize the determinant formulas for the solutions of p@xmath0 and p@xmath4 .we have the bilinear equations ( [ p2:bl1 ] ) and ( [ p2:bl2 ] ) for p@xmath4 . clearly , these bilinear equations correspond to equations ( [ bl1:kdv ] ) and ( [ bl2:kdv ] ) . thus , if we could obtain `` generalized '' bilinear equations which correspond to equations ( [ g - bl1:kdv ] ) and ( [ g - bl2:kdv ] ) , it might be possible to find the darboux transformation for p@xmath4 and p@xmath0 which leaves the linear equations ( [ lin1:p34 ] ) and ( [ lin2:p34 ] ) covariant .then we obtain a generalization of the determinant formulas . in the next section, we present our main results .in this section , we present a generalization of determinant formulas for the solutions of p@xmath0 and p@xmath4 .we start with a pair of linear equations ( [ lin1:p34 ] ) and ( [ lin2:p34 ] ) . as we mentioned in section [ classical ] , these equations are compatible provided that @xmath69 satisfies p@xmath79 $ ] . in other words , equation ( [ lin2:p34 ] ) follows if @xmath69 and @xmath80 are solutions of p@xmath11 $ ] and equation ( [ lin1:p34 ] ) , respectively. then we have the following theorem . [ main1 ]let @xmath69 be a solution of p@xmath11 $ ] , and @xmath81 be a solution of the linear equation ( [ lin1:p34 ] ) .we define two sequences @xmath82 , @xmath83 ( @xmath52 ) by @xmath84 we define hankel determinant @xmath85 , @xmath86 by @xmath87 then , @xmath88 satisfies @xmath89\psi_n = 0 , \label{lin3:p34}\\ \left(\frac{d^2}{dz^2}+w-2z\right)\psi_n=0 , \label{lin4:p34}\end{aligned}\ ] ] where @xmath90 thus , @xmath91 satisfies p@xmath92 $ ] .similar formula for p@xmath0 is obtained by applying the miura transformation ( [ miura ] ) . the linear equations ( [ lin1:p34 ] ) and ( [ lin2:p34 ] )are reduced to @xmath93 then we have the following theorem .[ main2 ] let @xmath94 be a solution of p@xmath9 $ ] , and @xmath85 be hankel determinant given by equation ( [ tau : p34 ] ) , where @xmath82 and @xmath83 are defined recursively by @xmath95 @xmath96 @xmath97 then , @xmath98 satisfies p@xmath99 $ ] . as mentioned in the previous section , theorems[ main1 ] and [ main2 ] is the direct consequence of the following proposition : [ bl ] the following bilinear equations hold .\tau_{n+1}\cdot\tau_n=0.\label{bl2}\end{aligned}\ ] ] in fact , dividing equations ( [ bl1 ] ) and ( [ bl2 ] ) by @xmath101 and using the formula @xcite @xmath102\psi_n,\end{aligned}\ ] ] we obtain the linear equations ( [ lin3:p34 ] ) and ( [ lin4:p34 ] ) in theorem [ main1 ] .similarly , dividing equations ( [ bl1 ] ) and ( [ bl2 ] ) by @xmath103 and using the formulas @xcite ,@xmath104 with @xmath105 we get @xmath106 eliminating @xmath107 , we obtain @xmath108 these results imply that even non - classical solutions possess the same determinant structure as the classical solutions .indeed , the known determinant expressions for the classical solutions are recovered as special cases .in fact , starting with a solution @xmath109 for p@xmath110 $ ] , the linear equation ( [ lin1:p34 ] ) yields @xmath111 from which we get @xmath112 without losing generality . then , we have from the recursion relations ( [ rec0:p34 ] ) , ( [ rec1:p34 ] ) and ( [ rec2:p34 ] ) , @xmath113 @xmath114 thus , we have a series of rational solutions given by equations ( [ tau : p34 ] ) , ( [ rec : rational : p34 ] ) , ( [ rec2:rational : p34 ] ) and @xmath115 the case of @xmath116 agrees with the hankel determinant expression for the rational solutions discussed in section [ classical ] . next , noticing that @xmath70 is a solution of p@xmath117 $ ] , we find from equation ( [ lin2:p34 ] ) that @xmath81 is determined by @xmath118 and the recursion relation ( [ rec1:p34 ] ) with equation ( [ rec0:p34 ] ) is reduced to @xmath119 thus , we have a series of solutions given by equations ( [ tau : p34 ] ) , ( [ airy2 ] ) , ( [ rec : airy ] ) and @xmath120 for @xmath116 , which agrees with the wronskian expression for the airy function type solutions discussed in section [ classical ]. however , the airy function type solutions are `` singular '' case in our formula . from equations ( [ rec0:p34 ] ) and ( [ rec2:p34 ] ) , we find that @xmath121 for all @xmath122 and thus @xmath123 for @xmath124 , which does not give meaningful result .we note that this phenomena is related to the symmetry of the @xmath125 sequence which will be mentioned in section [ conc ] , and the correct result for @xmath124 case is obtained by virtue of symmetry .we also note that this is the only singular case and our determinant formula works for other cases .in this section , we give the proof of the theorems [ main1 ] and [ main2 ] . since we see that these theorems follows immediately from proposition [ bl ] , it is sufficient to prove it .we have to prove both @xmath126 and @xmath124 cases , but we concentrate on the former case , since the latter case is proved in similar manner . the bilinear equations ( [ bl1 ] ) and ( [ bl2 ] ) are reduced to the plcker relations , which are quadratic identities of the determinants whose columns are shifted. thus , we first construct such differential formulas that shifted determinants are expressed by operating some differential operator on the original determinant .for this purpose , we introduce the following notation : let @xmath127 be a young diagram @xmath128 .then we define an @xmath129 determinant @xmath130 by @xmath131 then , we have the following differential formulas . [ shift ] @xmath132\tau_n,\label{shift5}\\ \hspace*{-40pt}\tau_n{}_{\young{3}}-\tau_n{}_{\young{12}}+\tau_n{}_{\young{111 } } = \left[2z\frac{d}{dz}+2n^2 + 4n(\alpha-1 ) -\frac{1}{4}\frac{dw}{dz}-3(\alpha-1)\right]\tau_n.\label{shift6}\end{aligned}\ ] ]the proof for proposition [ shift ] is an important step .however , since this requires straightforward but tedious calculations , we will give it in [ proof - shift ] .finally , we prove proposition [ bl ] . > from the plcker relations we have , @xmath133 @xmath134 by using proposition [ shift ] , we get the bilinear relations ( [ bl1 ] ) and ( [ bl2 ] ) from equations ( [ pl1 ] ) and ( [ pl2 ] ) , respectively , which is the desired result .in this article , we have presented determinant formulas for the solutions of p@xmath4 and p@xmath0 which are valid also for non - classical solutions by using the technique of the darboux transformation and bilinear formalism .the solutions of p@xmath92 $ ] and p@xmath99 $ ] , @xmath86 , are expressed by determinants whose entries are constructed from the solution of some linear equations .moreover , coefficients of those linear equations include the solution of p@xmath11 $ ] and p@xmath9 $ ] , respectively .we have also shown that known determinant expressions for classical solutions are recovered as special cases .this result implies that determinant structure of the classical solutions is universal among the solutions of p@xmath0 and p@xmath4 .finally , let us discuss the relation with the toda equation , which will be a key for generalization to other painlev equations .in general , the @xmath125 function for p@xmath0 is introduced through its hamiltonian @xcite , @xmath135 by @xmath136 we note that we obtain p@xmath9 $ ] for @xmath94 from the canonical equation , @xmath137 then it can be shown that @xmath8 , which is a solution of p@xmath9 $ ] , is expressed as @xmath138 by applying bt , we obtain a sequence of @xmath125 functions @xmath139 , where @xmath140 .okamoto has shown that bt of p@xmath0 is governed by the toda equation on the level of @xmath125 function : ( okamoto @xcite ) @xmath85 satisfies the toda equation , @xmath141 where @xmath142 is a non - zero constant .it is well - known that the solution of the toda equation ( [ toda ] ) is expressed by @xmath143 where @xmath39 is an arbitrary function in @xmath26 .equation ( [ toda : darboux ] ) is sometimes referred as `` darboux s formula '' .however , one should note that darboux s formula ( [ toda : darboux ] ) is valid under the condition , @xmath144 ( in practice , @xmath145 can be a constant . ) in @xcite , it is pointed out that the wronskian expression for the airy function type solutions of p@xmath0 is a consequence of the darboux s formula .we demonstrate how we could apply the darboux s formula for this case .we have a solution of p@xmath0 for @xmath146 , @xmath147 therefore , we can choose the @xmath125 functions as @xmath148 then we have a @xmath125 sequence for the airy function type solutions , @xmath149 now , as was mentioned in section [ classical ] , we have a reflection symmetry ( [ sym : u ] ) on @xmath94 which implies a symmetry on the @xmath125 function , @xmath150 in the case of the airy function type solutions , this symmetry induces a symmetry on the @xmath125 sequence as @xmath151 by virtue of this symmetry , we see that the @xmath125 sequence is divided into two parts as @xmath152 fortunately enough , since it can be shown that this @xmath125 sequence is governed by the toda equation ( [ toda ] ) with @xmath153 , we could apply darboux s formula for @xmath126 and @xmath124 separately. we note that if we prolong the @xmath125 sequence for @xmath124 following to the toda equation ( [ toda ] ) with the condition ( [ bc : darboux ] ) without taking the symmetry into account , we have @xmath123 for @xmath124 .this corresponds to the `` singular '' phenomena mentioned in section [ result ] . in the case of rational solutions ,the @xmath125 sequence is again separated into two parts .however , we can not apply the darboux s formula to this case .let us take a solution of p@xmath0 for @xmath154 , @xmath155 we can choose the @xmath125 function as @xmath156 and we have a @xmath125 sequence for the rational solutions . now the symmetry ( [ sym : tau ] ) implies a symmetry on the @xmath125 sequence , @xmath157 and thus we have , @xmath158 and again the @xmath125 sequence is separated into two parts .however in this case , these @xmath125 functions are shown to satisfy the toda equation of the form , @xmath159 thus , we can not apply darboux s formula for the rational solutions .of course , by introducing a gauge on the @xmath125 function as @xmath160 then @xmath161 satisfy the toda equation @xmath162 however , now @xmath163 is not a constant .there are two points for being able to apply the darboux s formula .the first is that @xmath125 sequence should have a symmetry which is induced from the painlev equation itself so that the @xmath125 sequence is separated into two parts , which is necessary to apply the darboux s formula without inconsistency .the second is that @xmath125 sequence should satisfy the toda equation of the form ( [ toda ] ) under the condition ( [ bc : darboux ] ) .both conditions are satisfied for the airy function type solution , but the second condition does not hold for rational solutions . now , for solutions which corresponds to generic @xmath14 , the symmetry on the @xmath125 function ( [ sym : tau ] ) does not induce any symmetry on the @xmath125 sequence .this observation shows that it is not trivial that generic @xmath125 function admit determinant formula , even if it satisfies the toda equation . despite of unavailability of the darboux s formula ,we could obtain the determinant formula for rational solutions .this is due to the determinant formula of the general solution of toda equation of c - type ( toda equation with the symmetry ( [ sym : c - toda ] ) ) obtained in @xcite .conversely , the general determinant formula for p@xmath0 strongly implies that it is possible to construct the determinant formula for the general solution of toda equation in general setting , i.e. , the toda equation admits the determinant formula for the general solution without any symmetries or gauge on @xmath125 sequence .then , our results might be regarded as the special case of such general solution for the toda equation .moreover , it is known that the bcklund transformations for the painlev equations ( except for p@xmath15 ) is governed by various types of toda equations @xcite .thus , it might be possible to present general determinant formula also for other painlev equations .we shall work out this problem in the next publication .the authors would like to thank dr .y. ohta for valuable suggestions .they also thank prof .h. umemura , prof .m. noumi and prof .y. yamada for interest to our work and discussions .one of the authors ( k k ) is supported by the grant - in - aid for encouragement of young scientists , the ministry of education , science and culture of japan , no . 09740164 .here , we give the proof for proposition [ shift ] .we first prove equation ( [ shift1 ] ) . notice that @xmath164 is expressed by @xmath165 where @xmath166 is the @xmath167-cofactor of @xmath85 and @xmath168 denotes a standard scalar product for @xmath169 matrices @xmath170 and @xmath171 which is defined as @xmath172 the first matrix of equation ( [ shift1-pro ] ) is rewritten by using the recursion relation ( [ rec1:p34 ] ) as @xmath173 the above matrix in the second term is separated as @xmath174 each of these terms gives zero contribution in equation ( [ shift1-pro ] ) . hence we obtain equation ( [ shift1 ] ) .next we prove equation ( [ shift2 ] ) .we consider @xmath175 where @xmath176 is @xmath167-cofactor of @xmath164 .the first matrix in the right - hand side is equal to @xmath177 taking the scalar product , the first and second terms give @xmath178 and @xmath179 , respectively , and the third term vanishes .hence we have equation ( [ shift2 ] ) .next we prove equation ( [ shift3 ] ) .we consider the following equality , @xmath180 the first matrix of right - hand side is rewritten as @xmath181 here we note that @xmath82 s also satisfy @xmath182 which is proved by induction from equations ( [ lin1:p34 ] ) , ( [ lin2:p34 ] ) and ( [ rec1:p34 ] ) . the first term of the right hand side of equation ( [ shift3:first ] )is rewritten using equation ( [ cont ] ) as @xmath183 applying the scalar product on these terms , we obtain equation ( [ shift3 ] ) .we get equations ( [ shift4])-([shift6 ] ) by similar calculations .99 see , for example , levi d and winternitz p 1992 _ painlev transcendents : their asymptotics and physical applications _( nato asi series b278)(new york : prenum ) umemura h and watanabe h 1997 _ nagoya math .j. _ * 148 * 151 umemura h _ irreducibility of the painlev equations evolution in the past 100 years _ , preprint , to appear in _ proceeding of the workshop on the painlev transcendents _( montreal , 1996 ) ince e l 1956 _ ordinary differential equations _( new york : dover ) matveev v b and salle m a 1991 _ darboux transformations and solitons _ ( berlin : springer - verlag ) satsuma j 1976 _ j. phys .jpn . _ * 46 * 359 ablowitz m j and segur h 1981 _ solitons and inverse scattering transform _( philadelphia : siam ) okamoto k 1987 _ ann . mat .pura appl . _* 146 * 337 .okamoto k 1987 _ japan j. math . _* 13 * 47 okamoto k 1986 _ math .ann . _ * 275 * 221 okamoto k 1987 _ funkcialaj ekvacioj _ * 30*(1987 ) 305 sato m and sato y 1981 _ nonlinear partial differential eequations in applied science _( tokyo : kinokuniya / north - holland ) p 259 ohta y , satsuma j , takahashi d and tokihiro t 1988 _ prog ._ * 94 * 210 kajiwara k and ohta y 1996 _ j. math. phys . _ * 37 * 4693 nakamura a 1992 _ j. phys .* 61 * 3007 hirota r and satsuma j 1976 _ prog .* 59 * 64 hirota r 1992 _ mathematical theory of solitons _ [ in japanese ] ( tokyo : iwanami )

<S> a generalization of determinant formulas for the classical solutions of painlev xxxiv and painlev ii equations are constructed using the technique of darboux transformation and hirota s bilinear formalism . </S><S> it is shown that the solutions admit determinant formulas even for the transcendental case . </S><S> [ section ] [ th]proposition [ th]lemma [ th]definition [ th]remark # 1#2#3 = # 1 = 1 = # 2pt by .#3pt # 1 [ determinant formulas for p@xmath0 and p@xmath1 </S>

astro-ph9707194

stephan s quintet ( sq ) epitomizes the problems that compact groups pose for our ideas about galaxy formation and evolution .it is one of the most luminous and high surface brightness aggregates included in the first reasonably complete catalog ( hickson 1982 ; hereafter hcg ) of compact groups ( hcg92 ) .questions center around how many of the galaxies are in close proximity to one another , as well as how long and violently they have been interacting .these questions are particularly relevant because one of the components , ngc 7320 , shows a discordant redshift . finally , there is the question of whether sq is representative of the compact group phenomenon .it is relevant for the latter question to point out that the accordant redshift part of sq satisfies the hcg selection criteria - stephan s quartet is also a compact group .sq is composed of a _ kernel _ of three galaxies ( ngc 7317 , 18a and 19 ) with very low velocity dispersion ( cz= 6563 , 6620 and 6650 respectively ; see moles et al .1997 ; hereafter paper i ) .ngc 7320 ( 800 ) and ngc 7318b ( 5765 ) complete the apparent compact group .another accordant redshift galaxy ( ngc 7320c , cz@xmath16000 ) lies 3 arcmin ene of ngc 7319 .ngc 7320c is sufficiently faint that its proximity does not violate the hcg isolation criteria .a much brighter spiral galaxy ngc 7331 lies 30 arcmin ne of sq and shows a redshift ( cz= 821 ) similar to ngc 7320 .all seven of these galaxies play a role in our interpretation of the dynamical state of sq .figure 1a shows a schematic of sq and environs , while figure 2a shows a wide band image of sq proper .wide field images can be found in the plates of arp and kormendy ( 1972 ) .we discuss new optical observations ( presented in paper i ) , as well as published x - ray and radio data .we use them to infer a dynamical history for sq .section 2 considers the relationship of discordant - redshift ngc 7320 and the remaining sq members .we consider the past and present dynamical states of the accordant group in sections 3 and 4 respectively . in section 5we consider the implications of sq as representative of the compact group phenomenon .several investigators have pointed out the low probability for ngc 7320 to be a chance projection on the accordant quartet in sq ( burbidge and burbidge 1961 ; arp 1973 ) .little weight would be given to such an _ a posteriori _ calculation were it not for the fact that discordant components are so numerous in compact groups ( @xmath143/100 groups in hcg contain at least one discordant - redshift member ; see sulentic 1994 ) .ngc 7320 shows some peculiarities ( see paper i ) that could be interpreted as evidence for interaction with the higher redshift sq members .the most significant involves a blue tidal tail that emerges from the se end of ngc 7320 ( arp and kormendy 1972 ) .if we assume that the tidal tail is evidence for recent galaxy - galaxy interaction , we are left with three possible interpretations : \(1 ) it involves only the high redshift members of sq . in this casengc 7320 is a foreground galaxy projected on one of the background manifestations of this interaction ; or \(2 ) it involves a past encounter between ngc 7320 and ngc 7331 . in this case , it is a foreground tail projected near the background high redshift quartet ; or \(3 ) it involves direct interaction between ngc 7320 and the accordant sq .this would require the assumption of non - doppler redshifts in one ( or four ) components because the velocity difference between the components is too high for a pure dynamical explanation .the tail emerging from ngc 7320 is parallel to a second narrower and brighter tail that emerges from one of the the spiral arms in ngc 7319 ( see figure 1b ) and extends in the direction of similar redshift ngc 7320c .this fact clearly favors hypothesis 1 because it relates the `` ngc7320 tail '' to one that is unambiguously related to interaction involving only higher redshift members .the tail emerging from ngc 7320 can actually be traced even farther than the brighter one , and it extends directly to ngc 7320c ( arp and lorre 1976 ) .ambiguity in the hypothesis 1 interpretation stems from the fact that ngc 7331 is located in the same direction , but beyond ngc 7320c .further support for hypothesis 1 comes from a low - redshift h@xmath2 image ( provided by w. keel ) shown in figure 2b .the distribution of hii regions in ngc 7320 is symmetric about its nucleus with no evidence for any low redshift h@xmath2 emission in the tidal tail .standard redshift - independent distance estimates for ngc 7320 as well as ngc 7317 and 18a , b have been attempted ( paper 1 ; see also kent 1981 ; sulentic 1994 ) .it is unclear whether techniques calibrated with normal galaxies can be reasonably applied to galaxies that are suspected to be , or are manifestly abnormal .the debate over the interpretation of the hi data illustrates this point very well ( sulentic 1994 ) . on the other hand ,we show in paper i that some of the galaxies in sq show only small disturbances ( including ngc 7320 , if the tail does not belong to it ) and , therefore employ normal distance estimators .indeed , those standard techniques yield distances consistent with the redshift - implied values , i. e. , @xmath1 10 mpc for the low- and @xmath1 65 mpc for the high - redshift components .an independent argument for two distances in sq involves consistency of sq redshifts with those measured in the surrounding field .galaxies projected within about one degree of sq show two preferred redshifts near cz= 800 and 6500 ( lynds 1972 ; materne and tammann 1974 ; allen and sullivan 1980 and shostak et al .the low redshift galaxies are shaded in figure 1a to distinguish them from galaxies with redshifts near 6500 .two additional galaxies with redshifts similar to ngc 7320 were identified in shostak et al .( 1984 ) and lie outside the field shown in figure 1a .the redshift data suggest that the two nearest supercluster structures in the direction of sq lie at distances of about 10 and 65 mpc respectively .rejection of the doppler interpretation for the high or low redshift parts of sq would require nearby galaxies to also show discordant redshifts or a fortuitous match of discordant redshifts in sq with cosmological shifts in the neighboring field .we interpret the kernel of three similar redshift members in sq as a physical triplet and the core of sq .these galaxies and ngc 7320c are hatched in figure 1b .the morphological evidence for physical membership in the triplet is least ambiguous for ngc 7319 .ngc 7317 and 7318a show early - type ( e2 ) morphologies and are therefore less sensitive to the effects of gravitational encounters .it is somewhat surprising to find an e - dominated triplet in the field at all , but this morphology might reflect past secular evolutionary effects within the group .however , neither of these ellipticals show an unusual color or signs of geometrical distortion ( schombert et al .1990 ; zepf et al 1991 ; paper i ) .the strongest evidence for physical membership of ngc 7317 and 18a comes from the diffuse optical light that surrounds them and that is probably caused by dynamical stripping processes ( arp 1973 ; arp and lorre 1976 ; schombert et al .1990 ; paper i ) .ngc 7318b with @xmath3v@xmath4850 relative to the triplet mean velocity is assumed to be a recent arrival and not relevant for the past dynamical history of sq ( see next section ) .the past dynamical history must account for evidence that points to past interactions , especially the following : ( 1 ) parallel tidal tails ( arp and kormendy 1972 ) ; ( 2 ) stripping of most of the hi from ngc 7319 ( shostak et al .1984 ) and ( 3 ) diffuse optical light that surrounds the triplet kernel .the first two observations are probably related because an extrapolation backwards of the brighter tail passes very close to the projected center of n7319 . the parallel extension of the tails and the fact that both point towards ngc7320c suggest a common origin involving the galaxy .simulations of galaxy interactions do not usually produce parallel tails ( e.g. howard et al . 1993 ) leading us to propose that we may be observing the remnants of two past encounters which suggest that ngc 7320c may be a loosely bound member of the system .the tails can reasonably be interpreted as mapping the trajectory of these past encounters if they were caused by ngc 7320c , an hypothesis that we favor here . in the recent past ngc7320c would have been inside the group and would have formed an accordant ( defined as @xmath3v@xmath510@xmath6 km s@xmath7from group median ) quartet with the triplet kernel .that accordant quartet would have satisfied the hcg selection criteria only if ngc 7320 were not superposed .we assume that ngc 7318b had not yet arrived on the scene .naturally the spiral member of the sq kernel ( ngc 7319 ) shows the most dramatic evidence for tidal disruption : 1 ) the spiral arms are asymmetric , with the eastern arm split into two concentric arcs and the western one ending in one of the tidal tails ; 2 ) the spiral arms show no hii regions on h@xmath2 images that show them in and near ngc 7318a , b and ngc 7320 ( arp 1973 ; paper i ) ; 3 ) ngc 7319 has been stripped of essentially all hi and 4 ) it shows a seyfert nucleus and associated radio / optical jet ( see paper i and aoki et al .figure 2c shows an h@xmath2 image ( again thanks to w. keel ) centered on ngc7319 that illustrates points 2 and 4 .high resolution 21 cm observations ( shostak et al . 1984 ) reveal two extended clouds of hi with redshift similar to the stable kernel ( see figure 1b ) .the total mass of stripped hi with velocity near 6500 km / s is 1.4@xmath810@xmath9 h@xmath7 m@xmath10 ( @xmath11 = h@xmath12/100 ) .this exceed the hi mass of luminous sb spirals like ngc 7319 by more than a factor of three and implies that some of the stripped gas may have come from other galaxies .the proposed intruder , ngc 7320c , shows a spiral or ringed morphology without evidence for any non - stellar material . the hi velocity listed for ngc 7320c in rc3 catalogis taken from old grid mapping with the arecibo telescope . higher resolution and sensitivity observations ( shostak et al .1984 ) report no hi at the position of ngc 7320c , therefore some of the stripped hi within sq may have originated from that galaxy as well .we conclude that most of the damage to ngc 7319 was caused by a direct collision with ngc 7320c ( see also shostak et al .. the twin tails and diffuse light in sq show blue colors consistent with recent or ongoing star formation .grid photometry obtained for sq by schombert et al .( 1990 ) find b@xmath13v @xmath14 0.57 for the brighter tail which is almost as blue as the foreground sd spiral ngc 7320 .the luminosity of this tail represents about 18% of the luminosity of ngc 7319 .the brightest part of the fainter tail was also detected and showed colors in the range 0.1 - 0.7 .similarly the diffuse light shows colors similar to the disks of spiral galaxies b - v= 0.5 - 0.7 . the interstellar matter ( ism ) in ngc 7319 andpossibly other components were stripped and heated to @xmath110@xmath15 k ( by analogy to the similar ongoing event involving ngc 7318b , see next section ) . the cloudswould then expand , recombine and cool over a time - scale @xmath15@xmath810@xmath0 years .deep images reveal condensations consistent with the size , color and brightness of hii regions ( or blue clusters ) scattered throughout the halo ( see e.g. arp and lorre 1976 ) .the data supports the hypothesis that much of the stripping occurred in the past few 10@xmath0 years and also that some tidally induced star formation is occurring or has recently occurred , in the halo and tails .if the tidal tails map the trajectory of ngc 7320c then it has traveled @xmath1105 h@xmath7 kpc on the plane of the sky since the most recent passage through sq .ngc 7320c shows an approximate line of sight velocity 700 lower than ngc 7319 ( a ratio of apo- to pericenter of 5 - 6 assuming ngc 7318a is near the group center of mass ) . assuming a transverse velocity equal to the line of sight value suggests that the collision occurred as recently as 1.5@xmath810@xmath0h@xmath7 years ago .the fainter tail is more diffuse ( 2@xmath8 broader than the brighter one ) which is consistent with the idea that it represents an earlier passage at least t@xmath16 5@xmath810@xmath0 years ago .the roles of ngc 7317 and 7318a in the past dynamical activity or even their past morphologies are impossible to ascertain .an earlier encounter involving ngc 7320c and the southwestern - most tail however would provide a mechanism to account for much of the @xmath16500 hi and blue halo condensations that are seen west of ngc 7319 .the implied trajectory of that tail passes under ngc 7320 and is consistent with a past encounter involving ngc 7318a .we can not rule out the possibility that ngc 7318a is the remnant bulge of an early spiral whose disk was disrupted in the past ( see paper i ) .in such a view the high early - type galaxy fraction would be a product of dynamical evolution in sq .the present dynamical history of sq centers on ngc 7318b which shows an 850 radial velocity difference relative to the sq kernel ( these four galaxies are hatched in figure 1c ) .this is so large that one might argue that this galaxy is unrelated to the group ( an accordant redshift projection ) or that it has been dynamically ejected by the triplet kernel .two observational clues , however , favor the interpretation that ngc 7318b is currently colliding with sq and for the first time .ngc 7320c is not actively involved in the dynamical evolution of the group at this time .first , we have reanalyzed the lower velocity hi clouds detected in sq .shostak et al .( 1984 ) found four , spatially - distinct hi clouds in sq with velocities near 5700 , 6000 and 6600 ( two clouds ) .the latter clouds were discussed in the previous section .we interpret the two lower velocity clouds as a single feature .they are adjacent to one another and are centered on the nucleus of ngc 7318b ( indicated schematically in figure 1c or see figure 4 in shostak et al .1984 and paper i ) .these clouds were previously assumed to be stripped material like the more massive and extensive @xmath16500 hi clouds .our reanalysis suggests that they are still associated with the spiral disk of ngc 7318b .that galaxy can plausibly be assigned an sb or sbb type and therefore would be expected to have a considerable ( @xmath15@xmath810@xmath17m@xmath10 ) hi mass .they show a velocity gradient consistent with rotation and a central hole coincident with the nuclear bulge region of that galaxy ( see figure 6 in the previous reference ) .such holes in the 2d distribution of hi in spiral galaxies with a prominent population ii bulge are often seen .the velocity profile of these two clouds taken together show a slightly double - horn structure characteristic of a rotating inclined spiral galaxy. the total hi mass would be approximately 4.1@xmath810@xmath17 h@xmath7 m@xmath10 with a fwzi velocity gradient of 400 km / s consistent with values for spiral galaxies of similar type and luminosity .the association of this much neutral gas with ngc 7318b suggests that it still retains a significant part of its ism and , therefore has not previously passed through sq .the second evidence for a direct collision between ngc 7318b and sq involves vla radio and recent x - ray data of sq .rosat - hri mapping ( pietsch et al . 1997 ) reveals an elongated structure on the east side of ngc 7318b partly coincident with one of its spiral arms .this structure shows a rather close spatial correspondence with a radio continuum arc ( van der hulst and rots 1981 ) .this structure appears to be sharply bounded and is most easily interpreted as a shock front ( it is indicated in figure 1c and is also well seen in the h@xmath2 image shown in figure 2c ) . the origin of the shock would most plausibly be ascribed to a high velocity ( @xmath3v@xmath1810@xmath6 ) collision between the ism in ngc 7318b and the stripped gas in sq .the collision must be ongoing and in its early stages because ngc 7318b retains most of its hi , many hii regions and a reasonably symmetric spiral pattern .time limits on the duration of this event come from two sources : 1 ) the lifetime of synchrotron electrons in the radio arc must be a few @xmath8 10@xmath19 years without a local source of acceleration ( van der hulst and rots 1981 ) and 2 ) the observed line of sight velocity difference between sq and ngc 7318b give a similar timescale for the duration of the intruder passage through the group . in our view , ngc 7318b has replaced ngc 7320c as a high velocity intruder in sq . the shock is assumed to be associated with the ongoing stripping of ngc 7318b .optical data also support this view , and the support becomes even more striking when the image of ngc 7318a is subtracted ( see paper i ) .most of the complex structure around ngc 7318a and b is consistent with spiral structure associated with the latter high velocity intruder .hii regions can also be seen in the spiral arms as a further indication that this galaxy has not previously sustained a major stripping encounter .the radio / xray feature interpreted as a shock front indicates the interface between the unstripped gas disk of ngc 7318b and the ism in sq .the narrowness of the shock front suggests that the collision is most easily modeled with ngc 7318b entering the far side of sq oriented nearly edge - on .the twin problems posed by compact groups center around the large number of discordant redshift components that they contain , and their gravitational stability . in sq , at least , the evidence favors the chance projection explanation for discordant ngc 7320 .the gravitational stability issue is a problem because attempts at modeling the groups ( e.g. mamon 1986 ; barnes 1989 ) indicate that they should be unstable to collapse on short time scales . in this casethe number of compact groups observed today implies that a large merger post - cursor population should exist .little or no evidence is found for either the mergers in progress ( zepf et al 1991 ; moles et al 1994 ) or for the merger post - cursor population ( sulentic and rabaa 1994 ) . this has led some to propose that the groups are not real physical systems at all ( e.g. mamon 1986 ; hernquist et al .the large volume of data for sq can shed some light on these problems , particularly , if our interpretation of its dynamical history is correct * and * if it is representative of the compact group phenomenon .our analysis indicates that sq is not simply a projection of accordant redshift galaxies .it is a dynamically active system and evidence for this activity is virtual proof that it is a physical aggregate .however the compact group aspect ( defined as n@xmath16 4 accordant members ) is transient in the sense that no fourth tightly bound component can be identified at this time .instead sq contains a kernel of three galaxies that must have sufficient mass to attract near neighbors into high velocity encounters .our best estimate ( paper 1 ) for the combined stellar ( galaxy+halo ) , hi and x - ray gas mass is m@xmath20@xmath1 0.75 - 1.5@xmath810@xmath21 h@xmath7m@xmath22 .infall velocities @xmath1800km s@xmath7 were found in n - body simulations of groups with total mass @xmath12@xmath810@xmath23m@xmath22 ( governato et al . 1996 ) .the infalling galaxies result in dynamical activity that is distinctly episodic .any catalog of isolated groups will be biased against many of the groups with a currently infalling intruder because the intruder will often `` bridge '' the group into its environment and prevent it from satisfying an isolation criterion .ngc7318b is an example of a `` safe '' ( because it is internal ) intruder and ngc 7320c is one that almost prevented sq from satisfying the hcg sample selection criteria. is sq a stable virialized system ?this question was first directed towards sq by limber and mathews ( 1960 ) before the redshift of ngc 7320 was known .if one repeats this calculation ( with m / l@xmath24= 13h for ngc 7317 , 18a and 20c as well as m / l@xmath24= 8h for ngc 7318b and 19 ) it suggests that the triplet kernel is stable with 2t/-@[email protected] but that inclusion of ngc 7318b and 20c yields 2t/-@xmath25@xmath1 11 .ngc7320c forms a marginally stable quartet with the triplet kernel with 2t/-@xmath25@xmath1 2 .these estimates do not take into account the large amount of ( baryonic ) mass present in the halo or the possible role of non - baryonic matter that might be needed to produce the observed infall velocities .the virial calculation and optical images ( the twin tidal tails ) interpreted as evidence for recent passage(s ) through the group suggests that ngc 7320c has been recently captured by the kernel .finally , ngc 7318b is almost certainly entering the group for the first time .thus , sq is probably a bound triplet that has captured a fourth member ( ngc 7320c ) within the past gyr and is now being visited by another ( ngc 7318b ) .governato et al .( 1996 ) have advanced one of the most complete scenarios for the formation and evolution of compact groups .they create groups in a critical universe by 1 ) seeding them with primordial merger events and 2 ) growing present day groups with secondary infall onto these seeds .our study of sq strongly supports the second part of this model including the conclusion that the infall will be high velocity and will resist rapid merging .it also supports the qualitative discussion of moore et al .( 1996 ) that sees high velocity intruders as an effective means to dynamically evolve galaxies and create diffuse halos .they consider random high velocity encounters in clusters while compact groups are found in non - cluster environments ( sulentic 1987 ) .while high relative velocities in clusters are due to the high internal velocity dispersion , the lower mass of small groups such as sq implies that only infalling galaxies can achieve velocities v@xmath26 500 km s@xmath7 . in our casethe seed is not a single merger remnant as proposed by governato et al .( 1996 ) but a triplet .the over - representation of luminous elliptical galaxies in the triplet suggests that it may also have experienced strong dynamical evolution as it formed perhaps by a random initial encounter between three spirals ( ? ) .it is not clear if dynamical evolutionary effects have significantly altered the properties of ngc 7317 or ngc 7318a because their morphologies , colors and kinematics appear rather normal .thus they could be ellipticals or remnant spiral bulges .existing images do not allow a distinction between these possibilities .an extrapolation of the tail under ngc 7320 is consistent with an encounter trajectory where ngc 7320c would have passed very near ngc 7318a as recently as @xmath1 1gyr ago . there is sufficient stripped gas to account for three or four spiral galaxies .the presence of radio continuum and x - ray emission in the nuclear regions of ngc 7318a may indicate an active past history .a luminous halo surrounds the triplet which indicates that it is a dynamically evolved physical system .our conservative estimate of the halo luminosity ( v band : see paper 1 ) gives m@xmath27= -20.9 + 5log(h)@xmath16 m@xmath28 .this is almost ten times the luminosity of the tail created in the most recent passage of ngc7320c .neither tail was included in this estimate but each will increase the halo luminosity by @xmath1 10% .it is the high velocity intruders that cause sq to grow ( in galaxy population and halo mass ) and that prevents the triplet kernel from coalescing by injecting kinetic energy into the group . in this viewsq must be a relatively young group unless ngc7320c has been perturbing it for a longer time ; otherwise the triplet should have merged .this would require the triplet to form and dynamically evolve in the past 1 - 2 gyr .assuming that the halo was created by similar processes ( a tidal tail at a time ) and at about the same rate ( one m@xmath29= -18.4 + 5log(h ) tail per 5@xmath810@xmath0 years ) , would imply an age of several gyr for its formation .sq may not be primordial but the stable - kernel part of it is at least 2 gyr old .neither of the elliptical components in the kernel show luminosities or other properties consistent with having been recent mergers . in this respect sqis typical of other compact groups where no evidence for ongoing merging is seen ( zepf and whitmore 1991 ; moles et al .if our view is correct , then rather than mergers , these ellipticals may be the dynamically evolved remnants of spiral galaxies .sq is also similar to other compact groups in the sense that its optical and fir emission properties indicate a lower level of current star formation activity than is observed in pairs ( sulentic and de mello rabaa 1994 ; moles et al .sq suggests that the lack of starburst activity in compact groups is due to the lack of bound gaseous disks in many component galaxies . in sqa major stripping event happened in the recent past and another is in progress .the gas is either stripped and neutral or shocked and hot .a large intergroup star forming region is observed in h@xmath2 emission within the group but outside of the galaxies ( see figure 2c ) .however so much of the gas in sq is stripped or shocked that it is too diffuse(cool or hot ) to form large numbers of stars .the fir emission from sq is not strong especially when allowances are made for possible contributions from ngc7320 and the seyfert nucleus in ngc7319 .the excess fir emission expected from these sources apparently can not compensate for the deficit emission from sq component galaxies .the transition from a normal star forming disk to the shocked state should be quite sudden given the high velocity of this intruder . given the efficiency and quasi - periodic nature of the tidal perturbation in compact group , it is tempting to ascribe the seyfert activity in ngc7319 to the past intruders as well. whatever gas was not stripped may have been rapidly channeled into the nucleus to fuel active galactic nucleus ( agn ) activity .sq suggests that compact groups consist of a tightly bound subsystem ( kernel or seed ) plus a loosely bound , or even unbound , population of infalling neighbors .the most common situation for hcg groups would involve a triplet or quartet acting as a seed plus 1 or two intruders .indeed , it is easier to form a bound pair than a triplet , but given the n@xmath164 number criterion used in hcg , triplets would be more often selected because it is easier to have a 3 + 1 rather than 2 + 2 configuration .the kind of dynamical encounters do not lead to rapid merging suggested by many models .the intruders are stripped which reduces their mass and cross section to frictional effects . at the same timethey inject energy into the kernel which sustains it against collapse .some stars form but not as many as commonly observed in more dynamically stable pairs .the more frequent and high impulse events in compact groups may also foster agn activity at a higher rate than pairs do .it is perhaps appropriate that the first compact group discovered more than a century ago provides the clearest clues to their origin and evolution .

<S> multiwavelength data for stephan s quintet ( sq ) are consistent with the following model for this compact galaxy group . </S><S> ( 1 ) discordant redshift ngc 7320 is an unrelated foreground galaxy . </S><S> ( 2 ) in the past sq was an accordant redshift quartet involving ngc 7317 , 18a , 19 and 20c . </S><S> ngc 7320c collided ( probably not for the first time ) with the group a few times 10@xmath0 years ago and stripped the interstellar matter from ngc 7319 . </S><S> ( 3 ) in the present sq is again an accordant quartet involving ngc 7317 , 18a , b , and 19 . </S><S> ngc 7318b is now entering the group at high velocity for the first time , giving rise to a shock zone . if most compact groups are like sq , then they are frequently visited by infalling neighbors that perturb the group and themselves . </S><S> sq represents strong evidence for secondary infall in a small group environment . </S><S> tidal stripping reduces the mass of the infalling galaxies , thereby increasing the timescale for their orbital decay . </S><S> there is little evidence that these high velocity `` intruders '' are rapidly captured and/or merge with the system . </S><S> instead they are the mechanism that sustains compact groups against collapse . </S><S> efficient gas stripping may account for the low star formation rate observed in compact groups and infall of residual gas into galactic nuclei may also foster the onset of active galactic nucleus activity . </S>

1201.4742

recent observational data indicate that our universe is currently in accelerated phase @xcite .since the discovery of the acceleration , a large number of possible mechanisms have been proposed to explain the origin of the dark energy .general relativity in its standard form can not explain the accelerated expansion without extra terms or components , which have been gathered under the name of dark energy .there are various proposals to construct an acceptable dark energy model , such as : a scalar field with quintessence - like or phantom - like behavior @xcite , spinor , ( non-)abelian vector theory , cosmological constant , fluid with a complicated equation of state , and higher dimensions @xcite . as a result, we have a number of competing scenarios which to describe late - time acceleration . working in framework of one of the above proposals ,we are forced to introduce the following extra cosmological components : inflaton , a dark component and dark matter .we concentrate on the modified gravity @xcite , which represents a classical generalization of general relativity ( modifications of the hilbert - einstein action by introducing different functions of the ricci scalar @xcite or gauss - bonnet invariant @xcite ) , should consistently describe the early - time inflation and late - time acceleration , without the introduction of any other dark component . in the present workwe will focus on the gauss - bonnet gravity , where the gravitational action includes functions of the gauss - bonnet invariant @xmath0 it is interesting , that the theory of this type is closely related to higher order string curvature corrections @xcite .the starting action is @xmath1\ .\ ] ] here @xmath2 is the gauss - bonnet invariant ( [ i2 ] ) and the scalar field @xmath3 is canonical in ( [ gbany1 ] ) .one may redefine the scalar field @xmath3 by @xmath4 .the action takes the following form @xmath5\ .\ ] ] here @xmath6 if a proper limit @xmath7 exists , the action ( [ gbany41 ] ) reduces to @xmath8\ .\ ] ] then @xmath9 is an auxiliary field . by variation of @xmath9, we find @xmath10 which may be solved with respect to @xmath9 as @xmath11 substituting ( [ gbany46 ] ) into the action ( [ gbany43 ] ) , we get for the @xmath12-gravity @xmath13\ , \quadf(g)\equiv \tilde v\left(\phi(g)\right ) - \tilde\xi_1\left(\phi(g)\right ) g\ .\ ] ] on the other hand , the action [ 1.1 ] is associated with non - local models @xcite let s start with the next action @xmath14 here @xmath15 is the matter lagrangian . by introducing the scalar field @xmath3 ,one may rewrite the action ( [ nlgb1 ] ) in a local form : @xmath16 in fact , from the @xmath3-equation we find @xmath17 . by substituting this expression into ( [ nlgb3 ] ), we reobtain ( [ nlgb1 ] ) . thus , by studying @xmath12 - gravity, we are also exploring several other models .note that in some cases , @xmath12 or even @xmath18 ) gravity maybe transformed to @xmath19 gravity @xcite .the five - year wilkinson microwave anisotropy probe ( wmap ) data @xcite give the bounds to the value of the equation of state ( eos ) parameter @xmath20 , which is the ratio of the pressure of the dark energy to the energy density of it , in the range of -1.11 < @xmath20 < -0.86 .this could be consistent if the dark energy is a cosmological constant with @xmath20 = -1 and therefore our universe seems to approach asymptotically de sitter universe . although the accelerating expansions seem to be de sitter type , the possibility that the current acceleration could be quintessence type , in which @xmath20 > -1 , or phantom type , in which @xmath20 < -1 , is not completely excluded .it is often assumed that the early universe started from the singular point often called big bang . however , if the current ( or future ) universe enters the quintessence / phantom stage , it may evolve to the finite - time future singularity depending on the specific model under consideration and the value of the effective eos parameter ( big rip ) @xcite .an elegant solution to this problem has been recently proposed @xcite under the form of the so - called little rip cosmology which appears to be a realistic alternative to the @xmath21cdm model .consider the following action ( see ref .@xcite ) : @xmath23 where @xmath24 being the newton constant , and the gauss - bonnet invariant is defined by([i2 ] ) . by varying the action ( [ gb11 ] ) over @xmath25 , are obtained : @xmath26 where @xmath27 and @xmath28 ( @xmath29 ) .we consider a spatially - flat universe @xmath30 here @xmath31 is the scale factor at cosmological time @xmath32 .it is easy to obtain the frw equations @xmath33@xmath34 here @xmath35 is the hubble rate ( @xmath36 , @xmath37 ) , @xmath38 is matter energy density : @xmath39 while @xmath2 , its derivative and @xmath40 can be defined as functions of the hubble parameter as @xmath41 it is easy to see that the second equation in ( [ gb13 ] ) is a differential consequence on the solution .the first equation ( [ gb13 ] ) can thus be expressed as follows @xmath42 by selecting certain of the hubble parameter @xmath35 we obtain a differential equation for the function @xmath43 .this is the method of reconstruction ( this method has been implemented in ref .@xcite for @xmath44 gravity and ref .@xcite for @xmath45 gravity ) .let us now consider a more general model for a kind of modified gauss - bonnet gravity .this can be described by the following action @xmath47 .\label{f(r , g)action}\ ] ] varying over @xmath25 the gravity field equations are obtained @xcite , @xmath48 in the case of a flat frw universe , described by the metric ( [ frw ] ) , the first frw equation yields @xmath49 for simplicity , we consider the following subfamily of functions @xmath50 correspondingly , the friedmann equation ( [ friedmeq ] ) can be split into two equations , as @xmath51in ref . @xcite is proposed little rip model on the basis of a viscous fluid . using the obtained in this study form the scale factor , we construct a model of the gb gravity .consider the scale factor as @xcite @xmath52 in this case the hubble parameter has the form @xmath53 then the derivative of the hubble parameter is proportional to itself , and @xmath54 frw equation for the @xmath55 theory gravity takes the form @xmath56 choose the energy density in the form @xmath57 we obtain the following solution @xmath58\right)}{2\beta^3 ( 1+\beta ) } \nonumber\end{aligned}\ ] ] here @xmath59 $ ] gives the exponential integral function @xmath60 $ ] , where @xmath61 .this solution can be illustrated by the graph ( we assume that @xmath62 ) : figures show that at the initial time all the solutions have the same behavior ( dominated by the first term in ( [ sol1 ] ) ) , then begins to play the role of the second term ( @xmath63 ) , and finally turned the third term ( @xmath64 ) , this is clearly seen in the figure 1 ., ( g,0,200 ) ( red line - @xmath65 , blue line - @xmath66).,scaledwidth=100.0% ] , ( g,0,200 ) ( red line - @xmath65 , blue line - @xmath66).,scaledwidth=100.0% ] we can consider the dependence of the @xmath12 . obviously not possible to do this , but for large @xmath35 , we can assume that @xmath67 .from ( [ sol1 ] ) one can see that there are 3 terms : the first dominates at small @xmath2 , then by @xmath68 included the last term ( @xmath64 and @xmath69 , figure 3 ) , then included the second term ( @xmath63 and @xmath70 , figure 4a ) and then @xmath71 ( figure 4b ) again dominated by the last term .this behavior is due to the fact that the last term consists of two parts with different signs . for small values of @xmath2 terms rather different , and with increasing @xmath2 , they become the same order and the difference becomes smaller than other terms .when @xmath2 tends to infinity , the difference between them begins to grow faster than the other terms . by selecting different values of the constants @xmath72 and@xmath73 can be different depending @xmath43 from @xmath2 .for example , we construct the graph for the following @xmath45 values of the constants of integration : @xmath74 , @xmath75 ( figure 5 ) . , ( g,0,70 ) ( @xmath74 , @xmath75).,scaledwidth=50.0% ] consider now the @xmath76 - gravity .let us solve the equations ( [ 3.1 ] ) @xmath77\right)\right)\nonumber\end{aligned}\ ] ] @xmath78+h^3 ( \beta+h ) \left(2c[1]+\alpha^2 c[2 ] \text{expintegralei}\left[\frac{h}{\beta}\right]\right)\right)\nonumber\end{aligned}\ ] ] using the @xmath79 we obtain the dependence of the function @xmath80 from @xmath40 @xmath81\right)\right)\nonumber\end{aligned}\ ] ] we can consider the dependence of the @xmath82 . obviously not possible to do this , but for large @xmath35 , we can assume that @xmath67 .then @xmath83\right)\end{aligned}\ ] ]consider a power - law solutions . choose the hubble parameter as @xmath84then the scale factor is given by @xmath85 this scale factor can meet the various phases of expansion , which are characterized by different values of the equation of state ( eos ) parameter @xmath20 . to easily find the value @xmath20 for ( [ big_rip1 ] ) .the equations frw have the form : @xmath86 comparing with the ( [ gb13 ] ) , we can write formally the same equations by removing all unnecessary to the effective pressure and energy density ; : @xmath87 so @xmath88 for ( [ big_rip ] ) have @xmath89 .consider the different values of @xmath20 .if @xmath20 = -1 this could be consistent if the dark energy is a cosmological constant and therefore our universe seems to approach to asymptotically de sitter universe .this value is not possible for the chosen scale factor .this value is the min in the permissible region , but the possibility that the current acceleration could be quintessence type , in which @xmath20 > -1 , or phantom type , in which @xmath20 < -1 , is not completely excluded . for quintessence type ,the value of @xmath90 is negative . for phantom type ,the value of @xmath90 is positive .it is easy to write of the equation ( [ fff1 ] ) for this cases as an equation for @xmath91 @xmath92 where @xmath93 .@xmath94 we now express @xmath32 through @xmath2 @xmath95 and we obtain the following form of the function @xmath45 @xmath96 it is seen that if @xmath97 the equation is considerably simplified @xmath98 now consider the space filled with dust @xmath99 .@xmath100 it is seen that if @xmath97 the equation is considerably simplified @xmath101 consider now the @xmath102 - gravity ( [ 3.1 ] ) .equations take the form @xmath103 @xmath104the solution of these equations have the form @xmath105 @xmath106 or finally get @xmath107 @xmath108de sitter solutions are described by an exponential expansion of the universe , where the hubble parameter and the scale factor are given by @xmath109 where @xmath110 is a constant .this kind of solutions are very important , as the observations suggest that the expansion of our universe behaves approximately as de sitter .it has been shown in ref .@xcite that de sitter points are critical points in @xmath44 gravity .it is straightforward to see that this is also the case in @xmath55 gravity .@xmath111 here , @xmath112 and we have ignored the contribution of matter . then , we have reduced the differential equation to an algebraic equation that can be resolved by specifying a function @xmath45 .consider the equation ( [ d9 ] ) as a differential equation for @xmath2 : @xmath113 in this case , the function @xmath12 takes the form @xmath114 we see that the function of the form @xmath115 describes the power solutions and de sitter solution ( @xmath116 ) .the general solution has the form @xmath117 here @xmath72 and @xmath73 -constants of integration .the case when @xmath118 corresponds to the de sitter solution . for @xmath119 gravitysimilarly , we have the following equations @xmath120 if we consider the case @xmath121 is not difficult to find a possible solution @xmath122 .we can find a little rip solution for gauss - bonnet gravity with dilaton for spaces of dimension @xmath123 . consider space of @xmath124 dimensions with two maximally symmetric subspaces : @xmath125-dimensional and @xmath126-dimensional .we write the lagrangian as @xmath127 here @xmath128 , @xmath129 are functions of dilaton @xmath9 . we choose them as @xmath130 variating the action with lagrangian ( [ egbd - lagr ] ) we get field equations @xcite .one of the possible solutions of these equations will be @xmath131 @xmath132 where @xmath133 and @xmath134 are arbitrary positive constants .@xmath135 are constants satisfying the following equations @xmath136@xmath137 @xmath138 @xmath139 therefore it is necessary to find solutions of this equations satisfied conditions @xmath140 this is easily done , for example @xmath141in this paper we construct a series of cosmological models which describe the accelerated expansion of the universe .the models in framework of the modified gauss - bonnet gravity are constructed by the reconstruction method .constructed models are of the effective phantom type and do not lead to a singularity in the future ( little rip models ) .an explicit form of the action leading to such a model within the framework of gravity @xmath55 and @xmath142 is obtained .the solution contains special function , which can be removed by choosing the constant of integration .the resulting action consists of three terms , each of which plays a role at the definite stage of the evolution of the universe . since the last term represents the difference between the functions of the same order , and this order of these functions is maximal , it has double the lead - with the average values and tends to infinity .in addition we construct a model describing the little rip in the multivariate gauss - bonnet theory with dilaton , when visible , and additional spaces are maximally symmetric .little rip models provide an evolution for the universe intermediate between asymptotic de sitter expansion and models with a big rip singularity .therefore , the method of reconstruction models were built describing the de sitter space and space with the metric [ big_rip1 ] , which refers to the phantom type and has a singularity of the future .for all constructed models @xmath143 ( except destter model ) and therefore they belong to the phantom type . for models of the effective phantom energy conditions are always broken .such behavior is typical for the little rip models @xcite we can continue to explore these theories and consider stability of solutions. however , as shown by work @xcite little rip solutions are stable and can be expected and received from our paper solutions will have the same property .riess , et al .j. * 116 * , 1009 ( 1998 ) ; astron .j. * 117 * , 707 ( 1999 ) ; s.perlmutter et al , nature * 391 * , 51 ( 1998 ) ; m. hicken et al ., astrophys .j. * 700 * , 1097 ( 2009 ) ; [ arxiv:0901.4804 [ astro-ph.co ] ] ; e. komatsu et al .[ wmap collaboration ] , astrophys .* 180 * , 330 ( 2009 ) ; arxiv:0803.0547 [ astro - ph ] ; w. j. percival et al . ,mon . not .. soc . * 401 * , 2148 ( 2010 ) ; arxiv:0907.1660 [ astro-ph.co ] .elizalde e. , nojiri s. , odintsov s.d ., phys.rev.d * 70 * 043539 ( 2004 ) [ arxiv : hep - th/0405034 ] ; elizalde e. , nojiri s. , odintsov s.d ., sez - gmez d. , faraoni v. , phys .d * 77 * 106005 ( 2008 ) [ arxiv:0803.1311 ] .e. elizalde , a.n .makarenko , v.v .obukhov , k.e .osetrin , a.e .filippov , phys.lett.b * 644 * , 1 - 6 ( 2007)[arxiv : hep - th/0611213 ] ; ilya v. kirnos , andrew n. makarenko , open astron.j . * 3 * 37 - 48 ( 2010 ) [ arxiv:0903.0083 ] . nojiri s. , odintsov s.d . , econf * c0602061 * , 06 ( 2006 ) [ int .phys . * 4 * , 115 ( 2007 ) ] [ arxiv : hep - th/0601213 ] ; nojiri s. , odintsov s.d ., phys.rept . * 505 * 59 - 144 ( 2011 ) [ arxiv:1011.0544 ] .s. capozziello , m. de laurentis and v. faraoni [ arxiv:0909.4672 ] ; s. capozziello and m. francaviglia , gen .* 40 * , 357 ( 2008 ) [ arxiv:0706.1146 ] ; s. capozziello , v. faraoni , `` beyond einstein gravity : a survey of gravitational theories for cosmology and astrophysics '' , springer , 2010 ; f. s. n. lobo [ arxiv:0807.1640 ] .m. sami , a. toporensky , p. v. tretjakov and s. tsujikawa , phys .b * 619 * , 193 ( 2005 ) [ arxiv : hep - th/0504154 ] ; h. m. sadjadi , arxiv:1009.2941 [ gr - qc ] ; m. alimohammadi and a. ghalee , phys .d * 79 * , 063006 ( 2009 ) [ arxiv:0811.1286 ] ; a. de felice and s. tsujikawa , phys .d * 80 * , 063516 ( 2009 ) [ arxiv:0907.1830 ] . nojiri s. , odintsov s.d . , physb * 631 * 1 ( 2005 ) [ arxiv : hep - th/0508049 ] ; nojiri s. , odintsov s.d ., sasaki m. , gauss - bonnet dark energy .d * 71 * 123509 ( 2005 ) [ arxiv : hep - th/0504052 ] ; ratbay myrzakulov , diego saez - gomez , anca tureanu , gen.rel.grav.*43 * , 1671 - 1684 ( 2011 ) ; e. elizalde , r. myrzakulov , v. v. obukhov and d. saez - gomez , class .* 27 * , 095007 ( 2010 ) [ arxiv:1001.3636 ] .gianluca allemandi , andrzej borowiec , mauro francaviglia , sergei d. odintsov , phys.rev .d * 72 * , 063505 ( 2005 ) [ arxiv : gr - qc/0504057 ] ; gianluca allemandi , andrzej borowiec , mauro francaviglia , phys.rev .d * 70 * , 043524 ( 2004 ) [ arxiv : hep - th/0403264 ] ; gianluca allemandi , andrzej borowiec , mauro francaviglia , phys.rev .d * 70 * , 103503 ( 2004 ) [ arxiv : hep - th/0407090 ] .

<S> we discuss the cosmological reconstruction in modified gauss - bonnet ( gb ) gravity . it is demonstrated that the modified gb gravity may describe the most interesting features of late - time cosmology . </S><S> we derive explicit form of effective phantom cosmological models ending by the finite - time future singularity ( big rip ) and without singularities in the future ( little rip ) . </S>

nucl-ex0207002

a central topic of hadron physics is to study the influence of the nuclear medium on elementary processes , for example by measuring meson production at projectile energies below the threshold for free @xmath5 collisions ( so - called subthreshold production ) .these processes necessarily involve collective effects of the nucleons inside the target nucleus .@xmath0-production is particularly well suited for such investigations since the meson is rather heavy , as compared to pions , so that its production requires strong medium effects .final state interactions of @xmath0 mesons in nuclei are generally considered to be rather small , due to their strangeness of @xmath10 . as a consequence ,the production of @xmath0-mesons in proton - nucleus collisions is of great importance to learn about either cooperative nuclear phenomena or high momentum components in the nuclear many - body wave function .the cooler synchrotron cosy - jlich @xcite , which provides proton beams in the range @xmath11 gev , is well suited for the study of @xmath0-production . in measurements with very thin windowless internal targets ,secondary processes of the produced mesons can be neglected and , simultaneously , sufficiently high luminosities of more than @xmath12 can be obtained .subthreshold @xmath0-production was a prime motivation for building the large - acceptance anke spectrometer @xcite within one straight section of the cosy ring .it consists of three dipole magnets , which separate forward - emitted reaction products from the circulating proton beam and allow to determine their emission angles and momenta . depending on the choice of the magnetic field in the spectrometer dipole , @xmath0-mesons in the momentum range @xmath13 mev / c can be detected .the layout of the device , including detectors and the data - acquisition system , was optimized @xcite to study @xmath0-spectra down to @xmath14 gev .this is a very demanding task because of the small @xmath0-production cross sections , e.g. 39 nb for @xmath15c collisions at 1.0 gev @xcite .thus , the kaons have to be identified @xcite in a background of secondary protons and pions which is up to @xmath16 times more intense .the measured cross sections for @xmath0 production in @xmath15c interactions are shown in fig .[ fig : cross_sections ] .the cross sections were normalized to known pion - production cross sections @xcite measured for similar kinematical conditions @xcite .for @xmath17 and 2.3 gev two data sets have been obtained for different settings of the field strength in the spectrometer dipole , @xmath18 t and 1.6 t. these two modes of operation allow to explore slightly different but overlapping regions of the kaon - momentum spectra . at @xmath19 gev ( @xmath18 t ) , anke for the first time allows to measure the full momentum spectrum of the produced kaons at deep - subthreshold energies @xcite .the geometry of the experimental setup as well as the analysis procedures are identical for both field values .[ fig : cross_sections ] shows that there are discrepancies between the two data sets for @xmath20 mev / c at @xmath21 gev , whereas at @xmath17 gev they coincide .we conclude that there is an as yet unknown source of systematic errors , which may yield an additional uncertainty of up to 30% at the higher beam energies .it should be noted that these uncertainties vanish when one calculates the cross - section _ ratios _ for different target nuclei @xcite .thus , these ratios are used for the discussion of reaction mechanisms and final - state interaction effects in sects .[ sec : mechanisms ] and [ sec : fsi ] .table [ tab : syst ] shows the available data on @xmath0-production in @xmath1 collisions .the different data sets were obtained for non - overlapping kinematical parameters ( i.e. beam energies , kaon emission angles and momenta ) which prevents a direct comparison with the data from anke and a test of the overall normalizations .c|c|c|c|c @xmath22 ( gev ) & targets & @xmath23 ( mev / c ) & @xmath24 ( @xmath25 ) & measured at ' '' '' + ' '' '' 0.842 0.99 & be pb & & pnpi @xcite + 2.1 & naf , pb&350750 & 1580 & lbl @xcite + 1.2 , 1.5 , 2.5 & c , pb&500700 & 40 & saturne @xcite + 1.2 & c & 165255 & 90 & celsius @xcite + 1.7 2.91 & be & 1280 & 10.5 & itep @xcite + 2.9 & be&545 & 17 & itep @xcite + 1.0 & c , cu , au&171507 & @xmath26 & * anke * @xcite + 2.5 , 3.5 & c , au & 3001050 & 40 & kaos , @xcite + 1.2 , 1.5 , 2.0 , 2.3 & c , cu , ag , au&171595 & @xmath26 & * anke * , _ prelim . _ in a recent publication @xcite we showed that the measured invariant cross sections @xmath27 follow an exponential scaling behavior when plotted as a function of the four - momentum transfer @xmath28 , which is illustrated in fig.[fig : systematics ] ( l.h.s . ) . apart from the data taken with anke at @xmath19 and 1.2 gev all spectra cover the range of negative @xmath28 .the sharp fall - off of the cross sections from anke towards positive values of @xmath28 was explained @xcite by the fact that the data were taken very close to the kinematical limit for hyper - nucleus formation at @xmath29 ( @xmath19 gev ) which is expected to be accompanied by very small cross sections .for @xmath30 the different data follow an exponential dependence of : @xmath31}\ , \label{eq : tdep1}\ ] ] with parameters @xmath32 and @xmath33 given in ref.@xcite .it has been speculated @xcite that deviations from the exponential behavior ( see , e.g. , the data from kaos @xcite in fig . [fig : systematics ] ) reflect a dependence on the available squared cm energy @xmath34 and the excitation energy of the target nucleus , @xmath35 ( @xmath36 and @xmath37 denote the mass of the target nucleus before and after the reaction process , respectively ) . based on regge phenomenologythe following formula has been suggested as a parameterization of the invariant cross section : @xmath38}\ .\label{regge}\ ] ] the r.h.s . of figure[ fig : systematics ] shows the invariant cross sections , divided by an average @xmath39 dependence , using @xmath40 ( indicated by the dashed line on the l.h.s . of fig.[fig: systematics ] ) and @xmath41 , as a function of the excitation energy @xmath42 .the solid lines in fig.[fig : systematics ] correspond to a parameterization of the invariant cross sections in @xmath43c interactions @xmath44 } = \sigma_0\cdot \delta m^{n_0 } \cdot ( s / s_0)^{b_0 t } \ , \ ] ] with the following fitted parameter sets : c||c|c|c|c beam energy & @xmath45 & @xmath46 @xmath47 & @xmath33 & @xmath48 + ' '' '' ( gev ) & ( @xmath49 ) & & ( gev@xmath50 ) & ( gev@xmath51 ) + ' '' '' @xmath52 & 25 & 17 ( 13 ) & & + & & & 2.2 & 100 + @xmath53 & 1000&9.5 ( 8) & & + the term @xmath54 reflects a phase space behavior @xmath55 with @xmath56 particles in the final state . as can be seen from fig.[fig : systematics ] , the parameterization from eq .( [ eq : parametrization ] ) can describe all available data on @xmath0-production in @xmath1 reactions obtained in largely different angular - momentum intervals within a factor @xmath57 .this parameterization is based on lorentz invariant variables and , thus , is independent of the choice of the reference system , @xmath58 or @xmath1 .the parameterization given by eq .( [ eq : parametrization ] ) not only supplies a useful description of all available data , but fig . [fig : systematics ] ( r.h.s . ) also reveals that 1 .the data from anke at @xmath19 gev were obtained down to @xmath59 gev / c@xmath51 , i.e. very close to the kinematical limit at @xmath60 gev / c@xmath51 . at this limitno energy can be transfered which excites the target nucleus or knocks out target nucleons .the target nucleus must take part in the reaction as a whole such that the effective target mass is @xmath61 .there are only two particles , @xmath0-meson and nucleus , in the final state ( @xmath62 ) .2 . within a phase - space treatment of the final state ,the steep rise ( @xmath63 , @xmath64 ) of the subthreshold data indicates that all 12 nucleons plus the @xmath65-hyperon carry away energy in the final state .this seems to be in contradiction with the above statement of @xmath62 .however , this might reflect that more and more nucleons in the _ initial state _ must take part in the kaon - production process when approaching the kinematical limit ( see discussion of the 1.0 gev data below ) .the different slope parameters @xmath46 below and above the free @xmath5 threshold ( @xmath6 gev ) indicate a change of the dominant @xmath0-production mechanism . this has been predicted by various model calculations @xcite andis confirmed by recent coincidence measurements with anke ( see sect . [sec : mechanisms ] for details ) .the anke data for @xmath0 production at @xmath19 gev are plotted as invariant cross sections in fig .[ fig:1.0gev ] , see also ref .as discussed above the high - momentum part of the kaon spectrum ( smallest values of @xmath42 ) is most sensitive to collective nuclear effects .for example , if the kaons are produced in a collision of the beam proton with a single nucleon , internal momenta of at least @xmath66 mev / c are needed in order to produce kaons in the forward direction with momenta of @xmath67 mev / c .such high momentum components above @xmath68 mev / c are essentially due to many - body correlations in the nucleus . to get a _rough estimate _ of the number of participating nucleons , one can describe the invariant cross section within a phase - space approximation .the invariant cross section for the @xmath69 reaction is then @xmath70m_n){+}2p_kp_pcos\theta_k\ , \end{aligned}\ ] ] where @xmath34 denotes the square of the cm energy of the incident proton and the @xmath71 target nucleons .the solid lines in fig .[ fig:1.0gev ] show the invariant cross section calculated with eq .( [ eq : kinem ] ) for @xmath72 .although this approach completely neglects the intrinsic motion of the @xmath71 target nucleons , it shows that kaon production at @xmath19 gev can only be understood in terms of cooperative effects , where the effective number of nucleons involved in the interaction is @xmath73 .it has been suggested @xcite that such effects can be described in terms of multi - step mechanisms ( see sect .[ sec : mechanisms ] ) or high - momentum components in the nuclear wave function .the dependence of meson - production cross sections on the target mass @xmath74 has frequently been used ( see ref .@xcite and references therein ) to obtain information about production mechanisms . at large beam energies ,@xmath0-mesons are supposed to be dominantly produced in a collision of the beam proton with a single target nucleon ( @xmath75 ) . in this caseone expects that the cross section scales as @xmath76 .below threshold multi - step production , in particular two - step reactions with the creation of intermediate pions ( @xmath77 , @xmath78 ) , should dominate , and the cross sections are then proportional to @xmath79 @xcite . figure [ fig : ratios ] shows the cross section ratios for au and c nuclei at @xmath19 gev and 2.3 gev .surprisingly , at both beam energies the momentum averaged cross - section ratios seem to be in line with @xmath80 .there is , at least , no increase of the average ratio towards the lower beam energy .the mass dependence of the 1.0 gev data from anke , selecting forward angles @xmath81 , is in strong contrast to the observed @xmath79 dependence ( @xmath82(au / c@xmath83 ) of the total cross sections at the same beam energy @xcite .the authors of ref .@xcite interpreted the latter as an indication for multi - step processes .a qualitative explanation @xcite of this discrepancy can be obtained if one assumes that in the subthreshold domain the large angle @xmath0-production is due to reactions of the projectile protons with more than a single nucleon .the @xmath74 dependence of the anke data at small angles indicates direct production of @xmath0-mesons through the interaction of incident protons with surface nucleons .we have to conclude , however , that the @xmath74 dependence of the differential spectra does not reveal the expected change from direct to multi - step kaon production with decreasing beam energy . according to model calculations ,the momentum distributions of protons and deuterons , measured in coincidence with the @xmath0-mesons , are sensitive to the different production mechanisms @xcite .a ( @xmath9)-pair in the final state can , for example , be observed in case of a particular two - step reaction with deuteron formation in the first step ( @xmath84 , @xmath85 ) @xcite , while @xmath8 pairs may come both from the two - step and direct production . on the other hand ,the influence of deuteron break - up in the nucleus might be strong , so that one does not observe any coincident deuterons .first test measurements of @xmath8 and @xmath9 coincidences have been performed at @xmath86 and 2.3 gev .the amount of detected deuterons is rather large ( comparable to the @xmath8 yield ) at @xmath17 gev and decreases towards larger energies .this is in line with the interpretation that the contribution of two - step mechanisms is large at 1.2 gev and almost negligible at the highest energy .the momentum spectrum of the correlated deuterons at @xmath17 gev is shown in fig .[ fig : k+d ] . a peak structure in the deuteron - momentum distribution is observed at @xmath87 mev / c , which is close to the expected value for on - shell two - body kinematics of the reaction @xmath88 with backward emission of the deuteron in the ( @xmath89 ) cms . in this collinear geometry ,pions with maximum momenta are produced which is energetically favorable for @xmath0-production in the second step .a similar peak structure has been predicted @xcite for the deuteron spectrum at @xmath19 gev and , thus , this spectrum can be considered as first direct experimental evidence of the two - step production below threshold .however , detailed model calculations still have to show whether the observed peak structure can not be faked by competing processes like coalescence @xcite or pick - up reactions .another source of @xmath9 pairs might be the kaon production on two- ( or multi- ) nucleon clusters @xcite which is expected to yield deuterons with similar momentum distributions as from the two - step mechanism .the ratios of kaon - production cross sections for cu / c , ag / c and au / c measured at 2.3 gev are presented in fig .[ fig : fsi ] . all data exhibit similar shapes , rising steadily with decreasing kaon momenta , passing a maximum and falling steeply at low momenta .the position of the maximum varies with the nucleus , a fit to the data gives @xmath90 , @xmath91 , and @xmath92 mev / c for au , ag , and cu , respectively .the same effect at low @xmath23 has been observed for beam energies down to 1.5 gev , suggesting that the phenomenon is independent of the @xmath0-production mechanism and , thus , due to final - state interaction of the @xmath0 with the residual nucleus .the peak structure seems not to be visible in the ratio au / c at @xmath19 gev shown on the l.h.s . of fig .[ fig : ratios ] . however , at this low beam energy the large background permitted us to extract the cross sections only down to @xmath93 mev / c , which is slightly below the peak positions observerd at @xmath94 gev . taking into account the large error bar of the lowest data point ,the apparent constant ratio at @xmath19 gev is not in contradiction with the strong decrease abserved at higher energies .[ fig : fsi ] shows that the position of the maximum in @xmath95/c ) increases with @xmath74 .the situation has a parallel in the well - known suppression of @xmath96 emission in heavy nuclei at low positron momenta , arising from the repulsive coulomb potential @xmath97 .positively charged particles , like @xmath0-mesons , produced at rest at some radius @xmath82 in the target nucleus , acquire a momentum @xmath98 , if other interactions are negligible .moreover , it is known from @xmath0 elastic scattering experiments at higher energies @xcite that the nuclear @xmath99 potential is mildly repulsive , in agreement with one - body optical potentials based upon low - energy @xmath100 scattering parameters @xcite . at normal nuclear density , @xmath101@xmath102 ,the predicted repulsive @xmath99 potential of strength @xmath103mev @xcite would shift the kaon momenta to higher values .results of calculations in the framework of a coupled channel transport model @xcite for @xmath82(au / c ) are shown in fig .[ fig : fsi ] ( r.h.s . ) . without including the coulomb and nuclear kaon potentials ( dashed line ) the ratio exhibits a smooth momentum dependence with a steady increase towards low momenta as a result of the stronger @xmath0 rescattering processes for the au target .coulomb interaction leads to a distortion of the momentum spectrum and provides a maximum at @xmath104 mev / c ( dotted line ) . an additional change is observed in the calculations when a repulsive kaon nuclear potential is also considered . using a kaon potential of @xmath105 mev in the calculations , a reasonable agreement with the datais achieved , with a maximum close to the experimental value of @xmath106 mev / c ( solid line ) .we expect that after refined model calculations the nuclear @xmath0-potential at normal nuclear density @xmath107 can be determined with an accuracy of better than @xmath7 mev .this is significantly better than the current knowledge from heavy - ion collisions ( @xmath108 ) where the uncertainties are of the order of 15 mev @xcite .the anke experimental program on inclusive production of @xmath0-mesons in @xmath1 reactions has been finalized .first test measurements of @xmath8 and @xmath9 coincidences have been performed in 2001 and show the predicted @xcite sensitivity to different reaction mechanisms leading to kaon production below and above threshold , see discussion in sect .[ sec : mechanisms ] .it is planned to continue these studies during the year 2003 .similar measurements have already been made with a hydrogen target at @xmath109 and 2.83 gev .first analyses of these data showed that anke allows to study the production of heavy hyperons , up to the @xmath65(1520 ) , in @xmath110 reactions @xcite . in spring 2002 a detection system for negatively charged particles , in particular @xmath111-mesons , has been commissioned at anke .it will be used to study the production of @xmath112 pairs in @xmath58 @xcite , @xmath113 @xcite and @xmath1 @xcite collisions . from the latter measurements one hopes to obtain information about the production and interaction of @xmath111-mesons in the nuclear medium .the nuclear potential of the @xmath111-mesons is supposed to be strongly attractive , in contrast to @xmath0-mesons where we found a repulsive potential , or mass shift , of @xmath103 mev .the measurements can be carried out in the end of 2003 or beginning of 2004 .this work profitted significantly from discussions with members of the anke collaboration , in particular w. cassing , v. koptev , z. rudy , a. sibirtsev and c. wilkin .we would like to thank b.l .ioffe and h. strher for carefully reading the manuscript .support from the following funding agencies was of indispensable help for building anke , its detectors and daq : georgia ( department of science and technology ) , germany ( bmbf : grants wtz - rus-649 - 96 , wtz - rus-666 - 97 , wtz - rus-685 - 99 , wtz - pol-007 - 99 , wtz - pol-015 - 01 , wtz - pol-041 - 01 ; dfg : 436 rus 113/337 , 436 rus 113/444 , 436 rus 113/561 , state of north - rhine westfalia ) , poland ( polish state committee for scientific research : 2 p03b 101 19 ) , russia ( russian ministry of science : fnp-125.03 , russian academy of science : 99 - 02 - 04034 , 99 - 02 - 18179a ) and european community ( intas-98 - 500 ) .v. koptev , _ proc. workshop on strangeness production in pp and pa interactions at anke _ , june 21/22 , 2001 , gatchina , russia ; _ berichte des forschungszentrums jlich _ , jl-*3922 * , issn 0944 - 2952 , editors m. bscher _ et al . _

<S> the production of @xmath0-mesons in @xmath1 ( @xmath2 c , cu , ag , au ) collisions has been investigated at the cooler synchrotron cosy - jlich at beam energies @xmath3 gev using the anke spectrometer . </S><S> anke allows to measure double differential cross sections for @xmath0-production at forward angles @xmath4 in a wide momentum range . </S><S> the cross sections obtained in these experiments together with those from literature show a uniform behavior as functions of momentum transfer and excitation energy of the residual nucleus . at the lowest measured beam energy , far below the free @xmath5-threshold ( @xmath6 gev ) , </S><S> the data reveal a high degree of collectivity in the target nucleus . </S><S> the spectra obtained above threshold should allow one to extract the nuclear @xmath0-potential with an accuracy of better than @xmath7 mev . </S><S> information about the mechanisms leading to @xmath0-production can be obtained from @xmath8 and @xmath9 coincidence measurements . </S><S> = by -1 </S>

1601.06265

singular points of a positive semi - definite metric @xmath3 are the points where the metric is not positive definite . in the authors previous work @xcite with hasegawa and saji ,a class of positive semi - definite metrics on @xmath4-manifolds called whitney metrics was given . singularities of whitney metrics are isolated and the pull - back metrics of cross caps in euclidean @xmath0-space @xmath5 are typical examples of whitney metrics .in @xcite with hasegawa , the authors gave three intrinsic invariants @xmath6 , @xmath7 and @xmath8 for cross caps . after that they were generalized in @xcite as invariants of whitney metrics .in this paper , we construct a series of invariants @xmath9 as an extension of @xmath10 and @xmath8 .this series of invariants can distinguish isometric classes of real analytic whitney metric completely ( see section [ sec:5 ] ) , and are related to the following problem : can each singular point of a whitney metric locally be isometrically realized as a cross cap in @xmath5 ? the authors expect the answer will be affirmative , under the assumption that the metric is real analytic .in fact , for real analytic cuspidal edges and swallowtails , the corresponding problems are solved affirmatively by kossowski @xcite ( see also @xcite ) . moreover , the moduli of isometric deformations of a given generic real analytic germ of cuspidal edge singularity was completely determined in @xcite . in this paper , we construct all isometric realizations of a given whitney metric germ at their singularities as formal power series solutions of the problem .the above family of invariants @xmath9 corresponds to the coefficients of the taylor expansion of a certain realization of the whitney metric associated to a given cross cap singular point .so we can give an explicit algorithm to compute the invariants ( cf .section [ sec:5 ] ) .although it seems difficult to show the convergence of the power series , we can approximate it by @xmath1 maps by applying borel s theorem ( cf .* lemma 2.5 in chapter iv ) ) , and get our main result ( cf .theorem [ thm : main ] ) .we recall fundamental properties of cross caps ( cf .let @xmath11 be a @xmath1 map , where @xmath12 is a domain in @xmath13 .a point @xmath14 @xmath15 is called a _ singular point _ if @xmath2 is not an immersion at @xmath14 .consider such a map given by @xmath16 which has an isolated singular point at the origin @xmath17 and is called the _ standard cross cap_. a singular point @xmath14 of the map @xmath11 is called a _cross cap _ or a _ whitney umbrella _ if there exist a local diffeomorphism @xmath18 on @xmath13 and a local diffeomorphism @xmath19 on @xmath5 satisfying @xmath20 such that @xmath21 and @xmath22 .let @xmath23 be a @xmath1 map such that @xmath24 is a cross cap singularity and @xmath25 . since cross cap singularities are co - rank one , @xmath26 .we call the line @xmath27 the _ tangential line _ at the cross cap .the plane passing through @xmath28 spanned by @xmath29 and @xmath30 is called the _ principal plane_. the principal plane is determined independently of the choice of the local coordinate system @xmath31 satisfying @xmath25 . by definition ,the principal plane contains the tangential line . on the other hand , the plane passing through @xmath28 perpendicular to the tangential lineis called the _normal plane_. the unit normal vector @xmath32 near the cross cap at @xmath33 can be extended as a @xmath1 function of @xmath34 by setting @xmath35 and @xmath36 , and the limiting normal vector @xmath37 lies in the normal plane .we have the following normal form of @xmath2 at a cross cap singularity : [ fact : w ] let @xmath38 be a germ of a cross cap singularity at @xmath39 .then there exist an orientation preserving isometry @xmath40 and a local diffeomorphism @xmath41 such that @xmath42 satisfies @xmath43 where @xmath44 and @xmath45 are smooth functions satisfying @xmath46 moreover , if we assume @xmath47 at @xmath39 , then the function germs @xmath48 , @xmath49 and @xmath50 are uniquely determined .this special local coordinate system @xmath51 is called the _ canonical coordinate system _ or the _ normal form _ of @xmath2 at the cross cap singularity . in particular , the function @xmath44 is called the _characteristic function _ associated to the cross cap @xmath2 .historically , west @xcite initially introduced this normal form of cross caps .unfortunately , it is difficult to have an access to the reference @xcite .an algorithm approach to determine the coefficients of the taylor expansions of @xmath44 and @xmath45 can be found in fukui - hasegawa ( * ? ? ?* proposition 2.1 ) , which we will apply at section [ sec:5 ] .for the sake of the later discussions , we give here a proof of the last assertion of fact [ fact : w ] as follows . without loss of generality, we may assume that @xmath52 .suppose that there exists another such normal form @xmath53 where @xmath54 .since @xmath52 , two isometries @xmath40 and @xmath55 can be considered as matrices in @xmath56 . by and , it holds that @xmath57 , where @xmath58 . since the tangential lines of @xmath2 and @xmath59coincide , we have @xmath60 hence , @xmath61 is an eigenvector of the matrix @xmath62 . on the other hand , by andagain , both of @xmath63 and @xmath64 must be proportional to @xmath65 .since the principal planes of @xmath2 and @xmath59 coincide , we have @xmath66 since we know that @xmath61 is an eigenvector of @xmath67 , we can conclude that @xmath68 is also an eigenvector of @xmath67 .thus @xmath69 is also an eigenvector of @xmath67 , and we can write @xmath70 then we get the expression @xmath71 comparing the first components , we have @xmath72 next , comparing the second components , we have @xmath73 substituting @xmath74 , we get @xmath75 , and therefore @xmath76 .so we can conclude that @xmath77 . by, we have @xmath78 . by comparing the third components, @xmath79 holds .hence we have @xmath80 since @xmath81 , we can conclude that @xmath82 .in particular , we have @xmath83 , @xmath84 , and @xmath45 coincides with @xmath85. then reduces to @xmath86 , proving the assertion . in the statement of fact[ fact : w ] , @xmath87 and @xmath88 can be taken as real analytic functions if @xmath2 is real analytic .the following assertion was proved in @xcite : the characteristic function @xmath44 vanishes identically if and only if the set of self - intersections of @xmath2 lies in the intersection of the principal plane and the normal plane .[ def : normal ] cross caps whose characteristic functions vanish identically are called _ normal cross caps _@xcite ) .let @xmath89 ( resp .@xmath90 ) be the set of @xmath1 function germs at the origin @xmath91 of the @xmath31-plane @xmath13 ( resp .the line @xmath92 ) .two functions @xmath93 , @xmath94 ( resp .@xmath95 ) are called _ equivalent _ ( denoted by @xmath96 ) if the taylor series of @xmath97 coincides with that of @xmath98 at the origin . by the well - known borel theorem ( cf .* lemma 2.5 in chapter iv ) ) , the quotient space @xmath99 ( resp .@xmath100 ) can be identified with the space @xmath101 $ ] ( resp .@xmath102 $ ] ) of formal power series in the variables @xmath103 ( resp .@xmath104 ) at the origin @xmath91 , that is , the formal power series @xmath105:=\sum_{k , l=0}^\infty \frac{\partial^{k+l } h(0,0 ) } { \partial u^k\partial v^l}\frac{u^kv^l}{k!l ! } \qquad \left(\mbox{resp.}~ [ h]:= \sum_{j=0}^\infty \frac{d^j h(0)}{dt^j}\frac{t^j}{j ! } \right)\ ] ] represents the equivalent class containing @xmath106 in @xmath99 ( resp . in @xmath100 ) . at the end of this subsection, we show that the following assertion is an immediate consequence of our main result ( theorem [ thm : main ] ) : [ prop : main ] let @xmath107 @xmath108 be two real analytic cross cap singularities such that the first fundamental form ( i.e. the pull back of the canonical metric of @xmath5 ) of @xmath109 coincides with that of @xmath110 .then , @xmath109 coincides with @xmath110 up to orientation - preserving isometries in @xmath5 if and only if the taylor series of their characteristic functions coincide .proposition [ prop : main ] tells us that an analytic isometric deformation of cross caps can be controlled by the corresponding deformation of characteristic functions .examples of isometric deformations of cross caps are constructed in @xcite ( cf .figure [ fig : deform ] ) . by the definition of normal cross caps ( cf. definition [ def : normal ] ) , we get the following corollary : [ cor : rigidity ] two germs of real analytic normal cross caps are congruent if and only if they have the same first fundamental form . corollary [ cor : rigidity ] suggests us the following : can a given cross cap germ in @xmath5 be isometrically deformed into a normal cross cap ?if the answer to the problem in the introduction is affirmative , so it is for the above question . since the standard cross cap ( cf . )is normal , the deformation of the standard cross cap in figure [ fig : deform ] can be re - interpreted as a normalization of the rightmost cross cap to the normal cross cap ( i.e. the leftmost cross cap ) .we give here another example : [ ex : normalize ] we set @xmath111 this gives the normal form of a cross cap at @xmath17 .( see figure [ fig : example ] , left ) .since @xmath112 , this cross cap is not normal .we suppose that there exists a real analytic germ @xmath109 of a normal cross cap which is isometric to @xmath113 . by corollary [ cor : rigidity ] , we know the uniqueness of @xmath109 . moreover ,for a given positive integer @xmath114 , we can determine the coefficients of its taylor expansion of order at most @xmath114 using our algorithm as in the proof of theorem [ thm : main ] .figure [ fig : example ] , right is an approximation of @xmath109 by setting @xmath115 .the main difference between the figures of @xmath113 and @xmath109 appears on the set of self - intersection .the set of self - intersection of the figure of @xmath109 consists of a straight line perpendicular to the tangential direction of the surface at @xmath17 . :the cross cap @xmath113 and its corresponding normal cross cap @xmath109.,title="fig:",height=75 ] : the cross cap@xmath113 and its corresponding normal cross cap @xmath109.,title="fig:",height=75 ] we fix a @xmath4-manifold @xmath116 , and a positive semi - definite metric @xmath3 on @xmath116 .a point @xmath117 is called a _ singular point _ of the metric @xmath3 if the metric is not positive definite at @xmath14 .let @xmath14 be a singular point of @xmath3 , and @xmath31 a local coordinate system centered at @xmath14 .we set @xmath118 the local coordinate system @xmath31 is called _ admissible _ if @xmath119 is a null direction of the metric @xmath3 , that is , it holds that @xmath120 at the origin .[ def : adms ] a singular point @xmath14 of the metric @xmath3 is called __ admissible__-_flatness_. our definition of admissibility is equivalent to the original one , see ( * ? ? ?* proposition 2.7 ) . ]if there exists an admissible local coordinate system @xmath31 centered at @xmath14 satisfying @xmath121 at the origin .if each singular point of @xmath3 is admissible , then @xmath3 is called _ admissible_. [ def : hessian ] let @xmath14 be a singular point of an admissible ( positive semi - definite ) metric @xmath3 on @xmath116 in the sense of definition [ def : adms ] .let @xmath31 be an admissible local coordinate system centered at @xmath14 and set @xmath122 where @xmath123 are functions satisfying .if the hessian @xmath124 does not vanish at @xmath14 , then @xmath14 is called an _ intrinsic cross cap _ of @xmath3 . moreover ,if @xmath3 admits only intrinsic cross cap singularities on @xmath116 , then it is called a _whitney metric _ on @xmath116 .the definition of intrinsic cross caps is independent of the choice of admissible coordinate systems .a gauss - bonnet type formula for whitney metrics is given in @xcite .[ def : inf ] two metrics @xmath125 ( @xmath126 ) defined on a neighborhood of @xmath117 are called _ formally isometric _ at @xmath14 if there exists a local coordinate system @xmath31 centered at @xmath14 such that ( see for the definition of the bracket @xmath127 $ ] ) @xmath128=[e_2],\quad [ f_1]=[f_2],\quad [ g_1]=[g_2]\ ] ] hold at @xmath17 , where @xmath129 we write @xmath130 if two metrics are formally isometric .two cross cap germs are said to be _ formally isometric _ if their induced metrics are formally isometric .the following is the main result of this paper : [ thm : main ] let @xmath14 be a singular point of a whitney metric @xmath3 . for any choice of @xmath1 function germ @xmath131satisfying @xmath132 , there exists a @xmath1 map germ @xmath2 into @xmath5 having a cross cap singularity at @xmath14 satisfying the following two properties : 1 . the first fundamental form of @xmath2 ( i.e. the pull - back of the canonical metric of @xmath5 by @xmath2 ) is formally isometric to @xmath3 at @xmath14 , 2 .the characteristic function of @xmath2 is equivalent to @xmath87 , that is , it has the same taylor expansion at @xmath133 as @xmath87 .moreover , such an @xmath2 is uniquely determined up to addition of flat functions function @xmath134 is called _ flat _ ( at @xmath14 ) if @xmath135 vanishes at @xmath14 for all non negative integers @xmath136 , @xmath137 . ] at @xmath14 . in other words ,the taylor expansion of @xmath2 gives a unique formal power series solution for the realization problem of the whitney metric @xmath3 as a cross cap .if the problem in the introduction is affirmative , then the set of analytic cross cap germs which have the same first fundamental form can be identified with the set of convergent power series in one variable . the uniqueness of @xmath2 modulo flat functions and the second assertion of theorem [ thm : main ] immediately imply proposition [ prop : main ] , by setting @xmath138 and @xmath139 , where @xmath140 is the canonical metric of the euclidean @xmath0-space @xmath5 . from now on ,we fix a whitney metric @xmath141 defined on a neighborhood @xmath12 of the origin @xmath142 in the @xmath31-plane @xmath13 .we suppose that @xmath91 is a singular point of @xmath3 .we set @xmath143,\qquad { \mathcal{f}}:=[f],\qquad { \mathcal{g}}:=[g],\ ] ] that is , @xmath144 , @xmath145 , @xmath146 are the formal power series in @xmath101 $ ] associated to the coefficients of the metric @xmath3 . [ def : order ] a formal power series @xmath147 in @xmath101 $ ] is said to be of _ order at least @xmath148 _ if @xmath149 we denote by @xmath150 the ideal of @xmath101 $ ] consisting of series of order at least @xmath148 . by definition ,@xmath151 $ ] . in @xcite ,the following assertion was given : [ fact : hhnsuy ] one can choose the local coordinate system @xmath31 centered at the singular point of @xmath3 so that @xmath152 ) & = 1 + & & a_{2,0}^2 u^2 + 2 a_{2,0}a_{1,1}u v + ( 1+a_{1,1}^2)v^2 & & + o_3(u , v),\\ \label{eq : comp - f } { \mathcal{f}}\ ( = [ f ] ) & = & & a_{2,0}a_{1,1 } u^2 + ( a_{2,0}a_{0,2}+a_{1,1}^2 + 1)u v + a_{1,1}a_{0,2}v^2 & & + o_3(u , v),\\ \label{eq : comp - g } { \mathcal{g}}\ ( = [ g ] ) & = & & ( 1+a_{1,1}^2 ) u^2 + 2 a_{1,1}a_{0,2}u v + a_{0,2}^2v^2 & & + o_3(u , v ) , \end{aligned}\ ] ] where @xmath153 , @xmath154 and @xmath155 are real numbers , @xmath154 and @xmath156 are invariants of the whitney metric @xmath3 . ] , and @xmath157 @xmath158 is a certain element of @xmath150 .so we can assume that our local coordinate system @xmath31 satisfies , and .we set @xmath159 we now fix a @xmath1 function germ @xmath160 satisfying @xmath132 . [lem : realize - zero ] let @xmath161 be a @xmath1 map having cross cap singularity at @xmath17 satisfying , @xmath162 and @xmath163 such that the first fundamental form of @xmath2 coincides with the whitney metric @xmath3 .moreover , suppose that @xmath31 satisfies , and .then , @xmath164 since @xmath162 , @xmath165 , and @xmath166 , we have @xmath167 . in particular , @xmath168 holds and @xmath169 . on the other hand, we have @xmath170 and we get @xmath171 . by , @xmath172 holds , proving the assertion . replacing @xmath31 by @xmath173 if necessary, we may assume that @xmath174 the map @xmath2 as in lemma [ lem : realize - zero ] satisfies @xmath175 where @xmath176 [ def : term ] we call @xmath177 the _ terms of @xmath178_. similarly , the _ terms of @xmath179 _ ( resp .the _ terms of @xmath180 _ ) are also defined .we consider the following three polynomials in @xmath103 : @xmath181 we set @xmath182 where @xmath183=\sum_{j=3}^{\infty}{b_j}t^j/{j!}$ ] . by definition ,each coefficient of @xmath184 is a polynomial in the coefficients of @xmath185 , @xmath186 and @xmath187 . to describe the key assertion ( cf .proposition [ prop : key ] ) , we prepare a terminology as follows : [ def : terms ] a triple of polynomials @xmath188 as in , and are called the _ @xmath148-th formal solution _ if they satisfy ( cf . and )@xmath189 and @xmath190+o_{m+1}(u , v ) , \\\label{eq : f0 } { \mathcal{f}}&=[f^m_u\cdot f^m_v]+o_{m+1}(u , v ) , \\\label{eq : g0 } { \mathcal{g}}&=[f^m_v\cdot f^m_v]+o_{m+1}(u , v ) , \end{aligned}\ ] ] where @xmath191 is a term belonging to @xmath192 . the key assertion , which we would like to prove in section [ sec : app ] , is stated as follows : [ prop : key ] let @xmath31 be a local coordinate system satisfying , and .then for each @xmath193 , there exists a unique @xmath148-th formal solution .we prove here the case @xmath194 of the proposition : [ lem:2 ] let @xmath31 be a local coordinate system satisfying , and .then there exists a unique second formal solution .more precisely it has the following expressions : @xmath195 by a straightforward calculation using @xmath132 , we have @xmath196= 1 + 2 u x(2,0)+2 v x(1,1)+o_2(u , v).\ ] ] since @xmath197+o_2(u , v)$ ] , we can conclude that @xmath198 .similarly , using @xmath199 , we have @xmath200= vx(0,2)+o_2(u , v).\ ] ] in particular , @xmath201 .using the fact @xmath202 ( @xmath203 ) , we have @xmath204= 1+u^2 \left(x(3,0)+4 y(1,0)^2+z(2,0)^2\right ) \\ + 2 u v \left(x(2,1)+2 y(0,1 ) y(1,0 ) + z(1,1)z(2,0)\right)\\ + v^2 \left(x(1,2)+y(0,1)^2+z(1,1)^2\right ) + o_3(u , v ) . \end{gathered}\ ] ] by , we have @xmath205 by , we have @xmath206 similarly , yields @xmath207 since @xmath208 and @xmath156 are positive ( cf .fact [ fact : hhnsuy ] and ) , reduces to @xmath209 . then yields that @xmath210 .moreover , reduces to @xmath211 because of @xmath212 ( cf . ) . on the other hand , implies @xmath213 .also @xmath214 follows from .then yields @xmath215 . finally , and reduce to @xmath216 so we have @xmath217 .moreover , yields @xmath218 .now we can prove theorem [ thm : main ] under the assumption that proposition [ prop : key ] is proved : by proposition [ prop : key ] , we get formal power series @xmath219 , @xmath220 , @xmath221 $ ] such that @xmath222 satisfies @xmath223 , \quad { \mathcal{f}}=[f_u\cdot f_v],\quad { \mathcal{g}}=[f_v\cdot f_v].\ ] ] in fact , the coefficients @xmath224 ( @xmath225 ) do not affect our computation , and @xmath226&=[f^m_u\cdot f^m_u]+o_{m+1}(u , v ) , \\[ f_u\cdot f_v]&=[f^m_u\cdot f^m_v]+o_{m+1}(u , v ) , \\[ f_v\cdot f_v]&=[f^m_v\cdot f^m_v]+o_{m+1}(u , v ) \end{aligned}\ ] ] hold , where @xmath191 is a term in @xmath192. then by borel s theorem , there exist @xmath1 functions @xmath227 , @xmath228 , @xmath88 whose taylor series are @xmath219 , @xmath220 , @xmath229 , respectively .so we set @xmath230 then the first fundamental form of @xmath2 is formally isometric to @xmath3 . by lemma [ lem:2 ] ,the map @xmath231 is a local diffeomorphism at the origin .taking @xmath51 as a new local coordinate system , we can write @xmath232 and @xmath233 . so @xmath51 gives the canonical coordinate system of the map @xmath2 .thus @xmath2 satisfies ( 1 ) and ( 2 ) of theorem [ thm : main ] .in this section , we prepare several properties of power series to prove the case @xmath234 of proposition [ prop : key ] . as in section [ sec : prelim ] , we denote by @xmath101 $ ] the ring of formal power series with two variables @xmath103 in real coefficients . each element of @xmath101 $ ] can be written as in .each @xmath235 ( @xmath236 ) is called the _@xmath237-coefficient _ of the power series @xmath238 .moreover , the sum @xmath239 is called the _ order _ of the coefficient @xmath235 . in particular , @xmath235 @xmath240 consist of all coefficients of order @xmath148 .the formal partial derivatives of @xmath238 denoted by @xmath241 are defined in the usual manner .[ lem : diff ] the @xmath237-coefficient of the ( formal ) partial derivatives @xmath242 and @xmath243 of @xmath238 are given by @xmath244 linear operations on power series also have a simple description as follows : [ lem : linear ] let @xmath238 , @xmath245 be two power series in @xmath101 $ ] , and let @xmath246 , @xmath247 .then @xmath248 the coefficient formula for products is as follows : [ formula : power - prod ] let @xmath249 be power series in @xmath101 $ ] .then @xmath250 if @xmath251 , and @xmath252 and @xmath253 , then the formula reduces to the following : @xmath254 moreover , setting @xmath245 to be the monomial @xmath255 or @xmath256 , we get the following : [ cor : power - prod ] @xmath257 hold , where coefficients with negative induces @xmath258 , @xmath259 @xmath260 are considered as @xmath133 . in definition [ def : order ] , we defined the ideal @xmath150 of @xmath101 $ ] consisting of formal power series of order at least @xmath148 .the following assertion is obvious : [ lem : order - sum ] if @xmath261 and @xmath262 , then @xmath263 , where @xmath264 .let @xmath265 ( @xmath266 ) be a power series in @xmath267 .for a given non - negative integer @xmath148 , we set @xmath268 roughly speaking , this number is an upper bound of the degree of the terms of @xmath265 to compute the @xmath148-th order term of @xmath269 , as follows .[ lem : order - prod ] let @xmath265 be a power series in @xmath267 @xmath270 .then the product @xmath269 belongs to the class @xmath271 .moreover , each @xmath237-coefficient @xmath272 @xmath240 of order @xmath148 can be written as a homogeneous polynomial of degree @xmath273 in the variables generated by @xmath274 the proof of this assertion is not so difficult , and so we leave it as an exercise . as a consequence , we get the following : [ cor : pi ] let @xmath275\to { \boldsymbol{r}}[[u , v]]/{\mathcal{o}}_{m+1}$ ] be the canonical homomorphism .the @xmath148-th order term of @xmath269 are determined by @xmath276 where @xmath277 in other words , to compute the @xmath148-th order terms of @xmath278 , we need only the at most degree @xmath279 terms of @xmath280 .[ eq : pq ] we set @xmath281 and @xmath282 as @xmath283 respectively .then it holds that @xmath284 to compute @xmath285 modulo @xmath286 , we need the information of @xmath287 and @xmath288 .so , we have that @xmath289 and so @xmath290 , where @xmath291 is an element of @xmath292 ( @xmath293 ) .the coefficients of the terms of order @xmath0 are @xmath294 they are homogeneous polynomials of degree @xmath4 in the variables @xmath295 , @xmath296 , @xmath297 , @xmath298 .let @xmath299 be power series in @xmath101 $ ] .we suppose that for each @xmath280 ( @xmath300 ) there exists a unique number @xmath301 such that @xmath302 for the sake of simplicity , we set @xmath303 in this case , each coefficient of the product @xmath269 can be expressed as coefficients of @xmath299 . for each non - negative integer @xmath148, we set @xmath304 roughly speaking , this number is an upper bound of the degree of the terms of @xmath305 appeared in @xmath280 to compute the @xmath148-th order term of @xmath269 .in fact , by applying lemma [ lem : order - prod ] and lemma [ lem : diff ] to this situation , we get the following : [ cor : order - prod ] under the conventions and , each @xmath237-coefficient @xmath306 @xmath240 of order @xmath148 is a homogeneous polynomial of degree @xmath273 in the variables generated by @xmath307 [ cor : order - prod2 ] the @xmath148-th order terms of @xmath269 are determined by @xmath308 where @xmath309 in other words , to compute @xmath269 , we need only the at most degree @xmath310 terms of @xmath305 .we set @xmath311 and @xmath312 , @xmath313 .then , @xmath314 since @xmath315 and @xmath316 , we have @xmath317 so we have @xmath318 on the other hand , @xmath319 so we have @xmath320 in this case , the upper bound @xmath321 of the order of @xmath220 for the contribution of the order @xmath4 coefficients of @xmath285 is not sharp .in fact , there are no order @xmath4 terms for @xmath285. also , this does not contradict proposition [ cor : order - prod ] , since @xmath133 can be considered as a homogeneous polynomial of order @xmath4 whose coefficients are all zero .let @xmath322 , @xmath323 , @xmath324 and @xmath160 be @xmath1 functions as in lemma [ lem : realize - zero ] , and @xmath238 a polynomial in @xmath325 each term ( cf . ) of @xmath326 , @xmath327 or @xmath328 is a typical example of such polynomials .we denote by @xmath329 the finite formal power series in @xmath103 ( i.e. a polynomial in @xmath103 ) that results after the substitutions @xmath330 into @xmath238 .a term @xmath40 of @xmath326 , @xmath327 or @xmath328 is called an _ @xmath148-ignorable term _ ( @xmath234 )if each @xmath148-th order coefficient @xmath331 does not contain any of the top term coefficients @xmath332 of @xmath185 , @xmath186 , @xmath187 ( cf . , and ) . a term which is not @xmath148-ignorableis called a _ leading term _ of order @xmath148 . for the computation of leading terms, we will use the following two convenient equivalence relations : let @xmath333 be the associative algebra generated by @xmath334 we denote by @xmath335 the ideal of @xmath333 generated by @xmath336 if two elements @xmath337 , @xmath338 satisfy @xmath339 , then we write @xmath340 let @xmath238 , @xmath245 be two polynomials in @xmath341 if all of the coefficients of @xmath342 ( as a polynomial in @xmath103 ) are contained in @xmath343 , we denote this by @xmath344 this notation is the same as the one used in , and this is rather useful for unifying the symbols . for example , if the term @xmath40 satisfies @xmath345 if and only if the term @xmath40 is @xmath148-ignorable . in the right hand sides of , and , terms containing @xmath347 appear , and they are @xmath348 respectively . in this subsection , we show that the terms as in , and are all @xmath148-ignorable . to prove this , we prepare the following lemma : [ lem : beta ] each @xmath148-th order coefficient of the power series associated to @xmath349can be expressed as a polynomial in the variables @xmath350 @xmath351 .we can write ( cf . ) @xmath352=\sum_{r=2}^\infty \frac{b(r)}{r ! } t^r.\ ] ] since @xmath353 and the index satisfies @xmath354 , we have @xmath355 by lemma [ lem : order - prod ] , each @xmath148-th order term of @xmath356 is a homogeneous polynomial of degree @xmath357 in variables @xmath350 ( @xmath358 ) for all @xmath354 . in particular , each @xmath148-th order term of the power series induced by @xmath359 can be expressed as a polynomial of degree at most @xmath148 in variables @xmath350 ( @xmath358 ) , which proves the assertion . in this situation, we set @xmath360 then , this number gives an upper bound of the degree of the terms of @xmath220 appeared in @xmath361 to compute the @xmath148-th order term of @xmath269 .the terms given in , and are all @xmath148-ignorable .we can categorize the terms in , and into three classes .one is @xmath362 and the other two classes are @xmath363 and @xmath364 respectively .we consider the term @xmath365 in .we have @xmath366 since @xmath367 is less than @xmath368 , @xmath361 does not effect the leading term ( cf .lemma [ lem : beta ] ) .similarly , we have that @xmath369 and we can conclude that @xmath370 is an @xmath148-ignorable term .similarly , one can also prove that other three terms in are also @xmath148-ignorable .we next consider the term @xmath371 in .we have @xmath372 hence , the coefficients of @xmath220 appeared in @xmath361 do not effect the leading term .similarly , the facts @xmath373 imply that the term @xmath374 is an @xmath148-ignorable term .similarly , other two terms in are also @xmath148-ignorable . finally , we consider the term @xmath375 in .since @xmath376 the coefficients of @xmath220 appearing in @xmath377 do not effect the leading term .we have @xmath378 and can conclude that @xmath375 is an @xmath148-ignorable term . similarly , other two terms in are also @xmath148-ignorable .the following terms @xmath379 appear in , and .we show the following : [ prop : rest ] the terms given in , and are all @xmath148-ignorable terms . since @xmath380, we have @xmath381 this implies that @xmath382 is @xmath148-ignorable . on the other hand , @xmath383 and @xmath384 both consist of two derivatives of @xmath227 and @xmath385 , andthe former term has lower total order , so if we show @xmath383 is @xmath148-ignorable , then so is @xmath384 . in fact , since @xmath386 , we have @xmath387 so @xmath383 is @xmath148-ignorable .finally , @xmath388 and @xmath389 both consist of @xmath390 and derivatives of @xmath391 .the term @xmath389 has lower total order .so if it is @xmath148-ignorable , then so is @xmath388 .the fact that @xmath392 is @xmath148-ignorable follows from the following computations : @xmath393 finally , @xmath394 is @xmath148-ignorable because @xmath395 .applying the computations in the previous section , we prove the following : let @xmath148 be an integer greater than @xmath4 , and @xmath136 , @xmath137 non - negative integers such that @xmath396 then the @xmath148-th order terms of the equalities , and reduce to the following relations : @xmath397 where @xmath398 removing the @xmath148-ignorable terms and from @xmath399 $ ] in , we have @xmath196 \equiv_m l_1,\qquad l_1:=x_u^2+y^2 x_u^2 + 2 xyx_uy_u+z_u^2.\ ] ] similarly , we get from , , and that @xmath400 \equiv_m l_2 , \qquad [ f_v\cdot f_v ] \equiv_m l_3,\ ] ] where @xmath401 the first term of @xmath402 is @xmath403. we can write @xmath404 ( @xmath405 ) . since @xmath406 we have @xmath407 and @xmath408 where we have applied lemma [ lem : diff ] .the second term of @xmath402 is @xmath409 . since @xmath410 and @xmath411 the @xmath148-th order terms of @xmath409 might not be @xmath148-ignorable .in fact , it can be written in terms of the @xmath412-st order coefficients of @xmath220 as follows . since @xmath413 ( @xmath414 ), we have @xmath415 .since @xmath416 ( @xmath417 ) and @xmath418 , @xmath419 is an @xmath148-ignorable term .thus , we have @xmath420 and so @xmath421 where we have applied corollary [ cor : power - prod ] .we examine the third term @xmath422 of @xmath402 .since @xmath423 the @xmath148-th order terms of @xmath422 can be written in terms of the coefficients of @xmath424 modulo @xmath343 .thus @xmath425 and @xmath426 the fourth term of @xmath402 is @xmath427 . since @xmath428 ,we have @xmath429 hence the @xmath148-th order terms of @xmath427 can be written in terms of the coefficients of @xmath430 modulo @xmath343 .if we write @xmath431 then @xmath432 is an @xmath148-ignorable term , and @xmath433 since @xmath434 for @xmath435 , we have @xmath436 by , , and , we have .we next prove .since @xmath437 @xmath438 contributes to the leading term . thus @xmath439 and @xmath440 on the other hand , since @xmath441 the coefficients of the factors @xmath220 and @xmath442 appear in the leading terms . thus @xmath443 and@xmath444 the third term of @xmath445 is @xmath446 . since @xmath447only the factor @xmath424 affects the computation of the leading term .so we have @xmath448 and @xmath449 the fourth term of @xmath445 is @xmath450 .since @xmath451 @xmath430 and @xmath452 both contribute to the leading terms , and @xmath453 so we have @xmath454 by , , and , we obtain .finally , we consider @xmath455 .the first term of @xmath455 is @xmath456 . since @xmath457 the coefficients of @xmath442 affect the leading term of @xmath456 .we have @xmath458 and @xmath459 the second term of @xmath455 is @xmath460 .since @xmath461 we have @xmath462 by and , we obtain .we prove the assertion by induction . by lemma [ lem:2 ], we have already determined the coefficients @xmath463 for the sake of simplicity , we set @xmath464 we say that @xmath465 , @xmath466 , @xmath467 is _@xmath148-fixed _ if it is expressed in terms of @xmath468 to prove the theorem , it is sufficient to show that @xmath469 are all @xmath148-fixed . by , and, we can write @xmath470 for @xmath471 , where @xmath472 , @xmath473 and @xmath474 are all previously @xmath148-fixed terms , by the inductive assumption . if we set @xmath475 in , we have @xmath476 where we used the fact that @xmath477 .if we next set @xmath478 in , then we have @xmath479 and @xmath480 hence @xmath481 is @xmath148-fixed ( cf . ) . on the other hand , for @xmath482 can be rewritten as @xmath483 for @xmath484 .if we set @xmath475 in , then we have @xmath485 thus @xmath486 can be @xmath148-fixed .on the other hand , for @xmath487 can be rewritten as @xmath488 where @xmath489 .subtracting from , we have @xmath490 for @xmath489 . by and, we have @xmath491 for @xmath484. thus @xmath492 are @xmath148-fixed. then @xmath493 are @xmath148-fixed by , and @xmath494 are also @xmath148-fixed by . finally , if we set @xmath495 in , then we have @xmath496 and @xmath497 is @xmath148-fixed . since the terms @xmath472 , @xmath473 and @xmath498 as in , and can be computed using the lower order term of @xmath185 , @xmath186 , @xmath187 and the informations of @xmath144 , @xmath145 and @xmath146 , we get a desired @xmath148-th formal solution @xmath499 satisfying , and .the uniqueness of @xmath184 is now obvious from our construction .if needed , we can explicitly write down the lower order terms @xmath472 , @xmath473 and @xmath498 as in , and , but we omit such formulas here , as they are complicated .let @xmath500 be the set of germs of whitney metrics at their singularities .two metric germs @xmath501 ( @xmath502 ) in @xmath500 are called _ isometric _ if there exists a local diffeomorphism germ @xmath18 such that @xmath503 is the pull - back of @xmath504 by @xmath18 .a map @xmath505 is called an _ invariant of whitney metrics _ if it takes a common value for all metrics in each isometric class . for a cross cap singularity, we can take a canonical coordinate system @xmath51 such that @xmath506 is expressed as ( cf . ) @xmath507 as shown in ( * ? ? ?* theorem 6 ) , the coefficients @xmath153 , @xmath154 and @xmath156 are intrinsic invariants . by , and , one can observe that these three invariants are determined by the second order jets of @xmath508 and @xmath509 .so one might expect that the coefficients of the taylor expansions of the functions @xmath123 are all intrinsic invariants of cross caps .however , for example , @xmath510 is not an intrinsic invariant of @xmath2 , since @xmath511 is changed by an isometric deformation of cross caps ( cf .* theorem 4 ) ) . in this section ,we construct a family of intrinsic invariants @xmath9 of whitney metrics ( @xmath512 and @xmath8 have been already defined in @xcite ) . when the metric is induced from a cross cap expressed by the canonical coordinate , then @xmath153 , @xmath154 and @xmath156 as in coincide with @xmath512 and @xmath8 for the induced whitney s metric .let @xmath3 be a whitney metric defined on a @xmath4-manifold @xmath116 , and @xmath117 a singular point of the metric .applying theorem [ thm : main ] for @xmath513 , there exists a @xmath1 map germ @xmath2 into @xmath5 defined on a neighborhood @xmath12 of @xmath14 having a cross cap singularity at @xmath14 satisfying the following two properties : 1 .the first fundamental form @xmath514 of @xmath2 is formally isometric ( cf. definition [ def : inf ] ) to @xmath3 at @xmath14 , 2 .the characteristic function of @xmath2 is a flat function at @xmath14 , that is , the taylor expansion at @xmath14 is the zero power series .if @xmath2 is real analytic , it is a normal cross cap ( cf . definition [ def : normal ] ) .however , we do not assume here the real analyticity of @xmath514 and @xmath2 .taking the normal form of @xmath2 , we may assume that @xmath2 is expressed as @xmath515 where @xmath516 are smooth functions defined on a neighborhood of @xmath517 .for each pair of integers @xmath518 satisfying @xmath519 and @xmath520 , there exists a unique assignment @xmath521 such that @xmath522= \sum_{n=2}^\infty \sum_{i=0}^n \frac{\alpha^{d\sigma^2}_{i , n - i}}{i!(n - i ) ! } x^i y^{n - i}.\ ] ] so we may regard the series @xmath523 as a family of invariants of @xmath3 . by theorem [ thm : main ] , we get the following assertion : in other words , @xmath246 is a family of complete invariants distinguishing the formal isometry classes of whitney metrics at @xmath14 .this family of invariants also induces a family of intrinsic invariants for cross caps in an arbitrarily given riemannian 3-manifold @xmath527 as follows .let @xmath528 be a @xmath1 map which admits only cross cap singularities .then the induced metric @xmath514 gives a whitney metric .let @xmath117 be a cross cap singularity of @xmath2 .then we set @xmath529 which can be considered as a family of intrinsic invariants of a germs of cross cap singularities .when @xmath527 is the euclidean @xmath0-space , we can give an explicit algorithm to compute the invariants as follows : 1 .take the @xmath530-st ( @xmath193 ) canonical coordinate system @xmath31 centered at @xmath14 , that is , @xmath2 has the following taylor expansion at @xmath517 : @xmath531=\left(u , uv+\sum_{n=3}^{m+1 } \frac{b_nv^n}{n ! } , \sum_{n=2}^{m+1 } \sum_{i=0}^n \frac{a_{i , n - i}}{i!(n - i ) ! } u^i v^{n - i}\right ) + o_{m+2}(u , v).\ ] ] such a coordinate system can be taken using fukui - hasegawa s algorithm given in @xcite .2 . using this coordinate system @xmath31, we can determine the coefficients of the following expansion up to @xmath530-st order terms because of the expression @xmath532 : @xmath533&=\sum_{i+j\le m+1}\frac{e(i , j)}{i!j!}u^iv^j + o_{m+2}(u , v),\\ [ f]&=\sum_{i+j\le m+1}\frac{f(i , j)}{i!j!}u^iv^j + o_{m+2}(u , v),\\ [ g]&=\sum_{i+j\le m+1}\frac{g(i , j)}{i!j!}u^iv^j + o_{m+2}(u , v ) , \end{aligned}\ ] ] where @xmath534 .3 . setting @xmath513 , we compute @xmath535 according to the algorithm given in the proof of theorem [ thm : main ] .we formally set @xmath536 and substitute them into the expansions @xmath537 then we can determine all of the coefficients @xmath538 5 . using them , we can finally determine all of the coefficients of the expansion @xmath539=\sum_{i+j\le m}\frac{a_{i , j}}{i!j!}x^iy^j + o_{m+1}(x , y),\ ] ] where @xmath540 .however , the uniqueness of the expression was already shown , and one can alternatively compute @xmath541 via any suitable method .we remark that the normal cross cap shown in the right - hand side of figure [ fig : example ] is drawn using the invariants @xmath542 for @xmath543 .one can get the following tables of intrinsic invariants ; @xmath544 the numerators of the above invariants have been computed in @xcite .the authors also computed the fourth order invariants @xmath542 ( @xmath545 ) , which are more complicated .for example , @xmath546 has the simplest expression amongst them , which is given by @xmath547

<S> two cross caps in euclidean @xmath0-space are said to be _ formally isometric _ if their taylor expansions of the first fundamental forms coincide by taking a suitable local coordinate system . for </S><S> a given @xmath1 cross cap @xmath2 , we give a method to find all cross caps which are formally isometric to @xmath2 . as an application , we give a countable family of intrinsic invariants of cross caps which recognizes formal isometry classes completely . </S>

cond-mat0401262

condensation transitions @xcite have been observed in a wide variety of nonequilibrium steady states @xcite . such a transition occurs when a single lattice site contains a finite fraction of particles of the system .condensation transitions are also manifested in a different guise in driven diffusive systems , where typically the particles obey hard - core exclusion . therethe condensation appears as a transition from a freely flowing fluid phase to a jammed phase @xcite .such jamming transitions have been reported , for example , in the context of traffic flow modelling @xcite .a simple lattice model known as the zero - range process ( zrp ) @xcite illustrates an equivalence between condensation and jamming transitions and provides a framework within which condensation transitions may be analysed @xcite . in the model particles hop between lattice sites with rates which depend on the number of particles at the departure site .if the rates decay to zero or if they decay slowly enough to a finite value as the particle number is increased , one finds a condensation transition whereby a single lattice site becomes macroscopically occupied .the connection between condensation and jamming transitions was first made in the context of the bus route model @xcite where an approximate description assigns hopping rates to buses that decay as a function of the distance to the next bus ahead .if the decay is suitably slow a jamming transition arises whereby the buses cluster together into a jam and there is a large region of the bus route devoid of buses .this approximate description of the bus route model maps exactly onto a zrp where the buses correspond to the sites of the zrp and the number of bus stops in front of a bus corresponds to the number of particles at a site of the zrp .phase separation in driven diffusive systems may also be related to condensation transitions and the zrp has been used to formulate a general criterion for phase separation in one dimension @xcite .in addition to the condensation mechanism in the zrp described above ( condensation associated with hopping rates that are decreasing functions of the occupation of a site ) one can have condensation associated with _@xcite . heredisorder refers to different sites having different hopping rates .the latter mechanism has been shown to be equivalent to bose - einstein condensation @xcite .a particularly simple system that exhibits bose - einstein - like condensation is where there is one slow defect site at which a condensate forms at high enough density . in the present work we study a simple model , with a single defect site , that exhibits features of both condensation mechanisms ( bus - route - like and disorder - induced ) mentioned above .the steady state can be solved and the critical density for condensation in the thermodynamic limit can be predicted .however , finite systems show significant deviations from these predictions . in particular , the low density , fluid phase appears to continue to higher densities than would be allowed in the thermodynamic limit .this results in an overshoot in the current , above the thermodynamic saturation value . related finite size effectshave been reported in traffic flow models @xcite .our aim in this paper is to analyse in detail how the finite size overshoot in the current occurs .now , since the zrp dynamics conserves particle number the natural ensemble to consider is the canonical ensemble .however , one finds that analysis of the model is more direct within the grand canonical ensemble ( i. e. where the particle number fluctuates ) .it has been shown that in the infinite system limit the two ensembles are equivalent @xcite . however , since we are interested in the behaviour of large but finite systems , we broach the issue of the inequivalence of the two ensembles on finite systems .we are able to analyse the single defect site model in both the canonical and grand canonical ensembles and find that they give different predictions for the finite size overshoot in the current .in particular , within the canonical ensemble the overshoot is predicted to peak at a density @xmath0 , where @xmath1 is the thermodynamic critical density and @xmath2 is the number of sites , and we confirm this by simulations .the paper is organised as follows : in section 2 we define the zrp and discuss the two types of condensation mechanism ; in section 3 we introduce the single defect site ( sds ) model and analyse the condensation transition within the grand canonical ensemble ; in section 4 we present exact numerical results and simulations that illustrate the overshoot of the low density fluid phase on a finite system in the canonical ensemble ; in section 5 we analyse the sds model in the canonical ensemble ; and we conclude in section 6 .we consider a one - dimensional , asymmetric zero - range process .we have a one - dimensional lattice of @xmath2 sites labelled @xmath3 , upon which reside @xmath4 indistinguishable particles , each site being able to hold any number ( @xmath5 ) of particles .particles move from site to site under a dynamics which conserves the number of particles and is described as follows : a particle from site @xmath6 hops to site @xmath7 with a rate , @xmath8 , which can depend on both the departure site , @xmath6 , and the number of particles on this site , @xmath9 .the boundary conditions are periodic ; a particle hopping from site @xmath2 will hop to site @xmath10 .the configuration of the system at any given time is specified by the set of occupancies of each site , @xmath11 .thus we consider the zero - range process on a finite system of size @xmath2 ; in particular we shall be interested in the limit of large @xmath2 where the particle density @xmath12 is held fixed .the one - dimensional zero - range process , with particles hopping to adjacent sites in one direction only , has a steady state which is straightforwardly solved @xcite . requiring that there be no net flow of probability to or from a configuration ,one finds the probability of a configuration , @xmath13 , in the steady state to be @xmath14 where the @xmath15 are given by @xmath16 and @xmath17 has been introduced to ensure the probabilities are correctly normalised .this normalisation is analogous to the canonical partition function from equilibrium statistical mechanics , being the sum of the un - normalised probability over all allowed configurations , @xmath18 here we have summed over all occupancies for each site , from zero to infinity , and included the delta function to ensure that only those configurations with the correct total number of particles contribute . thus @xmath19 is the normalisation in the canonical ensemble .this steady state probability distribution can be used to calculate other steady state properties of the system . for our purposesthe most important of these is the average hopping rate from a site , @xmath20 , which we refer to as the current . in the steady statethis is independent of site and is given by @xmath21 where in the final step we have used the fact that @xmath22 ( the departure rate from an empty site is zero ) and @xmath23 for @xmath24 .depending on the chosen hopping rates the system can exhibit a condensation transition in the thermodynamic limit @xmath25 with the density of particles fixed . in this case , at low density the system is in a _ fluid _phase , by which we mean that all sites have average occupancies that are infinitesimal fractions of the total number of particles . on the other hand , at high densitythe system is in a _ condensed _ phase wherein a single site holds a finite fraction of the total number of particles .we shall show that there are two distinct mechanisms by which this may occur .the simplest approach to analysing the condensation transition is to use the grand canonical ensemble .the functions @xmath26 given in ( [ marginalsf ] ) are only defined up to some multiplicative factor , @xmath27 , which we have tacitly set to one .reinstating @xmath28 , we can interpret @xmath28 as the fugacity .then we can get the equivalent of the grand canonical partition function in the usual way @xcite , by summing over all configurations , including those that violate the constraint on the total particle number , and then choose @xmath28 such that the average total particle number is equal to @xmath4 : @xmath29 the constraint on the average particle number is written as follows @xmath30 with the average occupation of each site , @xmath31 , being found as usual by : @xmath32 which yields @xmath33 where @xmath34 is the particle density @xmath35 . to see the two apparently different mechanisms we analyse the behaviour of the rhs of ( [ saddlepoint1 ] ) as @xmath36 where @xmath37 is the radius of convergence of ( [ bigf ] ) . as @xmath36one of two things can happen , either the rhs of ( [ saddlepoint1 ] ) will converge , or it will diverge . if the rhs of ( [ saddlepoint1 ] ) converges at @xmath38 , then this will give a critical density , @xmath1 , above which ( [ saddlepoint1 ] ) can no longer be satisfied .hence there is a phase transition , whereby the excess density condenses onto a single site .we refer to this as mechanism a. if the rhs of ( [ saddlepoint1 ] ) diverges at @xmath38 , then it is still possible to have a phase transition by a different mechanism . recall that each term in the sum in ( [ saddlepoint1 ] ) is the density of particles at a site .it is possible for the rhs of ( [ saddlepoint1 ] ) to diverge due to only a single term , say site 1 , of the sum diverging .this would imply that site 1 contains a finite fraction of the total number of particles , i. e. we have a condensate .we refer to this as mechanism b. to see that mechanism b is reminiscent of the bose - einstein condensation mechanism , one thinks of the particles as bosons and the sites of the zrp as bose states. then site 1 into which condensation may occur corresponds to the ground state of the bosonic system .the choice of hopping rates will determine which mechanism takes place .previously studied cases are : 1 .the hopping rates are the same for each site and decay with increasing particle number @xcite ; then we can have a condensation through mechanism a. 2 .the hopping rates do not depend on the number of particles at a site , but do depend on the site , i. e. the system is disordered @xcite .then we can have a condensation through mechanism b. one very simple realisation of this is a single defect site with a hopping rate less than the other sites . then at high density this site will support a condensate .one important difference between mechanisms a and b is that in a there can be a spontaneous symmetry breaking , in that the site supporting the condensate is selected at random . with mechanismb , on the other hand , the term in the sum which diverges specifies which site supports the condensate ( taken as site 1 above ) .if we have a mixture of decay and disorder in the hopping rates , then it is possible to have a phase transition through either mechanism . in the present workwe investigate this possibility . in the grand canonical ensemble the currentis given by @xmath39 where to proceed to the final line we have used the fact that @xmath22 and @xmath40 for @xmath24 .thus in the grand canonical ensemble we have an equivalence between the current , @xmath20 , and the fugacity , @xmath28 . from the discussion of the behaviour of @xmath28 in the previous subsection we deduce that , in the _ infinite _ system limit, @xmath20 increases with density in the fluid phase until at density @xmath41 the current saturates at @xmath42 and we enter the condensed phase .however , on a large but _ finite _ system , things are more subtle . in mechanismb one can always solve ( [ saddlepoint1 ] ) for @xmath43 .thus @xmath20 should always be less than its saturation value @xmath37 . in mechanisma , on the other hand , one can not satisfy ( [ saddlepoint1 ] ) for @xmath44 with @xmath45 .thus it is possible that @xmath20 overshoots the value @xmath37 @xcite . in simulations ( in the canonical ensemble ) of finite systems an overshoot in the current versus density diagram can be observed , i. e. the current appears to go above the expected maximum value before dropping back down at higher density ( see section [ sec : numerics ] ) .thus on a finite system there can appear to be a continuation of the fluid phase into the region where , on the infinite system , one would expect a condensate .we now restrict ourselves to a very simple zrp with all hopping rates constant and equal , except for a single slow defect site which has a hopping rate that decreases for increasing particle number .we refer to this as the single defect site ( sds ) model .we begin by analysing the model within the grand canonical ensemble .we consider the one - dimensional lattice of @xmath46 sites with periodic boundary conditions from section [ modeldefn ] and now choose the following hopping rates ( for @xmath47 ) @xmath48 \mathrm{u}_{\mu } ( l ) & = & 1\quad\quad\mbox{for}\quad \mu>1\ ; , \label{simplehopmu}\end{aligned}\ ] ] with @xmath49 and we generally take @xmath50 , giving site 1 a hopping rate slower than all the others for occupancies @xmath51 . then from ( [ marginalsf ] ) @xmath52 \mathrm{f}_{\mu}(n ) & = & 1\quad\quad\mbox{for}\quad \mu>1\ ; , \label{simplefmu}\end{aligned}\ ] ] where @xmath53 is the pochhammer symbol defined by @xmath54 also , from ( [ bigf ] ) we have @xmath55 where the hypergeometric function @xcite is defined as @xmath56 to see the equivalence in ( [ bigf1stih ] ) note that @xmath57 . if we look at the grand canonical analysis , specifically equation ( [ saddlepoint1 ] ), for the sds model the slow site gives a ratio of hypergeometric functions , while the ` non - slow ' sites , @xmath58 , give a ratio of two simple geometric sums @xmath59 & = & \frac{z}{m}\frac { { } _ { 2}\mathrm{f}_{1}(2,2\,;2+\lambda\,;z / p ) } { p(1+\lambda ) \ ; { } _ { 2}\mathrm{f}_{1}(1,1\,;1+\lambda\,;z / p ) } + \left ( 1 - \frac{1}{m } \right ) \frac { z } { 1-z } \\[0.5ex ] & = & \frac{\langle n_1(z)\rangle}{m } + \left ( 1 - \frac{1}{m } \right ) \langle n(z)\rangle\;. \label{saddlepointsimple}\end{aligned}\ ] ] in the last line @xmath60 means the average number of particles at site 1 , whereas @xmath61 means the average number of particles at a site @xmath62 .thus , as @xmath63 , the critical density will be @xmath64 to see this we take the limit @xmath65 and @xmath66 of the expression ( [ saddlepointsimple ] ) such that @xmath67 .this last condition ensures we are not in the condensed phase .however , for @xmath68 we must have @xmath69 finite to solve ( [ saddlepointsimple ] ) , thus we have a condensate on the slow site . in order to determine the finite size corrections to the current @xmath70 we analyse ( [ saddlepointsimple ] ) for @xmath2 large but finite .we invoke the following results @xcite @xmath71 \mbox{if}\quad c = a+b \quad \lim_{x\to 1 ^ - } \frac{{}_{2}\mathrm{f}_{1}(a , b\,;c\,;x ) } { & = & \frac { \gamma(a+b)}{\gamma(a)\,\gamma(b)}\\[1ex ] \mbox{if}\quad c< a+b \quad \lim_{x\to 1 ^ - } \frac{{}_{2}\mathrm{f}_{1}(a , b\,;c\,;x ) } { ( 1-x)^{c - a - b } } & = & \frac { \gamma(c)\,\gamma(a+b - c)}{\gamma(a)\,\gamma(b)}\;\end{aligned}\ ] ] where @xmath72 is the usual gamma function .thus @xmath73 \mbox{if}\quad \lambda = 2 & & \lim_{z\to p^- } \frac{\langle n_1(z)\rangle}{|\ln(1-z / p)| } = 1\\[1ex ] \mbox{if}\quad 1<\lambda < 2 & & \lim_{z\to p^- } \langle n_1(z)\rangle(1-z / p)^{2-\lambda } = \frac{(\lambda-1)^2\ , \pi}{\sin(\pi(\lambda-1))}\label{lambda22}\\[1ex ] \mbox{if}\quad \lambda = 1 & & \lim_{z\to p^- } \langle n_1(z)\rangle(1-z / p)|\ln(1-z / p)| = 1\\[1ex ] \mbox{if}\quad \lambda< 1 & & \lim_{z\to p^- } \langle n_1(z)\rangle(1-z / p ) = ( 1-\lambda)\;,\end{aligned}\ ] ] where to obtain ( [ lambda22 ] ) we have used the identity @xmath74 therefore when @xmath75 for @xmath76 one can satisfy ( [ saddlepointsimple ] ) by choosing @xmath77\\[1ex ] \mbox{if}\quad 1<\lambda < 2 & & 1-z / p \sim \left [ \frac { ( \lambda-1)^2 \pi}{\sin(\pi(\lambda-1 ) ) ( { \ensuremath{\varphi}}-{{\ensuremath{\varphi}}_{\rm c}})}\frac{1}{m } \right]^{1/(2-\lambda)}\\[1ex ] \mbox{if}\quad \lambda = 1 & & 1-z / p \sim \frac{1}{{\ensuremath{\varphi}}-{{\ensuremath{\varphi}}_{\rm c}}}\frac{1}{m\ln m}\\[1ex ] \mbox{if}\quad \lambda < 1 & & 1-z / p \sim \frac{(1-\lambda)}{{\ensuremath{\varphi}}-{{\ensuremath{\varphi}}_{\rm c}}}\frac{1}{m}\;. \label{zgc4}\end{aligned}\ ] ] for @xmath78 on the other hand , we can not satisfy ( [ saddlepointsimple ] ) as @xmath79 when @xmath80. instead we must consider @xmath81 .thus the grand canonical treatment implies an overshoot in the current @xmath20 ( @xmath82 ) for @xmath83 . in order to consider @xmath84 we impose cut - offs @xmath85 in the sums of ( [ zgce ] ) .this explicitly ensures that no site contains more particles than the total number @xmath4 .then @xmath86 for @xmath83 we write @xmath87 in the following form : @xmath88 to satisfy ( [ saddlepointsimple ] ) for @xmath44 we require @xmath89 .it turns out that as @xmath28 increases above @xmath90 this is first achieved when the numerator of ( [ n1co ] ) is @xmath91 and the denominator @xmath92 . in anticipation of thiswe write @xmath93 and @xmath94 , i. e. we have replaced the sums in ( [ n1co],[limitl ] ) by integrals over the asymptotic forms of the summands , retaining a finite error term @xmath95 in the denominator . then we require that @xmath96 and @xmath97 . now the asymptotic behaviour of @xmath98 is @xmath99 thus we require @xmath100 which implies @xmath101 we can now summarise the results of the grand canonical analysis for a large but finite system * for @xmath83 , @xmath102 remains finite as @xmath103 and we identify this case with condensation mechanism a of section [ mechsec ] . in the condensed phase , in order to satisfy ( [ saddlepointsimple ] ) @xmath104 overshoots its saturation value @xmath90 .the approach of @xmath20 to its saturation value as particle number @xmath4 increases is @xmath105 .* for @xmath76 , we can access the condensed phase through ( [ saddlepointsimple ] ) and @xmath20 approaches its saturation value from below as detailed in ( [ zgc1][zgc4 ] ) .we identify the case @xmath76 with mechanism b.in this section we will present results obtained from monte carlo simulations of the system and exact numerical calculations of @xmath17 in the canonical ensemble ( fixed particle number ) .what we shall find is that on finite systems the overshoot in the current , @xmath20 , is clearly present whenever @xmath106 .this is in contradiction to the grand canonical analysis of the previous section which predicts an overshoot only for @xmath107 . to observe the overshoot in the current , @xmath20 , exact numerical calculations of @xmath17 for systems of sizes up to @xmath108 , with up to @xmath109 particles were undertaken .these calculations are performed recursively @xcite .it is then possible to obtain @xmath20 via the relation ( [ vequals ] ) .the current , @xmath20 , is plotted against particle density at fixed system size for various values of @xmath110 and @xmath111 in figure [ vdifflambdafig ] . in this figurewe see that for low density the current increases whereas for high density it saturates . for @xmath106 , in between these two regimesis an overshoot in the current where we observe non - monotonic behaviour .remember that , for an infinite system , as the density is increased we would expect to see @xmath20 increase to a maximum value and stay there , yielding a current versus density curve that is non - analytic at the critical density .thus , for the infinite system , when @xmath20 is increasing the system is in a fluid phase , and when @xmath20 saturates the system is in a condensed phase . from figure [ vdifflambdafig ]we see that on a _ finite _ system , for @xmath106 , @xmath20 overshoots above its expected saturation value .this effect increases with increasing @xmath110 .notice also that for a finite system , @xmath20 crosses its predicted saturation value for an infinite system , in this case @xmath90 , close to the predicted critical density for an infinite system , given by @xmath112 from equation ( [ phic ] ) .the current , @xmath20 , is also shown in figure [ vdiffmfig ] , plotted against particle density , but for different system sizes , @xmath113 with @xmath114 and @xmath111 , from which it can clearly be seen that the overshoot decreases in severity for increasing system size . thus the current versus density curve will approach the expected infinite system result . , plotted against particle density for a system of size @xmath108 , with @xmath111 and @xmath115 from bottom to top .[ vdifflambdafig ] , width=377 ] , plotted against particle density for a system with @xmath116 , @xmath111 and varying system size @xmath113 from top to bottom .[ vdiffmfig],width=377 ] in order to investigate what is actually happening in the overshoot region , monte carlo simulations of the system were run at chosen points within the overshoot .the simulations were run in three key areas : firstly , between the predicted critical density on an infinite system ( [ phic ] ) and the maximum of the overshoot ; secondly , between the maximum of the overshoot and where @xmath20 drops down close to its expected maximum value of @xmath90 ; and thirdly where @xmath20 has dropped down close to @xmath90 . as we are interested in condensation - like phenomena , the simulations output graphs of the number of particles on the slow site with time .some typical examples of these graphs are shown in figure [ 3animouts ] ., @xmath111 with @xmath108 , obtained via monte carlo simulation .( a ) system is at @xmath117 , between expected critical density , in this case @xmath118 , and maximum of the overshoot .this is an extension of the fluid phase .( b ) system is at @xmath119 , just past the maximum of the overshoot .this is the coexistence region , a small condensate is seen to appear intermittently , but is not stable over long periods of time .( c ) system is at @xmath120 , beyond the ` end ' of the overshoot region where the current is approaching its expected maximum value @xmath121 . herethe large pile of particles at the slow site is stable and the system is in the condensed phase .[ 3animouts ] , width=453 ] of course , on a finite system , strictly one can not identify thermodynamic phases .however , it is useful to think of putative phases emerging in large systems . in the following andfigure [ 3animouts ] ` phase ' is often taken to mean such a putative phase . inspecting figure [ 3animouts ]we see that the overshoot is initially an extension of the fluid phase ( figure [ 3animouts]a ) ; up to the maximum of @xmath20 the system is still in the fluid phase . at the maximuma condensate begins to emerge on the slow site . in the range of densities just beyond the maximum of the overshoot the emergent condensate appears intermittently , but is not stable over long time periods ( figure [ 3animouts]b ) .we interpret this as ` coexistence ' between the fluid and condensed phases , i. e. at any given time the system is in one or the other .eventually this coexistence gives way to a clear condensate as @xmath20 drops back down to its predicted maximum value of @xmath90 ( figure [ 3animouts]c ) .having established what takes place in the overshoot region , we proceed in the next section to give an analytical account for this effect working within the canonical ensemble .the sds detailed in ( [ simplehop1 ] , [ simplehopmu ] ) , has a convenient property in that the canonical normalisation , @xmath17 ( [ bigzcanongeneral ] ) , can be reduced to a single sum .this makes it possible to analyse the model directly within the canonical ensemble .the form that @xmath17 takes for the sds model is @xmath122 here we have simply inserted the @xmath15 ( [ simplef1],[simplefmu ] ) into the general expression for @xmath17 ( [ bigzcanongeneral ] ) .the simple form of the hopping rates for sites @xmath123 ( [ simplehopmu ] ) allows the sum over @xmath124 to be performed easily , yielding the combinatoric factor which comes from the delta function in ( [ bigzcanongeneral ] ) .we also re - label @xmath125 as @xmath126 for convenience .the normalisation is the sum over all un - normalised probabilities of occupancies of the sites .this means that the magnitude of the @xmath127 term in the sum ( [ zcanon ] ) is proportional to the probability of the slow site having occupancy @xmath126 .thus if @xmath17 is dominated by a maximum at low @xmath126 , the corresponding system is in the fluid phase .conversely , if @xmath17 is dominated by a maximum at large @xmath126 , the corresponding system is in the condensed phase .if there are two maxima of similar magnitudes we interpret this as phase coexistence .thus investigating the maxima of this sum should furnish us with an explanation of the overshoot . to investigate the maxima of this sum, we look for its stationary points , i. e. points where the ratio of two consecutive terms tends to one . solving for these points gives a quadratic equation in @xmath126 , the solutions of which are : @xmath128 + \frac{a}{2 } { } \pm \frac{m}{2 } \left [ \left ( { \ensuremath{\varphi}}- { \ensuremath{\varphi}}_{\mathrm{c } } \right)^{2 } + \frac{1}{m } \left\ { 2 \left ( { \ensuremath{\varphi}}- { \ensuremath{\varphi}}_{\mathrm{c } } \right ) a - 4 \lambda { \ensuremath{\varphi}}_{\mathrm{c } } ( 1+{\ensuremath{\varphi } } ) \right\ } \right . \\ \left . { } + \frac{1}{m^{2 } } \left\ { a^{2 } + 4 \lambda { \ensuremath{\varphi}}_{\mathrm{c } } \right\ } \right]^{\frac{1}{2}}\ ; , \label{nturningpts}\end{gathered}\ ] ] where @xmath129 is the expected critical density and the constant @xmath130 .the nature of these turning points can be determined by considering the ratio of the first two terms and also of the last two in ( [ zcanon ] ) . in the followingwe also assume that @xmath50 for simplicity , although this turns out to make little difference to the overall picture .first we note that the roots in ( [ nturningpts ] ) become real at a density @xmath131 given by @xmath132^{\frac{1}{2 } } \;. \label{ph2}\ ] ] thus on increasing the density from zero the profile of the terms in the sum goes through the sequence illustrated in figure [ turningptsfig ] : ( a ) for @xmath133 there is a boundary maximum at @xmath134 ; ( b ) at @xmath135 a stationary point emerges ; ( c ) for @xmath136 a maximum emerges at large @xmath126 in addition to the local maximum at @xmath134 ; ( d ) the maximum at large @xmath126 increases and dominates , with @xmath134 eventually becoming a boundary minimum .( [ zcanon ] ) , for a system of size @xmath108 , with @xmath137 and @xmath138 .( a ) @xmath139 ( @xmath140 ) , boundary maximum at @xmath134 , no turning points .( b ) @xmath141 ( @xmath142 ) , boundary maximum at @xmath134 and stationary point at higher @xmath126 .( c ) @xmath143 ( @xmath144 ) , boundary maximum at @xmath134 with greater maximum at high @xmath126 .( d ) @xmath145 ( @xmath146 ) , boundary minimum at @xmath134 and maximum at high @xmath126 .[ turningptsfig],width=377 ] we interpret this sequence as follows . at low density ,@xmath147 there are no turning points in the physical region @xmath148 , with @xmath126 real .thus the values of @xmath126 near @xmath134 dominate and this corresponds to the fluid phase . if we were to consider the @xmath66 limit of ( [ ph2 ] ) we would obtain @xmath149 .however for @xmath2 finite , the square root part of ( [ nturningpts ] ) remains imaginary up to @xmath150 and the system remains in the fluid phase until this point .thus the maximum at high @xmath126 emerges beyond the expected critical density @xmath151 by an amount @xmath152 : @xmath153 when @xmath34 is just greater than @xmath154 we interpret the boundary maximum at @xmath134 and the turning point at large @xmath126 as coexisting fluid and condensed phases .finally , for @xmath34 much larger than @xmath131 the maximum at high @xmath126 dominates and we interpret this as the condensed phase .we expect that the maximum of the overshoot will be near to the point @xmath154 as this is where a putative condensate can begin to form and reduce the current .we see that an amount @xmath152 beyond @xmath1 is consistent with the data from exact numerical calculation of the current ( figure [ vdiffmfig ] ) .in fact , we compared the peak of the overshoot with @xmath154 for systems of size @xmath108 we found very good agreement for @xmath155 .outwith this range we believe that further finite size effects are becoming important .the form of @xmath17 , given in equation ( [ zcanon ] ) , also allows the behaviour of the current in the condensed phase to be analysed .one finds that the asymptotic behaviour of the current for large particle number ( i. e. where the sum in ( [ zcanon ] ) is dominated by a maximum at large @xmath126 ) is @xmath156\;. \label{vovercanon}\ ] ] thus for @xmath106 the current approaches its asymptotic value from above as @xmath157 . it should be noted that the second term in the expansion is small only for @xmath158 .in this work we have studied the condensation transitions of a zrp with a single defect site , in both the grand canonical and canonical ensembles .analysis in the grand canonical ensemble predicts the two different condensation mechanisms ( a and b ) discussed in section [ mechsec ] . for @xmath78 mechanisma applies which predicts a finite size overshoot in the current .the approach of @xmath20 to its saturation value @xmath90 is @xmath159 . for @xmath76 mechanism b applies and there should be no overshoot .however , simulations in the canonical ensemble reveal an overshoot for all @xmath106 ; only for @xmath160 do we see the expected mechanism b behaviour .analysis within the canonical ensemble confirms this and predicts that for @xmath106 the overshoot in the current peaks at a density @xmath161 .the approach of @xmath20 to its saturation value as particle number increases is @xmath162 .thus it is not clear to what extent the mechanisms a and b are relevant or well - defined within the canonical ensemble .moreover , from the point of view of finite size critical behaviour , the two ensembles are not equivalent. it would be interesting to further understand the finite size inequivalence of the ensembles . in order to shed some light onthis one can follow a standard approach @xcite to evaluating the canonical normalisation ( [ bigzcanongeneral ] ) in terms of the grand canonical partition function ( [ zgce ] ) .we use the integral representation of the delta function to write the canonical normalisation as a contour integral in the complex plane @xmath163 then for large system size and particle number , @xmath2 and @xmath4 respectively , this integral will be dominated by its saddle point , given by @xmath164 where @xmath34 is the particle density ( @xmath35 ) .this equation precisely recovers ( [ saddlepoint1 ] ) .however it must be borne in mind that ( [ sad ] ) holds only when the saddle point of ( [ zint ] ) is _ valid _ , i. e. when it dominates the integral ( [ zint ] ) .it would not be surprising for the saddle point to break down when it approaches the radius of convergence of @xmath165 in a way that depends on system size , as in the analysis of ( [ saddlepoint1 ] ) in section [ condgcesec ] .further analysis of this phenomenon would be enlightening .finally it would be of interest to perform simulations within the grand canonical ensemble to confirm our analytical predictions . to dothis one would introduce transitions whereby particles are created and annihilated with appropriate rates .we thank m. e. cates for useful discussions and comments .a. g. a. thanks the carnegie trust for the universities of scotland for a studentship .d. m. is supported by the israel science foundation ( isf ) and the einstein center and visits to edinburgh were funded by epsrc under grant number gr / r52497 .99 m. r. evans , braz .j. phys .* 30 * , 42 ( 2000 ) o. j. oloan , m. r. evans and m. e. cates , phys . rev .e * 58 * , 1404 ( 1998 ) s. n. majumdar , s. krishnamurthy and m. barma , phys .* 81 * , 3691 - 3694 ( 1998 ) d. chowdhury , l. santen , a. schadschneider , physics reports * 329 * , 199 ( 2000 ) f. spitzer , advances in math . * 5 * , 246 ( 1970 ) s. grosskinsky , g. m. schtz and h. spohn , j. stat .phys . *113 * , 389 ( 2003 ) y. kafri , e. levine , d. mukamel , g. m. schtz and j. trk , phys .* 89 * , 035702 , ( 2002 ) m. r. evans , europhys .lett . * 36 * , 13 ( 1996 ) j. krug and p. a. ferrari , j. phys .a * 29 * , l465 ( 1996 ) m. r. evans , j. phys .a * 30 * , 5669 ( 1997 ) r. barlovic , l. santen , a. schadschneider and m. schreckenberg , eur.phys .j. b * 5 * , 793 , ( 1998 ) k. huang , _ statistical mechanics _ , ( john wiley and sons , 1987 ) y. kafri , d. mukamel and l. peliti , eur .j. b * 27 * , 135 ( 2002 ) g. e. andrews , r. askey and p. roy , ed .rota , _ special functions _ ( encyclopedia of mathematics and its applications vol .71 ) , ( cambridge university press , 1999 ) p. bialas , z. burda and d. johnston , nucl .b * 493 * , 505 ( 1997 )

<S> condensation occurs in nonequilibrium steady states when a finite fraction of particles in the system occupies a single lattice site . </S><S> we study condensation transitions in a one - dimensional zero - range process with a single defect site . </S><S> the system is analysed in the grand canonical and canonical ensembles and the two are contrasted . </S><S> two distinct condensation mechanisms are found in the grand canonical ensemble . </S><S> discrepancies between the infinite and large but finite systems particle current versus particle density diagrams are investigated and an explanation for how the finite current goes above a maximum value predicted for infinite systems is found in the canonical ensemble . </S>

gr-qc0609126

the entropy associated to a black hole is given by the formula @xmath5 where @xmath6 is the horizon area @xcite .the statistical origin of @xmath7 is still not completely clear .however , its status as thermodynamical entropy is strongly supported by classical and semiclassical calculations in black hole backgrounds @xcite .these give support to the generalized second law of thermodynamics , which is the ordinary second law but where the black holes are considered as objects having the entropy ( [ bh ] ) @xcite . as a result, the association of area and entropy permeates to different situations including flat space ones .consider for example the following thought experiment @xcite .take a nearly flat region of space - time and imagine a spherical shell of matter collapsing over it in such a way to produce a black hole of area @xmath6 .thus , if the generalized second law is right ( and if the black hole really forms @xcite ) the original entropy present in the spherical region @xmath1 of boundary area @xmath8 satisfies the spherical bound @xcite @xmath9 this may indicate that the dimension of the hilbert space of physical states which can be localized in @xmath1 is finite and given by @xmath10 @xcite . in that caseany entropy @xmath11 for a region @xmath12 would also be bounded by @xmath13 .this is not simply a consequence of ( [ bo ] ) since in general the entropy is not monotonously increasing with size .this stronger form of the spherical bound can be called the holographic bound .different entropy bounds suggested by black hole thermodynamics have been proposed , and some covariant forms have been developed which are also applicable to regimes of strong gravity .in particular , the bousso entropy bound has ( [ bo ] ) as a special case @xcite .it involves the so called light - sheets .given a two dimensional spatial surface @xmath14 , a light - sheet @xmath15 for it is a null hypersurface orthogonal to @xmath14 which has everywhere non - positive expansion .the bound reads @xmath16 similarly , if a light - sheet @xmath17 starts at a spatial two dimensional surface @xmath18 and ends at a spatial surface of smaller area @xmath19 , the fmw bound reads @xcite @xmath20 this is stronger than the bousso bound and implies both , the generalized second law and the bekenstein bound @xcite .the bounds ( [ bousso ] ) and ( [ mfw ] ) hold in classical regimes , with the assumption that the entropy is extensive and can be computed out of some sort of entropy density @xcite .it is also required that the entropy density is bounded above by an expression involving the energy density and the planck mass . in this casethe reason behind the validity of the bounds turns out to be simply that the entropy requires energy , which in turn curves the space in such a way that ( [ bousso ] ) and ( [ mfw ] ) hold ( for example by ending a light sheet ) .thus , classical gravity is what keeps the bounds safe according to these examples in the classical regime @xcite .here we are interested in testing the entropy bounds in empty flat space .classically the entropy bounds are safe in this case , since no entropy is present . beyond the classical regime however , the vacuum fluctuations start to play a very relevant role .indeed , the quantum state @xmath21 relevant to the algebra of operators acting on a local region @xmath1 follows from the trace of the vacuum state over the exterior region hilbert space @xmath22 , @xmath23 the corresponding entropy ( entanglement entropy ) is given by @xmath24 and is divergent @xcite .the divergent terms are proportional to quantities which are local in the boundary of @xmath1 .that is , in four dimensions we have @xcite @xmath25 \,\epsilon^{-2 } + g_1[\partial v]\,\epsilon^{-1 } + g_0[\partial v]\,\log ( \epsilon)+ s_0(v)\ , , \label{div}\ ] ] where @xmath26 is a finite part , @xmath27 is a short distance cutoff , and the @xmath28 are local and extensive functions on the boundary @xmath29 , which are homogeneous of degree @xmath30 .the leading divergent term coefficient @xmath31 $ ] is proportional to the square of the size of @xmath1 , what is usually referred to as the area law for the entanglement entropy @xcite .however , @xmath32 and @xmath33 depend on the regularization procedure and @xmath32 is not proportional to the area if this later is not rotational invariant .these two terms are not physical within qft since they are not related to continuum quantities . in a series of recent papers ,this type of expansion was also found and exploited in the context of the ads - cft duality @xcite .the particular form of the divergent terms in ( [ div ] ) is due to both , the local nature of the ultraviolet divergences , and that the boundary is shared between @xmath1 and @xmath34 which have the same entropy ( for any pure global state ) @xmath35 on a technical level it is also the consequence of the fact that the entanglement entropy is the variation of the free energy under small conical singularities located on the boundary of @xmath1 @xcite .physically , the divergences have their origin in the same problem which obstructed the realization of a relativistic quantum mechanics of a single particle .the localization of the particle in a size smaller than its mass gives place to the creation of particle - antiparticle pairs , what in turn requires the simultaneous description of any number of particles .the localization in @xmath1 also produces pairs of particles , one member going inside @xmath1 and the other outside it , leading to an infinite amount of entanglement .this divergent entropy must be present in the localized entropy for any state , and also in curved space .in contrast , the finite value for the entropy of relativistic gases refers to the infinite volume limit . in that case oneusually considers the theory at a finite temperature to be defined in a box with some specific boundary conditions , avoiding the entanglement with the exterior region .any boundary condition for the finite volume theory , or a cutoff for the localized state in the unbounded theory , give the same result for the entropy per unit volume in the infinite volume limit , since the differences appear as area growing terms . however , the ambiguities are unavoidable when we need a prescription for calculating the entropy in a finite region , such as a black hole , the region inside a cosmological horizon , or the flat space region @xmath1 in the bound ( [ bo ] ) . in fact ,the same divergences show up in the qft calculation of the black hole entropy , where the horizon replaces the artificial division of inside and outside given by the boundary of @xmath1 @xcite . according to a widespread view, the finite result ( [ bh ] ) would be the manifestation of the quantum nature of space - time .quantum gravity would cutoff the ordinary qft calculation below a certain size of order of the planck length , where the description of space - time in terms of a classical manifold breaks down . in this scenarioit is expected that a fundamental cutoff regulates also the flat space entropy @xcite .in absence of a specific prescription for the physical cutoff on the entanglement entropy we have to deal with the qft results . in analogy with the calculation of the casimir energy , the worst effects of the vacuum fluctuations may be eliminated , but not at a zero cost .the subtraction of divergences must be done carefully and the entanglement entropy of the vacuum can not be simply put to zero .this is because there are finite quantities which are universal ( independent of regularization prescription ) and which can be derived from it .one such quantity is the mutual information @xcite @xmath36 between two non intersecting sets @xmath1 and @xmath3 .note that all divergent terms get subtracted in @xmath4 , since the contact terms also get subtracted away . ] .the mutual information is used in statistics as a measure of the information shared by two systems . in the present contextit can be thought as a correlator between non local objects , which may be defined for any qft irrespective of the field content . in special casesone can compute it explicitly . for large separating distances between @xmath1 and @xmath3it behaves as expected , is exponentially decreasing for massive theories and decreasing as a power of the distance for massless ones .we remark that @xmath4 is a well defined quantity at the axiomatic level in qft ( using an alternative expression rather than ( [ mutual ] ) ) , with the only requirement that there must be a non zero separation distance between @xmath1 and @xmath3 @xcite .imagine that @xmath3 gets larger and larger covering @xmath34 , and the boundary of @xmath3 approaches the one of @xmath1 from outside , encircling it . in the limit when @xmath1 and @xmath3 cover the hole space we would have heuristically @xmath37 , @xmath38 and @xmath39 thus , the mutual information diverges as the boundaries of the sets get closer to each other , somehow reproducing the divergences which were present in @xmath0 , but now in a universal way . in this sense @xmath4 is a kind of `` point splitting '' regularization of @xmath0 .the form of the leading term in the mutual information for two sets @xmath1 and @xmath3 approaching to each other in a conformal field theory can be inferred from dimensional analysis .if two parallel faces of area @xmath6 , one from @xmath1 and one from @xmath3 , are at small distance @xmath40 , we have in the conformal case @xmath41 where @xmath42 is the space - time dimension @xcite .we have computed explicitly the dimensionless constant @xmath43 for free fields @xmath44 where @xmath45 and @xmath46 are the multiplicity of the bosonic and fermionic degree of freedom , @xmath47 is the volume of a @xmath48 dimensional unit sphere , and @xmath49 and @xmath50 are the two dimensional entropic functions corresponding to boson and fermion fields @xcite .these are defined by the formula @xmath51 , where @xmath52 is the entanglement entropy corresponding to an interval of length @xmath53 in @xmath54 dimensions @xcite .the subindex @xmath55 corresponds to a two dimensional real scalar field and @xmath56 to a majorana fermion one , and @xmath57 is the field mass .we have in four dimensions the approximate result @xmath58 the mutual information @xmath4 can be thought as the localized form of entropy which replaces @xmath0 in the qft @xcite .it is also a positive quantity , and in addition , it has a very nice property which does not hold for the entropy : it is monotonically increasing with the size of the sets @xcite , @xmath59 this means that the mutual information is smoothly varying with the set , and that the sharpness in the definition of the set boundaries is not very relevant to it . using the inequality ( [ ine1 ] ) andthe fact that it is always possible to consider the density matrix @xmath60 as a partial trace of a pure state on a bigger space , we can show @xmath61 these mathematical properties of @xmath4 are reasonable , and make justice to the name of mutual information .the formula ( [ fofo ] ) is the starting point of our application of the mutual information to the entropy bounds .the mutual information @xmath4 is calculable in qft and in principle it might help to state and test the entropy bounds beyond the classical regime . in first placewe note that the combination of ( [ fofo ] ) with the actual results in qft ( see for example formula ( [ tuy ] ) ) show that the entanglement entropy in flat space , regularized by a cutoff at high energy , can not be renormalized to zero . in generalwe expect it to be very large , if actually finite ( see also the discussion at the end of @xcite ) .this is because we can use the qft safely at least up to separating distances which are large with respect to the planck length ( we elaborate more on this delicate point below ) .this big amount of entropy may pose difficulties to the entropy bounds , and at the same time make them more constraining and interesting . in particular, consider the case of the fmw bound in minkowski space .it gives zero entropy crossing for pieces of the null hyperplanes since the bounding areas @xmath18 and @xmath19 are equal in this case .imagine then a polyhedron @xmath62 included in a spatial hyperplane in flat space .it has a cauchy surface formed by several null surfaces @xmath63 included in null hyperplanes .each @xmath63 corresponds to one of the polyhedron faces , to which it is orthogonal .the total entropy has to be less than the sum of the entropies of the individual @xmath63 since the entropy is subadditive .thus we have according to the fmw bound @xmath64 however , @xmath62 has a non zero ( and in general very large ) entropy due to the vacuum fluctuations .we could try to reinterpret the fmw bound in such a way that this counterexample would not work .a possibility is to ask eq .( [ mfw ] ) to hold for the difference in the entropies crossing @xmath17 for a given global state and the vacuum state .this subtraction resembles the calculations done in the refs .@xcite in relation to the bekenstein bound ( which is related to the fmw bound ) .the problem is then to formulate this new version in a fully relativistic manner .consider now the holographic bound for a sphere @xmath1 .it reads in terms of the mutual information @xmath65 for any @xmath66 and any @xmath3 nonintersecting with @xmath67 .the case of a set @xmath3 with boundary approaching the one of @xmath1 shows that this bound is always violated naively in qft , since when the separating distance goes to zero @xmath4 tends to infinity .however , to violate the bound in this way we need to use planck scale distances between @xmath1 and @xmath3 .the space - time structure is uncertain at this scale , and the qft assumption presumably breaks down .then , consider instead the following construction .let @xmath67 be a set included in @xmath1 with large bounding area @xmath68 .let also @xmath3 be another set with boundary approximately parallel to the one of @xmath67 , that is , located at a distance from it which is kept around a fixed value @xmath40 ( i.e. the multi - layered sets in figure 1 ) .the area @xmath69 is then bounded as @xmath70 where @xmath71 is the sphere radius .according to eq .( [ tuy ] ) the mutual information between @xmath67 and @xmath3 can be very large , @xmath72 if one chooses @xmath73 with @xmath74 the planck mass , the mutual information @xmath75 exceeds the holographic bound @xmath76 , even if the separating distance @xmath40 is large in planck units .the equation ( [ volume ] ) gives a cutoff independent ( and infinite volume limit independent ) meaning to the statement that the number of degrees of freedom in a qft grows as the volume .thus , the `` area law '' for the entanglement entropy in qft does not refer to the number of internal degrees of freedom .it rather means simply that the number of degrees of freedom which are connected with the exterior grows as the area . on the other hand , once gravity is turned on one encounters objections to the application of the qft results even in this case involving distances which are large in planck units .the field modes through which the information is shared between @xmath67 and @xmath3 are of typical wavelength @xmath40 , and there are @xmath77 of these modes . if ( [ holds ] ) holds , the typical energy involved in the vacuum fluctuations is then of order @xmath78 which is greater than the mass of a black hole of radius @xmath71 . then , in principle it is possible that quantum gravity reduces the mutual information in this example through ( virtual ) black holes @xcite .naively one could have expected otherwise , that the gravitons add to the mutual information rather than diminish it .note that in any case , the physical `` ultraviolet '' cutoff must act paradoxically in a non local way in order to prevent the violation of the holographic bound .a covariant local cutoff to the entropy ( such as a pauli - villars regularization @xcite ) is not enough .on the other hand , the spherical bound and its generalization , the bousso bound , refer to the entropy , not to the number of degree of freedom , and are not in conflict with the construction above . the question is then if it is possible to violate this bound using ( [ fofo ] ) in a situation where the sets involved are at a distance to each other which is large in planck units .this is not possible for sets lying in the same spatial hyperplane ( for a reasonably small number of fields ) , but , strangely enough , can be done by locating them in space - time . in order to have large mutual information at long separation distances we consider a conformal field theory .take @xmath1 to be a sphere .since the mutual information is increasing under inclusion , we chose the set @xmath3 to be as large as possible being at a given distance from @xmath1 , as shown in the figure ( 2 ) .the conformal invariance of the vacuum state implies that the mutual information for this geometry is a function of the cross ratio @xmath79 with @xmath80 . then , in order to compute @xmath4 , we can put @xmath1 and @xmath3 in the same spatial plane , and if @xmath81 is small we have @xmath82 . in this casethe thin annulus between @xmath1 and @xmath3 tends to a long strip , and we use equation ( [ tuy ] ) to show that @xmath83 now consider the case @xmath84 .we have in this case @xmath85 and @xmath86 where @xmath87 is the spatial distance between the set edges . thus , given @xmath1we can keep @xmath40 fixed to any given distance and make the area of @xmath3 grow indefinitely in order to surpass the bound for @xmath1 . to do so according to ( [ bo ] ) and ( [ fofo ] ) we need @xmath88 where @xmath89 and @xmath90 are the radius of the spheres corresponding to the boundaries of @xmath1 and @xmath3 .this inequality can be satisfied for @xmath91 large enough , without requiring @xmath71 or @xmath40 to be short . however , as in the case of the holographic bound previously discussed , this violation of the bousso bound is not free from its own internal difficulties .for example , there are sub - planckian scales in this construction , such as @xmath92 .this is what invalidates the arguments which lead to ( [ esi ] ) in the case we model the theory in a square lattice of planck length spacing . in this casethe existence of a bound on the entropy ( different from ( [ bo ] ) ) for a finite volume set is evident .however , the physical cutoff should respect long distance covariance , and we could have used for @xmath81 the equivalent expression @xmath93 ( instead of ( [ eta ] ) ) , which does not involve any small distances . also , the field modes which interchange information between the sets are of large angular momentum @xmath94 .this can be seen by transforming back to the situation where the sets are in the same spatial hyperplane , where these modes have typical wavelength corresponding to the separation distance between @xmath1 and @xmath3 .this angular momentum gives one mode per solid angle @xmath95 , what represents more than one mode per planck area on the surface of @xmath1 .however , a direct angular discreteness is not compatible with large scale rotational invariance .this problem points out again at least an unusual way the physical cutoff is required to act to save the bound . maybe there is a hint in eq .( [ tara ] ) of a different way gravity can save the bound here . in fact , we do not have access to arbitrarily large flat space - times if the cosmological constant is present , what can indicate that we should take @xmath96 in order to apply the arguments above .then , equation ( [ tara ] ) resembles the one proposed by dirac in 1937 as an intriguing relation between the planck scale , the cosmological scales and some scale of particle physics @xcite .indeed , at low energies there is only the massless electromagnetic field ( we do not consider the graviton ) .however , due to the coupling with the electron , its self - interaction is important at energies higher than the electron mass .thus , the distances @xmath40 and @xmath71 can not be smaller than a typical size given by electrodynamics where the interactions between fotons break conformal invariance .let us take heuristically @xmath97 .the number of powers of @xmath98 can be different and we ignore the optimal numerical prefactor , which is not necessarily near @xmath99 .this gives an approximate bound @xmath100 which is surprisingly of the correct order of magnitude of the cosmological observations @xcite , @xmath101 a massless neutrino can surpass the bound , since it has much smaller interactions .this idea needs to be further explored using qed in a de sitter background .the argument above is connected to a theorem proved in ref .there it was shown that if the entanglement entropy is finite and lorentz invariant it must be exactly proportional to the boundary area .the proof is based on the strong subadditive property of the entropy , and involves crucially short and long distances of arbitrary size in minkowski space .an entropy proportional to the area means that the mutual information is identically zero , @xmath102 .as we have mentioned this simply can not hold since the mutual information , at least in certain situations , is a low energy quantity which is non zero in qft .of course , qft resolves this tension just making the entropy divergent .if we insist in keeping the entropy finite and covariant there must be a short distance cutoff invalidating the arguments involving arbitrarily short scales .but it is possible that this is not enough if the theory has a non trivial infrared fixed point . in this casewe could also need an infrared scale to keep @xmath4 for , say , human size configuration of sets @xmath1 and @xmath3 , fixed to a given positive value of order one .in conclusion we can say that the formulation and testing of the entropy bounds at a semiclassical level require treating qft in a bounded region .this means for example imposing a cutoff or an artificial boundary condition .these procedures necessarily give ambiguous values for the entropy because they contain input from external elements . in this sensethe mutual information offers a unique way to deal in a cutoff independent way with the localized entropy in qft . at the same time it maintains certain inequalities with the entropy which allow stating the entropy bounds unambiguously .it clarifies the role of the vacuum fluctuations showing that they lead to violations to the entropy bounds even in empty space .it also shows that if the entropy bounds hold in nature some essentially non local effects must be produced by the physical cutoff , even in very simple and specific flat empty space situations. these could be simpler scenarios that black holes in order to contrast ideas .i thank marina huerta for discussions and for long standing collaboration , on the results of which this paper is based . for a review on the entropy boundssee for example r. bousso , rev .phys . * 74 * , 825 ( 2002 ) [ arxiv : hep - th/0203101 ] .g. t hooft , arxiv : gr - qc/9310026 .r. bousso , jhep * 9907 * , 004 ( 1999 ) [ arxiv : hep - th/9905177 ] .s. ryu and t. takayanagi , phys .lett . * 96 * , 181602 ( 2006 ) [ arxiv : hep - th/0603001 ] ; s. ryu and t. takayanagi , jhep * 0608 * , 045 ( 2006 ) [ arxiv : hep - th/0605073 ] ; t. hirata and t. takayanagi , arxiv : hep - th/0608213 .h. casini and m. huerta , phys .b * 600 * , 142 ( 2004 ) [ arxiv : hep - th/0405111 ] .h. casini , c. d. fosco and m. huerta , j. stat . mech . * 0507 * , p007 ( 2005 ) [ arxiv : cond - mat/0505563 ] . h. casini and m. huerta , j. stat .mech . * 0512 * , p012 ( 2005 ) [ arxiv : cond - mat/0511014 ] .d. kabat , nucl .b * 453 * , 281 ( 1995 ) [ arxiv : hep - th/9503016 ] ; s. n. solodukhin , phys .d * 52 * , 7046 ( 1995 ) [ arxiv : hep - th/9504022 ] ; f. larsen and f. wilczek , nucl .b * 458 * , 249 ( 1996 ) [ arxiv : hep - th/9506066 ] ; m. hotta , t. kato and k. nagata , class .* 14 * , 1917 ( 1997 ) [ arxiv : gr - qc/9611058 ] .j. g. demers , r. lafrance and r. c. myers , phys .d * 52 * , 2245 ( 1995 ) [ arxiv : gr - qc/9503003 ]. h. araki , publ .kyoto univ . 809( 1976 ) , and kyoto univ - rims-191 ( 75,rec.oct ) ; o. bratteli and d. w. robinson , _ operator algebras and quantum statistical mechanics . vol . 2 _ , springer - verlag 1981 ; d. buchholz and e. h. wichmann , commun . math .phys . *106 * , 321 ( 1986 ) ; d. buchholz , c. dantoni and r. longo , commun .phys . * 129 * , 115 ( 1990 ) .h. casini , in preparation .d. marolf , d. minic and s. f. ross , phys .d * 69 * , 064006 ( 2004 ) [ arxiv : hep - th/0310022 ] ; d. marolf , arxiv : hep - th/0410168 .l. smolin , gen .* 16 * , 205 ( 1984 ) ; gen .* 17 * , 417 ( 1985 ) .

<S> we consider some formulations of the entropy bounds at the semiclassical level . </S><S> the entropy @xmath0 localized in a region @xmath1 is divergent in quantum field theory ( qft ) . instead of it </S><S> we focus on the mutual information @xmath2 between two different non - intersecting sets @xmath1 and @xmath3 . </S><S> this is a low energy quantity , independent of the regularization scheme . </S><S> in addition , the mutual information is bounded above by twice the entropy corresponding to the sets involved . </S><S> calculations of @xmath4 in qft show that the entropy in empty space can not be renormalized to zero , and must be actually very large . </S><S> we find that this entropy due to the vacuum fluctuations violates the fmw bound in minkowski space . </S><S> the mutual information also gives a precise , cutoff independent meaning to the statement that the number of degrees of freedom increases with the volume in qft . </S><S> if the holographic bound holds , this points to the essential non locality of the physical cutoff . </S><S> violations of the bousso bound would require conformal theories and large distances . </S><S> we speculate that the presence of a small cosmological constant might prevent such a violation . </S>

cond-mat0107225

the description of effects of disorder on the critical behavior of finite - size systems have attracted a lot of interest @xcite . up to nowa discussion takes place of whether the introduced disorder influences the finite - size scaling ( fss ) results@xcite , compared to the standard fss results , known for pure systems @xcite .a formulation of general fss concepts for the case of disorder is strongly complicated due to the additional averaging over the different random samples . for a random sample with volume @xmath3 , where @xmath0 is a linear dimension , any observable property @xmath4 , singular in the thermodynamic limit , has different values for different realizations of the randomness and can be considered as stochastic variable with mean @xmath5 and variance @xmath6 , where the over line indicates an average over all realizations of the randomness . here, an important theoretical problem of interest is related with the property of self - averaging ( sa ) @xcite .if the system does not exhibit sa a measurement performed on a single sample does not give a meaningful result and must be repeated on many samples .a numerical study of such a system also will be quite difficult .this point has been studied recently by means of fss arguments @xcite , renormalization group ( rg ) analysis @xcite and monte carlo simulations @xcite .the quantity under inspection is the relative variance @xmath7 .a system is said to exhibit `` strong sa '' if @xmath8 as @xmath9 .this is the case if the system is away from criticality , i.e. if @xmath10 . at the criticality ,i.e. when @xmath11 , the situation depends whether the randomness is irrelevant ( @xmath12 , e.g. `` pure '' , p - case ) or relevant ( @xmath13 , e.g. `` random '' r - case ) @xcite .one calls the former case `` weak sa '' , since @xmath14 , and the latter case `` no sa '' , since @xmath15 is fixed nonzero universal quantity even in the thermodynamic limit @xcite .the lack of sa in disordered systems implies that the standard fss breaks down .thus , one needs to formulate a fss theory , suitable for the case under consideration .there is an ongoing activity in this field but an understanding of the problems at a deeper level is still desirable @xcite .the successful application of the field theoretical methods @xcite for the analytic calculations of the size - dependent universal scaling functions for pure systems makes the possible extension of these methods very appealing also in the case of disordered systems .the appropriate field theory is the n - component @xmath16 theory with a `` random- @xmath1 '' term @xcite .the analytic difficulties raised by the disorder usually are avoided with the replica trick by solving an effective pure problem @xcite .however , the perturbative structure of the theory is still much more complicated than for the corresponding pure system , arising additional difficulties in the applicability of the ideas proposed in @xcite . some of them are of rather hard computational nature . between the generic models for magnetism , the most popular andrelatively well studied by means of monte carlo simulations is the ising model ( i.e. @xmath17 ) due to the fact that in three dimensions it is the model for which in accordance with the harris criterion , randomness is relevant .but its rg analysis is complicated as a result of the well known accidental degeneracy in the recursion relations that needs higher order in @xmath18 and so the use of the loop expansion to second order ( see e.g.@xcite ) .this leads to the apparent computational difficulties in the finite - size treatment of the system .the @xmath18- calculations based on the loop expansion to second order are not done even for the pure finite - size systems . in this situation ,more attractive for the finite - size rg study are the disordered xy ( @xmath19 ) and heisenberg ( @xmath20 ) cases as simpler and having the same qualitative features governed in @xmath2 dimensions by a random fixed point .other problems have more basic nature and are related with the breaking of the replica symmetry ( see @xcite ) . since its deeper understanding is still lacking , they are beyond of our interest in the present study . in this paper we analyze the finite - size properties of an n - component ( @xmath21 ) model of randomly diluted magnet with hypercubic geometry of linear size @xmath0 .exact calculations are performed in the mean - field regime @xmath22 , and up to the first order in @xmath18 near the upper critical dimension @xmath23 .although in this case , the problem of usefulness of the corresponding series expansions away from the dimension @xmath2 arises and is questionable ( see @xcite and refs . therein ) , we shall show that many generic fss properties of the model can be established and we hope this would have implications for more realistic cases .the paper is organized as follow . in sections[ secmodel ] and [ seceff ] we define the model and the effective hamiltonian . in section [ seczeromode ]we perform the analysis in the zero - mode approximation . in sections[ secshift ] and [ seccoupling ] we give the expressions for the shift of the critical temperature and the renormalized coupling constants in first order in @xmath18 . section [ secfss ] deals with the verification of the fss and the analysis of the problem of self - averaging is given in section [ secsa ] .finally in section [ discussion ] we present our main conclusions .we consider the `` random - @xmath1 '' ginzburg - landau - wilson model of disordered ferromagnets ( see , e.g. @xcite ) @xmath24,\ ] ] where @xmath25 is a n - component field with @xmath26 and the random variable @xmath27 has a gaussian distribution @xmath28}{\sqrt{2\pi\delta } } \ ] ] with mean @xmath29 and variance @xmath30 the over line in ( [ eqgu ] ) indicates a random average performed with the distribution @xmath31 . herewe will consider a system in a finite cube of volume @xmath3 with periodic boundary conditions .this means that the following expansion takes place @xmath32 and @xmath33 where @xmath34 is a discrete vector with components @xmath35 , @xmath36 , @xmath37 and a cutoff @xmath38 ( @xmath39 is the lattice spacing ) . in this paper , we are interested in the continuum limit , i.e. @xmath40 . in our case of quenched randomness one must average the logarithm of the partition function over the gaussian distribution ( [ eqgaus ] ) to produce the free energy @xmath41=-\int_{-\infty}^{\infty}d\varphi({\bbox x})p(\varphi({\bbox x } ) ) \ln { \cal z}_{r},\ ] ] where @xmath42.\ ] ] it is well known that the direct average of @xmath43 over the gaussian leads to equivalent results @xcite for the critical behavior as the @xmath44 limit of the following `` pure '' translationally invariant model @xcite @xmath45 \nonumber \\ & & + \frac{\delta}{8 } \sum_{\alpha , \beta=1}^{n } \int_{l^{d } } d^{d}x | \psi_{\alpha}({\bbox x})|^{2 } | \psi_{\beta}({\bbox x})|^{2 } ] .\end{aligned}\ ] ] here @xmath46 , @xmath47 ( @xmath48 beings the number of replicas ) are components of an @xmath49-components field @xmath50 .because of this equivalence , the model @xmath51 has been the object of intensive field - theoretical studies ( see @xcite and refs . therein ) in the bulk case . much lessis known for the equivalence of @xmath52 and the @xmath44-limit of @xmath51 in the finite - size case .problems may arise when finite - size techniques are used , since both procedures @xmath9 and removing of disorder by the `` trick '' @xmath53 may not commute .in this work we will use the rg technique introduced in @xcite and @xcite for studying pure systems with finite geometry .this technique permits explicit analytical calculations above and in the neighborhood of the upper critical dimension .the main idea is to expand the field in ( [ eqglw ] ) in fourier modes and then to treat the zero mode separately from the nonzero modes .the nonzero modes can be treated by the methods developed for the bulk systems ( e.g. loop expansion ) , while the zero mode , whose fluctuations are damped at the critical temperature , has to be treated exactly .this overcomes the problems due to the infra red ( ir ) divergences that take place in finite size systems ( see reference @xcite ) .in our more complicated case we have two possibilities : to consider the random model eq .( [ eqglw ] ) or to consider the replicated pure model eq .( [ glwcub ] ) . the last one is closer to the case treated in @xcite and @xcite by getting around the difficulties due to random average performed with @xmath31 and is used in the present study . for this case ,the replicated partition function is given by @xmath54.\ ] ] we decompose the field @xmath55 into a zero momentum component @xmath56 , which plays the role of the uniform magnetization and a second part depending upon the non - zero modes @xmath57 .after some algebra , the partition function can be expressed as @xmath58 \sigma_{\alpha}^2+{\rm higher\ powers\ of\ \sigma}\right\}.\end{aligned}\ ] ] here the terms involving @xmath59 vanishes since @xmath60 depends only on non zero modes .the terms containing @xmath60 are treated using diagram expansion , leading to the effective hamiltonian in the one - loop approximation @xmath61 where @xmath62 and @xmath63 will be presented bellow .with the help of the identity @xmath64 , \ ] ] we get @xmath65= \int_{-\infty}^{\infty}dy p_{n}\left(y\right ) \left[s_{n}\int_{0}^{\infty } d|\phi||\phi|^{n-1}\exp({\cal h}_{r}^{\rm eff}(n))\right]^{n},\ ] ] where @xmath66\ ] ] is an effective hamiltonian with a random variable @xmath67 with gaussian distribution ( depending on @xmath63 ) @xmath68 and @xmath69 is the surface of a @xmath70-dimensional unit sphere .let us note that the above mentioned equivalence between the models ( [ eqglw ] ) and ( [ glwcub ] ) in the used approximation may be mathematically expressed , within the used approximation by the following relation : @xmath71=-\frac{\partial}{\partial n } { \cal z}_{p}(n)\mid_{n=0}.\ ] ] from eqs .( [ ihs ] ) and ( [ meeq ] ) , and by using the identity @xmath72 for the free energy , we get @xmath73=-\int_{-\infty}^{\infty}dy p_{0}(y)\ln { \cal z}_{r}(0),\ ] ] where @xmath74\ ] ] is the partition function for the random system ( [ reff ] ) after taking the limit @xmath75 .the obtained effective `` random-@xmath1 model '' ( [ reff ] ) , distributed with gaussian weight ( [ eqgn ] ) , is the analytic basis of this paper .the effective constants @xmath76 , @xmath77 and @xmath63 involve @xmath48 and finite - size @xmath0 as parameters . for describing the finite - size properties of the initial model ( [ eqglw ] ) , as follows from eqs .( [ a1b2 ] ) and ( [ zet0 ] ) , it is necessary to set @xmath48 to zero . in the next sections we shall consider the results of this procedureif we neglect the loop corrections this corresponds to the mean - field approximation .then the zero mode playing the role of the uniform magnetization may be treated exactly . in this casethe effective hamiltonian ( [ effectiveham ] ) of the model reduces to @xmath78.\ ] ] now using an appropriate rescaling of the field @xmath79 and introducing the scaling variable @xmath80 for the partition function eq .( [ zet0 ] ) in the mean - field approximation we obtain @xmath81 where we have introduced the following auxiliary function @xmath82\right\}.\ ] ] from eqs .( [ a1b2 ] ) and ( [ alabala ] ) we get for the free energy @xmath83=-\frac{1 } { \sqrt{2\pi\delta}}\int_{-\infty}^{\infty}dy \exp\left(-\frac{1}{2}\frac{y^{2}}{\delta}\right ) \ln\left[(ul^d)^{-n/4}{\cal i}_{n}\left(\mu+y / u^{1/2}\right)\right].\ ] ] if we introduce a second scaling variable @xmath84 and using eqs .( [ nn ] ) ( see appendix ) , eq .( [ zbala ] ) takes its final form @xmath85= -\frac{1}{\sqrt{2\pi\lambda}}\int_{-\infty}^{\infty}dx e^{-(x-\mu)^{2}/(2\lambda ) } \ln\left[d_{-n/2}(\sqrt{3}x)\right]-\frac34(\lambda+\mu^2 ) + \frac n4\ln\left(\frac{ul^{d}}{12\pi^2}\right).\ ] ] for the ising case @xmath17 a similar expression for the quenched free energy in a slightly different context is obtained and its analytic structure is studied in references @xcite .obviously ( [ scba ] ) is well defined for any positive @xmath86 and in the limit @xmath87 we recover the well known result for the free energy of the pure model .in addition to the free energy , one also needs to know the correlation functions . within the replica method the averages of the fields @xmath88 are defined by ( see e.g. @xcite ) @xmath89,\ ] ] where @xmath90 note that the final result must be independent of replica index @xmath91 , because @xmath92 is invariant under permutation of the replicas . after taking the limit @xmath93, we end with the following expression : @xmath94 in a similar way @xmath95^{2 } e^{-(x-\mu)^{2}/2\lambda } .\ ] ] from eqs .( [ cum1 ] ) and ( [ cum2 ] ) , when @xmath96 and @xmath17 we obtain the results of ref .@xcite . in terms of the normalized magnetization @xmath97 the susceptibilityis given as @xmath98 another quantities of importance for numerical analysis is the binder cumulant defined by @xmath99 and the cumulant , specific for the random system defined as @xmath100 since the parameter @xmath101 is the relative variance of the observable ( the susceptibility ) , as we said in the introduction , it is a measure of the self - averaging in the random system .if self - averaging takes place this quantity should be zero in the thermodynamic limit .the perturbatively calculated parts of the free energy and cumulants , which contain contributions of all nonzero modes , depend , to one - loop order , on the shift of the critical temperature and on the renormalized coupling constants @xmath103 and @xmath104 . the application of the finite - size @xmath18-expansion to the model system ( [ glwcub ] ) requires the corresponding renormalization constants . to one loop order , using the minimal subtraction scheme , before taking the @xmath93 limit , we obtain : [ eqzzzg ] @xmath105 in eqs .( [ eqzzzg ] ) , [ eqzu ] @xmath106 the @xmath91 functions associated to @xmath107 and @xmath108 have the form [ bfc ] @xmath109 the fixed points of this system first have been studied in @xcite and for the purposes of the impurity problem in @xcite .the values of @xmath107 and @xmath108 in the fixed point , interesting in the random case , are [ zhfp ] @xmath110 the corresponding expression for the exponent @xmath111 up to the first order of @xmath18 , is : @xmath112 it should be noted here that we shall consider @xmath70-component fields with @xmath113 , where @xmath114 is the critical number of spin components that defines the stability of the random fixed point in the @xmath44 limit .the stability of the different fixed points of the model has been also considered in @xcite .the analysis of the ising case @xmath115 needs to perform a loop expansion to second order ( see introduction ) and is beyond the scope of the present study .as it was explained above , the loop corrections will be treated perturbatively on the nonzero @xmath34 modes . in the lowest order in @xmath18 ,this procedure generates a shift of the critical temperature @xmath116 ; @xmath117 where the term @xmath118 is coming from the one - loop counterterm ( see ( [ eqzzzg ] ) ) , and @xmath119\frac{1}{l^{d}}{\sum_k}'\frac{1}{\bbox k^{2}+t}\ ] ] is the finite - size correction .the two diagrammatic contributions for @xmath120 are shown in fig .both diagrams from the @xmath103 and the @xmath104 contributions differ only by their numerical factors .the prime in the @xmath121-fold sum in the above equations denotes that the term with a zero summation index has been omitted .after some algebra ( details for the pure case @xmath122 see in @xcite ) , near the upper critical dimension @xmath2 , we obtain @xmath123\left ( t\ln t + 4l^{-2}f_{4,2}(tl^{2})\right),\ ] ] where @xmath124,\ ] ] some particular values of the constant @xmath125 and a method of calculation are given in @xcite . at the fixed point @xmath126 , @xmath127 ,up to the first order in @xmath18 , the terms proportional to @xmath128 cancel and eq .( [ urtiq ] ) can be written in the following scaling form @xmath129\epsilon,\ ] ] where the scaling variable @xmath130 has been introduced . in the @xmath44 limit ,the expression for the exponent measuring the divergence of the correlation length is @xcite : @xmath131 instead of the critical exponent for the pure case @xmath132 the scaling form ( [ svqst ] ) can be written for this case as @xmath133\epsilon,\ ] ] where @xmath134 . from the above expressionone can obtain the large - l asymptotic form of the @xmath135 shift , i.e. @xmath136 in eq .( [ eqsh ] ) , @xmath137 denotes the average pseudo critical temperature ( @xmath138 is pseudo critical temperature for a specific random realization @xmath139 ) and @xmath140 .( [ eqsh ] ) was suggested in @xcite .combined with the fenomenological fss theory it gave rise to the lack of sa , and is confirmed by numerical studies ( see @xcite ) .here it is verified independently and directly .we perform the renormalization in a similar way also for @xmath141 and @xmath142 by taking into account the diagrammatic contributions , from @xmath103 and @xmath104 shown in fig .the result is @xmath143 where @xmath144 and @xmath145 are the one - loop counterterms for the coupling constants , and [ udre ]@xmath146 \frac{1}{l^{d}}{\sum_k}'\frac{1}{(k^{2}+t)^{2}}\\ \delta_{l}&= & -\left[u\delta\frac{n+2}{3}-\frac{8+nn}{2}\delta^{2}\right ] \frac{1}{l^{d}}{\sum_k}'\frac{1}{(k^{2}+t)^{2}},\end{aligned}\ ] ] are the corresponding finite - size corrections . as one can see the summand in eq .( [ udre ] ) can be expressed as the first derivative of the summand of eq .( [ eqnp ] ) with respect to @xmath147 .so , at the fixed point , algebraic transformations similar to those performed in the previous section lead to [ udsc ] @xmath148\\ \tilde{\delta}^{\star}(n)l^{\epsilon}&= & \delta^{\star}(n)\left[1+\frac{1}{2 } ( 1+\ln y)\epsilon + 2\epsilon f^{\prime}_{4,2}(y)\right],\end{aligned}\ ] ] where the prime indicates that we have derivative of the function @xmath149 with respect to its argument . the results for the disordered system simply follow by setting @xmath44 . from the results for the shift of the critical temperature ( [ svqst ] ) and the renormalization of the coupling constant @xmath103 , given by eq . ( [ udelta ] ), we reproduce the results for the pure fss case , by setting @xmath150 and @xmath44. moreover , this result still holds even if we find the fss corrections after the analytical continuation to @xmath44 , expressing the commutativity of the problem .let us consider the scaling variables @xmath151 at the fixed point they can be expressed in terms of scaling variable @xmath130 : @xmath152\epsilon \right.\nonumber\\ & & - \left.\frac{12n(n-1)}{16(n-1)-nn(n+8)}f_{4,2}(y)\epsilon - yf^{\prime}_{4,2}(y)\epsilon\right\}\end{aligned}\ ] ] and @xmath153 in the limit @xmath44 , eqs .( [ muzvg ] ) and ( [ lamzg ] ) yield the following scaling variables describing the disordered system ( [ eqglw ] ) : @xmath154\epsilon \right.\nonumber\\ & + & \left.\frac{3n}{4(n-1)}f_{4,2}(y)\epsilon - yf^{\prime}_{4,2}(y)\epsilon\right\},\end{aligned}\ ] ] where @xmath134 , and @xmath155 these equations verify the finite - size scaling hypotheses and show that we are really dealing with a one - variable problem , since the second variable @xmath156 is a fixed universal number . at the critical point @xmath157 , since the constant @xmath158 , see ref .@xcite , we have @xmath159 numerical values for different thermodynamic quantities can be obtained with the help of eqs .( [ lamz ] ) and ( [ hmu ] ) .note that the scaling variable @xmath160 is proportional to @xmath161 .consequently all the @xmath18-expansion results will be expressed in power of @xmath161 as it was the case for the pure systems ( see @xcite for example ) .@xcite , the binder cumulant @xmath162 and the the relative variance @xmath101 ( eqs .( [ bc ] ) and ( [ rnc ] ) ) are calculated analytically and numerically at the critical point @xmath163 in the asymptotic regime @xmath164 for @xmath17 and @xmath165 . in ref.@xcite ( see also @xcite ) , the same quantities were calculated numerically for @xmath17 and @xmath166 . in both cases the results showing that the system exhibits a lack of sa . in the remainder of this section we concentrate on the calculation of the cumulants @xmath162 and @xmath101 ( [ bc ] ) and ( [ rnc ] ) in cases @xmath167 and @xmath2 . the meaning to consider the case @xmath167 is in its simple analytical nonperturbative treatment .although the results based on the @xmath18 - expansion give only a qualitative description of the three dimensional physics , we hope that they shed some light at least on the applicability of the theory for studying diluted models .let us first note that if @xmath168 and @xmath2 , the case under consideration applies to the situation ( see eq .( [ nur ] ) ) where @xmath169 and randomness is relevant ( r - case ) .up to the first order in @xmath18 , due to the rg arguments , no sa must be expected near the critical point @xcite .this statement is supported also by our rg calculations . in table[ table1 ] and table [ table2 ] we present the corresponding universal numbers for b and r at @xmath170 and @xmath166 in the region @xmath171 , i.e. in the vicinity of the critical point .the calculations are performed with variable @xmath96 for @xmath172 and @xmath173 from eq .( [ hmu ] ) ( setting @xmath174 ) for @xmath166 , and with variable @xmath86 taken from eq .( [ lamz ] ) in both cases .the asymptotic behavior for small @xmath175 is presented in the appendix .the numerical values of b and r in the random case and for @xmath17 , presented in table [ table1 ] , are in full agreement with those obtained in ref .@xcite , while those of b for the pure case and @xmath17 ( table i and table [ table2 ] ) are in full agreement with ref .@xcite .the random case @xmath17 for @xmath176 can not be considered within the present expansion , because of the apparent divergence of @xmath160 that takes place to the used order in @xmath18 . up to nowthere are only numerical values @xmath177 , @xmath178 and @xmath179 obtained in @xcite through monte carlo simulations .what is possible to calculate here are the corresponding values of b and r very close to @xmath17 , e.g. for @xmath180 . for completenessthese results are presented in table ii .more general , one can see that if @xmath181 , then @xmath182 and @xmath183 , i.e. the system exhibits sa .this evident discrepancy with the reality is due to the wrong assumption that some information about the random case @xmath17 can be obtained from the above formulas in this limiting case .as it was pointed out the correct treatment of the case @xmath17 seems to be a more difficult computational problem .the finite size correction to the bulk critical behavior of the cumulants @xmath162 and @xmath101 in the region @xmath184 , i.e. away from the critical point , are obtained with the help of the asymptotics @xmath185 , given in the appendix ( eqs .[ am2 ] , [ am4 ] ) . according to the analysis presented there, we obtain for binder s cumulant @xmath186 + { \cal o}\left(\frac{1}{\mu^{3}}\right).\ ] ] for the cumulant @xmath101 we get @xmath187 the final results can be obtained by replacing @xmath86 and @xmath175 by their respective expressions evaluated at the fixed points of the model given in eqs .( [ muzv ] ) and ( [ lamz ] ) .so , to the lowest order in @xmath18 , we have @xmath188 and @xmath189 it is interesting to compare eq .( [ bdkt ] ) with the corresponding result for the pure system @xcite @xmath190 up to the lowest order in @xmath191 they coincide for @xmath168 , moreover for @xmath192 ( with @xmath193 ) , we have @xmath194 and @xmath195 . the result ( [ riz ] )confirms the statement @xcite that away from the critical point , a strong sa emerges in the system , as @xmath196 with @xmath197 .in the present paper we propose a general scheme for the fss scaling analysis of a finite disordered @xmath198 system .the method , we use here , is an extension of the field theoretical methods used to analyse fss properties in pure systems .the nature of the symmetry ( obtained as a consequence of the use of the replica trick , which removes the disorder ) of the model complicates the perturbative structure of the theory in comparison with the corresponding @xmath198 pure one .remind that , the final results for the disordered system are obtained by making the number of replicas vanishing .our results concern mainly systems with number of components larger than 1 i.e. non ising systems .their extension to ising systems requires higher loop calculations , because of the degeneracy of the one loop order rg equations .our main results are related to the formulation of the problem for some number of components @xmath70 of the fluctuating field for dimensions @xmath22 and @xmath2 . due to the presence of randomness, it is shown that we are dealing with two variables problem with scaling variables @xmath199 and @xmath200 . in the mean field regime@xmath22 our results are a generalization for @xmath21 of those obtained in @xcite for @xmath17 .evaluating numerically the corresponding analytic expressions for the binder s cumulant @xmath162 and the relative variance @xmath101 , we demonstrate a monotonic increase of @xmath162 as a function of @xmath70 in both pure and random cases and a monotonic decrease of @xmath101 ( to zero for @xmath201 ) in the random case ( see table [ table1 ] ) .the @xmath18-expansion to first order in @xmath18 , shows that close to the critical temperature , one can express the physical observables , the shift of the critical temperature and the renormalization of the coupling constants in terms of @xmath175 and @xmath86 .in the random fixed point the parameter @xmath86 takes an universal value ( see equation ( [ lamz ] ) ) .it is found that the distance , over which the bulk critical temperature is shifted , is proportional to @xmath202 , in agreement with the statement of ref .this result , combined with fenomenological fss , gives rise to the lack of sa .the scaling parameters @xmath175 and @xmath86 , also enter in the final expression for the binder cumulant ( [ bc ] ) and the relative variances ( [ rnc ] ) , giving explicit expressions for them in the different asymptotic regimes .the numerical calculation of the above parameters permits also the verification of the fss and sa ( see table [ table2 ] ) .the last is shown to be absent in the regime @xmath203 , where randomness are relevant and our analysis explicitly shows that in this regime the relative variance is always non - zero for the case of stability of the rg solutions , i.e. @xmath168 and @xmath2 .one can also try to repeat the analysis without the use of the replica - trick .for this aim , one needs to define in a proper way the procedure in the zero - mode approximation , when rescaling the random part of the hamiltonian for finite system .one can realize that this is the @xmath204 part of this term , which will give a contribution to the free - energy .this scheme permits formally to end with the same expressions for the shift of the critical temperature and the renormalization constants , as we did within the replica formalism . in our opinion, the present fss study can be also applied in the `` canonical '' case @xcite , where the disorder is characterized by a constant total number of the occupied sites ( or bonds ) , instead of the constant average density .we hope that results , similar to the bulk case , will also hold in the case of finite geometry , relating in this way our theoretical findings with the monte - carlo simulations .the authors aknowledge the hospitality of the abdus salam international centre for theoretical physics , trieste , where part of this work was written .e.k . and n.s.t .acknowledge also the financial support of associate & federation schemes of ictp .is also supported by the spanish dges contract no .pb97 - 0076 .in this appendix we present the mathematical details of how to obtain the asymptotic of the averages : where we have introduced the function @xmath207\right\}.\ ] ] the integral in the definition of function @xmath208 given by eq .( [ fi ] ) , may be evaluated in terms of parabolic cylinder functions @xmath209 using the identity @xcite @xmath210 the result is : @xmath211 _ suppose @xmath217 , @xmath218 and @xmath219 an analytic function in a neighborhood of @xmath220 , @xmath221 and that @xmath222,\ ] ] for positive constants @xmath223 , @xmath224 .then : @xmath225 as @xmath226 in the sector @xmath227. _ using the asymptotic of @xmath231 and @xmath232 for large @xmath175 , we can get the behavior of the cumulants @xmath101 and @xmath162 in both cases @xmath172 and @xmath2 .they are given by @xmath233 + { \cal o}\left(\frac{1}{\mu^{3}}\right).\ ] ] for the cumulant @xmath101 we get @xmath234 99 s. wiseman and e. domany , phys . rev .* e52 * ( 1995 ) 3469 .a. aharony and a.b .harris , phys .* 77 * ( 1996 ) 3700 .f. pzmndi , r.t .scalettor and g.t .zimnyi , phys .* 79 * ( 1997 ) 5130 .s. wiseman and e. domany , phys . rev .* e81 * ( 1998 ) 2938 .a. aharony , a.b .harris and s. wiseman , phys .* 81 * ( 1998 ) 252 . h.g .ballesteros , l.a .fernndez , v. martin - mayor , a.m. sudupe , g. parisi and j.j .ruiz - lorenzo , nucl .phys . * b512 * [ fs ] ( 1998 ) 681 .marques and j.a .gonzalo , phys . rev .* e60 * ( 1999 ) 2394 .k. bernardet , f. pzmndi and g.g .batruini , phys .* 84 * ( 2000 ) 4477 .m. barber in:_phase transitions and critical phenomena _ , eds . c. domb and j.l .lebowitz , academic , london , 1983 , vol.8 .v. privman , ed ., _ finite - size scaling and numerical simulations of statistical systems _ ,world scientific , singapore , 1990 .brankov , d.m .danchev and n.s .theory of critical phenomena in finite - size systems ; scaling and quantum effects _ , world scientific , singapore , 2000. k. binder and a.p .young , rev .* 58 * ( 1986 ) 801 .e. brzin and j. zinn - justin , nucl . phys . * b257 * [ fs14 ] ( 1985 ) 867 .j. rudnick , h. guo and d. jasnow , j. stat .* 41 * ( 1985 ) 751 .khemelnitskii , zh .* 68 * ( 1975 ) 1960 [ sov .jetp * 41 * ( 1975 ) 981 ] .t. lubensky , phys .* b11 * ( 1975 ) 3575 . g. grinstein and a. luther , phys .rev . * b13 * ( 1976 ) 1329 . c. jayaprakash and h.j .katz , phys . rev .* b16 * ( 1977 ) 3987 .shalaev , zh .* 73 * ( 1977 ) 2301 [ sovjetp * 46 * ( 1977 ) 1244 ] .g. jug , phys . rev . *b27 * ( 1983 ) 609 .e. korutcheva and d. uzunov , phys .lett * a106 * ( 1984 ) 175 .m. mzard , g. parisi and m .- a .virasoro , _ spin glass theory and beyond _ , singapore , world scientific , 1987 .a. pelissetto and e. vicari , phys .* b62 * ( 2000 ) 6393 .the technique that involves application of the rg transformations to the disordered system directly @xcite gives the same fixed points as the technique that involves first removing the disorder by averaging and then employing the rg transformation @xcite .for a direct nonperturbative derivation of this equivalence see @xcite .the indices `` r '' and `` p '' will refer to the `` random '' and the effective `` pure '' system .e. brzin , j. phys .* 43 * , 15 ( 1982 ) .a.j . bray , t. mccarthy , m.a .moore , j.d .reger and a.p .young , phys . rev .* b36 * ( 1987 ) 2212 .mckane , phys .* b49 * ( 1994 ) 12003 .g. lvarez , v. martin - mayor and j.j .ruiz - lorenzo , j. phys .gen . * 33 * ( 2000 ) 841 .e. brzin , j.c .le guillou and j. zinn - justin , phys . rev .* b10 * ( 1974 ) 892 .k. debell and d.j.w .geldart , phys . rev . * b32 * ( 1985 ) 4763 .h. chamati and n.s .tonchev , j. phys . a : math33 * ( 2000 ) l167 .ballesteros , l.a .fernndez , v. martin - mayor , a.m. sudupe , g. parisi and j.j .ruiz - lorenzo , phys .* b58 * ( 1998 ) 2740 .h. chamati and d.m .danchev , unpublished .i.s . gradsteyn and i. m. ryzhik , academic , new york and london , 1965 .fedoruk , _ asymptotics : integrals and series _ , nauka , moscow , 1987 .emery , phys . rev .* b11 * ( 1975 ) 239 .

<S> the critical behavior of a quenched random hypercubic sample of linear size @xmath0 is considered , within the `` random-@xmath1 '' field - theoretical model , by using the renormalization group method . </S><S> a finite - size scaling behavior is established and analyzed near the upper critical dimension @xmath2 and some universal results are obtained . </S><S> the problem of self - averaging is clarified for different critical regimes . </S>

1610.06313

for wide - field adaptive optics ( wfao ) systems , the turbulence is compensated in a large field of view ( fov ) using a tomographic reconstruction of the phase in the entire volume ( @xcite ) or in discrete directions ( @xcite ) , from wave - front sensor ( wfs ) measurements distributed over the system fov .knowledge of the statistical turbulence parameters is paramount , with several methods being dedicated to estimating them as slodar ( @xcite ) , mass ( @xcite ) and scidar ( @xcite ) techniques . as a moao on - sky demonstrator ,canary has been observing at the william herschel telescope ( wht ) since 2010 ( @xcite ) . since an open - loop configurationis adopted , the lack of feedback between the deformable mirror ( dm ) and wfs makes the control more sensitive to inaccuracies on the command matrix calibration .the recursive nature of a closed - loop tends to wash out these effects , which clearly is not the case here .vidal et al has proposed in @xcite the the learn & apply ( l&a ) technique that combines both turbulence profiling and system mis - registration identification / calibration .the efficiency of this method has been demonstrated on - sky with canary ( @xcite ) .the l&a consists in model - fitting the spatial covariance matrix of wfs measurements .the fitting is ensured thanks to the levenberg - marquardt iterative algorithm which estimates a number of model - parameters , as the @xmath0 profile , the outer scale profile and the stars position . in view of calibration ,the primary model proposed by vidal @xcite included system mis - registration : pupil shifts , rotation and magnification between wfs .we used in the past a model @xcite based on a fourier transformation of the von - krmn spectrum to get the bi - dimensional ( 2d ) spatial covariance of the wave - front slopes .the algorithm was particularly efficient when coupled to wave - front sensors with fixed sub - aperture sizes , because the computation using fast fourier transforms ( fft ) could be used to extract in one shot from the same 2d map all the covariance values between two wave - front sensors with arbitrary separation , without requiring any re - interpolation of the map -provided the sub - aperture size is a multiple of the fft sampling .now , the cone effect inherent to lgs induces layer - projected sub - aperture sizes and pupil sizes to vary with the altitude , which breaks the efficiency of the implementation when it s a matter of computing covariances between lgs and ngs .at least , an interpolation procedure could be implemented in order to access values that lay between covariance samples of the fft , but unfortunately this kind of approximation does not go well with a fitting procedure that internally requires to compute derivatives as finite differences of the covariance function .we propose thus in sect .[ s1 ] a fast - computable model that includes both turbulence characteristics and system mis - registration . to speed up the fitting algorithm, we propose in sect.[s : l3s ] an upgrade to the l&a , called the _ learn 3 steps ( l3s ) _ , that splits the identification of the ground layer parameters from those of altitude layers .the ground layer is known to dominate the turbulence strength with relatively slow time - variations ( @xcite ) .it appears thus that a large time - span of accumulated data would be required to obtain statistical convergence on the generation of the covariance matrices of phase ( @xcite ) , the latter being an input to the tomography . in splitting the ground and altitude parts, we mitigate the mis - convergence of statistics obtained from system telemetry .in addition , we have included in the l3s a separate estimation of the wind - speed profile . from the turbulence profile ,we reconstruct the phase at each retrieved altitude using an mmse approach .the wind - speeds on those layers is then determined from the full width at half maximum ( fwhm ) of the average temporal auto - correlation function of slopes .we present then in sect .[ s : stats ] a characterization of the turbulence along altitude above the wht from the canary on - sky data acquired in september 2013 .we have processed about 2,000 data sets with the l3s algorithm to present plots of distribution of the seeing , outer scale and wind - speed .we have also characterized the median values for the ground layer ( below 1.5 km ) the altitude layers ( between 1.5 and 20 km ) and the very high altitude layers ( above 20 km ) inaccessible to tomography with canary .we conclude in sect .[ s : conclusions ] .[ f1 ] lp6.5 cm variable & description + @xmath2 & coordinates of the sub - aperture @xmath3 in the pupil plane + @xmath4 & angular position of @xmath5 stars from the center of the fov + @xmath6 & altitude of the @xmath7 turbulent layer + @xmath8 & coordinates of the @xmath9 sub - aperture of the @xmath5 wfs at the altitude @xmath10 + @xmath11 & component of @xmath8 over the x - axis + @xmath12 & component of @xmath8 over the y - axis + @xmath13 & focus altitude of the @xmath5 wfs . + @xmath14 & size of the @xmath9 sub - aperture in the pupil plane + @xmath15 & pupil shifts for the @xmath5 wfs + @xmath16 & rotation shift for the @xmath5 wfs + @xmath17 & magnification for the @xmath5 wfs + [ t:2 ] in this proceeding , we review the calculations from @xcite in a more general way , in including system mis - registration .consider the @xmath9 sub - aperture of the @xmath5 wfs . in the pupil plane , we note @xmath2 the position of the @xmath9 sub - aperture which does not depend on the actual wfs since all are conjugated at the pupil plane .we include dm / wfs mis - registrations : for the @xmath5 wfs , we define @xmath18 and @xmath19 as the x and y pupil shifts , @xmath16 as the rotation and @xmath17 as the magnification .we note @xmath20 as the new @xmath9 sub - aperture s coordinates in the pupil : @xmath21 \times { \left[m_p\times { \mathbf{r}_{i}}\right ] } + \delta x_p + \delta y_p\ ] ] we note @xmath4 as the angular position of the @xmath5 wfs from the center of the fov .we will finally note @xmath13 the altitude focus of guide stars and @xmath8 as the coordinate of this sub - aperture , at the @xmath22 layer , in the pupil plane .we have : @xmath23 for the following theoretical developments , we define @xmath11 and @xmath12 the projection of @xmath8 on the reference pupil plane : @xmath24 where @xmath25 and @xmath26 are forming an orthonormal basis along directions @xmath27 and @xmath28 . for a shack - hartmann wfs, the 2d slopes map is obtained from the average phase gradient over the lenslet : @xmath29 where @xmath14 is the @xmath9 sub - aperture size , i.e. the edge to edge distance in the pupil plane considering all sub - apertures are squared , and @xmath30 is the square integration domain comprised between @xmath31 and @xmath32 in @xmath27 and @xmath28 directions .the integration variables @xmath33 @xmath34 are related to the @xmath9 sub - aperture and are given by @xmath35 and @xmath36 .the x - axis and y - axis slopes are the scalar product between @xmath37 and respectively @xmath25 and @xmath26 . in projecting the slopes alongx or y directions , the integral on sub - aperture surface in eq .[ e : s_dipl ] becomes integrals along sub - aperture side : @xmath38 we then define the spatial covariance @xmath39 , @xmath40 and @xmath41 as following : @xmath42 where @xmath43 is the separation between the @xmath9 sub - aperture of @xmath5 wfs and the @xmath44 sub - aperture of @xmath45 wfs , at altitude @xmath6 .we develop here the method to get @xmath46 only ; both @xmath47 and @xmath48 are similarly derived . noting the averaged gradient is the difference of phase along each side of the sub - aperture, we then get : @xmath49 we now use the identity : @xmath50 which allows to rewrite eq . [e : sxsx ] as follows : @xmath51 considering @xmath52 as the phase structure function ( sf ) , we have the final expression of the x - axis slopes covariance : @xmath53 where @xmath54 is a constant value to include the telescope tracking error and vibrations we assume to provide an additional isoplanatic tip - tilt to the turbulence .we now consider @xmath55 as the von - krmn spatial power spectral density ( psd ) given by : @xmath56 the phase structure function ( sf ) is related to the spatial psd @xmath55 by : @xmath57 where the phase sf following the von - krmn statistics is : @xmath58}^{5/6 } { \left(\dfrac{{l_0}({h_{l}})}{{r_0}({h_{l}})}\right)}^{5/3 } { \left[\frac{\gamma(5/6)}{2^{1/6}}- { \boldsymbol{\rho}}^{5/6 } k_{5/6}({\boldsymbol{\rho}})\right]},\ ] ] with @xmath59 . in the same way, we get : @xmath60 and @xmath61 to retrieve the solution , the integrals need be discretized along sub - apertures sides . in ( @xcite ) , this issue is solved using a mid - point version of the hudgin model : the phase gradient is calculated in taking only the mid - point edge - to - edge phase difference .in such a case , instead of having the integrand variables @xmath33 and @xmath34 ranging from @xmath31 to @xmath32 , they are evaluated at a single point , equating the integral to zero . inso doing we find again the analytic expression in ( @xcite ) .this approach allows to drastically reduce the computation time of analytic covariance matrices , few seconds with 8 7@xmath627 sh wfs on canary .the cost is an overestimation of 15% on the covariance peak ( @xcite ) , but the wings of the spatial covariance are still well reproduced .overall the covariance model depends on the following parameters : * @xmath63 : number of discrete layers .* @xmath6 : altitude layers . * @xmath64 : @xmath65 value for the discrete altitude at altitude @xmath6 * @xmath66 : @xmath67 value for the discrete altitude at altitude @xmath6 * @xmath68 , @xmath69 , @xmath16 and @xmath17 : system mis - registration .* @xmath54 , @xmath70 and @xmath71 : constant value on the covariance map for taking into account telescope tracking error and vibrations .the covariance of the aliased phase can be derived using eqs .[ e : gxx ] , [ e : gyy ] and [ e : gxy ] .these equations involve an expression of the phase sf which depends on the turbulence parameters and the system geometry [ e : dphi ] . to get the aliasing contribution in the covariance, we have to split the phase into a parallel part , i.e. compensated by the system , and an orthogonal part : @xmath72 the orthogonal phase contains only spatial frequencies higher than @xmath73 , the dm cut - off frequency . in the case of a fried geometry ( most frequent configuration using sh wfs ) , the wfs sub - aperture spacing is equal to the dm actuators pitch .considering then @xmath74 , the aliasing sf @xmath75 becomes @xmath76 using eq .[ e : dphialias ] to replace the sf expression in eqs .[ e : gxx ] , [ e : gxy ] and [ e : gxy ] , we get an analytic expression of the spatial covariance of the aliasing .for determining the model parameters listed in sect .[ ss : model ] , we use an extension of the learn&apply algorithm .this extension is called the _ learn 3 steps _ and operates similarly to the latter .+ the main idea of the learn technique is to least - squares fit the empirical covariance matrix from telemetry to the theoretical one . to mitigate the presence of noise we exclude the diagonal of these matrices during the minimization , for which we employ an iterative levenberg - marquardt algorithm ( lma ) .this method has been demonstrated to be powerful for performing the tomography on canary ( @xcite ) .+ with the l3s we try to reduce the sensitivity of the model - fitting procedure to the lack of convergence we have on the empirical covariance using short time - series of data .we aim to reach the same tomographic error on a shorter time series than we would with the retrieved parameters of the original method . from the following observation ( @xcite ) : the convergence effect is stronger when the turbulence gets a larger seeing and a slower wind , we take it that the ground layer is the one which contributes the most to this effect .we propose to retrieve it independently from the altitude layers .we report in fig .[ f : l3steps ] a diagram of the l3s algorithm : * * step 1 : * we start by removing the tip - tilt mode in measurement and model for each wfs. then we remove the mean slopes between all the wfs corresponding to the ground layer ( gl ) contribution in both measurement and model . in doingthat we suppress the gl contribution of the covariance matrices in order to focus on the altitude layer identification . by applying the lma on such matrices, we get an estimation of @xmath64 and @xmath77 in altitude . ** step 2 : * we get back to the initial values of covariance matrices , always tip - tilt removed , in keeping the same parameters in altitude retrieved during the previous step .the lma gives then an estimation of the ground layer seeing and outer scale .* * step 3 : * finally , we add back the tip - tilt components in the measurements and model covariance matrices .we then retrieve the telescope tracking error and vibrations coefficients @xmath78 and @xmath71 .they simulate an isoplanatic tip - tilt added to the turbulent one . during this third step , we have also the possibility to retrieve the system mis - registration @xmath68 , @xmath69 , @xmath17 and @xmath16 .x=1cm , y=1cm,>=stealth ( 0,0 ) rectangle ( 4,3.1 ) ; ( 0.1,3.1 ) node[below right , text width = 4 cm , color = blue ] * ngs / lgs cov. mat . * : + ngss slopes + lgss slopes + stars positions @xmath79 + lgss altitude @xmath13 + prior on @xmath63 + ; ( 4,2 ) ( 5,2 ) ; ( 5,2.75 ) rectangle node[text width = 1cm , text centered]rm .tt ( 6,1.25 ) ; ( 6,2 ) ( 7,2 ) ; ( 7,2.75 ) rectangle node[text width = 1cm , text centered]rm .mean slopes ( 8,1.25 ) ; ( 4,-1 ) ( 4.5,-1 ) ( 4.5,1.5 ) ( 5,1.5 ) ; ( 6,1.5 ) ( 7,1.5 ) ; ( 0,-.5 ) rectangle ( 4,-3.5 ) ; ( 0.1,-.5 ) node[below right , text width = 4 cm , color = red ] * cov . mat model * : + @xmath80,@xmath66 + @xmath78 and @xmath71 + @xmath81 + ; ( 8,2 ) ( 8.5,2 ) ( 8.5,0.75 ) ( 7.25,0.75 ) ( 7.25,0.5 ) ; ( 8,1.5 ) ( 8.25,1.5 ) ( 8.25,0.85 ) ( 7,0.85 ) ( 7,0.5 ) ; ( 5.25,1.25 ) ( 5.25,-.5);(5.5,1.25 ) ( 5.5,-.5 ) ; ( 6,0 ) ( 6,-.5);(6,-1 ) ( 6,-1.5 ) ; ( 4,-1.85 ) ( 5,-1.85 ) ; ( 4,.5 ) ( 4.25,.5 ) ( 4.25,-1.65 ) ( 5,-1.65 ) ; ( 5,0 ) rectangle nodelma ( 7.5,.5 ) ; ( 5,-1 ) rectangle nodelma ( 7.5,-.5 ) ; ( 5,-2 ) rectangle nodelma(7.5,-1.5 ) ; ( 8.75,-4 ) ( 8.75,4 ) ; ( 10.25,3.75 ) node*outputs * ; ( 6.25,3.75 ) node*learn 3 steps * ; ( 1.75,3.75 ) node*intputs * ; ( 4.25,-4 ) ( 4.25,4 ) ; ( 7.5,.25 ) ( 9,.25);(9,0 ) rectangle node[text width = 2.5cm , color = black!60!green]*step 1 * : + @xmath82 , @xmath83(12,1.5 ) ; ( 7.5,-.75 ) ( 9,-.75);(9,-1 . )rectangle node[text width=2.5cm , color = black!60!green]*step 2 * : @xmath84 ( 12,0 ) ; ( 7.5,-1.75 ) ( 9,-1.75);(9,-2.5 ) rectangle node[text width=2.5cm , color = black!60!green]*step 3 * : + @xmath78 and @xmath71 + @xmath81 + ( 12,-1 . ) ; ( 4.4,3.1 ) rectangle ( 8.6,-3.5 ) ; we compare the l3s approach with the initial l&a approach proposed by vidal ( @xcite ) . for only considering the gain on spitting the identification of relevant parameters in three steps ,we use exactly the same covariance matrix model in the both cases .we call thus _ learn 1 step _ ( l1s ) the version of the l3s where we retrieve all the parameters during a single step. we have selected three data sets acquired by canary in 2013 , comprising 10,000 frames sampled at 150 hz .data sets are the concatenation of slopes measured by three off - axis ngs wfs in a 2 field , one one - axis wfs and four rayleigh lgs wfs .these later were focused at 21 km and positioned at 23 off - axis in a square configuration ( @xcite ) + we report in tab .[ t : glbparam ] the observation conditions we got for these three cases . to be more accurate on the estimation of the turbulence parameters during the observation , all of these values have been evaluated on the entire 10,000 frames .+ . [ cols="^,^,^,^",options="header " , ] we have split each of these data sets into two parts : the 5,000 first frames form the _ calibration set _ while the 5,000 last form the _ test set_. we select @xmath85 frames in the calibration set to compute the empirical covariance matrix . in using either the l&a or l3s method ,we retrieve two different sets of model parameters .then , from the model - fitted covariance matrices , we derive a mmse reconstructor @xmath86 and then a wave front ( wf ) _ tomographic error _ in nm rms ( @xcite ) , in applying this reconstructor on the test set .we redo these operations for different values of @xmath85 in order to get the evolution of the tomographic error as function of @xmath85 .+ in fig .[ f : compar ] , we show the tomographic error versus @xmath85 for both l1s and l3s , for the three different sets .in addition in fig .[ f : compar ] , we compare turbulent profiles retrieved with @xmath87 frames for both methods .we get an improvement of the tomographic error on the three cases following two criteria : * * better robustness to the lack of convergence of empirical covariance :* tomographic errors with the l3s method are decreasing faster than the l1s ones .it means we reach the same level of accuracy on the turbulence parameters identification in using around five times less frames in nominal conditions . * * better accuracy :* as it appears in fig .[ f : compar ] , tomographic error plateaus when the time - series length is large enough .the final value of tomographic error we get with the l3s is smaller than the one given by the l1s method . for a time - series of 5,000 frames , we get an improvement on the quadratic error of 50% , 30% and 12% for sets 1 , 2 and 3 respectively .this gain is thus increasing with the seeing . indeed , for worse observation conditions than nominal ones , the mis - convergence of turbulence statistic parameters impacts more the l1s parameter retrieval .+ the same is confirmed by the turbulent profiles retrieved by the two approaches : for data set 1 , the seeing is worse than nominal conditions ( median of 0.7 " ) but the l&a has retrieved weak altitude layers . in comparing with the l3s results , they are fake layers coming from a lack of convergence of the lma which did not reach the optimal solution. it would be necessary to get many more frames to make the l1s algorithm converging towards this solution , which is in that case a very strong ground layer only .the l3s allows to identify this ground layer only from few thousands frames only .the temporal properties of the turbulence are required for data analysis , lqg control ( @xcite ) and psf reconstruction ( see paper 9909 - 64 _ psf reconstruction validated using on - sky canary data in moao mode _ in that conference ) . in that last case , considering we are seek to reconstructing a long - exposure psf , the knowledge of wind direction is no longer required .the long - exposure psf results from an average over all directions .we thus only need to estimate the wind - speed profile .we discuss here another approach than ones developed from external profiler as scidar ( @xcite ) or slodar ( @xcite ) based on a cross spatio - temporal covariance fitting .it is based on the temporal auto - covariance function of slopes acquired on ngs wfs in loop disengaged .such a technique has been also used on raven , the subaru telescope technical and science moao demonstrator ( @xcite ) .we start by reconstructing the wfs slopes layer - by - layer , using the knowledge of the vertical distribution of the turbulence we learn previously fron the l3s approach .then , we determine the fwhm of the temporal auto - correlation function of slopes , which is related to the wind speed according a calibrated relation .we note @xmath88 as the @xmath9 measurements provided by the @xmath5 wfs .it is a @xmath85-size vector where @xmath85 is the time - series length .the temporal auto - covariance function of @xmath88 is @xmath89 and comes from a two - dimensional fourier transform : @xmath90 we then average the @xmath91 functions over @xmath27 and @xmath28-axis measurements and ngs wfs to get the @xmath85-sized vector @xmath92 : @xmath93 where @xmath91 is denoised by the noise variance @xmath94 estimated by interpolating the auto - covariance peak .finally , we evaluate @xmath95 as the fwhm of @xmath96 .the wind - speed depends on the ratio @xmath97 where @xmath98 is the sub - aperture size in the pupil plane , actually 0.6 m for canary in phase b. our purpose is now to calibrate the relation between the real wind - speed on - sky assuming a taylor - like turbulence and the ratio @xmath97 . using the end - to - end simulation tool yao ( https://github.com/frigaut/yao.git ) ,we have generated phase screens with nominal seeing ( @xmath99 cm ) and outer scale ( @xmath100 m ) with different wind - speed values and wind direction comprised between 0 and 2@xmath101 .we use our routine to determine the ratio @xmath97 and compare this value to the set wind - speed in yao to get the factor to be applied to @xmath97 . in order to get as close as possible to the on - sky conditions ,we have created phase screens of size 50 m times 2048 frames(our acquisition time on - sky ) .we have performed this comparison for wind - speed spanning from 0 to 10 m.s@xmath1 , which corresponds to the range of values we have to deal with on - sky using canary . at each velocity value, yao has generated eight different phase screens from which we acquire system telemetry without noise .we expect to include the convergence effect of measurements into our calibration . using a proportional regression of the ratio @xmath97 versus @xmath102, we get the following calibration : @xmath103 we report in fig .[ f : velcal ] the wind - speed estimated using eq .[ e : vcal ] versus the set value in yao .we consider this calibration as acceptable for our next psf - r purpose . to get the wind - speed profile, we perform a mmse reconstruction of wfs slopes at a given layer .we need thus two covariance matrices to derive the reconstructor @xmath104 to get the phase at an altitude @xmath105 : @xmath106 where @xmath107 is the covariance matrix of all slopes , coming from both ngs / lgs wfs , for a sole turbulent layer at the altitude @xmath6 .it is computed analytically using the model presented in sect .[ ss : model ] .we then estimate the @xmath22 turbulent layer contribution to the wfs measurements . @xmath108 using the wind - speed estimation routine in sect .[ ss : fwhm ] on @xmath109 , we estimate the wind - speed at altitude @xmath6 .with the l3s method ( see sect . [ s1 ] ) and our wind - speed profile retrieval ( see . sect .[ s : windspeed ] ) , we have processed about 2,000 data sets acquired by canary in may 23 , 24 and 25 , in july 17,18,19 , 22 , 23 and 24 and in september 13 , 15 , 16 , 17 and 18 of september 2013 . see sect .[ ss33 ] and ( @xcite ) for more information about the canary design . on each data set, we have retrieved the seeing , outer scale and wind - speed profiles over five layers .the left panels of fig .[ f : cnhprof ] show the distribution of the turbulence along the altitude as a kind of a probability density .each point is a retrieved layer . using the auto - correlation of slopes , it is possible to get an estimation of the seeing above 20 km , the maximal altitude we can reconstruct by tomography with canary .we have gathered the seeing of such very high altitude layers into the 20 km plateau in fig .[ f : cnhprof ] . in addition, we report on the right panels of fig .[ f : cnhprof ] histograms of seeing , outer scale and wind - speed .we have split the turbulence into a ground part , altitude below 1 km , and an altitude part . for each retrieved turbulent profile ,we have then averaged the ground and altitude values to analyze the turbulence behavior , differently if it comes from the ground or the altitude . to do that, we define the _ effective outer scale _ as : @xmath110 and the _ coherent wind - speed _ , coming from the coherent time of the turbulence , as : @xmath111 where the summation is done over all layers or altitude layers ( above 1 km ) or ground layers only .[ e : l0eff ] and eq .[ e : veff ] gives integrated values of outer scale and wind speed in averaging the profile along layers in a same way than the @xmath65 computation ( @xcite ) .the @xmath112 is retrieved for any layers using the l3s approach .we report in tab .[ t : statturbu ] average values for seeing , layers contribution , effective outer scale and coherent wind - speed . ' '' '' + global seeing & 0.699 `` & 0.35 '' ' '' '' + ground seeing ( [email protected] km ) & 0.552 `` & 0.32 '' ' '' '' + altitude seeing ( 1.5 km@xmath114h@xmath11320 km ) & 0.263 `` & 0.23 '' ' '' '' + high altitude seeing ( h@xmath11420 km , appears in 16% of processed cases ) & 0.187 `` & 0.13 '' ' '' '' + ground contribution & 76.8% & 17% ' '' '' + altitude contribution & 22.2% & 16.4% ' '' '' + high altitude contribution & 7.46% & 14.9% ' '' '' + global @xmath115 & 8.9 m & 8.4 m ' '' '' + ground @xmath115 & 9.2 m & 7.9 m ' '' '' + altitude @xmath115 & 10.3 m & 9.7 m ' '' '' + high altitude @xmath115 & 12.9 m & 13.1 m ' '' '' + global @xmath116 & 3.07 m.s@xmath1 & 1.9 m.s@xmath1 ' '' '' + ground @xmath116 & 2.89 m.s@xmath1 & 1.9 m.s@xmath1 ' '' '' + altitude @xmath116 & 5.22 m.s@xmath1 & 2.78 m.s@xmath1 ' '' '' + as expected , the ground layer is the most dominant one with a median value of 76.8% of the total turbulence , while altitude layers , below 20 km represent about 22% of the turbulence .we get on average a ground seeing of 0.67 " , which is close to the reported value at the wht ( http://www.ing.iac.es/astronomy/telescopes/wht/ ) .according to histograms in fig .[ f : cnhprof ] , the ground layer also dominates the seeing variability .altitude layers below 20 km reach an average seeing of 0.26 " and stay relatively constant in strength with time .we also notice negative altitude layers .they may appear between the wht primary and secondary mirror , but we can not determine if they really exist or if this is a lma artifact . note the very altitude layers , above 20 km and too high for being reconstructed by tomography with canary , reach 7.5% of the total turbulence in median value , with 0.187 " of seeing .however this only occurs in 16% of cases .it means this phenomena is not so frequent , but significant in terms of seeing .[ f : cnhprof ] displays the vertical outer scale distribution .the relevance of the outer scale for tomography and psf reconstruction is one of the actual preoccupation ( see .paper 9909 - 16 _ review on ao real - time turbulence estimation _ in that conference ) .the plot is less dense than the ones from the seeing and wind - speed distribution since 55% of samples have reached more than 100 m of outer scale .this is a software breakdown in the l3s code : above 100 m of outer scale , we no longer consider these is any physical meaning . from the outer scale distribution ,we draw two other observations : firstly , in more than 55% the l3s ( and generally the l&a approach ) retrieves an outer scale larger than 100 m. this is a software breakdown we have implemented into the spatial covariance model : if the lma reaches this limit , we stick the value to 100 m , we consider there is no more physical meaning on that . inexcluding those 55% of cases , we have 6% of cases that reach more than 20 m , 2.7% with more than 30 m and only 1% of cases with an outer scale greater than 50 m. the question now is why in 55% of cases is the l&a approach retrieved a fake value of the outer scale ? a main reason is we are observing with a 4.2 m telescope and we are not sensitive to large values of outer scale .but now , we can wonder whether we are reaching the software breakdown of 100 m because the aperture - size of the telescope or because any other reason to be determined using end - to - end simulation .+ secondly , this distribution of @xmath117 is quite constant around twice the telescope diameter and is constrained by the ground layer .[ t : statturbu ] shows @xmath118 becomes larger in altitude compared to the ground value .the conclusion on that is the estimation of the outer scale we have is biased by the finite aperture size from which we observe the turbulence . in altitude , because the stars angular separations , the meta - pupil size is increasing with the altitude .it is exactly like the aperture - size through which we observe the turbulence is increasing as well , as the outer scale in altitude looks to do according to fig . [f : cnhprof ] . from those two observations ,we conclude the estimation of the outer scale we are doing on a 4.2 m telescope is not directly related to a physical quantity .the problem we have now is to determine whether taking into account the entire outer scale profile is relevant for tomography and psf - reconstruction on canary and demonstrate and understand why , using end - to - end simulations , the outer scale , as we have defined in eq .[ e : l0eff ] , is twice the meta - pupil size .figure [ f : cnhprof ] illustrates the wind - speed profile as well .it appears the ground turbulence is the slowest one , with 2.89 m.s@xmath1 , against 5.22 m.s@xmath1 in altitude .we get a very strong variability of wind - speed values , especially in altitude , it makes thus the l3s approach more valuable .+ finally , the distributions of seeing , outer scale and wind - speed in fig .[ f : cnhprof ] are denser around particular altitudes .these are the most contributing layers to the turbulence and are located at the ground , around 2 km , 6 km , 10 km and 16 km .the tomographic resolution we have on canary allows to retrieve on average five turbulent layers .we have discussed on the l3s approach , an upgrade of the learn & apply algorithm , which dissociates the identification of the altitude layers from the ground layer . under nominal observation conditions , the parameters it estimates lead to the same level of tomographic error than the l&a approach in using five time less temporal frames than the latter .moreover , we can gain up to 30% on the tomographic error using the l3s .the l3s technique has been applied over a large set of canary data to characterize the turbulence above the william herschel telescope ( wht ) : we find 0.67``/8.9m/3.07m.s@xmath1 of total seeing / outer scale / wind - speed against 0.55''/9.2m/2.89m.s@xmath1 below 1.5 km and 0.263``/10.3m/5.22m.s@xmath1 between 1.5 and 20 km .we have also determined that the high altitude layers above 20 km , that can not be compensated by canary , appear only 16% of time with a median seeing of 0.187 '' .we have observed the outer scale values we can measure with canary are limited by around twice the meta - pupil size at a given altitude . the research leading to these results received the support of the a*midex project ( no .anr-11-idex-0001 - 02 ) funded by the investissements davenir french government program , managed by the french national research agency ( anr ) .this work is supported by cnrs , insu , observatoire de paris , universit paris diderot - paris 7 and the european commission ( fp7 infrastructures 2012 - 1 , opticon grant 312430 , wp1 ) . ,e. , vidal , f. , brangier , m. , nd morris , t. , hubert , z. , basden , a. , rousset , g. , myers , r. , chemla , f. , longmore , a. , butterley , t. , dipper , n. , dunlop , c. , geng , d. , gratadour , d. , henry , d. , laporte , p. , looker , n. , perret , d. , sevin , a. , talbot , g. and younger , e. , _ moao first on - sky demonstration with canary _ , astron . & astrophys . , 2011 , ,e. , morel , c. , osborn , j. , martin , o. , gratadour , d. , vidal , f. , le louarn , m. and rousset , g. , _ robustness of tomographic reconstructors versus real atmospheric profiles in the elt perspective _ , proc .spie , 2014 ., _ on - sky validation of an optimal lqg control with vibration mitigation : from the canary multi - object adaptive optics demonstrator to the gemini multi - conjugated adaptive optics facility ._ , american astronomical society meeting abstracts , 2015 ,

<S> canary is the multi - object adaptive optics ( moao ) pathfinder for the future moao - assisted integral - field units ( ifu ) proposed for extremely large telescopes ( elt ) . </S><S> the moao concept relies on tomographically reconstructing the turbulence using multiple measurements along different lines of sight . </S><S> tomography requires the knowledge of the statistical turbulence parameters , commonly recovered from the system telemetry using a dedicated profiling technique . for demonstration purposes with the moao pathfinder canary </S><S> , this identification is performed thanks to the learn & apply ( l&a ) algorithm , that consists in model - fitting the covariance matrix of wfs measurements dependant on relevant parameters : @xmath0 profile , outer scale profile and system mis - registration . </S><S> we explore an upgrade of this algorithm , the learn 3 steps ( l3s ) approach , that allows one to dissociate the identification of the altitude layers from the ground in order to mitigate the lack of convergence of the required empirical covariance matrices therefore reducing the required length of data time - series for reaching a given accuracy . for nominal observation conditions </S><S> , the l3s can reach the same level of tomographic error in using five times less data frames than the l&a approach . </S><S> the l3s technique has been applied over a large amount of canary data to characterize the turbulence above the william herschel telescope ( wht ) . </S><S> these data have been acquired the 13th , 15th , 16th , 17th and 18th september 2013 and we find 0.67``/8.9m/3.07m.s@xmath1 of total seeing / outer scale / wind - speed , with 0.552''/9.2m/2.89m.s@xmath1 below 1.5 km and 0.263``/10.3m/5.22m.s@xmath1 between 1.5 and 20 km . </S><S> we have also determined the high altitude layers above 20 km , missed by the tomographic reconstruction on canary , have a median seeing of 0.187 '' and have occurred 16% of observation time . </S>

quant-ph9906080

quantum entanglement represents a resource that can be used for many applications , such as teleportation [ 1 ] , superdense coding [ 2 ] , quantum cryptographic key distribution [ 3 ] and entanglement enhanced communication complexity [ 4 ] .much progress has been made in the analysis of bipartite entanglement of pure and mixed states .the characterization of multiparticle entangled states is , however , much harder to attain , as some key theorems that apply to the bipartite states can not be easily extended to the multipartite case . in the case of bipartite pure statesa single satisfactory measure of entanglement has been found [ 5,6 ] : the partial von neumann entropy of one of the parties .this quantity equals both the states s distillable entanglement ( the number of epr pairs that can be asymptotically obtained from it using only classically coordinated local operations ) and the state s entanglement of formation ( the number of epr pairs necessary to asymptotically build the state with classically coordinated local operations ) .the characterization of entanglement in multiparticle states is still an elusive goal .there seems to be qualitatively different kinds of entanglement in n - partite states , making it hard to even find higher bounds for the number of different kinds of entangled states necessary to build n - partite states irreversibly [ 7 ] . in this workwe describe a protocol for creating any n - partite pure state @xmath0 using only local operations and classical communication between the n parties ( n - locc ) , and a certain number of epr pairs . as has been discussed before[ 8 ] , any measure of entanglement must be non - increasing under n - locc .this will enable us to identify the number of epr pairs used as an upper bound for the bipartite entanglement of formation @xmath1 .it is likely that multipartite states will be built from epr pairs in future practical applications of multiparticle entanglement ; our protocol would then provide bounds for the efficiency of such a procedure .we also hope that our approach may help clarify the relations between bipartite and multipartite entanglement , still a problematic theoretical issue .our protocol needs to rely on an asymptotically large number of epr pairs , creating a correspondingly large number of copies of the desired n - partite state @xmath2 .@xmath1 will then be bounded from above by the number of epr pairs used per copy of @xmath0 generated .we note that in general the procedure will be irreversible .in this section we describe a protocol that builds a large number of copies of an arbitrary n - partite pure state using n - locc and epr pairs only .let us first see how it would work for an arbitrary 3-qubit pure state @xmath3 .the basic idea is very simple and consists of distributing the state using a series of teleportations , keeping track of the amount of epr pairs used in each step .let us suppose that we start with a large number @xmath4 of copies of @xmath2 at charlie s laboratory .let us label the three qubits according to their von neumann entropies , qubit 1 being the one with least entropy @xmath5 , and so on in ascending order .the first step is to do quantum data compression [ 9 ] on the sub - ensembles corresponding to qubits 1 and 2 , resulting in two groups of particles asymptotically containing @xmath6 and @xmath7 maximally compressed qubits. now charlie teleports the set corresponding to qubit 1 to alice and the other set to bob .thanks to the data compression we performed , these two teleportation steps will require only @xmath8 epr pairs shared with alice and @xmath7 with bob . to end the process we decompress the sets at alice and bob , asymptotically recovering @xmath4 qubits at each party with negligible error .the resulting state consists of @xmath4 copies of @xmath3 distributed among the 3 parties .the above discussion shows us that an arbitrary 3-partite pure state can be built using 3-locc and a bipartite entanglement of @xmath9 epr pairs , if we consider the manipulation of a large number of copies of the state . this protocol can be simply generalized for any number of parties @xmath10 . to see how ,let us first recall the sub - additivity property of von neumann entropies : if a and b are subsystems of a larger system ab , their entropies must obey the inequality @xmath11 in order to create the n - partite distributed state @xmath12 using our protocol we will need to perform @xmath13 teleportations .each teleportation takes the compressed sub - ensemble corresponding to subsystem @xmath14 ( which may include one or more of the final parties subsystems we want to distribute ) from one party to another , using @xmath15 epr pairs in the process . due to the sub - additivity property of entropy ,the most economical series of teleportations will , in general , involve teleporting larger subparts of @xmath2 first , in order to save on the number of epr pairs used in total . as an example , in order to distribute a state among 4 parties a , b , c and d , it may be advantageous to first teleport the compressed subsystem cd from site a to site c , and then teleport subsystems b and d to their respective sites .we are interested in finding the best series of teleportations from a list of all possible sequences of @xmath13 teleportations that distribute the n - partite state @xmath2 from a single location .this optimal series of teleportations can be found through an exhaustive search and constitutes what we shall call * protocol p1 * for creating @xmath16 ; the number of epr pairs used per copy of state @xmath3 created will be called * * p1(**@xmath17 * ) * for short .an example of the use of protocol p1 for a simple 4-partite system can be seen in fig . 1 .epr pairs per copy .protocol p1 invokes the most economical sequence of teleportations , one of which in this case is : first subsystem ab from c , and then the other 2 teleportations , spending the optimal amount of @xmath18 epr pairs.,title="fig : " ]in this section we calculate @xmath19 for some multipartite states of interest .we will see that protocol p1 is not optimal , at least for some classes of states .@xmath19 will then represent an upper bound for @xmath1 .@xmath20 will also be an upper bound for the amount of bipartite entanglement distillable from @xmath2 , as this quantity is bounded from above by @xmath21 .the tripartite case is especially simple : if the three sub - systems of @xmath0 have entropies @xmath22 and @xmath23 ( labeling the systems such that @xmath24 ) , then @xmath25 @xmath9 epr pairs . in the case of an n - partite state @xmath26 ( @xmath10 )there is not such a closed formula for p1(@xmath26 ) ; as we discussed above , in order to find p1(@xmath26 ) we will need to find the most economical series of @xmath13 teleportations that will distribute @xmath27 .let us apply protocol p1 to build a generalized n - ghz state @xmath28 which is a maximally entangled state of n qubits , also known as a n - cat state .if x is a non - trivial subset of the parties , then the reduced density matrix of subset x is defined as @xmath29 the n - ghz states have the property that for every non - trivial subset x we have @xmath30 .this means that we use exactly 1 epr pair in each teleportation step , no matter which subsystem of @xmath31 we choose to teleport . therefore , all sequences of teleportations that distribute the state @xmath32 will use exactly the same amount of epr pairs , this number being @xmath33 since all teleportation sequences are equivalent , we may as well choose one in particular : let us consider the one that distributes each subsystem directly to its definitive party , thus teleporting @xmath13 qubits from a single location . since each qubit is in a maximally mixed state , the quantum data compression would be of no use , making it unnecessary for us to operate on multiple copies of the state .this establishes the correspondence between protocol p1 and the one by zeilinger _[ 10 ] for building a single copy of a n - ghz state of n qubits from @xmath13 epr pairs .protocol p1 is as efficient as zeilinger s , but has a different physical meaning : we are simply teleporting each party s subsystem from a single location , using a total of @xmath13 epr pairs .the above discussion generalizes trivially for the case of n - ghz states in which each party possesses k - level subsystems instead of qubits .incidentally , the n - ghz states represent the worst case scenario for protocol p1 , in the sense that a single copy of _ any _ n - partite state of n qubits can be prepared locally and distributed using @xmath34 epr pairs .it is not known whether p1 is optimal for n - ghz states .these are states which can be written in schmidt form @xmath35 where the @xmath36 represent an orthonormal basis in party x s hilbert space .note that the coefficients @xmath37 may be made to be real by convenient redefinition of the basis vectors .for these states all reduced density matrices @xmath38 have the same von neumann entropy of @xmath39 , and therefore @xmath40 bennett _ et al . _[ 7 ] showed that a state @xmath41 can ( in the asymptotic limit of many copies ) be transformed reversibly by n - locc into a set of @xmath42 n - ghz states of n qubits .thus we may achieve the creation of @xmath41 by two different routes : either using p1 directly or first creating @xmath42 n - ghz states with p1 and then using bennett _et al . _ s procedure to turn them into @xmath41 .we have seen above that @xmath43 epr pairs , which means that the two routes demand the same number of epr pairs for creating the state . in view of this, we see that p1 s optimality for creating n - ghz states ( if it can be proved ) would mean p1 would be an optimal protocol for schmidt decomposable n - partite states as well .let us consider the n - partite state @xmath45 consisting of @xmath46 shared epr pairs , one between each pair of parties .this state has been named a n - toast state after the custom of clinking glasses during a toast [ 7 ] .we choose these states here as an example of the inefficiency of p1 for some classes of states .the most economical series of teleportations that builds @xmath47 consists again of teleporting each party s subsystem directly from an initial party . in order to see this ,let us calculate the entropies of the reduced density matrices corresponding to all non - trivial sets of parties @xmath48 .let @xmath38 denote the reduced density matrix corresponding to an arbitrary set of m parties ( @xmath49 ) .given this partition of the set of parties , the @xmath44 epr pairs in state @xmath47 can be grouped in three categories : @xmath50 shared between parties in @xmath48 ; @xmath51 shared between parties in @xmath52 ; and @xmath53 shared between a party in @xmath48 and another in @xmath52 . the trace operation over the @xmath54 parties will result in a @xmath14 that is a tensor product of the @xmath50 epr pairs within @xmath48 and the identity matrix corresponding to the epr pairs that were ` broken ' between @xmath48 and @xmath55 .thus we have @xmath56 the entropy of @xmath38 can then be easily evaluated to be @xmath57 .@xmath15 corresponds to the number of epr pairs necessary to teleport subsystem @xmath38 as part of protocol p1 .since p1 consists of @xmath58 teleportations , we can not hope to perform p1 and use less than @xmath59 epr pairs ( where the minimum is taken over all possible @xmath48 ) .this is actually achievable , as we may teleport each party s subsystem directly from an initial one , using a total number of epr pairs equal to @xmath60 the fact that state @xmath45 consists of a collection of @xmath44 epr pairs makes it obvious that @xmath61 , which is clearly below @xmath62 . for these states@xmath63 which shows that our protocol is particularly inefficient for @xmath47 states @xmath64 is as high as twice the actual bipartite entanglement of formation for large @xmath65 .we have just seen that protocol p1 is not optimal .this inefficiency prompts us to obtain bounds for @xmath21 using other methods , as a way of assessing the protocol s shortcomings .this can be done for the specific case of n - ghz states by appealing to a recent theorem proved by julia kempe [ 11 ] : * definition * : _ two n - partite states are said to be n - locc incommensurate when they can not be transformed into each other either way by any n - locc protocol . _ * theorem * ( kempe ) : _ two n - partite states _@xmath3 _ _ and _ _ @xmath26 _ _ are n - locc incommensurate if and only if they are not equivalent under local unitary transformations at each party . _ _kempe applied this theorem to prove that @xmath13 n - ghz states and an @xmath47 state are n - locc incommensurate . in both casesthe density matrices of each party are identical and equal to @xmath66 the @xmath45 state has inseparable @xmath38 for any choice of x other than a single party ( see eq .( [ rox ] ) ) .the state consisting of @xmath13 n - ghz s has separable @xmath38 for all choices of x. using kempe s theorem and the fact that local unitaries can not increase entanglement between parties , we prove that the states under consideration are n - locc incommensurate .a similar result was also obtained by bennett _et al . _ in [ 7 ] . we will now argue that this result holds true for exact transformations between arbitrarily large numbers of copies of n - ghz and epr states .let us consider the two following n - partite systems : the state @xmath67 consisting of a tensor product of @xmath68 n - ghz states of n qubits , each party possessing a subsystem with @xmath68 qubits in state @xmath69 and state @xmath70 consisting of the tensor product @xmath71 , with each party s subsystem in state @xmath72 we will have equal density matrices at each party for @xmath67 and @xmath70 if @xmath73 making it possible for us to apply the same reasoning as we did above for the case with 2 ghz and 3 epr states .thus we have established that @xmath74 and @xmath75 are incommensurate states . in particular , this means that we need strictly more than @xmath76 epr pairs to create one n - ghz state , no matter how clever our n - locc protocol is . in all of the abovewe have been considering exact transformations between the states . in references[ 5,7 ] the authors define larger equivalence classes between n - partite states , allowing for slightly imperfect fidelity @xmath77 in the transformation and for @xmath78 that deviates from 1 by a small quantity @xmath79 . the question of whether the bound of @xmath80 epr pairs per n - ghz state created holds for the transformations considered in [ 5,7 ] is open . combining the lower bound of @xmath80 epr pairs per n - ghz with the higher bound given by protocol p1 , we can write : @xmath81 it is interesting to note that the above result implies that there is a fundamental irreversibility when we transform n - partite entanglement into bipartite and vice - versa , using exact transformations of a large number of copies .for example , a 3-ghz can not be asymptotically converted into @xmath82 of an epr pair , yet we need strictly more than @xmath82 of an epr pair to asymptotically recreate a 3-ghz .p1 is obviously optimal for a trivial set of states , such as that on fig .1 . it is , however , inefficient for states like that of section [ sec sym epr ] above . for n - ghz states ,as far as we know , there is no known protocol more efficient than p1 , and inequality ( [ bound for ghz ] ) represents what seems to be the best bounds as yet for their bipartite entanglement of formation .if it can be proved that p1 is optimal for n - ghz states , then it would be optimal for schmidt decomposable n - partite states as well , as we discussed in section [ schmidt ] above .protocol p1 is not the most general protocol using quantum data compression and teleportation to build multipartite states . in this sectionwe will briefly describe three successive generalizations .these generalizations all involve taking advantage of the fact that each party s system may beconsidered as consisting of subsystems which can be compressed and teleported independently .first we describe protocol p2 .take an n - partite state in the hilbert space @xmath83 ( where @xmath84 are the parties ) .now , it may be the case that some or all of the spaces @xmath85,@xmath86 can themselves be written as tensor products of smaller spaces .for example , we may write the space @xmath85 as a tensor product of @xmath87 spaces @xmath88 writing the hilbert space in this way corresponds to dividing each party s system into subsystems .the most general way of doing this for a party @xmath89 is the following .we first write @xmath90 as the product of its prime factors .if there are @xmath87 such prime factors then we can regard @xmath89 as consisting of @xmath87 subsystems , each having the dimension of one of the prime factors .all possible subsystems of system @xmath89 consist either of these @xmath91 prime - dimensional subsystems , or of the composite - dimensional subsystems obtained by grouping these in different ways . to obtain the most general way of identifying subsystems we can consider performing unitary transformations at each party , either passively ( on the hilbert space ) or actively on the system itself ( in which case we would need to undo the transformation at a later stage ) .now let us see the steps that constitute protocol p2 .we start by preparing many copies of the state we wish to build at one location .then we consider all possible ways in which the systems @xmath84 ( which must be distributed to each party ) can be decomposed in a tensor product of prime - dimensional subsystems .the idea now is to analyze all possible ways in which these subsystems , or groups of them , can be successively compressed at one party , teleported and decompressed at another party .protocol p2 is taken to be the most efficient way of building the n - partite state by such a series of teleportations of subsystems .this protocol will be optimal for any state that consists of a tensor product of shared bipartite states , in particular for @xmath65-toast states.this becomes clear if we consider a simple example . take a 3-toast state @xmath92 we have seen that p1 is inefficient at creating n - toast states ; p1 would take 4 epr pairs to create this tensor product of 3 epr pairs .we write system @xmath89 as being composed of two 2-dimensional subsystems @xmath93 and @xmath94 , @xmath95 composed of subsystems @xmath96 and @xmath97 , and @xmath98 of @xmath99and @xmath100 .this state can be built in three teleportation steps .we start with many copies of @xmath101 at location @xmath102 first we teleport system @xmath99 to location @xmath98 ( since subsystem @xmath99 is maximally mixed there is no need to compress it ) .then we take subsystems @xmath103 and @xmath100 and , regarding them as one subsystem , we compress and teleport them to location @xmath95 , where they can be decompressed .finally , we teleport subsystem @xmath100 from location @xmath95 to location @xmath98 ( again , there is no need for compression ) .each of these three steps requires one epr pair per copy of @xmath101 and hence this procedure is optimal , constituting an example of p2 .it is clear that this protocol will be optimal for any state consisting of shared bipartite states .protocol p2 reduces to p1 in the case where all systems @xmath84 have a prime number of dimensions ( thus being not decomposable ) .this observation motivates protocol p3 . in protocol p3we allow the introduction of ancillary systems @xmath104 prepared in some known state .let system @xmath105 consist of @xmath89 and @xmath106 and similarly for @xmath107 .protocol p3 consists of applying p2 to the primed systems and optimizing also over all possible ancilla choices .the introduction of ancillae allows for greater flexibility of manipulation of the prime - dimensional subsystems , as unprimed systems with prime dimension may result in primed decomposable systems .it is easy to find examples of states for which protocol p3 is more efficient than both p1 and p2 .for example , let us consider the following tripartite state closely resembling the @xmath108 state , but in a 5x5x5 dimensional hilbert space : @xmath109 \notag \\ & + \sqrt{\varepsilon}\left| 4,4,4\right\rangle , \notag\end{aligned}\ ] ] where we have rewritten the @xmath101 state in a different form , adding a term proportional to @xmath110 and orthogonal to the space spanned by the @xmath101 state. eigenvalues of density matrices that are continuous functions of a parameter @xmath111 are themselves continuous functions of @xmath111 [ 12 ] .hence , the entropies of the partial density matrices obtained from @xmath112 will differ from those of @xmath113 by arbitrarily small amounts , for suitable values of @xmath111 .we are now dealing with subsystems in a prime - dimensional ( 5 ) hilbert space , which means that we are unable to identify subsystems in each party s system and apply p2 .therefore @xmath114 which is arbitrarily close to 4 , for small @xmath115 .protocol p3 improves on this by adding a 3-qubit ancilla to each 5-level system .we then map the state of each 5-level system into its ancilla through the unitary operation that takes ( for alice s system ) @xmath116 and similarly for bob s and charlie s systems .this amounts basically to increasing the dimensionality of each system from 5 to 8 , as we may discard the original systems and deal with the ancillae instead . since 8 is a composite number , now we can apply protocol p3 , which consists of finding the optimal way of distributing this 9-qubit system among alice , bob and charlie through teleportation . it suffices to say that the optimal way will be similar to the one we adopted when we applied protocol p2 to the @xmath117 state , with the difference that now we have one extra qubit at each party ( like ancillary bit @xmath118 in the example above ) which is minimally entangled with the rest of the system . by the same continuity argument we see that p3(@xmath26 ) is as close as we like to p2(@xmath101)@xmath119 , for small @xmath115 .this provides us with an example for which protocol p3 is better than both p1 and p2 . the introduction of the ancillae in protocol p3 was not done in the most general way possible .we considered local operations at each party that entangled ancillae to the systems corresponding to a single copy of the state we want to prepare .as we have argued , this allows for a more general manipulation of the state , and perhaps to a more efficient series of teleportations that build it .we may , however , operate with a large number of copies of the state , introducing ancillae which couple to a certain number @xmath120 of copies , instead of a single one .then our basic units would consist of one ancilla for each set of @xmath120 copies of system @xmath89 , one for each @xmath120 copies of system @xmath95 , and so on up to the last system @xmath121.protocol p4 can then be defined as the application of p2 to these larger units , optimizing over @xmath120 and ancilla sizes .p4 is more general than p3 and may be more efficient for some states .we have described a protocol that can be used ( in the asymptotic limit of many copies ) to build any n - partite pure quantum state .protocol p1 uses quantum data compression and teleportation to distribute a large number of copies of the state among the n parties , consuming bipartite entanglement in the form of epr pairs .the efficiency of the protocol depends on the state we want to create .it is optimal for some simple states ( such as that in fig .1 ) , but can be shown to be inefficient for n - toast states , discussed in sec .[ sec sym epr ] .its efficiency for arbitrary states is hard to evaluate , as there are not many results for the exact bipartite entanglement of formation @xmath21 of n - partite states .lower bounds for @xmath21 of n - ghz states are presented for comparison with p1 s results .we also show that proof of p1 s optimality for n - ghz states would imply its optimality for schmidt decomposable states as well .we consider some generalizations of protocol p1 , which can be shown to be more efficient than p1 for some classes of states ( such as n - toast states ). it would be interesting to investigate in more depth the bounds for @xmath21 obtained by the proposed generalizations . [3 ] c. h. bennett and g. brassard `` quantum cryptography : public key distribution and coin tossing '' , proceedings of ieee international conference on computer systems and signal processing , bangalore , india , december 1984 , pp 175 - 179 ; d. deutsch , a. ekert , r. jozsa , c. macchiavello , s. popescu , and a. sanpera , phys .lett . * 77 * , 2818 ( 1996 ) ; _ erratum _ * 80 * , 2022 ( 1998 ) ; d. mayers , lanl e - print quant - ph/9802025 ; h. -k .lo , h. f. chau , lanl e - print quant - ph/9803006 .

<S> we describe a protocol which can be used to generate any n - partite pure quantum state using einstein - podolsky - rosen ( epr ) pairs . </S><S> this protocol employs only local operations and classical communication between the n parties ( n - locc ) . </S><S> in particular , we rely on quantum data compression and teleportation to create the desired state . the protocol can be used to obtain upper bounds for the bipartite entanglement of formation of an arbitrary n - partite pure state , in the asymptotic limit of many copies . </S><S> we apply it to a few multipartite states of interest , showing that in some cases it is not optimal . </S><S> generalizations of the protocol are developed which are optimal for some of the examples we consider , but which may still be inefficient for arbitrary states . </S>

1108.5915

the theoretical investigations are based on density functional theory @xcite , which is first - principles in that it requires no experimental input other than the nuclear charges .we apply the generalized gradient approximation @xcite to the exchange - correlation potential and use the full - potential linearized augmented plane wave method ( flapw ) as implemented in the fleur code @xcite .the self - consistent calculations with soc were carried out with the cut - off parameter k@xmath66 of 4.0 bohr@xmath67 and 144 @xmath49-points ( for the 4@xmath84 cases ) in the full two - dimensional brillouin zone .the muffin - tin radius of 2.7 bohr ( 1.2 bohr ) was used for tm ( carbon ) atoms . to calculate the transverse anomalous hall conductivityaccurately , we applied the wannier interpolation technique , using the ` fleur ` and ` wannier90 ` codes to construct the maximally - localized wannier functions @xcite .for all systems , the distance between the tm atoms and the rigid graphene layer was structurally optimized by total energy minimization neglecting soc .we confirmed that the relaxations of the carbon atoms within the graphene layer are indeed very small , and neglecting them does not lead to any changes in the calculated quantities .tms bond covalently with graphene @xcite , thus we suspect that the van der waals interaction plays a minor role .we acknowledge discussions with mojtaba alaei , frank freimuth , klaus koepernik , pengxiang xu , tobias burnus , gustav bihlmayer , marjana leai and nicolae atodiresei .this work was supported by the hgf - yig programme vh - ng-513 and by the dfg through research unit 912 and grant he3292/7 - 1 .computational resources were provided by jlich supercomputing centre . bode , m. , heide , m. , von bergmann , k. , ferriani , p. , heinze , s. , bihlmayer , g. , kubetzka , a. , pietzsch , o. , blgel , s. , & wiesendanger , r. chiral magnetic order at surfaces driven by inversion asymmetry ._ nature _ * 447 * , 190193 ( 2007 ) .piotrowski , m.j ., piquini , p. , & da silva , j.l.f .density functional theory investigation of _ 3d _ , _ 4d _ , and _5d _ 13-atom metal clusters . _ phys .b _ * 81 * , 155446 ( 2010 ) . pfandzelter , r. , steierl , g. , & rau , c. evidence for _ 4d _ ferromagnetism in 2d systems : ru monolayers on c(0001 ) substrates .lett . _ * 74 * , 34673470 ( 1995 ) .novoselov , k.s . ,geim , a.k . ,morozov , s.v . ,jiang , d. , zhang , y. , dubonos , s.v . ,grigorieva , i.v . , & firsov , a. a. electric field effect in atomically thin carbon films ._ science _ * 306 * , 666669 ( 2004 ) .novoselov , k.s . , geim , a.k . ,morozov , s.v . , jiang , d. , katsnelson , m.i . ,grigorieva , i.v . ,dubonos , g.v . , & firsov , a.a .two - dimensional gas of massless dirac fermions in graphene ._ nature _ * 438 * , 197200 ( 2005 ) .cretu , o. , krasheninnikov , a. v. , rodrguez - manzo , j. a. , sun , l. , nieminen , r. m. , & banhart , f. migration and localization of metal atoms on strained graphene .lett . _ * 105 * , 109102 ( 2010 ) .xiao , r. , fritsch , d. , kuzmin , m.d . ,koepernik , k. , eschrig , h. , richter , m. , vietze , k. , & seifert , g. co dimers on hexagonal carbon rings proposed as subnanometer magnetic storage bits .* 103 * , 187201 ( 2009 ) .

<S> the combination of the unique properties of graphene with spin polarization and magnetism for the design of new spintronic concepts and devices has been hampered by the small coulomb interaction and the tiny spin - orbit coupling of carbon in pristine graphene . </S><S> such device concepts would take advantage of the control of the spin degree of freedom utilizing the widely available electric fields in electronics or of topological transport mechanisms such as the conjectured quantum anomalous hall effect . here </S><S> we show , using first - principles methods , that @xmath0 transition - metal ( tm ) adatoms deposited on graphene display remarkable magnetic properties . all considered </S><S> tm adatoms possess significant spin moments with colossal magnetocrystalline anisotropy energies as large as 50 mev per tm atom . </S><S> we reveal that the magneto - electric response of deposited tm atoms is extremely strong and in some cases offers even the possibility to switch the spontaneous magnetization direction by a moderate external electric field . </S><S> we predict that an electrically tunable quantum anomalous hall effect can be observed in this type of hybrid materials </S><S> . spin - orbit interaction , which couples the spin degree of freedom of electrons to their orbital motion in the lattice , leads to many prominent physical phenomena , such as the antisymmetric exchange interaction @xcite , the colossal magnetic anisotropy @xcite , or the anomalous hall effect @xcite in conventional ferromagnets . </S><S> it is also the key interaction in the newly - found quantum topological phase in topological insulators , where the quantum spin hall effect was observed experimentally @xcite , and the quantum anomalous hall effect ( qahe ) was predicted to exist @xcite . since the orbital motion can be manipulated with external electric fields , spin - orbit coupling ( soc ) opens a route to dissipationless transport and to electrical control of magnetic properties @xcite , playing a crucial role in future spintronics applications . owing to its strong spin - orbit coupling , heavy @xmath1 and @xmath0 tms display fascinating physical properties for desirable spintronic applications , especially when combined with non - vanishing magnetization . </S><S> however , magnetism of @xmath0 tms proved difficult to achieve due to their more delocalized valence @xmath2 wavefunctions and the smaller intra - atomic exchange integrals compared to @xmath3 tms such as fe , co , or ni . </S><S> low - dimensional structures , such as monolayers @xcite , atomic wires @xcite , and clusters @xcite , were proposed to facilitate the stability of @xmath0 magnetism by reducing the @xmath2-@xmath2 hybridization . </S><S> nevertheless , when deposited on substrates of noble or late transition metals , such as cu , ag , au , and pt , interdiffusion at interfaces is inevitable , and strong hybridization between the @xmath2 states of the substrate and of the adatoms is also destructive for @xmath0 magnetism . from this point of view , using @xmath4 substrates is more promising . </S><S> in fact , @xmath1 ferromagnetism was first observed in a ru monolayer deposited on the graphite ( 0001 ) surface @xcite , which is close in its chemical and physical properties to graphene . </S><S> since the day graphene was isolated and produced as a two - dimensional material @xcite , it abruptly altered the research direction of material science @xcite with the aim of exploring its fascinating transport properties . in a sense </S><S> , graphene serves as a prototype of topological insulators @xcite . </S><S> for instance , the berry phase of @xmath5 of electronic states in graphene induces a half - integer quantum hall effect @xcite , while the existence of the quantum spin hall effect was first suggested for pure graphene when soc is taken into account @xcite . </S><S> one common feature for those transport properties is the non - trivial topological origin , resulting in dissipationless charge or spin current carried by edge states with conductivity quantized in units of @xmath6 . </S><S> however , from the application point of view , a large external magnetic field is required to obtain the quantum hall effect , and the spin - degeneracy in the quantum spin hall effect makes it hard to manipulate the spin degree of freedom by controlling external fields . </S><S> to avoid such constraints while keeping the benefit of topologically protected quantized topological transport , the long - sought qahe is a perfect solution . </S><S> the essence of the qahe lies in the quantization of the transverse charge conductivity in a material with intrinsic non - vanishing magnetization . </S><S> the fact that the magnetization in ferromagnets can be much easier handled experimentally than large magnetic fields makes the qahe extremely attractive for applications in spintronics and quantum information . however , at present the qahe is merely a generic theoretical concept for magnetically doped topological insulators @xcite . </S><S> recently , qiao _ et al . _ suggested that the qahe could also occur at comparatively low temperatures in graphene decorated with fe adatoms @xcite . </S><S> here , we demonstrate based on first - principles theory that @xmath0 tms deposited on graphene are strongly magnetic , provide colossal magneto - crystalline anisotropy energies and exhibit topologically non - trivial band gaps due to very strong spin - orbit interaction . </S><S> a generic representative of this hybrid class of materials has a magneto - crystalline anisotropy energy of 10@xmath730 mev per tm and a quantum anomalous hall gap in its electronic spectrum with the size of 20@xmath780 mev . in connection with a large magneto - electric response of the deposited adatoms , </S><S> we predict that within this class of systems an electrically tunable qahe at room temperature could be achieved experimentally . </S><S> we have performed first - principles calculations of @xmath0 tm ( hf , ta , w , re , os , ir , and pt ) adatoms on graphene in 2@xmath82 , 3@xmath83 , and 4@xmath84 supercell geometries corresponding to the deposition density of 4.7 , 2.1 and 1.2 atoms / nm@xmath9 , respectively ( see the method part for details ) . throughout this work </S><S> the tm atoms are placed at the hollow sites of graphene . according to previous theoretical studies @xcite </S><S> , it is the most favorable absorption site with the exceptions of pt @xcite and ir @xcite , for which the bridge site is more preferable . among all systems </S><S> studied we selected one prototype system , w on graphene , that we discuss in more detail . in the 4@xmath84 geometry , </S><S> the optimized distance between w adatoms on the hollow site and the c plane is about 1.74 , and the hollow site is about 0.14 ev ( 0.41 ev ) per tm atom lower in energy than the bridge ( top ) site , in agreement with previous observations @xcite . </S><S> applying an electric field , the relaxed positions change by at most 0.01 bohr radii , and the hollow site remains the preferred adsorption site . </S><S> turning now to magnetism , and looking first at tm atoms adsorbed in the 4@xmath84 supercell , we find that most of the @xmath0 tm atoms ( except for pt ) are magnetic with sizable magnetic moments ( see fig . </S><S> 1a ) ranging between 0.5 and 2@xmath10 . </S><S> this is consistent with the large magnetization energy calculated , defined as the total energy difference between the nonmagnetic and ferromagnetic state , reaching e.g. for ta and w values as high as 0.56 ev and 0.34 ev , respectively . </S><S> surprisingly , the magnetic moments across the series of the rather isolated tm atoms do not follow hund s first rule , i.e. the magnetic moments do not increase in steps of 1@xmath10 from hf to re , reaching a maximum at the center of the tm series and decreasing again from re to pt . on the contrary we find the minimal spin moment for re . </S><S> this demonstrates the role of the hybridization of the tm @xmath2-orbitals with graphene . increasing the deposition density studying a 3@xmath83 supercell , the results remain basically unaltered , indicating that the @xmath2-@xmath2 hybridization between neighboring adatoms is week and that the physics is determined by the local tm - graphene bonding . </S><S> this changes in the case of the 2@xmath82 geometry , when the adatoms are in close vicinity to each other and the magnetism survives only for ta , w , re , and ir atoms . </S><S> unique to the @xmath0 tms is the property that the soc and the intra - atomic exchange are of the same magnitude . from this </S><S> we anticipate a significant influence of soc on the electronic properties , leading to strong anisotropies of the calculated quantities upon changing the direction of magnetization @xmath11 relative to the graphene plane . </S><S> the most impressive manifestation of soc in the tm - graphene hybrid materials are the colossal values of the magneto - crystalline anisotropy energy ( mae ) , defined as the total - energy difference between the magnetic states with spin moments aligned in the plane and out of the graphene plane . </S><S> the mae presents one of the most fundamental quantities of any magnetic system as its sign defines the easy magnetization axis and its magnitude gives an estimate on the stability of the magnetization with respect to temperature fluctuations @xmath7 a key issue for magnetic storage materials . </S><S> a large mae makes the magnetization very stable but also difficult to manipulate . </S><S> the values of the mae obtained in 3@xmath83 and 4@xmath84 geometry , presented in fig . 1b , </S><S> lie in the range of 10 to 50 mev per tm . </S><S> these are orders of magnitude larger than the values for @xmath3 tms , irrespective of whether they are arranged in bulk , thin films or multilayers . </S><S> similar colossal mae magnitude has been reported for the free - standing @xmath1 and @xmath0 tm chains @xcite as well as dimers of tms @xcite . </S><S> characteristic for the colossal mae @xcite is the significant contribution from the variation of magnetic moments and corresponding magnetization energies when changing the direction of @xmath11 , a contribution that is absent for conventional @xmath3 magnets . </S><S> for instance , for 4@xmath84 w on graphene , this variation of the spin moment reaches as much as 0.1@xmath10 ( fig . </S><S> 2a ) . by looking at the mae values as the function of the band filling and deposition density in fig . 2b , we find that , in contrast to the spin moments , the mae exhbits a much more irregular behavior as a function of the band filling and depends much more sensitively on the adatom density . e.g. for w the mae changes sign depending on the density of adatoms , and has a small value in the 4@xmath84 geometry . </S><S> such a situation provides a possibility to manipulate the magnetization direction with weak external perturbations , which can be particularly important for transport applications as we demonstrate below . in search for ability to manipulate the magnetic properties of adatoms by external fields , </S><S> we show in fig . </S><S> 2 the dependence of the spin moments and the mae of w adatoms in 4@xmath84 geometry on the strength of the electric field @xmath12 , applied perpendicularly to the graphene layer ( along the @xmath13-axis ) . </S><S> remarkably , the spin moment of w displays a strong dependence on the field strength , especially when the magnetization points out of plane . </S><S> qualitatively , this response can be characterized by a magnetoelectric coefficient @xmath14 , which relates the change in the spin moment to the strength of the @xmath12-field in the zero - field situation : @xmath15(w ) @xmath16 , where @xmath17 is the vacuum magnetic permeability constant . </S><S> for 4@xmath84 w on graphene and an out - of - plane magnetization , @xmath18 amounts to about @xmath19 g@xmath20cm@xmath9/v , which is one order of magnitude larger than that in fe thin films @xcite . </S><S> for an in - plane magnetization , the magnetoelectric coefficient @xmath22 is only @xmath23 g@xmath20cm@xmath9/v , i.e. one order of magnitude smaller than @xmath18 , which underlines the strong anisotropy of this quantity . </S><S> we observe similarly large variations of the magnetoelectric coupling strength in other considered @xmath0 tms . </S><S> the variation of orbital moment with varying @xmath12-field is negligible for all systems . </S><S> adatoms on graphene . </S><S> ( a ) magnetic moments due to electron spin , @xmath24 , calculated without soc and ( b ) the mae of @xmath0 tm adatoms on graphene in 2@xmath82 , 3@xmath83 and 4@xmath84 superlattice geometry . </S><S> positive ( negative ) values of mae imply an out - of - plane ( in - plane ) easy axis of the spin moments , i.e. perpendicular to ( in ) the graphene plane.,width=317 ] w on graphene on the strength of an external electric field . </S><S> ( a ) the spin moment of w adatoms @xmath24 ( in @xmath10 ) as a function of the strength of an external electric field @xmath12 , perpendicular to graphene surface , for the out - of - plane and in - plane magnetization , respectively . </S><S> negative values of @xmath12 correspond to the electric field in @xmath25-direction , i.e. the direction from graphene towards the adatoms . </S><S> ( b ) dependence of the mae on the strength of an external electric field @xmath12 , sign convention of mae is consistent to fig . </S><S> 1 . inset displays the difference of spin densities @xmath26 for the system without electric field and the case with a negative electric field of 0.13 v / .,width=317 ] striking is the effect of the electric field on the magnetization direction ( see fig . </S><S> 2b ) . at zero field ( @xmath27 ) </S><S> the magnetization is in - plane . applying a negative field of magnitude of [email protected] v / , values typical in graphene field effect transistor structures @xcite </S><S> , the sign of the mae is changed . </S><S> this means that the equilibrium direction of the magnetization can be switched from in - plane ( @xmath28 , for @xmath27 ) to out - of - plane ( @xmath29 for @xmath30 ) . </S><S> supposing that the @xmath12-field is completely screened in our system by forming a screening charge @xmath31 , the variation of the mae with respect to @xmath31 , @xmath32 , reaches as much as 28 mev / e for negative electric fields , which is more than three times larger than that on the surface of copt slabs @xcite , and one order of magnitude larger than that of fe slabs @xcite . with a moderate out - of - plane electric field of @xmath330.13 v / switching of the magnetization can be also achieved in 4@xmath84 hf , and the mae can be significantly altered in 4@xmath84 os on graphene ( by @xmath34 10 mev ) and 4@xmath84 ir on graphene ( by @xmath34 20 mev ) . </S><S> this establishes @xmath0 tm adatoms on graphene as a class of magnetic hybrid materials with high susceptibility of magnetic properties to the electric field , thus making the electrical control of magnetism in these systems possible . </S><S> the reason for such a strong magnetoelectric response of @xmath0 tms on graphene can be exemplified for the case of w in 4@xmath84 geometry . </S><S> the local w @xmath35- and @xmath2-decomposed density of electron states without soc , grouped into @xmath36 , @xmath37 and @xmath38 contributions , is presented in fig . </S><S> 3a . </S><S> as we can see , the w spin moment of about 1.6@xmath10 originates from two occupied spin - up and two unoccupied spin - down @xmath39-states , while the exchange - split @xmath40- and @xmath41-states are situated above and below @xmath42 , respectively , and almost do not contribute . upon </S><S> including soc ( e.g. for out - of - plane spins ) a strong hybridization between the @xmath43 , and @xmath40 and @xmath41 states of both spin occurs around @xmath42 ( see also fig . </S><S> 3b ) , which results in a formation of hybrid bands of mixed spin and orbital character , while the @xmath44 band remains mainly non - bonding ( see fig . </S><S> 3c ) . </S><S> when an electric field is applied along the @xmath13-axis , the states which experience most influence of the corresponding potential change are the @xmath44-states , directed perpendicularly to the graphene plane . </S><S> corresponding modification of the band structures due to the electric fields can be seen in fig . 3c , where the @xmath44 band is shifted downwards ( upwards ) by negative ( positive ) applied @xmath12 fields , while the hybrid bands of mixed character remain almost unaffected . </S><S> this causes the redistribution of the electrons in the @xmath44-states , its hybridization with the hybrid band below the fermi level and hence the variation of the magnetic moment . </S><S> this can be visualized by the plot of the spin density difference @xmath45 for the case with @xmath12 = 0 and @xmath12 = @xmath70.13 v / ( inset in fig . </S><S> 2b ) . </S><S> it is obvious that , upon applying an electric field , a certain amount of spin density is transferred from the @xmath39 state of @xmath46 character to the @xmath40 states . </S><S> owing to the difference in the hybridization with the graphene states , seen from the width of the corresponding peaks in the density of states , the @xmath39 and @xmath40 states have different localization inside the w atoms thus leading to a change in the spin moment . </S><S> at last we turn to the prediction of a stable qahe , a manifestation of the quantization of the transverse anomalous hall conductivity . </S><S> it is motivated by the observation that for 4@xmath84 w on graphene with an out - of - plane magnetization , @xmath40 and @xmath41 bands of opposite spin cross ( see circles in fig . </S><S> 3b ) and hybridize under the presence of soc , forming a global band gap across the brillouin zone ( bz ) . </S><S> thus , 4@xmath84 w becomes an insulator upon a spin - orbit driven metal - insulator transition . </S><S> we compute the anomalous hall conductivity of this system , given by @xmath47 , with @xmath48 as the chern number of all occupied bands that can be obtained as a @xmath49-space integral @xmath50 . </S><S> the integrand , @xmath51 , is the so - called berry curvature of all states below the fermi level : @xmath52 where @xmath53 is the spinor bloch wavefunction of band @xmath54 with corresponding eigen energy @xmath55 , @xmath56 is the @xmath57th cartesian component of the velocity operator . the berry curvature in reciprocal space , presented in fig . </S><S> 3d , displays a comparatively complex pattern , with positive contributions concentrated along the bz boundary , and large negative dips at certain isolated @xmath49-points away from the high - symmetry points . </S><S> such non - trivial distribution of the berry curvature , which plays a role of an effective magnetic field in the @xmath49-space , is typical for complex transition - metal compounds @xcite . </S><S> the calculated anomalous hall conductivity as a function of the electron filling expressed as energy @xmath58 relative to the fermi level is presented in the inset of fig . </S><S> 3c for an out - of - plane magnetization , @xmath59 , and zero @xmath12-field . </S><S> we indeed find that the chern number of all occupied states acquires an integer value of @xmath602 . </S><S> according to the physics of the quantum hall effect , this will result in two dissipationless and topologically protected edge states on each side of a finite graphene ribbon with w adatoms . </S><S> remarkably , another 32 mev large qahe gap with the @xmath61 can be observed at the energy of @xmath70.27 ev below @xmath42 , originating from the soc - mediated hybridization between the @xmath40 and @xmath41 bands of the same spin - up character , see figs . </S><S> 3b and 3d . </S><S> such a peculiar situation suggests that the topological properties of the qahe state in 4@xmath84 w on graphene can be controlled by tuning the position of @xmath42 e.g. via deposition on an appropriate substrate . </S><S> more importantly , at equilibrium the magnetization direction of 4@xmath84 w lies in - plane , rendering @xmath62 owing to the antisymmetric nature of the anomalous hall conductivity with respect to the magnetization direction . from our previous observations it follows , however , that the direction of @xmath11 can be conveniently switched to out - of - plane by applying a moderate electric field without any significant modification of the the electronic structure . </S><S> the gaps remain intact ( cf fig . </S><S> 3c ) , its size unchanged , and the value of the anomalous hall conductivity remains quantized at both energies . </S><S> such non - trivial topological qahe states occur also for all other @xmath0 adatoms ( except pt ) on graphene in 4@xmath84 geometry , and several of them in 2@xmath82 geometry , with the qahe gaps of comparable magnitude , although positioned away from the fermi energy . </S><S> for example , graphene with re deposited in 4@xmath84 geometry exhibits three @xmath63 qahe gaps at @xmath71 , @xmath70.15 and @xmath600.67 ev , with the corresponding gap size of 38 , 63 and 98 mev , respectively . also , </S><S> 4@xmath84 ir and os on graphene show two qahe gaps ( @xmath63 for ir and @xmath64 for os ) at 0.24 and 0.47 ev for ir , and @xmath70.38 and 0.15 ev for os , with corresponding gap size in between 11 and 80 mev . </S><S> the fact that several qahe gaps with different chern numbers at different energies can occur in the same material makes these systems a rich playground for topological transport studies . </S><S> characteristic of such systems is the leading role of the @xmath0 orbitals and strong soc of heavy adatoms in determining the topological properties : as we can see from fig . </S><S> 3b , the creation of the qahe gaps can occur at the points in the brillouin zone away from high - symmetry points due to hybridization of bands with a strong tm character , which are significantly altered by the tms soc . </S><S> this is in contrast to the cases considered previously @xcite , in which the role of the magnetic adatoms is rather to insert a rashba spin - orbit or exchange fields on the dirac states at the fermi energy and high - symmetry points in the brillouin zone . </S><S> most importantly , however , the strong soc of @xmath0 tms leads to an increase of the qahe gap by an order of magnitude , when compared to all qahe systems suggested before , guaranteeing a strong topological protection against defects , structural disorder or thermal fluctuations , and thus opening a route to qahe at room temperature . </S><S> to summarize , we predict that graphene decorated with @xmath0 transition - metals is a hybrid material displaying remarkable magnetic and transport properties , which may be utilized experimentally for the purposes of topological transport applications at reasonable temperatures and conditions . </S><S> the fact that heavy transition - metals deposited on graphene display very stable magnetism due to colossal magneto - crystalline anisotropies with exciting possibilities in engineering a robust quantum anomalous hall effect with a large band gap stems from the strong spin - orbit interaction of these elements . </S><S> the extreme sensitivity of these hybrid systems to an electric field that we predicted is likely to be enhanced when graphene is deposited on a ferroelectric substrate . </S><S> thus @xmath0-graphene hybrid materials provide a unique functionality in tailoring the required magnetic and topological properties , not likely to be achieved in conventional magnetic materials based on @xmath3 transition - metals . considering the breadth of this material class including other possible heavy adatoms , such as the @xmath1 tms or @xmath65-elements </S><S> , we anticipate that room temperature qahe may become possible in these systems and we encourage intense experimental research on these materials . </S>

0908.0874

the total and differential yields of hadrons in leptonuclear reactions reflect the space - time structure of the quark string fragmentation and the formation of hadrons , both produced directly or as a result of secondary intranuclear interactions or originated from the decays of resonances .the latters play a significant role in the production of stable hadrons ( pions and kaons ) .for example , in the case of neutrino - induced reactions the fraction of pions from the decays of light meson resonances ( up to masses @xmath20 1 gev/@xmath21 ) composes about 20 - 40% , depending on the pion type and the neutrino energy @xmath22 ( see @xcite and references therein ) .the information on the space - time structure of the hadron formation based only on the data on stable hadrons would , therefore , be rather incomplete and should be complemented by the data on the resonance production .the latter data themselves provide more direct information on the quark string fragmentation , since they are , in general , expected to be less contaminated by a contribution from the decays of the higher - mass resonances .this contribution can be rather different for various resonances and in various kinematical regions of their production .in particular , in neutrino - induced reactions the production of a favorable resonance , which can carry a large fraction of the current quark energy and hence occupies the forward hemisphere in the hadronic c.m.s .( with @xmath2 , @xmath3 being the feynman variable ) , is contributed by the decay of the higher - mass resonances mainly at @xmath23 , but to a much smaller extent at @xmath2 .the situation with unfavorable resonances ( not containing the current quark ) can be rather different .an interesting example is @xmath15 meson .+ note first of all , that the neutrinoproduction of direct @xmath15 mesons , both unfavorable and especially favorable , is expected to be strongly suppressed as compared to other favorable vector mesons , for example , @xmath24 meson ( with the same spin and isospin ) which can be produced via the main subprocess @xmath25 followed by the recombination of the current @xmath26- quark with a @xmath27- antiquark created at the string breaking .the production of a favorable @xmath15 meson , which proceeds via the subprocess @xmath28 followed by the recombination with a @xmath29- quark , is triple - suppressed as compared to the @xmath24 production due to : i ) the suppressed magnitude of the @xmath30 element of cabibo - kobayashi - maskawa matrix @xcite ; ii ) the strangeness suppression in the quark string fragmentation @xcite ; iii ) the smallness of the fraction of the nucleon momentum ( integrated over the bjorken @xmath31 variable ) carried by the @xmath27- antiquark as compared to that for the @xmath32- quark ( see e.g. @xcite ) .the product of these suppression factors is about @xmath33 .even when taking into account the contribution from the competing recombination processes , e.g. @xmath34 for the case of @xmath24 and @xmath35 for the case of @xmath0 , the expected ratio of the yields of favorable @xmath0 and @xmath24 mesons remains very small , no more than @xmath36 .+ this value is in strict contrast with experimentally estimated ratio @xmath37 at @xmath2 @xcite , indicating on the dominant role of indirect processes in the @xmath15 neutrinoproduction .one of them is , probably , related to the production of the charmed , strange @xmath1 meson via the subprocess on a strange sea quark , @xmath38 , followed by the recombination of the charm quark with the sea @xmath39- antiquark from the nucleon remnant , @xmath40 . since hereno creation of an extra @xmath41 pair is necessary , the @xmath1 meson can carry the overwhelming fraction @xmath42 of the current @xmath11- quark energy and then transfer a significant part of the latter to the @xmath15 meson in @xmath43 decays ( which occur with a summary rate [email protected]% @xcite ) .an alternative subprocess , leading to the @xmath1 production ( with approximately the same probability as in the previous case ) , is @xmath45 followed by the recombination of the @xmath11- quark with a @xmath39- antiquark created at the string breaking . in this case, however , @xmath1 carries on an average a lesser energy fraction @xmath42 .+ this work is devoted to the experimental study of the neutrinoproduction of @xmath15 and @xmath1 mesons , with a particular aim to check the aforesaid mechanism of the indirect @xmath15 production . in section 2 , the experimental procedure is described .section 3 presents the experimental data on the total and differential yields of @xmath15 meson . in section 4 ,several decay modes of @xmath1 meson are analyzed and an estimation of its yields is inferred for the case of the leading @xmath1 meson carrying the overwhelming fraction @xmath46 of the current quark energy .the results are summarized in section 5 .the experiment was performed with skat bubble chamber @xcite , exposed to a wideband neutrino beam obtained with a 70 gev primary protons from the serpukhov accelerator .the chamber was filled with a propane - freon mixture containing 87 vol% propane ( @xmath47 ) and 13 vol% freon ( @xmath48 ) with the percentage of nuclei h : c : f : br = 67.9:26.8:4.0:1.3 % .a 20 kg uniform magnetic field was provided within the operating chamber volume .+ charged current ( @xmath49 ) interactions containing a negative muon with momentum @xmath500.5 gev / c were selected .other negatively charged particles were considered to be @xmath51 mesons , except for the cases explained below . protons with momentum below 0.6 gev@xmath52 and a fraction of protons with momentum 0.6 - 0.85gev@xmath52 were identified by their stopping in the chamber .non - identified positively charged particles were considered to be @xmath53 mesons , except for the cases explained below .events in which errors in measuring the momenta of all charged secondaries and photons were less than 60% and 100% , respectively , were selected .the mean relative error @xmath54 in the momentum measurement for muons , pions and gammas was , respectively , 3% , 6.5% and 19% .each event is given a weight which corrects for the fraction of events excluded due to improperly reconstruction .more details concerning the experimental procedure , in particular , the estimation of the neutrino energy @xmath55 and the reconstuction of @xmath56 and neutral strange particle decays can be found in our previous publications @xcite .+ the events with @xmath57 30 gev were accepted , provided that the reconstructed mass @xmath9 of the hadronic system exceeds @xmath58 = 2 gev .no restriction was imposed on the transfer momentum squared @xmath59 .the number of accepted events was 4577 ( 5784 weighted events ) .the mean values of the kinematical variables were @xmath60 = 10.7 gev , @xmath61 = 3.0 gev , @xmath62 = 9.6 gev@xmath63 , @xmath64 = 2.8 ( gev / c)@xmath63 . + about 8% of neutrino interactionsoccur on free hydrogen .this contribution was subtracted using the method described in @xcite .+ when considering the production of resonances decaying into charged kaons , the @xmath65 and @xmath66 hypothesis was applied , respectively , for negatively charged particles and non - identified positively charged particles with momenta @xmath67 gev@xmath52 , introducing thereat proper corrections for the momentum of these particles .it has been checked that the choice of lower values of @xmath68 does no practically influence the results presented in next sections .the @xmath69 effective mass distributions for three different ranges of @xmath3 are plotted in figure 1 .the main problem in the @xmath0 signal separation is the large background from misidentified @xmath70 pairs having small effective masses .the background related to the correlated low - mass @xmath70 from @xmath71 , @xmath72 and @xmath73 decays was subtracted from experimental distributions using the data on these resonances obtained in @xcite . to describe the remaining ( combinatorial ) background ,two methods were applied : + i)the mixed event method , in which the shape of the background distribution was determined combining kaons from different events of the same topology and ranges of global kinematical variables ( @xmath22 , @xmath9 ) .the normalization of the background distribution was kept as a free parameter when fitting the experimental mass distribution ; + ii ) the fitted background method , in which the background was approximated by @xmath74 where @xmath75 is the charged kaon mass , while @xmath76 , @xmath77 and @xmath78 are free parameters ( in some cases @xmath78 was fixed to zero ) .+ the signal distribution was parametrized by a gaussian form @xmath79 \ , \ , , \ ] ] with the fixed pole mass @xmath80 = 1019 mev and the experimental resolution @xmath81 = 4 mev ( estimated by simulations , resulting in a fwhm value @xmath82 10 mev exceeding largely the @xmath0 natural width @xmath83 4.3 mev ) .+ the mass distributions shown in figure 1 were fitted as a sum of the signal and background distributions .the fit results for both methods i ) and ii ) turned out to be practically the same . as it is seen , the data exhibit a clear @xmath0 signal at @xmath84 , with the corresponding yield @xmath85 ( corrected for unobserved @xmath0 decay modes ) . no @xmath0 production is observed in the backward hemisphere ( @xmath23 ) , where the data are consistent with the background distribution . as a result , the @xmath0 signal in the full @xmath3- range turns out to be less expressed as compared to that at @xmath2 , leading to a less accurate estimation of the total yield : @xmath86 . +we have also obtained the @xmath58- dependence of @xmath87 and found it to increase continuously from @xmath88 at @xmath58 = 1.8 gev @xcite to @xmath89 and @xmath90 at @xmath58 = 2.6 gev and 2.9 gev , respectively .+ the differential spectrum @xmath91 , compared to that for @xmath92 meson @xcite ( at @xmath93 gev ) , is presented in figure 2 , from which two observations emerge .firstly , unlike @xmath15 meson , the yield of @xmath92 in the backward hemisphere is not much smaller than that in the forward hemisphere .this fact can be explained @xcite by a contribution from secondary intranuclear interactions of pions , @xmath94 , resulting in a @xmath92 multiplicity gain , mainly at @xmath23 .nothing similar happens with @xmath0 , due to a much smaller probability of the @xmath0 production in secondary interactions ( see section 5 for details ) .secondly , the yield of @xmath92 is rather prominent at large @xmath95 @xmath96 a characteristic feature for the direct leptoproduction of a favorable hadron . on the contrary ,the yield of @xmath0 is more concentrated around moderate positive values of @xmath3 , indicating on its origin from the decay of a favorable leading resonance .a probably candidate for the latter is , as it was mentioned in introduction , the charmed , strange @xmath1 meson the production of which is considered in the next section .as it was mentioned above , the @xmath3- dependence of the @xmath0 meson yield indicates on its origin from the decay of leading @xmath1 mesons which can be produced , for example , in exclusive reactions @xmath97 and @xmath98 with the vector @xmath99 state decaying into @xmath100 ( 94.2% ) or @xmath101 ( 5.8% ) , a nucleon remnant @xmath102 ( turning into a nucleon or a low - state resonance ) , and a hyperon @xmath103 .+ in reactions ( 4 ) and ( 5 ) , most of @xmath1 mesons carry the overwhelming fraction @xmath42 of the current @xmath11- quark energy .the mean yield of the leading @xmath104 vector meson ( with @xmath105 ) was measured earlier in @xmath106- interactions , resulting in @xmath107 @xcite .one can assume , that the mean yield of the pseudoscalar @xmath108 ( direct production ) composes 1/3 of that for @xmath109 , i.e. the summary yield of @xmath108 is expected to be about @xmath110 , leading to the mean yield of the decay @xmath0 mesons @xmath111 per @xmath49 @xmath112- interaction .+ below we consider several hadronic decay modes of @xmath1 with branching ratios exceeding a few percent ( see table ) .a rather severe cut was applied on @xmath42 ( @xmath113 0.85 ) in order to reduce the contribution from the background processes .the events with @xmath114 2.9 gev , i.e. below the threshold of the @xmath1 production , were excluded ( it should be , however , noted , that the experimental resolution of @xmath9 is about 10% in this experiment , and hence the events with the estimated value of @xmath9 in between @xmath115 gev can also have a small contribution to the @xmath1 production ) .the experimental mass resolutions ( estimated by simulations ) for considered channels are quoted in table .the considered @xmath1 decay modes , experimental mass resolutions , estimated yields at @xmath116 and decay fractions . [ cols="<,^,^,^ " , ] the main channels with a @xmath117 pair in the final states are : a ) @xmath118 and b ) @xmath119 .the contamination from @xmath70 pairs is expected @xcite to be more prominent at large ( in the absolute value ) negative values of @xmath120 , @xmath121 being the angle between the @xmath69 direction in the @xmath1 candidate rest frame and the direction of the lorentz boost from the lab system to the @xmath1 rest frame .we excluded the combination with @xmath122 where a significant enhancement of the number of combinations was observed . for channelb ) containing a @xmath123 , a cut 115 @xmath124 155 mev was applied for the @xmath125 effective mass ( the same cut was also applied for channels c)@xmath96f ) containing one or two neutral pions in the final state ) .the effective mass distributions for channels a ) and b ) are plotted in figure 3 . for both channels ,an excess of events around the @xmath1 mass is observed above the combinatorial background which can be satisfactorily described using the mixed event method ( the dashed curves ) .the experimental distributions were fitted by a sum of a gaussian ( with the width quoted in table ) and the mixed background distribution .the resulting yields , corrected for the applied cuts and the contamination from the background @xmath125 combinations ( see @xcite for details ) , are quoted in table .+ a rough estimation for the yield of the channel c ) was extracted from the excess of events above the combinatorial background in the @xmath1 mass region ( figure 3 c ) .the combinatorial background was obtained with the mixed event method and normalized to the experimental distribution above 2.15 gev@xmath126 ( dashed curve ) . the mean yield , obtained after introducing corrections related to the @xmath123 reconstruction ,is quoted in table .+ to select events - candidates to channel d ) @xmath127 , the following cuts were applied : the effective mass of @xmath70 from the @xmath92 decay is enclosed in the range [email protected] gev@xmath126 ; the effective mass of @xmath128 from the @xmath129 decay is enclosed in a slightly wider ( as compared to @xmath92 ) range of [email protected] gev@xmath126 , in view of worse mass resolution ( @xmath130 mev ) as compared to @xmath131 mev ( see @xcite for details ) ; the effective mass of @xmath132 from the @xmath133 decay is enclosed in the range of @xmath134 , with the mass resolution @xmath135 mev .the effective mass distribution for channel d ) plotted in figure 3 d is fitted as a sum of a gaussian ( with the width quoted in table ) and the mixed background distribution . the resulting yield , corrected for the applied cuts ,is quoted in table .+ to select events - candidates to channels e ) @xmath136 and f ) @xmath137 the following cuts were applied : the effective masses of @xmath138 and @xmath125 from the @xmath139 decay are enclosed in the ranges of @xmath140 and @xmath141 , with the mass resolutions , respectively , @xmath142 = 28 mev @xcite and @xmath143 = 62 mev .after these cuts , only one event - candidate to channel e ) and two event - candidates to channel f ) survived .all three events had effective masses compatible with the @xmath1 mass within @xmath44 1.5 mass resolution for the corresponding channel ( see table ) .the corresponding yields , corrected for the applied cuts are quoted in table .+ finally , we looked for decay channels with a @xmath144 in the final state : @xmath145 , @xmath146 and @xmath147 , with the summary branching ratio ( 4.09@xmath148 % @xcite .no combination with @xmath116 was found in the @xmath1 mass region , probably due to smallness of their branching ratios and the restricted statistics of @xmath144 mesons in our experiment .+ the summary experimental yield of channels a)@xmath96f ) ( with the summary branching ratio ( 27.7@xmath149 % ) is equal to @xmath150 , resulting to the total yield of leading @xmath1 mesons equal to @xmath151 gev @xmath152 .this value can be compared with the mean yield of the vector @xmath153 in @xmath106- interactions measured at slightly different kinematics : @xmath154 gev @xmath155 @xcite .the inclusive production of @xmath0 mesons in neutrinonuclear interactions is studied for the first time .it is found that the @xmath0 production occurs predominantly in the forward hemisphere of the hadronic c.m.s ., with the mean yield varying from @xmath156 at @xmath93 gev up to @xmath157 at @xmath158 gev and @xmath159 at @xmath12 gev .the measured yields are much larger than expected for the case of the direct @xmath0 neutrinoproduction , hence indicating on a dominant role of indirect mechanisms . + a possible candidate to the latter are the secondary intranuclear interactions like @xmath160 where @xmath161 denotes pions , kaons , non - strange @xmath162 and strange @xmath163 vector mesons , as well as @xmath139 and @xmath133 mesons .to estimate the multiplicity gain @xmath164 of @xmath0 meson in reactions ( 5 ) , we used a simple model described and applied in @xcite .the main quantities determining @xmath164 are the mean multiplicities @xmath165 of intermediate mesons produced in @xmath166- interactions at @xmath2 and the cross sections @xmath167 of reactions ( 5 ) .+ the values of @xmath168 were taken from @xcite , subtracting the contribution from the decays of mesonic resonances @xcite .the values of @xmath169 were taken from @xcite and averaged over the pion momentum range from @xmath170 gev@xmath52 up to @xmath171 gev@xmath52 ( above which the pion yield is negligible in this experiment ) .the mean value of @xmath172 averaged over the pion and target nucleon species is in the range of @xmath173 mb , resulting in @xmath174 .the cross sections @xmath175 and @xmath176 are expected to be slightly larger as compared to @xmath177 due to larger mean momenta of @xmath178 and @xmath24 .nevertheless , due to the smallness of their mean multiplicities @xcite , the contribution to the @xmath0 production is approximately by one order of the magnitude smaller as compared to @xmath179 .+ the averaged cross section for kaon - induced reactions , estimated from the data compiled in @xcite , is in the range of @xmath180 mb .the mean multiplicity of charged kaons is assumed to be equal to that of neutral ones taken from @xcite .the mean multiplicities of vector mesons @xmath181 were taken from @xcite , while the cross section @xmath182 was assumed to be equal to @xmath183 . as a result, the summary contribution @xmath184 was estimated to be @xmath185 .+ the mean multiplicities of @xmath139 and @xmath133 were taken from @xcite . in view of the @xmath186 content of @xmath139 and @xmath133 ( with probabilities 2/3 and 1/3 , respectively ) ,the @xmath0 production cross sections were assumed to be related to that of @xmath92 production by pions : @xmath187 mb and @xmath188 mb ( at @xmath189 around a few gev@xmath52 ) . under this assumption ,one obtains @xmath190 and @xmath191 , where the uncertainty in cross sections is not included in the quoted errors .the summary contribution from all aforesaid processes does not exceed @xmath192 .this value , even when assuming that all produced @xmath0 mesons acquire @xmath193 0 ( which is not the case ) , composes only a minor fraction of the measured yield @xmath194 at @xmath93 gev .an appreciable contribution to the latter can be provided if only the unknown cross sections @xmath195 and @xmath196 were a few times larger than supposed above .it might be worthwhile to note , that an information about the role of secondary intranuclear interactions can be , in principle , inferred from the data ( not yet existing ) on the @xmath197- dependence of the @xmath0 yield in neutrinonuclear reactions . +another source of indirect @xmath0 production at @xmath2 are the decays of leading @xmath1 mesons carrying the overwhelming fraction @xmath42 of the hadronic energy . to look the neutrinoproduction of @xmath1 mesons , six different final states of @xmath1 decaysare considered requiring @xmath116 . for all of them ,an excess of events is observed at the @xmath1 mass region , allowing to infer , for the first time in neutrino - induced reactions , an estimation for its yield : @xmath198 gev @xmath152 .+ the events - candidates of the @xmath1 production were collected to a subsample used furthermore for simulation of the @xmath199 and @xmath200 decays composing the overwhelming fraction ( 85% ) of the @xmath201 decays @xcite .the simulated differential spectrum @xmath202 turned out to be strongly shifted towards the forward hemisphere in the hadronic c.m.s . , providing 85.4% of @xmath0 mesons to have @xmath2 and resulting in the expected mean yield of decay @xmath0 mesons equal to @xmath203 gev @xmath204 .this composes more than ( or about ) the half of the measured value @xmath205 gev @xmath206 .+ in order to make a comparison with the measured differential spectrum at @xmath93 gev ( plotted in figure 2 ) , the yield of @xmath1 was properly renormalized to the number of events with @xmath207 gev , assuming that the @xmath1 production at @xmath208 gev can be neglected .this lead to @xmath209 gev @xmath210 , resulting in the expected yields of the decay @xmath0 mesons @xmath211(all @xmath212 gev @xmath213 and @xmath214 gev @xmath215 .these compose less than ( or about ) the half of experimentally estimated values , respectively , @xmath216(all @xmath217 gev @xmath218 and @xmath219 gev @xmath220 . as a result, the magnitude of the predicted spectrum @xmath202 approximately twice underestimates that for the measured one ( figure 2 ) .the shape of the predicted spectrum is , however , compatible with the experimental one . * acknowledgement . *the authors from yerphi acknowledge the supporting grants of calouste gulbenkian foundation and swiss fonds `` kidagan '' .the activity of one of the authors ( h.g . ) is supported by cooperation agreement between desy and yerphi signed on december 6 , 2002 .00 = 0.pt = 0.pt = 0.pt n.m.agababyan et al .( skat coll . ) , yerphi preprint -1619 ( 2008 ) , yerevan ; arxiv : 0811 . 2343 [ hep - ex ] , 14 nov .2008 c.amsler et al . , phys .b * 667 * , 1 ( 2008 ) w.hofman , nucl . phys .a * 479 * , 337c ( 1988 ) g.t.jones et al .( wa21 coll . ) , z. phys .c * 57 * , 197 ( 1993 ) v.v.ammosov et al .chastits at .yadra * 23 * , 648 1992 [ sov .* 23 * , 283 ( 1992 ) ] n.m.agababyan et al .( skat coll . ) , yad .fiz . * 66 * , 1350 ( 2003 ) [ phys . of at .nucl . * 66 * , 1310 ( 2003 ) ] d.s.baranov et al .( skat coll . ) , yad .fiz . * 41 * , 1520 ( 1985 ) [ sov . j. nucl. phys . * 41 * , 963 ( 1985 ) ] v.v.ammosov et al .( skat coll . ) , z. phys .c * 30 * , 183 ( 1986 ) n.m.agababyan et al .( skat coll . ) , yerphi preprint-1578 ( 2002 ) , yerevan n.m.agababyan et al .( skat coll . ) , yad .* 68 * , 1241 ( 2005 ) [ phys . of at .nucl . * 68 * , 1160 ( 2005 ) ] n.m.agababyan et al .( skat coll . ) , phys . of at .nucl . * 70 * , 1898 ( 2007 ) a.e.asratyan et al .b * 257 * , 525 ( 1991 ) n.m.agababyan et al .( skat coll . ) , yerphi preprint -1614 ( 2008 ) , yerevan ; arxiv : 0801 . 3590 [ hep - ex ] , 23 jan .2008 n.m.agababyan et al .( skat coll . ) , phys . of at .nucl . * 70 * , 1731 ( 2007 ) v.flaminio et al ., cern - hera 83 - 01 , 1983 v.flaminio et al . , cern - hera 79 - 02 , 1979

<S> the neutrinoproduction of @xmath0 and @xmath1 mesons is studied , using the data obtained with the skat bubble chamber at the serpukhov accelerator . </S><S> it is found that the @xmath0 production occurs predominantly in the forward hemisphere of the hadronic c.m.s . </S><S> ( at @xmath2 , @xmath3 being the feynman variable ) , with the mean yield strongly exceeding the expected yield of directly produced @xmath0 mesons and varying from @xmath4 at @xmath5 2 gev up to @xmath6 at @xmath7 2.6 gev and @xmath8 at @xmath7 2.9 gev , @xmath9 being the invariant mass of the hadronic system . the yield of leading @xmath1 mesons carrying more than @xmath10 of the current @xmath11- quark energy is estimated : at @xmath12 gev , @xmath13 </S><S> . it is shown , that the shape of the @xmath0 meson differential spectrum on @xmath3 is reproduced by that expected from the @xmath14 decays which , however , can account for only the half of the measured @xmath0 yield . </S><S> 1.0 cm * a study of the neutrinoproduction of @xmath15 and @xmath16 mesons * skat collaboration n.m . </S><S> agababyan@xmath17 , v.v . </S><S> ammosov@xmath18 , m. atayan@xmath19 , + n. grigoryan@xmath19 , h. gulkanyan@xmath19 , a.a . </S><S> ivanilov@xmath18 , + zh . </S><S> karamyan@xmath19 , v.a . </S><S> korotkov@xmath18 @xmath17 joint institute for nuclear research , dubna , russia @xmath18 institute for high energy physics , protvino , russia @xmath19 yerevan physics institute , armenia * yerevan 2009 * </S>

gr-qc9409016

in this contribution , i would like to present a particular speculation in a quite heuristic manner .it concerns the stabilization of scale factors belonging to internal spaces .the underlying physical problem is provided by the typical classical behaviour of scale factors in the very simplest higher - dimensional cosmological models ( see e.g. refs.@xcite ) : some blow up , while the others usually collapse , either approaching zero asymptotically or running into a singularity after a finite amount of proper time ( crack of doom ) . the papers @xcite and @xcite , as well as the references contained therein provide examples for these two types . even those rare classical solutions which evolve towards finite values of the internal scale factors are unstable against small perturbations in contrast to what we observe ( clap your hands , and you will certainly not cause an internal space to collapse ) .stabilization of an internal space can be achieved in more sophisticated models in which a particular interaction between gravity and matter may capture some of the scale factors near a ( possibly local ) minimum of an effective potential .( maybe the most popular approaches are those inspired by higher - dimensional supergravity theories ) . for a selection of some papers on this issue ,some authors considered an effective @xmath1-action for gravity @xcite , the use of finite temperature quantum field theoretic methods ( see e.g. ref.@xcite ) , or cosmological models based on the theory of superstrings @xcite .it is worth , however , thinking about the possible existence of pure gravity mechanisms causing a comparable effect in a logically and technically simpler way ( by pure i mean that matter enters a model in a very simple or even a more or less symbolic form , e.g. as a massive scalar field or a cosmological constant ) .one such possibility might be provided by the gravitational casimir effect @xcite , but this is not very much clear by now .another area in which a pure gravity solution to the stabilization problem may be looked for is marked by the minisuperspace approaches to quantum cosmology @xcite , based on a ( possibly path integral motivated ) wheeler - dewitt equation and boundary conditions for the wave function of the universe .concerning these attempts , one must say that only few solutions to few models are known .there is especially one model maybe the most natural to consider in this context that is remarkably resistent against analytical methods : a ten - dimensional space - time with @xmath0 as spatial sections , and a positive cosmological constant @xmath2 .( i will definitely refuse to assume the existence of a supergravity - inspired rank six antisymmetric tensor field which would generate stability of the @xmath3 by virtue of a freund - rubin mechanism @xcite ; see also ref .this model i will talk about , sometimes using @xmath4 for comparison .by the shorthand notations used above i mean the class of metrics @xmath5 where @xmath6 is the metric on the round unit three - sphere , and @xmath7 is the metric on the round unit six - sphere ( or on the flat unit six - torus , in which case the internal dimensions do not contribute a curvature term in the action ) .this model ( and , more general , @xmath8 with @xmath9 ) showed up in some of the lists given by other contributors at this conference ( e.g. sascha zhuk ) , and it was implicitly declared `` non - fully - integrable '' in the sense of not being reducible to an ( integrable ) toda system .we shall see that this characterization is possibly connected with physically desirable properties .the mathematical problem is that one has to treat the two scale factors @xmath10 and @xmath11 on essentially the same footing there seem to be very few possibilities to get rid of one degree of freedom ( e.g. by perturbative approximations ) .this is in contrast , for example , to the friedmann - robertson - walker ( frw ) models with a single scalar field @xmath12 @xcite , where , in a first approximation , @xmath12 appears only through the effective cosmological constant @xmath13 , and the two dimensions of minisuperspace are reduced to one . as a consequence ,the application to ( [ metric ] ) of methods developed for simpler models @xcite is likely to be problematic , and we are far from knowing the set of all physically reasonable quantum states . some material on this model may be found in refs .@xcite ) . in order to develop my speculation ,let me outline briefly what one usually does in minisuperspace quantum cosmologies , using ( [ metric ] ) as a starting point .( for an excellent introduction to this subject see ref.@xcite . )the einstein - hilbert action for ten - dimensional gravity , including a ( positive ) gravitational constant and the usual boundary term that subtracts a divergence , is given by @xmath14 where @xmath15 @xmath16 the metric induced by @xmath17 on the boundary @xmath18 and @xmath19 the trace of its extrinsic curvature .the above choice of @xmath20 ensures the correct gravitational constant today . inserting ( [ metric ] ), one obtains an action @xmath21 that we will not display ( see e.g. ref.@xcite ) , but just mention that the total volume of the @xmath22 space sections is given by @xmath23 where @xmath24 in the @xmath3 case ( then we define @xmath25 ) , and @xmath26 in the @xmath27 case ( @xmath28 ) . next we change variables according to @xmath29 and make the definition ( using @xmath30 and @xmath31 ) @xmath32 the last identity being valid only in the @xmath25 case when @xmath11 compactifies .( let us remark here that one could consider any @xmath33 ( @xmath34 ) instead of @xmath3 , but then the prefactors and exponents in these definitions would be even more messy ) . in these new variables ,the action ( [ einsteinhilbert ] ) becomes @xmath35 where the potential is given by @xmath36 the corresponding euclidean action ( obtained by sending @xmath37 , thus changing ( [ metric ] ) into the riemannian metric @xmath38 ) reads @xmath39 the wheeler - dewitt metric reduces to @xmath40 , which shows that @xmath41 and @xmath42 are lightlike coordinates in minisuperspace ( both ranging from @xmath43 to @xmath44 ) .we define the timelike direction in minisuperspace by @xmath45 .the hamiltonian corresponding to ( [ s ] ) is the well - known energy constraint @xmath46 that has to vanish ( on the constrained phase space or on physical quantum states , respectively ) .the classical ( lorentzian ) solutions of this system must obey ( in the gauge @xmath47 ) @xmath48 with the + signs ( the first of these just meaning @xmath49 ) ; the corresponding euclidean solutions following from variation of @xmath50 obey ( [ eqmotion ] ) with the signs .lorentzian and euclidean solutions ( trajectories in minisuperspace ) describe ten - dimensional geometries when re - transformed and inserted into ( [ metric ] ) and its euclidean counterpart , respectively .the variables chosen are such that the form of the classical equations of motion is formally independent of @xmath2 and @xmath20 .the transition to quantum mechanics is achieved by replacing @xmath51 in @xmath52 , thus leading to the wheeler - dewitt equation @xmath53 the operator ordering has been chosen in the only way that is consistent with general covariance under transformations of all three variables , including the lapse function @xcite .( let us note in brackets that by choosing other variables @xmath54 , one can achieve various forms of the potential @xmath55 by transforming the lapse as @xmath56 .however , the choice ( [ w ] ) seems to me to be the best one .having set up convenient variables , let us now look at some basic features of the model .( clearly , in view of ( [ eqmotion ] , [ wdw ] ) , all such features are contained in the potential @xmath57 , because the rest is fairly trivial ) .first we note that there is a curve @xmath58 ( which may be found by solving ( [ w ] ) with respect to @xmath42 ) , lying entirely in the interior of minisuperspace ( see fig.1 ) . in the region @xmath59( which extends to the axes @xmath60 and @xmath61 ) , the euclidean solutions are necessarily timelike curves ( with respect to the wheeler - dewitt metric ) , the lorentzian ones are spacelike curves ( cf . the first equation of ( [ eqmotion ] ) ) .the opposite is of course true if @xmath62 .an escape of both scale factors @xmath63 to large values is only possible inside the @xmath62 region which one is tempted to call the classical one .the region @xmath59 could then be called the quantum or euclidean regime .note that the two asymptotic branches of the zero - potential curve ( @xmath64 for the upper and @xmath65 for the lower branch in fig.1 ) approach constant @xmath66 and @xmath67 , respectively ( which is , if @xmath68 , what one expects from a classical / quantum transition regime ) .moreover , the curve @xmath58 provides all points in which a classical euclidean solution may be matched smoothly to a lorentzian one .it is not quite clear which role euclidean trajectories should play in a sensible quantum cosmology , but in some of the most prominent approaches they provide a key of finding the state of the universe .euclidean trajectories describing regular ten - geometries emerge from the origin ( @xmath69 ) and behave there like @xmath70 or @xmath71 .one usually singles out two particluar euclidean solutions as instantons : one being along the curve @xmath72 ( in @xmath59 ) and the other along @xmath73 ( in @xmath59 , too ) .they give rise to the riemannian ten - geometries @xmath74 and @xmath75 , respectively and are the only euclidean solutions starting at the origin and having a turning point ( @xmath76 , i.e. zero extrinsic curvature ) at @xmath58 .they are usually considered the preferred candidates describing the coming - into-(classical)existence of the universe by quantum tunneling @xcite . in fig.1 , these instantons as well as several other typical euclidean trajectories are displayed . when expressed in terms of their proper - time ( @xmath77 ) , the lorentzian solutions either expand in both scale factors @xmath10 and @xmath11 exponentially ( thus describing inflation without compactification ) , or they expand in one scale factor and contract in the other , thereby entering the quantum region @xmath59 and finally collapsing towards one of the axes in a kasner - type singularity ( @xmath78 while @xmath79 , or vice versa ) .the only solutions that compactify are the lorentzian analogues of the two instantons ( lying on the same curves as these , but now in the region @xmath62 ) , and even these are unstable against small perturbations . each of these two very special solutions are characterized by one scale factor being actually constant ( @xmath80 and @xmath81 , respectively ) . in the approaches that connect the instantons to the most probable classical evolution of the universe@xcite , this last number is interpreted as the actual kaluza - klein scale factor . in fig.2, some lorentzian trajectories are shown . comparing this structure to the @xmath28 case, we find that there the zero - potential curve @xmath58 is just given by @xmath82 , and the lower asymptotic branch has disappeared . can the transition from the classical trajectories satisfying ( [ eqmotion ] ) to the wheeler - dewitt equation ( [ wdw ] ) improve likelihood and stability of compactification ?let me as a starting point adopt the prescription of hartle and hawking @xcite for finding the no - boundary-solution of ( [ wdw ] ) . in this approach ,the wave function is considered as a path integral @xmath83)$ ] over compact euclidean ten - geometries having the argument of @xmath84 ( a nine - geometry or , here , just a point @xmath85 ) as their boundary . the wkb - approximation scheme @xcitethen tells us to compute @xmath86 for points @xmath85 near ( or on ) the zero - potential curve , where the euclidean action @xmath87 is taken along the euclidean trajectories .@xmath88 is a prefactor that may be estimated by wkb - methods , once @xmath50 is known .we would then have to take this @xmath89 as a boundary condition for the wheeler - dewitt equation ( [ wdw ] ) and evolve it into the region @xmath62 .there , @xmath84 is expected to develop a form @xmath90 .the phase @xmath91 is the action of a family of classical solutions that one may find using the hamilton - jacobi formalism @xcite .the probability measure is provided by the prefactor @xmath92 , and @xmath84 can be viewed as describing a superposition of a family of classical universes ( a particular one of course being ours ) .some of these universes inflate in both scale factors , and some will eventually re - enter the quantum region .only two classical paths ( namely the ones emerging from the two instantons by smooth continuation ) will undergo compactification of one scale factor and although @xmath84 can be expected to be highly peaked around these ( as dominant contributions to the path integral ) , the situation is unsatisfactory as far as stability is concerned .( remember that i refuse the introduction of a stabilizing supergravity - inspired matter coupling ) .however , one encounters some complications in this model : to begin with , let us note that the euclidean action @xmath87 is not unique , since there are several euclidean trajectories connecting a given @xmath85 with the origin .( choosing the largest or the smallest of these would presumably not result into an approximate solution of the wheeler - dewitt equation at all , due to the effects at those points where two trajectories give equal @xmath50 ) .moreover , the interpretation of the region @xmath59 according to the standard methods is questionable : since in a large portion of this region the curves @xmath93 are timelike with respect to the wheeler - dewitt metric , @xmath84 would have to be expected oscillatory there rather than exponential @xcite .this is however contrary to one s intuition about what should happen when internal spaces approach planck scale size .the question i would like to focus on is the following : what happens to the universe if it is described in the classical ( @xmath62 ) region by a trajectory that approaches the zero - potential curve , enters the region of negative @xmath57 and when evolved further classically recollapses ?if @xmath84 in the region @xmath59 ( for @xmath41 and @xmath42 large ) is indeed a semi - classical state , one should expect the universe to run into a singularity .there are some remarks about similar situations in the literature , stating either that the universe actually _ will _ recollapse classically @xcite or that there might possibly occur some kind of collapse by tunneling @xcite . in any case ,such trajectories are usually not considered as important in the description of our universe .now , let us proceed ( very ) heuristically and retain against possible objections the notion that a physically sensible state has to qualify the region @xmath59 as a quantum or a classically forbidden domain ( at least if @xmath94 ) , irrespective of the fact that classical solutions do exist there formally .then the question is : what happens , if the universe re - enters into a classically forbidden region of superspace ?the intuitive answer is : tunneling .since i have no better recipe at hand ( maybe a path integral formalism would provide one ) , and since the euclidean trajectories are in general considered as a viable tool describing tunneling , i try to _ match a euclidean trajectory smoothly to the re - entering lorentzian one_. this euclidean trajectory will in general return to the curve @xmath58 and generate another semiclassical universe ( which might recollapse again and so forth ) .hence , we arrive at a procedure which tells us to evolve trajectories according to the classical equations of motion ( [ eqmotion ] ) , but with the respective signs in each region .these ( mixed ) trajectories correspond to metrics undergoing an infinite succession of signature changes .there are certainly many reasonable objections against such a procedure playing any role in the interpretation ( or construction ) of a state @xmath84 .one such objection is provided by the fact that just matching trajectories smoothly as described above does not automatically lead to smooth ten - geometries displaying signature change ( nor to a stationary point of the action @xmath95 or @xmath50 ) .optimal matching would rather demand vanishing extrinsic curvature @xcite at @xmath58 ( and this is in turn only possible for the two instantons , due to their turning point structure ) .but even this point is not so clear , because the recent discussion on signature change shows that weaker junction conditions may allow for reasonable classical evolution @xcite too .i will return to this in the very last remark of my talk .however , even if the mixed trajectories do not represent admissable geometries , it may well be that they are linked with dominant contributions to the path integral with respect to some measure .( recall that the structure of euclidean trajectories in the minisuperspace of the model we consider is rather involved , and to find the dominant ones for a given point @xmath85 is not a simple matter , even in the standard hartle - hawking approach @xcite ) .moreover , it is of course necessary to ask for the significance of the time variable @xmath96 along the euclidean pieces , and in which way such a construction is linked to physical predictions .however , there might be a more accurate version of this procedure , where the matching of trajectories is not understood individually ( one would expect a whole family of such mixed trajectories to build up a quantum state anyway ) .the weakest question one may make out of this is : is there a solution of the wheeler - dewitt equation ( [ wdw ] ) which drags the universe approximately along such paths ?this could happen either in terms of some oscillating wkb - type wave function ( which is less likely ) or in terms of a non - oscillating @xmath84 displaying huge amplitudes in the according region of minisuperspace .( recall that except in the semiclassical wkb - approximation there is no viable and commonly accepted interpretation of solutions to ( [ wdw ] ) , especially when such high peaks occur in the absence of oscillations .this touches upon the most fundamental conceptual problems of quantum gravity @xcite , and therefore i can not even give a precise statement which mathematical properties such a wave function should have . )it is time now to look at the particular way , the euclidean / lorentzian mixed trajectories behave in the model we are considering . using simple numerical techniques , it turns out that there is a preferred region on the curve @xmath58 from which such trajectories emerge ( this region seems to consist of two pieces which are located near the turning points of the two instantons , and consist of points such that two different euclidean trajectories starting from the origin and meeting there have approximately the same action @xmath50 ) .fig.3 and fig.4 show two examples for mixed trajectories .they undergo oscillations , thereby wandering along one of the two asymptotic branches of @xmath58 .one of the two scale factors thereby approaches a finite limiting value in each case , the other one blows up . in fig.5 ,the data from fig.4 are re - expressed in terms of @xmath97 , where the time coordinate @xmath96 refers to the gauge @xmath47 .fig.6 provides a magnification of the same plot , showing that the oscillation of @xmath11 is actually a dampted one .hence , mixed trajectories correspond to metrics that oscillate in signature and display as far as the values of the scale factors are concerned perfect kaluza - klein behaviour .so , can the universe reappear due to quantum tunneling ?it is a subject for further research to look for solutions of ( [ wdw ] ) describing such a behaviour .let me however conclude with some remarks .first , such a compactification and stabilization mechanism would be quite general and independent of most features of matter coupling .it just requires the internal space to have non - vanishing ricci curvature .( note that in the @xmath4 model compactification of the @xmath27 can not happen in this way because the lower asymptotic branch of @xmath58 is missing ) .moreover , the compactified scale factor approaches @xmath98 , hence the whole mechanism works only if @xmath2 does not decay .it must be a fundamental part of the action ( and not just mimicking a @xmath13 which would disappear after an inflationary phase @xcite ) .a possible success of computations along these lines could certainly be connected with a new aspect in the interpretational problems of the state @xmath84 , because in some sense the universe would remain permanently a quantum one .one would have to work out how the arrow of time comes about , and how the universe as a whole is experienced .naively , one would expect that a state in which @xmath11 compactifies ( cf .fig.5 ) describes an effectively three - dimensional world expanding as @xmath99 .but why do we observe a classical space - time with lorentzian signature , and how does @xmath10 evolve with respect to an ( effectively ) lorentzian time ? let me add a comment that emerged in discussions at this meeting : when trying to compute @xmath87 along euclidean trajectories , one would as already mentioned encounter different , competing contributions @xmath100 and @xmath101 , say .this degree of non - uniqueness can by virtue of rather involved branching phenomena result into the mathematical structure necessary to develop non - expected compactification solutions and at the same time into the feature of being non - integrable in some sense .hence , the difficulties in finding analytic solutions may possibly be compensated by new physical effects . in this concern , there is also still something to learn about the @xmath28 model ( where the @xmath102 would compactify along the lines described above ) .my last remark concerns the form of the wheeler - dewitt equation ( [ wdw ] ) .one might consider the euclidean solutions as physical ( i.e. classical ) ones ( and not just tools to integrate ( [ wdw ] ) or to evaluate a path integral ) .in other words , a physical signature change would actually be a thing to happen . in this case, it is worth noting that the mixed trajectories extremize the action @xmath103 , obtained from ( [ s ] ) after replacing @xmath57 by its absolute value @xmath104 . towhat extent they represent physical solutions is not totally clear , because the question which junction conditions should apply at the surface of signature change is still a bit open @xcite . in any case , the classical signature change approach favours weaker conditions ( continuity of the extrinsic curvature instead of vanishing ) , and the mixed trajectories i described are likely to be accepted as classically sensible solutions .clearly , such an approach alters the standard foundations of quantum gravity . following the spirit of my speculations , one would modify the wheeler - dewitt equation accordingly , and thus arrive at @xmath105 instead of ( [ wdw ] ) . expressed briefly , treating signature change as a classical effect, the substitution @xmath51 in the euclidean hamiltonian @xmath106 leads to an equation like ( [ wdw ] ) but with @xmath107 replaced . in contrast ,the usual quantum cosmology approach amounts to perform the euclideanized substitution @xmath108 in @xmath109 , which leads back to ( [ wdw ] ) .the statement that the use of either lorentzian or euclidean geometries makes no difference in deriving the wheeler - dewitt equation @xcite is only true if the latter are of no classical significance . postulating euclideansignature in the quantum domain of minisuperspace ( which may well be a classical domain in this context , especially if @xmath110 ) , leaves the zero - potential curve as the set of points where signature change _ may _ happen .assuming further that it _ does _ happen there , one arrives at ( [ wdwmod ] ) .a modified wheeler - dewitt equation as in ( [ wdwmod ] ) is exactly what has been proposed recently by martin @xcite . in his paper ,martin also shows ( in a frw plus scalar field minisuperspace model ) that the modified wheeler - dewitt equation allows less physically sensible boundary conditions than the standard one .if such an approach turns out to be fruitful for quantum cosmology , the speculations made here would turn out to be much less speculative than in the usual context , but would simply be a heuristic description of a particular candidate semiclassical state @xmath111 or @xmath112 .maybe the following ( very last ) speculation is true : states built around mixed trajectories appear in the standard cosmological approach as highly peaked in the @xmath113 region of superspace , whereas in the classical signature change picture they emerge as oscillating wave functions .+ + + this work was supported by the austrian academy of sciences in the framework of the austrian programme for advanced research and technology .also , i would like to express my thanks to the organizers of this meeting .+ + + + this plot displays , using a @xmath85-coordinate system of minisuperspace , the zero - potential curve ( dashed ) , with its two asymptotic branches that are called upper and lower in the text , as well as some classical euclidean trajectories .the instantons are represented by those two curves which appear to end at @xmath58 ( they actually have turning points there ) .the arrows indicate the direction off the regular zero - geometry .recall that the timelike direction in minisuperspace is given by @xmath114 .this may be considered defining a light - cone that is rotated by @xmath115 against the usual one in the special relativistic @xmath116-diagrams .+ + + some lorentzian ( physical ) trajectories are drawn .the only solutions describing compactification ( i.e. the counterparts of the two instantons ) are those curves which end at their turning points on the ( dashed ) zero - potential line .all other trajectories displayed were ( without loss of much generality ) chosen without turning point and are extended maximally in both directions of their time parameter .concerning the fact that some of them hit the axes , recall from ( [ transf ] ) that , for example , @xmath117 at finite @xmath41 means @xmath78 while @xmath79 .this behaviour is referred to as kasner - type singularity in the text .a nice example is provided by the trajectory starting from @xmath118 , evolving into the region @xmath62 , re - entering the @xmath59 domain at @xmath119 and running back into the @xmath41-axis at @xmath120 ( outside the plot ) .a similar fate occurs to the trajectory starting at @xmath121 which re - enters the @xmath59 domain at @xmath122 and finally collapses towards the @xmath41-axis at @xmath123 .+ + + a particular trajectory changing its character between euclidean and lorentzian ( denoted mixed trajectory in the text ) is displayed .it oscillates along the upper branch of the @xmath58 curve ( dashed ) and hence approaches asmmptotically a constant value @xmath124 .the asymptotic boundary condition near the origin ( from where the numerical evolution starts ) has been chosen as @xmath125 .+ + + this plot shows another mixed trajectory , now oscillating around the lower branch of the zero - potential curve .the scale factor belonging to @xmath3 approaches @xmath126 .the asymptotic behavour near the origin is @xmath127 .+ + + the graphs of the functions @xmath99 and @xmath128 for the trajectory displayed in fig.4 are shown .the gauge condition defining @xmath96 is @xmath47 . in order to get definite units ,we have chosen @xmath129 , hence @xmath130 .+ + + this provides a magnification of fig.5 in order to show how @xmath128 performs damped oscillations arounds its limiting value .

<S> it is argued heuristically using an @xmath0 minisuperspace model that there might be a fundamental quantum gravity effect stabilizing internal spaces with non - vanishing ricci curvature . </S>

1401.7808

molecular junctions , where a single molecule is bound to two electrodes , provide interesting systems to study basic mechanisms of non - equilibrium charge transport at the nanoscale and are promising candidates for the development of nanoelectronic devices @xcite .the possibility of using single molecules as electronic components in electronic circuits has motivated intensive experimental @xcite and theoretical @xcite research on the conductance properties of these systems and revealed a multitude of interesting transport phenomena .in particular , it has been demonstrated that a molecular junction may be used as a nano switch if the molecule bridge has two ( or more ) stable states with different conductance , which can be reversibly transformed into each other @xcite . in recent years , several optical and non - optical mechanisms for molecular switches in junctionshave been investigated @xcite .many of the optical mechanisms rely on conformational changes of the molecule triggered by light absorption ( such as light induced _ cis_-_trans _ isomerization processes @xcite ) to facilitate reversible switching between different conductance states .non - optical mechanisms investigated include electrochemical switching , where a reversible redox pair is generated using voltage pulses @xcite or mechanical switching , where external mechanical forces are applied to produce the reversible change between states with different conductance @xcite . in recent theoretical work by some of the authors@xcite , it was shown that it is possible to use a photoinduced excited state hydrogen transfer reaction between the keto and enol tautomers of a molecule to realize a molecular switch .as was demonstrated for an example of a polyene functionalized with keto and amino groups , this switching mechanism relies on the fact that keto and enol tautomeric forms often exhibit different conjugation properties , which are associated with localized or delocalized charge distributions on the molecular bridge , and thus realize different conductance states . in this contribution, we build on these results and propose a new non - optical switching mechanism based on a ground state proton transfer reaction triggered by an external electrostatic field . as an example to demonstrate the feasibility of this mechanism , we study the ground state proton transfer reaction between oligomeric forms of p2hp and p2py .specifically , we investigate the proton transfer reaction between the enol tris[2,5-(4-hydroxypyridine ) ] ( t2hp ) and the corresponding keto form tris\{2,5-[4(1h)-pyridone ] } ( t2py ) ( for binding to the au leads , the oligomers have been appropriately functionalized at their ends as indicated in fig . [ fig2 ] ) .we show that both tautomers exhibit different current - voltage characteristics providing the `` on '' and `` off '' states of a molecular switch and provide a proof of principle that both forms can be reversibly interconverted using an external electrostatic field .keto tris\{2,5-[4(1h)-pyridone ] } ( t2py , top ) and enol tris[2,5-(4-hydroxypyridine ) ] ( t2hp , bottom ) tautomers of the molecular switch investigated in this work .both forms have been functionalized for binding to the au leads with 3,4-dimercapto-(1h)-pyridine and 3,4-dimercaptobenzenone ( keto form , top ) and 3,4-dimercaptopyridine and 3,4-dimercaptophenol ( enol form , bottom).,scaledwidth=55.0% ]to investigate the transport properties of the molecular junctions , we have used a methodology which combines electronic structure calculations employing density functional theory ( dft ) and landauer transport theory .the method has been described in detail previously @xcite . here, we only provide the main aspects and some details specific to the current application . in the electronic structure calculations, we have modelled the molecular junction using the bridge molecule and two clusters containing three layers of au atoms ( comprising a total of 88 au atoms ) .the equilibrium structures of the keto and enol tautomers have been optimized using dft methods .specifically , as described elsewhere @xcite , we have carried out a partial geometry optimization of a cluster consisting of the respective tautomer and the first layer of the au atoms of each lead in order to determine realistic tautomer - lead binding geometries .subsequently , two extra layers of au atoms per lead were added to the optimized structures to obtain the models investigated in this work ( see fig .[ fig2 ] ) . following previous work @xcite, all calculations were carried out using the b3lyp hybrid correlation - exchange functional together with a sv(p ) basis set @xcite ( au atoms have been described using the ecp-60-mwb pseudopotential @xcite ) as implemented in turbomole @xcite . to describe the effect of infinite leads we have added the surface self - energy of a au(111 ) surface to the atomic orbital energies of the au atoms of the outer layer of both leads @xcite .the current for a given voltage was obtained from the landauer formula @xcite @xmath0de\ ] ] where @xmath1 and @xmath2 are the fermi distributions of the left and right lead , respectively @xcite , and @xmath3 is the transmission function , which depends on the applied bias voltage .@xmath3 contains the information about the junction and can be expressed as @xcite @xmath4\ ] ] here , @xmath5 denotes the molecular green s function and the functions @xmath6 describe broadening of molecular states due to coupling with the metallic leads . the methodology to obtain these functionsis described in ref .fig . [ fig3 ] shows the current - voltage characteristics calculated for t2hp and t2py ( see fig .[ fig2 ] ) .the results reveal very different transport properties of the two tautomers .while the keto form exhibits a rather low current , the enol form facilitates a significantly larger current for bias voltages @xmath7 v. thus , for a given bias voltage , the two tautomers realize different conductance states of the molecular junction and can be associated with `` on '' ( enol ) and `` off '' ( keto ) states of a molecular nano switch . to obtain insight into the mechanism underlying the very different conductance properties of the two tautomers , we have analyzed their transmission functions @xmath3 ( see eq .( 2 ) ) , depicted for zero bias voltage in fig .the current is determined by the values of the transmission function close to the fermi energy .both tautomers exhibit low values of transmission in the immediate vicinity of the fermi energy resulting in low current for bias voltages @xmath8 v. significant values of transmission are observed for energies @xmath9 ev @xmath10 ev , below the fermi energy .the structures of the transmission function in this energy range reveal a pronounced difference between the two tautomers .in particular , the transmission peaks of the enol form are significantly broader , indicating a stronger coupling of the molecular bridge to the leads and , thus , resulting in the larger current observed for bias voltages @xmath7 v ( cf .[ fig3 ] ) .current - voltage characteristics of the two tautomeric forms of the molecular junction depicted in fig .the black and the red lines depict the current of the enol and keto tautomers , respectively.,scaledwidth=70.0% ] transmission as a function of electron energy ( relative to the fermi level ) at zero bias voltage .the black and red lines correspond to the enol and keto forms , respectively .the inset depicts the region close to the fermi level.,scaledwidth=70.0% ] this finding is corroborated by an analysis of the molecular orbitals associated to the most relevant structures in the transmission function , depicted in fig .[ fig5 ] . in the case of the enol form, the peak at @xmath11 ev is associated to the @xmath12-like orbitals ( a ) and ( b ) that extend over the molecule . on the other hand , the orbital ( c ) in fig .[ fig5 ] . which is responsible for the peak at @xmath13 ev in the keto tautomer , is more localized and has most of its density in the center of the molecule .the difference in the extension of the orbitals of the enol and keto forms can be explained in terms of the degree of conjugation present in both systems . whereas the enol form is mainly characterized by a delocalized valence bond structure , the contribution of this class of valence bond structures to the keto tautomercan be expected to be less important .the results presented above suggest that t2ph and t2py may be used as the `` on '' and `` off '' states in a molecular nano switch . to achieve this functionality , a mechanism for a reversible transformation between the two tautomers , i.e. a proton transfer ,is required . for this purpose ,the different charge distributions of the two tautomers may be exploited .as is well known , proton transfer reactions may be triggered by an external electrical field @xcite . to investigate this possibility, we have studied the relative stability of both tautomers in the presence of a constant external electrostatic field oriented in a direction along the axis of the junction ( see fig . [ fig6 ] ) . as a measure of the relative stability of the keto and enol forms , fig .[ fig6 ] depicts the difference of their ground state energies as a function of the strength of the applied external field .the results show that without external field , the most stable tautomer is the enol form .this is due to the stabilization obtained by aromaticity and conjugation .application of an external electric field modifies the situation . upon a gradual increase in the strength of the field the energy difference between the enol and the keto forms decreases and for values larger than @xmath14 @xmath15 an inversion of the relative stabilities of the tautomerscan be obtained .this provides a proof of principle that a reversible interconversion between the t2hp and t2py using an external electrostatic field is possible . in a more general context, the possibility to control the conductance by an external electrostatic field , may also be used as a mechanism for a single molecule transistor .+ + we finally mention some possible extensions of the present study .first , for the specific system investigated , the largest difference in current occurs only for larger voltages .this is due to the fact that the most relevant molecular orbitals have energies further away from the fermi energy .an optimization of the functionality can thus be achieved by using chemical substitutions that shift the energies closer to the fermi energy @xcite .second , in the present study we have focused on the conductance properties of the two tautomers at fixed nuclear geometry .an investigation of the mechanism of switching requires the modelling of the potential energy surfaces along the reaction coordinates and the nuclear dynamics under bias .third , the external electrostatic field may also influence the conductance , e.g. by shifting the energy levels of the molecular bridge .thus a more complete description has to include the electric fields of the junction and the external electric field in the transport calculations .this will be the subject of future work .energy difference between t2py and t2hp for different values of the applied external electrostatic field .the inset shows the direction of the field.,scaledwidth=70.0% ]we have explored the possibility to use a proton transfer reaction as a mechanism for switching a molecular junction between different conductance states . as a specific example, we have analyzed the conductance properties of t2hp and t2py enol - keto tautomers .we have shown that both forms exhibit very different electrical currents , which allows their use as the `` on '' and `` off '' states of a molecular nano switch .the origin of this difference has been explained in terms of the different electronic structures characterizing both tautomers .furthermore , we have demonstrated that switching between the two conductance state may be achieved by an external electrostatic field which triggers a proton transfer reaction that interconverts the two tautomers reversibly .the results of this theoretical study suggest that keto - enol tautomers can be used as building blocks of a nanoscale molecular switch or , more generally , a molecular transistor .this work has been supported by the the german - israeli foundation for scientific development ( gif ) , the deutsche forschungsgemeinschaft ( dfg ) through the cluster of excellence `` engineering of advanced materials '' ( eam ) , sfb 953 and a research grant , as well as projects , ctq2012 - 36966 ( micinn ) , and uah2011/exp-041 ( uah ) .als acknowledges the research grant of the national science centre of poland 2011/01/m / st2/00561 .generous allocation of computing time at the computing centers in erlangen ( rrze ) , munich ( lrz ) , and jlich ( jsc ) is greatly acknowledged .park , h. ; park , j. ; lim , a. k. l. ; anderson , e. h. ; alivisatos , a. p. ; mceuen , p. l. nanomechanical oscillations in a single - c60 transistor ._ nature _ * 2000 * , _ 407 _ , 5760 elbing , m. ; ochs , r. ; koentopp , m. ; fischer , m. ; von hnisch , c. ; weigend , f. ; evers , f. ; weber , h. b. ; mayor , m. a single - molecule diode .u. s. a. _ * 2005 * , _ 102 _ , 88158820 reed , m. a. ; zhou , c. ; muller , c. j. ; burgin , t. p. ; tour , j. m. conductance of a molecular junction ._ science _ * 1997 * , _ 278 _ , 252254 reichert , j. ; ochs , r. ; beckmann , d. ; weber , h. b. ; mayor , m. ; lhneysen , h. v. driving current through single organic molecules .lett . _ * 2002 * , _ 88 _ , 176804 smit , r. h. m. ; noat , y. ; untiedt , c. ; lang , n. d. ; van hemert , m. c. ; van ruitenbeek , j. m. measurement of the conductance of a hydrogen molecule ._ nature _ * 2002 * , _ 419 _ , 906909 qiu , x. h. ; nazin , g. v. ; ho , w. vibronic states in single molecule electron transport .lett . _ * 2004 * , _ 92 _ , 206102 tao , n. j. electron transport in molecular junctions ._ nature nanotechnology _ * 2006 * , _ 1 _ , 173181 ioffe , z. ; shamar , t. ; ophir , a. ; noy , g. ; yutsis , i. ; kfir , k. ; cheshnovsky , o. ; selzer , y. detection of heating in current - carrying molecular junctios by raman scattering ._ nature nanotechnology _ * 2008 * , _ 3 _ , 727732 meded , v. ; bagrets , a. ; arnold , a. ; evers , f. molecular switch controlled by pulsed bias voltages . _ small _ * 2009 * , _ 5 _ , 22182223 van der molen , s. j. ; liao , j. ; kudernac , t. ; agustsson , j. s. ; bernard , l. ; calame , m. ; van wees , b. j. ; feringa , b. l. ; schnenberger , c. light - controlled conductance switching of ordered metal molecule metal devices ._ nano letters _ * 2009 * , _ 9 _ , 7680 van der molen , s. j. ; liljeroth , p. charge transport through molecular switches. _ j. phys .: condens . matter _* 2010 * , _ 22 _ , 133001 ballmann , s. ; hrtle , r. ; coto , p. b. ; elbing , m. ; mayor , m. ; bryce , m. r. ; thoss , m. ; weber , h. b. experimental evidence for quantum interference and vibrationally induced decoherence in single - molecule junctions ._ * 2012 * , _ 109 _ , 056801 ballmann , s. ; hieringer , w. ; hrtle , r. ; coto , p. b. ; bryce , m. r. ; grling , a. ; thoss , m. ; weber , h. b. the role of vibrations in single - molecule charge transport : a case study of oligoynes with pyridine anchor groups .status solidi b _ * 2013 * , hnggi , p. ; ratner , m. ; yaliraki , s. ; eds , processes in molecular wires .* 2002 * , _ 281 _ nitzan , a. ; ratner , m. a. electron transport in molecular wire junctions ._ science _ * 2003 * , _ 300 _ , 13841389 cuniberti , g. , fagas , g. , richter , k. , eds ._ introducing molecular electronics _; lectures notes in physics ; springer - verlag berlin heidelberg , 2005 ; vol . 680 joachim , c. ; ratner , m. a. molecular electronics : some views on transport junctions and beyond . _u. s. a. _ * 2005 * , _ 102 _ , 88018808 datta , s. _ quantum transport : atom to transistor _ ; cambridge university press , cambridge , 2nd edition , 2005 galperin , m. ; nitzan , a. ; ratner , m. a. the non - linear response of molecular junctions : the polaron model revisited . _ j. phys . : condens. matter _ * 2008 * , _ 20 _ , 374107 nazarov , y. v. ; blanter , y. m. _ quantum transport : introduction to nanoscience _ ; cambridge university press , cambridge , 2009 cuevas , j. c. ; scheer , e. _ molecular electronics : an introduction to theory and experiment _ ; world scientific publishing co. pte .singapore , 2010 hrtle , r. ; thoss , m. resonant electron transport in single - molecule junctions : vibrational excitation , rectification , negative differential resistance , and local cooling . _b _ * 2011 * , _ 83 _ , 115414 zimbovskaya , n. a. ; pederson , m. r. electron transport through molecular junctions . _rep . _ * 2011 * , _ 509 _ , 187 wang , h. ; thoss , m. numerically exact , time - dependent study of correlated electron transport in model molecular junctions ._ j. chem .phys . _ * 2013 * , _ 138 _ , 134704 hrtle , r. ; butzin , m. ; thoss , m. vibrationally induced decoherence in single - molecule junctions .b _ * 2013 * , _ 87 _ , 085422 pshenichnyuk , i. a. ; coto , p. b. ; leitherer , s. ; thoss , m. charge transport in pentacene - graphene nanojunctions . _ j. phys . chem . lett . _ * 2013 * , _ 4 _ , 809814 aviram , a. ; ratner , m. a. molecular rectifiers ._ chemical physics letters _ * 1974 * , _ 29 _ , 277283 metzger , r. m. unimolecular electronics ._ j. mater .chem . _ * 2008 * , _ 18 _ , 43644396 vuillaume , d. molecular - scale electronics ._ comptes rendus physique _ * 2008 * , _ 9 _ , 7894 choi , b .- y . ;kahng , s .- j . ; kim , s. ; kim , h. ; kim , h. w. ; song , y. j. ; ihm , j. ; kuk , y. conformational molecular switch of the azobenzene molecule : a scanning tunneling microscopy study ._ * 2006 * , _ 96 _ , 156106 del valle , m. ; gutirrez , r. ; tejedor , c. ; cuniberti , g. tuning the conductance of a molecular switch . _nature nanotechnology _ * 2007 * , _ 2 _ , 176179 mendes , p. ; flood , a. ; stoddart , j. nanoelectronic devices from self - organized molecular switches .phys . a : mater .sci . process . _* 2005 * , _ 80 _ , 11971209 prasongkit , j. ; grigoriev , a. ; ahuja , r. mechano - switching devices from carbon wire - carbon nanotube junctions .b _ * 2013 * , _ 87 _ , 155434 sobolewski , a. l. reversible molecular switch driven by excited - state hydrogen transfer . _ phys .phys . _ * 2008 * , _ 10 _ , 12431247 benesch , c. ; rode , m. f. ; ek , m. ; hrtle , r. ; rubio - pons , o. ; thoss , m. ; sobolewski , a. l. switching the conductance of a single molecule by photoinduced hydrogen transfer ._ j. phys .* 2009 * , _ 113 _ , 1031510318 benesch , c. ; ek , m. ; klime , j. ; kondov , i. ; thoss , m. ; domcke , w. vibronic effects in single molecule conductance : first - principles description and application to benzenealkanethiolates between gold electrodes ._ j. phys .c _ * 2008 * , _ 112 _ , 98809890 benesch , c. ; ek , m. ; thoss , m. ; domcke , w. vibronic effects on resonant electron conduction through single molecule junctions ._ * 2006 * , _ 430 _ , 355360 schfer , a. ; horn , h. ; ahlrichs , r. fully optimized contracted gaussian basis sets for atoms li to kr . _phys . _ * 1992 * , _ 97 _ , 25712577 andrae , d. ; huermann , u. ; dolg , m. ; stoll , h. ; preu , h. energy - adjusted ab initio pseudopotentials for the second and third row transition elements . _. acta _ * 1990 * , _ 77 _ , 123141 , a development of university of karlsruhe and forschungszentrum karlsruhe gmbh , 1989 - 2007 , turbomole gmbh , since 2007 ; available from http://www.turbomole.com .fisher , d. s. ; lee , p. a. relation between conductivity and transmission matrix .b _ * 1981 * , _ 23 _ , 68516854 in the calculations we have used @xmath16 ev meir , y. ; wingreen , n. s. landauer formula for the current through an interacting electron region . _ phys .lett . _ * 1992 * , _ 68 _ , 25122515 arabi , a. a. ; matta , c. f. effects of external electric fields on double proton transfer kinetics in the formic acid dimer .* 2011 * , _ 13 _ , 1373813748

<S> a novel mechanism for switching a molecular junction based on a proton transfer reaction triggered by an external electrostatic field is proposed . as a specific example to demonstrate the feasibility of the mechanism , the tautomers [ 2,5-(4-hydroxypyridine ) ] and \{2,5-[4(1h)-pyridone ] } </S><S> are considered . employing a combination of first - principles electronic structure calculations and landauer transport theory , </S><S> we show that both tautomers exhibit very different conductance properties and realize the `` on '' and `` off '' states of a molecular switch . </S><S> moreover , we provide a proof of principle that both forms can be reversibly converted into each other using an external electrostatic field . </S>

0704.0319

recently the spin - orbit(so ) interaction in semiconductor mesoscopic system has attracted a lot of interest@xcite . due to the coupling of electron orbital motion with the spin degree of freedom ,it is possible to manipulate and control the electron spin in so coupling system by applying an external electrical field or a gate voltage , and it is believed that the so effect will play an important role in the future spintronic application . actually , various interesting effects resulting from so coupling have already been predicted , such as the datta - das spin field - effect transistor based on rashba so interaction@xcite and the intrinsic spin hall effect@xcite . in this paperwe shall focus our attention on the persistent charge current and spin current in mesoscopic semiconductor ring with so interaction .the existence of a persistent charge current in a mesoscopic ring threaded by a magnetic flux has been predicted decades ago@xcite , and has been extensively studied in theory@xcite and also observed in various experiments@xcite .the reason that a persistent charge current exists may be interpreted as that the magnetic flux enclosed by the ring introduces an asymmetry between electrons with clockwise and anticlockwise momentum , thus leads to a thermodynamic state with a charge current without dissipation . for a mesoscopic ring with a texture like inhomogeneous magnetic field , d. loss et al.@xcite predicted that besides the charge current there are also a persistent spin current .the origin of the persistent spin current can be related to the berry phase acquired when the electron spin precesses during its orbital motion .the persistent spin current has also been studied in semiconductor system with rashba so coupling term@xcite .recently it is shown that a semiconductor ring with so coupling can sustain a persistent spin current even in the absence of external magnetic flux@xcite .for the system of a mesoscopic ring with a magnetic impurity , the persistent charge current has been investigated in the context of a mesoscopic ring coupled with a quantum dot@xcite , where the quantum dot acts as an impurity level and will introduce charge or spin fluctuations to the electrons in the ring .the kondo effect arising from a localized electron spin interacting with a band of electrons will be essential in the charge transport in the ring .but to our knowledge in these systems the so effect hasnt been considered .it might be expected that the interplay between the kondo effect and the so coupling in the ring can give new features in the persistent currents . in this paperwe shall address this problem and investigate the so effect on persistent charge and spin currents in the ring system with an anderson impurity .the anderson impurity can act as a magnetic impurity when the impurity level is in single electron occupied state and as well as a barrier potential in empty occupied regime .the outline of this paper is as follows . in sectionii we introduce the model hamiltonian of the system and also the method of calculation by finite - u slave boson approach@xcite . in section iii the results of persistent charge current and spin current are presented and discussed . in sectioniv we give the summary .the electrons in a closed ring with so coupling of rashba term can be described by following hamiltonian in the polar coordinates@xcite @xmath1\;,\ ] ] where @xmath2 , @xmath3 is the radius of the ring .@xmath4 will characterize the strength of rashba so interaction .@xmath5 is the external magnetic flux enclosed by the ring , and @xmath6 is the flux quantum .we can write the above hamiltonian in terms of creation and annihilation operators of electrons in the momentum space , @xmath7\;,\ ] ] where @xmath8 , @xmath9,(@xmath10 ) with @xmath11 .one can see that the so interaction causes the @xmath12 mode electrons coupled with @xmath13 and @xmath14 mode electrons and spin - flip process .we consider the system with a side - coupled impurity which can be described by the anderson impurity model , @xmath15 the tunneling between the impurity level and the ring are given by @xmath16 then the total hamiltonian for the system should be @xmath17 in order to treat the strong on - site coulomb interaction in the impurity level .we adopt the finite - u slave boson approach@xcite .a set of auxiliary bosons @xmath18 is introduced for the impurity level , which act as projection operators onto the empty , singly occupied(with spin up and spin down ) , and doubly occupied electron states on the impurity , respectively. then the fermion operators @xmath19 are replaced by @xmath20 , with @xmath21 . in order to eliminate un - physical states ,the following constraint conditions are imposed : @xmath22 , and @xmath23 .therefore , the hamiltonian can be rewritten as the following effective hamiltonian in terms of the auxiliary boson @xmath18 and the pesudo - fermion operators @xmath24 : @xmath25 \nonumber\\ & + & \sum_{\sigma}\epsilon_d f^\dagger_{\sigma}f_{\sigma}+ ud^\dagger d \nonumber\\ & + & \sum_{m,\sigma } ( t_dz^\dagger_{\sigma } f^\dagger_{\sigma}c_{m\sigma}+h.c . )+ \lambda^{(1)}(\sum_{\sigma } p^\dagger_{\sigma}p_{\sigma}+e^\dagger e+d^\dagger d-1 ) \nonumber\\ & + & \sum_{\sigma}\lambda^{(2)}_{\sigma}(f^\dagger_{\sigma}f_{\sigma}-p^\dagger_{\sigma}p_{\sigma}-d^\dagger d ) \;,\end{aligned}\ ] ] where the constraints are incorporated by the lagrange multipliers @xmath26 and @xmath27 .the first constraint can be interpreted as a completeness relation of the hilbert space on the impurity level , and the second one equates the two ways of counting the fermion occupancy for a given spin . in the framework of the finite - u slave boson mean field theory@xcite , the slave boson operators @xmath28 and the parameter @xmath29 are replaced by real c numbers . thus the effective hamiltonian is given as @xmath30 \nonumber\\ & + & \sum_{\sigma}{\tilde\epsilon_{d\sigma}}f^\dagger_{\sigma}f_{\sigma } + \sum_{m\sigma } ( { \tilde t_{d\sigma } } f^\dagger_{\sigma}c_{m\sigma}+h.c.)+e_g\;,\end{aligned}\ ] ] where @xmath31 represents the renormalized tunnel coupling between the impurity and the mesoscopic ring .@xmath29 can be regarded as the wave function renormalization factor .@xmath32 is the renormalized impurity level and @xmath33 is an energy constant . in this mean field approximationthe hamiltonian is essentially that of a non - interacting system , hence the single particle energy levels can be calculated by numerical diagonalization of the hamiltonian matrix .then the ground state of this system @xmath34 can be constructed by adding electrons to the lowest unoccupied energy levels consecutively . by minimizing the ground state energy with respect to the variational parameters a set of self - consistent equationscan be obtained as in ref.[27,28 ] , and they can be applied to determine the variational parameters in the effective hamiltonian .in this section we will present the results of our calculation of the persistent charge current and spin current circulating the mesoscopic ring .since there is still some controversial in the literature for the definition of the spin current operator in the ring system with so coupling term@xcite .we give both the formula of charge and spin currents used in this paper explicitly .it is easy to obtain that the @xmath35 component of electron velocity operator in this so coupled ring is @xmath36\;.\ ] ] thereby the charge current operator is define as @xmath37 , and in terms of creation and annihilation operator it can be written as @xmath38\;.\ ] ] at zero temperature , the persistent charge current is given by the expectation value of the above charge current operator in the ground state , @xmath39 , and it can also be calculated from the expression @xmath40 where @xmath41 is the ground state energy . in fig.1 the persistent charge current vs. the enclosed magnetic fluxis plotted for a set of values for the so coupling strength .here we have taken the model parameters @xmath42 and the total number of electrons @xmath43 is around @xmath44 . in this caseone can obtain the fermi energy of the system @xmath45 and the level spacing @xmath46 around the fermi surface .we consider the energy level of the anderson impurity is well below the fermi energy ( with @xmath47 ) , therefore the anderson impurity is in the kondo regime .one can see in fig.1 that the characteristic features of persistent charge current depends on the parity of the total number of electrons(@xmath43 ) , and can be distinguished by two cases with @xmath43 odd and @xmath43 even .this is attributed to the different occupation patterns of the highest occupied single particle energy level in the mean field effective hamiltonian .the persistent charge current for the system with @xmath48 electrons is different from that with @xmath43 electrons by a @xmath49 phase shift @xmath50 . in case ( i ) where the electron number is odd(@xmath51 and @xmath52 ) , one electron is almost localized on the impurity level and forming a singlet with electron cloud in the conducting ring .this phenomena leads to the well known kondo effect .fig.1 shows that the kondo effect decreases the magnitude of the persistent charge current , and also makes its curve shape resemble sinusoidal . in the presence of finite so coupling(@xmath53 ) ,the spin - up and spin - down electrons are coupled and it causes the splitting of the twofold degenerated energy levels in the effective hamiltonian .it turns out that the kondo effect is suppressed and the abrupt jumps of the persistent charge current with similarity to that of ideal ring case appears .it is explained in ref.[14 ] that the jumps of the persistent charge current in the case of odd number of electrons are due to a crossing of levels with opposite spin . in case ( ii )where @xmath43 is even ( @xmath54 and @xmath55 ) , the kondo effect is manifested that the magnitude of persistent charge current is significantly suppressed compared with ideal ring case and the rounding of the jumps of persistent charge current due to the level crossing . in the presence of finite so coupling ,the persistent charge current decreases with increasing the so coupling strength when @xmath53 .fig.2 displays the persistent charge current as a function of the so coupling strength @xmath4 at different enclosed magnetic flux .the persistent charge current exhibits oscillations with increasing the value of @xmath4 for both the systems with even or odd number of electrons .therefore by tuning the so coupling strength , the magnetic response of this system can change from paramagnetic to diamagnetic and vice versa .it indicates that so coupling can play a important role in electron transport in this mesoscopic ring .the curve of the persistent charge current for odd number of electrons shows discontinuity in its derivation , this can be attributed the level crossing in the energy spectrum by changing @xmath4 .it is also noted that the position of this discontinuity for odd @xmath43 also corresponds to the peak or valley in even n case .since the electron has the spin degree of freedom as well as the charge , the electron motion in the ring may give rise to a spin current besides the charge current .now we turn to study the persistent spin current in the ground state .the spin current operator is defined by @xmath56 , which can be written explicitly as @xmath57\}\;,\ ] ] therefore the three component of spin current operator in terms of creation and annihilation operators are given by @xmath58\;,\ ] ] @xmath59\;,\ ] ] @xmath60\;,\ ] ] the expectation value of the spin current @xmath61 . in our calculationwe find that only the @xmath62 component of the spin current is nonzero in the ground state .fig.3 shows the persistent spin current @xmath63 vs. magnetic flux at different so coupling strength .the persistent spin current is a periodic function of the magnetic flux @xmath64 , which has the even parity symmetry @xmath65 and also an additional symmetry @xmath66 .it is noted that the persistent spin current has quite different dependence behaviors on magnetic flux compared with the persistent charge current in fig.1 . in the presence of finite so coupling ,the persistent spin current is nonzero both for the systems with odd @xmath43 and even @xmath43 at zero magnetic flux , it indicates that a persistent spin current can be induced solely by so interaction without accompany a charge current .this phenomena is also shown in ref.[17 ] where a so coupling / normal hybrid ring was considered .in fig.4 the persistent spin current @xmath63 as a function of so coupling strength is plotted . in the absence of so coupling @xmath67 ,the persistent spin current is exactly zero for both even and odd number electron system . in the presence of so coupling ,the persistent spin current becomes nonzero and shows oscillations with increasing @xmath4 .it can change from positive to negative values or vice versa by tuning the so coupling strength .the sign of the persistent spin current also shows dependence on the enclosed magnetic flux . for the system with odd @xmath43, there is abrupt jumps in the curve of persistent spin current at certain value of @xmath4 , the reason for the jump is the same as that in the charge current , and is due to the level crossing in the energy spectrum .it is noted that the position of the jump coincides with that of the persistent charge current .this kind of characteristic feature of the persistent currents might provide a useful way to detect the so coupling effects in semiconductor ring system .in summary , we have investigated the rashba so coupling effect on the persistent charge current and spin current in a mesoscopic ring with an anderson impurity .the anderson impurity leads to the kondo effect and decreases the amplitude of the persistent charge and spin current in the ring . in the semiconducting ring with so interaction , the persistent chargecurrent changes significantly by tuning the so coupling strength , e.g. from the paramagnetic to diamagnetic current . besides the persistent chargecurrent , there also exists a persistent spin current , which also oscillates with the so coupling strength .it is shown that at zero magnetic flux a persistent spin current can exist even without the charge current .since the persistent spin current can generate an electric field@xcite , one might expect that experiments on semiconductor ring with rashba so coupling can detect the persistent spin current .i. zutic , j. fabian , and s. das sarma , rev .mod . phys . * 76 * , 323 ( 2004 ) .s. datta and b. das , appl .. lett . * 56 * , 665(1990 ) .s. murakami , n. nagaosa , and s. c. zhang , science * 301 * , 1348 ( 2003 ) ; j. sinova , d. culcer , q. niu , n. a. sinitsyn , t.jungwirth , and a. h. macdonald , phys .lett . , * 92 * , 126603(2004 ) .m. b@xmath68ttiker , y. imry , and r. landauer , phys . lett.*96a * , 365 ( 1983 ) .h. f. cheung , y. gefen , e. k. riedel , and w. h. shih , phys .b * 37 * , 6050 ( 1988 ) .d. loss and p. goldbart , phys .b * 43 * , 13762 ( 1991 ) .g. montambaux , h. bouchiat , d. sigeti , and r. friesner , phys .b * 42 * , 7647 ( 1990 ) .y. meir , y. gefen , and o. entin - wohlman , phys .lett . * 63 * , 798 ( 1989 ) .b. l. altshuler , y. gefen , and y. imry , phys .lett . * 66 * , 88(1991 ) .l. p. l@xmath69vy , g. dolan , j. dunsmuir , and h. bouchiat , phys .lett . * 64 * , 2074 ( 1990 ) .v. chandrasekhar , r. a. webb , m. j. brady , m. b. ketchen , w. j. gallagher , and a. kleinsasser , phys .lett . * 67 * , 3578 ( 1991 ) .d. mailly , c. chapelier , and a. benoit , phys .lett . * 70 * , 2020 ( 1993 ) .d. loss , p. goldbart , and a. v. balatsky , phys .lett . , * 65 * , 1655 ( 1990 ) ; d. loss and p. m. goldbart , phys .b * 45 * , 13544 ( 1992 ) .j. splettstoesser , m. governale , and u. z@xmath68licke , phys .b * 68 * , 165341 ( 2003 ). j. s. shen and k. chang , phys .b * 74*,235315(2006 ) .r. citro and f. romeo , phys .b*75*,073306(2007 ) . q. f. sun , x. c. xie , and j. wang , cond - mat/0605748 .m. b@xmath68ttiker and c. a. stafford , phys .lett . * 76 * , 495 ( 1996 ) . v. ferrari , g. chiappe , e. v. anda , and m. a. davidovich , phys .rev . lett .* 82 * , 5088 ( 1999 ) .i. affleck and p. simon , phys .rev . lett.*86 * , 2854 ( 2001 ) ; phys .b*64 * , 085308 ( 2001 ) .k. kang and s. c. shin , phys .lett . * 85 * , 5619 ( 2000 ) .s. y. cho , k. kang , c. k. kim , and c. m. ryu , phys .b * 64 * , 033314 ( 2001 ) h. p.eckle , h. johannesson , and c. a. stafford , phys .rev . lett . * 87 * , 016602 ( 2001 ) .h. hu , g. m. zhang , and l. yu , phys .lett . * 86 * , 5558 ( 2001 ) .g. kotliar and a. e. ruckenstein , phys .* 57 * , 1362 ( 1986 ) .v. dorin and p. schlottmann , phys .b * 47 * , 5095 ( 1993 ) .b. dong and x. l. lei , phys .b * 63 * , 235306 ( 2001 ) ; phys . rev .b * 65 * , r241304 ( 2002 ) .g. h. ding and b. dong , phys .b * 67 * , 195327 ( 2003 ) .f. e. meijer , a. f. morpurgo , and t. m. klapwijk , phys . rev .b * 66 * , 033107 ( 2002 ) .q. f. sun and x. c. xie , phys .b * 72 * , 245305(2005 ) .f. meir and d. loss , phys . rev .lett . * 90 * , 167204 ( 2003 ) .( solid line),@xmath70(dashed line ) , @xmath71(dotted line),@xmath72(dash - dotted line ) ) .the total number of electrons @xmath73 ( a ) , @xmath44(b ) , @xmath74(c ) , @xmath75(d ) .we take the other parameters @xmath76 in the calculation .the persistent charge current is measured in units of @xmath77 . ] vs. magnetic flux for a set of values for the spin - orbit coupling strength ( with @xmath80(solid line),@xmath81(dashed line ) , @xmath72(dotted line ) ) .the panel ( a ) , ( b ) , ( c ) and ( d ) corresponds the system with total number of electrons @xmath82 , respectively .the persistent spin current is measured in units of @xmath83 , and we have taken the other parameter values the same as that in fig.1 . ]

<S> based on the finite @xmath0 slave boson method , we have investigated the effect of rashba spin - orbit(so ) coupling on the persistent charge and spin currents in mesoscopic ring with an anderson impurity . </S><S> it is shown that the kondo effect will decrease the magnitude of the persistent charge and spin currents in this side - coupled anderson impurity case . in the presence of so coupling , </S><S> the persistent currents change drastically and oscillate with the strength of so coupling . </S><S> the so coupling will suppress the kondo effect and restore the abrupt jumps of the persistent currents . </S><S> it is also found that a persistent spin current circulating the ring can exist even without the charge current in this system . </S>

1210.1321

the physics of atomic nuclei and nuclear systems is very rich but extremely complicated to describe .it unites many of the toughest aspects of the quantum many body problem in a single spot . to name only a few, we can mention the existence of four different types of fermions ( proton / neutron - spin up / down ) , the relevance of pairing and quartet correlations in the dynamics , or the role of the mean field approach as a first order approximation while quantum fluctuations are very important as a consequence of the smallness of nuclei .moreover , the nucleon - nucleon ( nn ) force is very complicated with its hard core , tensor ( dipolar ) components and strong spin orbit interactions .it is remarkable that nuclear theory has nowadays mastered a great deal of those challenges and many of the experimental data are very well described either on a phenomenological basis , or , more rarely , with microscopic approaches . in this workwe try to progress with the latter strategy .theories aiming towards a global description of low energy nuclear data always start with a mean field description and a semi - phenomenological nuclear energy density functional . in a recent paper @xcite , we proposed an approach slightly different from the usual for the establishment of a nuclear energy density functional .we tried to follow as much as possible the kohn - sham ( ks ) @xcite density functional theory ( dft ) , which is based on the hohenberg - kohn theorem @xcite , and is of common use , e.g. , in atomic and molecular physic @xcite .the ks method introduces a set of auxiliary single - particle states and takes for the kinetic energy the slater form in this basis . in the ks method the energy density functional is written as a sum of two terms .one of them corresponds to the uncorrelated kinetic energy density plus the direct contribution derived from the underlying two body ( coulomb ) interaction .the other piece contains an unknown term for the exchange and correlation energies .for a practical implementation of the method , the latter term has to be guessed and therefore many strategies have been developed along the years to describe different physical phenomena .a popular strategy in atomic and molecular physics is to use accurate theoretical calculations ( mainly by monte - carlo techniques ) as a function of the density in simplified models where finite size effects are neglected. the results of those calculations are used to make an educated guess of the unknown exchange and correlation part of the functional in the real system . in this waythe minimization procedure gives hartree - like equations for the single - particle states , where the potential part includes in an effective way the overall exchange and correlation contribution .the application of the ks - dft scheme to self - bound systems like atomic nuclei is not straightforward , because of the absence of a external confining potential .the usual procedure used with skyrme functionals has been justified recently @xcite by reformulating the problem in the intrinsic " system . in nuclear physics to build up the exchange and correlation contribution amounts to a detailed determination of the ground state of symmetric nuclear matter and of neutron matter to be used as a guide for finite nuclei energy density functionals .accurate nuclear matter calculations are very complicated and not as much advanced as for electron systems of condensed matter .nonetheless , in recent years , quite some progress has been achieved and we will build our ks - dft approach on the equation of state ( eos ) of baldo et al . , developed in refs .@xcite , which is based on the well known hole line expansion .this eos is calculated from realistic two body and three body interactions treated at the brueckner hartree - fock level .it is compatible with phenomenological constraints coming from data of heavy ion reactions @xcite as well as from the analysis of astrophysical observations @xcite .it is worth mentioning that only two works exist which tried to follow the same strategy as here , one by fayans @xcite and the other by steiner et al @xcite .however , they are based on a somewhat older eos @xcite .the functional obtained will be used to describe finite nuclei at the mean field level concentrating on the binding energy systematic . therefore , it is important to include beyond mean field effects having a strong impact on binding energies and not present in nuclear matter .the most obvious one is the correlation energy associated to symmetry restoration that will be accounted for in the case of translational and rotational invariance . a different approach to nuclear energy density functional ( edf )was developed on the basis of the chiral two- and three body forces @xcite .the final form of the edf contains a set of contact interactions , similar to the skyrme functionals , with a corresponding set of about a dozen of parameters .this part is supplemented by a contribution coming from pion exchanges , which is non - local and treated through the density matrix expansion procedure .preliminary calculations @xcite on a selected set of nuclei show a slight improvement with respect to the standard skyrme functionals and open a promising prospect for large - scale fitting procedures .the reason for such an intense focus on an accurate theoretical description of bulk properties of nuclei all over the nuclide chart is discussed for instance in @xcite : many physical scenarios involve neutron rich nuclei which are far from reach from an experimental point of view and therefore reliable theoretical predictions are the only possible option at present .the present status is that good agreement between different theoretical models is to be expected as long as we do not move towards the neutron drip line , a region of relevance in stellar nucleo - synthesis processes .therefore , a new model based on somewhat different ideas closer to the ks - dft approach than the more traditional skyrme - like edf ( inspired by contact central potentials ) , can help to clarify the uncertainties associated to present - day models . in the present work ,a ks approach based on the same microscopic bulk input as in @xcite will be used for the particle- hole channel .however , we will be able to reduce the number of parameters by two without losing accuracy .as before , we will consider an additional bulk parameter for a precise adjustment of the numerically obtained e / a value .however , for the surface , we will reduce the number of parameters from three to one by the condition that , in infinite symmetric and infinite pure neutron matter , the strength parameters of the finite range term reproduce the coefficients of the quadratic terms of the bulk energy density obtained from the microscopic calculation .therefore , no subtraction term in the finite range part as in @xcite is necessary .the range @xmath0 of the surface term remains the only adjustable parameter , see the more detailed discussion below . although there are , together with the strength of the spin - orbit term , three parameters in the particle - hole channel , we want to emphasize that the most relevant are the ones from bulk and surface .we stress this point , because it seems to us a reduction to two basic physical inputs to the binding energy of nuclei : energy per particle of infinite matter and surface energy ! that this is possible is as much surprising as it is gratifying !it also should be mentioned that the adjustment of those two parameters is extremely sensitive : both have to be fine tuned to the order of @xmath1 ! this sensitivity points to a well defined physical content of the parameters . it should be pointed out , however , that ks - dft addresses only the ground state and is , in principle , not tailored to describe excited states of the system .nevertheless , we will also use it for the description of the giant monopole resonances ( gmr s ) .the paper is organized as follows : the following section is devoted to briefly recall our previous bcp functional . in the third section , the new energy density functional built up in this paper is discussed .the results obtained with this improved functional , that we will call bcpm ( barcelona - catania - paris - madrid ) , are also presented in the same section .the predictive power of the bcpm functional regarding other observable such as charge radii , quadrupole and octupole deformations and fission barriers are discussed in the fourth section . the ability of the bcpm functional for describing the isoscalar giant monopole resonance is the subject of the fifth section .finally , the summary and conclusions are given in the last section .the former bcp functional was proposed in ref .this and subsequent refinements @xcite are based on the kohn - sham density functional theory ( ks - dft ) where the one body density @xmath2 plays a central role . in the ks - dft theory an auxiliary set of @xmath3 orthonormal wave functions @xmath4 , where @xmath3 is the mass number , is introduced to express formally the density as if it were obtained from a slater determinant as a sum of the product of single particle wave functions @xmath5 with the @xmath6 s determined from the minimization of the ground state energy . in condensed matter and atomic physicsthe energy density functional @xmath7 $ ] is usually split into two parts @xcite : @xmath8 + w[\rho].\ ] ] the first piece @xmath9 $ ] corresponds to the uncorrelated kinetic energy written in the usual manner as @xmath10 the other piece , @xmath11 $ ] , contains the potential energy and the correlated part of the kinetic energy . in the bcp family of functionalsthis part is given as the sum of spin - orbit energy , the coulomb part and the nuclear energy term which depends on the neutron and proton densities @xmath12 and @xmath13 , respectively .in nuclear physics , contrary to the situation in condensed matter and atomic physics , the contribution of the spin - orbit interaction to the energy functional is very important .non - local contributions have been included in dft in several ways long ago ( see ref .@xcite for a review of this topic ) . in line with this , we split the spin - orbit part into an uncorrelated part @xmath14 plus a remainder not treated explicitly here .the form of the uncorrelated spin - orbit part is taken exactly as in typical skyrme @xcite or gogny forces @xcite .another piece that we explicitly split off is the coulomb part @xmath15 .we treat this contribution at lowest order , i.e. the direct term plus the exchange contribution in the slater approximation , that is @xmath16 and @xmath17 with @xmath18 . the nuclear energy functional contribution @xmath19 $ ] contains the nuclear potential energy as well as additional correlations .this contribution is divided into a finite range term @xmath20 $ ] to account for correct surface properties and a bulk correlation part @xmath21 $ ] that we take from one of the most advanced microscopically determined eos existing so far in the literature @xcite as mentioned before .collecting all these contributions our final ks - dft functional reads @xmath22 the functional is supplemented by a density dependent pairing interaction and some beyond mean field corrections ( see below ) .one of the prominent results of our recent work within the ks - dft bcp density functional scheme @xcite was that , with a very reduced number of adjustable parameters to finite nuclei , rms values for binding energies and radii came out to be of the same quality as the ones of some skyrme @xcite or gogny @xcite functionals .the reduction of the number of parameters stemmed from the fact that , prior to a fit of the functional to nuclear masses , the bulk part of the functional @xmath23 was adjusted to the microscopic eos obtained in refs . @xcite by fitting polynomials in the total density @xmath24 to microscopic results in symmetric and pure neutron matter , followed by a quadratic interpolation in the asymmetry parameter @xmath25 between these two limits .the curve obtained in the polynomial fit can be slightly changed without changing significantly the mean square deviation to the microscopic eos . on the other hand , although the microscopic eos is state - of - the - art , the uncertainties in the underlying interaction and the ones attributed to the approximations involved translate to minor uncertainties in the computed points of the microscopic eos that will eventually lead to minor changes in the polynomial curve .we will exploit this limited freedom to produce polynomial fits with slightly different values of e / a at saturation ( around e / a=16 mev ) .the resulting edfs are then used in the finite nuclei fit ( see below ) and the one leading to the smallest binding energy rms is chosen . only changes of the order of a few hundred of kev in e / a are considered .the additional surface term in @xcite was of the form @xmath26 = & \frac{1}{2 } & \sum_{t , t'}\!\!\iint \!\ !d { \bf r } d { \bf r}'\rho_{t}({\bf r})v_{t , t'}({\bf r}-{\bf r}')\rho_{t'}({\bf r } ' ) \\ \nonumber & - & \frac{1}{2}\sum_{t , t'}\gamma_{t , t'}\int d { \bf r}\rho_{t}({\bf r})\rho_{t'}({\bf r})\end{aligned}\ ] ] where the index @xmath27 is the label for neutron and proton , i.e. @xmath28 , and @xmath29 the volume integral of @xmath30 .the subtraction term was introduced not to contaminate the bulk part , determined from the microscopic infinite matter calculation . for the finite range form factor @xmath30a simple gaussian ansatz @xmath31 was made .the strength parameters were chosen to distinguish only between like and unlike particles , i.e. @xmath32so that the finite range term contained three adjustable parameters : @xmath33 , and @xmath34 .together with the adjustment of the bulk and the spin - orbit strength , the functional in @xcite has all in all five adjustable parameters , even though they are not all to be considered on the same level of significance .we will come back to these considerations later in the text where we also will show how to avoid the subtraction term in the finite range part of the energy . in order to describe properly open shell nucleione has to introduce pairing correlations .the formal way of including pairing within the dft is also known since long from the generalization of the hohenberg - kohn theorem to paired systems @xcite . in @xcite we have introduced pairing in our functional through the bcs approach using a zero - range density dependent interaction adjusted to reproduce the neutron gaps of the gogny force in symmetric nuclear matter with @xmath35 @xcite .the free parameters of the energy density functional reported in @xcite were fitted either to reproduce the binding energies of a few spherical nuclei or binding energies and charge radii of the same set of spherical nuclei .we called these functionals bcp1 or bcp2 ( barcelona - catania - paris ) , respectively .the predictive power of these functionals was illustrated by computing the binding energies and charge radii of 161 spherical nuclei between @xmath36ne and @xmath37u .we found that the rms in energy and charge radii are similar to the ones obtained with successful mean - field models such us the gogny d1s force @xcite , the skyrme sly4 interaction @xcite and the relativistic mean field parametrization nl3 @xcite . in subsequent publications we explored the properties of bcp1 and bcp2 in describing quadrupole and octupole deformation ground state properties , fission , excited octupole states , etc @xcite . in dealing with deformed nucleithe most relevant beyond mean field effect , namely the rotational energy correction has been considered in an approximate way @xcite .also some other parametrizations of the functional based on fitting to the binding energy of deformed nuclei instead of the fitting to spherical nuclei , have been considered @xcite .we would like to point out that for other energy density functionals , like the ones of the skyrme , gogny , or relativistic mean field type , the number of parameters seems much higher , typically more than ten . however , many of those parameters are implicitly used to get a reasonable nuclear matter and neutron matter equation of state .the advantage of our ks - dft procedure is that one clearly separates the tasks of reproducing the nuclear matter eos from the most prominent finite size effects , namely the surface and coulomb energies and the spin - orbit potential .following @xcite we write the bulk part of the energy density functional as @xmath38=\int\ !d{\bf r } \big[p_{s}(\rho)(1-\beta^{2 } ) + p_{n}(\rho)\beta^{2}\big]\rho\label{eq : eq4a},\ ] ] where the interpolating polynomials for symmetric and pure neutron matter , @xmath39 and @xmath40 respectively , read @xmath41 where the reference densities are @xmath42=0.16 @xmath43 and @xmath44=0.155 @xmath43 , respectively .as compared to the former version of the bcp energy density functional , the fit of the microscopic eos has been redone in order to avoid some rather strong oscillatory behavior in the density - dependent incompressibility of the nuclear matter ( see section iv below ) when plotted as a function of density . to cure this unwanted effect ,the number of theoretical points is increased and a fifth order polynomial in the density is chosen for fitting the microscopic eos in symmetric nuclear matter in a wider range of densities , up to 0.625 @xmath43 , instead of the prescription given in @xcite , see also @xcite .furthermore special care has been paid to the smoothness not only of the fitted eos but also of the density - dependent incompressibility .this has been achieved by considering only polynomial fits with smooth high order derivatives with respect to the density . in fig .[ fig : eos ] the microscopic eos for nuclear and neutron matter as well as the corresponding polynomial fits are shown as function of the density . in the lower panel ,the symmetry energy computed using the polynomial fits as a function of the density is also displayed . as discussed in @xcite , the low density behavior of the microscopic nuclear matter eoshas a characteristic trend , usually not reproduced by skyrme and gogny functionals ( see also ref .@xcite ) , missing there quite a substantial part of binding .since we want to construct the ks - dft functional on the basis of the microscopic calculations , the bulk part @xmath23 of the functional , directly related to a realistic eos , is determined once and for all as in eq .( [ eq : eq4a ] ) together with the local density approximation .however , as mentioned before slightly different polynomial interpolations with different values of e / a at saturation have been considered to improve the finite nuclei results for the binding energies .the saturation density has been kept fixed at the nominal value of 0.16 @xmath43 . for e / a at saturationwe have explored the interval between 15.97 and 16.03 mev in steps of 0.01 mev . for each value of e/ a the complete fitting process for finite nuclei has been performed ( see below ) .the optimal value for the saturation energy per particle turns out to be 15.98 mev .the fine tuning of e / a at saturation is to be expected because of the extreme sensitivity of the mass of heavy nuclei to this parameter .in addition , as the fine tuning involves finite nuclei calculations where the surface term is included it can be considered as a way to account for the coupling of volume and surface in finite nuclei ..parameters ( in mev ) of the polynomial fits , @xmath45 and @xmath46 for symmetric and neutron matter , respectively .the reference densities are @xmath47 @xmath43 and @xmath48 @xmath43 .two sets of parameters for symmetric matter are given , they correspond to eos with a minimum at the given values of @xmath49 .the parameters are given with six digits , enough to obtain binding energies with an accuracy of a couple of ev .the value @xmath50 mev has been used .[ tab : polpar ] [ cols="^,^,^,^",options="header " , ] experimental information about the excitation energy of isoscalar giant resonances in medium and heavy nuclei is obtained through inelastic scattering of @xmath51 particles of few hundreds of mev measured at extremely forward angles . the multipole decomposition analysis allows to extract from the scattering data the strength associated to different multipolarities , in particular those corresponding to monopole and quadrupole . from these strengths ,the position and width of the isgmr and isgqr peaks can be determined @xcite .we analyze first the excitation energies of isgmr ( @xmath52 ) of several nuclei for which the experimental values are known @xcite .we compare these values with the theoretical predictions @xmath53 and @xmath54 provided by the bcpm energy density functional .the relevant information about these excitations energies is collected in tables [ t1 ] and [ t2 ] .as far as we are including in our study tin and cadmium ( neglecting the effects of the small deformation ) isotopes and @xmath55sm , for which pairing correlations are essential to describe accurately the ground state , we discuss first the role of pairing in the sum rule approach used to estimate @xmath52 .the generalization of rpa to the case of nuclei with open shells when pairing correlations are active , is the quasi - particle rpa ( qrpa ). the so - called dielectric theorem , which justifies constrained calculations within qrpa to obtain the @xmath56 sum rule , has been proven recently by col and collaborators in @xcite .also khan et al have shown the validity of the thouless theorem for the energy weighted sum rule in the case of self - consistent qrpa based on hfb @xcite . as far as a simple generalization of the thouless theorem allows to obtain the @xmath57 sum rule @xcite, it can be expected that this also will be the case in the qrpa framework .in addition we have also numerically checked that the virial theorem , i.e. the stability against scale transformations of the wave - functions in eq .( [ eq2 ] ) , is also fulfilled in the hfb approximation . from this discussionwe may conclude that the sum rule approach can also be applied in the qrpa case . in table[ t1 ] we display the theoretical scaled and constrained estimates ( including pairing ) of @xmath52 of the nuclei @xmath55sm , @xmath58pb , @xmath59zr , @xmath60sn , and @xmath61cd together with the corresponding experimental values . these estimates are given by @xmath53 and @xmath54 , respectively , defined previously , which , as mentioned , fulfill @xmath62 ( see appendix ) .the escape width ( [ eq21 ] ) lies in a window of 2 - 3 mev , pointing out that the monopole strength of the considered nuclei is mainly concentrated around well defined peaks .the influence of pairing correlations on @xmath52 has been discussed in earlier literature ( see @xcite and references therein ) .we have repeated the scaled and constrained estimates of the @xmath52 switching off pairing and the corresponding results are reported in table [ t2 ] . comparing tables [ t1 ] and [ t2 ]we notice that , when pairing correlations are taken into account ( table [ t1 ] ) , the @xmath54 and @xmath53 values in open shell nuclei slightly decrease as compared with the values obtained without pairing ( table [ t2 ] ) which in agreement with the results of refs .the effect of pairing correlations is , in general , more important in the constrained calculations than in the scaled ones , especially in half - filled major shell nuclei as it is the case of @xmath63sn , @xmath64sn , @xmath65sn , @xmath66cd , @xmath67cd , @xmath63cd .when the occupancy of the shell increases , as it happens in @xmath68sn , @xmath69sn , @xmath70sn , @xmath71sn , @xmath64cd and @xmath65cd , the influence of pairing is actually very small as it can be seen in table [ t2 ] . in order to compare with the experiment , we will use as representative the constrained energy @xmath54 , which estimates the energy of the centroid in a very precise way at least in the case of the nucleus in @xmath58pb @xcite ( see also table iv of @xcite in this respect ) . from table[ t1 ] , we see that the experimental @xmath52 in @xmath58pb is very well reproduced by @xmath54 obtained using the bcpm energy density functional .the experimental excitation energies of @xmath59zr and @xmath55sm are slightly overestimated by the theoretical constrained calculation. however , our theoretical calculation are less successful in the case of the cadmium and tin isotopes .the predicted @xmath54 values are shifted by around 1 mev up as compared with the experimental results @xcite .this is a general tendency shown by mean - field models of different nature @xcite .several attempts to explain this behavior have been proposed in the past such as the density dependence of the incompressibility in neutron rich matter @xcite or different types of pairing interactions @xcite . however , no clear explanation has been found to date and the question why sn ( and cd ) isotopes are so soft is still an open problem .it may be that nonlinear effects coming from particle vibration coupling play a role in such nuclei as sn isotopes , since we reproduce the stiff doubly magic nucleus of @xmath58pb and not the semi - magic ones as , e.g. , the sn isotopes . from the previous discussion it seems that the @xmath52 estimated with the bcpm functional are somewhat high as compared with the experiment in spite of its relatively low incompressibility modulus @xmath72=216 mev ( see table [ nmatter ] ) .the fact that effective mean field models with different values of @xmath72 can give similar values of @xmath52 in @xmath58pb has actually been a general puzzle in the past ( see @xcite and references therein ) .the reason is the proportionality between @xmath52 and the square root of @xmath73 found by blaizot @xcite for gogny forces .for example rmf approaches systematically lead to @xmath52 comparable to the ones obtained with non - relativistic theories like skyrme or gogny , in spite of the fact that @xmath73 values are appreciably larger in relativistic than in non relativistic approaches .some time ago col et al .@xcite showed that one can construct skyrme functionals that give reasonable values for @xmath52 but with relatively high values of @xmath73 i.e. values which approach the ones of rmf theories .as discussed in @xcite , the underlying reason of this fact can be attributed to the different density dependence of the symmetry energy at , and around , saturation . in the same reference @xciteit is concluded that within non - relativistic models there is not a unique relation between the value of @xmath73 associated to a given model and the energy and the @xmath52 predicted by the same model . to clarify this latest problem a very recent proposal of khan , margueron and vidaa @xcite may be useful .these authors analyzed a relatively large set of skyrme , gogny and rmf interactions finding that the corresponding density dependent incompressibilities @xmath74 defined in eq .( [ incomp ] ) show a crossing point at a density of about 75% of the saturation density .this crossing density corresponds roughly to the average density of a heavy nucleus defined as _ av = ^2 ( ) d .the authors claim that the excitation energy of finite nuclei is more properly correlated with the slope of incompressibility @xmath74 computed at the crossing density than with the incompressibility modulus computed at saturation density @xmath72 .this scenario is similar to the one found for the nuclear symmetry energy in bulk matter , which also shows a crossing point for a similar value of the density ( @xmath750.11 @xmath43 ) @xcite . in this caseit is also found @xcite that the slope of the symmetry energy at the crossing density is linearly correlated with the neutron skin of a heavy nucleus , e.g. @xmath58pb . in general , the existence of a crossing pointis actually related to the fact that the parameters of successful mean- field models have been fitted to reproduce properties of finite nuclei whose densities are , on the average , smaller than the saturation density .therefore , if an observable measured in finite nuclei is related to some property of nuclear matter , the correlation is better described at the average ( crossing ) density than at the saturation one @xcite . following the suggestion of ref .@xcite , we have analyzed the behavior of @xmath52 of @xmath58pb and @xmath59zr calculated using different mean field models as a function of the parameter @xmath76 defined as m = 3 _ _ av , which is directly related to the slope of the density dependent incompressibility @xmath74 at the crossing density @xmath77 . in our study in addition of the bcpm energy density functional , we have also considered the d1 and d1s gogny forces , the skm@xmath78 and t6 skyrme interactions and the rmf parametrizations nl1 , nl3 and nlsh . using these models the correlation between the scaled and constrained estimates of @xmath52 s and the parameter @xmath76are displayed in fig .[ f2 ] for the two aforementioned nuclei .the data of the @xmath52 predicted by gogny forces and rmf parametrizations are taken from refs . @xcite and @xcite respectively . from this figureit can be seen that the @xmath52 follows a linear trend with the parameter @xmath76 ( the correlation parameter is r=0.985 in all the considered cases ) with high accuracy , in agreement with the suggestion of @xcite . consequently , and as it is claimed in @xcite , the density dependence of @xmath74 can be constrained by the @xmath52 measured in finite nuclei .the same correlation is also fulfilled by the nucleus @xmath59zr with @xmath76 evaluated at the same average density .this fact points out that the dependence on the mass number of the average density is actually rather weak for medium and heavy nuclei @xcite . from the lower panel of this figurewe also see explicitly that considering globally mean field models of different nature , the linearity between @xmath52 and the square root of the incompressibility modulus at saturation @xcite is lost in some cases , in agreement with the conclusion of ref .@xcite . to investigate this aspect in more detail , we plot in the upper panel of fig .[ f3 ] the bulk incompressibility of eq .( [ incomp ] ) as a function of the density . in the lower panelwe show , magnified , the relevant region of the upper panel . for densities close to 0.72@xmath42 the density dependent incompressibility @xmath74 computed with skyrme and gogny forces and the rmf takes similar values around 40 mev .however , the bcpm functional predict relatively larger values of @xmath74 at this average density .the reason of that lies , probably , in the fact that @xmath74 grows more linearly with the density than the other considered models .as far as the slope of @xmath74 has a value consistent with @xmath52 , as it can be seen in fig .[ f2 ] , the value of @xmath74 at 0.72 @xmath42 is larger than the value predicted by the other considered mean field models . from the analysis of fig .[ f3 ] one derives two important conclusions .first , the crossing density actually disappears when one considers mean field models of different nature .what is actually relevant , is to compute the slope of @xmath74 at _ the average density_. the same happens in the analysis of the symmetry energy using different mean field models @xcite .second , the small differences of @xmath74 at the average density may explain why @xmath79 and @xmath52 are not always linearly correlated .this happens , e.g. , with the nl1 parametrization with @xmath80=211 mev , a value similar to the one of the skm@xmath78 force but predicting @xmath52 clearly smaller , or the case of bcpm discussed previously . in what concerns the isgqr , we are well aware of the fact that a theory with @xmath81 , as is the case of the bcpm functional , underestimates the collective quadrupole excitation energy . in table [ t1 ] we display the scaled energies of the quadrupole excitation computed with bcpm for several nuclei together with the available experimental values @xcite .insofar as the focus with a ks - dft approach is on ground state properties , we may not worry about the failure in the estimate of the isgqr .in general , for the description of excitation energies , the ks - dft should be generalized to include non - localities .attempts in this direction exist in condensed matter. it may be worth to look into that in the future also in the nuclear context .in this work we further elaborated on the previously established kohn - sham dft functional , bcp @xcite , based on advanced microscopic nuclear and neutron matter calculations .the bcp contained by large 5 adjustable parameters , three for finite size , one for spin - orbit and one for bulk .we reduced this number by approximately a factor of two coming up with a ks - dft functional containing only three adjustable parameters without deteriorating the excellent results .one of these parameters , the strength of the spin - orbit force , has the usual value , the results being quite insensitive to its variation within a relatively large margin .we would like to call this a rather weak , i.e. almost predetermined parameter ( in addition it may be possible to extract it from the g - matrix @xcite ) .the other two parameters , on the contrary , are fixed within extremely tight margins of the order of 10@xmath82 .they concern the bulk energy per particle , e / a , on the one hand and the surface energy on the other hand . indeed in the polynomial fit to the microscopically determined eos ,a single fine tuning parameter had to be introduced in order to get a minimal rms value for nuclear masses as well as a drift free ( i.e. scatter around zero ) difference of theoretical and experimental binding energies as a function of mass number .this parameter turned out to pin down the binding energy to the value , e / a = 15.98 mev .we emphasize that the fourth digit is significant here .for the second parameter , we made the simplifying ansatz to introduce a finite range gaussian convoluting the two densities in the @xmath83 terms of nuclear and neutron matter fits , that is @xmath84 .the pre - factor is determined from the bulk , so that the only free parameter is the range @xmath34 which evidently is responsible for the surface energy .the minimization of the rms of masses ( see main text for details ) again is determined within the very narrow margin of 10@xmath82 and turned out to have the value @xmath85 fm . with this , a rms value for the masses of 1.58 mev for 579 nuclei and of 0.027 fm for the radii with 313 nuclei is obtained .those values are as good as , e.g. , the ones of the gogny d1s and d1 m forces for the same set of nuclei .this very encouraging result calls for some comments .it is , indeed , quite surprising but on the other hand very satisfying , that the two basic nuclear quantities , energy per particle of the bulk and surface energy essentially suffice to determine a ks - dft functional with excellent ground state properties of nuclei .we should point out that the ks - dft approach is tailored for the ground state and that a time dependent extension is not evident .we nevertheless tried to determine at least the gmr energies with the sum rule technique and found that the results are in the same ball park as the ones of other skyrme , relativistic , or gogny type approaches . on the other handthe energies of the gqr seem to be underestimated systematically by about 1 mev .this is not so surprising , since a particularity of the ks - dft approach is that the bare mass is used while it is known that the gqr is sensitive to the value of the effective mass .future developments concern eventually the introduction of an effective density dependent mass @xmath86 .whether this is possible with results as satisfying as here , remains to be seen . a further point which needs to be improvedis pairing . at the momentwe treat it at a rather elementary level with a zero range force and cut off. a finite range force of the gogny type in the pairing channel would certainly be preferable .promising developments in this direction have recently been published @xcite . at the endlet us mention that our theory fulfills all the eleven criteria put forward in @xcite besides one .our functional does not agree with the constraint concerning the value of the volume isospin incompressibility coefficient that appears in the leptodermous expansion of the finite nucleus incompressibility @xcite .this quantity is defined as @xmath87 the bcpm functional predicts @xmath88=-195.2 mev which is outside of the range proposed in @xcite , -760 @xmath89 -372 mev .however , two comments are in order .first the extraction of this coefficient from the experimental excitation energies of the gmr in finite nuclei @xcite is not a simple task at all and may give erroneous estimates of the coefficients of the leptodermous expansion of the finite nucleus incompressibility as it has been discussed in the past ( see for example refs .second , there are other estimates of the @xmath88 coefficients available in the literature , as for instance in ref .@xcite , that predict -370 @xmath90 120 mev which embraces the prediction of the bcpm functional .in general , the criteria and their limitations in @xcite can certainly be subjected to debate .for example there is no reason to exclude functionals with @xmath35 as we have demonstrated in this work .the fact that our results comply with the criteria of @xcite is an indication that nuclear density functionals should be based as much as possible on a solid microscopic input . in this respectit is to be noted that out of the five functionals which survived the criteria in @xcite , three are also based on a microscopic input .we are grateful to p. danielewicz for useful discussions .we would like to thank n. schunck for a careful reading of the manuscript and many useful suggestions .work supported in part by micinn grants nos . fpa2009 - 08958 , fis2009 - 07277 , and fpa2008 - 03865-e / in2p3 and by the consolider - ingenio 2010 program cpan csd2007 - 00042 and multidark csd2009 - 00064 .x. v. also acknowledges the support from fis2008 - 01661 ( spain and feder ) and 2009sgr-1289 ( spain ) .support by compstar , a research networking programme of the european science foundation is also acknowledged .references m. baldo , p. schuck and x. vias , phys .* b663*,390 ( 2008 ) .w. kohn and l.j .sham , phys .rev . * 140 * , a1135 ( 1965 ) .p. hohenberg and w. kohn , phys .rev . * 136 * , b864 ( 1964 ) .r. o. jones and o. gunnarsson , rev .phys . * 61 * , 689 ( 1989 ) .h. eschrig , _ the fundamentals of density functional theory _ ( b. g. teubner , stuttgart , 1996 ) .j. p. perdew and k.schmidt , _ density functional theory and its applications to material _ , v. van doren et al .cp577 ( american institute of physics , melville ny , 2001 ) .a. e. mattsson , science * 298 * , 759 ( 2002 ) .jianmin tao , j. p. perdew , v. n. staroverov , and g. e. scuseria , phys .lett . * 91 * , 146401 ( 2003 ) .jianmin tao and j. p. perdew , phys .rev . lett.*95 * , 196403 ( 2005 ) .j. messud , m. bender , and e. suraud phys . rev . *c80 * , 054314 ( 2009 ) .j. engel , phys . rev . * c75 * , 014306 ( 2007 ) .n. barnea , phys .c76 , 067302 ( 2007 ) .m. baldo , i. bombaci and g.f .burgio , astronomy and astrophysics * 328 * , 274 ( 1997 ) .m. baldo , a. fiasconaro , h.q .song , g. giansiracusa , and u. lombardo , phys .* c 65 * , 017303 ( 2001 ) .m. baldo , g. giansiracusa , u. lombardo and h.q .song , phys .* b 473 * , 1 ( 2000 ) .zhou , g.f .burgio , u. lombardo , h .- j .schulze and w. zuo , phys .* c 69 * , 018801 ( 2004 ) .li , u. lombardo , h .- j .schulze , w. zuo , l.w .chen and h.r .ma , phys .rev . * c74 * , 047304 ( 2006 ) .m. baldo , c. maieron , p. schuck and x. vias , nucla736 * , 241 ( 2004 ) .m. baldo and c. maieron , j. of phys .* g34 * , r1 ( 2007 ) .m. baldo and g.f .burgio , rep.prog.phys .* 75 * , 026301 ( 2012 ) .g. taranto , m. baldo and g.f .burgio phys .rev . * c87 * , 045803 ( 2013 ) . s.a .fayans , jetp lett . *68 * , 169 ( 1998 ) .steiner , m. prakash , j.m .lattimer and p.j .ellis , phys .rep . * 411 * , 325 ( 2005 ) b. friedman and v.r .pandharipande , nuc . phys . *a361 * , 502 ( 1981 ) ; r.b .wiringa , v. fiks and a. fabrocini , phys . rev . *c38 * , 1010 ( 1988 ) .holt , n. kaiser , and w. weise , phys . rev .* c 81 * , 024002 ( 2010 ) .holt , n. kaiser and w. weise , eur .j. * a47 * , 128 ( 2011 ) .n. kaiser , eur .j * a48 * , 36 ( 2012 ) .b. gebremariam , t. duguet , and s. k. bogner phys . rev .* c 82 * , 014305 ( 2010 ) .m. stoitsov , m. kortelainen , s. k. bogner , t. duguet , r. j. furnstahl , b. gebremariam , and n. schunck phys . rev . *c 82 * , 054307 ( 2010 ) .j. erler , n. birge , m. kortelainen , w. nazarewicz , e. olsen a. perhac , and m. stoitsov , nature * 486 * , 509 ( 2012 ) .m. baldo , l. robledo , p. schuck and x. vias , j. phys .g , nucl . and part .physics * 37 * , 064015 ( 2010 ) .j. engel , lecture notes in physics * 620 * , 56 ( 2003 ) , c. fiolhais , f. nogueira and m. marques eds .( springer - verlag , berlin , 2003 ) .e. chabanat , p. bonche , p. haensel , j. meyer and r. schaeffer , nucl . phys . * a 635 * , 231 ( 1998 ) ; nucl. phys . * a 643 * , 441 ( 1998 ) .s s goriely , n n chamel and j m jm pearson , phys .lett . * 102 * , 152503 ( 2009 ) .j. decharg , d. gogny , phys .rev . * c21 * , 1568 ( 1980 ) . j.f .berger , m. girod , d. gogny , nucl .phys , * a428 * , 23c ( 1984 ) .oliveira , e.k.u .gross and w. khon , phys . lett . * 60 * , 2430 ( 1988 ) .e. garrido , p. sarriguren , e. moya de guerra , and p. schuck , phys .rev . * c60 * , 064312 ( 1999 ) . g.a .lalazissis , j. konig , and p. ring , phys . rev .* c55 * , 540 ( 1997 ) .robledo , m. baldo , p. schuck , and x. vias , phys . rev . *c77 * , 051301(r ) ( 2008 ) .x. vias , l.m robledo , p. schuck , and m. baldo , intl .j. of mod .e18 * , 935 ( 2009 ) .robledo , m. baldo , p. schuck , and x. vias , phys .* c81 * , 034315 ( 2010 ) .j.l egido and l.m .robledo , lecture notes in physics * 641 * , 269 ( 2004 ) .a. akmal , v. r. pandharipande and d. g. ravenhall , phys* c58 * , 1804 ( 1998 ) .robledo , intl .j. of mod .e16 * 337 ( 2007 ) .robledo , jour . of phys.*g37* 064020 ( 2010 ) .j. margueron , e. van dalen and c. fuchs , phys . rev . *c76 * , 034309 ( 2007 ) .e. khan , j. margueron , g. col , k. hagino and h. sagawa phys .rev . * c82 * , 024322 ( 2010 ) . w.d .myers and w.j .swiatecki , ann .( n.y . ) * 55 * , 395 ( 1969 ) .o. bohigas , a.m. lane and j. martorell , phys . rep . * * 5**1 , 267 ( 1979 ) .blaizot , phys . rep . * 64 * 171 ( 1980 ) .m. dutra , o. loureno , j.s .s martins , a. delfino , j.r . stone and p.d .stevenson , phys . rev . *c85 * , 035201 ( 2012 ) .tsang et al . , arxiv 1204.0166 .brown , phys .lett . * 85 * , 5296 ( 2000 ) .m. warda , x. vias , x. roca - maza and m. centelles , phys . rev . *c80 * 024316 ( 2009 ) .m. centelles , x. roca - maza , x. vias and m. warda , phys .lett . * 102 * , 122502 ( 2009 ) .nayak , j.m .pearson , m. farine , p. gleissl , and m. brack , nucl .a516 * , 62 ( 1990 ) .chossy and w. stocker , phys . rev .* c56 * , 2518 ( 1997 ) .t. niksic , d. vretenar and p. ring phys . rev . *c78 * , 034318 ( 2008 ) . p. ring and p. schuck , _the nuclear many body problem _ , ( springer , berlin ) ( 1980 ) .s. hilaire , m. girod , eur .j * a33 * , 237 ( 2007 ) .robledo and g.f .bertsch , phys .rev c84 , 014312 ( 2011 ) . c. titin - schnaider and p. quentin , phys* b * 49 , 397 ( 1974 ) .m. anguiano , j. l. egido , and l. m. robledo , nucl . phys . * a * 683 , 227 ( 2001 ) .butler , d.w.l .sprung , j. martorell nucl .phys . * a * 422 , 157 ( 1984 ) .soubbotin , v.i .tselyaev , x. vias phys . rev .* c 67 * , 014324 ( 2003 ) .g. audi and a.h .wapstra , nucl . phys . *a729 * , 337 ( 2003 ) .m.centelles , m.del estal , and x.vinas , nucl.phys .* a635 * 193 ( 1998 ) .gogny d1s web mass table _ http://www-phynu.cea.fr/hfb-gogny.htm _g. f. bertsch , b. sabbey , and m. uusnkki , phys . rev .* c 71 * , 054311 ( 2005 ) .f. chappert , m. girod , and s. hilaire , phys . lett . *b668 * , 420 ( 2008 ) .s. goriely , s. hilaire , m. girod , and s. per , phys . rev . lett . *102 * , 242501 ( 2009 ) .robledo and g.f .bertsch , phys .rev * c84 * , 054302 ( 2011 ) .m. kortelainen , t. lesinski , j. mor , w. nazarewicz , j. sarich , n. schunck , m. v. stoitsov , and s. wild , phys . rev .* c 82 * , 024313 ( 2010 ) .n. nikolov , n. schunck , w. nazarewicz , m. bender , and j. pei , phys . rev .* c 83 * , 034305 ( 2011 ) .m. kortelainen , j. mcdonnell , w. nazarewicz , p .-reinhard , j. sarich , n. schunck , m. v. stoitsov , and s. m. wild , phys . rev .* c 85 * , 024304 ( 2012 ) .p. klpfel , p .-reinhard , t.j .brvenich , and j. a. maruhn phys .rev * c79 * , 034310 ( 2009 ) .reinhard and b.k .agrawal int .jour . of mod .e20 * , 1379 ( 2011 ) .i. angeli , at .data nucl .data tables * 87 * , 185 ( 2004 ) .f. buchinger et _ al .rev . * c41 * , 2883 ( 1999 ) .p. campbell et _ al .lett . * 89 * , 082501 ( 2002 ) . c. amsler et _ al ._ ( particle data group ) , physb667 * , 1 ( 2008 ) .m. dupuis , s. karataglidis , e. bauge , j.p .delaroche , and d.gogny , phys . rev .* c 73 * , 014605 ( 2006 ) .mang , phys .rep . * 18 * , 325 ( 1975 ) .robledo , m. baldo , p. schuck and x. vias , jour . of phys . : conference series * 321 * , 012015 ( 2011 ) .m. warda , j.l .egido , l.m .robledo and k. pomorski , phys . rev . * c66 * , 014310 ( 2002 ) . m. hunyadi et _al _ , physb505 * , 27 ( 2001 ) .youngblood , y .- w .lui , h.l .clark , b. john , y. tokimoto and x. chen , phys .rev . * c69 * , 034315 ( 2004 ) . d.h .youngblood , y .- w .lui , b. john , y. tokimoto , h.l .clark and x. chen , phys .rev . * c69 * , 054312 ( 2004 ) .t. li et al . ,lett . * 99 * , 162503 ( 2007 ) ; phys . rev . *c81 * , 034309 ( 2010 ) .d. patel et al . , arxiv:1209.0681 .l. capelli , g. col and jun li , phys . rev . *c79 * , 054329 ( 2009 ) .e. khan , n. sandulescu , m. grasso and n. v. giai , phys .rev . * c66 * , 024309 ( 2002 ) .v. tselyaev , j. speth , s. krewald , e. litvinova , s. kamerdzhiev , n. lyutorovich , a. avdeenkov and f. grmmer , phys . rev . *c79 * , 034309 ( 2009 ) .g. col , n. v. giai , j. meyer , k. bennaceur and p. bonche , phys . rev . *c70 * , 024307 ( 2004 ) .e. khan , phys .rev . * c80 * , 011307 ( 2009 ) .g. col and n. v. giai , nucl . phys . *a731 * , 15 ( 2004 ) .d. vretenar , t. niksic , and p. ring phys . rev . *c68 * , 024310 ( 2003 ) .j. piekarewicz and m. centelles , phys . rev .* c79 * , 054311 ( 2009 ) .blaizot , j.f .berger , j. decharg and m. girod , nucl .a591 * , 435 ( 1995 ) .e. khan , j. margueron and i. vidaa , phys .lett . * 109 * , 092501 ( 2012 ) .brown , phys .lett . * 85 * , 5296 ( 2000 ) .m. centelles , x. roca - maza , x. vias and m. warda , phys .lett . * 102 * , 122502 ( 2009 ) .soubbotin , v.i .tselyaev and x. vias , phys . rev .* c69 * , 064312 ( 2004 ) .patra , x. vias , m. centelles and m. del estal , nucl .a703 * , 240 ( 2002 ) .lui , d. h. youngblood , y. tokimoto , h. l. clark , and b. john phys . rev . *c69 * , 034611 ( 2004 ) .brieva , j.r .rook , nucl . phys . * a 297 * 206 ( 1978 ) .m. kohno , arxiv:1209.5048 .j. dukelsky , s. lerma h. , l. m. robledo , r. rodriguez - guzman , and s. m. a. rombouts phys .c 84 , 061301 ( 2011 ) .patra , m. centelles , x. vias , and m. del estal , phys . rev . * 65 * , 044304 ( 2002 ) . l.w .chen , b.j .cai , c.m .ko , bao - an li , c. shen , and j. xu , phys . rev . * 80 * , 014322 ( 2009 ) .m. centelles , x. vias , s.k .patra , j.n. de and tapas sil , phys . rev . *c72 * , 014304 ( 2005 ) .p. gleissl , m. brack , j. meyer and p. quentin , ann .( n.y . ) * 197 * , 205 ( 1990 ) .in this appendix we briefly summarize the basic theoretical aspects of the sum rule approach to obtain moments of the rpa strength functions . for more detailswe refer the reader to refs .@xcite .in the sum rule approach it is enough to know few low energy weighted moments of @xmath91 ( the so - called sum rules ) to estimate the excitation energies of the giant resonances .the sum rules are defined as m_k = _n e_n^k | n |q| 0 |^2 , [ eq01 ] where @xmath92 is the operator associated to the excitation , @xmath93 and @xmath94 are the exact ground and excited states respectively and @xmath95 the excitation energies . if @xmath96 is an odd integer , the sum rules @xmath97 can be computed as the expectation values in the ground state @xmath98 of some commutators which involve the hamiltonian @xmath99 and the operator @xmath92 ( assumed to be a hermitian and one - body operator ) @xcite , as for example m_1 = 0 | | 0 [ eq06 ] and m_3 = 0 | | 0 , [ eq07 ] from where an average excitation energy can be estimated as the full calculation of the sum rules is still a complicated task because the exact ground - state wavefunction is , in general , unknown .however , if the moments are computed within the 1p1h rpa , it is possible to replace the exact ground - state wavefunction by the uncorrelated hf one to obtain the sum rules ( [ eq06 ] ) and ( [ eq07 ] ) @xcite . in spite of all these simplifications ,the calculation of the sum rules is only easy to handle in few particular cases , namely the isgmr and isgqr ones . in these cases the sum rules needed to estimatethe average excitation energy of these resonances can be simplified as we will discuss below .for the isoscalar monopole and quadrupole oscillations the corresponding excitation operators are taken as q_m = _i=1^a r_i^2 q_q = _ i=1^a ( r_i^2 -3 z_i^2 ) .[ eq02 ] the underlying effective interaction associated to the bcpm energy density functional commutes with the operator @xmath92 and , therefore , the only contribution to the commutator @xmath100 $ ] in eq .( [ eq06 ] ) comes from the kinetic energy operator .consequently @xcite m_1 = a r^2 ( l=0 ) m_1 = a r^2 ( l=2 ) , [ eq09b ] where the expectation value of the operator @xmath101 is calculated with the hf wave functions .the direct evaluation of the commutators entering in eq . ( [ eq07 ] ) to compute the @xmath57 sum rule is a rather cumbersome task , that in the case of the monopole and quadrupole oscillations can be avoided with the help the so - called scaling method @xcite .we start from the scaled ground - state wave function that in the case of the monopole reads _^m = ^3 /2 _ 0(x , y , z ) , [ eq2 ] where @xmath102 is an arbitrary scaling parameter . in this casethe @xmath57 sum rule can be expressed by means of the second derivative of the scaled ground - state energy @xcite , that is , m_3 = _ = 0 .[ eq3 ] in the scaling approach the bcpm energy density functional reads e ( ) = t ( ) + e^_int ( ) + e^fr_int ( ) + e_c ( ) + e^s.o ( ) , [ eq6a ] where the scaled kinetic , spin - orbit and coulomb energies can be found in ref .the scaled bulk contribution can be easily obtained from the scaling of the particle density ( @xmath103 ) .the scaling of the finite range contribution has been described in refs .it is easy to show that the scaling of the hartee term reduces to a renormalization of the range @xmath104 of the interaction , or in other words that the scaled relative coordinate is @xmath105 . in this way ,the second derivative with respect to the scaling parameter @xmath102 , needed to compute the @xmath57 sum rule , finally reads @xcite : _ = 1 = d d ( ) ( ) [ eq6d ] in the bcpm energy density functional , the form factor @xmath106 is of gaussian type with a range @xmath104 . in this casethe second derivative of the scaled finite range contribution can be recast as @xcite _ = 1 = ^2 .[ eq6e ] finally the @xmath57 for the monopole oscillation computed with the bcpm energy density functional can be written as @xmath107\end{aligned}\ ] ] as explained in detail in ref .@xcite , a pure hartree term scales in the quadrupole case as e^h_q ( ) = d d ( ) ( ) v(s_q , ) , [ eq7d ] where @xmath108 .after some algebra the surface contribution to the @xmath57 sum rule in the scaling approach for the quadrupole oscillation can be written as ( see ref.@xcite for details ) _ = 1 = d d ( ) ( ) , that in the case of the bcpm functional owing to the gaussian character of the form factor @xmath109 can also be written as : _ = 1 = .this is due to the fact of the volume conservation in the quadrupole scaling , the bulk interaction part of the bcp energy density functional does not contribute to the @xmath57 sum rule in this case .the contribution from kinetic and spin - orbit energies can be easily obtained from the scaling transformation of the single - particle wave functions assuming a spherically symmetric ground - state @xcite .the coulomb contribution can also be obtained from eq .( [ eq7d ] ) using the coulomb form factor @xmath110 @xcite . collecting all these contributions , the @xmath57 sum rule for the quadrupole oscillation can finally be written as @xmath111 \right].\end{aligned}\ ] ] once the @xmath57 sum rule has been obtained through the scaling method , an estimate of the excitation energy of the isgmr and isgqr can be calculated with the help of eqs .( [ eq04 ] ) and([eq09b ] ) e^s_m = .[ eq19a ] let us now consider a nucleus , described by a hamiltonian @xmath99 , under the action of a weak one - body field @xmath112 .assuming @xmath113 sufficiently small , so that the perturbation theory holds , the variation of the expectation value of @xmath92 and @xmath99 are directly related to the @xmath56 moment @xcite , m_-1 = _ n = - _ = 0 = _ = 0 .[ eq12b ] therefore at rpa level the average energy of the giant resonances can also be estimated performing constrained hf calculations , i.e. looking for the hf solutions of the constrained hamiltonian h ( ) = h - q [ eq17 ] where @xmath92 is the collective monopole ( quadrupole ) operator defined previously ( see eq .( [ eq02 ] ) . from the constrained hfwave - function @xmath114 solution of ( [ eq17 ] ) , the rpa @xmath56 moment can be written as @xmath115_{\eta=0 } \\\nonumber & = & \frac{1}{2 } \left [ \frac{\partial^2}{\partial \eta^2 } \langle \phi(\eta)\vert h \vert \phi(\eta ) \rangle \right]_{\eta=0},\end{aligned}\ ] ] from where another estimate of the average energy of the isoscalar giant resonances is given by |e_1 = .[ eq19 ] due to the fact that the rpa moments fulfill @xmath116 @xcite , the average energies @xmath53 and @xmath54 values are an upper and lower bound of the mean energy of the resonance and their difference is related with the variance of the strength function ( resonance width ) = . [ eq21 ]

<S> a new version of the barcelona - catania - paris ( bcp ) energy functional is applied to a study of nuclear masses and other properties . </S><S> the functional is largely based on calculated ab initio nuclear and neutron matter equations of state . </S><S> compared to typical skyrme functionals having 10 - 12 parameters apart from spin - orbit and pairing terms , the new functional has only 2 or 3 adjusted parameters , fine tuning the nuclear matter binding energy and fixing the surface energy of finite nuclei . </S><S> an energy rms value of 1.58 mev is obtained from a fit of these three parameters to the 579 measured masses reported in the audi and wapstra 2003 compilation . </S><S> this rms value compares favorably with the one obtained using other successful mean field theories , which range from 1.5 - 3.0 mev for optimized skyrme functionals and 0.7 - 3.0 for the gogny functionals . </S><S> the other properties that have been calculated and compared to experiment are nuclear radii , the giant monopole resonance , and spontaneous fission lifetimes . </S>

1107.2918

colour and absorption line gradients are markers of stellar population variations within galaxies , providing important clues to galaxy evolution ( e.g. @xcite ; @xcite ; @xcite ; @xcite ; ( * ? ? ?* herafter t+10 ) , @xcite ) .although spectral features are less ambiguous probes of stellar population properties than colours are ( e.g. @xcite ; @xcite ; @xcite ) , there are only small samples available outside the nuclear regions of galaxies .fortunately , studies based on colours turn out to provide , on average , similar age and metallicity estimates to spectral studies ( e.g. t+10 ) , allowing for much larger surveys of radially - extended stellar populations . in t+10 we analyzed colour gradients within an effective radius ( ) for an extensive sample of low - redshift galaxies based on sdss data ( @xcite ) , including both early - type ( etgs ) and late - type ( ltgs ) galaxies .we carried out simple stellar populations analyses , fitting synthetic models to the observed colours to derive estimated age and metallicity gradients , and stellar masses ( see t+10 and later in the paper for further details ) .the main results found can be summarized as follows : 1 .colour gradients are , on average , negative ( i.e. galaxies are redder in the centre ) , except for galaxies with lower stellar masses ( @xmath1 ) which tend to have null or positive gradients . at fixed mass ,ltgs have steeper gradients than etgs , whose colour profiles in general show milder variations . 2 .the colour gradients of ltgs decrease systematically with mass while the trend for etgs changes near a mass of @xmath2 : systematically decreasing and increasing at lower and higher masses , respectively .3 . the primary driver of the colour gradients , mainly for galaxies with older centers , is metallicity variations , with age gradients playing a secondary role . for ltgs ,the metallicity gradients tend to be steeper for higher masses , the same is true for etgs except that for masses above @xmath3 , the ( negative ) gradients are fairly invariant or slightly increasing with mass .4 . at fixed mass ,galaxies with older centres have the shallowest metallicity and age gradients .we found that these results were generally consistent with others in the literature ( e.g. @xcite ; @xcite ) once different galaxy selection effects were taken into account .the overall observational picture can be framed within the predictions of hydrodynamical and chemodynamical simulations of galaxy formation ( @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) which include various physical processes that can influence the gradients : supernovae and agn feedback , galaxy mergers , interactions with environment , etc . a fresh and qualitative overview of this physical scenario is given by fig .[ fig : fig1 ] , where we show the isodensity contours for the colour - mass diagram ( which we have already discussed in t+10 ) and superimpose some lines ( in the inset ) and arrows which indicate the efficiency of the highlighted physical processes .in particular , the observations for high - mass etgs seem to generally suggest the influence of mergers and agn feedback ( @xcite ; @xcite ; @xcite ) , with an efficiency possibly increasing with mass ( see red arrows and lines in the inset panel ) . on the other hand , the population gradients of ltgs and less massive etgs may be mainly driven by infall and supernovae feedback ( @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) , which are more efficient at very low masses ( see blue lines and arrows ) .in this paper we consider the implications of the observed colour , metallicity and age gradients from t+10 for variations of stellar _ mass - to - light ratio _ @xmath4 with radius .this issue is relevant for understanding the detailed distribution of mass within galaxies : both the stellar mass and the dark matter , whose distribution is inferred by substracting the stellar mass from the total dynamical mass .there has been scarcely any literature to date that focussed on this topic , except for a study in spiral galaxies by @xcite . as a first approximation ,one could use single - colour semi - empirical models such as those of @xcite and ( * ? ? ?* b03 hereafter ) to go directly from colour gradients to gradients .however , we wish to make use of multiple colours , to allow for departures of individual galaxies from mean trends , to explore the systematics thoroughly , and to trace the dependence on stellar parameters , like age and metallicity . this paper is organized as follows . the sample and the spectral analysisare discussed in [ sec : sample ] , in where we also show the ability of our fitting procedure to recover the correct gradients , and we examine possible systematics .the results are shown in [ sec : results ] , while [ sec : conclusions ] is devoted to a discussion of the physical implications and conclusions .our database consists of @xmath5 low redshift ( @xmath6 ) galaxies in the nyu value - added galaxy catalog extracted from sdss dr4 ( ( * ? ? ? * hereafter b05 ) ) , recently analyzed in t+10 .following t+10 , we have identified etgs as those systems with a r - band s@xmath7rsic index satisfying the condition @xmath8 and with a concentration index @xmath9 .the final etg sample consists of @xmath10 galaxies . of the remaining entries of our database , @xmath11 are ltgs , defined as objects with @xmath12 , @xmath13 , and @xmath14 .we have used the multi - band structural parameters given by b05 to derive the colour profile @xmath15 of each galaxy as the differences between the ( logarithmic ) surface brightness measurements in the two bands , @xmath16 and @xmath17 .we have defined the colour gradient as the angular coefficient of the relation @xmath18 vs @xmath19 , @xmath20 , measured in mag dex@xmath21 ( omitted in the following ) , where is the r - band effective radius . by definition , a positive gradient , @xmath22 ,means that a galaxy is redder as @xmath23 increases , while it is bluer outward for a negative gradient .the fit of synthetic colours is performed on the colours at @xmath24 and @xmath25 and on the total integrated colours .the stellar population properties ( age , metallicity and ) are derived by fitting ( * ? ? ?* hereafter bc03 ) `` single burst '' synthetic stellar models to the observed colours . age and metallicity are free to vary , and a @xcite initial mass function ( imf ) is assumed .stellar parameter gradients are defined as @xmath26 , where @xmath27 are the estimated age , metallicity and .because of the definitions adopted for @xmath28 and @xmath29 , an easier and equivalent definition can be used , in fact , we define @xmath30 $ ] , @xmath31 $ ] and @xmath32 $ ] where ( age@xmath33 , @xmath34 , @xmath35 ) , with @xmath36 , are the estimated parameters at @xmath28 and @xmath29 , respectively - band , but for simplicity we omit the subscript @xmath37 . ] . in the present paperwe will concentrate on the analysis of the gradients , , in terms of stellar mass and velocity dispersion .the dataset , including main stellar parameters and gradients , is available at : http://www.itp.uzh.ch/@xmath38ctortora/gradient_data.html . before proceeding, we check for systematic effects on our estimates from the stellar population fits . because of the well known age - metallicity degeneracy ( @xcite , @xcite , @xcite ) , the stellar population parameters and the gradients might be biased when using the optical colours only , as we do now . widening the wavelength baseline to include near - infrared ( nir ) colours should ameliorate the degeneracy . in t+10 we checked age and metallicity inferences using optical versus optical+nir constraints , and found little systematic difference . herewe will carry out a similar analysis of the gradients , using a suite of monte carlo simulations .we extract some sets of simulated galaxy spectra from our bc03 spectral energy distribution libraries with random , uncorrelated stellar parameters and gradients .we start from generating data with fully random ages and metallicities , and then we also analyze the cases when the stellar parameters are constrained within fixed intervals , generating not - zero sample average gradients .we extract the input galaxy colours and add simulated measurement errors , as randomly extracted steps from the interval ( @xmath39,@xmath40 ) , with @xmath41 .we apply our fitting procedure , searching for the best model in our synthetic library , which reproduces the colours of each of the simulated galaxies and then compare the output parameter estimates to the input model values .we perform the fit ( 1 ) using only the optical sdss bands @xmath42 and ( 2 ) adding nir photometry ( @xmath43 , @xmath44 and @xmath45 ; @xcite ) and ( 3 ) ultraviolet photometry ( nuv and fuv ; @xcite ) .we define the median error in the gradients as @xmath46 , where @xmath47 and @xmath48 are the input and fitted gradients with @xmath49 , @xmath50 and @xmath51 , which we show in tables [ tab : tab1 ] and [ tab : tab1bis ] .starting from the purely random case shown in table [ tab : tab1 ] , we have found that , adding the nir or uv data , both the values and their gradients are perfectly recovered with a very little scatter . with optical data only , the uncertainties are doubled ( around 0.1 in @xmath52 ) and there is a small systematic shift in the gradients ( of @xmath53 for the worst case analyzed ) , but the results are fairly well recovered with no spurious trends expected .we notice that the median results are quite independent of the range adopted for the input @xmath54 , and only slight differences in the scatter are found .although we will be mainly interested in the trends of with mass and velocity dispersion , we have also checked the systematics in age and metallicity gradients .we find similar results to the case of discussed above , but slightly larger shifts in the median and larger scatters ( @xmath55 in the worst cases ) when only the optical is used . in table[ tab : tab1bis ] , calculating the sample averages , we start with the input values @xmath56 and @xmath57 . here the median errors and the scatters increase , but the conclusions discussed above are qualitatively confirmed .when optical data are used , for high @xmath54 we have the worst discrepancies in the estimated gradients amounting to @xmath58 , @xmath59 . for age and metallicity gradients, we could underestimate ( overestimate ) the age ( metallicity ) gradients of @xmath60 , which is reduced to @xmath61 when near - ir or uv are included .thus , in many cases we have found that the median errors are acceptable and within the typical sample scatter and the uncertainties due to the adopted stellar population prescription .we will consider these results as qualitative indications of the effects on estimated gradients caused by systematics .in this section we discuss first the gradients in terms of colour , age and metallicity gradients and then as a function of stellar mass and velocity dispersion , paying attention to the role of central age in the observed correlations .we start by plotting in fig .[ fig : fig2 ] the gradients as a function of @xmath62 colour gradient ( @xmath63 ) for etgs and ltgs .the two quantities are strongly and positively correlated and no statistically significant difference between the morphological types is found . while such a correlation is not surprising owing to the well known - correlations between colours and ( @xcite ; b03 ) , which should apply to all radii thus producing the relationship between and , there are important details to consider .@xcite have adopted different evolution models for ltgs to establish some linear correlations between integrated total colours and the corresponding determining some best fit relations between these two quantities , which have been proven to work for etgs as well ( @xcite ) . adopting the solar metallicity models in @xcitewe have found a good agreement with @xcite or @xcite synthetic prescriptions in etgs with high , while at low @xcite overestimates the ( see @xcite for further details ) . in the left panel of fig . [fig : fig2 ] we compare our - trends with the ones ( solid and dashed lines ) derived from the b03 best fitted relations , finding a good agreement .the black and green lines plotted in fig . [ fig : fig2 ] are derived by the relations between the g - band and the colours ( @xmath64 , @xmath62 , @xmath65 and @xmath66 ) from b03 .we insert in these relations the measured colour for each galaxy in our sample , deriving the corresponding .this operation is made for both the radii @xmath28 and @xmath29 , thus we can calculate the gradients .finally , the plotted lines are obtained , first calculating medians of in bins of and then fitting a straight line .we see that the best fitted relations depend on the colours adopted in the colour relations from b+03 to derive the gradients .for instance , we observe a flattening of the relation when using colours probing redder wavelengths , i.e. @xmath65 and @xmath66 , but a shallower trend is obtained when @xmath64 is adopted .the agreement is quite good for moderate colour gradients , while for very steep positive or negative gradients some departures are observed . despite the good agreement , the very simplified approach in @xcite and b03has some strong limitations .in fact , a ) the b03 relations hold for integrated quantities , thus the radial variation of the @xmath4 can be derived if one assumes that the same relations apply at different apertures and in general , the zero - point and the slope of these fitted linear relations would depend on the radius adopted and b ) these relations do not take into account the stellar parameters ( e.g. age and metallicity at each radius ) thus they do not allow to catch the full physics behind these correlations .these shortcomings are overcome by our multi - colour direct stellar population analysis which connects the gradients self - consistently to the underlying stellar population properties ( such as age and metallicity ) .this also allows us to track the physical reasons for the variations , which in turn should be reproduced by galaxy formation models . differently from what happens for colour gradients which are mostly driven from metallicity gradients , from the right column of fig .[ fig : fig2 ] it is evident that the relation is driven mainly by _ age _ , in both ltgs and etgs .the trends of with stellar mass and velocity dispersion are shown in the top panels of fig .[ fig : fig3 ] . as with the trend of colour and metallicity gradients with mass ( see comments in [ sec : intro ] also t+10 ) , we find that ltgs have a decreasing trend with mass and show @xmath67 for @xmath68 and @xmath69 at larger masses .in contrast , etgs have positive @xmath70 at low masses , which decreases down to a minimum of @xmath71 at @xmath72 and then invert the trend , with the most massive etgs having gradients around zero or slight positive . when plotted as a function of velocity dispersion , the run with @xmath73is generally flatter , with ltgs having no trend and etgs showing a two - fold trend with a minimum around @xmath74 km / s and a mild increase toward the low and high @xmath73 sides , always staying consistent with @xmath75 .since galaxies are morphologically selected by using criteria on concentration and s@xmath76rsic index , the steeper gradients found for ltgs when compared with etgs indicate a sequence in terms of such structural parameters .in fact , at fixed mass , galaxies with lower @xmath77 or @xmath78 have steeper colour and gradients .as observed for the stellar population gradients ( see t+10 ) , the central age is responsible for much of the scatter of the trends versus mass and velocity dispersion , which is illustrated by the median distribution for centrally old ( @xmath79 ) and young ( @xmath80 ) galaxies in fig .[ fig : fig3 ] . in particular , young ltgs have almost null gradients , while the oldest ones have very steep values of @xmath81 to @xmath82 .etgs show a similar but offset effect .older etgs have slightly negative values of @xmath83 , while the younger ones have steep positive @xmath84 at very low masses , decreasing to @xmath60 at larger masses . as a comparison, we also show in fig . [fig : fig3 ] ( top panels ) the results when galaxy colours are fitted to a synthetic spectral model which has no age gradient ( fixed @xmath85 ) . in this case , ltgs have which are null up to @xmath86 , and decrease at larger masses , while gradients in etgs show a milder trend when compared with the reference fit . when plotted versus both ltgs and etgs have similar @xmath83 for all velocity dispersions .note that these results are consistent with what was found for the older galaxies in the reference fit .we also compare our results with g - band s derived from b03 . as an example we plot the median lines derived using the @xmath87 and @xmath88 relations for both etgs and ltgs . the trends with and velocity dispersion are qualitatively unchanged , but some differences emerge since the s from each b03 relation are derived using different colours , which probe different wavelength spectral regions .the s obtained using the @xmath62 colours seems to resemble our trends for ltgs , while the agreement is poorer when @xmath64 is used .for etgs our trend is between the two b03 results at low mass , while the from @xmath64 is a little bit better at large mass .we have also checked the effect of the waveband adopted for the ratios in the bottom panels of fig .[ fig : fig3 ] . here , the @xmath37-band gradients are compared to the ones in @xmath89- and @xmath90-band . except for the most massive etgs , in generalthe gradients tend to become shallower ( closer to zero ) when using redder bands . also , the trends with mass and velocity dispersion become weaker .thus , redder bands , which are less affected by dust extinction and trace the older and metal - richer stellar populations in galaxies , tend to be less dependent on the existence of gradients in colour and stellar population parameters . extending such analyzing into the nir bandsshould find gradients increasingly approaching zero . from the preceding results, we can see that although the age gradient is generally the driving parameter behind the gradient , this conclusion can vary depending on galaxy type , mass and central age .we summarize the gradients for different galaxy subsamples in table [ tab : tab2 ] , noting also which of the stellar parameters drives the gradient . [ cols="<,<,<",options="header " , ]in this paper we have investigated the correlation between gradients and stellar mass or velocity dispersion for a large sample of local galaxies from sdss .the gradients have been derived by the fitting of synthetic spectral models to observed optical colours .we have used monte carlo simulations to check that the optical - only wavelength baseline should not dramatically affect the results compared to including near - infrared and ultraviolet data .this conclusion is encouraging in view of future studies of higher redshift galaxies which are generally observed in rest - frame optical bands . anyway , future analysis in local samples , relying on near - ir and/or uv data would be able to give further details about the degeneracies and systematics affecting stellar fitting .we have found that there exists a tight positive correlation between and colour gradients which is not surprising because of the well known -colour correlations ( e.g. , @xcite ; b03 ) .we have found that is a decreasing function of stellar mass in ltgs , while a two - fold trend is found for etgs ( for both stellar mass and velocity dispersion ) , in agreement with the trends of colour gradients discussed in t+10 .similar results are obtained using the -colour relations from b03 , which , although easy to use , do not give information about stellar population parameters .thus , the novelty of our approach resides in our ability to check how gradients are driven by variations in metallicity or age . in general, we have found that age is most important , but that metallicity becomes more important in many high - mass galaxies .this picture does not change dramatically if gradients are calculated in redder bands ( @xmath89- and @xmath90-bands rather than @xmath37 ) . in these cases are shallower and the trends with stellar mass and velocity dispersion become milder ( fig .[ fig : fig3 ] ) .these results suggest that estimates from red or infrared bands would not depend on galactocentric radius , since the radiation emitted by stellar populations in these spectral regions is more homogenous .this is the first work to derive gradients 1 ) from observations of a large galaxy sample and 2 ) using a direct stellar population analysis approach .the only point of comparison is the work by @xcite in a sample of spiral galaxies , where negative gradients were found to be important in modelling rotation curves .we may explain some of our observed trends through physical mechanisms as follow .starting from low - mass systems , centrally - young ltgs have age increasing with radius while metallicity decreases , which can be due to the presence of a gas inflow which allows the formation of younger stars in the centre with a smaller .centrally - old ltgs have age decreasing with radius which implies recent outer star formation with a similar metallicity to the centre . herethe central star formation might have been quenched by sne which have expelled metals to larger radii ( where the younger stars drive a smaller ) .centrally - young etgs have positive age , metallicity and gradients , and may hosting expanding shells ( @xcite ) .centrally - old etgs show mild negative age gradients and very shallow positive metallicity gradients , similarly to the old ltgs thus again suggesting sne having played a role .the more massive centrally - young ltgs have decreasing with radius , but also strong negative , thus the new stars formed inside the core ( @xmath67 ) have higher metallicity and .this might be due to recycled gas from sne or agn which is falls back to the centre due to the deep potential wells and produces a rejuvenated stellar population . in centrally - young etgs ,on the other hand , the younger stars formed later in the centres ( i.e. with positive age gradients ) and with higher metallicities , possibly because of a central burst of star formation in `` wet '' mergers ( e.g. @xcite ; @xcite ; @xcite ; t+10 ) . centrally - old ltgs and etgs show very similar behaviors , which seem compatible with merging events that largely wash out any age gradients ( and the in etgs ) , and may produce coeval metallicities in the outer regions of the ltgs . in both cases the net effect is that metallicity is the main driver of the strong negative in ltgs and of the weak negative of the massive etgs ( in any ) .thus , the physical mechanisms that can account for the observations outlined in tab .[ tab : tab2 ] seem consistent with the evolutionary scenario discussed in [ sec : intro ] .in particular , the two - fold trend of etgs in fig . [fig : fig3 ] , with a minimum around @xmath91 is mirrored by the colour and stellar population gradients as in fig .[ fig : fig1 ] ( and t+10 ) and found also in other works ( @xcite ; @xcite ) .the characteristic mass agrees well with the typical mass scale break for star forming and passive systems ( @xcite ) , for `` bright '' and `` ordinary '' galaxies ( @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) , or for cold and hot flow ( shock heating ; @xcite ) . following the main conclusions in t+10, quasi - monolithic collapse and supernovae feedback could be the main drivers of the decreasing - trends in ltgs and low mass etgs ( @xcite ; @xcite , @xcite ) , while the increasing trends in massive etgs could be explained by the heightened importance of mergers and agn feedback ( @xcite ; @xcite ; @xcite ) .gradients are expected from simulations of galaxy formation and evolution .e.g. @xcite found in chemo - dynamical simulations of spiral galaxies that the central regions ( @xmath92 ) have steep negative gradients , becoming flatter at larger radii .high - redshift ( @xmath93 ) merger remnants simulated by @xcite showed time - dependent negative as a consequence of redder cores in redder galaxies ( similarly to observations of local galaxies in t+10 ) .gradients were also predicted for merger remnants by @xcite , who stressed that the effective radius derived from the @xmath37-band light would increase with time ( by @xmath94 10% ) simply because of stellar population gradients .this effect is important for conclusions about stellar densities and related dynamical implications ( e.g. for the fundamental plane ) .similarly @xcite argued that the existence of such gradients can imply that the observed quiescent galaxies at large redshift may be of an order or magnitude more dense than previously found from observations . the gradients present a general complication for estimates of the dark matter distribution in galaxies .@xcite found that for spiral galaxies , including gradients would change the fits to circular velocity curves but not produce strong effects on the inferred dark matter density parameters . to our knowledge , this issue has not been addressed in the literature for early - type galaxies , and it is beyond the scope of the present paper to do so in any detail . in particular , we have studied the stellar population gradients only inside 1 , while the gradients out to @xmath94 5 could be important for dark matter analyses ( e.g. @xcite ) . assuming that the gradients do not change radically outside @xmath94 1 , for the massive etgs generally studied with lensing or dynamics ( @xmath95 150 km s@xmath21 ), we would expect the stellar gradients to be mild ( see fig .[ fig : fig3 ] ) , and to not strongly affect the dark matter estimates .a final issue is that until now , we have assumed that the imf is invariant .if there are galaxy - to - galaxy variations in the imf ( see e.g. @xcite ) , then the stellar values are changed only by a constant multiplicative value , and the implied gradients remain unchanged . however , imf variations if real would then likely exist _ within _ galaxies owing to variations in stellar populations .indeed , there have been many suggestions that stars forming at early times had a non - standard imf ( e.g. @xcite ) .one study of nearby etgs suggested that systems with older stars within 1 have `` lighter '' ( closer to chabrier than salpeter ) imfs @xcite .this would imply that positive and negative age gradients would decrease and increase the stellar gradients , respectively . revisiting fig .[ fig : fig3 ] , the gradients for the centrally - young objects would decrease and the centrally old objects would increase , possibly reducing and even eliminating the age - related scatter in the gradient trends .another scenario would be if the central regions of massive etgs have `` higher - mass '' imfs than standard ( @xcite ) , while the outer regions have accreted galaxies with more `` normal '' imfs ( e.g. @xcite or @xcite ) .this would decrease these galaxies stellar gradients , although it is not clear how this effect would relate to age and metallicity trends . in future analyses, we plan to analyze stellar profiles out to a few , and analyze the impact of gradients on dark matter profile inferences , which may be compared to @xmath96cdm ( e.g. @xcite ). we will also consider the ( probably small ) adjustments implied for central dark matter inferences ( e.g. @xcite ) .dust effects will further be examined : dust is not important for metallicity gradients ( see t+10 ) but could affect age and thereby gradients , particularly for intermediate - mass etgs .we thank the referee for his suggestions which helped to improve the paper .ct was supported by the swiss national science foundation .ajr was supported by national science foundation grants ast-0808099 and ast-0909237 .

<S> we analyze the stellar mass - to - light ratio ( ) gradients in a large sample of local galaxies taken from the sloan digital sky survey , spanning a wide range of stellar masses and morphological types . as suggested by the well known relationship between mass - to - light ( ) ratios and colors </S><S> , we show that gradients are strongly correlated with colour gradients , which we trace to the effects of age variations . </S><S> stellar gradients generally follow patterns of variation with stellar mass and galaxy type that were previous found for colour and metallicty gradients . in late - type galaxies gradients are negative , steepening with increasing mass . in early - type galaxies </S><S> gradients are shallower while presenting a two - fold trend : they decrease with mass up to a characteristic mass of @xmath0 and increase at larger masses . </S><S> we compare our findings with other analyses and discuss some implications for galaxy formation and for dark matter estimates . </S><S> [ firstpage ] dark matter galaxies : evolution galaxies : galaxies : general galaxies : elliptical and lenticular , cd . </S>

1408.4221

the global organization of complex molecular interactions within and across cells is being disclosed by the graph - theoretic approaches @xcite .the obtained cellular networks exhibit universal topological features which are rarely found in random networks , such as broad degree distributions @xcite and high modularity @xcite .their origins and implications to cellular and larger - scale functions have thus been of great interest .diverse network models based on simple mechanisms of adding and removing nodes and links have been proposed @xcite .those models capture the common aspects , like the preferential attachment @xcite , of biological processes such as the duplication , divergence , and recruitment of genes , proteins , and enzymes , and successfully reproduce the empirical features of biological networks , suggesting that the former can be the origin of the latter .yet it remains to be explored what drives such construction and remodeling of biological networks functioning in living organisms .a population of living organisms find the typical architecture and function of their cellular networks changing with time .such changes on long time scales are made by the organisms of different traits , giving birth to their descendants with different chances , that is , by evolution @xcite .therefore , it is desirable to investigate how the generic features of evolution lead to the emergence of the common features of biological networks .living organisms are required to possess adaptability and stability simultaneously @xcite . to survive and give birth to descendants in fluctuating environments ,the ability to adjust to a changed environment is essential @xcite , which leads to , e.g. , phenotypic diversity and the advantage of bet - hedging strategy @xcite . at the same time , the ability to maintain the constant structure and perform routine important functions regularly , such as cell division and heat beats , is highly demanded .therefore , in a given population , the cellular networks supporting higher adaptability and stability are more likely to be inherited , which leads the representative topology and function of the cellular networks to evolve over generations . herewe study how such evolutionary pressure shapes the biological networks .we propose a network model , in which links are rewired such that both adaptability and stability are enhanced .the dynamics of the network is simply represented by the boolean variables assigned to each node regulating one another @xcite .the boolean networks have been instrumental for studying the gene transcriptional regulatory networks @xcite and the metabolic networks @xcite .this model network is supposed to represent the network structure typical of a population .the evolution of boolean networks towards enhancing adaptability @xcite , stability @xcite , or both @xcite has been studied , mostly by applying the genetic algorithm or similar ones to a group of small networks . in particular , the model networks which evolve by rewiring links towards local dynamics being neither active nor inactivehave been shown to reproduce the critical global connectivity and many of the universal features of real - world biological networks @xcite , demonstrating the close relation between evolution and the structure of biological networks .however , the evolutionary evaluation and selection are made for each whole organism , not for part of it . in the simulated evolution of our model ,the adaptability and the stability of the _ global _ dynamical state are evaluated in the wild - type network and its mutant network , differing by a single link from each other , and the winner of the two becomes the wild - type in the next step .the study of this model leads us to find that sparse and heterogeneous connectivity patterns emerge , which are consistent with the gene transcriptional regulatory networks and the metabolic networks of diverse species .the scaling behavior of stability with respect to system size suggests that the evolved networks are critical , lying at the boundary between the inflexible ordered phase and the unstable chaotic phase .our study also shows how the nature of fluctuations and correlations changes by evolution .the extent of perturbation spread characterizing the system s stability fluctuates over different realizations of evolution .the fluctuation turns out to scale linearly with the mean in the stationary state of evolution while the square - root scaling holds in the transient period .we argue that this dynamic crossover is rooted in the variation of the combinatorial impacts of the structural fluctuation , driven by evolution , and the internal stochasticity .the scaling of the correlation volume , representing the typical number of nodes correlated with a node , is another feature of the evolved networks .our results thus show the universal impacts of biological evolution on the structure and function of biological networks and illuminate the nature of correlations and fluctuations in such evolving systems distinguished from randomly - constructed or other artificial systems .the paper is organized as follows .the network evolution model is described in detail in sec .[ sec : model ] .the emergent structural and functional features are presented in sec .[ sec : evolution ] . in sec .[ sec : generalized ] , we represent the hamiltonian approach to a generalized model , including our model in a limit , and show the robustness of the obtained results .the scaling behaviors of the fluctuation of perturbation spread and the correlation volume are analyzed in secs .[ sec : fluctuation ] and [ sec : correlation ] , respectively .we summarize and discuss the results of our study in sec . [sec : discussion ] .we consider a network in which the node activities are regulated by one another .the network may represent the transcriptional regulatory network of genes , in which the transcription of a gene is affected by the transcriptional factors encoded from other genes , or the metabolic network of metabolites and reactions , the concentrations and fluxes of which are correlated .various cellular functions are based on those elementary regulations .the model network does not mean that of a specific organism but is representative of the cellular networks of a population of organisms , which evolve with time . in our modelthe network evolution is made by adding or removing links , representing the establishment of new regulatory inputs or the loss of existing targets possibly caused by point mutations in the regulatory or coding regions of dna @xcite . to be specific , we consider a network @xmath0 of @xmath1 nodes which are assigned boolean variables @xmath2 for @xmath3 .@xmath4 represents whether a node @xmath5 is active or inactive in terms of the transcription of the messenger rna , the flux of the corresponding chemical reaction , or the concentration of the metabolite . the global dynamical state is represented by @xmath6 .initially @xmath7 directed links are randomly wired and @xmath4 s are set to @xmath8 or @xmath9 randomly . a link from node @xmath10 to node @xmath5 , with the adjacency matrix @xmath11 , indicates the regulation of the activity of @xmath5 by @xmath10 @xcite .@xmath12 of node @xmath5 at the microscopic time step @xmath13 is determined by its regulators at @xmath14 as @xmath15 where @xmath16 is the time - constant regulation function for node @xmath5 , taking a value @xmath8 or @xmath9 for each of all the @xmath17 states of @xmath18 regulators with @xmath19 .a target state @xmath20 is demanded of the network by the environment and the distance between @xmath21 and @xmath22 quantifies the adaptation to the environment .is generated by adding or removing a link randomly in the wild - type @xmath0 , here between nodes @xmath5 and @xmath10 .( b ) the transition from @xmath0 to @xmath23 happens if @xmath24 or if @xmath25 and @xmath26 .a new target state @xmath27 is generated if @xmath28 . ]the dynamical state @xmath29 is updated every microscopic time step @xmath14 as in eq .( [ eq : boolean_evol ] ) . also the structure of the network @xmath0 , including its adjacency matrix @xmath30 and the regulating functions @xmath31 , evolves on a longer time scale as follows . at @xmath32 with @xmath33 the macroscopic time step and @xmath34 a time constant , a mutant network @xmath23 is generated , which is identical to the wild - type @xmath0 except that it has one more or less link with a different regulation function ( see fig .[ fig : model ] ) .then we let the dynamical state @xmath35 evolve on @xmath0 and @xmath23 , respectively , for @xmath36 . due to their structural difference, the @xmath35 s may evolve differently although they are set equal initially at @xmath37 . at @xmath38 ,the adaptability and the stability of the time trajectories @xmath39 on @xmath0 and @xmath23 are evaluated in terms of the hamming distances , @xmath40 , and @xmath41 , where the first two characterize the adaptation to the environment and the latter two represent the typical extent of perturbation spread .the winner of @xmath0 and @xmath23 is determined in the way detailed below , which then becomes the wild - type @xmath0 for @xmath42 competing with its mutant .these procedures are repeated for @xmath33 .the adaptability of a boolean network @xmath0 at time @xmath43 is here quantified by the average hamming distance between @xmath35 and a given target state @xmath22 @xcite over a microscopic time interval as @xmath44 where @xmath45 is the kronecker delta function .@xmath46 is a microscopic - time constant such that the hamming distance @xmath47 is stationary for @xmath48 .another constant @xmath34 is set to @xmath49 , which is found to range from @xmath50 to @xmath51 for @xmath52 in our simulations .if smaller values of @xmath34 and @xmath46 were used , @xmath53 in eq .( [ eq : htarget ] ) would not represent the adaptability of the network in the stationary state of the boolean dynamics .the smaller @xmath53 is , the closer the dynamical state on @xmath0 is likely to approach the target state , implying that @xmath0 is more adaptable to a given environment .we compute @xmath54 in the same way as in eq .( [ eq : htarget ] ) .the stability in performing routine processes is another key requirement of life .given that local perturbations can spread globally , the ability to suppress such perturbation spread can be a measure of stability @xcite . to quantify the stability of @xmath0 at time @xmath43, the difference between the original state @xmath35 and the perturbed state @xmath55 is measured .the perturbed state is obtained by flipping the states of @xmath56 randomly - selected @xmath57 s in @xmath35 at @xmath37 and then letting it evolve on @xmath0 for @xmath58 .then we count the number of perturbed nodes , having @xmath59 , as @xmath60 with the hamming distance @xmath61 defined in eq .( [ eq : htarget ] ) .@xmath62 represents the typical fraction of perturbed nodes ; the smaller @xmath62 is , the more stable the network @xmath0 is against dynamical perturbations .the stability of the mutant @xmath23 is also computed in the same way .we remark that the number of initial flipped variables can be changed over a significant range without changing the main results .the mutant @xmath23 becomes the winner ( i ) if @xmath63 ( @xmath23 is more adaptable than @xmath0 ) or ( ii ) if @xmath64 ( @xmath23 is more stable than @xmath0 ) and @xmath65 . if @xmath65 and @xmath66 , the winner is chosen at random .examples of the transition from @xmath0 to @xmath23 are depicted in fig .[ fig : model ] . finally , to model the changes of the environment , a new target state @xmath67 is generated if @xmath68 of the winner is zero .therefore our network evolution model represents the co - evolution of the structure and dynamics of the boolean network on different time scales in a changing environment .is set to @xmath69 or @xmath56 giving @xmath70 or @xmath71 . for each @xmath1 and @xmath7 ,we run @xmath72 independent simulations , each for @xmath73 , where @xmath74 ranges from @xmath75 to @xmath76 and @xmath77 for @xmath78 , @xmath79 for @xmath80 , and @xmath81 for @xmath82 .@xmath83 indicates the ensemble average .( a ) the plots of the mean connectivity @xmath84 for @xmath85 .it converges to a constant irrespective of the initial value , which is evaluated as @xmath86 .the networks at selected times are presented .( b ) the @xmath1-dependence of the stationary - state mean connectivity . @xmath87( solid line ) fits reasonably the model results ( circles ) .the mean connectivity @xmath88 of the transcriptional regulatory networks of four species ( triangles ) @xcite and of the bipartite metabolic networks of 506 species ( crosses ) @xcite are shown .the fitting line ( dotted ) given by @xmath89 fits the data of the metabolic networks with @xmath1 the number of reactions and metabolites .( c ) the cumulative distributions of the in - degree , @xmath90 at @xmath91 ( initial state ) and @xmath92 ( stationary state ) for @xmath85 . the distribution in the random networks of @xmath85 nodes and @xmath93 links is also shown for comparison .] the simulation of the proposed model shows a variety of interesting features of evolving networks .most of all , we find that the mean connectivity @xmath94 , with @xmath95 the in - degree or the number of regulators of node @xmath5 and @xmath96 the total number of links at time @xmath43 , converges to a constant @xmath97 , which depends only on @xmath1 regardless of @xmath98 [ fig . [fig : evol_structure ] ( a ) ] .the mean connectivity has been shown to converge to @xmath99 in some evolution models @xcite , which is the critical point distinguishing the ordered and the chaotic phase in random boolean networks @xcite .different values of @xmath100 have been reported in other models @xcite , where @xmath101 , implying a fundamental difference between the evolved networks and random networks . in our model ,@xmath100 ranges from @xmath102 to @xmath103 for @xmath52 and the data are fitted by a logarithmic growth with @xmath1 as @xmath104 [ see fig . [fig : evol_structure ] ( b ) ] .this suggests that @xmath100 would remain small for @xmath1 reasonably large , e.g. , @xmath105 for @xmath106 .such sparse connectivity is identified in real biological networks @xcite .the mean connectivities of the transcriptional regulatory networks are between @xmath8 and @xmath107 while the number of nodes ranges from hundreds to thousands .the mean connectivities of the metabolic bipartite networks also range between @xmath8 and @xmath107 .furthermore , they show logarithmic scaling with @xmath1 in agreement with our model [ see fig . [fig : evol_structure ] ( b ) ] .the number of regulator nodes ( in - degree ) @xmath108 is broadly distributed in the evolved network compared with the poissonian distribution of the random networks as seen in fig .[ fig : evol_structure ] ( c ) .such broad distributions are universally observed in real - world networks @xcite . the cumulative in- degree distribution @xmath109 , with @xmath110 the heavisde step function , appear to take the form of an exponential function , which is in agreement with the transcriptional regulatory networks of _ s. cerevisiae _ @xcite .this is , however , inconsistent with the previous studies on the real metabolic networks @xcite or other model networks evolving via node duplication and divergence @xcite , which display power - law degree distributions .it is known that the node duplication @xcite or the preferential attachment of links @xcite may lead to such power - law degree distributions , which is missing in our model . in ref .@xcite , the functional form of the degree distributions of some real metabolic networks are hard to point out .in contrast to the broad in - degree distributions , the out - degree @xmath111 in the evolved networks of our model is found to follow the poisson distribution as in the random networks .it is known that the out - degree distribution is irrelevant to the determination of the dynamical phase - ordered or chaotic - of random boolean networks @xcite . for @xmath112 and @xmath85 with the initial mean connectivity @xmath113 and @xmath114 .@xmath77 simulations are run .the color varies with the evolution time @xmath43 and the arrows indicate the direction of increasing time .( inset ) the scaling behavior of the stationary - state hamming distance @xmath115 with respect to the number of nodes @xmath1 .@xmath116 ( dashed line ) fits the data .( b ) plots of @xmath117 versus the mean connectivity @xmath84 for the evolving networks and the random networks of @xmath85 . ] as evolution proceeds , it is more facilitated for the evolving network to get close to or reach a given target state .such adaptability is quickly acquired , as implied in the rapid decrease of @xmath118 with increasing @xmath43 [ fig .[ fig : evol_dynamics ] ( a ) ] .we remark that @xmath119 may increase with @xmath43 even in a single realization of evolution , since the target state , the state demanded by the environment , may change with time .the extent of perturbation spread @xmath117 also decreases rapidly by evolution .its stationary - state value @xmath120 shows the following scaling behavior with @xmath1 : @xmath121 this implies an intermediate level of stability of the evolved networks compared with the following networks .the random boolean networks with the mean connectivity at the threshold @xmath122 find the perturbation spread scale similarly to eq .( [ eq : hpertscal ] ) but with a smaller scaling exponent ranging between @xmath123 and @xmath124 , depending on the functional form of the in - degree distribution @xcite .therefore , the perturbation spread in those critical random networks is much larger than that in the evolved networks for large @xmath1 .figure [ fig : evol_dynamics ] ( b ) shows that during the whole period of evolution , the evolving networks have smaller spread of perturbation than the random networks with the same mean connectivity @xmath125 .on the other hand , in a variant of our model , the stability - only " model , in which only the stability of the wild - type and the mutant is evaluated for selection , the perturbation spread scales as @xmath126 [ fig .[ fig : relax ] ( b ) ] .the original networks allow larger spread of perturbation than the stability - only model in order to facilitate adaptation to a fluctuating environment .the mean connectivity @xmath100 is also subject to such a balance constraint .as the opposite to the stability - only model , we can consider the adaptation - only " model in which only the adaptability of the wild - type and the mutant is considered .we found that the mean connectivity is much larger than in the original model .a large number of links make more and larger attractors in the state space , which can be helpful for adaptation . in the stability - only model , on the contrary , we find that the mean connectivity is much smaller than that of the original model [ fig .[ fig : relax ] ( a ) ] , suppressing the transitions between attractors .all these characteristics demonstrate that the structure and dynamics of the evolved networks are at the boundary between the stable and robust phase and the flexible and adaptable phase @xcite .is related to the relative importance of adaptability with respect to stability as in eq .( [ eq : rdef ] ) .( a ) plots of the stationary - state mean connectivity @xmath100 versus the system size @xmath1 .@xmath100 increases slowly with @xmath1 for all @xmath127 except for the stability - only model .( b ) plots of the perturbation spread @xmath120 versus @xmath1 .the scaling behavior @xmath128 is observed for all the considered cases .( inset ) the scaling exponent @xmath129 decreases from @xmath8 to @xmath130 with increasing @xmath131 . ] in this section , we represent our model in the hamiltonian approach , which offers a natural extension of the model allowing us to check the robustness of the obtained results .the evolution trajectory of the model network corresponds to a path in the space of networks @xmath0 .a system of @xmath1 nodes changes its location in the @xmath0 space in the stochastic way as described in sec .[ sec : model ] .therefore , a generalized evolution model can be introduced by specifying the transition probability @xmath132 from @xmath0 to @xmath23 for a given dynamical state @xmath21 @xcite .note that the dynamical state evolves with microscopic time @xmath14 in a deterministic way as long as the network structure @xmath0 is fixed .suppose that the transition probabilities satisfy the relation @xmath133 where the hamming distances are computed by eqs .( [ eq : htarget ] ) and ( [ eq : hpert ] ) with @xmath134 and two temperatures @xmath135 and @xmath136 are introduced .transitions to the networks with smaller @xmath68 and @xmath137 are preferred to the extent depending on the two temperatures .our model corresponds to the limit @xmath138 since the transition from @xmath0 to @xmath23 is made only if @xmath139 or @xmath140 and @xmath141 . in case @xmath142 and @xmath143 , the transition to a less adaptable ( @xmath68-larger ) or less stable ( @xmath137-larger ) network can be made with non - zero probability contrary to that of our model .the adaptability - only model corresponds to the limit @xmath144 and @xmath145 and the stability - only model to @xmath146 and @xmath147 . with the transition probabilities satisfying eq .( [ eq : boltzmann ] ) , each network @xmath0 appears in the stationary state with probability @xmath148 with the two hamming distances playing the role of hamiltonians coupled with two temperatures . to investigate the robustness of the results obtained in sec . [sec : evolution ] , we investigate this generalized model with the temperature ratio @xmath131 positive , @xmath147 , and @xmath144 . for @xmath127 ,the transition from @xmath0 to @xmath23 is available if and only if @xmath149 . @xmath131 controls the relative importance of @xmath68 with respect to @xmath137 .simulations show that @xmath100 displays similar @xmath1-dependent behaviors for all @xmath127 ; it increases with @xmath1 slowly [ see .[ fig : relax](a ) ] . on the contrary , in the stability - only model , the mean connectivity decreases with @xmath1 .this highlights the crucial role of adaptation in shaping the architecture of regulatory networks . secondly , as shown in fig .[ fig : relax ] ( b ) , @xmath150 with @xmath151 is observed not only for @xmath152 but also for sufficiently large @xmath131 , in the range @xmath153 . for small @xmath131 , roughly @xmath154 and in the stability - only model , @xmath155 , implying that stronger stability is achieved than for large @xmath131 .the scaling exponent @xmath129 decreases from @xmath8 to @xmath130 with @xmath131 increasing in the range @xmath156 .such robustness of the structural and functional properties for all large @xmath131 makes our model ( @xmath152 ) appropriate for modeling the evolutionary selection requesting both adaptability and stability .as the initial randomly - wired networks evolve , many of their properties change with time , the investigation of which may illuminate the mechanisms of evolution by which living organisms optimize their architecture for acquiring adaptability and stability .evolution is accompanied by fluctuations .environments are different for different groups of organisms and vary with time as well even for a given group .mutants are generated at random and thus the specific pathway of evolution becomes stochastic .the studied networks also display fluctuations over different realizations of evolution @xmath157 for each quantity @xmath30 . among others , here we investigate such ensemble fluctuation of perturbation spread characterizing the system s stability @xmath158 .while the evolutionary pressure results in enhancing stability ( reducing @xmath159 ) , its fluctuation , normalized by the mean @xmath159 , is stronger and the whole distribution is broader , respectively , than those of random networks as shown in fig .[ fig : fluctuation ] ( a ) .such enhancement of fluctuations helps the evolving network search for the optimal topology under fluctuating environments @xcite .it is observed for a wide range of real - world systems that the standard deviation @xmath160 and the mean @xmath161 of a dynamic variable show the scaling relation @xmath162 with the scaling exponent @xmath163 reflecting the nature of the dynamical processes : for instance , @xmath164 in the case of no correlations among the relevant variables and their distributions having finite moments as in the conventional random walk while the widely varying external influence may make such significant correlations as leading to @xmath165 @xcite . such scaling relation has been observed for the gene expression level or the protein concentration that fluctuates over cells and time @xcite .also in our model the mean @xmath117 and the fluctuation @xmath166 of perturbation spread at different times @xmath43 satisfy the scaling relation @xmath167 interestingly , the scaling exponent @xmath163 changes with evolution [ fig .[ fig : fluctuation ] ( b ) ] ; @xmath168 with @xmath169 for @xmath170 and @xmath171 for @xmath172 during transient period but @xmath173 with @xmath174 in the stationary state .such crossover in @xmath163 is robustly observed for all @xmath1 and @xmath7 as shown in fig .[ fig : fluctuation](c ) and [ fig : fluctuation](d ) . what is the origin of such dynamic crossover in @xmath163 ?it has been shown that the interplay of exogenous and endogenous dynamics may affect the scaling exponent @xmath163 in systems under the influence of external environments @xcite . in our evolution model ,the extent of perturbation spread depends on the initial perturbation and on the network structure .the network structure is the outcome of the specific evolution pathway affected by the changing environment .the location of initial perturbation is determined on a random basis in our model , modeling the stochasticity of the internal microscopic dynamics in real systems .therefore the perturbation spread can be considered as a function of the internal dynamics component @xmath175 and the network structure @xmath176 , i.e. , @xmath177 .then the fluctuation of @xmath137 is represented as @xmath178 , where @xmath179 and @xmath180 represent the average over @xmath175 and @xmath176 as @xmath181 and @xmath182 and decomposed into the internal and the external fluctuation as @xcite : @xmath183 the internal fluctuation @xmath184 denotes the structural average of the internal - dynamics fluctuation of @xmath137 . on the other hand ,the external fluctuation @xmath185 is the structural fluctuation of the internal - dynamics average of @xmath137 . in simulations ,the quantities @xmath186 are obtained simply by the ensemble averages @xmath83 . to obtain @xmath187, we use the relation @xmath188@xcite , where @xmath189 and @xmath190 are the perturbation spreads from two different initial perturbations on the same network and are computed by eq .( [ eq : hpert ] ) with different perturbed states @xmath191 and @xmath192 from two initial perturbations . inserting @xmath193 and @xmath194 in eq .( [ eq : sigmas ] ) , one finds that the internal fluctuation is represented as @xmath195 and the external fluctuation is @xmath196 .the external fluctuation @xmath197 is found to be much larger than @xmath198 for all @xmath43 [ figure [ fig : fluctuation ] ( e ) ] , implying the wide variation of the network structure arising from exploiting differentiated pathways of evolution in changing environments .moreover , the external fluctuation displays a similar crossover behavior to @xmath166 , that is , @xmath199 with @xmath200 increasing from @xmath201 , a value close to @xmath123 , in the transient period to a value @xmath202 in the stationary state [ fig .[ fig : fluctuation ] ( f ) ] . on the other hand, the internal fluctuation behaves as @xmath203 with @xmath204 remaining close to @xmath123 , like in the diffusion process [ fig .[ fig : fluctuation](f ) ] . which is dominant of the internal andthe external fluctuation has been investigated for various complex systems @xcite .contrary to the static ( nature of ) systems of the previous works , the evolving networks in our model display a dynamic crossover in the fluctuation scaling while the external fluctuation is always dominant . to decipher the mechanism underlying this phenomenon , we begin with assuming that in the scaling regime the perturbation spread @xmath205 is small and factorized as @xmath206 where @xmath207 and @xmath208 are the components reflecting the dependence of perturbation spread on the location of initial perturbation and on the global network structure , respectively . @xmath207 and @xmath208 are expected to be independent .we assume that their fluctuations scale as @xmath209 and @xmath210 with @xmath211 and @xmath212 time - independent constants .then the mean of the perturbation spread should be given by @xmath213 and the internal and the external fluctuation in eq .( [ eq : sigmas ] ) are represented as @xmath214 using eqs .( [ eq : hdecomp0 ] ) and ( [ eq : fluctdecomp0 ] ) , we can analyze the scaling behaviors of fluctuations as follows . in the transient period before entering the stationary state , the network structure is transformed significantly , making the structural component @xmath215 essentially govern the perturbation spread in its time - dependent behavior , yielding @xmath216 this is supported by the similarity of the temporal patterns of @xmath117 and the mean connectivity @xmath84 in fig . [fig : fluctuation ] ( g ) .therefore one can relate the external fluctuation to the mean of perturbation spread as @xmath217 comparing this with the simulation results in fig .[ fig : fluctuation](f ) , we find that @xmath218 .that is , @xmath219 .the estimated value @xmath212 is also consistent with the simulation result @xmath220 , since @xmath221 , with @xmath222 given @xmath159 small in the scaling regime . in the stationary state ,the network structure varies little with time ; @xmath84 rarely varies ( fig .[ fig : fluctuation ] ( h ) ) .in contrast , @xmath117 fluctuates significantly on short time scales .this suggests that randomly - selected locations of initial perturbation , having no correlations at different time steps , drive such time - dependent behaviors of @xmath117 .therefore , from eqs .( [ eq : hdecomp0 ] ) and ( [ eq : fluctdecomp0 ] ) , the mean and the fluctuation of perturbation spread are represented as @xmath223 regardless of the value of @xmath211 , the external fluctuation is proportional to @xmath117 , @xmath224 in agreement with the observation @xmath225 in fig . [fig : fluctuation ] ( f ) .the internal fluctuation is expected to scale as @xmath226 , which allows us to find @xmath227 .therefore @xmath228 like @xmath219 .the above arguments following eqs .( [ eq : hdecomp0 ] ) and ( [ eq : fluctdecomp0 ] ) with @xmath229 illustrate why the internal fluctuation always scale as @xmath230 while the external fluctuation shows the dynamic crossover from @xmath231 to @xmath232 . combined with the observation that the external fluctuation makes a dominant contribution to @xmath166 ,the arguments explain the crossover in the fluctuation scaling of perturbation spread shown in fig .[ fig : fluctuation ] ( b ) .our results can be compared with the other cases showing a crossover in the fluctuation scaling driven by the change of the dominant fluctuation between @xmath233 and @xmath234 @xcite . on the other hand , @xmath234 is always dominant in our model .the time - varying perturbation spread is dominantly governed by the structure component @xmath208 in the transient period and the internal dynamics component @xmath207 in the stationary state , which underlies the crossover of @xmath163 from @xmath123 to @xmath8 in our model .the rapid and significant changes of the structure of the evolving networks are identified only in the transient period , and the internal stochasticity dominates the statistics of stability in the stationary state of evolution .therefore the nature of fluctuations is fundamentally different between the evolved networks and the random network or those which are not sufficiently evolved .the evolved networks in our model are more stable than random networks but less stable than the stability - only networks as shown by the scaling behaviors of @xmath120 in sec .[ sec : evolution ] . such balance between robustness and flexibilityis hardly acquired unless the relevant dynamical variables , the spread of perturbation in our case , at different sites are correlated with one another . for a quantitative analysis ,let us consider the local perturbation @xmath235 at node @xmath5 and time @xmath43 defined as @xmath236,\ ] ] denoting whether the activity of node @xmath5 is different between the original state @xmath21 and the perturbed state @xmath237 .notice that the stability hamming distance @xmath205 in eq .( [ eq : hpert ] ) is the spatial average of the local perturbations , @xmath238 .if node @xmath10 tends to have larger perturbation than its average when node @xmath5 does , @xmath239 , their local perturbations can be considered as correlated , meaning that local fluctuations at @xmath240 are likely to spread to node @xmath241 . in that case , we can expect that @xmath242 .therefore we define the correlation volume as @xmath243 which represents how many nodes are correlated with a node in the perturbation - spreading dynamics .for instance , @xmath244 if @xmath245 for all @xmath5 and @xmath10 ( perfect correlation ) and @xmath246 if the @xmath247 s are completely independent of one another such that @xmath248 . one can find that the variance of the perturbation spread @xmath249 is decomposed into the local variance @xmath250 and the correlation volume @xmath251 as @xmath252 where @xmath250 is defined in terms of the variance of @xmath235 as @xmath253 the decomposition in eq .( [ eq : variance ] ) allows us to see that the fluctuation of perturbation spread depends on the magnitude of local fluctuations , @xmath250 , and how far the local fluctuation propagates to the system , characterized by the correlation volume @xmath251 in eq .( [ eq : c ] ) .if the @xmath235 s are independent , the local fluctuation does not spread , as @xmath246 , and the whole variance @xmath254 is identical to the local variance @xmath255 . on the contrary , if the @xmath235 s are perfectly correlated , the correlation volume is @xmath256 and the whole variance @xmath254 is @xmath1 times larger than the local variance as @xmath257 , representing that local fluctuations spread to the whole system . .( a ) plots of @xmath251 versus time @xmath43 for @xmath85 .( b ) the initial correlation volume @xmath258 at @xmath91 and the stationary - state one @xmath259 averaged over the stationary period ( @xmath260 ) are plotted as functions of the system size @xmath1 for @xmath113 ( upper ) and for @xmath261 ( lower ) .@xmath259 scales with @xmath1 as @xmath262 or @xmath263 while @xmath258 does not increase with @xmath1 . ] in fig .[ fig : corrvol ] ( a ) , the correlation volume is shown to be larger in the stationary state than in the initial state .the correlation volume averaged over the stationary period , @xmath259 , is about @xmath264 while that in the initial state , @xmath258 , ranges between @xmath265 and @xmath107 for @xmath85 .the dependence of @xmath251 on the system size @xmath1 is different between the initial and the stationary states .furthermore , the correlation volume in the stationary state increases with @xmath1 as @xmath266 while the correlation volume of the initial network @xmath258 does not increase with @xmath1 [ fig .[ fig : corrvol ] ( b ) ] .such a scaling behavior is not seen in the whole fluctuation @xmath254 even in the evolved networks . therefore , the scaling behavior of the correlation volume in eq .( [ eq : cscaling ] ) can be another hallmark of the evolved systems and can be related to the system s capacity to be stable and adaptable simultaneously .in this work we have introduced and extensively investigated the characteristic properties of an adaptive network model capturing the generic features of biological evolution . in reality , the evolutionary selections are made for a population of heterogeneous living organisms , as adopted by the genetic algorithm , but here we considered a simplified model , where a single network , representing the network structure typical of a population of organisms , add or remove a link depending on whether that change improves its fitness or not .the fitness of a network is evaluated in terms of its adaptability to a changing environment and the stability against perturbations in the dynamical state , which look contradictory to each other but essential for every living organism . despite such simplification, the model network reproduces many of the universal network characteristics of evolving organisms , including the sparsity and scaling of the mean connectivity , broad degree distributions , and stability stronger than the random boolean networks but weaker than the networks evolved towards stability only , implying the simultaneous support of adaptability and robustness .fluctuations and correlations display characteristic scaling behaviors in the stationary state of evolution contrasted to those in the transient period or in the initial random - network state .the evolutionary pressure drives the regulatory networks towards becoming highly stable by exploiting different pathways from realization to realization in the rugged fitness landscape , which results in a large fluctuation .the presence of two distinct components in the perturbation - spread dynamics , related to the different network structure depending on the evolution pathway and the location of random initial perturbation , respectively , is shown to bring the dynamic crossover in the fluctuation scaling .such evolution makes large correlations as well .the proposed model is simple and generic allowing us to understand the evolutionary origin of the universal features of diverse biological networks .it illuminates the nature of dynamic fluctuations and correlations in evolving networks that are continuously influenced by the changing environments .the ensemble of those evolving networks can be formulated by the hamiltonian approach , which depends on a time - varying external environment , and thus it opens a way to study biological evolution from the viewpoint of statistical mechanics .given the increasing importance of the capacity to manipulate biological systems , natural or synthetic , our understanding of biological fluctuations can be particularly useful .the strong interaction with environments , like the natural selection in evolution , is common to diverse complex systems and thus the theoretical framework to deal with multiple components of dynamics presented here can be of potential use in substantiating the theory of complex systems .61ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ] + 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty http://dx.doi.org/10.1038/nrg1272 [ * * , ( ) ] @noop * * , ( ) http://dx.doi.org/10.1038/nature04209 [ * * , ( ) ] link:\doibase 10.1093/bioinformatics/19.2.270 [ * * , ( ) ] http://dx.doi.org/10.1038/35036627 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) http://www.karger.com/doi/10.1159/000067642 [ * * , ( ) ] http://dx.doi.org/10.1038/nrm2787 [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/s0378-4371(99)00291-5 [ * * , ( ) ] http://books.google.co.kr/books?id=st4lidk5no4c[__ ] ( , ) http://dx.doi.org/10.1038/nrg1523 [ * * , ( ) ] @noop _ _( , , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) \doibase http://dx.doi.org/10.1016/j.jtbi.2005.04.025 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) \doibase http://dx.doi.org/10.1016/j.jtbi.2010.08.006 [ * * , ( ) ] http://www.scopus.com/inward/record.url?eid=2-s2.0-0029661331&partnerid=40&md5=8d8ba66617ff4e7467a9d74b1e3148cf [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) \doibase http://dx.doi.org/10.1016/j.jtbi.2008.03.003 [ * * , ( ) ] http://stacks.iop.org/1367-2630/11/i=3/a=033005 [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/j.biosystems.2009.09.002 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physreve.81.021908 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1371/journal.pcbi.1002669 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1093/nar / gki892 [ * * , ( ) ] @noop link:\doibase 10.1103/physrevlett.108.128702 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1137/070710111 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( )

<S> diverse biological networks exhibit universal features distinguished from those of random networks , calling much attention to their origins and implications . here </S><S> we propose a minimal evolution model of boolean regulatory networks , which evolve by selectively rewiring links towards enhancing adaptability to a changing environment and stability against dynamical perturbations . </S><S> we find that sparse and heterogeneous connectivity patterns emerge , which show qualitative agreement with real transcriptional regulatory networks and metabolic networks . </S><S> the characteristic scaling behavior of stability reflects the balance between robustness and flexibility . </S><S> the scaling of fluctuation in the perturbation spread shows a dynamic crossover , which is analyzed by investigating separately the stochasticity of internal dynamics and the network structures different depending on the evolution pathways . </S><S> our study delineates how the ambivalent pressure of evolution shapes biological networks , which can be helpful for studying general complex systems interacting with environments . </S>

math0404112

in many problems one is led to consider in the euclidean plane lines joining a fixed point @xmath18 ( which is not necessarily an integer lattice point ) with a set of integer lattice points .a natural way of measuring the distribution of directions @xmath19 , @xmath20 , is via correlations and consecutive spacings .when the fixed point is the origin , the problem is related to the distribution of farey fractions with multiplicities , each fraction @xmath21 in @xmath22 being counted @xmath23 $ ] times .the consecutive @xmath24-level spacing measures of customary farey fractions were computed for @xmath25 in @xcite and for @xmath26 in @xcite . limiting correlations of farey fractionswere shown to exist and computed recently in @xcite . when the fixed point is not an integer lattice point , the problem of existence of limiting correlations / consecutive spacings is considerably more difficult .it is therefore natural to try to prove first some averaging results , letting the fixed point to vary in a given region . in the first part of this paperwe derive such a result for the limiting pair correlation measure .the limiting average pair correlation function is constant , as in the poisson case .what is striking however is that this constant is not @xmath27 , as in the poisson case , but @xmath28 .we now give a mathematical formulation of the problem . for each @xmath29 , let @xmath30 denote the set of integer lattice points in the square @xmath6 ^ 2 $ ] , and set @xmath31 . for each point @xmath32 , we consider the finite sequences @xmath33 , @xmath34 large integer , of angles between the line @xmath2 and the horizontal direction .the pair correlation of this finite sequence is defined as @xmath35 where @xmath36 denotes the measure of the angle @xmath37 . throughout the paperwe shall consider a fixed disc @xmath9 of center @xmath38 and radius @xmath39 .we are interested in the asymptotic behavior of the average @xmath40 of @xmath41 over @xmath9 , for fixed @xmath42 and @xmath10 .the first three sections are concerned with the proof of the following result .[ t1.1 ] for every @xmath43 and @xmath44 @xmath45 if one replaces @xmath9 by a vertical or horizontal segment of length one , an identical asymptotic formula as in turns out to be true .this can be proved by similar techniques as in this paper or using erd " os - tur ' an type discrepancy estimates , and suggests that may be true regardless of the shape of the range of the fixed point .the behavior of higher level correlations appears to be different . in the last sectionwe prove that the @xmath15-level correlations diverge for every fixed point .when @xmath46 , the repartition of the @xmath1-level correlation measure of the finite sequence @xmath47 is defined for each vector @xmath48 by @xmath49 for randomly chosen directions one would expect to obtain the poissonian limit @xmath50 = 2^{\nu-1 } \lambda_1 \dots\lambda_{\nu-1 } .\ ] ] it turns out however that fails in this situation .more precisely , we will show that if @xmath51 , then for every point @xmath14 and for every @xmath52 , @xmath53 .this is a consequence of [ t1.2 ] for every @xmath14 , every @xmath54 , and every @xmath55 , for @xmath34 large enough in terms of @xmath56 and @xmath57 @xmath58 as in theorem [ t1.2 ] one can prove [ c1.3 ] the @xmath15-level correlations of angles of directions @xmath2 , where @xmath5 is a lattice point inside an expanding region @xmath59 , diverges as @xmath10 whenever @xmath60 is a convex domain in @xmath61 which contains the origin .the phenomenon is similar to the one encountered in the problem of the distribution of fractional parts of polynomials .there , one can handle the pair correlation problem generically ( see @xcite,@xcite ) .moreover , in the case of the sequence @xmath62 one is able to solve the problem for all @xmath63-level correlations for a large class of irrational numbers @xmath64 ( see @xcite,@xcite ) .however , as shown in @xcite , there are irrational numbers @xmath64 for which the @xmath65-level correlation of fractional parts of @xmath66 , @xmath67 , diverges to infinity as @xmath68 .this occurs as a result of the presence of large clusters of such fractional parts . in the case of theorem [ t1.2 ] above, large clusters of elements of the given sequence are responsible , too , for the divergence of the @xmath15-level correlations , and hence of any other higher level correlations .for obvious practical reasons , we try to replace from the beginning @xmath70 by one of its trigonometric functions in the definition of @xmath71 .suppose that two distinct points @xmath72 , @xmath73 , are such that @xmath74 and @xmath75 .then for sufficiently large @xmath34 ( depending only on @xmath76 ) we have @xmath77 ^ 2}\vert \theta_{p , p^\prime}(x , y)\vert \geq \arcsin \frac{1}{\sqrt{q^2 + 1 } } > \frac{2\pi \lambda}{n } \ .\ ] ] as a result , we may only consider in the definition of @xmath7 points from the same quadrant .thus if we set @xmath78 and @xmath79 then @xmath80 for @xmath81 , @xmath82 , we define @xmath83 then @xmath84 for each @xmath85 , consider the weight @xmath86 from we infer that @xmath87 denote @xmath88 and define for every @xmath89 @xmath90 in the remainder of this section we show that the asymptotic of @xmath91 as @xmath10 is closely related to that of @xmath92 . for fixed @xmath93 , denote by @xmath94 the angle between the line @xmath95 determined by @xmath5 and @xmath96 and the horizontal direction .consider also the lines @xmath97 , parallel to @xmath95 and such that @xmath98 .the equation of @xmath95 is given by @xmath99 while the equation of @xmath97 is given by @xmath100 we see that @xmath101 the set whose area defines @xmath102 is the intersection of the strip bounded by @xmath103 and @xmath104 and the disc @xmath9 , thus @xmath105 we also have @xmath106 [ l2.1 ] let @xmath107 $ ] .let @xmath108 be a compact set in @xmath109 .then for all @xmath110 and all @xmath111 @xmath112 _ proof . _ the estimate reads as @xmath113 . combining it with we see that it suffices to show that @xmath114 taking @xmath115 , we gather @xmath116 ^ 2 \ , : \ , \vert a^{\prime \prime } q - aq^{\prime \prime}\vert \ll_c \| op^{\prime \prime}\| \big\ } .\end{split}\ ] ] the two conditions on @xmath117 above yield that @xmath117 should belong to the intersection of a strip of width @xmath118 bounded by the lines @xmath119 with the square @xmath120 ^ 2 $ ] .the number of integer lattice points inside this region is of order @xmath121 , thus @xmath122 since @xmath123 , this gives @xmath124 as desired .[ l2.2 ] for every compact set @xmath125 and every @xmath110 , there exist constants @xmath126 such that @xmath127 _ proof . _the trivial estimate @xmath128 and yield for all @xmath85 , @xmath129 , @xmath130 , that @xmath131 we first analyze the case where @xmath132 . in this case@xmath133 , and the relation above and the definitions of @xmath134 and @xmath135 , yield @xmath136 such that @xmath137 when @xmath138 , we take @xmath139 in lemma [ l2.1 ] . since @xmath140 as @xmath10 , we get @xmath141an immediate consequence of and is that the contribution to @xmath142 of pairs of points @xmath143 with @xmath144 or with @xmath145 is negligible . indeed , we see from that when @xmath146 , the term @xmath102 is zero unless @xmath147 , thus the total contribution of such points to @xmath142 is @xmath148 the contribution of pairs of points @xmath143 with @xmath149 to @xmath150 is @xmath151 similar estimates in the case @xmath145 show that @xmath152 as a result , we shall subsequently assume that @xmath153 and @xmath154 .we now set @xmath155 the remainder of this section is elementary and is concerned with putting @xmath142 in a tidy form , suitable for a precise estimation which will be completed in the next section .let @xmath156 denote the center of @xmath9 , let @xmath157 be the line passing through @xmath156 and perpendicular to @xmath95 , and denote by @xmath158 and @xmath159 the intersections of @xmath157 with the circle @xmath160 , by @xmath161 the intersection of @xmath157 and @xmath95 , and by @xmath162 the intersection of @xmath157 with @xmath97 .direct computation gives @xmath163 and @xmath104 with the disc @xmath9 . ]while ordering the points @xmath164 and @xmath165 the following situations may occur : * case 1 .* @xmath166 , that is @xmath167 this gives @xmath168 , hence @xmath169 suppose first that @xmath170 . by we know that for fixed @xmath171 , the expression @xmath172 will only take values between @xmath173 and @xmath174 .hence the number of solutions @xmath175 of @xmath176 is of order @xmath177 , where @xmath178 is the greatest common divisor of @xmath179 and @xmath180 . but@xmath181 , hence this is actually @xmath182 and the contribution to @xmath183 is @xmath184 the case @xmath185 is settled similarly by first summing over @xmath175 . * case 2 .* @xmath186 , that is @xmath187 or equivalently @xmath188 the change of variables @xmath189 , @xmath190 gives @xmath191 so , keeping @xmath192 and @xmath193 fixed , the range of @xmath194 has cardinality @xmath182 . now the equation @xmath195 has either no solution @xmath117 when @xmath196 does not divide @xmath197 , or has @xmath198 solutions @xmath117 when @xmath178 divides @xmath197 .thus the contribution of terms @xmath135 with @xmath199 is @xmath200 \\\gcd ( q_0^{\prime \prime},a_0^{\prime \prime})=1 } } \frac{q}{q_0^{\prime \prime } } \cdot \frac{1}{\sqrt{q } } \ \leq \ \frac{1}{q\sqrt{q } } \sum\limits_{d=1}^q \frac{q}{d } \sum\limits_{0<\vert q_0^{\prime \prime}\vert \leq \big [ \frac{q}{d}\big ] } \frac{1}{q_0^{\prime \prime } } \ \ll \\frac{\ln^2 q}{\sqrt{q } } \\ll_\delta q^{-\frac{1}{2}+\delta } .\ ] ] the contribution of terms @xmath135 with @xmath201 is @xmath202 } \frac{q}{q_0^{\prime \prime } } \cdot \frac{1}{dq_0^{\prime \prime } } \\frac{1}{q } \sum\limits_{d , q_0^{\prime \prime}=1}^\infty \frac{\sqrt{q}}{d^2 q_0^{\prime \prime } } \ \ll \ q^{-\frac{1}{2 } } \ .\ ] ] * case 3 . *@xmath203 , that is @xmath204 we infer as in case 2 that the contribution of @xmath135 is in this case too @xmath205 .* @xmath206 , that is @xmath207 or equivalently @xmath208 denote @xmath209 and @xmath210 .the interval @xmath211 $ ] has length @xmath212 .hence we find that the contribution of terms @xmath135 for which @xmath213 to @xmath183 is @xmath214 } \frac{1}{k_0 \sqrt{k_0 ^ 2+\ell_0 ^ 2 } } \ll \frac{1}{q } \sum\limits_{d=1}^q \frac{1}{d } \sum\limits_{k_0=1}^{\big [ \frac{q}{d}\big ] } \frac{1}{k_0 } \sum\limits_{\ell_0=1}^{\big [ \frac{q}{d}\big ] } \frac{1}{\ell_0 } \\ & \ll \frac{\ln^3 q}{q } \ . \end{split}\ ] ] one shows similarly that the contribution of points @xmath93 for which @xmath215 is of the same order .therefore by and the previous considerations we infer that @xmath216 next @xmath217 is approximated by elementary calculus .[ lii1 ] the area of the region inside the circle of radius @xmath39 centered at the origin and inside the strip bounded by the vertical lines @xmath218 and @xmath219 , @xmath220 , is given , for small @xmath221 , by @xmath222 _ proof . _the error is seen to be given ( see figure 2 ) by @xmath223 it is @xmath224 as a result of @xmath225 we take @xmath226 notice now that @xmath227 denote @xmath209 , @xmath210 , @xmath228 , \ ] ] and find that the contribution of the error provided by lemma [ lii1 ] in is @xmath229 } \frac{qd}{dk_0 } \cdot d(k_0 ^ 2+\ell_0 ^ 2)^{\frac{1}{2 } } \cdot \frac{1}{d^{3/2 } ( k_0 ^ 2+\ell_0 ^ 2)^{3/4 } } \\ & = \frac{1}{q } \sum\limits_{d=1}^q \frac{1}{d^{1/2 } } \sum\limits_{k_0,\ell_0=1}^{\big [ \frac{q}{d}\big ] } \frac{1}{k_0(k_0 ^ 2+\ell_0 ^ 2)^{1/4 } } \\ & \leq \frac{1}{q } \sum\limits_{d=1}^q \frac{1}{d^{1/2 } } \sum\limits_{k_0}^{\big [ \frac{q}{d}\big ] } \frac{1}{k_0 } \sum\limits_{\ell_0=1}^{\big [ \frac{q}{d}\big ] } \frac{1}{\ell_0^{1/2 } } \\ & \ll \frac{\ln q}{q } \sum\limits_{d=1}^q \frac{1}{d^{1/2 } } \cdot \frac{q^{1/2}}{d^{1/2 } } \\ & \ll \frac{\ln^2 q}{q^{1/2 } } \ . \end{split}\ ] ] therefore by , lemma [ lii1 ] , and we find that @xmath230 where @xmath231 denotes the contribution of the main term in lemma [ lii1 ] to , that is @xmath232 finally we show that , if @xmath233 , one can replace @xmath234 by @xmath235 in and . since @xmath236 , then @xmath237 , and @xmath238 this gives @xmath239 and similarly @xmath240 hence one can replace @xmath241 in by @xmath242 at the cost of an error which is @xmath243 } \frac{q}{k_0 } \ll \frac{1}{q^2 } \sum\limits_{d=1}^q \frac{q}{d}\cdot \ln q = \frac{\ln^2 q}{q } \ . \end{split}\ ] ] in summary , we have shown for any @xmath244 in a fixed compact set @xmath125 , that @xmath245 as @xmath10 .by reflecting @xmath9 about the axes and about the line @xmath247 , we see that it suffices to only estimate the contribution @xmath248 to @xmath150 of points @xmath249 ^ 2 $ ] with @xmath250 .we thus consider @xmath251 where @xmath252 then we gather from and the above formula for @xmath246 that @xmath253 changing @xmath179 to @xmath254 and @xmath255 to @xmath256 , we may write @xmath257 putting @xmath258 and taking @xmath259 as in , that is @xmath260,\ ] ] we get @xmath261 \\a^\prime \in [ a , q ] \\ aq^\prime -a^\prime q = d } } \frac{(q^\prime -q)q^\prime \sqrt{r_0 ^ 2(q^2+a^2)-(y_0 q - x_0 a+d)^2}}{q^2 } \ .\ ] ] we will prove the following result .[ p3.1 ] @xmath262 from this and we infer the following [ c3.2 ] @xmath263 theorem [ t1.1 ] now follows combining corollary [ c3.2 ] with lemma [ l2.2 ] .we now start the proof of proposition [ p3.1 ] .we first lay out some notation and prove an elementary calculus lemma .fix @xmath264 and consider the function @xmath265 and the domain @xmath266 and @xmath267.,title="fig : " ] and @xmath267.,title="fig : " ] consider also the projection @xmath268 of @xmath269 on the second coordinate , the @xmath270-section @xmath271\ , : \, \phi(t , x)\geq 0\},\ ] ] and the @xmath272-section @xmath273 define the function @xmath274 by @xmath275 [ l3.0 ] for every @xmath264 one has @xmath276 @xmath277 _ proof . _( i ) by the definition of @xmath278 it is seen that @xmath279 is the union of one or two intervals @xmath280 $ ] , where @xmath281 and @xmath282 are equal to @xmath283 , @xmath27 , or a root of @xmath284 . in all these cases @xmath285 and as a result we get @xmath286 and @xmath287 using the triangle inequality and the change of variables @xmath288 we obtain @xmath289 \(ii ) the same change of variable as in ( i ) gives @xmath290 we start to evaluate @xmath246 . if @xmath291 , then @xmath292 .the equality @xmath293 is equivalent to @xmath294 .hence in the inner sum we should sum over @xmath295 $ ] such that @xmath296 and @xmath297 or equivalently @xmath298 next we show that the bulk of the contribution to @xmath246 of only comes from @xmath299 to see this , notice first that for @xmath180 fixed , the relations @xmath300 and @xmath301 imply that @xmath302 takes at most @xmath303 $ ] values .so the total contribution of terms with @xmath304 $ ] to @xmath246 is @xmath305)\cdot \frac{(q^\prime - q)q^\prime}{q^2}\cdot \sqrt{2 } r_0 q\ll_{r_0 } \frac{1}{q } \sum\limits_{q=1}^q \frac{1}{q }\ll \frac{\ln q}{q}\ .\ ] ] when @xmath306 $ ] , we get @xmath307)q\leq ( q - a)q$ ] , thus @xmath308 .suppose now that @xmath180 is between @xmath179 and @xmath309 .owing again to @xmath310 , it follows that @xmath311 for some integer @xmath312 .but the range of @xmath180 is an interval of length @xmath313 , hence the range of @xmath314 is an interval of length @xmath315 $ ] .thus @xmath312 , and consequently @xmath180 , take at most @xmath316 values .besides , in this case we have @xmath317 .hence the contribution to @xmath246 of terms with @xmath180 between @xmath179 and @xmath309 is @xmath318 therefore we have shown that we can replace the summation conditions in the inner sum from by @xmath319 $ ] and @xmath320 .we write @xmath321 and @xmath322 .take @xmath323 , @xmath324 and @xmath325 , unless otherwise specified , and note that @xmath326 then and the previous considerations lead to @xmath327 \\aq^\prime = d \hspace{-8pt}\pmod{q } } } \frac{(q^\prime -q)q^\prime}{q}\ \sqrt{\phi \bigg ( \frac{a}{q},\frac{d}{r_0q}\bigg)}\ + o(q^{-1}\ln q ) \\ & = \frac{2r_0}{q^4 } \sum\limits_{q=1}^q \frac{1}{q } \sum\limits_{d\in r_0 q\mathrm{pr}_2 { { \mathcal{d}}}}\sum\limits_{\substack{q^\prime \in [ q , q ] \\ a\in qi_{\hspace{-2pt}\frac{d}{r_0q } } \\ aq^\prime = d\hspace{-8pt}\pmod{q } } } ( q^\prime -q)q^\prime \sqrt{\phi \bigg(\frac{a}{q},\frac{d}{r_0 q}\bigg ) } \ + o(q^{-1}\lnq ) , \end{split}\ ] ] where @xmath279 and @xmath328 are defined as in and .take @xmath329 and write @xmath330 , @xmath331 .then @xmath178 divides @xmath302 , so @xmath332 .the congruence @xmath333 is equivalent to @xmath334 , and we may write @xmath335 } \frac{d}{q_0 } \sum\limits_{d_0\in r_0 q_0\mathrm{pr}_2 { { \mathcal{d } } } } \hspace{-10pt } \sum\limits_{\substack{q_0^\prime \in \big [ q_0,\frac{q}{d}\big ] \\\gcd ( q_0^\prime , q_0)=1 \\dq_0 i_{\hspace{-3pt } \frac{d_0}{r_0q_0 } } \\ aq_0^\prime = d_0 \hspace{-8pt } \pmod{q_0 } } } \hspace{-15pt } ( q_0^\prime -q_0)q_0^\prime \sqrt{\phi \bigg ( \frac{a}{dq_0},\frac{d_0}{r_0q_0}\bigg ) }\ + o(q^{-1}\ln q).\ ] ] to estimate the inner sum above , we need some information about the distribution of solutions of the congruence @xmath336 .we shall employ the following result , which is a consequence of proposition [ pa2 ] from the appendix .[ l3.3 ] assume that @xmath337 and @xmath24 are two given integers , @xmath338 and @xmath339 are intervals , and @xmath340 is a @xmath341 function .then for every integer @xmath342 and every @xmath44 @xmath343 where @xmath344 .we now return to the formula for @xmath246 given in and first give an upper bound for the contribution to @xmath246 of quadruples @xmath345 for which @xmath346 with @xmath347 to be chosen later .[ l3.4 ] let @xmath348 with @xmath349 .then for any @xmath197 and @xmath350 @xmath351 _ proof . _using @xmath352 we see that it suffices to consider the case @xmath353 . in this casethe statement follows from the fact that the double inequality @xmath354 is equivalent to @xmath355 and from the inequality @xmath356 suppose that @xmath357 is fixed and consider the following two cases : * case 1 ) * @xmath358 . by and lemma [ l3.4 ]the range of @xmath255 is the union of at most two intervals of length @xmath359 .hence @xmath255 can only assume @xmath360 values .but , for each @xmath255 , @xmath361 belongs to @xmath362 $ ] and is subject to the condition @xmath363 .hence @xmath361 takes @xmath364 values . thus the contribution to @xmath246 of quadruples@xmath345 which satisfy is @xmath365 * case 2 ) * @xmath366 thus @xmath367 .in this case we collect directly from @xmath368 hence @xmath255 can only assume @xmath369 values and we find , arguing as in case 1 , that the contribution to @xmath246 of quadruples @xmath345 which satisfy is @xmath370 next we investigate the case @xmath371 we consider the range of @xmath361 : @xmath372 , \ ] ] the range of @xmath255 ( which is the union of at most two intervals ) : @xmath373\ , : \ , \phi \bigg ( \frac{y}{dq_0},\frac{d_0}{r_0q_0}\bigg ) \geq\frac{l^2}{r_0 ^ 2d^2 q_0 ^ 2}\bigg\ } \subseteq dq_0 i_{\hspace{-3pt}\frac{d_0}{r_0q_0 } } \ , \ ] ] and the functions @xmath374 , \\ f(x , y)=f_{q_0,d_0,d}(x , y ) & : = g(x ) \psi(y ) , \qquad ( x , y)\in { \mathcal{i}}_{q_0,d } \times { \mathcal{j}}_{q_0,d_0,d , l } .\end{split}\ ] ] with this notation the following estimates hold on @xmath375 : @xmath376 applying proposition [ l3.3 ] we find that @xmath377 \\ \gcd ( q_0^\prime , q_0)=1 \\ a\in { \mathcal{j}}_{q_0,d_0,d , l } \\ aq_0^\prime = d_0 \hspace{-8pt}\pmod{q_0 } } } \hspace{-15pt } ( q_0^\prime - q_0)q_0 & \sqrt{\phi \bigg ( \frac{a}{dq_0},\frac{d_0}{r_0q_0}\bigg ) } = \sum\limits_{\substack{q_0^\prime \in \big [ q_0,\frac{q}{d}\big ] \\\gcd ( q_0^\prime , q_0)=1 \\ a\in { \mathcal{j}}_{q_0,d_0,d , l } \\ aq_0^\prime = d_0 \hspace{-8pt}\pmod{q_0 } } } \hspace{-15pt } g_{q_0}(q_0^\prime ) \psi_{q_0,d_0,d}(a)\\ & = \frac{\varphi(q_0)}{q_0 ^ 2 } \int\limits_{q_0}^{\frac{q}{d } } g_{q_0}(x)\ , dx \hspace{-10pt } \int\limits_{{\mathcal{j}}_{q_0,d_0,d , l } } \hspace{-15pt } \psi_{q_0,d_0,d}(y)\ , dy + { \mathcal{e}}_{q_0,d_0,d , l } , \end{split}\ ] ] where @xmath378 using @xmath379 we see that @xmath380 taking @xmath381 $ ] and @xmath382 we find that @xmath383 thus the total contribution of @xmath384 to @xmath246 is @xmath385 moreover , the quantities in and become @xmath386 and respectively @xmath387 thus we gather @xmath388 with @xmath389 } \frac{d}{q_0 } \sum\limits_{d_0 \in r_0 q_0 \mathrm{pr}_2 { { \mathcal{d } } } } \frac{\varphi(q_0)}{q_0 ^ 2 } \int\limits_{q_0}^{\frac{q}{d } } g_{q_0}(x)\ , dx \hspace{-15pt } \int\limits_{{\mathcal{j}}_{q_0,d_0,d , q_0^{9/10 } } } \hspace{-20pt } \psi_{q_0,d_0,d}(y)\ dy.\ ] ] next we show that we can replace @xmath390 by @xmath391 in the last integral . clearly @xmath392 and by lemma [ l3.4 ] we have @xmath393 thus @xmath394 and as a result the error that results from replacing @xmath395 by @xmath396 in @xmath397 is @xmath398 } \frac{d\varphi(q_0)}{q_0 ^ 3 } \cdot \frac{q_0^{4/5}}{d } \int\limits_{q_0}^{\frac{q}{d } } g_{q_0}(x)\ , dx.\ ] ] on the other hand we find that @xmath399 where @xmath400 in particular , the integral of @xmath401 on @xmath402 $ ] is @xmath403 , and we find that the total cost of replacing @xmath390 by @xmath404 in @xmath397 is @xmath398 } \frac{\varphi(q_0)}{q_0 ^ 3}\cdot q_0^{\frac{4}{5 } } \cdot \frac{q^3}{d^3 } \leq \frac{1}{q } \sum\limits_{d=1}^\infty \frac{1}{d^3 } \sum\limits_{q_0=1}^\infty q_0^{-\frac{6}{5 } } \ll \frac{1}{q } \ .\ ] ] thus we infer that @xmath405 } \frac{d}{q_0}\cdot\frac{\varphi(q_0)}{q_0 ^ 2 } \cdot q_0 ^ 3 \ , g\bigg ( \frac{q}{dq_0}\bigg ) \hspace{-2pt}\sum\limits_{d_0 \in r_0q_0 \mathrm{pr}_2 { { \mathcal{d}}}}\ \int\limits_{dq_0i_{\hspace{-3pt}\frac{d_0}{r_0q_0 } } } \hspace{-10pt } \psi_{q_0,d_0,d}(y)\ , dy + o(q^{-1 } ) \\ & = \frac{2r_0}{q^4 } \sum\limits_d \sum\limits_{q_0=1}^{\big [ \frac{q}{d}\big ] } d\varphi(q_0)\ , g\bigg ( \frac{q}{dq_0}\bigg ) \sum\limits_{d_0\in r_0q_0 \mathrm{pr}_2 { { \mathcal{d}}}}\ \ \int\limits_{dq_0i_{\hspace{-3pt}\frac{d_0}{r_0q_0 } } } \hspace{-10pt } \sqrt{\phi \bigg ( \frac{y}{dq_0},\frac{d_0}{r_0q_0}\bigg)}\ dy + o(q^{-1 } ) \\ & = \frac{2r_0}{q^4 } \sum\limits_d \sum\limits_{q_0=1}^{\big [ \frac{q}{d}\big ] } d^2 q_0 \varphi(q_0 ) \ , g\bigg ( \frac{q}{dq_0}\bigg ) \sum\limits_{d_0\in r_0q_0 \mathrm{pr}_2 { { \mathcal{d}}}}\ \int\limits_{i_{\hspace{-3pt}\frac{d_0}{r_0q_0 } } } \hspace{-7pt } \sqrt{\phi \bigg ( t,\frac{d_0}{r_0 q_0}\bigg ) } \ dt + o(q^{-1 } ) \\ & = \frac{2r_0}{q^4 } \sum\limits_d \sum\limits_{q_0=1}^{\big [ \frac{q}{d}\big ] } d^2 q_0 \varphi(q_0 ) \ , g\bigg ( \frac{q}{dq_0}\bigg ) \sum\limits_{d_0\in r_0q_0 \mathrm{pr}_2 { { \mathcal{d } } } } \hspace{-8pt } \psi \bigg ( \frac{d_0}{r_0q_0}\bigg ) + o(q^{-1 } ) . \end{split}\ ] ] by euler - maclaurin summation and lemma [ l3.0 ] the inner sum above is given by @xmath406 inserting this back into we obtain @xmath407 } \varphi(q_0 ) ( dq_0)^2 g\bigg ( \frac{q}{dq_0}\bigg ) + o\bigg ( \frac{\ln q}{q}\bigg ) \\\big ( q = dq_0\in [ 1,q]\big ) \qquad & = \frac{4\pi r_0 ^ 2}{3q^4 } \sum\limits_{q=1}^q q^2 g\bigg( \frac{q}{q}\bigg ) \sum\limits_{q_0 \vert q } \varphi(q_0 ) + o\bigg ( \frac{\ln^2 q}{q}\bigg ) \\ & = \frac{4\pi r_0 ^ 2}{3q^4 } \sum\limits_{q=1}^q q^3 g\bigg ( \frac{q}{q}\bigg ) + o\bigg ( \frac{\ln^2 q}{q}\bigg ) \\ &= \frac{4\pi r_0 ^ 2}{3 } \int\limits_0 ^ 1 \bigg ( \frac{(1-x)^3}{3}+\frac{x(1-x)^2}{2 } \bigg ) dx+o\bigg ( \frac{\ln^2 q}{q } \bigg ) \\ & = \frac{\pi r_0 ^ 2}{6}+o\bigg ( \frac{\ln^2 q}{q}\bigg ) .\end{split}\ ] ] proposition [ p3.1 ] now follows from and .in this section we prove theorem [ t1.2 ] .we first prove a counting result .[ l4.1 ] let @xmath408 be positive integers with @xmath409 .let @xmath410 denote the largest divisor of @xmath178 which is relatively prime with @xmath411 , and put @xmath412. then @xmath413 _ proof ._ using m " obius inversion we express the left - hand side of as @xmath414 note that if @xmath302 does not divide @xmath410 , then there is no @xmath63 for which @xmath415 . indeed , if for some @xmath63 we have @xmath416 , then @xmath417 for some @xmath418 .hence @xmath255 is divisible by @xmath419 .then @xmath419 divides @xmath409 , so @xmath420 , and by the definition of @xmath410 it follows that @xmath302 divides @xmath410 .therefore @xmath421 for @xmath422 we have @xmath420 and there is a unique solution @xmath423 to the congruence @xmath424 .thus there are exactly @xmath425 values of @xmath63 in @xmath426 for which @xmath415 . as a resultwe infer that @xmath427 which proves the lemma . for any positive integers @xmath428 , consider the set @xmath429 [ l4.2 ] for @xmath179 large and @xmath430 such that @xmath431 @xmath432 * remark . *if in the definition of @xmath433 we took the range of @xmath434 and @xmath435 to be @xmath436 $ ] instead of @xmath437 $ ] , the cardinality of @xmath433 would be much smaller .for example , if @xmath438 , @xmath439 , and @xmath440 , then @xmath441 , and in the range @xmath442 this forces @xmath443 .then the only pair @xmath444 with @xmath445 is @xmath446 , so @xmath433 will only contain one element .lemma [ l4.2 ] shows a sudden increase in the cardinality of @xmath433 when the range of @xmath434 and @xmath435 increases by a factor @xmath447 ._ proof of lemma [ l4.2 ] ._ to make a choice , assume in what follows next that @xmath448 .then @xmath449 since @xmath450 .it follows that for any solution @xmath444 to the congruence @xmath451 , @xmath434 has to be divisible by @xmath178 .write @xmath452 and @xmath453 , so @xmath454 .denote by @xmath455 the multiplicative inverse of @xmath456 in the interval @xmath457 $ ] .note that since @xmath179 is divisible by the product @xmath458 , we have @xmath459. therefore @xmath460 .so @xmath461 is large for large @xmath179 , and by the prime number theorem we know that @xmath462 for any fixed prime @xmath463 with @xmath464 , we count the solutions @xmath465 of the congruence @xmath466 this is equivalent to @xmath467 since @xmath468 , this congruence has a unique solution modulo @xmath461 , namely @xmath469 .denote by @xmath470 the solution to which belongs to the interval @xmath471 $ ] .then will have @xmath472 solutions , given by @xmath473 it remains to be seen how many of the numbers @xmath435 from are relatively prime with @xmath474 .note first that at most two such numbers @xmath435 are divisible by @xmath463 .assume that @xmath475 with @xmath476 .since @xmath477 , @xmath463 does not divide @xmath461 .then it follows from that @xmath478 here at least one of the differences @xmath479 , @xmath480 is less than @xmath178 , and @xmath481 , so it is impossible that both divisibilities in hold true . therefore at most two numbers from are divisible by @xmath463 .note also , by the same reasoning , that for smaller values of @xmath178 - more precisely for @xmath482 , that is @xmath483 - one has @xmath484 .then one concludes that at most one number @xmath435 from can be a multiple of @xmath463 .we now count the numbers @xmath435 from which are relatively prime with @xmath178 .we claim that @xmath485 .indeed , let us assume this fails and choose a prime divisor @xmath486 of @xmath487 .since @xmath488 , we have @xmath489 and @xmath490 . recall that @xmath470 satisfies , hence @xmath491 for some @xmath418 . here@xmath492 , @xmath493 , so @xmath486 must also divide the left side of .the inequalities @xmath494 show that @xmath495 , so @xmath463 divides @xmath411 .but then @xmath496 , and we obtain a contradiction .this shows that @xmath497 .then lemma [ l4.1 ] is applicable to @xmath498 .if @xmath410 denotes the largest divisor of @xmath178 which is relatively prime with @xmath461 , and @xmath412 , then lemma [ l4.1 ] provides @xmath499 it follows in particular that there are always at least two numbers @xmath435 as in for which @xmath500 , and as soon as @xmath501 or @xmath502 , there are at least four such numbers . since at most two numbers @xmath435 as in are divisible by @xmath463 , and for small values of @xmath178 we know that at most one number @xmath435 as in is divisible by @xmath463 , we conclude that in all cases we have @xmath503 combining with we infer that @xmath504 and the lemma is completed using the inequalities @xmath505 let now @xmath179 be a large positive integer , let @xmath506 such that @xmath431 , and let @xmath34 be a positive integer larger than @xmath179 . in our applications @xmath34 will be at least of the order of magnitude of @xmath507 .we will construct some sets of lattice points inside the square @xmath508 ^ 2 $ ] , indexed by the set @xmath433 .precisely , we select a positive integer @xmath509 , which will be chosen later to be the integer part of a certain fractional power of @xmath34 , and for each pair @xmath510 we construct a set @xmath511 as follows . fix @xmath512 . to make a choice assume that @xmath513 .let @xmath108 be the integer defined by the equation @xmath514 denote by @xmath515 the unique integer satisfying @xmath516 where @xmath517 is the multiplicative inverse of @xmath435 modulo @xmath434 , and put @xmath518 then @xmath519 is an integer .also , from the inequalities @xmath520 it follows that @xmath521 .hence @xmath522 so that @xmath523 .thus @xmath524 , since @xmath519 is an integer .let now @xmath525 $ ] , and define @xmath526 to be the set of lattice points given by @xmath527 note that the case @xmath443 can only occur when @xmath528 , and in this situation we also get @xmath529 , @xmath530 , and so @xmath531 . we have constructed @xmath532 sets of the form @xmath526 , each set @xmath526 consisting of @xmath509 lattice points .note that @xmath533 in the definition of @xmath526 depend on the pair @xmath444 . in what followswe assume that @xmath509 satisfies the inequality @xmath534.\ ] ] define also @xmath535 some properties of these sets are collected in the following lemma .[ l4.3 ] @xmath536 ^ 2,{\widetilde{{\mathcal{m}}}}_{a , b , q}\big)\geq \displaystyle \frac{q}{3}\ , .$ ] @xmath537 ^ 2 $ ] .the sets @xmath526 are disjoint . _ proof ._ ( i ) owing to and to the inequality @xmath538-\bigg [ \frac{x}{2}\bigg ] \geq \frac{x}{2}-1,\ ] ] we have for any @xmath512 and any point @xmath539 @xmath540-\bigg [ \frac{q}{4q}\bigg ] \bigg ) a\geq \bigg ( \bigg [ \frac{q}{a}\bigg ] -\bigg [ \frac{q}{2a}\bigg ] \bigg ) a\geq \frac{q}{2}-a\geq \frac{q}{2}-2q.\ ] ] recall that @xmath34 is much larger than @xmath179 .it follows that the distance between any two points @xmath541 ^ 2 $ ] and @xmath542 satisfies @xmath543 \(ii ) for any @xmath512 with , say , @xmath544 , and any point @xmath545 , one has @xmath546 . also , @xmath547b\leq q$ ] , since @xmath513 .hence all points @xmath548 lie inside the square @xmath508 ^ 2 $ ] .\(iii ) assume that there is a lattice point @xmath549 which belongs to two sets @xmath526 and @xmath550 with @xmath551 and @xmath552 .assume first that @xmath513 and @xmath553 .then @xmath554 for some @xmath555 , and similarly @xmath556 for some @xmath557 , with @xmath558 given by appropriate definitions .we compute the ratio @xmath559 in two ways .first , by , and the equality @xmath560 we have @xmath561 by a similar computation we also have @xmath562 since @xmath563 are all positive and @xmath564 , this forces @xmath565 and @xmath566 .in general we get by the same argument @xmath567 and @xmath568 , thus @xmath569 . if @xmath570 and @xmath571 , we get @xmath572 , hence @xmath573 and @xmath574 .fix now a point @xmath575 ^ 2 $ ] and a pair @xmath510 .also , choose any two points @xmath576 , say @xmath577 and @xmath578 , with @xmath579 . consider the angle @xmath580 .then by @xmath581 next , using the equality @xmath560 , we rewrite as @xmath582 since @xmath583 and @xmath584 , we see that @xmath585 . if @xmath509 satisfies , then @xmath586 . as a consequence of wealso have @xmath587 from we also infer that @xmath588 by and we derive that @xmath589 as a consequence of and of the inequalities @xmath590 , we have @xmath591 uniformly for all pairs @xmath512 and all pairs of points @xmath592. _ proof of theorem [ t1.2 ] . _fix @xmath593 ^ 2 $ ] , @xmath594 with @xmath595 , and @xmath44 .the case when both @xmath270 and @xmath596 are rational numbers is clear . in this case , if we fix an integer @xmath597 for which both @xmath598 and @xmath599 are integers , and consider the set of lattice points @xmath600\big\}$ ] , then all these points lie on the same line , that passes through @xmath3. then all the @xmath15-tuples of distinct elements from @xmath601 will contribute to @xmath16 .since @xmath602 $ ] , it follows that @xmath603 note that in this case the @xmath604-level correlations already diverge as @xmath10 .consider now the case when at least one of @xmath605 is irrational . with@xmath56 and @xmath57 fixed , choose a large positive integer @xmath34 .let @xmath606 be a parameter , whose precise value will be chosen later and will be the integer part of a fractional power of @xmath34 . by minkowski s convex body theorem ( see ( * ? ? ?* thm.6.25 ) for the formulation used here ) , there exists an integer @xmath607 for which @xmath608 where @xmath609 denotes here the distance to the closest integer .let @xmath255 and @xmath411 denote the closest integers to @xmath610 and @xmath611 respectively .then @xmath612 , @xmath613 , and gives @xmath614 dividing if necessary @xmath255 , @xmath411 and @xmath179 by @xmath615 , we may assume in what follows that @xmath616 .we will have @xmath617 as @xmath10 , and since at least one of @xmath605 is irrational , this forces @xmath618 as @xmath10 .then all our previous results valid for large @xmath179 are applicable .let @xmath509 be a positive integer satisfying , whose precise order of magnitude will be chosen later .consider the disjoint subsets @xmath526 of @xmath30 , with @xmath510 .by we know that for any @xmath510 and any @xmath619 , the measure of the angle @xmath37 satisfies @xmath620 plugging in we find that @xmath621 if we take the order of magnitude of @xmath509 to be slightly smaller than that of both @xmath622 and @xmath623 , for instance @xmath624,\bigg [ \frac{\sqrt{t}}{\ln q}\bigg]\bigg\ } , \ ] ] then @xmath509 will satisfy on the one hand , and on the other hand we will have @xmath625 it follows that for @xmath34 large enough in terms of @xmath626 , all the @xmath15-tuples @xmath627 of distinct points from @xmath526 will contribute to @xmath16 . therefore , since @xmath628 for each @xmath512 , we derive that @xmath629 here @xmath630 , and from and lemma [ l4.2 ] it follows that @xmath631 we now choose @xmath632 $ ] .then , no matter how small @xmath179 might be , we have @xmath633 also , since @xmath634 , it follows that @xmath635 now follows from , and .for a fixed integer @xmath636 , we consider for any integers @xmath637 and any sets @xmath638 the kloosterman sum @xmath639 the incomplete kloosterman sum @xmath640 and the set @xmath641 here @xmath642 denotes the multiplicative inverse of @xmath643 . _ proof ._ we write @xmath655 where the main term can be expressed , as a result of lemma [ la1 ] , as @xmath656 the error is given by @xmath657 owing to , lemma [ la2 ] , and to the inequality @xmath658 , we have @xmath659 writing @xmath660 , with @xmath661 , we eventually get @xmath662 } \frac{1}{m } \ll \gcd ( h , q)^{\frac{1}{2 } } q^{\frac{1}{2}+2{\varepsilon } } \log q\ll \gcd ( h , q)^{\frac{1}{2 } } q^{\frac{1}{2}+3{\varepsilon}}.\ ] ] [ pa2 ] assume that @xmath337 and @xmath24 are two given integers , @xmath338 and @xmath339 are intervals of length lesser than @xmath179 , and @xmath663 is a @xmath341 function .then for any integer @xmath342 and any @xmath44 @xmath664

<S> let @xmath0 denote the repartition of the @xmath1-point correlation measure of the finite set of directions @xmath2 , where @xmath3 is the fixed point @xmath4 and @xmath5 is an integer lattice point in the square @xmath6 ^ 2 $ ] . </S><S> we show that the average of the pair correlation repartition @xmath7 over @xmath8 in a fixed disc @xmath9 converges as @xmath10 . </S><S> more precisely we prove , for every @xmath11 and each @xmath12 , the estimate @xmath13 we also prove that for each individual point @xmath14 , the @xmath15-level correlation @xmath16 diverges at any point @xmath17 as @xmath10 , and give an explicit lower bound for the rate of divergence . </S>

1101.3008

the first potentially real - life application of the emerging field of quantum information is arguably quantum key distribution ( qkd ) which allows two distant parties to establish a secret key over an _ a priori _ unsecure communication channel @xcite .the importance of a qkd protocol is usually measured through three main criteria : its _ practicality _ , its _ performance _ and its _ security_. among all existing protocols , those based on continuous variables appear quite appealing from a _practical _ point of view @xcite .for instance , the protocol gg02 @xcite simply requires the preparation of coherent states and their detection via homodyne ( or heterodyne ) detection .moreover , the _ security _ of this protocol is established against collective attacks @xcite which are optimal in the asymptotic limit @xcite . thereexist different ways to quantify the _ performance _ of a given protocol , but the usual figure of merit is the secret key rate that can be achieved as a function of the distance .continuous - variable protocols such as gg02 perform quite well for short distances , but seem unfortunately limited to much shorter distances than their discrete - variable counterparts ( see fig . 4 of ref .@xcite for a comparison of the performance of various protocols ) .in particular , while discrete - variable protocols allow one to distribute secret keys over distances larger than 100 km , gg02 has only be implemented for a distance of 25 km @xcite and there is not much hope of increasing its range well beyond 50 km @xcite . in this paper, we introduce continuous - variable ( cv ) qkd protocols which share the practicality and security of gg02 , but greatly improve its performance , especially in terms of achievable distance . in section [ philosophy ] , we quickly review the status of current cv protocols . in section [ new_schemes ], we present the specific modulation schemes of our protocols .we then introduce the concept of _ decoy states _ for cv qkd protocols in section [ section_decoy ] . in section [ security ] , we prove the security of our protocols against collective attacks and we finally discuss their performance in section [ performance ] .proving the security of a qkd protocol is usually a difficult task , but the situation is even worse for continuous - variable qkd protocols , because the relevant hilbert space is infinite dimensional .the task is difficult for ( at least ) one specific reason : the bipartite state @xmath0 shared by alice and bob ( in the entanglement - based version of the protocol ) has an infinite number of degrees of freedom .this means that a full tomography of this state is hopeless , even in the case of collective attacks , where alice and bob share many copies of the same state .fortunately , one can take advantage of extremality properties of gaussian states @xcite to show that the eavesdropper s information is upper - bounded by the information she would have if alice and bob shared instead the state @xmath1 , the gaussian state with the same first two moments as @xmath0 @xcite . hence, knowing the first two moments of the state @xmath0 , that is , a _ finite _ number of parameters , is sufficient to bound eve s information .furthermore , taking advantage of specific symmetries of the protocols in phase - space @xcite , it is possible to reduce this number to only three , namely the variances of alice and bob s reduced states and the covariance .that so few parameters are indeed sufficient to bound eve s information is quite remarkable , but it should be noted that this is also a necessity if one wants to take finite - size effects into account .indeed , it was shown in @xcite that estimating these parameters with a precision compatible with a secret key rate is already very challenging in terms of resources . because of the constraints imposed by finite - size effects , the only theoretical tool presently available to prove the security of a continuous - variable qkd protocol is therefore this gaussian optimality .this technique unfortunately comes at a price , namely that if the true quantum state @xmath0 is not sufficiently close to a gaussian state , then the bound on eve s information will not be tight enough to still get secret bits at the end of the protocol . by construction , this bound is indeed only tight for gaussian states , and it turns out that it degrades very rapidly as the non - gaussianity of the state ( and consequently of the protocol ) increases .this observation leads us to the unavoidable conclusion that with the theoretical techniques available today , protocols using a gaussian modulation ( _ i.e. _ gg02 , possibly with a heterodyne detection @xcite ) are optimal among all continuous - variable qkd protocols , and using any other modulation scheme can only lead to worse performances .a natural question then arises : why should one consider new protocols involving specific non - gaussian modulations since _ theoretically _ , they can not beat a gaussian modulation ?the reason is that , _ in practice _ , protocols with a gaussian modulation do not perform as well as expected , especially with regards to long distance communication .this is due to the fact that error - correction is very hard to implement for a gaussian modulation , thereby making the effective secret key rate drop to zero at distances of the order of 50 km , even if the theoretical key rate is strictly positive .the protocols we introduce in the present paper outperform previous protocols in terms of maximal range .this is achieved thanks to a combination of two facts : first , the specific modulation schemes allow one to extract information much more efficiently than with a gaussian modulation ; second , there is a regime ( for the variance of the modulation ) where the protocols are still significantly close to a gaussian protocol , hence making the bound on eve s information tight enough to be useful .the main argument for switching from a gaussian modulation to a non - gaussian one is _ not _ because it might make the eavesdropping more difficult ( available theoretical tools are not powerful enough to answer this question ) .the reason is that present coding techniques do not allow for a very efficient reconciliation procedure at ( very ) low signal - to - noise ratio ( snr ) when alice s data are gaussian .remember that a qkd protocol typically consists of three phases : first , alice and bob exchange quantum signal , perform measurements and obtain correlated classical data , say vectors * x * and * y * ; second , in the _ reconciliation _ step , they use classical error correction techniques to agree on a common ( errorless ) bit string * u * , and finally , they apply _ privacy amplification _ to obtain a secure key from * u*. the importance of the reconciliation phase is specific to continuous - variable protocols for two reasons .first , one has to deal with continuous values for _ both _ alice and bob s variables instead of bits , which is rather unusual in the field of digital communication .indeed , even when analog signals are used to transmit information on noisy ( classical ) channels , the modulation is almost always discrete and not continuous .the second reason is that contrary to discrete - variable qkd protocols where the error rate is always below some small constant ( to guarantee the security of the protocol ) , the snr can be arbitrarily small for continuous - variable protocols ( and is actually very small as one tries to increase the range of the protocol and the transmission of the channel is @xmath2 ( in fiber if @xmath3 is the channel length in kilometers ) , then the snr is roughly given by @xmath4 .the modulation variance on alice s side , @xmath5 is a free parameter of the protocol which can be optimized for a quantum channel , but numerical simulations show that the optimal value for @xmath5 practically does not depend on the transmission @xmath6 of the channel .for this reason , the snr is roughly proportional to the transmission of the channel , and is consequently very small for long distances . ] ) . for continuous - variable qkd protocols ,information is encoded in phase space , in general in the quadratures of coherent states .more precisely , if the classical information she wants to send is described by a vector @xmath7 , then alice prepares the @xmath8 coherent states @xmath9 and sends them through the quantum channel .there are two possibilities concerning the detection : bob can perform either homodyne or heterodyne measurements . in the case of a homodyne detection , for each state , bob chooses randomly which quadrature to measure .he then obtains an @xmath8-dimensional classical vector @xmath10 and later informs alice of his choices of quadratures . in the case of a heterodyne detection ,bob ends up with a @xmath11-dimensional vector @xmath10 .hence , alice and bob share twice as many data for the heterodyne protocol than for the homodyne one .however , a heterodyne detection adds 3 db of noise to the data , and in practice the performances of both schemes are quite similar . in order to complete the key distribution ,two additional steps are required .first alice and bob proceed with the reconciliation of their classical data in order to agree on a common bit string @xmath12 . here, we restrict ourselves to a _reconciliation @xcite , meaning that only bob can send classical information to alice ( in contrast to _ direct _ reconciliation where alice would send some side - information to help bob correct his errors ) .second , they use privacy amplification in order to obtain a secret key from @xmath12 : this can only be done once they have an upper bound on eve s information about @xmath12 . in this paper , we consider protocols for which the reconciliation can be performed efficiently ( which is not the case of the protocols using a gaussian modulation ) , and for which eve s information can be bounded if she is restricted to collective attacks .security against general attacks ( unconditional security " ) is then obtained in the asymptotic limit , thanks to a de finetti representation theorem for infinite - dimensional quantum systems @xcite . to be more specific ,the protocols considered here are characterized as the ones for which the _ reverse _ reconciliation problem can be reduced to the channel coding problem for the binary - input additive white gaussian noise ( bi - awgn ) channel , which can itself be tackled with standard techniques ( efficient error correcting codes ) .one such protocol is the four - state protocol considered in @xcite , but it turns out that other , more efficient , _ continuous _ modulation schemes are also possible .let us consider protocols involving a homodyne detection . in this case , the protocols of interest display a discrete modulation with either 2 or 4 states , and a basic problem is how to evaluate the transmission channel , and thereby eve s information .such a problem does not arise for gaussian modulations , because the variances and covariances directly measured by alice and bob give a covariance matrix , which is all that is needed to characterize the worst possible attack by eve , which is a gaussian attack according to the gaussian optimality theorem . on the other hand , for a non - gaussian modulation , the noises and correlations measured by alice and bob can not be directly connected to a relevant covariance matrix , and the gaussian optimality theorem does not apply directly . among possible approaches ,two - state protocols were studied in @xcite , and the authors proved the security of the protocols against any collective attacks , but with the _ caveat _ that a complete tomography of the state was required . in @xcite ,the authors considered the noises and correlations measured by alice and bob from their non - gaussian modulation , and then considered the gaussian attacks associated with these values as optimal .however , this approach has the implicit assumption that the transmission channel can be considered as _ linear _ , which means that it is intrinsically characterized by a transmission @xmath6 and an excess noise @xmath13 ( see details in appendix [ linear ] ) .gaussian channels are obvious examples of linear channels , but a linear channel may also be non - gaussian . since the proofs of @xcite only involve estimating the second moments of alice and bob s correlations , they are compatible with a practical implementation taking into account finite - size effects @xcite , but they are not fully general . in the present paper , we extend the security proof of discrete modulation protocols against arbitrary collective attacks .our proof still only requires us to estimate two quantities : the transmission @xmath6 and excess noise @xmath13 of the quantum channel , without carrying out a full channel tomography . but achieving this estimation requires the use of decoy states , as will be explained in section [ section_decoy ] .since the four - state protocol always outperforms the two - state protocol , we will only discuss the case of the former in the present paper . in this protocol , the modulation consists of four coherent states : @xmath14 where @xmath15 is a positive number .the modulation variance @xmath5 is given by @xmath16 .the practical implementation of the reverse reconciliation problem for the four - state protocol is discussed in details in appendix [ reduction ] . before introducing the modulation schemes compatible with a heterodyne detection , let us say a few words concerning the reconciliation procedure .the main difficulty here lies in the fact that we need a reverse reconciliation .indeed , the side information sent by bob must help alice without giving eve any relevant information .the only schemes where side information seems to have these properties ( while being efficiently computable by bob ) are when it describes rotations in particular dimensions , namely dimensions 1 , 2 , 4 and 8 @xcite . this surprising result is a consequence of the fact that the only real division algebras are the real numbers ( @xmath17 ) , the complex numbers ( @xmath18 ) , the quaternions ( @xmath19 ) and the octonions ( @xmath20 ) . indeed, one can show that the possibility of an efficient reconciliation protocol ( in terms of computation complexity as well as classical communication ) is intimately connected with the existence of a division operation for the data considered .in particular , identifying vectors in @xmath17 , @xmath21 , @xmath22 or @xmath23 with units of the real numbers , the complex numbers , the quaternions or the octonions allows one to take benefit of the division structure naturally associated with these ensembles .for instance , the reconciliation protocol of the four - state protocol exploits this property in dimension 1 .the modulation schemes we consider now exploit it for dimensions 2 , 4 and 8 .first , note that in the case of the four - state protocol , alice chooses the value of each quadrature uniformly on the ( 0-dimensional ) sphere @xmath24 in dimension 1 .( this value is then appropriately rescaled to obtain the desired variance of modulation . ) for this reason , we will sometimes refer to the four - state protocol as the 1-dimensional protocol in the rest of the paper .the modulation schemes we consider now are simply the generalizations to dimensions 2 , 4 and 8 . hence , in the @xmath3-dimensional protocol( with @xmath25 ) , alice draws random variables uniformly on the sphere @xmath26 in dimension @xmath3 . for @xmath27 , alice draws @xmath8 points on the unit circle : @xmath28 , and sends the @xmath8 coherent states @xmath29 , where the variables are rescaled by a factor @xmath15 . for @xmath30, alice draws @xmath31 points on the sphere@xmath32 : @xmath33 , and sends the @xmath8 coherent states @xmath34 , where the variables are rescaled by a factor @xmath35 . for @xmath36 , alice draws @xmath37 points on the sphere @xmath38 : @xmath39 , @xmath40 , and sends the @xmath8 coherent states @xmath41 , where the variables are rescaled by a factor @xmath42 .the procedure to reduce the reverse reconciliation problem in these three scenarios to the usual problem of channel coding for the bi - awgn channel is explained in detail in appendix [ reduction ] .let us say a few words about what we mean by _ efficient _ reconciliation procedure in the context of qkd .usually in the field of computer science , an algorithm is said to be efficient if the amount of resources ( _ e.g. _ running time , randomness generation , classical communication , etc ) it consumes grow at most polynomially with the natural size of the problem . in the case of a reconciliation procedure ,the problem size is given by the length @xmath8 of the vectors alice and bob try to agree on .as @xmath8 usually takes very large values ( for instance @xmath43 or @xmath44 ) , an algorithm requiring resources scaling as @xmath45 or @xmath46 is obviously unacceptable : only a linear scaling is compatible with a practical implementation .moreover , the factor of proportionality should be small enough . for @xmath47 or 8 , the reconciliation procedure introduced in @xciterequires that bob draws one random bit and transmits classically one real number to alice per exchanged signal .the computational complexity of the protocol ( not including the decoding of the error correcting code ) is linear with @xmath8 . for values of @xmath3strictly greater than 8 , the naive approach ( which consists in drawing random transformations uniformly in the orthogonal group @xmath48 and transmitting them to alice ) requires that bob draw @xmath3 random variables from a normal distribution ( instead of only 1 bit ) and to send @xmath3 real values to alice for each exchanged quantum signal , which is prohibitive for a realistic implementation .it should be emphasized that the higher the dimension , the higher the secret key rate of the qkd protocol .the reason for this is the very specific technique that we use to bound eve s information .our bound depends only on the covariance matrix of alice and bob s bipartite state in the entanglement - based version of the protocol , and therefore is tight only when the state is gaussian .fortunately , if the state is almost gaussian , then the bound is good enough for our purpose .because of this , we want to use a protocol as gaussian as possible .it turns out that considering modulations in higher and higher dimensions brings us closer and closer to the gaussian modulation for which the bound is tight .indeed , a gaussian modulation of variance 1 can be seen as drawing uniformly a random point of the sphere of radius @xmath49 in @xmath50 as @xmath3 tends to infinity .hence , the gg02 protocol with a gaussian modulation can be seen as the @xmath3-dimensional protocol with @xmath51 .unfortunately , for @xmath51 , efficient reconciliation techniques at low snr are not known .as we already mentioned , it is crucial that the security proof of cv qkd protocols requires the estimation of only a few parameters .ideally , one would like a gg02-type security proof where only the transmission @xmath6 and the excess noise @xmath13 of the quantum channel need to be estimated . as we will discuss in section [ security ] , the security proof of our protocols indeed relies on the fact that one can estimate the covariance matrix of alice and bob s bipartite state in the entanglement - based version of the protocol .a difficulty lies in the fact that alice and bob do not perform the entanglement - based version of the protocol ( in which case the covariance matrix is directly accessible in the experiment ) but use instead the _ prepare and measure _ version .hence , if alice encodes the variable @xmath52 in the quadrature of a state and bob obtains the result @xmath53 when measuring this quadrature , they can estimate the three following moments of order 2 : alice s variance @xmath54 , bob s variance @xmath55 and the covariance @xmath56 . whereas alice and bob s variance in the prepare and measure scenario are directly related to the respective variances in the entanglement - based scenario ,the same is not true for the covariances .there are two cases where the covariance matrix in the prepare and measure protocol allows one to recover the covariance matrix of the state in the entanglement - based scenario , namely when the quantum channel is linear ( see appendix [ linear ] ) , for instance a gaussian channel , and when the modulation is gaussian . in refs .@xcite , the security of the protocols considered in the present paper was established in the case of linear channels . here, we wish to get rid of this hypothesis ( which can never be perfectly verified in practice with a finite number of samples ) , and for this reason , it is necessary to use a gaussian modulation for the parameter estimation procedure .unfortunately , it is not _ a priori _ possible to use two different modulations for key distribution and parameter estimation , because an eavesdropper could use a different strategy in each case .the solution is to add a _ third _ modulation consisting of decoy states .let us call `` key '' , `` decoy '' and `` g '' the modulations corresponding respectively to states used for the key distillation , decoy states and states used for parameter estimation purposes ( a gaussian , in fact thermal , distribution ) .one can define the three following states : @xmath57 where @xmath58 for the @xmath3-dimensional protocol .( in particular , @xmath59 refers here to @xmath60 coherent states . ) in these expressions , the probability distribution @xmath61 is the uniform measure of the sphere @xmath26 ( with radius @xmath62 ) and @xmath63 is the gaussian distribution @xmath64 in @xmath3 dimensions . in other words, @xmath61 corresponds to the modulation schemes described in section [ new_schemes ] , and @xmath63 is the gaussian distribution of the gg02 protocol . if the probability distribution @xmath65 is chosen such that @xmath66 where @xmath67 is a weight between 0 and 1 , then the state sent by alice to use for parameter estimation is indistinguishable from that used to distill a key ( or as a decoy ) .the idea is that after the exchange of quantum states is complete , alice can announce to bob which states he can use for the key , which states he can discard ( decoys ) and which states should be used for parameter estimation . if in principle , the form of @xmath68 does not matter ( it could be any state , not necessarily of the form [ decoy_def ] ) , this is no longer true if we require the protocol to be practical .indeed , for this reason , we impose the extra constraint that @xmath68 should be obtained as a mixture of coherent states , _i.e. _ be of the form [ decoy_def ] .we discuss this in details in appendix [ decoys_coherent ] where we describe two techniques for finding the appropriate decoy states .to summarize , the modulation that is used in the protocols is a mixture of three particular modulations .let us note @xmath69 the fraction of states that alice and bob want to use for parameter estimation purposes ( this fraction can be optimized numerically but in a typical scenario , its value can be around 50 @xmath70 ) .then for each state she sends , alice will choose either the modulation @xmath71 with probability @xmath72 or modulation @xmath73 with probability @xmath69 or send a decoy state with probability @xmath74 .in this paper , we restrict ourselves to the case of collective attacks since they are optimal in the asymptotic limit @xcite .the ( asymptotic ) secret key rate @xmath75 is then given by @xcite : @xmath76 where @xmath77 is the reconciliation efficiency , @xmath78 is the classical mutual information between alice and bob s data ( for the data corresponding to the modulation @xmath71 ) and @xmath79 is the holevo quantity : @xmath80 where @xmath81 is the von neumann entropy , @xmath53 is bob s measurement result obtained with probability @xmath82 , @xmath83 is the corresponding state of eve s ancilla and @xmath84 is eve s partial state . note that this rate should be modified to take finite - size effects into account .for simplicity , we only consider the asymptotic rate here , but a complete analysis of finite - size effects can be found in ref .@xcite .since the quantity @xmath85 is directly observable in practice , the goal of the security proof consists in deriving an upper bound for the quantity @xmath79 which should be a function of parameters accessible in an experiment . in our case , we will find a bound which only depends on three parameters : the variance of modulation @xmath5 which is chosen by alice , the transmission @xmath6 and the excess noise @xmath13 of the quantum channel which can be estimated with the technique described in ref .@xcite .we now consider the entanglement - based version of our protocols . in this scenario, alice prepares @xmath8 two - mode squeezed vacuum states , keeps one half of each state and sends the second half to bob through the quantum channel .let us introduce some notations . in the following, we will consider bipartite states , either before or after the quantum channel .we use the superscript 0 to denote states before the quantum channel .moreover , the action of the quantum channel can be described by a map @xmath86 where the identity acts on the first part of the state ( namely , alice s state ) and the quantum channel @xmath87 acts non - trivially only on the second part of the state .the three states of interest are @xmath88 , @xmath89 and @xmath90 , which are the schmidt purifications of the states @xmath91 , @xmath92 and @xmath68 , respectively .( note for instance that @xmath88 is a two - mode squeezed vacuum : @xmath93 . ) after the quantum channel , these three states become respectively @xmath94 , @xmath95 and @xmath96 .the main idea of the security proof is that one can bound eve s information by a function of the covariance matrix of the state used to distill the key : @xmath97 . in the protocol , alice always starts with a two - mode squeezed state @xmath88 but she can choose between two measurement strategies depending on whether a given state will be used for key distillation or for parameter estimation .we also introduce a general measurement acting on alice s part @xmath98 such that , when applied to the two - mode squeezed vacuum @xmath88 , the result corresponding to @xmath99 prepares the state @xmath89 used in the @xmath3-dimensional protocol while the second result prepares @xmath90 ( see appendix [ measurement ] for a description of this measurement ) . for each state, alice chooses randomly between key distillation and parameter estimation .the fraction of each task should be optimized taking into account all finite - size effects .if a state is dedicated to parameter estimation , alice simply performs a heterodyne detection on her part and bob proceeds as usual . at the end of the protocol, alice and bob can compare their statistics and compute the covariance matrix @xmath100 of the state @xmath94 .for this , we do not need to make any assumption ( for instance of linearity ) concerning the quantum channel .if a state is to be used for key distillation , then alice performs the generalized measurement @xmath98 on her half of the state , thus preparing @xmath89 with probability @xmath67 and @xmath90 with probability @xmath101 .states corresponding to the result @xmath102 will later be discarded .finally , only the states @xmath89 are used for key distillation .let us note @xmath103 ( @xmath104 ) the covariance matrix of these states before ( after ) the quantum channel .it was proven in @xcite that the quantity @xmath79 can always be upper bounded by a function of the covariance matrix of alice and bob s bipartite state . here , the state used for key distillation is @xmath95 , meaning that one can bound @xmath79 with a ( known ) function of @xmath105 ( see ref .@xcite for the precise form of this function ) .all that is left to do is therefore to compute the covariance matrix @xmath105 .note that the covariance matrix @xmath106 of the state before the quantum channel can be computed and only depends on the modulation variance ( as well as the dimension of the modulation scheme ) .details on how to compute this covariance matrix are given in appendix [ covariance ] .the covariance matrix @xmath106 has the following form ( with the convention @xmath107 ) : @xmath108 where @xmath109 .here @xmath110 is alice s modulation variance in the prepare and measure version of the protocol and @xmath111 is a function of @xmath5 and the dimension @xmath3 of the protocol .the two - mode squeezed vacuum , corresponding to a modulation on a sphere whose dimension tends to infinity , has the same form with @xmath112 . a comparison of @xmath113 ( four - state protocol ), @xmath114 with the maximal value @xmath115 is displayed on figure [ z ] .( color online ) comparison of the covariance coefficient @xmath116 for the states @xmath117 ( top solid black curve ) and @xmath118 for the four - state protocol ( bottom short - dashed blue curve ) and the 8-dimensional protocol ( middle long - dash red curve ) as a function of the variance of modulation @xmath5 . ]one also knows the covariance matrix @xmath100 of the state @xmath94 _ after _ the quantum channel : @xmath119 where @xmath6 and @xmath13 refer respectively to the transmission and excess noise of the channel and can be estimated experimentally @xcite .the last part of the argument consists in proving that the covariance matrix @xmath105 has the same form ( or at least can be safely considered to have the same form ) as @xmath100 after the quantum channel , if one simply replaces @xmath115 by @xmath111 : @xmath120 this is clear if the quantum channel @xmath87 is linear , for instance gaussian but the argument is more involved in the case of an arbitrary quantum channel .in fact , if the channel is linear , alice and bob can directly compute the covariance matrix @xmath105 from the data corresponding to the modulation @xmath71 and the gaussian modulation or the decoy states are not required in that case @xcite . because the modulation considered here is indistinguishable from a gaussian modulation ,the parameter estimation is performed in such a way that there are no privileged direction in phase space . however , this is formally not enough to warrant that eve s attack has the same symmetry .we therefore provide the following two strategies : either one assumes that the symmetry is not broken by eve ( which in theory could be checked by performing a tomography of the state ) , or one does not want to make such an assumption and prefers to actively symmetrize the protocol . in the first scenario , under the assumptionthat the symmetry of the quantum channel is not broken , the security protocol with decoy states presented here is at least as good as in the case where the channel is linear . indeedthe quantum state shared by alice and bob is invariant under the group of conjugate passive symplectic operations applied on alice s @xmath8 modes and bob s @xmath8 modes , which means that their state can be safely considered to be gaussian if the analysis is restricted to collective attacks @xcite .however , if one does not want to rely of the assumption that the symmetry is not broken , it is possible to remove this assumption thanks to an active symmetrization of the protocol .this is described in detail in appendixes [ active ] and [ symmetrized_protocol ] .this additional step shows that the state @xmath121 after the quantum channel is rotationally invariant in phase space .therefore , when restricting ourselves to collective attacks , we conclude using the technique presented in ref .@xcite that the state @xmath121 can be safely considered to be gaussian . in particular , the covariance matrix given in eq .[ gamma_key ] can be used for the security analysis , with the same values of @xmath6 and @xmath13 as the ones in @xmath100 , obtained from the parameter estimation step .finally , using the covariance matrix @xmath105 , one can compute the quantity @xmath79 using for instance the formalism detailed in @xcite .the ( asymptotic ) secret key rate of the protocols reads : @xmath122 the idea in our protocols is to use a modulation scheme which is compatible with a very efficient reconciliation , thereby greatly increasing the quantity @xmath85 in comparison with gaussian modulation protocols .the price to pay is that this non gaussianity makes our bound on @xmath79 less tight .this is because the correlation @xmath111 of the state @xmath118 is strictly less than @xmath115 for a given variance of modulation .interestingly , this discrepancy can be interpreted in terms of excess noise : the fact that @xmath118 displays smaller correlations than the two - mode squeezed state has the same effect as some virtual excess noise .in particular , the value of @xmath79 one obtains in the @xmath3-dimensional protocol corresponds to the value one would obtain for a gaussian modulation protocol ( gg02 ) with a quantum channel characterized by a transmission @xmath123 , and an excess noise @xmath124 , where @xmath125 . since one has @xmath126 for reasonable values of @xmath5 ( see figure [ z ] ) , the main effect of the non - gaussian modulation is the _ equivalent excess noise _ @xmath127 .figure [ deltaxi ] displays this equivalent excess noise in the case of the four - state protocol ( @xmath128 ) and the 8-dimensional protocol .( color online ) equivalent excess noise @xmath129 due to the non - gaussian modulation as a function of the variance of modulation @xmath5 .upper curve refers to the four - state protocol , lower curve to the 8-dimensional protocol . by definition , a protocol with a gaussian modulationdoes not display any equivalent excess noise .an excess noise of one unit of shot noise corresponds to an entanglement - breaking channel , therefore no security is possible with such a level of noise .this figure clearly shows that the 8-dimensional protocol outperforms significantly the four - state protocol . ] in state - of - the - art implementations @xcite , the excess noise is typically less than a few percent of the shot noise .this gives an approximate limit for the value of the equivalent excess noise that is acceptable .in particular , for the four - state protocol , one needs to work with modulation variances below 0.5 units of shot noise . on the contrary, it becomes possible to work with much higher variances in the case of 8-dimensional protocol . in fig .[ k4octo_asympt ] , we display the asymptotic secret key rate of the four - state and the 8-dimensional protocols .the various parameters are chosen conservatively : a quantum efficiency of @xmath130 , a reconciliation efficiency of @xmath131 and an excess noise of 0.005 or 0.01 units of shot noise .the superiority of the 8-dimensional protocol is quite clear : the secret key rate is higher by nearly an order of magnitude , and one can work with significantly larger modulation variances ( the optimized variances are @xmath132 for the four - state protocol and @xmath133 for the 8-dimensional protocol ) .( color online ) asymptotic secret key rate @xmath75 for the 8-dimensional protocol ( solid lines ) and the four - state protocol ( dashed lines ) as a function of the distance ( assuming transmission through a standard telecommunications fiber with 0.2 db of loss per kilometer ) .the various parameters are an excess noise of 0.005 ( upper red lines ) or 0.01 ( lower blue lines ) and a quantum efficiency of the detectors @xmath134 of @xmath130 .reconciliation efficiency is supposed to be a conservative @xmath131 . ] to confirm the robustness of the 8-dimensional protocol , we display in fig .[ koctofinite ] the secret key rate when finite - size effects are taken into account .the secret key rate is computed against collective attacks , as detailed in ref . @xcite . among various finite - size effects@xcite , the most crucial ones for continuous - variable protocols are clearly the imperfect reconciliation efficiency ( which prevents the protocol with a gaussian modulation to achieve key distribution over large distances ) and the parameter estimation .while the reconciliation efficiency is taken care of by the 8-dimensional continuous modulation , the parameter estimation is quite sensitive for continuous - variable protocols .in fact , the real problem lies in the estimation of the excess noise @xmath13 , which is very small compared to the shot noise , and thus hard to evaluate accurately .( color online ) non - asymptotic secret key rate @xmath75 for the 8-dimensional protocol , obtained for realistic values : excess noise @xmath135 , security parameter @xmath136 , quantum efficiency of the detectors @xmath137 , reconciliation efficiency @xmath131 for the bi - awgn channel and transmission through a telecommunications fiber with 0.2 db of loss per kilometer .half the samples are used for parameter estimation . from left to right , the block length is equal to @xmath138 and @xmath139 . ] in fig .[ koctofinite ] , all such finite - size effects are taken into account @xcite .the results are rather pessimistic , but remember that this is also true for all discrete - variable protocols @xcite , and the protocols presented here perform reasonably well in comparison . while exchanging @xmath139 quantum signalsis rather unrealistic , exchanging @xmath43 or even @xmath140 signals is not completely out of reach of today s technology .hence , our protocols allow the distribution of secret keys over distances of the order of 100 km , taking into account all finite - size effects .in conclusion , we introduced continuous - variable qkd protocols with non - gaussian modulations .we established the security of these protocols against arbitrary collective attacks , which implies their unconditional security in the asymptotic limit . the four - state ( the 8-dimensional ) protocol appears optimal _ from a practical point of view _ among all protocols using a homodyne ( heterodyne ) detection in the sense that it allows an efficient reconciliation while remaining as close as possible to the theoretically optimal gaussian protocols @xcite .the main open questions concern the status of the decoy states .first , it should be possible to prove the security of the protocols considered here without requiring any decoy states .this might come at the price of slightly degraded bounds ( to take into account possible non gaussian attacks ) .second , without removing the decoy states , an important question is how well they should approximate the gaussian distribution .put otherwise , how indistinguishable should the distribution corresponding to key distillation be from the one used for parameter estimation ?in particular , how should this distinguishability be taken into account in the overall security parameter of the protocol ? finally , the most outstanding problem that remains for continuous - variable qkd protocol concerns general attacks .this question was partially answered with the derivation of a de finetti - type theorem for quantum systems of infinite dimension @xcite. however , the bounds obtained there are not good enough to be used in practice .hence , it seems crucial to see if the post - selection technique introduced in @xcite can be adapted to continuous variables , since this technique is already known to provide much better ( almost tight ) bounds than the de finetti theorem in the case of discrete variables @xcite .the authors acknowledge fruitful discussions with frdric grosshans , norbert ltkenhaus and renato renner .this work was carried out in the framework of the anr project sequre ( anr-07-sesu-011 - 01 ) .al received financial support from the eu erc starting grant percent . 32 natexlab#1#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 , , , , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , , , , , , , , , * * , ( ) . , , , , , , * * ( ) . , , , , , * * , ( ) . , , , * * , ( ) . , , , ,* * , ( ) . , , , ( ) ., , , , , , * * , ( ) . , , , , , , * * , ( ) . , * * , ( ) . , , , , * * , ( ) . ,( ) . , in _ _( ) , vol . , pp . . , , , , , * * , ( ) . , * * , ( ) . , * * ( ) . , , , * * , ( ) . , , , * * , ( ) ., , , * * , ( ) , issn . , * * , ( ) . , , , * * , ( ) . , pp . ( ) . , ph.d .thesis , ( ) , http://tel.archives-ouvertes.fr/tel-00451021 . , * * , ( ) . ,we shall define a linear quantum channel by the input - output relations of the quadrature operators in heisenberg representation : @xmath141 where the added noises @xmath142 , @xmath143 are uncorrelated with the input quadratures @xmath144 , @xmath145 .such relations have been extensively used for instance in the context of quantum non - demolition ( qnd ) measurements of continuous variables @xcite , and they are closely related to the linearized approximation commonly used in quantum optics .gaussian channels ( channels that preserve the gaussianity of the states ) are usual examples of linear quantum channels. however , linear quantum channels may also be non - gaussian , this will be the case for instance if the added noises @xmath142 , @xmath143 are non - gaussian . for our purpose ,the main advantage of a linear quantum channel is that it will be characterized by transmission coefficients @xmath146 , @xmath147 , and by the variances of the added noises @xmath142 and @xmath143 .these quantities can be determined even if the modulation used by alice is non - gaussian , with the same measured values as when the modulation is gaussian ( because these values are intrinsic properties of the channel ) .the relevant covariance matrix can then be easily determined , and eve s information can be bounded by using the gaussian optimality theorem .this justifies the approach taken in refs .@xcite , but unfortunately this is not fully general , contrary to the proof of the present paper .the goal of this section is to explain how an efficient reconciliation can be achieved for the various modulation schemes considered in this article , for arbitrarily low snr .reconciliation in a qkd protocol is very similar to the problem of channel coding ( that is , transmitting information efficiently and reliably on a noisy communication channel ) with the additional constraint that the input of the channel is not chosen by alice but instead randomly picked from a given probability distribution corresponding to the modulation scheme . in particular ,the usual task is the following : alice and bob are given two @xmath8-dimensional real vectors @xmath148 and @xmath10 and their goal is to agree on a common bit string @xmath149 .a supplementary constraint when dealing with reverse reconciliation is that @xmath149 should be a function ( possibly randomized ) of @xmath10 , and that all public communication should be from bob to alice .there exists a standard technique for reducing the problem of reconciliation to the one of channel coding , namely coset coding introduced by wyner @xcite .the idea is that bob will use an additional public ( but authenticated ) channel to describe a function @xmath150 such @xmath151 .alice can then apply this function to her vector and obtain @xmath152 . in the case of coset coding ,the description of @xmath150 is simply a translation corresponding to the syndrome of @xmath53 for a linear error correcting code @xmath153 .this gives rise to a virtual communication channel with input @xmath149 and output @xmath154 for which one can apply standard channel coding techniques .one case where reconciliation can be performed very efficiently occurs when the virtual channel is a binary - input additive white gaussian noise ( bi - awgn ) channel meaning that the coordinates of @xmath149 and @xmath154 are related through @xmath155 where @xmath156 and @xmath157 is a centered normal random variable . for experimental realizations of continuous - variable qkd, the quantum channel always behaves in very good approximation like a gaussian channel and the bi - awgn channel is therefore the model of interest here . in this case , it is possible to show that the existence of an efficiently computable function @xmath150 is possible only for very specific modulation schemes , namely the cases where @xmath3-uplets of @xmath158 are distributed uniformly on the unit sphere in dimension 1 , 2 , 4 or 8 @xcite .the case @xmath128 corresponds to a binary modulation , that is , the four - state protocol ( which indeed displays a binary modulation for _ each _ quadrature ) ; the case @xmath36 corresponds to the 8-dimensional modulation .let us therefore consider @xmath3-uplets @xmath159 and @xmath160 , with @xmath161 and @xmath162 with @xmath163 .without loss of generality , we restrict our attention to the case where the transmission is 1 . for dimensions 1 , 2 , 4 and 8, the unit sphere @xmath26 has a division algebra .in particular , a @xmath3-dimensional vector can be identified with an element of @xmath164 , that is , a real number ( @xmath128 ) , a complex number ( @xmath27 ) , a quaternion ( @xmath30 ) or an octonion ( @xmath36 ) .therefore , both multiplication and division are well defined in this context .bob chooses a random element @xmath165 with the uniform distribution on the @xmath3-dimensional hypercube and sends the variable @xmath166 to alice ( through the classical channel ) .alice computes @xmath167 which is possible because @xmath26 is a division algebra .we now wish to prove that the channel @xmath168 is a bi - awgn channel .let us characterize the noise @xmath169 on this virtual channel : @xmath170 since @xmath171 and @xmath172 are simply rotations on @xmath26 , one concludes that @xmath173 , which proves that the virtual channel @xmath174 is indeed a bi - awgn channel , for which efficient error correcting codes are available . if @xmath175 , alice and bob simply divide their data into @xmath176 @xmath3-uplets and proceed as described above .all is left to do is to use coset coding to finish the reconciliation : bob sends the syndrome of @xmath177 for a linear code alice and he agreed on beforehand .alice simply decodes her word @xmath178 in the coset code defined by the syndrome .this can be done very efficiently with capacity approaching codes such as low - density parity - check ( ldpc ) codes @xcite .if the snr is very low , then one can work with a concatenation of a low rate ldpc code ( such as a multi - edge ldpc code for instance @xcite ) with a repetition code .this simple technique allows one to obtain good error correcting codes of arbitrarily low rate ( see chapter 5.2 of ref .@xcite for more details concerning this concatenation technique ) .in the entanglement - based version of the protocol , one has to apply the generalized measurements @xmath98 . in the prepare and measure scenario , it is therefore necessary for alice to send states which are compatible with these measurements .the states corresponding to the operator @xmath99 are not a problem , since by construction , they correspond to the modulation used to distill the key , that is , coherent states drawn uniformly on the sphere in @xmath3 dimensions .the states corresponding to the operator @xmath102 might be a little bit more problematic in the sense that they are not usually easy to produce experimentally .ideally , one would like to be able to produce these states simply by modulating coherent states ( which are the only states simple enough to allow for a realistic qkd protocol ) .in particular , if alice applies the measurements @xmath98 , then one obtains the following relation : @xmath179 with @xmath180 . however , in this case , the state @xmath68 does not have a positive @xmath181-function , meaning that it can not be obtained as a mixture of coherent states .we now present two different techniques to deal with this problem : either one replaces @xmath92 with a noisy version ( see appendix [ noisy ] ) or one relaxes equation [ def_decoy ] and considers instead an approximate version of the decoy states ( see appendix [ optimization ] ) .let us consider the prepare and measure version of the @xmath3-dimensional protocol . in this case, the gaussian modulation can be seen as sending @xmath60 coherent states @xmath182 such that the random variables @xmath183 are independent and identically distributed centered normal variables . without loss of generality , we consider variables with variance @xmath184 .taking advantage of the rotational invariance of the gaussian distribution , one can equivalently choose the random vector @xmath185 by first picking uniformly a random point of the unit sphere @xmath26 in @xmath3 dimensions , and drawing the radius @xmath186 of the vector @xmath187 from a chi distribution with @xmath3 degrees of freedom .in particular , the probability density function of @xmath188 is : @xmath189 the probability distributions corresponding to @xmath190 and @xmath191 are displayed on figure [ chi_distrib ] .in particular , it should be noted that they become more and more peaked as the dimension @xmath3 increases . in this picture , the probability distribution corresponding to the key , that is @xmath192 , is by construction a dirac distribution centered in 1 .( color online ) probability density functions for the radius of @xmath187 in the @xmath3-dimensional protocol for a gaussian modulation . from least peaked to most peaked , @xmath190 and @xmath191 . ]the first approach we investigate aims at satisfying equation [ def_decoy ] exactly while allowing for a positive ( and non - negligible ) probability @xmath67 of sending a signal state .this is done by considering slightly noisy versions of the true signal modulation. in particular , one chooses two parameters @xmath193 $ ] and @xmath194 and defines the states used for key distillation as the ones with a radius bounded by these two parameters : @xmath195 .the decoy states then simply correspond to the remaining states .provided that @xmath196 and @xmath197 are close enough to 1 , the penalty imposed by this noise , compared to the ideal case where the key modulation is strictly equal to 1 , is negligible in terms of reconciliation efficiency . on the other hand , one should not choose values too close to 1 , otherwise the probability @xmath67 that a given state can be used for key distillation will be very small : @xmath198 hence , optimizing the values of @xmath196 and @xmath197 should be seen as a tradeoff between the penalty imposed on reconciliation efficiency and the probability that a given state can be used for key distillation .note also that with this approach , the state @xmath192 is not longer described as in appendix [ covariance ] .in particular , the covariance matrix of the new state is a little bit different from the one presented in appendix [ covariance ] .we do not give an explicit derivation of the new covariance matrix here , but we point out that because the new state used for the key is actually closer to a gaussian modulation , the holevo information between bob and eve can still be safely bounded using the covariance matrix given in eq . [ gamma_key ] .we now give a second approach to the problem of approximating decoy states with coherent states .our goal is still to achieve the following equality : @xmath199 but this time , without considering a noisy version of @xmath200 .if one chooses @xmath180 as defined in appendix [ measurement ] , then the state @xmath68 does not have a positive @xmath181-function , meaning that it can not be obtained as a mixture of coherent states .fortunately , @xmath181-functions can be regularized rather well , and for our purpose , it is sufficient to find a state @xmath68 with a non negative @xmath181-function such that equation [ def_decoy2 ] only holds _approximately_. here , approximately should be understood in terms of the trace distance .more precisely , we are interested in finding a probability distribution ( hence non negative ) @xmath201 such that @xmath202 for a value of @xmath203 sufficiently small . if eq .[ approx ] holds and alice uses the modulation @xmath201 for the decoy states , then the probability that eve can distinguish the states used for key distillation from the ones used for parameter estimation is upper bounded by @xmath203 . here , we do not give a solution for the problem of finding the best distribution @xmath181 compatible with a success probability @xmath67 and the error @xmath203 , but we point out that the usual optimization tools ( for instance the optimization toolbox of matlab ) allow one to find very good instances of @xmath181 . for example , for the 2-dimensional protocol and @xmath204 , we could obtain a value of @xmath203 less than @xmath205 for @xmath206 with a distribution @xmath181 corresponding to a sum of 6 dirac distributions , that is a mixture of 6 particular coherent states .the natural question that arises here is how good should the approximation be ? is a trace distance equal to @xmath205 sufficient to guarantee a reasonable level of security ? or should one aim for a value of @xmath207 ?note that if the approximation is not perfect , it means that eve might have a ( very ) small probability to distinguish between the states used for the key distillation and those used for parameter estimation .however , discriminating between these two modulations does not appear to be a good solution for eve as this would induce a lot of phase noise in the signal : indeed , because all the modulations considered here are phase - invariant , the optimal discrimination procedure consists in projecting the states onto fock states , thereby erasing all the phase information .for this reason , a trace distance of @xmath205 between @xmath96 and its approximation is very likely to be sufficient for any practical implementation .we now describe the general measurement @xmath98 performed by alice to prepare the state @xmath118 from a two - mode squeezed vacuum .the state to which this measurement is applied is a ( possibly multimodal ) two - mode squeezed vacuum as described in table [ table_resource ] ..parameterization of the various protocols .parameter @xmath3 corresponds to the number of quadratures that should be processed together . [cols="^,^,^",options="header " , ] [ table_resource ] the operator @xmath208 is defined as @xmath209 with @xmath210 and @xmath211 the states @xmath212 are defined in appendix [ covariance ] and @xmath213 with @xmath214 and @xmath215 .when performing the general measurement @xmath216 , conditioned on the result corresponding to @xmath208 , the state @xmath217 is transformed into @xmath218 : @xmath219 let us consider the state @xmath220 with @xmath221 conditioned on the result @xmath208 , one obtains @xmath222 with @xmath223 the condition @xmath224 leads to @xmath225 with @xmath226 the probability of obtaining the result corresponding to @xmath208 , meaning successfully creating a state @xmath227 is @xmath228 and is displayed on figure [ proba_pi ] as a function of @xmath15 . for @xmath229 , one has : @xmath230 where @xmath231 and @xmath232 @xmath233 is a negative binomial distribution @xmath234 .let us define the operators @xmath235 and @xmath236 as @xmath237 where @xmath238 is given by : @xmath239 ensuring that @xmath99 is a genuine povm element .it is straightforward to check that the probability of success of the measurement is @xmath240 ( color online ) probability of success of the measurement @xmath99 as a function of @xmath15 . from bottom to top , @xmath190 and @xmath191 . ]in this appendix , we derive the covariance matrices of the states corresponding to the four - state protocol , which is optimal with a homodyne detection , and the 8-dimensional protocol which is optimal for a heterodyne detection .the covariance matrices corresponding to the other ( suboptimal ) choices of modulation can be found with a similar technique .let us note @xmath241 for @xmath242 the state used for the key distillation in each protocol , _i.e. _ @xmath243 and @xmath227 is the initial bipartite state for the four - state protocol whereas @xmath244 corresponds to the 8-dimensional protocol .the two - mode squeezed vacuum is noted as @xmath245 .[ covariance_matrix ] let us use the notation @xmath246 for @xmath247 and @xmath248 .in the four - state protocol , the state @xmath118 is a pure state @xmath227 defined as @xmath249 where latexmath:[\[\begin{aligned } for @xmath247 and @xmath251 let us note @xmath252 and @xmath253 the annihilation operators on the two modes .applying @xmath252 to @xmath254 gives : @xmath255 for @xmath256 and @xmath257 let us compute the covariance matrix @xmath258 of the bipartite state @xmath227 .it has the following form : @xmath259 one has : @xmath260 the correlation term @xmath113 of the covariance matrix is given by @xmath261 one has : @xmath262 where addition should be understood modulo 4 .finally , we obtain : @xmath263 it may be noticed that at the lowest order in @xmath15 , the states @xmath254 are simply the number states @xmath264 for @xmath265 , 1 , 2 , 3 , which are independent of @xmath15 .the states @xmath212 are four orthogonal linear combinations of these four number states , with coefficients of the form @xmath266 , where @xmath67 is an integer .in addition , the state @xmath267 is simply @xmath268 , which is also the lowest - order ( gaussian ) epr state .correspondingly , @xmath269 , in the limit where @xmath15 tends to 0 . since the entangled state is already gaussian in this regime , no decoy states are needed , and ref .@xcite can be used directly to establish the unconditional security of the protocol .unfortunately , this approach is restricted to values of @xmath15 which are too small to be useful in practice ; this is why the more powerful proof presented in the present paper is needed .the partial trace @xmath270 is defined by the modulation scheme : it is the uniform mixture of quadrimodal coherent states over a real 7-dimensional sphere : @xmath271 where @xmath272 and the sphere @xmath273 is defined as @xmath274 and @xmath275 is the haar measure on @xmath273 . because this state is a four - mode orthogonally invariant state ( by construction ), it can be written as @xcite : @xmath276 where @xmath277 in order to determine the @xmath278 , we compute the probability @xmath279 of finding @xmath280 photons in the four - mode state @xmath281 : @xmath282 since @xmath283 .because the coherent state @xmath284 , which refers to the vacuum , does not contain any photon , one has : @xmath285 we therefore get the expression : @xmath286 finally @xmath287 is defined as : @xmath288 where @xmath289let us compute the covariance matrix @xmath290 of @xmath244 .it has the form @xmath291 where @xmath292 where @xmath293 and @xmath294 refer to alice and bob s annihilation operators relative to the first mode .tracing @xmath295 over the last three modes gives @xmath296 : @xmath297 one immediately has : @xmath298 then , @xmath299 which gives @xmath300 .+ let us now compute @xmath301 .first , one notes that @xmath302 except if @xmath303 .some combinatorics shows that @xmath304 using the expression of @xmath244 , one obtains @xmath305 and finally @xmath306a quantum key distribution protocol is described as a map @xmath307 @xcite : @xmath308 where @xmath309 is the @xmath8-mode bipartite state shared by alice and bob at the end of the distribution phase ( in the entanglement - based protocol ) , @xmath310 and @xmath311 are respectively alice and bob s final keys and @xmath312 is a transcript of all classical communication as well as alice s and bob s raw data . a protocol @xmath307 is said to be invariant under some set of transformations @xmath313 if for any element @xmath314, there exists a cptp map @xmath315 such that : @xmath316 let us consider the uniform measure @xmath317 on @xmath313 .if the protocol @xmath307 is invariant under the set @xmath313 , then it is sufficient to prove the security of @xmath307 for states displaying the same symmetry .in particular , it is sufficient to consider states of the form : @xmath318 for any @xmath309 and where @xmath319 is the image of the mode state @xmath0 by @xmath320 .if we consider here for @xmath313 the group of conjugate passive symplectic operations applied on alice s @xmath8 modes and bob s @xmath8 modes ( in phase space , such operations are simply conjugate orthogonal transformations ) , then for a given operation @xmath320 applied to the state , the map @xmath315 is obtained by applying the orthogonal transformations corresponding to @xmath320 on the classical data measured by alice and bob . if the protocol is invariant under this whole group , then it is sufficient to look at gaussian states to prove security against collective attacks @xcite . in the prepare and measure version of the protocol , this group becomes the orthogonal group @xmath321 .one simple way to ensure that a protocol is indeed invariant under a set @xmath313 of transformations is for alice and bob to actively apply random transformations of @xmath313 to their states . in particular , if alice and bob both apply random orthogonal transformations to their classical vectors in the prepare and measure protocol , then the security analysis can be done assuming that they share a gaussian state in the entangled version of the protocol . for our security proof ,the goal of the symmetrization is to make sure that the state shared by alice and bob is as isotropic as possible .indeed , remember that we need to estimate the covariance matrix of the state shared by alice and bob after alice s generalized measurement @xmath98 .the only problem that could potentially happen would be that eve guesses which states might be used for key distillation and which ones might be used for parameter estimation and that she manages to somehow play with alice and bob s correlations in order to fool them into overestimating their correlations .to totally prevent such a ( quite unrealistic ) scenario , it is sufficient to symmetrize the state so that there are no privileged directions in phase space that eve could exploit .hence , in practice , the symmetrization does not require to apply random conjugate passive symplectic operations chosen with the uniform measure over the whole set of such operations : a smaller subset should be efficient . a very conservative quantitative criterium to evaluatethe quality of such a set would be for instance the distance between the partial trace of @xmath322 once we trace out @xmath323 modes and the gaussian state with the same first two moments . to make sure that the protocol is indeed invariant under specific transformations, we apply an active symmetrization step to the state @xmath0 before applying the protocol .the transformations we consider are conjugate passive symplectic operations applied on alice s @xmath8 modes and bob s @xmath8 modes , which therefore correspond to orthogonal transformations applied to their classical vectors in the prepare and measure protocol . for simplicity , we restrict the discussion to this prepare and measure scheme in the following .the active symmetrization requires us to choose a subset @xmath324 of the orthogonal group and for alice and bob , and to apply the same element @xmath325 ( chosen uniformly at random ) to their data before starting the postprocessing . as we stressed above , taking for @xmath324the whole orthogonal group is not necessary in practice .hence , we want @xmath324 to be a subset of the orthogonal group with the following properties : drawing a random element @xmath150 from the uniform measure on @xmath324 should be doable with resources ( time and alea generation ) scaling at most linearily in @xmath8 , the description of @xmath150 should also be at most linear in @xmath8 and applying @xmath150 ( or @xmath326 ) to a random vector of @xmath327 should also be at most linear in @xmath8 .these conditions ensure that the protocol with the active symmetrization remains practical .moreover , the symmetrization should work as well as possible , meaning that @xmath324 should symmetrize the state as much as possible .we give examples of such possible subsets @xmath324 in the next subsection .let us describe a recursive algorithm that allows one to draw an orthogonal transformation with the haar measure on @xmath328 .if we assume that we already drew a random transformation @xmath329 from the haar measure on @xmath330 , then let us note @xmath331 the orthogonal transformation in @xmath328 acting as the identity on the first element of the canonical basis of @xmath327 and as @xmath329 on the last @xmath323 elements of the basis .then , let us draw uniformly at random a unit vector @xmath332 on the sphere @xmath333 and define the following householder reflection @xmath334 : @xmath335 note that drawing @xmath332 can be done in linear time simply by drawing @xmath8 random normal variables and normalizing the obtained vector . also , applying @xmath334 to any vector can be done in linear time in @xmath8 .finally , one can show that random orthogonal transformation @xmath336 follows the haar distribution of @xmath328 @xcite .in particular , one has : @xmath337 where one defines @xmath338 .hence , drawing , describing and applying a random orthogonal transformation from @xmath328 are tasks with complexity quadratic in @xmath8 .we define the subset @xmath339 of the orthogonal group as corresponding to the set of compositions of @xmath280 such householder reflections , that is the last @xmath280 steps of the algorithm described above .for instance , @xmath340 corresponds to the set of householder reflections with respect to a hyperplane of @xmath341 , and @xmath342 is the orthogonal group @xmath328 .one can also define a family of measures @xmath343 on @xmath328 corresponding to the uniform measures of @xmath344 .a complete symmetrization would imply performing orthogonal transformations on both alice and bob s data chosen randomly with the measure @xmath345 , but for all practical purposes , it seems that @xmath346 already provides a high level of symmetrization .we present here two different schemes , depending on the choice of modulation which can be either fully gaussian or consists of a non - gaussian modulation supplemented by appropriate decoy states .the former modulation is compatible with both a homodyne or a heterodyne detection and corresponds to the technique detailed in appendix [ noisy ] while the latter , which is more efficient in terms of resources , is only compatible with a heterodyne detection ( and is detailed in appendix [ optimization ] ) .these schemes include the active symmetrization introduced in appendix [ active ] .* alice draws @xmath11 random variables @xmath347 from a centered normal distribution with the appropriate variance . these form a vector @xmath348 . *alice sends the states @xmath349 to bob , with @xmath350 .* bob receives the states after the quantum channel and measures them , with either a homodyne detection or a heterodyne detection . in the case of a heterodyne detection, he obtains a @xmath11-dimensional vector @xmath53 . in the case of a homodyne detection, he obtains an @xmath8-dimensional vector @xmath53 , then informs alice about his choices of measured quadratures ( @xmath52 or @xmath67 for each state ) ; alice only keeps the relevant coordinates in her data in order to form a new @xmath8-dimensional vector @xmath52 . *alice randomly draws a random orthogonal transformation @xmath351 from the orthogonal group @xmath321 ( or @xmath328 for a homodyne detection ) . in theory , to achieve a perfect symmetrization of the state , alice should draw @xmath351 with the haar measure on @xmath321 .however , in practice , @xmath351 can be chosen uniformly in a well - chosen subset of @xmath321 which has the advantage of allowing for efficient descriptions of its elements , such as one of the measures @xmath352 defined in appendix [ active ] . *alice describes @xmath351 to bob through the classical communication channel , and both parties apply @xmath351 to their respective vector , hence obtaining @xmath353 and @xmath354 . *alice chooses randomly @xmath355 coordinates that are used for parameter estimation . *the next step is where lies the novelty of our protocols .for instance , in the so - called four - state protocol , alice considers the coordinates @xmath356 which were not used for parameter estimation and keeps only the ones such that @xmath357 is sufficiently close to a predetermined value . in the case of the eight - dimensional protocol ( with heterodyne detection for instance ) , alice divides her data into blocks of size 8 and keeps the blocks for which the euclidean norm is close to a predetermined value ( see appendix [ noisy ] for details ) .alice informs bob of the indices that she keeps .the rest of the data are discarded . at this point ,alice and bob have classical correlations for which an efficient reconciliation protocol is available ( see appendix [ reduction ] ) .* alice draws @xmath8 8-dimensional random vectors , each chosen from one of the three following distributions : random vectors on the 7-dimensional sphere with the appropriate radius ( these data correspond to the non - gaussian modulation which will be used for the key distillation ) , random vectors on the 7-dimensional sphere with an appropriately fluctuating radius ( these are the decoy states which will be discarded at the end of the protocol : the mixture of these states with the previous ones should be indistinguishable from a true gaussian distribution ) or gaussian vectors which are used for parameter estimation .alice hence obtains an @xmath358-dimensional vector @xmath52 for which each subset of length 8 corresponds either to legitimate information , decoy data that will be discarded or data used for parameter estimation . *alice randomly draws a random orthogonal transformation @xmath351 from the orthogonal group @xmath359 . in theory , to achieve a perfect symmetrization of the state , alice should draw @xmath351 with the haar measure on @xmath359 .however , in practice , @xmath351 can be chosen uniformly in a well - chosen subset of @xmath321 which has the advantage of allowing for efficient descriptions of its elements , such as one of the measures @xmath352 defined in appendix [ active ] . *alice computes the vector @xmath360 , which is the image of @xmath52 by the orthogonal transformation @xmath351 and uses this vector for her modulation .hence she sends the states @xmath361 to bob , with @xmath362 .* bob receives the states after the quantum channel and measures them , with a heterodyne detection .he obtains an @xmath358-dimensional vector @xmath363 .* alice describes @xmath351 to bob through the classical communication channel .bob applies @xmath364 to his vector @xmath363 and obtains @xmath365 .* alice reveals which subsets of length eight should be kept for the key distillation , which ones should be discarded ( as they correspond to decoy states ) and which ones should be used for parameter estimation .* at this point , alice and bob have classical correlations for which an efficient reconciliation protocol is available ( see appendix [ reduction ] ) .

<S> in this paper , we consider continuous - variable quantum key distribution ( qkd ) protocols which use non - gaussian modulations . </S><S> these specific modulation schemes are compatible with very efficient error - correction procedures , hence allowing the protocols to outperform previous protocols in terms of achievable range . in their simplest implementation </S><S> , these protocols are secure for any linear quantum channels ( hence against gaussian attacks ) . </S><S> we also show how the use of decoy states makes the protocols secure against arbitrary collective attacks , which implies their unconditional security in the asymptotic limit . </S>

1002.1334

ba@xmath0mn@xmath1o@xmath2 is a novel layered spin - dimer compound , comprising magnetic dimers of mn@xmath6 ions arranged on triangular planes@xcite .the mn ions occupy equivalent sites with a distorted tetrahedral coordination , resulting in a quenched orbital angular momentum and a total spin @xmath7 @xcite .antiferromagnetic intradimer exchange leads to a singlet ground state with excited triplet and quintuplet states .weaker interdimer exchange leads to dispersion of these excitations , as previously revealed by inelastic neutron scattering @xcite .the rhombohedral r@xmath8 m crystal structure of ba@xmath0mn@xmath1o@xmath2 comprises staggered hexagonal planes ( fig .[ cryst](a ) ) , leading to geometric frustration both within individual planes , and also between adjacent planes .magnetic fields can be used to close the spin gap , and the competition between interdimer interaction on this highly frustrated lattice , and the uniaxial single ion anisotropy associated with the @xmath7 ions , leads to a very complex phase diagram with at least three distinct ordered states @xcite . despite the complexity of the phase diagram of ba@xmath0mn@xmath1o@xmath2, we have previously shown that for fields oriented along the principal crystalline axes the high - field behavior of this material can be described by a remarkably simple spin hamiltonian , including terms representing superexchange between nearest and next - nearest mn ions within and between planes , the zeeman energy and uniaxial single ion anisotropy @xcite .these terms have been determined through a combination of inelastic neutron scattering ( ins ) and electron paramagnetic resonance ( epr ) .ins studies revealed the dominant exchange within a dimer as @xmath9 = 1.642(3 ) mev ; the nearest and next nearest out - of - plane exchanges , @xmath10 = 0.118(2 ) mev and @xmath11 = 0.037(2 ) mev ; and the dominant in - plane exchanges @xmath12 = 0.1136(7 ) mev ( fig . [ cryst ] ) @xcite .epr measurements in the diluted compound ba@xmath0(v@xmath13mn@xmath14)@xmath1o@xmath2 revealed a nearly isotropic @xmath15-tensor , with @xmath16 = 1.96 and @xmath17 = 1.97 , and an easy axis single ion anisotropy @xmath18 = -0.024 mev @xcite .similar measurements in the pure ba@xmath0mn@xmath1o@xmath2 compound revealed a zero field splitting of the triplet states characterized by @xmath18 = - 0.032 mev @xcite , the difference reflecting the additional effect of intradimer dipolar coupling in the undiluted compound .mn@xmath1o@xmath2 ( ba and o ions are not shown ) .values for the exchange constants are given in the main text ( b , c ) phase diagram of singlet - triplet ordered states for fields aligned along the @xmath3 and @xmath4 axes , respectively , taken from our previous work@xcite .i , ii and pm mark phase i , phase ii and the paramagnetic phase as described in the main text .solid and open symbols represent data points obtained via heat capacity and mce , respectively.,width=302 ] at low temperatures , the field dependence of the magnetization of ba@xmath0mn@xmath1o@xmath2 reveals two regions of linearly increasing moment as first the @xmath19 triplet ( between @xmath20 t and @xmath21 t ) and then @xmath22 quintuplet ( between @xmath23 t and @xmath24 t ) states are polarized , separated by a plateau at half the saturation magnetization @xcite . herewe concentrate solely on the first region of increasing magnetization , which can be accessed by moderate laboratory fields . in this fieldrange the ground state can be approximated as a coherent mixture of singlet and @xmath25=1 triplet states .previous measurements revealed a distinct anisotropy in the phase diagram depending on the field direction @xcite .for fields applied along the @xmath3 axis , a single ordered state was found in this regime ( fig .[ cryst](b ) ) .analysis of the minimal spin hamiltonian suggests that this state is an incommensurate xy antiferromagnet with wavevector shifted slightly away from the 120@xmath26 structure favored for a 2d triangular layer@xcite .nmr measurements appear to confirm this conclusion@xcite , but to date the magnetic structure has not been explicitly solved .alternatively , fields applied perpendicular to the @xmath3 axis revealed _ two _ ordered states in this field regime ( fig .[ cryst](c ) ) .analysis of the minimal spin hamiltonian for this field direction yields two distinct incommensurate modulated states - an ising phase stabilized closed to @xmath5 and @xmath27 , as well as a canted xy state in between .the modulation in these two phases is stabilized by the uniaxial anisotropy associated with the zero field splitting of the triplets .similar to the situation for fields oriented along the @xmath3 direction , the exact form of the magnetic structure has not been explicitly solved , and it remains to be seen whether the minimal spin hamiltonian that has been used to describe this system so far is complete .to this end , careful measurements of the phase diagram of ba@xmath0mn@xmath1o@xmath2 for fields at intermediate angles between the @xmath4 and @xmath3 axes have the potential to determine the presence or absence of additional terms . in this paper we present results of heat capacity and torque magnetometry measurements revealing how the two distinct ordered states for fields perpendicular to the @xmath3 axis evolve into a single phase for fields along the @xmath3 axis . through analysis of the previously established minimal spin hamiltonianwe can quantitatively account for the angular dependence of @xmath5 solely via consideration of the triplet dispersion .however , the same analysis , incorporating the predicted magnetic structures , fails to quantitatively account for the angular dependence of the transition between the two ordered states .we discuss the implications of this observation .single crystals of ba@xmath0mn@xmath1o@xmath2 were grown from a naoh flux according to our previously published procedure@xcite .heat capacity ( @xmath28 ) studies were performed on a quantum design physical properties measurement system ( ppms ) using standard thermal relaxation - time calorimetry .these measurements were performed in fields up to 14 t and temperatures down to 0.35 k. the sample was mounted on angled brackets made from oxygen - free high conductivity copper and the field was oriented in the [ 100]-[001 ] plane .cantilever torque magnetometry experiments were performed at the national high magnetic field laboratory ( nhmfl ) in a superconducting magnet for fields up to 18 t in a dilution refrigerator .a crystal was mounted on one face of a capacitance cantilever which was attached to a rigid plate rotatable about an axis parallel to the torque axis and perpendicular to the magnetic field .the magnetic field was aligned away from the principal crystalline axes yielding a finite torque .-@xmath3 plane . labels indicate the angle between the field and the @xmath3 axis .successive data sets are offset vertically by 1.6 j / molk for clarity.,width=264 ] representative heat capacity measurements , taken at 12 t for several angles , are shown in figure [ cpref ] .these data show a single peak for fields aligned along the @xmath3 axis , a peak and a shoulder for fields 15@xmath26 from the @xmath3 axis and two peaks for larger angles .significantly , comparison of the data at 75@xmath26 and 90@xmath26 degrees shows that the 75@xmath26 data has both a slightly higher critical temperature between the paramagnetic phase and phase ii ( @xmath29 ) and also a substantially lower critical temperature between phase ii and phase i ( @xmath30 ) than the 90@xmath26 data .the phase diagram derived from the complete set of @xmath28 measurements , shown in figure [ cp](a ) , reveals the evolution as a function of angle of the two distinct singlet - triplet ordered states for fields in the [ 100]-[001 ] plane .the data show a single transition for all fields for @xmath31 and two transitions for @xmath32 .the extent in temperature of phase ii , @xmath33 , is shown in figure [ cp](b ) .@xmath34 increases as a function of angle as the field is rotated away from the @xmath3 axis , reaches a maximum at 75@xmath26 , and decreases at the @xmath4 axis ( 90@xmath26 ) .for example , for a field of 11 t , @xmath34 is @xmath35 0.07 k larger at 75@xmath26 than at 90@xmath26 . ) , and between phase ii and phase i ( @xmath36 ) , as a function of temperature and angle in the [ 100]-[001 ] plane for various fields , where @xmath37 indicates the angle between the field and the @xmath3 axis .( b ) width of phase ii , @xmath38 , as a function of angle for 11 t , 12 t , 13 t and 14 t ( black circles , red up triangles , green down triangles , and blue squares , respectively.),width=321 ] torque magnetometry measurements , taken at 25 mk , are shown in figure [ tscans ] for three representative angles .the sample was inclined slightly so that the field did not exactly rotate within the @xmath4-@xmath3 plane such that a finite torque was generated for all angles studied .angles are quoted in terms of the angular position with respect to the closest approach to the @xmath3 axis , but it is important to note that the field was never less than @xmath39 from the @xmath3 axis .consequently two phase transitions are observed for all angles studied .critical fields , marking the transition between the paramagnetic phase and phase ii ( @xmath5 ) , and between phase ii and phase i ( @xmath40 ) , were determined from maxima and minima in the second field derivative of the torque divided by field @xcite and are marked with dashed vertical lines .the sign of the peak in the second derivative changed with the evolution of angle , reflecting a change in anisotropy for the two different phases .this leads to minor discontinuities in the determination of the phase boundary ( dashed lines in fig .[ torq ] ) ., 55.0@xmath26 and 88.0@xmath26 with respect to the nearest approach to the @xmath3 axis ( see main text ) .the critical fields were estimated from peaks in the 2@xmath41 derivative as shown by the dashed lines.,width=321 ] similar to the phase diagram obtained from heat capacity measurements ( fig .[ cp ] ) , the phase diagram obtained from torque measurements at 25mk reveals a non - monotonic angle dependence ( fig .[ torq ] ) .the maximum value of @xmath40 occurs between 65 - 75@xmath26 from the closest approach to the @xmath3 axis .this is in agreement with the heat capacity data , for which the smallest @xmath42 occurs at 75@xmath26 from the @xmath3 axis .additionally , the field extent of phase ii , @xmath43 , is largest at 75@xmath26 , and decreases by @xmath350.05 t from the maximum at the highest angles .the previously established minimal spin hamiltonian for arbitrarily oriented field direction in ba@xmath0mn@xmath1o@xmath2 is : @xmath44 where @xmath45 , @xmath46 , @xmath47 is the diagonal gyromagnetic tensor with components @xmath17 , @xmath48 , and @xmath37 is the angle between the applied field and the @xmath3-axis . the quantization @xmath49 axis is set along the field direction .here @xmath50 , @xmath51 designate the dimer coordinates , @xmath52 , @xmath53 denote each of the two s=1 spins in each dimer .the various exchange constants are shown in fig .[ cryst](a ) and are defined as follows : the exchange within a dimer is @xmath54 ; the dominant out - of - plane exchange is @xmath55 for @xmath56 nearest neighbor dimers between planes ; the dominant in - plane exchanges between dimers is @xmath57 and @xmath58 for @xmath56 in plane nearest neighbor dimers and @xmath59 ; and finally the second largest out - of - plane exchange is @xmath60 for @xmath56 next nearest neighbor dimers between planes. axis as described in the main text .red dotted line shows the calculated transition between the paramagnetic and ordered phases as described in the main text.,width=321 ] using this spin hamiltonian and the measured values of @xmath9-@xmath11 and @xmath18 given in the introduction , we have previously been able to quantitatively account for the observed critical fields and magnetization of the ordered states of ba@xmath0mn@xmath1o@xmath2 for fields oriented along the principal axes @xcite .the calculation is based on a generalized spin - wave approach in which we only keep the singlet and the three triplet states of each dimer @xcite .the critical field @xmath61 corresponds to the value for which the energy of lowest energy triplet mode becomes equal to zero .the softening of this triplet mode signals the onset of the magnetic instability towards an ordered state ( phase ii ) . for field directions along the principal axes, it is possible to obtain simple analytical expressions for the critical field .the expression for @xmath31 is : @xmath62 while for @xmath63 the expression is : @xmath64 where @xmath65 is the minimum of the interdimer exchange portion of the dispersion and is fully described in equation ( 3 ) of reference [ ] .the difference in the first two terms between these expressions stems from a change in the zero field splitting of an isolated dimer depending on quantization direction expressed in the reduced basis of dimer states . for the quantization axis along the @xmath3 axis ,the zero field gap of an isolated dimer between the @xmath19 triplet and the singlet is @xmath66 , while for the quantization axis along the @xmath4 axis the equivalent zero field gap is @xmath67 .the two additional terms of @xmath5 for fields perpendicular to the @xmath3 axis arise from a second order process mixing singlets and triplets described in our previous work @xcite .this state mixing causes the gap between the @xmath19 triplet and singlet states to close as @xmath68 , as expected for an ising - like qcp . evaluating these expressions using the values of the exchanges and single ion anisotropy described earlier yield values for the critical fields which are in good accord with the measured data .numerical calculation of @xmath5 for arbitrary field orientations in the [ 001]-[100 ] plane yields the red dotted curve shown in fig .the calculation was performed using the values of @xmath18 and the interdimer couplings given in the introduction while @xmath9 was allowed to vary leading to a fit value of 1.567 mev .the calculated values agree well with the measured data up to the inherent uncertainty associated with the misalignment of the sample in the torque measurements described above .the single - ion anisotropy term of @xmath69 is : @xmath70 .\label{single}\end{aligned}\ ] ] the last term of eq .( [ single ] ) is zero for fields along the @xmath4 and @xmath3 axes , but adds a small contribution for intermediate angles .in particular , this term grows linearly in small deviations of @xmath37 from @xmath71 , @xmath72 , while the other two terms vary quadratically in @xmath73 .thus , the last term of eq .( [ single ] ) determines the shape of the boundary between phase i and phase ii slightly away from @xmath74 , and could be responsible for the striking non - monotonic behavior observed for @xmath40 in fig .[ torq ] as we describe in greater detail below . to understand the effect that the last term of eq .( [ single ] ) has on the ground state , it is convenient to analyze the effective low - energy hamiltonian , @xmath75 , introduced in ref .the low - energy hamiltonian results from projecting the original hamiltonian @xmath76 onto the low - energy subspace generated by the singlet and the @xmath19 triplet states of each dimer .this two - level hilbert space is described by a local pseudo - spin @xmath77 in each dimer , @xmath78 , such that @xmath79 if the dimer @xmath50 is in the @xmath19 triplet state and @xmath80 if it is in the singlet state . in our earlier work we provided the expression of @xmath75 for @xmath81 and@xmath82 @xcite .in particular , for @xmath82 , we showed that the second term of eq .( [ single ] ) generates an effective exchange anisotropy that is responsible for the emergence of phase ii . according to our analysis, this exchange anisotropy favors an ising - like phase in which the transverse spin components ( perpendicular to the applied field ) are aligned along the easy @xmath3-axis .in contrast , phase i is an elliptical spiral phase in which the transverse spin components of adjacent dimers rotate around the field axis . the effect the last term of eq .( [ single ] ) has on @xmath75 at intermediate angles can be determined using second order degenerate perturbation theory .such an analysis yields an effective dzyaloshinskii - moriya ( dm ) interaction between dimers on adjacent bilayers connected by the @xmath10 and @xmath11 exchange constants : @xmath83 with @xmath84 , @xmath85 and @xmath86 .the arrow indicates how the bonds @xmath87 are oriented ( @xmath50 always denotes the dimer in the lower bilayer ) .microscopically , this process turns one singlet into an @xmath19 triplet or vice versa .this effective dm coupling between pseudo - spins results from two important symmetry considerations .first , the singlet and the triplet states of a given dimer have opposite parity under exchange of the two sites of the dimer : @xmath88 .second , the @xmath10 and @xmath11 terms of @xmath76 are invariant under the inversion symmetry transformation around the center of the corresponding bonds : @xmath88 and @xmath89 ( see fig . [ twod](a ) ) .this implies that a dm term is allowed between pseudo - spins connected by the @xmath10 and @xmath11 exchanges .in contrast , the effective dm interaction can not occur for pairs of dimers within a plane because the @xmath90 and @xmath91 terms of @xmath76 are invariant under the symmetry transformation @xmath89 , while the effective dm interaction changes sign under such transformation .these symmetries are fundamentally equivalent to the selection rules governing the dm vector in the effective lattice , where each dimer constitutes a single , unique site . in such a lattice, there is a center of inversion symmetry at the midpoint of the effective @xmath90 and @xmath91 exchange , precluding a dm term by the normal selection rules , while no such inversion symmetry exists at the midpoints of the effective @xmath10 and @xmath11 exchanges .this is contrast to the symmetries of the @xmath92 lattice , where there are inversion symmetries at the midpoint of the @xmath10 and @xmath11 exchanges but none at the midpoint of the @xmath90 and @xmath91 exchanges .that leads to eq .( [ efff2 ] ) .the @xmath3 axis is vertical , and the two dimers belong to adjacent planes .@xmath93 represents either @xmath10 or @xmath11 .( b ) ( i ) , ( ii ) schematic diagram for two pseudo - spins on adjacent dimers connected by the @xmath93 interaction for arbitrary field direction in the @xmath4-@xmath3 plane .red arrows show uniform moment , while green arrows show ordered moment .field direction along @xmath49 axis , @xmath94 along @xmath95 axis , and @xmath3 axis vertical.,width=302 ] this effective dm interaction between dimers on adjacent layers is frustrated in both ordered states at a mean field level ( the mean value of eq .( [ efff2 ] ) is zero for the semi - classical states associated with phases i and ii ) .therefore , the small contribution of the effective dm interaction to the ground state energy must be produced by quantum fluctuations . because the dm vectors point along the @xmath95 direction ( perpendicular to the applied field and to the easy @xmath3 axis ) this contribution term will favor phase ii , for which the pseudo - spins only have @xmath96 and @xmath49 components ( fig .[ twod](b)(ii ) ) , as opposed to phase i for which the pseudo - spins have an additional third component along the @xmath95 direction ( fig .[ twod](b)(i ) ) @xcite .thus this contribution strengthens phase ii relative to phase i near @xmath82 and leads to a small non - monotonic behavior of the @xmath40 curve ( see fig .[ torq ] ) .although this simple analysis captures the qualitative non - monotonic behavior of the @xmath40 curve , it can not account for the magnitude of the observed effect .the amplitude @xmath97 of the effective dm interaction is of order 100 mk for @xmath98 . because the interaction mixes the singlet and @xmath19 triplet dimer states , the mean value of the dm term is less than @xmath99 for any state with magnetization @xmath100 ( the mean density of @xmath19 triplets ) .noting that @xmath101 at @xmath40 , the upper bound on the dm term of @xmath102 mk is the order of magnitude of the observed non - monotonic effect of 5 - 10 mk in the @xmath40 curve .given that @xmath103 is much weaker than the dominant terms of @xmath76 , it is clear that the effective dm term can only explain the magnitude of the non - monotonic effect if it gives a first order contribution to the energy of phase ii .however , as established above for the proposed ordered states , the effective dm interaction contributes via a second order correction and must therefore be considerably smaller .this leaves us with two possibilities : a ) the magnetic structure of phase ii is different from the simple ising phase proposed in ref . [ ] in such a way that the mean value of the effective dm term is non - zero , or b ) the non - monotonic effect is caused by a term that has not been included in @xmath76 . at presentit is impossible to distinguish between these possibilities , but ongoing efforts to experimentally determine the magnetic structure have the potential to directly address option ( a ) , while epr experiments should , at least in principle , be able to determine the energy scale of additional interactions not considered in the minimal spin hamiltonian ( eq .( [ ham ] ) ) .in summary , via heat capacity and torque magnetometry measurements we have established the angular dependence of the phase boundary for singlet - triplet ordered states of ba@xmath0mn@xmath1o@xmath2 .the data reveal a striking non - monotonicity of the phase boundary as the field is rotated between the principal axes .the angle - dependence of @xmath5 can be quantitatively understood in terms of the original minimal spin hamiltonian that we had proposed for this material .this quantity does not depend on details of the magnetically ordered states but only on the minimum of the triplet dispersion .however , the observed non - monotonicity in @xmath104 is at least an order of magnitude larger than anticipated based on this model and assuming the magnetic structures previously proposed .this indicates that a complete theoretical description of this material requires either subtle changes in the proposed magnetic ordered structures or an additional low - energy term in the hamiltonian .work at stanford university is supported by the division of materials research , national science foundation under grant no .lb is supported by doe - bes and the nhmfl - ucgp program .a portion of this work was performed at the national high magnetic field laboratory , which is supported by the national science foundation under cooperative agreement no .dmr-0084173 , by the state of florida , and the department of energy . m. t. weller and s. j. skinner , acta crystallogr . , sect .c : cryst .struct . commun . * 55 * , 154 ( 1999 ) .m. uchida , h. tanaka , h. mitamura , f. ishikawa , and t. goto , phys .b * 66 * , 054429 ( 2002 ). m. b. stone , m. d. lumsden , s. chang , e. c. samulon , c. d. batista , and i. r. fisher , phys .lett . * 100 * , 237201 ( 2008 ) .e. c. samulon , y .- j .jo , p. sengupta , c. d. batista , m. jaime , l. balicas , and i. r. fisher , phys .b * 77 * , 214441 ( 2008 ) .e. c. samulon , y. kohama , r. d. mcdonald , m. c. shapiro , k. a. al - hassanieh , c. d. batista , m. jaime , and i. r. fisher , phys .103 * , 047202 ( 2009 ) .m. h. whitmore , h. r. verdn , and d. j. singel , phys .b * 47 * , 11479 ( 1993 ). s. hill , private communication . _ frustrated spin systems _ ed h. t. diep ( world scientific , singapore , 2004 ) . s. suh , k. a. al - hassanieh , e. c. samulon , j. s. brooks , w. g. clark , p. l. kuhns , l. l. lumata , a. reyes , s. e. brown , c. d. batista , arxiv:0905.0718 . while the @xmath15 factor will also play a role in the angle dependence of phase ii , it is nearly isotropic and will vary monotonically as a function of angle. k. a. al - hassanieh and c. d. batista , in preparation .

<S> heat capacity and magnetic torque measurements are used to probe the anisotropic temperature - field phase diagram of the frustrated spin dimer compound ba@xmath0mn@xmath1o@xmath2 in the field range from 0 t to 18 t . for fields oriented along the @xmath3 axis a single magnetically ordered phase </S><S> is found in this field range , whereas for fields oriented along the @xmath4 axis two distinct phases are observed . </S><S> the present measurements reveal a surprising non - monotonic evolution of the phase diagram as the magnetic field is rotated in the [ 001]-[100 ] plane . </S><S> the angle dependence of the critical field ( @xmath5 ) that marks the closing of the spin gap can be quantitatively accounted for using a minimal spin hamiltonian comprising superexchange between nearest and next nearest mn ions , the zeeman energy and single ion anisotropy . </S><S> this hamiltonian also predicts a non - monotonic evolution of the transition between the two ordered states as the field is rotated in the @xmath4-@xmath3 plane . </S><S> however , the observed effect is found to be significantly larger in magnitude , implying that either this minimal spin hamiltonian is incomplete or that the magnetically ordered states have a slightly different structure than previously proposed . </S><S> -15 mm </S>

quant-ph9906084

a quantum turing machine ( qtm ) is a theoretical model of quantum computers , which expands the classical model of a turing machine ( tm ) by allowing quantum interference to take place on their computation paths . designing a qtm in general , however ,is significantly harder than that of a classical tm because of its _ well - formedness _ condition as well as its halting condition , known as the _ timing problem_. recently bernstein and vazirani @xcite initiated a study of quantum complexity theory founded on a restrictive model : a one - head , multi - track , stationary , dynamic , normal form , unidirectional qtm ( for definitions , see section [ sec : basics ] ) that prohibits a tape head to stay still .we call such a restrictive qtm _ conservative _ for convenience. one may find easier to program a less restrictive qtm when he wishes to solve a problem on a quantum computer . in this paperwe wish to introduce a qtm as general as possible . in section [sec : basics ] , we introduce a multi - tape , multi - head qtm with multi final states that also allows tape heads to stay still .although many variations of qtms are known to be polynomially equivalent @xcite , unsolved is the question of what is the degree of polynomials of these simulation overhead . as we will show in section [ sec : simulation ] , any multi - tape , multi - head , well - formed qtm can be effectively simulated by a conservative qtm with only cubic polynomial slowdown .our primary goal is to contribute to the foundation of programming a handy qtm . in section [ sec : lemmas ], we will prove two fundamental lemmas : well - formedness lemma and completion lemma , which are important tools in constructing a qtm .the lemmas expand the results of bernstein and vazirani @xcite , who considered mostly conservative qtms . using the lemmas, we will show that any computation of a well - formed qtm can be reversed on a well - formed qtm with quadratic polynomial slowdown .we will also address the timing problem in section [ sec : simulation ] . in section [ sec : oracle ] , we will focus on an oracle qtm with multi query tapes and multi oracles . for any oracle qtm @xmath0, we can build an oracle qtm , similar to the classical case , that simulates @xmath0 with a fixed number of queries of fixed length on every computation path .this section briefly describes the formal definition of quantum turing machines . for our purpose, we wish to make the definition as general as possible . herewe present a definition that is slightly more general than the one given in @xcite .a @xmath1-tape _ quantum turing machine _ ( qtm ) @xmath0 is a quintuple @xmath2 , where each @xmath3 is a finite alphabet with a distinguished blank symbol @xmath4 , @xmath5 is a finite set of internal states including an initial state @xmath6 and @xmath7 , a set of final states , and @xmath8 is a multi - valued , _ quantum transition function _ from @xmath9 to @xmath10 .( note that @xmath11 must be defined . ) for brevity , write @xmath12 for @xmath13 .a qtm has two - way infinite tapes of cells indexed by @xmath14 and read / write tape heads that move along the tapes . directions @xmath15 and @xmath16 mean that a head steps right and left , respectively , and direction @xmath17 mean that a head makes no movement .we say that all tape heads move _ concurrently _ if they move in the same direction at any time ( in this case , , we write @xmath18 instead of @xmath19 ) .we call a qtm _ dynamic _ if its heads never stay still .a qtm is _ unidirectional _ if , for any @xmath20 , @xmath21 , and @xmath22 , @xmath23 implies @xmath24 .we assume the reader s familiarity with the following terminology : a _ time - evolution operator _ , a _ configuration _ and _ final configuration _ , a_ superposition _ and a _ final superposition _well - behaved _ and _ stationary _ qtm , and the _ acceptance probability _ of a qtm . for their definitions , see @xcite . hereare ones different from @xcite .a qtm is in _ normal form _ if , for every @xmath25 , there exists a direction @xmath26 such that @xmath27 for any @xmath28 .a qtm @xmath0 is called _ synchronous _ if , for every @xmath29 , any two computation paths of @xmath0 on @xmath29 reach final configurations at the same time .the _ running time _ of @xmath0 on @xmath29 is defined to be the minimal number @xmath30 such that , at time @xmath30 , all computation paths of @xmath0 on @xmath29 reach final configurations .we write @xmath31 to denote the running time of @xmath0 on @xmath29 if one exists ; otherwise , it is undefined .we say that @xmath0 on input @xmath29 _ halts in time @xmath30 _ if @xmath31 exists and @xmath32 .a qtm is _ well - formed _ if its time - evolution operator preserves the @xmath33-norm .a multi - tape qtm is said to be _ conservative _ if it is a well - formed , stationary , dynamic , unidirectional qtm in normal form with concurrent head move .we write @xmath34 to denote the probability that @xmath0 accepts input @xmath29 . throughout this paper, @xmath30 denotes a function from @xmath35 to @xmath36 .in this section , we will prove two lemmas that are essential tools in programming a well - formed qtm : _ well - formedness lemma _ and _ completion lemma_. for convenience , the head move directions @xmath15 , @xmath17 , and @xmath16 are identified with @xmath37 , @xmath38 , and @xmath39 , respectively . [ [ well - formedness - lemma . ] ] * well - formedness lemma . * + + + + + + + + + + + + + + + + + + + + + + + + one of the most significant feature of a qtm is the well - formedness condition on its quantum transition function that reflects the unitarity of their corresponding time - evolution operators .here we present in lemma [ lemma : general - formedness ] three local requirements for a quantum transition function whose associated qtm is well - formed .let @xmath40 be a @xmath1-tape qtm . recall that @xmath12 stands for @xmath13 .we introduce the notation @xmath41 $ ] .let @xmath42 , @xmath43 , and @xmath44 .let @xmath45 and @xmath46 .define @xmath47 and @xmath48 , where @xmath49 and @xmath50 .let @xmath51 , where @xmath52 if @xmath53 and @xmath54 otherwise . finally , we define @xmath41 $ ] as follows : @xmath41 = \sum_{q\in q } \sum_{{\mbox{\boldmath $ { } _ { d}$}}\in d_{{\mbox{\boldmath $ { } _ { \epsilon}$ } } } } \delta(p,{\mbox{\boldmath $ \sigma$}},q,{\mbox{\boldmath $ \tau$}},{\mbox{\boldmath $ d$}})|e_{{\mbox{\boldmath $ { } _ { d}$}}}|^{-1/2 } { | q \rangle } { | h_{{\mbox{\boldmath $ { } _ { d}$}},{\mbox{\boldmath $ { } _ { \epsilon}$ } } } \rangle}$ ] .[ lemma : general - formedness ] ( well - formedness lemma ) a @xmath1-tape qtm @xmath40 is well - formed iff the following three conditions hold . 1. ( unit length ) @xmath55 for all @xmath56 .2 . ( orthogonality ) @xmath57 for any distinct pair @xmath58 .3 . ( separability )@xmath59\cdot \delta[p_2,{\mbox{\boldmath $ \sigma$}}_2,{\mbox{\boldmath $ \tau$}}_2|{\mbox{\boldmath $ \epsilon$}}']=0 $ ] for any distinct pair @xmath60 and for any pair @xmath61 .the proof of the lemma is similar to that of theorem 5.3 in @xcite .note that , since any two distinct tapes do not interfere , a @xmath1-tape qtm must satisfy the @xmath1 independent conditions for the case @xmath62 .we leave the detail to the reader . [[ sec : completion ] ] * completion lemma . * + + + + + + + + + + + + + + + + + + + a quintuple @xmath40 is called a _ partial _ qtm if @xmath8 is a partial quantum transition function that is defined on a subset @xmath63 of @xmath64 .if @xmath8 satisfies the three conditions of lemma [ lemma : general - formedness ] on all entries of @xmath8 , then we call @xmath0 a _ well - formed partial _ qtm @xcite .completion lemma says that any well - formed partial qtm can be expanded to a well - formed qtm .[ lemma : completion](completion lemma ) for every @xmath1-tape , well - formed partial qtm with quantum transition function @xmath8 , there exists a @xmath1-tape , well - formed qtm with the same state set and alphabet whose transition function @xmath65 agrees with @xmath8 whenever @xmath8 is defined . to show the lemma , we first consider how to change the basis of a given qtm .let @xmath40 be a given qtm .we first partition @xmath66 into mutually orthogonal spaces @xmath67 such that ( i ) @xmath68 and ( ii ) for any @xmath46 and any @xmath45 , @xmath41\in { \mathbb{c}}_{{\mbox{\boldmath $ { } _ { \epsilon}$}}}$ ] .note that if @xmath69 then @xmath70 . for each @xmath46 ,let @xmath71 be an orthonormal basis for @xmath72 .let @xmath73 be the union of all such @xmath71 s .we assume that , at time @xmath74 , @xmath0 in state @xmath75 scans symbol @xmath76 and that @xmath8 maps @xmath77 to@xmath78 .define the change of basis from @xmath79 to @xmath80 by mapping @xmath81 into @xmath82 .let @xmath83 denote this transform .this matrix @xmath83 is unitary because @xmath84 , which implies @xmath85 .it is known in @xcite that @xmath83 preserves the @xmath33-norm iff @xmath83 is unitary .let @xmath86 denote @xmath87 for any @xmath88 .in what follows , we show that @xmath65 is `` unidirectional '' in the sense that if @xmath89 then @xmath90 .let @xmath91 and @xmath92 be distinct and in @xmath93 .note that the separability condition ensures that @xmath41\cdot \delta[p',{\mbox{\boldmath $ \sigma$}}',{\mbox{\boldmath $ \tau$}}'|{\mbox{\boldmath $ \epsilon$}}']=0 $ ] for any @xmath94 .since @xmath41 = \sum_{v\in b } \delta'(p,{\mbox{\boldmath $ \sigma$}},v,{\mbox{\boldmath $ \tau$}},{\mbox{\boldmath $ \epsilon$ } } ) { | v \rangle}\in { \mathbb{c}}_{{\mbox{\boldmath $ { } _ { \epsilon}$}}}$ ] , @xmath95 for any @xmath96 if @xmath97 .therefore , @xmath65 is `` unidirectional . ''the transform @xmath83 is useful to show completion lemma .we go back to the formal proof of completion lemma .lemma [ lemma : completion ] let @xmath40 be a given qtm .let @xmath83 be defined as above . as shown above, @xmath83 is unitary . as a result , @xmath98 is a set of orthonormal vectors since so is @xmath99 . for each @xmath100 , let @xmath101 be @xmath91 such that @xmath102 for some @xmath45 if any , andlet @xmath103 otherwise .since @xmath65 is `` unidirectional '' , @xmath101 is uniquely determined .this implies that we can define the vector @xmath104 as @xmath105 .now we expand @xmath106 to @xmath107 by adding arbitrarily extra orthonormal vectors associated with elements in @xmath108 .let @xmath109 be such an expansion of @xmath106 .we define @xmath110 by @xmath111 .we then apply the inverse transform @xmath112 to @xmath113 and let @xmath114 be the result obtained .define @xmath115 .since @xmath83 is unitary , @xmath116 must be well - formed .completion lemma also enables us to use a @xmath1-tuple of a single alphabet , @xmath117 , instead of @xmath118 . in the following sections, we will deal only with a @xmath1-tape qtm with tape alphabets @xmath117 .in this section we demonstrate several simulation results using the main lemmas in section [ sec : lemmas ] . since we are interested only in the acceptance probability of a qtm , the `` simulation '' of a qtm @xmath0 by another qtm @xmath119 in this paper regards with the statement that @xmath17 produces the same acceptance probability as @xmath0 does .more formally , we say that @xmath119 _ simulates @xmath0 with slowdown @xmath120 _ if , for every @xmath29 , @xmath121 and @xmath122 .assume that @xmath0 is a @xmath1-tape well - formed qtm running in time @xmath123 on input @xmath29 .for @xmath124 , let @xmath125 denote the @xmath126-tape qtm that , on input @xmath29 in tapes 1 to @xmath1 and @xmath127 in tape @xmath128 and empty elsewhere , behaves like @xmath0 on input @xmath29 except that the heads in tapes @xmath128 to @xmath129 idle in the start cells .the tape alphabets of @xmath125 for tapes 1 to @xmath1 are the same as @xmath0 s . for convenience ,let @xmath130 denote the final superposition of @xmath0 that starts with superposition @xmath131 . in the case where @xmath131 is an initial configuration with input @xmath29 , we write @xmath132 for @xmath130 . for any pair @xmath133 and @xmath134, @xmath135 denotes the @xmath126-tuple @xmath136 .in particular , we write @xmath137 for @xmath138 and @xmath139 for @xmath140 . [ [ simulation - by - synchronous - machines . ] ] * simulation by synchronous machines . * + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we show how to transform any well - formed qtm into a well - formed , synchronous qtm with a single final state with the help of the information on its running time . [lemma : final - state ] let @xmath0 be a @xmath1-tape , well - formed qtm that halts in time @xmath123 on any @xmath1-tuple input string @xmath29 .then , there exists a @xmath141-tape , well - formed , synchronous qtm @xmath119 with a single final state such that , on input @xmath142 , it halts in time @xmath143 , the last two tape heads move back to the start cells , leaving @xmath127 unchanged , and @xmath144 .if @xmath0 already has a single final state , then @xmath119 needs only @xmath128 tapes and satisfies @xmath145 .let @xmath146 be a given qtm with @xmath147 . by completeness lemma , it suffices to build a partial qtm @xmath119 that satisfies the lemma .assume that @xmath29 is given in tapes @xmath148 to @xmath1 and @xmath127 is in tape @xmath128 .tape @xmath149 is initially empty .the qtm @xmath119 simulates each step of the computation of @xmath0 using tapes @xmath148 to @xmath1 , together with stepping right in tape @xmath128 , which counts the number of steps executed by @xmath0 . when @xmath119 arrives at any final configuration of @xmath0 with final state @xmath150 , @xmath151 , at time exactly @xmath123 , @xmath119 deposits the number @xmath152 ( as a single tape symbol ) onto tape @xmath149 , freeing itself from state @xmath150 .then , @xmath119 moves its @xmath128st tape head back to the start cell in @xmath153 steps and enters its own final state @xmath154 .thus , the running time of @xmath119 is exactly @xmath143 .it is not difficult to check the well - formedness of @xmath119 using well - formedness lemma .note that the acceptance probability of @xmath0 does not change during the above simulation process .thus , @xmath155 .if @xmath0 already has a single final state @xmath154 , we modify the above procedure in the following fashion .firstly , we replace every occurrence of @xmath154 in @xmath8 by @xmath156 .secondly , we apply the above simulation procedure .thirdly , after the simulation , we force @xmath119 to enter @xmath154 as its final state exactly when the @xmath128st tape head returns to the start cell . in this case , we do not need the @xmath149nd tape at all .[ [ simulation - by - machines - with - concurrent - head - move . ] ] * simulation by machines with concurrent head move .* + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the simulation of a multi - tape qtm by a single tape qtm is a central subject in this subsection .we show that any multi - tape , well - formed qtm can be simulated by a certain well - formed , well - behaved qtm with concurrent head move .the simulation overhead here is a quadratic polynomial .this result makes it possible to simulate a multi - tape qtm by a single tape qtm with quadratic polynomial slowdown .[ prop : concurrent ] let @xmath0 be a @xmath1-tape , well - formed qtm that halts in time @xmath123 on input @xmath29 .there exists a @xmath141-tape , well - formed , well - behaved qtm @xmath119 with concurrent head move such that @xmath119 , on input @xmath29 in tapes 1 to @xmath1 and empty elsewhere , simulates @xmath0 in time @xmath157 .moreover , if @xmath0 is synchronous , dynamic , unidirectional , or normal form , so is @xmath119 . in particular ,when @xmath0 is synchronous , @xmath119 can be made stationary with extra @xmath158 steps . given a qtm @xmath0, we construct a new qtm @xmath119 that simulates in @xmath159 steps the @xmath160th step of @xmath0 by moving its heads back and forth in all tapes concurrently and by expanding the simulation area by 2 .thus , the new qtm needs @xmath161 steps ( with an additional pre - computation of 4 steps ) to complete this simulation on input @xmath29 .let @xmath146 be a given qtm .the desired @xmath119 , starting from state @xmath162 , works as follows .initially , in four steps we mark $ in the start cell in tape @xmath149 and we set up the simulation area of three cells ( which are indexed @xmath37 , @xmath38 , @xmath148 ) in tape @xmath128 , each of which holds the record of the head position of @xmath0 .we will maintain this record in tape @xmath128 by updating a symbol @xmath163 in each cell , where @xmath164 means that the @xmath152th tape head rests in the current cell .finally , @xmath119 enters state @xmath165 , where @xmath166 . at round @xmath160 ,@xmath167 , we simulate the @xmath160th step of the computation of @xmath0 in @xmath159 steps .we start with state @xmath168 , provided that @xmath75 is a current state of @xmath0 . moving the head rightward along all tapes toward the end of the simulation area , we collect the information on a @xmath1-tuple @xmath169 of tape symbols being scanned by @xmath0 at time @xmath160 and we then remember it by changing our internal state from @xmath168 to @xmath170 .after the head arrives at the first blank cell , by applying the transition @xmath171 , we change @xmath170 into @xmath172 if @xmath173 is non - zero . to end this simulation phase ,we update the head position marked in tape @xmath128 ( by using @xmath174 ) and tape symbols ( by using @xmath175 ) by moving the head leftward to the first blank cell in tape @xmath128 . whenever the head reaches an end of the simulation area, we expand this area by 1 by writing the symbol @xmath176 in its boundary blank cell .after the simulation phase , @xmath119 enters state @xmath177 .suppose that @xmath0 is in normal form .it is easy to verify that no well - formed qtm in normal form has more than two final state .let @xmath154 be a single final state of @xmath0 .adding the rule @xmath178 makes @xmath119 be in normal form .if @xmath119 is synchronous , then @xmath119 can use the marker $ in tape @xmath149 to move its head back to the start cell and erase $ from the tape in @xmath158 steps .this last movement forces @xmath119 to be stationary .any qtm with concurrent head move can reduce the number of tapes by merging a @xmath1-tuple of tape symbols which the head is scanning , into a single tape symbol .[ lemma : one - tape ] let @xmath179 .let @xmath0 be a @xmath1-tape , well - formed qtm with concurrent head move that , on input @xmath29 in tapes @xmath148 to @xmath180 and empty elsewhere , halts in time @xmath123 .there exists an @xmath180-tape , well - formed qtm @xmath119 such that , on input @xmath29 , halts in time @xmath123 and simulates the computation of @xmath0 on @xmath29 . if @xmath0 is dynamic , synchronous , stationary , unidirectional , or normal form , so is @xmath119 . [[ sec : dynamic ] ] * simulation by dynamic machines . * + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + this subsection is devoted to show that any well - formed qtm can be simulated by a certain conservative qtm with quadratic polynomial slowdown .[ prop : dynamic ] let @xmath0 be a @xmath1-tape , well - formed , synchronous qtm that halts in time @xmath123 on input @xmath181 . there exists a @xmath182-tape , well - formed , stationary , synchronous , unidirectional , dynamic qtm @xmath119 such that , on input @xmath29 in tape 1 to @xmath1 and empty elsewhere , @xmath119 simulates @xmath0 in time @xmath183 .if @xmath0 has a single final state , then @xmath119 is further in normal form .the proof uses an idea of yao @xcite .let @xmath184 be a given qtm with @xmath147 .we define the desired partial @xmath119 so that it simulates the @xmath160 step of @xmath0 by a round of @xmath185 steps with all the heads moving concurrently .since @xmath0 requires @xmath123 steps , @xmath119 needs @xmath186 steps together with a pre- and post - computation of @xmath187 steps , which gives the desired running time .we first show the proposition for the special case @xmath62 .let @xmath188 be an input given in tape 1 . in the initial phase, we create in four steps the configuration @xmath189 , where @xmath190 and @xmath191 is a distinguished state of @xmath119 and symbol $ is in the cell indexed @xmath37 .to understand the simulation phase , we associate a configuration @xmath192 of @xmath0 with a certain configuration @xmath193 of @xmath119 defined in the following way .assume that @xmath194 , where @xmath0 in state @xmath195 scans symbol @xmath196 in the cell indexed @xmath1 and @xmath197 is the content of the tape .at the beginning of round @xmath160 , @xmath198 , we create the configuration @xmath193 of @xmath119 which is of the form @xmath199 , where @xmath200 is written in tape 2 with $ in the cell indexed @xmath37 ( which marks the simulation area ) and @xmath201 is identical to @xmath197 execpt that the cell indexed @xmath1 has symbol @xmath202 instead of @xmath196 . to disregard any head direction that results from an application of @xmath8 , we treat as a single symbol the three consecutive symbols , where the head of @xmath0 scans the middle symbol .in the course of the simulation , wefirst search in tape @xmath203 the three consecutive symbols @xmath204 , where @xmath0 in state @xmath195 scans @xmath205 , and encode them into the single symbol @xmath206 by moving the head back and forth .we then apply @xmath8 to this symbol with stepping right .this makes @xmath119 dynamic and also unidirectional .finally , we decode the result and update the content of tape @xmath203 . for each configuration at time @xmath160 of @xmath0 on input @xmath207 , at the end of the simulation , @xmath119 produces its associated configuration .therefore , when @xmath0 enters a final configuration at time @xmath208 , @xmath119 reaches a configuration in which a tape symbol of the form @xmath209 is found in tape 2 .when @xmath119 finds such a symbol , it enters its own final state @xmath156 in exactly @xmath123 steps .let @xmath210 be the transition function for @xmath119 .for a general case @xmath211 , let @xmath212 , @xmath213 , and @xmath214 .we first produce @xmath215 in tapes @xmath128 to @xmath182 and enters state @xmath216 with changing symbol @xmath217 in the start cell in tape @xmath152 into @xmath218 .we then define @xmath219 to be the product @xmath220 .clearly , this qtm is well - formed , stationary , and unidirectional .note that the running time of the @xmath1-tape qtm @xmath119 does not depend on the number of tapes .since @xmath0 halts at time @xmath123 , @xmath119 finally enters state @xmath221 for some @xmath1-tuple @xmath222 at time @xmath183 . in the case where @xmath0 has a single final state @xmath154, we can add the new transition rule : @xmath223 , where @xmath224 , which makes @xmath119 be in normal form .since the proposition regards with a unidirectional qtm , it also gives an extension of unidirection lemma in @xcite to multi - tape qtms .simply combining propositions [ prop : concurrent ] and [ prop : dynamic ] and lemmas [ lemma : final - state ] and [ lemma : one - tape ] , we obtain the following corollary . let @xmath0 be a @xmath1-tape , well - formed qtm that , on input @xmath207 in tape 1 and empty elsewhere , runs in time @xmath208 .there exist a quartic polynomial and a two - tape conservative qtm @xmath119 such that , on input @xmath225 , @xmath119 halts in time @xmath226 and satisfies @xmath227 .note that , by modifying the simulation given in the proof of proposition [ prop : concurrent ] ( with @xmath228 slowdown , where @xmath229 is any space bound of @xmath0 ) , we can achieve a much tighter @xmath230 time bound .the detail is left to the reader .[ [ reversing - a - computation . ] ] * reversing a computation . * + + + + + + + + + + + + + + + + + + + + + + + + + + first recall definition 4.11 in @xcite that defines the notion : @xmath231 _ reverses the computation of _ @xmath232 .different from @xcite , we only assume that @xmath232 and @xmath231 are well - formed qtms ( whose tape alphabets may differ ) and that @xmath232 has a single final state .we show below that we can reverse the computation of any well - formed qtm with quadratic polynomial slowdown .[ theorem : reverse ] let @xmath0 be a @xmath1-tape , well - formed qtm with a single final state that halts in time @xmath123 on input @xmath29 . there exist a quadratic polynomial @xmath75 and a @xmath233-tape , well - formed , synchronous , dynamic qtm @xmath234 in normal form that , on input @xmath29 in tapes @xmath148 to @xmath1 and @xmath127 in tape @xmath128 and empty elsewhere , reverses the computation of @xmath235 in time @xmath236 .let @xmath237 be a well - formed qtm .by lemma [ lemma : final - state ] , we have a @xmath238-tape , well - formed , synchronous qtm @xmath232 running in time @xmath143 on input @xmath142 that satisfies @xmath239 . by modifying the proof of proposition [ prop : dynamic ] , we can show the existence of a @xmath233-tape , well - formed , stationary , synchronous , unidirectional , dynamic qtm @xmath231 in normal form such that ( i ) @xmath231 on input @xmath142 halts in time @xmath240 , ( ii ) when @xmath231 halts , tape @xmath128 consists only of its input @xmath127 and tapes @xmath149 to @xmath241 are empty , and ( iii ) @xmath242 is identical to @xmath243 when tapes @xmath149 to @xmath241 are ignored .it is easy to extend reversal lemma in @xcite to any multi - tape qtm .let @xmath234 be the qtm ( as constructed in @xcite ) that reverses the computation of @xmath231 with extra two steps . since @xmath231 is well - formed , synchronous , and dynamic , so becomes @xmath234 because of its construction .since any final superposition of @xmath231 is identical to that of @xmath235 , the theorem follows .theorem [ theorem : reverse ] leads to the following lemma .the proof of the lemma also uses an argument similar to that of theorem 4.14 in @xcite .[ lemma : squaring](squaring lemma ) let @xmath244 .let @xmath0 be a @xmath1-tape , well - formed qtm with a single final state which , on input @xmath29 , outputs @xmath245 in the start cell of tape @xmath1 in time @xmath123 with probability @xmath246 .there exist a quadratic polynomial @xmath75 and a @xmath247-tape , well - formed , stationary , normal form qtm @xmath119 such that , on input @xmath248 , @xmath119 reaches in time @xmath236 the configuration in which @xmath119 is in a single final state with @xmath127 in tape @xmath148 , @xmath29 in tapes @xmath203 to @xmath1 , @xmath249 in tape @xmath128 , and empty elsewhere , with probability @xmath250 .let @xmath0 be a given qtm . by theorem [ theorem : reverse ], there exists a @xmath233-tape , well - formed , synchronous , dynamic , normal form qtm @xmath234 that , on input @xmath248 , reverses the computation of @xmath235 in time @xmath240 .we define the desired qtm @xmath119 as follows .let @xmath248 be any input .starting with its initial configuration @xmath251 , @xmath119 runs @xmath235 with ignoring tape @xmath252 .consider the final superposition @xmath253 .when @xmath235 halts , @xmath119 copies the content of the start cell in the output tape into tape @xmath252 in two extra steps . now we have the superposition @xmath254 , where @xmath255 is the content of tape @xmath252 and @xmath256 ranges over all configurations excluding the status of tape @xmath252 .next , @xmath119 runs @xmath234 starting with @xmath131 with ignoring tape @xmath252 .note that @xmath257 for the superposition @xmath258 . by a simple calculation, we have @xmath259 , which equals @xmath246 since @xmath260 outputs @xmath249 with probability @xmath246 . since @xmath234 preserves the inner product , @xmath261 , which is the amplitude in @xmath262 of @xmath263 .thus , the squared magnitude of amplitude of @xmath263 is exactly @xmath250 . [ [ sec : timing ] ] * timing problem . *+ + + + + + + + + + + + + + + + + let @xmath237 be a @xmath1-tape , partial , well - formed , normal form qtm .we assume that any computation path of @xmath0 on input @xmath29 is completely determined by @xmath8 and ends with final state @xmath154 and that the length of any computation path of @xmath0 on @xmath207 does not exceed @xmath123 .we modify @xmath8 by forcing @xmath264 to be @xmath265 for any @xmath266 and let @xmath267 denote this modified @xmath8 .this @xmath268 makes @xmath0 halt within time @xmath123 . for clarity ,let @xmath269 be the qtm defined by @xmath268 .although @xmath269 may not be well - formed , when the final superposition has unit @xmath33-norm , we can still consider the acceptance probability of @xmath269 as before .can we simulate @xmath269 on a well - formed qtm ? for convenience ,we say that @xmath0 is _ well - structured _ if ( 1 ) it is well - formed , ( 2 ) any computation path of @xmath0 on input @xmath207 is completely determined by @xmath8 and ends with a single final state , and ( 3 ) any final superposition of @xmath269 on each input has unit @xmath33-norm . for simplicity , we write @xmath34 to denote the acceptance probability of @xmath269 on input @xmath29 .[ lemma : timing ] let @xmath0 be a @xmath1-tape , well - structured , partial qtm in normal form such that the length of any computation path of @xmath0 on each input @xmath29 is less than @xmath123 .there exists a @xmath270-tape , well - formed qtm @xmath119 such that , on input @xmath29 in tapes @xmath148 to @xmath1 and @xmath127 in tape @xmath128 and empty elsewhere , it halts in time @xmath240 and satisfies @xmath271 .let @xmath0 be a given qtm .we first construct a well - formed qtm @xmath232 running in time @xmath240 on input @xmath142 such that the probability that @xmath0 halts in an accepting configuration in which tape @xmath149 consists only of symbols @xmath272 , is @xmath273 .we produce in tape @xmath274 the `` reversed '' binary representation of @xmath123 in exactly @xmath275 steps . using this representation , we produce @xmath276 bit zeros ( following control bit @xmath148 ) in tape @xmath149 .we simulate @xmath0 s move by incrementing two _ counters_.the first counter is in tape @xmath128 , of unary form , and the second one is a binary counter in tape @xmath149 . at each round of simulating a single step of @xmath0, @xmath232 also increments the unary counter by stepping right and increments the binary one ( using control bit @xmath148 ) in exactly @xmath277 steps .when @xmath0 terminates , @xmath232 keeps incrementing the unary counter but idles on the binary counter for each @xmath277 steps ( using control bit @xmath38 in tape @xmath149 for reversibility ) . \3 .after @xmath123 rounds , we apply a hadamard transform , with stepping right , to the content of the binary counter except its control bit ( , @xmath278 , where @xmath279 is in tape @xmath149 ) .since the length of this counter is @xmath280 , we can observe symbols @xmath281 in tape @xmath149 with amplitude @xmath282 .hence , the probability that @xmath0 reaches an accepting configuration with @xmath272 in tape @xmath149 is @xmath283 .we design @xmath119 so that the heads in tapes @xmath128 to @xmath274 return to the start cells ( using @xmath127 in tape @xmath128 ) and the rest of heads stay in the same cells as @xmath0 s .it is easy to see that @xmath232 is in normal form if we add the rule : @xmath284 .moreover , if @xmath0 is stationary , @xmath232 is also stationary . for the desired machine @xmath119 , we design it to accept input @xmath142 exactly when @xmath232 reaches an accepting configuration with @xmath272 written in tape @xmath149 .it thus follows that @xmath285 .lemma [ lemma : timing ] solves the timing problem for any quantum complexity class whose acceptance criteria is invariant to a polynomial fraction of acceptance probability .unlike the previous sections , we will focus on an oracle qtm , which is a natural extension of a classical oracle tm with the help of a set of oracles .formally , we define a @xmath126-tape _ oracle qtm @xmath0 with @xmath180 query tapes _ to be a septuple @xmath286 , where @xmath5 includes @xmath287 , a set of pre - query states , and @xmath288 , a set of post - query states , and the transition function @xmath8 is defined only on @xmath289 .we assume the reader s familiarity with an _oracle query_. for its definition , see @xcite .conventionally , we assume that every alphabet @xmath290 , @xmath151 , includes binary bits @xmath291 .let @xmath292 be a series of oracles such that each @xmath293 is a subset of @xmath294 .note that query states @xmath295 and @xmath296 correspond only to the @xmath152th query tape and the @xmath152th oracle @xmath293 .we can reduce the number of query tapes by combining a given set of oracles into a single oracle together with copying a query word written in one of query tapes into a single query tape .when we copy a query word @xmath297 from the @xmath152th query tape , we pad the suffix @xmath298 ( between @xmath256 and @xmath299 ) to make the copying process reversible .[ lemma : number - states ] let @xmath300 .let @xmath0 be a @xmath126-tape , well - formed , oracle qtm with @xmath180 query tapes that halts in time @xmath123 on input @xmath181 .let @xmath292 be a series of oracles .there exists a @xmath301-tape , well - formed , oracle qtm @xmath119 with a single query tape such that , on input @xmath142 , halts in time @xmath302 and @xmath303 , where @xmath304 .let @xmath0 be a given qtm . at the end of each round , in which a new qtm @xmath119 simulates a single step of the computation of @xmath0 , we force @xmath119 to make a query ( of the form @xmath305 ) in 6 steps if @xmath0 does not query .when @xmath0 invokes an oracle query , we force @xmath119 to idle for 6 steps instead of making a query of @xmath305 .this proves the lemma below .[ lemma : query - number ] let @xmath0 be a @xmath238-tape , well - formed , oracle qtm in normal form with a single query tape that halts in time @xmath123d , on input @xmath181 .let @xmath306 be an oracle .there exist a @xmath141-tape , well - formed , oracle qtm @xmath119 with two query tapes running in time @xmath307 on input @xmath29 such that @xmath119 makes exactly @xmath123 queries along each computation path and @xmath308 .[ lemma : query - length ] let @xmath0 be a @xmath238-tape , well - formed , oracle qtm in normal form with a single query tape that halts in time @xmath123 on input @xmath181 .let @xmath306 be an oracle set .there exists a @xmath270-tape , well - formed , oracle qtm @xmath119 such that , for every input @xmath142 , it halts in time @xmath312 , the length of any query word is exactly @xmath313 on any computation path , and it satisfies @xmath314 , where @xmath315 .

<S> the notion of quantum turing machines is a basis of quantum complexity theory . </S><S> we discuss a general model of multi - tape , multi - head quantum turing machines with multi final states that also allow tape heads to stay still . to appear in the proceedings + of the 24th international symposium + on mathematical foundations + of computer science , september , 1999 + a foundation of + programming a multi - tape quantum turing machine + + tomoyuki yamakami + department of computer science + princeton university + 35 olden street , princeton , nj 08544 </S>

hep-ph0208230

because of its large mass , the top quark has rich decay products and this opens a new window to test the standard model ( sm ) and to get some clues about the new physics , beyond . in the literaturethere are various studies in the sm and beyond @[email protected] rare flavor changing transitions @xmath5 have been studied in @xcite , @xmath6 in @xcite and @xmath7 in @xcite .the sm predictions of the branching ratio ( @xmath8 ) of the process @xmath5 is @xmath9 @xcite , and @xmath6 is at the order of the magnitude of @xmath10 , in the sm @xcite , which are not measurable quantities even at the highest luminosity accelerators .possible new physics effects are the candidates for the enhancement of the @xmath8 s of the above processes . @xmath6 and @xmath7 decays have been analysed in @xcite and @xcite , in the framework of the general two higgs doublet model ( model iii ) . in these studiesit has been observed that there could be a strong enhancement in the @xmath8 , almost seven orders larger compared to the one in the sm for the decay @xmath6 ; a measurable @xmath8 , at the order of the magnitude of @xmath11 for the decay @xmath7 . in @xcite @xmath12 decayhas been analysed in the topcolor assisted technicolor theory .the charged @xmath13 transitions exist in the sm model and have been studied in the literature extensively .the top decay @xmath14 has been analysed ( see @xcite and references therein ) in the two higgs doublet model and @xmath15 decay has been studied in @xcite .the present work is devoted to the analysis of the charged @xmath16 and @xmath17 decays in the framework of the general two higgs doublet model ( model iii ) .this decay occurs in the tree level with the extended higgs sector since the scalar bosons @xmath18 and @xmath19 exist in the new sector .we study the @xmath8 of the above decays and obtain a measurable quantities , at the order of the magnitude of @xmath20 and @xmath21 , respectively .furthermore , we search the possible cp violating effects . to obtain a nonzero cp asymmetry@xmath22 we take yukawa coupling for @xmath23 transition complex and introduce a new complex parameter , where its complexity comes from some type of radiative corrections , due to the model beyond the model iii ( see section ii ) .we obtain a measurable @xmath22 , at the order of the magnitude of @xmath4 and observe that these physical quantities can give valuable information about physics beyond the sm , the free parameters existing in these models .the paper is organized as follows : in section 2 , we present the @xmath8 and @xmath22 of the decay @xmath24 in the framework of model iii .section 3 is devoted to discussion and our conclusions .if one respects the current mass values of @xmath25 , namely @xmath26 ( @xmath27 ) , the charged @xmath28 is kinematically possible and does not exist in the sm model . with the minimal extension of the higgs sector the cp odd new higgs scalar@xmath19 arises and the @xmath1 decay in the tree level is permitted . in this model ,@xmath0 decay is possible in the tree level , where @xmath18 is the new cp even higgs scalar and , in general , it mixes with the sm one , @xmath29 . in this section, we study the @xmath8 in the general two higgs doublet model , so called model iii and the possible cp violating asymmetry , beyond .the @xmath28 decay is created by the charged @xmath30 process and the neutral @xmath31 or @xmath32 processes , which are controlled by the yukawa interaction @xmath33 where @xmath34 and @xmath35 denote chiral projections @xmath36 , @xmath37 for @xmath38 , are the two scalar doublets , @xmath39 are left handed quark doublets , @xmath40 are right handed up ( down ) quark singlets , with family indices @xmath41 . the yukawa matrices @xmath42 and @xmath43 have in general complex entries . by considering the gauge and @xmath44 invariant higgs potential which spontaneously breaks @xmath45 down to @xmath46 as @xmath47 ^ 2 + c_4 [ ( \phi_1^+ \phi_1 ) ( \phi_2^+ \phi_2)-(\phi_1^+ \phi_2)(\phi_2^+ \phi_1 ) ] \nonumber \\ & + & c_5 [ re(\phi_1^+ \phi_2)]^2 + c_{6 } [ im(\phi_1^+ \phi_2)]^2 + c_{7}\,\ , .\label{potential}\end{aligned}\ ] ] and choosing the parametrization for @xmath48 and @xmath49 as @xmath50\ , ; \phi_{2}=\frac{1}{\sqrt{2}}\left(\begin{array}{c c } \sqrt{2 } h^{+}\\ h_1+i h_2 \end{array}\right ) \,\ , . \label{choice}\end{aligned}\ ] ] with the vacuum expectation values , @xmath51 the @xmath52 and @xmath53 becomes the mass eigenstates @xmath18 and @xmath19 respectively since no mixing occurs between two cp - even neutral bosons @xmath29 and @xmath18 , in tree level .this scenerio permits one to collect sm particles in the first doublet and new particles in the second one .furthermore the flavor changing ( fc ) interaction can be obtained as @xmath54 with the couplings @xmath55 for the fc charged interactions @xmath56 where @xmath57 is defined by the expression @xmath58 notice that the index `` n '' in @xmath57 denotes the word `` neutral '' . using the relevant diagrams for the @xmath28 decay which are given in fig [ fig1 ] and taking into account only the real yukawa couplings @xmath59 , the matrix element square @xmath60 ( @xmath61 ) reads @xmath62 where @xmath63 here the functions @xmath64 are @xmath65 with weak coupling constant @xmath66 , @xmath67 , @xmath68 and @xmath69 and @xmath70 , @xmath71 , @xmath72 and @xmath73 are four momentum of @xmath74 quark , @xmath75 quark , w boson and higgs scalar @xmath25 , respectively . finally , using the well known expression defined in the @xmath74 quark rest frame @xmath76 and the total decay width @xmath77 as @xmath78 , we get the @xmath8 for the decay @xmath28 .now , we would like to study a possible cp violating effects , which can give comprehensive information about the free parameters of the model used .for the process under consideration , the cp violation can be obtained by choosing the complex yukawa couplings in general , namely , taking the parametrizations @xmath79 however , this choice is not enough to get non - zero @xmath22 @xmath80 where @xmath81 is the decay width for the cp conjugate process .this forces one to go beyond the model iii and try to obtain a new complex quantity so that its complexity does not come from the yukawa couplings but from some radiative corrections . under the light of this discussion, we introduce an additional complex correction @xmath82 to @xmath83 transition , which may come from the new model beyond the model iii as @xmath84 here we take the magnitude of @xmath82 at most @xmath85 , which is more than one order smaller compared to the vertex due to model iii . in this case , we take the correction to the @xmath86 transition small since the strength of @xmath86 transition is weaker compared to strength of the @xmath83 transition , with respect to our choice ( see discussion section ) . at this stage , we introduce a model beyond the model iii as follows : the multi higgs doublet model which contains more than two higgs doublets in the higgs sector can be one of the candidate .the choice of three higgs doublets brings new yukawa couplings which are responsible with the interactions between new higgs particles and the fermions .the yukawa lagrangian in three higgs doublet model ( 3hdm ) reads @xmath87 where @xmath88 is the new coupling and @xmath89 can be chosen as @xmath90 with vanishing vacuum expectation value .the fields @xmath91 and @xmath92 represent the new charged and cp even ( odd ) higgs particles , respectively . notice that the other yukawa couplings and higgs particles in eq .( [ lagrangian3h ] ) are the ones existing in the model iii .now , we choose the additional yukawa couplings @xmath88 real and take into account the radiative corrections to the @xmath83 transition which comes from the contributions of third higgs doublet for the decay under consideration . herethe complexity of the parameter should come from the radiative corrections but not from the new yukawa couplings .we can take this complex contribution as a source for the additional part @xmath82 .since the number of free parameters , namely masses of new higgs particles @xmath93 and the new yukawa couplings @xmath88 , increases , there arises a difficulty to restrict them .however , the overall uncertainity coming from these free parameters lies in the contribution @xmath82 and it can be overcome by the possible future measurement of the cp violation for our process .finally , by using the definition @xmath94 we obtain the differential @xmath95 for the process @xmath28 as @xmath96 where @xmath97 with @xmath98 , @xmath99 .notice that we do not present the functions @xmath100 and @xmath101 since their explicit expressions are long . herethe functions @xmath102 and @xmath103 are given in eq.([s1s2 ] ) .this section is devoted to the analysis of the @xmath8 and @xmath22 of the decay @xmath0 and @xmath1 in the framework of model iii and beyond . in our numerical analysiswe use the form of the coupling @xmath104 , which is defined as @xmath105 .since the model iii contains large number of free parameters such as yukawa couplings , @xmath106 , the masses of new higgs bosons , @xmath107 , @xmath18 and @xmath19 , we try to restrict them by using experimental measurements . in our calculations , we neglect all the yukawa couplings except @xmath108 and @xmath109 , due to their their light flavor contents .in addition to this we neglect the off diagonal coupling @xmath110 , since it is smaller compared to @xmath111 ( see @xcite ) .one of the most important experimental measurement for the prediction of the constraint region for the couplings @xmath108 and @xmath109 is the the cleo measurement @xcite @xmath112 and our procedure is to restrict the wilson coefficient @xmath113 which is the effective coefficient of the operator @xmath114 ( see @xcite and references therein ) , in the region @xmath115 , where the upper and lower limits were calculated using eq .( [ br2 ] ) and all possible uncertainities in the calculation of @xmath113 @xcite . in the case of the calculation of @xmath22 , @xmath109 ( @xmath108 )is taken complex ( real ) and a new small complex parameter @xmath82 , due to the physics beyond model iii , is introduced . in the following ,we choose @xmath116 .notice that , in figures , the @xmath8 and @xmath22 are restricted in the region between solid ( dashed ) lines for @xmath117 ( @xmath118 ) . here , there are two possible solutions for @xmath113 due to the cases where @xmath119 and @xmath120 . in the case of complex yukawa couplings , only the solutions obeying @xmath119 exist . in fig .[ brc1h0 ] , we plot the @xmath121 with respect to @xmath122 for @xmath123 , @xmath124 . as shown in this figure , the @xmath8 is at the order of the magnitude of @xmath20 and it increases with the increasing values of the @xmath122 .its magnitude ( the restriction region ) is larger ( broader ) for @xmath117 compared to the one for @xmath118 .[ brc1a0 ] is devoted to the same dependence of the @xmath125 for @xmath123 , @xmath126 . for this processthe @xmath8 is at the order of the magnitude of @xmath21 , almost 2 order smaller compared to the @xmath121 .it increases with the increasing values of the @xmath122 and its magnitude ( the restriction region ) is larger ( broader ) for @xmath117 compared to the one for @xmath118 .furthermore , the restriction region is sensitive to the parameter @xmath122 and for @xmath118 , upper and lower bounds almost coincide .[ brmh0 ] ( [ brma0 ] ) represents @xmath127 with respect to @xmath128 for @xmath123 and @xmath129 . herethe @xmath8 increases with the decreasing values of @xmath128 .this can give a powerfull information about the lower limit of the mass value @xmath130 with the help of the possible future experimental measurement of the process under consideration .notice that with the increasing values of @xmath128 the restriction regions for @xmath117 and @xmath118 become narrower and coincide .now , we would like to analyse the cp asymmetry @xmath22 of the decay @xmath28 . to obtain a nonzero @xmath22 we take the coupling @xmath109 complex and introduce a new complex parameter @xmath82 due to the physics beyond the model iii ( see section ii ) .in fig [ acpsinbbh0 ] ( [ acpsinbba0 ] ) we present the @xmath131 dependence of @xmath132 for @xmath133 , @xmath134 , the intermediate value of @xmath135 and @xmath136 .@xmath22 is at the order of the magnitude of @xmath137 and slightly larger for @xmath118 compared to the one for @xmath117 .fig [ acpsinkaph0 ] ( [ acpsinkapa0 ] ) represents the @xmath138 dependence of @xmath132 for @xmath133 , @xmath134 , the intermediate value of @xmath139 and @xmath140 .the behavior of @xmath22 is similar to the one obtained in fig .[ acpsinbbh0 ] ( [ acpsinbba0 ] ) . as shown in these figures, @xmath22 is a mesurable quantity , which gives strong clues about the possible physics beyond the sm .now we will summarize our results : * the @xmath8 of the process @xmath28 is at the order of @xmath20 ( @xmath21 ) in the model iii and it can be measured in the future experiments .this ensures a crucial test for the new physics beyond the sm . *the @xmath8 is sensitive to @xmath109 and the mass value @xmath141 ( @xmath142 ) .this is important in the prediction of the lower limit of the mass @xmath128 with the possible future experimental measurement of the process under consideration .* @xmath22 is at the order of the magnitude of @xmath4 for the intermeditate values of @xmath131 and and @xmath138 .this a mesurable quantity , which gives a strong clue about the possible physics beyond the sm and also more beyond .therefore , the experimental investigation of the process @xmath143 will be effective for understanding the physics beyond the sm .this work was supported by turkish academy of sciences ( tuba / gebip ). 1 g.mahlon and s.parke , _ phys ._ * b347 * ( 1995 ) 394 ; e. jenkins , _ phys ._ * d56 * ( 1997 ) 458 ; d. atwood and m. sher , _ phys .lett . _ * b411 * ( 1997 ) 306 .g.eilam , j.l.hewett and a.soni , _ phys .* d44 * ( 1991 ) 1473 , erratum _ phys .* d59 * 039901 ( 1999 ) .c.s.li , r.j.oakes and j.m.yang , _ phys .rev . _ * d49 * ( 1994 ) 293 ; j.m.yang and c.s.li , _ phys ._ * d49 * ( 1994 ) 3412 ; g.couture , c.hamzaoui and h.knig , _ phys .* d52 * ( 1995 ) 171 ; j.l.lopez , d.v.nanopoulos and r.rangarajan , _ phys .rev . _ * d56 * ( 1997 ) 3100 ; g.couture , m.frank and h.knig , _ phys .* d56 * ( 1997 ) 4213 ; g.m.de divitiis , r.petronzio , l.silvestrini , _ nucl .phys . _ * b504 * ( 1997 ) 45 .c. yue , g. lu , g. liu and q. xu , _ phys ._ * d64 * ( 1995 ) 095004 ; t.han , r.d.peccei , and x.zhang , _ nucl ._ * b454 * ( 1995 ) 527 ; t.han , r.d.peccei , and x.zhang , _ nucl ._ * b454 * ( 1995 ) 527 ; t.han , k.whisnant , b .- l.young , and x.zhang , _ phys ._ * d55 * ( 1997 ) 7241 ; _ phys . lett ._ * b385 * ( 1996 ) 311 ; m.hosch , k.whisnant , and b .- l.young , _ phys .rev . _ * d56 * ( 1997 ) 5725 ; e. malkawi and t. tait , _ phys . rev ._ * d54 * ( 1996 ) 5758 ; t. tait and c.p .yuan , _ phys .rev . _ * d55 * ( 1997 ) 7300 ; t. han , m. hosch , k. whisnant , b .-young and x. zhang , _ phys ._ * d58 * ( 1998 ) 073008 ; f. del aguila and j.a .aguilar - saavedra , _ phys ._ * b462 * ( 1999 ) 310 ; j. l. diaz - cruz , m. a. perez , g. tavares - velasco , j .j. toscano , _ phys ._ * d60 * ( 1999 ) 115014 ; f. del aguila and j.a .aguilar - saavedra , _ nucl .phys . _ * b576 * ( 2000 ) 56 ; c. yue , g. lu , q.xu , _ phys .* b508 * ( 2001 ) 290 .w.s.hou , _ phys ._ * b296 * ( 1992 ) 179 ; k.agashe , m.graesser , _ phys ._ * d54 * ( 1996 ) 4445 ; m.hosch , k.whisnant , b.l.young , _ phys ._ * d56 * ( 1997 ) 5725 ; j.guasch , hep - ph/9710267 ; l.t.handoko and j.hashida , _ phys* d58 * ( 1998 ) 094008 .g.eilam , b.haeri and a.soni , phys .d41 ( 1990 ) 875 .b. mele , s. petrarca and a. soddu , _ phys .lett . _ * b435 * , ( 1998 ) 401 .t. han and jing jiang , b. mele , _ phys .lett . _ * b516 * , ( 2001 ) 337 .e. iltan , _ phys .rev . _ * d65 * , ( 2002 ) 075017 .e. iltan , i. turan _ phys ._ hep - ph/0207087 . c. yue , g. lu , g. liu , q. xu , _ phys .rev . _ * d64 * , 095004 ( 2001 ) .a. denner , a. h. hoang , _ nucl_ * b397 * , ( 1993 ) 483 ; t. hasuike , t. hattori , t. hayashis .wakaizumi , _ z. phys . _ * c76 * ( 1997 ) 127 .( 2002 ) 483 ; e. jenkins , _ phys .rev . _ * d56 * , ( 1997 ) 456 ; j. l. d. cruz , d. a. l. falcon , _ phys ._ * d61 * , 095004 ( 2000 ) 051701 ; g. alterelli , l. conti , v. lubicz , _ phys . lett . _ * b502 * , ( 2001 ) 125 .t. m.aliev , e. iltan , _ j. phys ._ * g25 * ( 1999 ) 989 .s. chen , et .al . , cleo collaboration , _ phys .* 87 * ( 2001 ) 251807 .

<S> we study the charged @xmath0 and @xmath1 decays in the framework of the general two higgs doublet model , so called model iii and beyond . here , </S><S> we take the yukawa couplings complex and introduce a new complex parameter due to the physics beyond the model iii , to switch on the cp violating effects . </S><S> we predict the branching ratios as @xmath2 and @xmath3 . </S><S> furthermore , we observe a measurable cp asymmetry , at the order of @xmath4 , for both decays . </S>

1001.1769

for different values of @xmath12 .dotted lines include only an expansion of the axion potential up to the term @xmath13,scaledwidth=50.0% ] since axions are real scalar particles , it is very useful to remember how self - gravitating systems made of spin zero particles are constructed .we will follow the method developed by ruffini and bonazzola @xcite .the self - gravitating system arise as a solution of the ekg equations : @xmath14 @xmath15 where @xmath16 $ ] and @xmath17 is the scalar field potential . here@xmath18 denotes an average over the ground state of a system of many particles .its presence refers to the fact that we are working in the semi - classical limit of the einstein s equations .we will work with units where @xmath19 . in the case of a spherically symmetric , static space - time described by @xmath20 it has been shown that such self - gravitating systems are fully characterized by the scalar field properties , i.e. the mass @xmath0 of the scalar field and its energy density potential @xmath21 @xcite .the total mass of the resulting object and the typical radius depend mainly on these two properties of the scalar field .the axion is no exception . to deal with the quantum nature of the axion field, we have to compute the average @xmath22 in eq .( [ einstein ] ) . what is usually done is to quantize the scalar field @xmath23 where @xmath24 and @xmath25 are the usual creation ( annihilation ) operators for a particle with angular momentum @xmath26 , azimuthal momentum @xmath0 and energy @xmath27 .these operators satisfy the usual commutation relations @xmath28=[\mu_{nlm}^-,\mu_{n'l'm'}^-]=0 $ ] and @xmath29=\delta_{n n'}\delta_{l l'}\delta_{m m'}$ ] . with the operator @xmath30, it is now possible to construct the energy - momentum tensor operator @xmath31 just by inserting the operator @xmath32 into the classical expression for the energy - momentum tensor @xmath33 the average @xmath34 is done by considering a state @xmath35 for which all @xmath10 particles are in the ground state ( @xmath36 , @xmath37 ) .the ground state satisfies @xmath38 .this procedure , as pointed out in @xcite , cancels all time dependence on the vacuum expectation value @xmath34 and , for the case of a free scalar field ( @xmath39 ) , the real quantized scalar field yields the same field equations as those obtained by using a classical complex scalar field . at this level , the self - gravitating system for a real quantized scalar field does not differ from a complex classical scalar field , hence , a real quantized scalar field does not produce the so called `` oscillatons '' @xcite , which are time - dependent .in our case we are interested in the axion potential ( [ potential ] ) . in order to compute @xmath40, we perform a taylor expansion of ( [ potential ] ) , i.e. @xmath41 we will show that the final results do not depend strongly on the number of terms considered in the taylor expansion of ( [ potential ] ) . the relevant term that should be considered is the self - interacting term @xmath13 and the sign it carries with itself , which differs from the one considered in boson stars ( bs ) @xcite . with the potential ( [ taylor ] ) , it is possible to compute @xmath42 by performing the quantization and averaging procedure previously discussed .the computed average of the stress energy tensor is @xmath43 we have dropped all sub - indexes since , as we have already pointed out , we will assume that the axion is in its ground state .we can observe that there is no time dependence in ( [ vacuum ] ) .furthermore , there are new numerical factors that appear due to the averaging performed in @xmath44 @xcite . for instance@xmath45 and @xmath46 , in such a way that we can not recover the original @xmath47 from which we departed .following a similar procedure but now applied to the scalar wave equation ( [ kg ] ) , with a potential ( [ taylor ] ) and the spherically symmetric metric ( [ metric ] ) , the einstein - klein - gordon system is obtained : @xmath48&=&0\,.\end{aligned}\ ] ] following standard definitions @xcite , we rewrite the system ( [ sistema ] ) in dimensionless variables : @xmath49 , @xmath50 and @xmath51 , and we have found it convenient to define the dimensionless self - interaction term @xmath52 by imposing regularity at the origin and flatness at infinity , system ( [ sistema ] ) is solved using a shooting method . even though the set of equations ( [ sistema ] ) is very similar to the case for typical bs , @xcite ,the behavior we found for the family of solutions with zero - nodes is completely different . a full set of equilibrium configurationsis shown in fig .[ masses ] , where the gravitational mass is plotted for different values of @xmath53 and @xmath12 .the equilibrium configurations have a maximum mass @xmath54 at some @xmath55 for each value of @xmath12 .but the switch in the potential sign of the @xmath13 term produces a significant change in the behavior of @xmath54 in comparison with standard bs @xcite .the relation @xmath56 is no longer satisfied .instead of increasing @xmath54 as we increase the value of @xmath12 , we found a decreasing @xmath54 . and compactness for different configurations .the dependence of @xmath57 on @xmath12 is negligible . for a given value of @xmath53 , the compactness decreases as @xmath12 increases.,scaledwidth=50.0% ] this effectis expected since instead of adding a repulsive interaction between the particles of the system , the change in the sign due to the cosine - like potential ( [ potential ] ) implies an attractive potential .thus the total number of particles needed to form an equilibrium configuration that balance the gravitational collapse against the quantum pressure is lower than the case of a repulsive potential .one can think that this effect is apparent and as soon as the complete potential ( [ potential ] ) is implemented in the ekg system , a different behavior would be seen .but the decrease in the mass of the equilibrium configurations is a robust behavior .the masses of equilibrium configurations including up to the fourth power of @xmath58 in the taylor series are plotted also in fig .[ masses ] , illustrating this robustness .the bigger the value of @xmath12 , the lower the differences on the masses .this is because when @xmath12 is increased , @xmath59 decreases ( and equivalently @xmath60 where we are interested ) .then , the true expansion parameter of ( [ potential ] ) is @xmath61 , an it is always a small parameter .another interesting issue is that the dependence of the radius @xmath57 ( defined as the radius where @xmath62 of the total gravitational mass is reached ) on the value of @xmath12 is weak , as it is shown in upper panel of fig .[ compactness ] . combining the invariance of the radius as @xmath12 increases , with the decrease in the mass, means that the self - gravitating system made of a scalar field that has an axion - like potential has a lower compactness ( @xmath63 ) as the self - interaction term increases .this `` newtonization '' of the system is shown on the lower panel of fig .[ compactness ] .and an axion mass of @xmath64ev.,scaledwidth=50.0% ] the previous results where obtained assuming arbitrary values of the mass @xmath0 of the scalar field associated with the axion as well as free values for the decay constant @xmath1 .but the mass of the axion is constrained by astrophysical and cosmological considerations to lie in the range @xmath65 and the decay constant is related to the axion mass by @xmath66 @xcite with these two restriction we have @xmath67 and then , the previous selection of dimensionless variables @xmath68 ) is now inadequate in order to numerically solve the system ( [ sistema ] ) .after some frustrated attempts , we found that a more suitable set of variables to solve the system ( [ sistema ] ) is the following : @xmath69 since @xmath70 , it is natural to think that the resulting axion star will have a small compactness and low mass .so , besides the change in variables ( [ variables ] ) , it is convenient to solve for @xmath71 . solving the system ( [ sistema ] ) for the new set of variables @xmath72 we obtained typical nodeless configurations for these axion stars .some of them are shown in fig .[ axionstar ] , where we have taken an axion mass @xmath73ev .a possible scenario emerges with the hypothesis that dm is mainly composed by axions . as was already pointed out by kolb and tkachev @xcite, nonlinear effects in the evolution of the axion field in the early universe may lead to the formation of `` axion miniclusters '' .these miniclusters may relax , due to the collisional @xmath74 process or by gravitational cooling @xcite , and they will evolve to a bs . in the present work we have constructed those bss for axion particles by solving the ekg system for a real quantized scalar field which is regulated by an axion - potential ( [ potential ] ) .these self - gravitating systems , the axion stars , have very small masses and radii of meters ( table [ physicalvalues ] ) and consequently very low compactnesses .the resulting densities are not enough to produce stimulated decays of the axion to photons since they occur when @xmath75 which implies densities @xmath76 kg/ m@xmath77 for @xmath73ev @xcite .typical densities for axion stars are shown in table [ physicalvalues ] .the galactic halo will be an ensemble of axion stars and this picture is not in contradiction with observations since the size of axions stars fit into the limits coming from microlensing or gravothermal instability @xcite .previous studies that construct models for galactic dark matter halos out of scalar fields assume that each galactic halo is a spherical bose - einstein condensate made of an ultra - light scalar field @xcite . in the present work, the axion stars will play the role of the scalar field mini - machos , that is , a scenario where the scalar field ( the axion ) form a large number of stable asteroid - sized scalar condensations which end up clustering into structures similar to cdm halos with all their advantages .the stability of structures made of scalar fields have been extensively studied @xcite , and those analysis could be extended to our case with potential energy density given by ( [ potential ] ) .if dm is distributed as axion stars , their detection will be very difficult .the proposed femptolensing to detect axion compact objects @xcite is close to its lower detectable limit .another related issue is the low number of axion stars around the earth .assuming for instance a navarro - frenk - white profile for the galactic halo , and a local halo density of @xmath78cm@xmath77 around the sun , there will be @xmath80 axion star in the volume cover between jupiter and the sun . nevertheless, another axion properties can shed light on the axion , such as the conversion of axions into photons in the presence of strong magnetic fields @xcite .collisions of axion stars with neutron stars @xcite will produce flashes of light that could be detected by gamma ray observatories @xcite . a more detailed analysis of these ideas together with a more detailed study of the stability of the axion stars could help us to determine if dm is mainly composed by scalar field particles as the axions .+ * acknowledgments : * we would like to thank tonatiuh matos , luis urena , shin yoshida and carlos palenzuela for very enlightening discussion and aaryn tonita for careful reading of the manuscript .we thank l. rezzolla for his support and hospitality at the aei .this work was supported in part by the conacyt and conacyt - sni .99 w. j. g. de blok , s. s. mcgaugh , a. bosma and v. c. rubin , astrophys .j. * 552 * , l23 ( 2001 ) h. v. peiris _ et al . _[ wmap collaboration ] , astrophys .j. suppl .* 148 * , 213 ( 2003 ) a. refregier , ann .astrophys . * 41 * , 645 ( 2003 ) m. tegmark _ et al ._ [ sdss collaboration ] , phys . rev .d * 69 * , 103501 ( 2004 ) m. taoso , g. bertone and a. masiero , jcap * 0803 * , 022 ( 2008 ) w. l. freedman and m. s. turner , rev . mod .phys . * 75 * , 1433 ( 2003 ) p. sikivie , lect . notes phys .* 741 * , 19 ( 2008 ) j. barranco and a. bernal , aip conf .proc . * 1026 * , 245 ( 2008 ) .j. barranco and a. bernal , aip conf .proc . *1083 * , 13 ( 2008 ) .e. takasugi and m. yoshimura , z. phys .c * 26 * , 241 ( 1984 ) .e. w. mielke and f. e. schunck , nucl .b * 564 * , 185 ( 2000 ) e. w. mielke and f. e. schunck , gen .grav . * 33 * , 805 ( 2001 ) .m. colpi , s. l. shapiro and i. wasserman , phys .* 57 * , 2485 ( 1986 ) .g. g. raffelt , lect .notes phys .* 741 * , 51 ( 2008 ) f. e. schunck and e. w. mielke , class .grav . * 20 * , r301 ( 2003 ) x. hernandez , t. matos , r. a. sussman and y. verbin , phys .d * 70 * , 043537 ( 2004 ) e. w. kolb and i. i. tkachev , phys .lett . * 71 * , 3051 ( 1993 ) c. j. hogan and m. j. rees , phys .b * 205 * , 228 ( 1988 ) .s. ghigna , b. moore , f. governato , g. lake , t. r. quinn and j. stadel , astrophys .j. * 544 * , 616 ( 2000 ) j. f. navarro _ et al ._ , monthly notices of the royal astronomical society * 402 * , 21 ( 2010 ) r. ruffini and s. bonazzola , phys .rev . * 187 * , 1767 ( 1969 ) .f. e. schunck and d. f. torres , int .j. mod .d * 9 * , 601 ( 2000 ) l. a. urena - lopez , class .* 19 * , 2617 ( 2002 ) j. w. ho , f. c. khanna and c. h. lee , arxiv : gr - qc/0207073 . c. amsler_ et al . _[ particle data group ] , phys .b * 667 * , 1 ( 2008 ) .e. seidel and w. m. suen , phys .lett . * 72 * , 2516 ( 1994 ) i. i. tkachev , phys. lett .b * 191 * ( 1987 ) 41 .s. j. sin , phys .d * 50 * , 3650 ( 1994 ) e. w. kolb and i. i. tkachev , astrophys .j. * 460 * , l25 ( 1996 ) d. lai , w. c. g. ho , m. van adelsberg and j. s. heyl , arxiv:0906.4717 [ astro-ph.he ] .a. iwazaki , phys .b * 489 * , 353 ( 2000 ) p. sreekumar _ et al ._ [ egret collaboration ] , astrophys .j. * 494 * , 523 ( 1998 )

<S> we show that the inclusion of an axion - like effective potential in the construction of a self - gravitating system of scalar fields decreases its compactness when the value of the self - interaction coupling constant is increased . by including the current values for the axion mass @xmath0 and decay constant @xmath1 , </S><S> we have computed the mass and the radius for self - gravitating systems made of axion particles . </S><S> it is found that such objects will have asteroid - size masses and radii of a few meters , thus a self - gravitating system made of axions could play the role of scalar mini - machos and mimic a cold dark matter model for the galactic halo . the necessity of introducing dark matter ( dm ) as the main component of galactic matter has become a solid fact due to its observational support @xcite . </S><S> nevertheless , the nature of dm is one of the most intriguing mysteries in physics . </S><S> a large variety of particles have been considered as the main component of dm in the universe and only a few of them are still considered as good prospects since they must fulfill several requirements @xcite . among the survivors , </S><S> the neutralino and the axion are leading candidates @xcite . </S><S> the question we address in the present work is the following : if dm is mainly composed of axions , what type of astrophysical objects will the axions form ? in order to answer this question , we have solved the einstein - klein - gordon ( ekg ) equations in the semiclassical limit . </S><S> the source for the einstein equations is the mean value of the energy - momentum tensor operator @xmath2 of a real , quantized scalar field constructed with potential energy density given by @xcite @xmath3 \ , .\ ] ] it is found that the resulting self - gravitating system , the axion star , will have an asteroid - size mass ( @xmath4 ) and radius of a few meters . </S><S> this work improves our previous attempts @xcite by solving numerically the ekg system for current allowed values for the axion mass @xmath0 and decay constant @xmath1 without the necessity of any interpolation . </S><S> our findings differ from previous estimates where the effect of the potential energy density was either neglected @xcite or was taken into account with a wrong sign in the self - interacting term of the potential @xcite . in the first case , it is known that there is a maximum mass for such self - gravitating system given by @xmath5 , where @xmath0 is the mass associated with the scalar field and @xmath6 is the planck mass . for the allowed values of the axion mass , @xmath7 @xcite , the maximum mass for a self - gravitating system with the potential energy density ( [ potential ] ) neglected lies in the range @xmath8 . on the other hand , </S><S> when the axion is considered to have a repulsive self - interacting term , instead of the attractive one given by ( [ potential ] ) , the maximum mass will be as big as @xmath9 . </S><S> here we solve the ekg system including a taylor expansion of the potential energy density ( [ potential ] ) and we observe that its inclusion tends to decrease the mass and consequently the compactness of the self - gravitating system made of axions . due to the smallness of the axion star s masses they could play the role of scalar field mini - machos @xcite and they will be the final state of axion miniclusters @xcite originated in the early universe at the qcd epoch @xcite . assuming that the axion is the main component of dm </S><S> , the galactic halo will be a collisionless ensemble of axion stars and will be indistinguishable to the standard cdm scenario since @xmath10-body simulations of cdm with ultra - high resolution are insensitive to particle mass granularity smaller than @xmath11 @xcite . </S><S> the paper is organized as follows : in section [ ekg ] , the ekg equations for a real , quantized scalar field with a taylor expansion of the potential energy density ( [ potential ] ) are obtained and are solved for arbitrary values of the axion mass @xmath0 and the decay constant @xmath1 . in sec . </S><S> [ main ] we include the current values of @xmath0 and @xmath1 and we obtain the mass and radius of the axion - stars . </S><S> we finish section [ main ] by commenting on some consequences derived in the case that the axion - star has the properties calculated here . </S>

0804.2664

a topological insulator is a material with a bulk electronic excitation gap generated by the spin orbit interaction , which is topologically distinct from an ordinary insulator @xcite .this distinction , characterized by a @xmath3 topological invariant , necessitates the existence of gapless electronic states on the sample boundary . in two dimensions ,the topological insulator is a quantum spin hall insulator@xcite , which is a close cousin of the integer quantum hall state .the edge states predicted for this phase have recently been observed in transport experiments on hgcdte quantum wells@xcite . in three dimensionsthere are four @xmath3 invariants characterizing a time reversal invariant band structure@xcite .one of these distinguishes a strong topological insulator , which is robust in the presence of disorder .the strong topological insulator is predicted to have surface states whose fermi surface encloses an odd number of dirac points and is associated with a berry s phase of @xmath7 .this defines a topological metal surface phase , which is predicted to have novel electronic properties@xcite in ref .we predicted that the semiconducting alloy bi@xmath0 sb@xmath1 is a strong topological insulator using a general argument based on the inversion symmetry of bulk crystalline bi and sb .the surface states of bi have been studied for several years .experimentally there are several photoemission studies of bi crystals and films which have probed the surface states @xcite .there are fewer studies of bi@xmath0 sb@xmath1@xcite , but in very recent work , hsieh et al .@xcite have mapped the ( 111 ) surface states , and verified the topological structure predicted for a strong topological insulator .first principles calculations provide a clear picture of the surface state structure of bi@xcite , which captures many of the experimental features , including their spin structure@xcite . for the alloy , bi@xmath0 sb@xmath1 ,one expects the surface states to evolve smoothly from bi , at least for small @xmath8 .the alloy presents two difficulties for these calculations , though .first , since these calculations must be done on relatively thin slabs , features near the small band gap are inaccessible because finite size quantization mixes the bulk and surface states .moreover , describing the alloy would require some kind of mean field treatment of the substitutional disorder .in this paper we study the surface states of bi@xmath0 sb@xmath1 first by developing a phenomenological tight binding model which can be solved numerically and then by developing general arguments that exploit spatial symmetries and explain a number of model independent features of the surface states .our phenomenological tight binding model is based on an interpolation of a model developed by liu and allen@xcite .this model has the advantage that it can be solved in a semi infinite geometry , which allows the surface state features near the small band gap to be calculated .our aim is not to perform a quantitatively accurate calculation of the surface states , but rather to provide a concrete calculation in which robust , model independent features of the surface states can be identified and characterized . herewe list our main conclusions : \(1 ) we find that the fermi surface of the 111 surface of bi@xmath0 sb@xmath1 consists of an electron pocket centered around the @xmath9 point and six elliptical hole pockets centered a point in between @xmath9 and the @xmath10 point .( here the bar refers to symmetry points in the 111 surface brillouin zone ) .this is similar to the surface states in bi . unlike the alloy , however , bi has bulk states at the fermi energy : hole states near @xmath9 and electron states near @xmath10 .this calculation verifies the topological structure of the surface states predicted in ref . .in that work we showed that the four @xmath3 invariants @xmath11 characterizing the valence bands of pure bi and sb are @xmath12 and @xmath2 respectively .the semiconducting alloy bi@xmath0 sb@xmath1 was argued to be in the same class as sb , which is a strong topological insulator .these invariants determine the number of surface bands crossing the fermi energy modulo 2 between each pair of time reversal invariant momenta ( trim ) in the surface brillouin zone .specifically , it predicts that for the 111 surface an odd number of fermi surface lines separate the @xmath9 point from the three equivalent @xmath10 points .this is consistent with both our calculation and with experiment@xcite .\(2 ) we will show that for crystals with inversion symmetry there is additional topological structure in the bulk band structure , which further constrains the surface band structure . at each of the 8 trim , @xmath13 in the bulk brillioun zone, the product of the parity eigenvalues of the occupied bands defines a _ parity invariant _ @xmath14 , which is a topological invariant in the space of inversion symmetric hamiltonians .the four @xmath3 invariants , which require only time reversal symmetry are determined by these 8 signs , and determine the number of fermi surface lines separating two surface trim .they do not , however , specify which of the trim are _ inside _ of the surface fermi surface and which are _outside_. we will show that the bulk parity invariants @xmath14 provide that information . specifically , for each surface trim we will define the _ surface fermion parity _ as the parity of the number of fermi lines that enclose that trim .this distinguishes the trim that are outside the fermi surface from those that are inside a ( single ) electron or hole pocket . for a crystal terminated on an inversion planewe will establish a theorem which relates the surface fermion parity to the bulk parity invariants .thus , for inversion symmetric crystals , the 8 bulk parity invariants provide _ more _ information about the surface states than just the four @xmath3 invariants . we will give a simple proof of this theorem in appendix a , which establishes a more direct connection between the bulk parity eigenvalues and the surface state structure than that presented in ref . .for the 111 surface of bi@xmath0sb@xmath1 our general theorem is consistent with both our surface state calculation and with experiment .we will also apply this result to make predictions about the other surfaces of bi@xmath0sb@xmath1 .in addition , our theorem has implications for inversion symmetric crystals which are ordinary insulators .in particular , we will show that it has non trivial implications for the surface states of pure bi , whose valence band is in the trivial @xmath12 topological class .\(3 ) in addition to inversion symmetry , the crystal lattices of bi and sb have a mirror symmetry .we will show that the presence of mirror symmetry leads to a further topological classification of the bulk band structure in terms of an integer , @xmath4 , which we refer to as a _ mirror chern number_. this integer is similar to the spin chern number , which occurs in the quantum spin hall effect when spin is conserved@xcite , and its parity is related to the @xmath3 invariant@xcite . the valence band of pure bi , which has the ( 0;000 ) @xmath3 class@xcite , has @xmath15 .the semiconducting alloy is a topological insulator with @xmath3 class @xmath2 .there are two possibilities for the mirror chern number @xmath16 , however , which correspond to topologically distinct phases .we will show that the sign of @xmath4 in the topological insulator phase further constrains the behavior of the surface states .the transition between the @xmath12 and @xmath2 classes in bi@xmath0sb@xmath1 occurs for small @xmath17 because pure bi is very close to a band inversion transition where the @xmath18 valence band and @xmath19 conduction band cross .the @xmath20 theory of these states has been studied extensively in the literature @xcite and has the form of a nearly massless _ three dimensional _ dirac point .we will show that the _ change _ @xmath21 in the mirror chern number at the band inversion transition is determined by a previously unexplored parameter in that theory : a sign @xmath22 which we will refer to as the _ mirror chirality_. @xmath6 is related to the sign of the @xmath23 factor , which relates the magnetic moment to the angular momentum in a particular direction .for @xmath24 the @xmath23 factor is like that of a free electron , while for @xmath25 it is anomalous .we will use this result to interpret our surface state calculation and to provide guidance for how @xmath6 can be measured .in addition to the dirac point enclosed by the surface fermi surface at @xmath9 , our tight binding surface band calculations for both pure bi and bi@xmath0sb@xmath1 predict that the 6 hole pockets also enclose dirac points which reside at points along the line between @xmath26 and @xmath10 . unlike the dirac points at the surface trim , the degeneracy at these dirac points is not protected by time reversal symmetry , but rather by mirror symmetry .this prediction is inconsistent with first principles calculations of the surface states in bi@xcite , which do not find a band crossing inside the hole pocket .since the dirac point occurs above the fermi energy arpes experiments do not directly probe this issue .nonetheless , spin resolved arpes experiments on bi provide evidence that the surface band structure is consistent with the first principles calculations@xcite .we will show that this inconsistency can be traced to the mirror chirality and the mirror chern number .the mirror chirality in the topological insulator phase of bi@xmath0sb@xmath1 can be determined from the structure of the @xmath5 perturbation theory of the energy bands in the vicinity of the @xmath27 point in pure bi .we find the liu - allen model predicts that @xmath28 .this value implies that the surface state bands in the alloy cross in such a way as to establish the presence of the dirac points in the hole pockets in agreement with our surface state calculation .in contrast , we find that an earlier , but more fundamental pseudopotential calculation by golin@xcite predicts that @xmath29 .this value predicts that the bands do not cross , and that there are no extra dirac points , which is consistent with the presently available experimental results as well as first principles calculations@xcite .the liu allen tight binding parameters were chosen to reproduce the _ energy _ of the bands computed using first principles calculations , incorporating available experimental constraints .therefore , there is no reason to expect that it gets @xmath4 right .we conclude that the inconsistency in our surface state calculation is an artifact of the liu allen tight binding model , which culd be corrected with a suitable choice of new parameters .the outline of the paper is as follows . in sectionii we will review the salient features of bulk bi@xmath0 sb@xmath1 and describe our phenomenological tight binding model . in sectioniii we will describe our surface state calculations for bi@xmath0 sb@xmath1 . in section iv we will establish the relationship between the surface fermion parity and the bulk parity eigenvalues and use that result to analyze the surfaces of bi@xmath0 sb@xmath1 . in section v we will discuss the mirror chern number , and show it is related to the mirror chirality of the @xmath5 theory of pure bi . in sectionvi we will conclude with a discussion of the relevance of our results to existing and future experiments .finally , in appendix a we provide a simple proof of the theorem relating the surface fermion parity to the bulk parity eigenvalues .bismuth and antimony are group v semimetals .they have the rhombohedral a7 structure shown in fig . [ crystal](a ) , which can be viewed as a distorted simple cubic lattice in which the triangular ( 111 ) lattice planes ( which we will refer to as monolayers )are paired to form bilayers .the trivalent @xmath30 atoms tend to form strong covalent bonds directed to the three nearest neighbors within a bilayer .different bilayers are more weakly coupled .the primitive unit cell consists of two atoms in different monolayers , and each bilayer has a structure similar to a honeycomb lattice . the brillouin zone for this lattice is shown in fig .[ crystal](b ) .it contains 8 special points which are invariant under inversion and time reversal , denoted by @xmath31 , @xmath32 and 3 equivalent @xmath27 and @xmath33 points .both bi and sb have a finite direct energy gap throughout the brillouin zone , but they have a negative indirect gap . in bi the conduction band minimum at @xmath27 is below the valence band maximum at @xmath32 , which gives rise to an anisotropic hole pocket and three electron pockets with small effective masses@xcite . at @xmath27 the conduction band minimum , which has even parity @xmath18 symmetry , nearly touches the valence band maximum , with odd parity @xmath19 symmetry , forming a three dimensional dirac point with a small mass gap @xmath34 mev . in sb ,the conduction band minimum at @xmath27 has @xmath19 symmetry , and is below the valence band maximum at the lower symmetry @xmath35 point .the alloy bi@xmath0 sb@xmath1 retains the rhombohedral a7 crystal structure .the evolution of its band structure has been studied experimentally @xcite . as @xmath8is increased from zero two things happen .first , the small gap at @xmath27 closes and then reopens .the @xmath18 and @xmath19 bands switch places , and the mass of the three dimensional dirac point changes sign .second , the top of the valence band at @xmath32 descends below the bottom of the conduction band , resulting in a semimetal - semiconductor transition .for @xmath36 the alloy is a direct gap semiconductor a gap of order 30 mev at the l points .time reversal invariant band structures are classified topologically by four @xmath3 invariants @xcite . in ref .we exploited inversion symmetry to show that these four invariants can be determined by the parity @xmath37 of the occupied bands at the 8 trim @xmath13 , via the quantities @xmath38 which we will refer to as _parity invariants_. here the product includes each kramers pair ( which satisfy @xmath39 ) only once . for an inversion symmetriccrystal , all 8 of the parity invariants are topological invariants in the following sense .if the crystal hamiltonian is smoothly deformed , _ preserving the inversion symmetry _ , then the only way any of the @xmath14 s can change is if the gap at @xmath13 goes to zero , so that states with opposite parity can be exchanged between the conduction and valence band .if inversion symmetry is relaxed , then the 8 invariants lose their meaning .however , in ref .we showed that provided _ time reversal symmetry _ is preserved four combinations of the @xmath14 remain robust and define the four @xmath3 invariants denoted by @xmath11 .the most important invariant , @xmath40 , distinguishes the strong topological insulator , and survives even in the presence of disorder@xcite .@xmath41 is given simply by the product of all 8 @xmath14 .pure bi and sb have inversion symmetry .the parity eigenvalues for inversion about the point o in fig .[ crystal](a ) are tabulated in the literature@xcite .based on this data we display @xmath14 in table [ deltatab ] , along with the predicted @xmath3 invariants for pure bi , pure antimony and the alloy .the valence band of pure bi is characterized by the trivial class @xmath12 , while antimony has the @xmath2 class .the difference is due to the inversion of the @xmath18 and @xmath19 bands , which changes the sign of @xmath42 .the alloy inherits its topological class from antimony , and is a strong topological insulator ..parity invariants @xmath14 and @xmath3 topological invariants @xmath11 for bismuth , antimony , and bi@xmath0sb@xmath1 determined from the product of parity eigenvalues @xmath37 at each bulk trim @xmath13 . [ cols="^,^,^,^,^,^",options="header " , ] [ tab1 ] .[ kptable ] in the previous section we showed that provided there is only a single transition between pure bi and the topological insulator phase of bi@xmath0sb@xmath1 , the mirror chirality deduced from the pure bi band structure determines the mirror chern number in the topological insulator .this , in turn , determines the direction of propagation of the @xmath43 and @xmath44 states along the line @xmath45 .the surface state structure predicted by the tight binding model was shown in fig .[ tbsurface](c ) .the crossing of the @xmath46 band is consistent with @xmath24 .this crossing guarantees that there is a dirac point enclosed by the hole pocket .this can be probed either by inverse photoemission or by spin polarized photoemission . in the latter case, the presence of the dirac point would lead to a change in the sign of the spin on either side of the hole pocket .it will be interesting to experimentally determine this property for bi@xmath0sb@xmath1 using spin polarized arpes . currently available spin polarized photoemission data on the 111 surface of pure bi@xcite provide an indirect probe of @xmath6 .hole pockets are observed along the line from @xmath9 to @xmath10 in both bi@xmath0sb@xmath1 and pure bi . provided we make the plausible assumption that no additional level crossings occur near the transition to the topological insulator , then the presence or absence of dirac points in the hole pockets should be the same on both sides of the transition . in ref . , the spin in either side of the hole pocket was found to point in the same direction , which indicates that in pure bi , the hole pockets do not enclose a dirac point .this conclusion was supported by first principles surface state calculations , which also find no crossing@xcite .this suggests that in the alloy , it should be the @xmath47 band which connects the conduction and valence bands , which is consistent with @xmath25 .it thus appears likely that the mirror chirality in bi has the anomalous sign , @xmath25 .this conclusion contradicts the prediction of the tight binding model , but it is supported by ( 1 ) the pseudopotential band structure of pure bi and ( 2 ) the observed and calculated surface state structure of pure bi .spin polarized arpes experiments on the topological insulator bi@xmath0sb@xmath1 could more directly determine this sign by probing the mirror chern number @xmath4 .in this paper we have analyzed the surface state structure of the topological insulator bi@xmath0sb@xmath1 . using a simple tight binding model based on liu and allen s tight binding parameterization we confirmed that the surface states have the signature of the strong topological insulator by showing that the surface fermi surface encloses an odd number of dirac points .the tight binding model also predicts that the surface is semi metallic , with an electron pocket centered on @xmath9 along with 6 hole pockets . using general arguments based on inversion symmetry, we showed that the location of electron and hole pockets in the surface brillouin zone is constrained by a quantity which we defined as the surface fermion parity .this quantity is determined by the parity invariants of the bulk band structure , and for a given surface it determines which surface trim are enclosed by an odd number of fermi surface lines .this argument establishes a simple and direct connection between the bulk electronic structure and the surface electronic structure for crystals with inversion symmetry . using this general principle , we predicted the structure of the surface states for several different faces of bi@xmath0sb@xmath1 .for the 111 face , these predictions agree both with our surface state calculations and with experiment. it will be interesting to test these predictions experimentally on other faces of bi@xmath0sb@xmath1 .finally , we showed that the mirror symmetry present in the rhombohedral a7 lattice leads to additional topological structure in the bulk energy bands .we defined an integer mirror chern number @xmath4 , whose value is nonzero in the topological insulator phase .the sign of @xmath4 determines the direction of propagation of each of the surface states along the mirror plane , and thus determines which surface states connect the conduction and valence bands .we find that the crossing of the @xmath46 band predicted by the tight binding model , which leads to a dirac point in the hole pockets , disagrees with the natural extrapolation of experiments and first principles calculations on pure bi , which find no dirac point in the hole pockets .we traced this discrepancy to a previously unexplored property of the @xmath5 band structure of pure bi , which we defined as the mirror chirality , @xmath6 .we showed that @xmath6 in pure bi determines @xmath4 in the topological insulator .moreover , we showed that the liu allen model predicts the conventional value @xmath24 , while an earlier pseudopotential calculation by golin predicts the anomalous value @xmath25 .the latter value is consistent with the available experimental data on bi , though the connection is rather indirect .a more direct test would be to directly measure the mirror chirality @xmath4 in the topological insulator by probing the surface states with spin polarized arpes. it would be interesting to check that the value of @xmath6 predicted by more accurate first principles calculations of bi agrees with the pseudopotential prediction .since the tight binding model was designed only to get the energies of the bands right , there is no reason to expect that it would get @xmath6 right .it should be possible to come up with a new parameterization of the liu allen model which would have @xmath25 .we expect that the surface states computed within this model would have band crossings which agree with experiment and first principles calculations , though of course a quantitative description of the surface states requires an accurate description of the surface potential .an important lesson to be learned from this paper is that in addition to time reversal symmetry , spatial symmetries can play an important role in topologically constraining bulk and surface band structures .our analysis of these symmetries has not been exhaustive . a complete theory of _ topological band theory _ , which accounts for the full point group symmetry of a crystalis called for .we thank gene mele for helpful discussions and zahid hasan and david hsieh for sharing their experimental results prior to publication .this work was supported by nsf grant dmr-0605066 , and by acs prf grant 44776-ac10 .in this appendix we show that for an inversion and time reversal invariant crystal the surface fermion number @xmath48 discussed in section iv is an integer , whose parity is determined by the product of bulk parity invariants @xmath49 , which are products of parity eigenvalues given in eqs .[ delta ] and [ pi ] .the simple proof outlined here provides a direct connection between the topological structure of the surface states and the parity eigenvalues characterizing the bulk crystal .the bloch hamiltonian @xmath50 describes a parity and time reversal invariant one dimensional system . in the followingwe will suppress the dependence on @xmath51 and consider a purely one dimensional system . to determine the end charge @xmath52 we introduce the cutting procedure " depicted in fig .[ 1dfig](a ) .we begin with a large but finite system with periodic boundary conditions .we then replace the hopping amplitudes @xmath53 for all bonds that cross the cleavage plane @xmath54 by @xmath55 , where @xmath56 is real .provided @xmath54 corresponds to an inversion plane , the one dimensional hamiltonian retains inversion and time reversal symmetry for all @xmath56 .the fully cleaved crystal corresponds to @xmath57 . for @xmath58the system is translationally invariant , so the excess charge near @xmath54 is @xmath59 . since the insulator can have no bulk currents , the only way @xmath60 can changeis if a state localized near @xmath54 crosses the fermi energy .thus @xmath61 will be the difference between the number of states that cross @xmath62 from above and from below for @xmath63 $ ] .kramers theorem requires that every state is at least twofold degenerate , so the number of states crossing @xmath62 will be an _ even _ integer .since the charge will be divided evenly between the two sides , @xmath64 is an integer , which may be written @xmath65 where @xmath66 is the number of _ kramers pairs _ that cross @xmath62 from above or below .we now relate the parity of @xmath52 to the bulk parity eigenvalues . to this endit is useful to consider the evolution of the spectrum for @xmath67 $ ] and to define @xmath68 as the product of the parities of all of the occupied states , where each kramers pair @xmath69 is included only once .this quantity is well defined because @xmath70 .our proof consists of two steps .we will first show that @xmath71 we will then show that @xmath72 \equiv \pi .\label{pp2}\ ] ] here @xmath73 are the parity of the bloch states in the @xmath74th kramers degenerate band at the trim @xmath75 , and again each kramers pair is included only once .@xmath76 is the number of occupied kramers degenerate bands .taken together , ( [ pp1 ] ) and ( [ pp2 ] ) establish the relationship summarized by eqs .[ delta ] and [ pi ] between the bulk parity eigenvalues and the surface fermion parity .[ pp1 ] follows from the symmetry of the end state spectrum about @xmath77 .the hamiltonian @xmath78 differs @xmath79 only by a phase twist of @xmath7 across @xmath54 .this twist can be spread over the entire circumference @xmath27 by performing the gauge transformation @xmath80 for @xmath81 . when @xmath82 the hamiltonian for @xmath83 near @xmath54 becomes identical to @xmath79 .thus every bound state @xmath84 satisfies @xmath85 . since ( a5 ) changes the parity , @xmath84 and @xmath86 have opposite parity .it follows that every kramers pair that crosses the @xmath62 at @xmath87 $ ] has a partner with opposite parity that crosses @xmath62 in the opposite direction at @xmath88 as shown in fig [ 1dfig](b ) .thus between @xmath58 and @xmath89 the conduction and valence band exchange two kramers pairs with opposite parity , leading to a change in the relative sign between @xmath90 and @xmath91 .we conclude that @xmath92 , which leads directly to ( [ pp1 ] ) .. [ pp2 ] follows from a consideration of the parities of the bloch wavefunctions .consider first the simplest case where there is a single kramers degenerate occupied band , as shown in fig .[ 1dfig](c , d ) . at @xmath58the single particle states are labeled by momentum @xmath93 with @xmath94 , where @xmath95 is the number of unit cells . at the twotrim @xmath96 , @xmath97 the parity eigenvalues are @xmath98 .every other @xmath99 has a partner @xmath100 , and even and odd parity combinations of the two can be formed .the @xmath101 @xmath102 pairs thus each contribute @xmath103 to the product in ( [ pprod ] ) .therefore , @xmath104 for @xmath89 the gauge transformation ( [ gauge ] ) leads to a periodic hamiltonian identical to @xmath105 , but with momenta shifted by @xmath106 , as shown in fig .[ 1dfig](d ) .thus all the momenta are paired , so that @xmath107 combining ( [ p1 ] ) and ( [ p2 ] ) leads directly to ( [ pp2 ] ) , which is straightforwardly generalized to the case of @xmath76 kramers degenerate bands .

<S> we study the electronic surface states of the semiconducting alloy bismuth antimony ( bi@xmath0sb@xmath1 ) . using a phenomenological tight binding model </S><S> we show that the fermi surface for the 111 surface states encloses an odd number of time reversal invariant momenta ( trim ) in the surface brillouin zone . </S><S> this confirms that the alloy is a strong topological insulator in the @xmath2 @xmath3 topological class . </S><S> we go on to develop general arguments which show that spatial symmetries lead to additional topological structure of the bulk energy bands , and impose further constraints on the surface band structure . </S><S> inversion symmetric band structures are characterized by 8 @xmath3 parity invariants " , which include the 4 @xmath3 invariants defined by time reversal symmetry . </S><S> the extra invariants determine the surface fermion parity " , which specifies which surface trim are enclosed by an odd number of electron or hole pockets . </S><S> we provide a simple proof of this result , which provides a direct link between the surface state structure and the parity eigenvalues characterizing the bulk . using this result </S><S> we make specific predictions for the surface state structure for several faces of bi@xmath0sb@xmath1 . </S><S> we next show that mirror invariant band structures are characterized by an integer mirror chern number " , @xmath4 , which further constrains the surface states . </S><S> we show that the sign of @xmath4 in the topological insulator phase of bi@xmath0sb@xmath1 is related to a previously unexplored @xmath3 parameter in the l point @xmath5 theory of pure bismuth , which we refer to as the mirror chirality " , @xmath6 . the value of @xmath6 predicted by the tight binding model for bismuth disagrees with the value predicted by a more fundamental pseudopotential calculation . </S><S> this explains a subtle disagreement between our tight binding surface state calculation and previous first principles calculations of the surface states of bismuth . </S><S> this suggests that the tight binding parameters in the liu allen model of bismuth need to be reconsidered . </S><S> implications for existing and future angle resolve photoemission ( arpes ) experiments and spin polarized arpes experiments will be discussed . </S>

1212.4944

as time goes by , our confidence in what we are doing seems to be dwindling - so to look for `` superstring '' , etc ., as alternatives . nevertheless , the language which was built up during the last century , namely , renormalizable quantum field theory , may still be the language underlying the final standard model. it will be the language of this paper . usually in a textbook, the qcd chapter precedes the one on glashow - weinberg - salam ( gws ) electroweak theory .nothing is wrong with it but the basic units ( or the building blocks ) are further divided into the left - handed and right - handed components. it would be nice ( in helping us in thinking ) if the framework is formulated all at once - in an extended standard model we could see everything consistent with one another .then , the questions which we pose could have broader meanings and implications . thus , this is what we wish to do . we shall work with the lie group @xmath4 as the gauge group .thus , the basic units are made up from quarks ( of six flavors , of three colors , and of the two helicities ) and leptons ( of three generations and of the two helicities ) , together with all originally massless gauge bosons and the somewhat hidden induced higgs fields . in view of the search over the last forty years , we could assume `` minimum higgs hypothesis '' as the working rule . if we look at the basic units as compared to the original particle , i.e. the electron , the starting basic units are all `` point - like '' dirac particles .dirac invented dirac electrons eighty years ago and surprisingly enough these `` point - like '' dirac particles are the basic units of the standard model .thus , we call it `` dirac similarity principle '' - a salute to dirac ; a triumph to mathematics .our world could indeed be described by the proper mathematics .the proper mathematics may be the renormalizable quantum field theory , although our confidence in it sort of fluctuates in time .there is no way to `` prove '' the above two working rules - `` dirac similarity principle '' and `` minimum higgs hypothesis '' .it might be associated with the peculiar property of our lorentz - invariant space - time . to use these two working rules ,we could simplify tremendously the searches for the new extended standard models .so far , we have decided on the basic units - those left - handed and right - handed quarks and leptons ; the gauge group is chosen to be @xmath0 . in the gauge sector , the lagrangian is fixed if the gauge group is given ; only for a massive gauge theory , higgs fields are called for and we postpone its discussions until we have spelled out the fermion sector . for the fermion sector , the story is again fixed if the so - called `` gauge - invariant derivative '' , i.e. @xmath5 in the kinetic - energy term @xmath6 , is given for a given basic unit @xcite . thus , we have , for the up - type right - handed quarks @xmath7 , @xmath8 , and @xmath9 , @xmath10 and , for the rotated down - type right - handed quarks @xmath11 , @xmath12 , and @xmath13 , @xmath14 on the other hand , we have , for the @xmath15 quark doublets , @xmath16 for the lepton side , we introduce the family triplet , @xmath17 ( column ) , under @xmath18 .since the minimal standard model does not see the right - handed neutrinos , it would be a natural way to make an extension of the minimal standard model .we propose that neutrinos are only species seeing the family gauge sector . or , we have , for @xmath19 ( @xmath20 ) , @xmath21 and , for the left - handed @xmath18-triplet and @xmath15-doublet @xmath22 ( all columns ) ( @xmath23 ) , @xmath24 if the right - handed charged leptons were singlets under @xmath18 , then the mass - generation terms for charged leptons would involve the cross terms , such as @xmath25 , which is not acceptable at all .thus , the right - handed charged leptons have to form another triplet @xmath26 under @xmath18 . in other words ,the quark masses are given by the higgs mechanism in the minimal standard model while the masses of charged leptons are determined by @xmath27 , only a universal number in the leading - order sense .to make a reasonable theory @xcite , we have to make certain that all gauge bosons and the residual family higgs particles are very massive , i.e. @xmath28 a few tev . as slightly differing from the previous effort @xcite , we would like to write down the @xmath29 standard model _ all at once_. we introduce the neutrino mass term as follows : @xmath30 where @xmath31 are the lepton multiplets just mentioned above ( with the first number for @xmath18 and the second for @xmath15 ) .the cross ( curl ) product is somewhat new @xcite , referring to the singlet combination of three triplets in @xmath32 - an @xmath32 operation ( and not a matrix product ) .the higgs field @xmath33 is new in this effort , because it carries some nontrivial @xmath15 charge .neutrinos have masses , the tiny masses far below the range of the masses of the quarks and charged leptons .neutrinos oscillate among themselves , giving rise to a lepton - flavor violation ( lfv ) .there are other oscillation stories , such as the oscillation in the @xmath34 system , but there is a fundamental difference here - the @xmath34 system is composite while neutrinos are `` point - like '' dirac particles .it is true that neutrino masses and neutrino oscillations may be regarded as one of the most important experimental facts over the last thirty years @xcite .in fact , certain lfv processes such as @xmath35 @xcite and @xmath36 are closely related to the most cited picture of neutrino oscillations so far @xcite . in recent publications by one @xcite of us , it was pointed out that the cross - generation or off - diagonal neutrino - higgs interaction may serve as the detailed mechanism of neutrino oscillations , with some vacuum expectation value of the new higgs field(s ) . in the other words ,the first term in the last equation [ eq . ( 6 ) ] can be used as the basis to analyze the various lepton - flavor - violating decays and reactions . to illustrate the point further , we calculate the golden lepton - flavor - violating decay @xmath35 as the celebrated example .we show in figures 1(a ) , 1(b ) , and 1(c ) three leading basic feynman diagrams . herethe conversion of @xmath37 into @xmath38 is marked by a cross sign and it is a term from the off - diagonal interaction given above with the higgs vacuum expectation value @xmath39 . herethe higgs masses are assumed to be very large , i.e. , greater than a few @xmath40 , as in @xmath18 .the only small number ( coupling ) is @xmath41 , consistent with the tiny masses of neutrinos ..,width=384 ] using feynman rules from wu and hwang @xcite , we write , for fig .1(a ) , @xmath42 with @xmath43 . on the other hand ,feynman rules yield , for fig .1(b ) , @xmath44 and a similar result for fig .the four - dimensional integrations can be carried out , via the dimensional integration formulae ( e.g. ch .10 , wu / hwang @xcite ) , especially if we drop the small masses compared to the w - boson mass @xmath45 in the denominator . in this way, we obtain @xmath46 it is interesting to note that the wave - function renormalization , as shown by figs . 1(b ) and 1(c ) , yields @xmath47 noting that @xmath48 and @xmath49 would make the contribution from figs .1(b ) and 1(c ) to be of the opposite sign to that from fig .it is interesting to note that the leading terms all cancel , a result of gauge invariance .we have computed some next - order terms but a complete result seems to be rather difficult to obtain . in a normal treatment, one ignores the wave - function renormalization diagrams 1(b ) and 1(c ) in the treatment of the decays @xmath35 , @xmath50 , and @xmath51 . comparing this to the dominant mode @xmath52 @xcite, we could obtain the branching ratio . even though the decay rate for @xmath53 would be of the order @xmath54 , which is extremely small .note that the cancelation does not exist for @xmath55 , nor for the conversion processso , the rates would be expected to be much larger .the off - diagonal mass matrix would be modified by the self - energy diagram since the neutrinos form a triplet under @xmath18 .it is presumed that these self - energy diagrams , after the ultraviolet divergences get subtracted , lead to masses of the right order .if the off - diagonal mass matrix is diagonalized alone , the three roots would be two negative and one positive , adding up to zero .this seems like one ordering in the masses of neutrinos - one up and two downs . besides the golden decay @xmath53 ( much too small ) and neutrino oscillations ( already observed ) , violation of the @xmath57 universalityis also expected and might be observed . as the baryon asymmetry is sometime attributed to the lepton - antilepton asymmetry , the current scenario for neutrino oscillations @xcite seems to be inadequate .if we take the hints from neutrinos rather seriously , there are so much to discover , even though the minimal standard model for the ordinary - matter world remains to be pretty much intact .let us come back to look at the statement of the extended standard model .we choose the basic units at first and then the gauge group .the higgs mechanism would be in the last step .if that is the case , we have some difficulty in writing down the left - right model @xcite .if we need to assign a certain left - handed or right - handed spinor into two basic units simultaneously , then the kinetic term appears twice - our language does not go ; we believe that a lagrangian should have only one kinetic term .so , our first question would be : could the above rationale rule out the right - handed sector , since the simultaneous presence of the left - handed and right - handed basic units as @xmath58 doublets are excluded ? experimentally , we should check this point .as long as we could argue , we note that , as long as the left - handed and right - handed components are split in the basic units , parity has to be violated , either v - a or v+a . in a slightly different context @xcite ,it was proposed that we could work with two working rules : `` dirac similarity principle '' , based on eighty years of experience , and `` minimum higgs hypothesis '' , from the last forty years of experience .using these two working rules , the extended model mentioned above becomes rather unique - so , it is so much easier to check it against the experiments .the model stated in the paper is yet to be completed , in view of the `` minimum higgs hypothesis '' .the higgs mechanism in the previous @xmath18 family gauge theory is complete since the theory is treated _alone_. with @xmath18 and @xmath15 ( 3 , 2 ) higgs multiplet mentioned above plus one ( 3 , 1 ) higgs triplet , is it sufficient to do the higgs - mechanism job - no `` unwanted '' massless particles ?we would like to list this `` mathematical '' question as the second important question .we would be curious about how the dark - matter world looks like , though it is difficult to verify experimentally. the first question would be : the dark - matter world , 25 % of the current universe ( in comparison , only 5 % in the ordinary matter ) , would clusterize to form the dark - matter galaxies , maybe even before the ordinary - matter galaxies .the dark - matter galaxies would then play the hosts of ( visible ) ordinary - matter galaxies , like our own galaxy , the milky way .note that a dark - matter galaxy is by our definition a galaxy that does not possess any ordinary strong and electromagnetic interactions ( with our visible ordinary - matter world ) .this fundamental question deserves some thoughts , for the structural formation of our universe .of course , we should remind ourselves that , in our ordinary - matter world , those quarks can aggregate in no time , to hadrons , including nuclei , and the electrons serve to neutralize the charges also in no time. then atoms , molecules , complex molecules , and so on .these serve as the seeds for the clusters , and then stars , and then galaxies , maybe in a time span of @xmath59 ( i.e. , the age of our young universe ) .the aggregation caused by strong and electromagnetic forces is fast enough to help giving rise to galaxies in a time span of @xmath59 . on the other hand, the seeded clusterings might proceed with abundance of extra - heavy dark - matter particles such as familons and family higgs , all greater than a few @xmath40 and with relatively long lifetimes ( owing to very limited decay channels ) .so , further simulations on galactic formation and evolution may yield clues on our problem . finally , coming back to the fronts of particle physics , neutrinos , especially the right - handed neutrinos , might couple to the dark - matter particles .any further investigation along this direction would be of utmost importance .it may shed light on the nature of the dark - matter world .the minimum higgs picture is as follows : the standard - model higgs doublet @xmath60 , the purely family higgs triplet @xmath61 , and the mixed family higgs @xmath33 . the triplet @xmath61 and the neutral part @xmath62 undergoes the spontaneous symmetry breaking as in @xcite . in the u - gauge ,the neutral part @xmath63 gets projected out and the mass term becomes negative while there remains to have no ssb in the charged part @xmath64 . for details , please consult @xcite .over the years , pauchy hwang would like to thank jen - chieh peng and tony zee for numerous interactions , those , plus a lot of ( unspoken ) personal thoughts , lead to this paper .this work is supported in part by national science council project ( nsc99 - 2112-m-002 - 009-my3 ) .ta - you wu and w - y .pauchy hwang , `` relatistic quantum mechanics and quantum fields '' ( world scientific 1991 ) ; francis halzen and alan d. martin , `` quarks and leptons '' ( john wiley and sons , inc .1984 ) ; e.d .commins and p.h .bucksbaum , `` weak interactions of leptons and quarks '' ( cambridge university press 1983 ) .we use the first book for the notations and the metrics .pauchy hwang , nucl .* a844 * , 40c ( 2010 ) ; w - y .pauchy hwang , international j. mod .* a24 * , 3366 ( 2009 ) ; the idea first appeared in hep - ph , arxiv : 0808.2091 ; talk presented at 2008 cospa symposium ( pohang , korea , october 2008 ) , intern .series * 1 * , 5 ( 2011 ) ; plenary talk at the 3rd international meeting on frontiers of physics , 12 - 16 january 2009 , kuala lumpur , malaysia , published in american institute of physics 978 - 0 - 7354 - 0687 - 2/09 , pp .25 - 30 ( 2009 ) .pauchy hwang , arxiv:1207.6443v1 [ hep - ph ] 27 jul 2012 ; w - y .pauchy hwang , hyperfine interactions * 215 * , 105 ( 2013 ) ; arxiv:1207.6837v1 [ hep - ph ] 30 jul 2012 ; w - y .pauchy hwang , arxiv:1209.5488v1 [ hep - ph ] 25 sep 2012 .

<S> we would like to discuss the language to write an extended standard model - using renormalizable quantum field theory as the framework ; to start with certain basic units together with a certain gauge group . </S><S> specifically we use the left - handed and right - handed spinors to form the basic units together with @xmath0 as the gauge group . </S><S> we could write down the extended standard model , though the details of the higgs mechanism remains to be worked out . </S><S> the same general quest appeared about forty years ago - the so - called `` how to build up a model '' . </S><S> it is timely to address the same question again especially since we could now put together `` dirac similarity principle '' and `` higgs minimum hypothesis '' as two additional working rules . </S><S> pacs indices : 12.60.-i ( models beyond the standard model ) ; 98.80.bp ( origin and formation of the universe ) ; 12.10.-g ( unified field theories and models ) . * leading questions in an extended standard model * .5 cm w - y . </S><S> pauchy hwang@xmath1 12 april 2013 ; to be published in `` the universe '' . ] and tung - mow yan@xmath2 + _ @xmath3asia pacific organization / center for cosmology and particle astrophysics , + institute of astrophysics , center for theoretical sciences , + and department of physics , national taiwan university , taipei 106 , taiwan + @xmath2department of physics , cornell university , ithaca , n.y . </S><S> 14850 _ .2 cm ( december 21 , 2012 ) </S>

hep-ph0004257

the dominant weak decays of hadrons containing a heavy quark , @xmath1 or @xmath0 , are caused by the decay of the heavy quark . in the limit of a very large mass @xmath2 of a heavy quark @xmath3the parton picture of the hadron decay should set in , where the inclusive decay rates of hadrons , containing @xmath3 , mesons ( @xmath4 ) and baryons ( @xmath5 ) , are all the same and equal to the inclusive decay rate @xmath6 of the heavy quark .yet , the known inclusive decay rates @xcite are conspicuously different for different hadrons , especially for charmed hadrons , whose lifetimes span a range of more than one order of magnitude from the shortest @xmath7 ps to the longest @xmath8 ps , while the differences of lifetime among @xmath0 hadrons are substantially smaller .the relation between the relative lifetime differences for charmed and @xmath0 hadrons reflects the fact that the dependence of the inclusive decay rates on the light quark - gluon ` environment ' in a particular hadron is a pre - asymptotic effect in the parameter @xmath2 , which effect vanishes as an inverse power of @xmath2 at large mass .a theoretical framework for systematic description of the leading at @xmath9 term in the inclusive decay rate @xmath10 as well as of the terms relatively suppressed by inverse powers of @xmath2 is provided @xcite by the operator product expansion ( ope ) in @xmath11 .existing theoretical predictions for inclusive weak decay rates are in a reasonable agreement , within the expected range of uncertainty , with the data on lifetimes of charmed particles and with the so far available data on decays of @xmath12 mesons .the only outstanding piece of present experimental data is on the lifetime of the @xmath13 baryon : @xmath14 , for which ratio a theoretical prediction , given all the uncertainty involved , is unlikely to produce a number lower than 0.9 .the number of available predictions for inclusive decay rates of charmed and @xmath0 hadrons is sufficiently large for future experimental studies to firmly establish the validity status of the ope based theory of heavy hadron decays , and , in particular , to find out whether the present contradiction between the theory and the data on @xmath15 is a temporary difficulty , or an evidence of fundamental flaws in theoretical understanding .it is a matter of common knowledge that application of ope to decays of charmed and @xmath0 hadrons has potentially two caveats .one is that the ope is used in the minkowsky kinematical domain , and therefore relies on the assumption of quark - hadron duality at the energies involved in the corresponding decays . in other words, it is assumed that sufficiently many exclusive hadronic channels contribute to the inclusive rate , so that the accidentals of the low - energy resonance structure do not affect the total rates of the inclusive processes .theoretical attempts at understanding the onset of the quark - hadron duality are so far limited to model estimates @xcite , not yet suitable for direct quantitative evaluation of possible deviation from duality in charm and @xmath0 decays .this point presents the most fundamental uncertainty of the ope based approach , and presently can only be clarified by confronting theoretical predictions with experimental data .the second possible caveat in applying the ope technique to inclusive charm decays is that the mass of the charm quark , @xmath16 , may be insufficiently large for significant suppression of higher terms of the expansion in @xmath17 .the relative lightness of the charm quark , however , accounts for a qualitative , and even semi - quantitative , agreement of the ope based predictions with the observed large spread of the lifetimes of charmed hadrons : the nonperturbative effects , formally suppressed by @xmath18 and @xmath19 are comparable with the ` leading ' parton term and describe the hierarchy of the lifetimes .another uncertainty of a technical nature arises from poor knowledge of matrix elements of certain quark operators over hadron , arising as terms in ope .these can be estimated within theoretical models , with inevitable ensuing model dependence , or , where possible , extracted from the experimental data . with these reservations spelled out , we discuss here the ope based description of inclusive weak decays of charm and @xmath0 hadrons , with emphasis on specific experimentally testable predictions , and on the measurements , which would less rely on model dependence of the estimates of the matrix elements , thus allowing to probe the ope predictions at a fundamental level .the optical theorem of the scattering theory relates the total decay rate @xmath20 of a hadron @xmath21 containing a heavy quark @xmath3 to the imaginary part of the ` forward scattering amplitude ' . for the case of weakdecays the latter amplitude is described by the following effective operator l_eff=2 i m , [ leff ] in terms of which the total decay rate is given by . ]_ h = h_q | l_eff | h_q .[ lgam ] the correlator in equation ( [ leff ] ) in general is a non - local operator .however at @xmath22 the dominating space - time intervals in the integral are of order @xmath11 and one can expand the correlator in @xmath23 , thus producing an expansion in inverse powers of @xmath2 .the leading term in this expansion describes the parton decay rate of the quark .for instance , the term in the non - leptonic weak lagrangian @xmath24 with @xmath25 being the appropriate combination of the ckm mixing factors , generates through eq.([leff ] ) the leading term in the effective lagrangian l^(0)_eff , nl = |v|^2 g_f^2 m_q^5 64 ^3 _ nl ( q q ) , [ lef0 ] where @xmath26 is the perturbative qcd radiative correction factor .this expression reproduces the well known formula for the inclusive non - leptonic decay rate of a heavy quark , associated with the underlying process @xmath27 , due to the relation @xmath28 , which is valid up to corrections of order @xmath29 .one also sees form this example , that in order to separate individual semi - inclusive decay channels , e.g. non - leptonic with specific flavor quantum numbers , or semi - leptonic , one should simply pick up the corresponding relevant part of the weak lagrangian @xmath30 , describing the underlying process , to include in the correlator ( [ leff ] ) .the general expression for first three terms in the ope for @xmath31 has the form @xmath32 where the superscripts denote the power of @xmath11 in the relative suppression of the corresponding term in the expansion with respect to the leading one , @xmath33 is the gluon field tensor , @xmath34 stand for light quarks , @xmath35 , and , finally , @xmath36 , @xmath37 denote spin and color structures of the four - quark operators .the coefficients @xmath38 depend on the specific part of the weak interaction lagrangian @xmath30 , describing the relevant underlying quark process .one can notice the absence in the expansion ( [ first3 ] ) of a term suppressed by just one power of @xmath11 , due to non - existence of operators of suitable dimension .thus the decay rates receive no correction of relative order @xmath11 in the limit of large @xmath2 , and the first pre - asymptotic corrections appear only in the order @xmath29 . the mechanisms giving rise to the three discussed terms in ope are shown in figure 1 .the first , leading term corresponds to the parton decay , and does not depend on the light guark and gluon ` environment ' of the heavy quark in a hadron .the second term describes the effect on the decay rate of the gluon field that a heavy quark ` sees ' in a hadron .this term in fact is sensitive only to the chromomagnetic part of the gluon field , and contains the operator of the interaction of heavy quark chromomagnetic moment with the chromomagnetic field .thus this term depends on the spin of the heavy quark , but does not depend on the flavors of the light quarks or antiquarks .therefore this effect does not split the inclusive decay rates within flavor su(3 ) multiplets of heavy hadrons , but generally gives difference of the rates , say , between mesons and baryons .the dependence on the light quark flavor arises from the third term in the expansion ( [ first3 ] ) which explicitly contains light quark fields .historically , this part is interpreted in terms of two mechanisms @xcite : the weak scattering ( ws ) and the pauli interference ( pi ) . the ws corresponds to a cross - channel of the underlying decay , generically @xmath39 , where either the quark @xmath40 is a spectator in a baryon and can undergo a weak scattering off the heavy quark : @xmath41 , or an antiquark in meson , say @xmath42 , weak - scatters ( annihilates ) in the process @xmath43 .the pauli interference effect arises when one of the final ( anti)quarks in the decay of @xmath3 is identical to the spectator ( anti)quark in the hadron , so that an interference of identical particles should be taken into account .the latter interference can be either constructive or destructive , depending on the relative spin - color arrangement of the ( anti)quark produced in the decay and of the spectator one , thus the sign of the pi effect is found only as a result of specific dynamical calculation . in specific calculations , however , ws and pi arise from the same terms in ope , depending on the hadron discussed , and technically there is no need to resort to the traditional terminology of ws and pi . 0.9 mm ( 140.00,84.00)(10,0 ) ( 10.00,70.00)(1,0)20.00 ( 30.00,70.00)(1,0)30.00 ( 60.00,70.00)(1,0)20.00 ( 30.00,70.00)(45.00,84.00)(60.00,70.00 ) ( 30.00,70.00)(45.00,56.00)(60.00,70.00 ) ( 30.00,70.00 ) ( 60.00,70.00 ) ( 20.00,72.00)(0,0)[cb]@xmath3 ( 70.00,71.00)(0,0)[cb]@xmath3 ( 49.00,78.00)(0,0)[cb]@xmath44 ( 49.00,71.00)(0,0)[cb]@xmath45 ( 49.00,62.00)(0,0)[ct]@xmath46 ( 10.00,43.00)(1,0)20.00 ( 30.00,43.00)(1,0)30.00 ( 60.00,43.00)(1,0)20.00 ( 30.00,43.00)(45.00,57.00)(60.00,43.00 ) ( 30.00,43.00)(45.00,29.00)(60.00,43.00 ) ( 30.00,43.00 ) ( 60.00,43.00 ) ( 20.00,45.00)(0,0)[cb]@xmath3 ( 70.00,44.00)(0,0)[cb]@xmath3 ( 49.00,51.00)(0,0)[cb]@xmath44 ( 49.00,44.00)(0,0)[cb]@xmath45 ( 49.00,35.00)(0,0)[ct]@xmath46 ( 45.00,36.00)(0,-1)2.00 ( 45.00,33.00)(0,-1)2.00 ( 45.00,30.00)(0,-1)2.00 ( 47.00,28.00)(0,0)[cc]@xmath47 ( 10.00,12.00)(1,0)20.00 ( 30.00,12.00)(1,0)30.00 ( 60.00,12.00)(1,0)20.00 ( 30.00,12.00)(45.00,26.00)(60.00,12.00 ) ( 30.00,12.00 ) ( 60.00,12.00 ) ( 20.00,14.00)(0,0)[cb]@xmath3 ( 70.00,13.00)(0,0)[cb]@xmath3 ( 49.00,20.00)(0,0)[cb]@xmath44 ( 49.00,13.00)(0,0)[cb]@xmath45 ( 10.00,5.50)(3,1)20.00 ( 60.00,12.00)(4,-1)20.00 ( 20.00,7.00)(0,0)[ct]@xmath40 ( 70.00,8.00)(0,0)[ct]@xmath40 ( 87.00,70.00)(1,0)11.00 ( 87.00,43.00)(1,0)11.00 ( 87.00,12.00)(1,0)11.00 ( 110.00,70.00)(1,0)30.00 ( 116.00,71.00)(0,0)[cb]@xmath3 ( 134.00,71.00)(0,0)[cb]@xmath3 ( 110.00,43.00)(1,0)30.00 ( 116.00,44.00)(0,0)[cb]@xmath3 ( 134.00,44.00)(0,0)[cb]@xmath3 ( 125.00,43.00)(0,-1)2.00 ( 125.00,40.00)(0,-1)2.00 ( 125.00,37.00)(0,-1)2.00 ( 127.00,35.00)(0,0)[cc]@xmath47 ( 110.00,12.00)(1,0)30.00 ( 116.00,13.00)(0,0)[cb]@xmath3 ( 134.00,13.00)(0,0)[cb]@xmath3 ( 110.00,7.00)(3,1)15.00 ( 125.00,12.00)(3,-1)15.00 ( 117.00,9.00)(0,0)[lt]@xmath48 ( 132.00,9.00)(0,0)[rt]@xmath48 ( 125.00,62.00)(0,0)[cc ] @xmath49 ( 125.00,27.00)(0,0)[cc ] @xmath50 ( 125.00,0.00)(0,0)[cc ] @xmath51 ( 123.50,70.00 ) ( 123.50,43.00 ) ( 123.50,12.00 ) [ fig : ope ] in what follows we discuss separately the effects of the three terms in the expansion ( [ first3 ] ) and their interpretation within the existing and future data .the leading term in the ope amounts to the perturbative expression for the decay rate of a heavy quark . in @xmath0 hadronsthe contribution of the subsequent terms in ope is at the level of few percent , so that the perturbative part can be confronted with the data in its own right . in particular , for the @xmath52 meson the higher terms in ope contribute only about 1% of the total non - leptonic as well as of the semileptonic decay rate .thus the data on these rates can be directly compared with the leading perturbative term in ope .the principal theoretical topic , associated with this term is the calculation of qcd radiative corrections , i.e. of the factor @xmath26 in eq.([lef0 ] ) and of a similar factor , @xmath26 , for semileptonic decays .it should be noted , that even at this , perturbative , level there is a known long - standing problem between the existing data and the theory in that the current world average for the semileptonic branching ratio for the @xmath12 mesons , @xmath53 , is somewhat lower than the value @xmath54 preferred from the present knowledge of theoretical qcd radiative corrections to the ratio of non - leptonic to semileptonic decay rates ( see e.g. @xcite ) .however , this apparent discrepancy may in fact be due to insufficient ` depth ' of perturbative qcd calculation of the ratio @xmath55 . in order to briefly elaborate on this point , we notice that the standard way of analyzing the perturbative radiative corrections in the nonleptonic decays is through the renormalization group ( rg ) summation of the leading log terms and the first next - to - leading terms @xcite in the parameter @xmath56 . forthe semileptonic decays the logarithmic dependence on @xmath57 is absent in all orders due to the weak current conservation at momenta larger than @xmath58 , thus the correction is calculated by the standard perturbative technique , and a complete expression in the first order in @xmath59 is available both for the total rate @xcite and for the lepton spectrum @xcite . in realityhowever the parameter @xmath60 is not large , and non - logarithmic terms may well compete with the logarithmic ones .this behavior is already seen from the known expression for the logarithmic terms : when expanded up to the order @xmath61 the result of ref.@xcite for the rate of decays with single final charmed quark takes the form = 1 + _ s + _s^2 ^2 , [ as2l ] where , in terms of notation of ref.@xcite , @xmath62 .the behavior of the function @xmath63 is known explicitly @xcite and is quite weak : @xmath64 , @xmath65 , and @xmath66 for the realistic mass ratio @xmath67 .one can see that the term with the single logarithm @xmath68 contributes about two thirds of that with @xmath69 in the term quadratic in @xmath59. under such circumstances the rg summation of the terms with powers of @xmath68 does not look satisfactory for numerical estimates of the qcd effects , at least at the so far considered level of the first next - to - leading order terms , and the next - to - next - to - leading terms can be equally important as the two known ones , which would eliminate the existing impasse between the theory and the data on @xmath70 .one can present some arguments @xcite that this is indeed the case for the @xmath0 quark decay , although a complete calculation of these corrections is still unavailable .the corrections suppressed by two powers of @xmath11 to inclusive decay rates arise from two sources @xcite : the @xmath71 corrections to the matrix element of the leading operator , @xmath72 , and the second term in ope ( [ first3 ] ) containing the chromomagnetic interaction .the expression for the matrix element of the leading operator with the correction included is written in the form h_q| qq | h_q = 1- _ ^2(h_q)-_g^2(h_q ) 2 m_q^2 + , [ qbarq ] where @xmath73 and @xmath74 are defined as @xmath75 with @xmath76 being the qcd covariant derivative .the correction in equation ( [ qbarq ] ) in fact corresponds to the time dilation factor @xmath77 , for the heavy quark decaying inside a hadron , where it has energy @xmath78 , which energy is contributed by the kinetic part ( @xmath79 ) and the chromomagnetic part ( @xmath80 ) .the second term in ope describes the effect of the chromomagnetic interaction in the decay process , and is also expressed through @xmath74 .the explicit formulas for the decay rates , including the effects up to the order @xmath29 are found in @xcite and for decays of the @xmath0 hadrons read as follows .for the semileptonic decay rate _sl(h_b)=|v_cb|^2 g_f^2 m_b^5 192 ^3 h_b | b b |h_b _ sl i_0 ( x,0,0 ) , [ gamsl1 ] and for the non - leptonic decay rate _nl(h_b)=|v_cb|^2 g_f^2 m_b^5 64 ^3 h_b | b b| h_b \{_nl i ( x)-8 _ 2 _ g^2 m_b^2 i_2(x ) } . [ gamnl1 ] these formulas take into account only the dominant ckm mixing @xmath81 and neglect the small one , @xmath82 .the following notation is also used : @xmath83 , @xmath84 stands for the kinematical suppression factor in a three - body weak decay due to masses of the final fermions .in particular , @xmath85 furthermore , @xmath86 , and @xmath87 finally , the qcd radiative correction factor @xmath88 in eq.([gamnl1 ] ) is known in the leading logarithmic approximation and is expressed in terms of the well known coefficients @xmath89 and @xmath90 in the renormalization of the non - leptonic weak interaction: @xmath91 with c_-()=c_+^-2()=^4/b , [ cpm ] and @xmath0 is the coefficient in the qcd beta function . the value of @xmath0 relevant to @xmath0 decays is @xmath92 .numerically , for @xmath93 , the expressions for the decay rates can be written as @xmath94 where @xmath95 is the perturbation theory value of the corresponding decay rate of @xmath0 quark .the matrix elements @xmath73 and @xmath74 are related to the spectroscopic formula for a heavy hadron mass @xmath96 , m(h_q)=m_q + ( h_q)+_^2(h_q)-_g^2(h_q ) 2 m_q + [ massf ] being combined with the spin counting for pseudoscalar and vector mesons , this formula allows to find the value of @xmath74 in pseudoscalar mesons from the mass splitting : _g^2(b)=3 4 ( m_b^*^2-m_b^2 ) 0.36 gev^2 .[ mugn ] the value of @xmath73 for @xmath12 mesons is less certain .it is constrained by the inequality @xcite , @xmath97 , and there are theoretical estimates from the qcd sum rules @xcite : @xmath98 and from an analysis of spectroscopy of heavy hadrons @xcite : @xmath99 . in any event ,the discussed corrections are rather small for @xmath0 hadrons , given that @xmath100 .the largest , in relative terms , effect of these corrections in @xmath12 meson decays is on the semileptonic decay rate , where it amounts to 4 5 % suppression of the rate , which rate however is only a moderate fraction of the total width . in the dominant non - leptonic decay ratethe effect is smaller , and , according to the formula ( [ gamnum ] ) amounts to about 1.5 2 % .the effect of the @xmath29 corrections can be evaluated with a somewhat better certainty for the ratio of the decay rates of @xmath13 and @xmath12 mesons .this is due to the fact that @xmath101 , since there is no correlation of the spin of the heavy quark in @xmath13 with the light component , having overall quantum numbers @xmath102 .then , applying the formula ( [ gamnum ] ) to @xmath12 and @xmath13 , we find for the ratio of the ( dominant ) non - leptonic decay rates : = 1 -_^2(_b)-_^2(b ) 2 m_b^2 + 0.5 _ g^2(b ) m_b^2 .[ lambdab ] the difference of the kinetic terms , @xmath103 , can be estimated from the mass formula : _ ^2(_b)-_^2(b)=2 m_b m_c m_b - m_c = 0 0.04 gev^2 , [ difmu ] where @xmath104 is the spin - averaged mass of the mesons , e.g. @xmath105 .the estimated difference of the kinetic terms is remarkably small .thus the effect in the ratio of the decay rates essentially reduces to the chromomagnetic term , which is also rather small and accounts for less than 1% difference of the rates . for the ratio of the semileptonic decay rates the chromomagnetic term is approximately four times larger , butthen the contribution of the semileptonic rates to the total width is rather small .thus one concludes that the terms of order @xmath106 in the ope expansion for the decay rates can account only for about 1% difference of the lifetimes of @xmath13 and the @xmath12 mesons .the significance of the @xmath29 terms is substantially different for the decay rates of charmed hadrons , where these effects suppress the inclusive decays of the @xmath76 mesons by about 40% with respect to those of the charmed hyperons in a reasonable agreement with the observed pattern of the lifetimes .it should be emphasized once again that the @xmath29 effects do not depend on the flavors of the spectator quarks or antiquarks .thus the explanation of the variety of the inclusive decay rates within the flavor su(3 ) multiplets , observed for charmed hadrons and expected for the @xmath0 ones , has to be sought among the @xmath107 terms .although the third term in the expansion ( [ first3 ] ) is formally suppressed by an extra power of @xmath11 , its effects are comparable to , or even larger than the effects of the second term .this is due to the fact that the diagrams determining the third term ( see fig .1 ) contain a two - body phase space , while the first two terms involve a three - body phase space .this brings in a numerical enhancement factor , typically @xmath109 .the enhanced numerical significance of the third term in ope , generally , does not signal a poor convergence of the expansion in inverse heavy quark mass for decays of @xmath0 , and even charmed , hadrons the numerical enhancement factor is a one time occurrence in the series , and there is no reason for similar ` anomalous ' enhancement among the higher terms in the expansion . herewe first present the expressions for the relevant parts of @xmath108 for decays of @xmath0 and @xmath1 hadrons in the form of four - quark operators and then proceed to a discussion of hadronic matrix elements and the effects in specific inclusive decay rates .the consideration of the effects in decays of charmed hadrons is interesting in its own right , and leads to new predictions to be tested experimentally , and is also important for understanding the magnitude of the involved matrix elements using the existing data on charm decays .we start with considering the term @xmath108 in @xmath0 hadron non - leptonic decays , @xmath110 , induced by the underlying processes @xmath111 , @xmath112 , @xmath113 , and @xmath114 .unlike the case of three - body decay , the kinematical difference between the two - body states @xmath115 and @xmath116 , involved in calculation of @xmath110 is of the order of @xmath117 and is rather small . at present level of accuracy in discussing this term in ope, one can safely neglect the effect of finite charmed quark mass . in this approximation the expression for @xmath110 reads as @xcite @xmath118 \right \}~,\end{aligned}\ ] ] where the notation @xmath119 is used , the indices @xmath120 are the color triplet ones , @xmath121 , and @xmath122 is the color current of the light quarks with @xmath123 being the generators of the color su(3 ) .the notation @xmath124 , is used as shorthand for the short - distance renormalization coefficients @xmath125 at @xmath126 : @xmath127 . the expression ( [ l3nlb ] ) is written in the leading logarithmic approximation for the qcd radiative effects in a low normalization point @xmath128 such that @xmath129 ( but still , at least formally , @xmath130 ) . for such @xmath128 there arises so called ` hybrid ' renormalization @xcite , depending on the factor @xmath131 .the coefficients @xmath132 with @xmath133 in eq.([l3nlb ] ) have the following explicit expressions in terms of @xmath124 and @xmath134 : @xmath135 ~ , \nonumber \\ & & { \tilde c}_4=-{1 \over 4 } \ , \kappa^{1/2 } \ , ( 5{\tilde c}_+^2+{\tilde c}_-^2 + 6{\tilde c}_+{\tilde c}_-)~ , \nonumber \\ & & { \tilde c}_5=-{1 \over 4 } \ , \left [ ( { \tilde c}_++{\tildec}_-)^2 + { 1 \over 3 } ( 1-\kappa^{1/2 } ) ( 5{\tilde c}_+^2+{\tilde c}_-^2 - 6{\tilde c}_+{\tilde c}_- ) \right]~ , \nonumber \\ & & { \tilde c}_6=-{1 \over 4 } \ , \kappa^{1/2 } \ , ( 5{\tilde c}_+^2+{\tilde c}_-^2 - 6{\tilde c}_+{\tilde c}_-)~. \label{coefs}\end{aligned}\ ] ] the expression for the ckm dominant semileptonic decays of @xmath0 hadrons , associated with the elementary process @xmath136 does not look to be of an immediate interest .the reason is that this process is intrinsically symmetric under the flavor su(3 ) , and one expects no significant splitting of the semileptonic decay rates within su(3 ) multiplets of the @xmath0 hadrons. the only possible effect of this term , arising through a penguin - like mechanism can be in a small overall shift of semileptonic decay rates between @xmath12 mesons and baryons . however , these effects are quite suppressed and are believed to be even smaller than the ones arising form the discussed @xmath106 terms. for charm decays there is a larger , than for @xmath0 hadrons , variety of effects associated with @xmath108 , that can be studied experimentally , and we present here the relevant parts of the effective lagrangian . for the ckmdominant non - leptonic decays of charm , originating from the quark process @xmath137 , the discussed term in ope has the form @xmath138 \right \}~,\end{aligned}\ ] ] where , @xmath139 is the cabibbo angle , and the coefficients without the tilde are given by the same expressions as above for the @xmath0 decays ( i.e. those with tilde ) with the replacement @xmath140 .the part of the notation in the superscript @xmath141 points to the selection rule for the dominant ckm unsuppressed non - leptonic decays .one can rather realistically envisage however a future study of inclusive rates for the once ckm suppressed decays of charmed hadrons , satisfying the selection rule @xmath142 and associated with the quark processes @xmath143 and @xmath144 .the corresponding part of the effective lagrangian for these processes reads as @xmath145 \right \}~ \label{l3nl1}\end{aligned}\ ] ] where again the notation @xmath119 is used .the semileptonic decays of charm , the ckm dominant , associated with @xmath146 , and the ckm suppressed , originating from @xmath147 , contribute to the semileptonic decay rate , which certainly can be measured experimentally .the expression for the part of the effective lagrangian , describing the @xmath107 terms in these decays is @xcite @xmath148 + \right .\nonumber \\ & & \sin^2 \theta_c \ , \left [ l_1\ , ( \overline c \gamma_\mu c + { 2 \over 3}\overline c \gamma_\mu \gamma_5c ) ( \overline d \gamma_\mu d)+l_2 \ , ( \overline c_i \gamma_\mu c_k + { 2 \over 3}\overline c_i \gamma_\mu \gamma_5 c_k ) ( \overline d_k \gamma_\mu d_i ) \right ] - \nonumber \\ & & \left . 2 \ , \kappa^{1/2 } \ , ( \kappa^{-2/9}-1 ) \ , ( \overline c \gamma_\mu t^a c + { 2 \over 3}\overline c \gamma_\mu \gamma_5 t^ac ) j_\mu^a \right \ } ~ , \label{l3sl}\end{aligned}\ ] ] with the coefficients @xmath149 and @xmath150 found as l_1=(^1/2 - 1 ) , l_2 = - 3 ^1/2 .the expressions for the terms in @xmath108 still leave us with the problem of evaluating the matrix elements of the four - quark operators over heavy hadrons in order to calculate the effects in the decay rates according to the formula ( [ lgam ] ) . in doing so only few conclusions can be drawn in a reasonably model independent way , i.e. without resorting to evaluation of the matrix elements using specific ideas about the dynamics of quarks inside hadrons .the most straightforward prediction can in fact be found for @xmath0 hadrons .namely , one can notice that the operator ( [ l3nlb ] ) is symmetric under the flavor u spin ( an su(2 ) subgroup of the flavor su(3 ) , which mixes @xmath151 and @xmath152 quarks ) .this is a direct consequence of neglecting the small kinematical effect of the charmed quark mass .however the usual ( in)accuracy of the flavor su(3 ) symmetry is likely to be a more limiting factor for the accuracy of this symmetry , than the corrections of order @xmath153 .modulo this reservation the immediate prediction from this symmetry is the degeneracy of inclusive decay rates within u spin doublets : ( b_d)=(b_s ) , ( _ b)=(_b^0 ) , [ upred ] where @xmath154 stands for the average rate over the two eigenstates of the @xmath155 oscillations. the data on decay rates of the cascade hyperon @xmath156 are not yet available , while the currently measured lifetimes of @xmath52 and @xmath157 are within less than 2% from one another .theoretically , the difference of the lifetimes , associated with possible violation of the su(3 ) symmetry and with breaking of the u symmetry of the effective lagrangian ( [ l3nlb ] ) , is expected to not exceed about 1% . for the non - vanishing matrix elements of four - quark operators over pseudoscalar mesonsone traditionally starts with the factorization formula and parametrizes possible deviation from factorization in terms of ` bag constants ' . within the normalization convention adopted here the relations used in this parametrization read as @xmath158 where @xmath159 stands for pseudoscalar meson made of @xmath3 and @xmath160 , @xmath161 is the annihilation constant for the meson , and @xmath12 and @xmath162 are bag constants .the parameters @xmath12 and @xmath163 generally depend on the normalization point @xmath128 for the operators , and this dependence is compensated by the @xmath128 dependence of the coefficients in @xmath108 , so that the results for the physical decay rate difference do not depend on @xmath128 .if the normalization point @xmath128 is chosen at the heavy quark mass ( i.e. @xmath126 for @xmath12 mesons , and @xmath164 for @xmath76 mesons ) the predictions for the difference of total decay rates take a simple form in terms of the corresponding bag constants ( generally different between @xmath12 and @xmath76 ) : @xmath165 \nonumber \\ & \approx & - 0.025 \ , \left ( { f_b \over 200 \ , mev } \right ) ^2 \ , ps^{-1}~ , \label{bpred}\end{aligned}\ ] ] @xmath166 \nonumber \\ & \sim & - 0.8 \ , \left ( { f_d \over 200 \ , mev } \right ) ^2 \ , ps^{-1}~ , \label{dpred}\end{aligned}\ ] ] where the numerical values are written in the approximation of exact factorization : @xmath167 , and @xmath168 .it is seen from the numerical estimates that , even given all the theoretical uncertainties , the presented approach is in reasonable agreement with the data on the lifetimes of @xmath76 and @xmath12 mesons . in particular , this approach describes , at least qualitatively , the strong suppression of the decay rate of @xmath169 mesons relative to @xmath170 , the experimental observation of which has in fact triggered in early 80-s the theoretical study of preasymptotic in heavy quark mass effects in inclusive decay rates . for the @xmath12 mesons the estimate ( [ bpred ] ) is also in a reasonable agreement with the current data for the discussed difference ( @xmath171 ) .the weakly decaying heavy hyperons , containing either @xmath1 or @xmath0 quark are : @xmath172 , and @xmath173 .the first three baryons form an su(3 ) ( anti)triplet .the light diquark in all three is in the state with quantum numbers @xmath174 , so that there is no correlation of the spin of the heavy quark with the light component of the baryon . on the contrary , in @xmath175 the two strange quarks form a @xmath176 state , anda correlation between the spins of heavy and light quarks is present .the absence of spin correlation for the heavy quark in the triplet of hyperons somewhat reduces the number of independent four - quark operators , having nonvanishing diagonal matrix elements over these baryons .indeed , the operators entering @xmath108 contain both vector and axial bilinear forms for the heavy quarks. however the axial part requires a correlation of the heavy quark spin with that of a light quark , and is thus vanishing for the hyperons in the triplet .therefore only the structures with vector currents are relevant for these hyperons .these structures are of the type @xmath177 and @xmath178 with @xmath48 being @xmath152 , @xmath151 or @xmath179 .the flavor su(3 ) symmetry then allows to express , for each of the two color combinations , the matrix elements of three different operators , corresponding to three flavors of @xmath48 , over the baryons in the triplet in terms of only two combinations : flavor octet and flavor singlet .thus all effects of @xmath108 in the triplet of the baryons can be expressed in terms of four independent combinations of matrix elements .these can be chosen in the following way : @xmath180 \right \rangle_{\xi_q^{(d)}-\lambda_q } = \left \langle { 1 \over 2 } \ , ( \overline q \ , \gamma_\mu \ , q ) \left [ ( \overline s \ , \gamma_\mu s ) - ( \overline d \ , \gamma_\mu d ) \right ] \right \rangle_{\lambda_q - \xi_q^{(u)}}~ , \\\nonumber y&=&\left \langle { 1 \over 2 } \ , ( \overline q_i \ , \gamma_\mu \ , q_k ) \left [ ( \overline u_k \ , \gamma_\mu u_i ) - ( \overline s_k \ , \gamma_\mu s_i ) \right ] \right \rangle_{\xi_q^{(d)}-\lambda_q } = \left \langle { 1 \over 2 } \ , ( \overline q_i \ , \gamma_\mu \ , q_k ) \left [ ( \overline s_k \ , \gamma_\mu s_i ) - ( \overline d_k \ , \gamma_\mu d_i ) \right ] \right \rangle_{\lambda_q - \xi_q^{(u)}}~,\end{aligned}\ ] ] with the notation for the differences of the matrix elements : @xmath181 , for the flavor octet part and the matrix elements : @xmath182 for the flavor singlet part , where @xmath21 stands for any heavy hyperon in the ( anti)triplet . the initial , very approximate , theoretical estimates of the matrix elements @xcite were essentially based on a non - relativistic constituent quark model , where these matrix elements are proportional to the density of a light quark at the location of the heavy one , i.e. in terms of the wave function , proportional to @xmath183 .using then the same picture for the matrix elements over pseudoscalar mesons , relating the quantity @xmath183 to the annihilation constant @xmath161 , and assuming that @xmath183 is approximately the same in baryons as in mesons , one arrived at the estimate y =- x = x_s =- y_s 0.006 gev^2 , [ xyold ] where the sign relation between @xmath23 and @xmath184 is inferred from the color antisymmetry of the constituent quark wave function for baryons . since the constituent picture was believed to be valid at distances of the order of the hadron size , the estimate ( [ xyold ] )was applied to the matrix elements in a low normalization point where @xmath185 . for the matrix elements of the operators , containing @xmath151 quarks over the @xmath175 hyperon , this picture predicts an enhancement factor due to the spin correlation : _ q | ( q _ q ) ( s _ s ) |_q = - _ q | ( q_i _ q_k ) ( s_k _ s_i ) |_q = 10 3 y [ omegaold ] although these simple estimates allowed to correctly predict @xcite the hierarchy of lifetimes of charmed hadrons prior to establishing this hierarchy experimentally , they fail to quantitatively predict the differences of lifetimes of charmed baryons .we shall see that the available data indicate that the color antisymmetry relation is badly broken , and the absolute value of the matrix elements is larger , than the naive estimate ( [ xyold ] ) , especially for the quantity @xmath23 .it should be emphasized that in the heavy quark limit the matrix elements ( [ defxy ] ) and ( [ xys ] ) do not depend on the flavor of the heavy quark , provided that the same normalization point @xmath128 is used .therefore , applying the ope formulas to both charmed and @xmath0 baryons , one can extract the values for the matrix elements from available data on charmed hadrons , and then make predictions for @xmath0 baryons , as well as for other inclusive decay rates , e.g. semileptonic , for charmed hyperons .the only data available so far , which would allow to extract the matrix elements , are on the lifetimes of charmed hyperons .therefore , one has to take into account several essential types of inclusive decay , at least those that contribute to the total decay rate at the level of about 10% .here we first concentrate on the differences of the decay rates within the su(3 ) triplet of the hyperons , which will allow us to extract the non - singlet quantities @xmath23 and @xmath184 , and then discuss the su(3 ) singlet shifts of the rates .the differences of the dominant cabibbo unsuppressed non - leptonic decay rates are given by @xmath186~ , \nonumber \\\delta_2^{nl , \,0 } \equiv \gamma^{nl}_{\delta s = \delta c}(\lambda_c)-\gamma^{nl}_{\delta s = \delta c}(\xi_c^+ ) = \cos^4 \theta_c \ , { g_f^2 \ , m_c^2 \over 4 \pi } \left [ ( c_3 - c_1 ) \ , x + ( c_4 - c_2 ) \ , y \right ] ~. \label{dnl0}\end{aligned}\ ] ] the once cabibbo suppressed decay rates of @xmath187 and @xmath188 are equal , due to the @xmath189 property of the corresponding effective lagrangian @xmath190 ( eq.([l3nl1 ] ) ) .thus the only difference for these decays in the baryon triplet is ^nl,1 ^nl_s = 0 ( _ c^0)-^nl_s = 0 ( _ c ) = ^2 _ c ^2 _ c g_f^2 m_c^2 4 .[ dnl1 ] the dominant semileptonic decay rates are equal among the two @xmath191 baryons due to the isotopic spin property @xmath192 of the corresponding interaction lagrangian , thus there is only one non - trivial splitting for these decays : ^sl,0 ^sl_s = -1(_c ) - ^sl_s = -1 ( _ c ) = - ^2 _ c g_f^2 m_c^2 12 .[ dsl0 ] finally , the cabibbo suppressed semileptonic decay rates are equal for @xmath187 and @xmath193 , due to the @xmath194 property of the corresponding interaction .thus the only difference for these is ^sl,1 ^sl_s = 0(_c ) - ^sl_s = 0(_c^+ ) = - ^2 _ c g_f^2 m_c^2 12 . [ dsl1 ]using the relations ( [ dnl0 ] ) - ( [ dsl1 ] ) on can find expressions for two differences of the measured total decay rates , @xmath195 and @xmath196 , in terms of the quantities @xmath23 and @xmath184 : @xmath197 + \right .\nonumber \\ & \left ., \left [ \cos^2 \theta \ , ( c_6-c_4 ) + \sin^2 \theta \ , ( 2 \ , c_6 - c_2 -c_4)- { 2 \over 3 } \ , l_2\right ] \right \}~ , \label{d1}\end{aligned}\ ] ] and @xmath198 + \right .\nonumber \\ & \left .y \left [ cos^4 \theta \ , ( c_4-c_2 ) + { 2 \over 3 } \ , ( \cos^2 \theta - \sin^2 \theta ) \ , l_2 \right ] \right \}~ . \label{d2}\end{aligned}\ ] ] by comparing these relations with the data , one can extract the values of @xmath23 and @xmath184 . using the current data for the total decay rates : @xmath199 , @xmath200 , and the updated value @xcite @xmath201, we find for the @xmath128 independent matrix element @xmath23 x = -(0.04 0.01 ) gev^3 ( 1.4 gev m_c ) ^2 , [ resx ] while the dependence of the thus extracted matrix element @xmath184 on the normalization point @xmath128 is shown in fig .2 , @xmath202 , are shown only for illustrative purpose .the coefficients in the ope , leading to the equations ( [ d1],[d2 ] ) , are purely perturbative .thus , formally , they correspond to @xmath203 , i.e. to @xmath204 . ] .[ fig : xy ] notably , the extracted values of @xmath23 and @xmath184 are in a drastic variance with the simplistic constituent model : the color antisymmetry relation , @xmath205 , does not hold at any reasonable @xmath128 , and the absolute value of @xmath23 is substantially enhanced once the non - singlet matrix elements are determined , they can be used for predicting differences of other inclusive decay rates within the triplet of charmed hyperons as well as for the @xmath0 baryons .due to correlation of errors in @xmath23 and @xmath184 it makes more sense to express the predictions directly in terms of the total decay rates of the charmed hyperons .the thus arising relations between the rates do not depend on the normalization parameter @xmath128 . in this way one finds @xcite for the difference of the cabibbo dominant semileptonic decay rates between either of the @xmath191 hyperons and @xmath187 : _sl(_c)-_sl(_c ) ^sl , 0= 0.13 _ 1 - 0.065 _ 2 0.59 0.32 ps^-1 .[ dslr ] when compared with the data on the total semileptonic decay rate of @xmath187 , @xmath206 , this prediction implies that the semileptonic decay rate of the charmed cascade hyperons can be 23 times larger than that of @xmath187 .the predictions found in a similar way for the inclusive cabibbo suppressed decay rates are @xcite : for non - leptonic decays ^nl,1=0.082 _ 1 + 0.054 _ 2 0.55 0.22 ps^-1 [ dnl1r ] and for the semileptonic ones ^sl,1=^2 _ c ^sl , 0 0.030 0.016 ps^-1 .[ dsl1r ] for the only difference of the inclusive rates in the triplet of @xmath0 baryons , @xmath207 , one finds an expression in terms of @xmath23 and @xmath184 , or alternatively , in terms of the differences @xmath208 and @xmath209 between the charmed hyperons , @xmath210 \approx \nonumber \\ & & |v_{bc}|^2 \ , { m_b^2 \over m_c^2 } \ , ( 0.85 \ , \delta_1 + 0.91 \ , \delta_2 ) \approx 0.015 \ , \delta_1 + 0.016 \ , \delta_2 \approx 0.11 \pm 0.03 \ , ps^{-1}~. \label{dbres}\end{aligned}\ ] ] when compared with the data on the total decay rate of @xmath13 this result predicts about 14% longer lifetime of @xmath211 than that of @xmath13 .the singlet matrix elements @xmath212 and @xmath213 ( cf .eq.([xys ] ) ) are related to the shift of the average decay rate of the hyperons in the triplet : _q= 1 3 ( ( _ q ) + ( _ q^1)+(_q^2 ) ) .[ avgam ] for the charmed baryons the shift of the dominant non - leptonic decay rate is given by @xcite _nl^(3,0 ) _ c=^4 g_f^2 m_c^2 8 ( c_+^2 + c_-^2 ) ^5/18 ( x_s-3 y_s ) , [ d3gc ] while for the @xmath0 baryons the corresponding expression reads as ^(3 ) _b= |v_bc|^2 g_f^2 m_b^2 8 ( c_+ - c_-)^2 ^5/18 ( x_s-3 y_s ) .[ d3 gb ] the combination @xmath214 of the su(3 ) singlet matrix elements cancels in the ratio of the shifts for @xmath0 hyperons and the charmed ones : ^(3 ) _b= |v_bc|^2 ^4 m_b^2 m_c^2 ( c_+ - c_-)^2 c_+^2 + c_-^2 ^5/18 _ nl^(3,0 ) _ c 0.0025 _ nl^(3,0 )[ rbc ] ( one can observe , with satisfaction , that the dependence on the unphysical parameter @xmath128 cancels out , as it should . ) this equation shows that relatively to the charmed baryons the shift of the decay rates in the @xmath0 baryon triplet is greatly suppressed by the ratio @xmath215 , which parametrically is of the second order in @xmath59 , and numerically is only about 0.12 .an estimate of @xmath216 from eq.([rbc ] ) in absolute terms depends on evaluating the average shift @xmath217 for charmed baryons .the latter shift can be conservatively bounded from above by the average total decay rate of those baryons : @xmath218 , which then yields , using eq.([rbc ] ) , an upper bound @xmath219 .more realistically , one should subtract from the total average width @xmath220 the contribution of the ` parton ' term , which can be estimated from the decay rate of @xmath221 with account of the @xmath222 effects , as amounting to about @xmath223 .( one should also take into account the semileptonic contribution to the total decay rates , which however is quite small at this level of accuracy ) .thus a realistic evaluation of @xmath216 does not exceed @xmath224 , which constitutes only about 1% of the total decay rate of @xmath13 .thus the shift of the total decay rate of @xmath13 due to the effects of @xmath108 is dominantly associated with the su(3 ) non - singlet difference ( [ dbres ] ) .the shift of the @xmath13 decay rate with respect to the average width @xmath225 due to the non - singlet operators is one third of the splitting ( [ dbres ] ) , i.e. about 5% . adding to this the 1% shift of the average width and another 1% difference from the meson decays due to the suppression of the latter by the @xmath106 chromomagnetic effects , one concludes that at the present level of theoretical understanding it looks impossible to explain a more than 10% enhancement of the total decay rate of @xmath13 relative to @xmath52 , where an ample 3% margin is added for the uncertainties of higher order terms in ope as well as for higher order qcd radiative effects in the discussed corrections . in other words , the expected pattern of the lifetimes of the @xmath0 hyperons in the triplet , relative to @xmath52 , is ( _ b^0 ) ( _ b ) < ( b_d ) < ( _ b^- ) , [ pattern ] with the best " theoretical estimate of the differences to be about 7% for each step of the inequality . for the double strange hyperons @xmath226 and @xmath227 there is presently no better approach to evaluating the four - quark matrix elements , than the use of simplistic relations , like ( [ omegaold ] ) based on constituent quark model .such relations imply that the effects of the strange quark , ws and pi , in the @xmath175 baryons are significantly enhanced over the same effects in the cascade hyperons . in charmed baryonsa presence of strange spectator quark enhances the decay through positive interference with the quark emerging from the @xmath228 transition in the decay . for @xmath226this implies a significant enhancement of the total decay rate @xcite , which is in perfect agreement with the data on the @xmath226 lifetime .also a similar enhancement is expected for the semileptonic decay rate of @xmath226 . in @xmath0 baryons , on the contrary , the interference effect for a spectator strange quark is negative .thus the non - leptonic decay rate of @xmath227 is expected to be suppressed , leaving @xmath227 most probably the longest - living particle among the @xmath0 baryons .rather unexpectedly , the problem of four - quark matrix elements over heavy hyperons is related to decays of the type @xmath230 . the mass difference between the charmed cascade hyperons @xmath191 and @xmath187 is about 180 mev .the expected analogous mass splitting for the @xmath0 hyperons should be very close to this number , since in the heavy quark limit m(_b)-m(_b)=m(_c)-m(_c ) + o(m_c^-2-m_b^-2 ) .[ massdiff ] therefore in both cases are kinematically possible decays of the type @xmath229 , in which the heavy quark is not destroyed , and which are quite similar to decays of ordinary ` light ' hyperons .surprisingly , the rate of these decays for both @xmath191 and @xmath231 is not insignificantly small , but rather their branching fraction can reach a level of few per mill for @xmath191 and of one percent or more for @xmath231 @xcite .the transitions @xmath229 are induced by two underlying weak processes : the ` spectator ' decay of a strange quark , @xmath232 , which does not involve the heavy quark , and the ` non - spectator ' weak scattering ( ws ) s c c d [ ws ] trough the weak interaction of the @xmath233 and @xmath234 currents .one can also readily see that the ws mechanism contributes only to the decays @xmath235 and is not present in the decays of the @xmath0 cascade hyperons .an important starting point in considering these transitions is that in the heavy quark limit the spin of the heavy quark completely decouples from the spin of the light component of the baryon , and that the latter light component in both the initial an the final baryons forms a @xmath102 state with quantum numbers of a diquark .since the momentum transfer in the considered decays is small in comparison with the mass of the heavy quark the spin of the amplitudes with spin flip of the heavy quark , and thus of the baryon , are suppressed by @xmath11 . in terms of the two possible partial waves in the decay @xmath229 , the @xmath236 and @xmath237, this implies that the @xmath237 wave is strongly suppressed and the decays are dominated by the @xmath236 wave .according to the well known current algebra technique , the @xmath236 wave amplitudes of pion emission can be considered in the chiral limit at zero four - momentum of the pion , where they are described by the pcac reduction formula ( pole terms are absent in these processes ) : _ q _ i ( p=0 ) | h_w | _ q = f _ _ q || _ q , [ pcac ] where @xmath238 is the pion triplet in the cartesian notation , and @xmath239 is the corresponding isotopic triplet of axial charges .the constant @xmath240 , normalized by the charged pion decay , is used here , hence the coefficient @xmath241 in eq.([pcac ] ) .the hamiltonian @xmath242 in eq.([pcac ] ) is the non - leptonic strangeness - changing hamiltonian : @xmath243 + \right .\nonumber \\ & & \left .\left ( c_+ - c_- \right ) \ , \left [ ( \overline d_l \ , \gamma_\mu \ , s_l)\ , ( \overline u_l \ , \gamma_\mu \ , u_l ) - ( \overline d_l \ ,\gamma_\mu \ ,s_l)\ , ( \overline c_l \ , \gamma_\mu \ , c_l ) \right ] \right \}~. \label{hw}\end{aligned}\ ] ] in this formula the weak hamiltonian is assumed to be normalized ( in llo ) at @xmath244 .the terms in the hamiltonian ( [ hw ] ) without the charmed quark fields describe the ` spectator ' nonleptonic decay of the strange quark , while those with the @xmath1 quark correspond to the ws process ( [ ws ] ) .it is straightforward to see from eq.([pcac ] ) that in the pcac limit the discussed decays should obey the @xmath245 rule .indeed , the commutator of the weak hamiltonian with the axial charges transforms under the isotopic su(2 ) in the same way as the hamiltonian itself . in other words , the @xmath246 part of @xmath242 after the commutation gives an @xmath246 operator , while the @xmath247 part after the commutation gives an @xmath247 operator .the latter operator however can not have a non vanishing matrix element between an isotopic singlet , @xmath248 , and an isotopic doublet , @xmath249 .thus the @xmath250 part of @xmath242 gives no contribution to the @xmath236 wave amplitudes in the pcac limit .once the isotopic properties of the decay amplitudes are fixed , one can concentrate on specific charge decay channels , e.g. @xmath251 and @xmath252 .an application of the pcac relation ( [ pcac ] ) with the hamiltonian from eq.([hw ] ) to these decays , gives the expressions for the amplitudes at @xmath253 in terms of baryonic matrix elements of four - quark operators : @xmath254 + \nonumber \\ & & \left ( c_+ - c_- \right ) \, \left [ ( \overline d_l \ , \gamma_\mu \ , s_l)\ , ( \overline u_l \ , \gamma_\mu \ , d_l ) - ( \overline u_l \ ,\gamma_\mu \ ,s_l)\ , ( \overline u_l \ , \gamma_\mu \ , u_l ) \right ] | \ , \xi_b^- \rangle = \nonumber \\ & & { \sqrt{2 } \over f_\pi } \ , g_f \ , \cos \theta_c \ , \sin \theta_c \ , \langle \lambda_b \,| c_- \ , \left [ ( \overline u_l \ , \gamma_\mu \ , s_l)\ , ( \overline d_l \ , \gamma_\mu \ , d_l ) - ( \overline d_l \ ,\gamma_\mu \ ,s_l)\ , ( \overline u_l \ , \gamma_\mu \ , d_l ) \right ] - \nonumber \\ & & { c_+ \over 3 } \left [ ( \overline u_l \ , \gamma_\mu \ ,s_l)\ , ( \overline d_l \ , \gamma_\mu \ , d_l ) + ( \overline d_l \ , \gamma_\mu \ , s_l)\ , ( \overline u_l \ , \gamma_\mu \ , d_l ) + 2 \ , ( \overline u_l \ , \gamma_\mu \ , s_l)\ , ( \overline u_l \ , \gamma_\mu \ , u_l ) \right ] | \ , \xi_b^- \rangle~ , \label{xib}\end{aligned}\ ] ] where in the last transition the operator structure with @xmath255 giving a vanishing contribution is removed and only the structures with explicitly @xmath245 are retained , and @xmath256 in the latter formula the first term on the r.h.s . expresses the fact that in the heavy quark limit the ` spectator ' amplitudes do not depend on the flavor or the mass of the heavy quark . the rest of the expression ( [ xic ] ) describes the ` non - spectator ' contribution to the amplitude of the charmed hyperon decay . using the flavor su(3 ) symmetry the latter contributioncan be related to the non - singlet matrix elements ( [ defxy ] ) ( normalized at @xmath164 ) as @xmath257~. \label{das}\end{aligned}\ ] ] furthermore , with the help of the equations ( [ d1 ] ) and ( [ d2 ] ) relating the matrix elements @xmath23 and @xmath184 to the differences of the total decay widths within the triplet of charmed hyperons , one can eliminate @xmath23 and @xmath184 in favor of the measured width differences .the resulting expression has the form @xmath258 = \nonumber \\ -10^{-7 } \left [ 0.97 \left ( \gamma(\xi_c^0)-\gamma(\lambda_c ) \right ) + 0.09 \left ( \gamma(\lambda_c)-\gamma(\xi_c^+ ) \right ) \right ] \ , \left ( { 1.4 \ , gev \over m_c } \right ) ^2 \ , ps ~ , \label{dasm}\end{aligned}\ ] ] where , clearly , in the latter form the widths are assumed to be expressed in @xmath259 , and @xmath260 is used as a ` reference ' value for the charmed quark mass .it is seen from eq.([dasm ] ) that the evaluation of the difference of the amplitudes within the discussed approach is mostly sensitive to the difference of the decay rates of @xmath193 and @xmath187 , with only very little sensitivity to the total decay width of @xmath188 . using the current data the difference @xmath261is estimated as a= -(5.4 2 ) 10 ^ -7 , [ dasn ] with the uncertainty being dominated by the experimental error in the lifetime of @xmath193 .an amplitude @xmath262 of the magnitude , given by the central value in eq.([dasn ] ) would produce a decay rate @xmath263 , which result can also be written in a form of triangle inequality + .[ tri ] although at present it is not possible to evaluate in a reasonably model independent way the matrix element in eq.([xib ] ) for the ` spectator ' decay amplitude , the inequality ( [ tri ] ) shows that at least some of the discussed pion transitions should go at the level of @xmath264 , similar to the rates of analogous decays of ` light ' hyperons .we summarize here specific predictions for the inclusive decay rates , which can be argued with a certain degree of theoretical reliability , and which can be possibly experimentally tested in the nearest future .shifman and m.b .voloshin , ( 1981 ) unpublished , presented in the review v.a.khoze and m.a .shifman , .m.a . shifman and m.b .voloshin , .m.a . shifman and m.b .voloshin , .b. chibisov , r.d .dikeman , m.a .shifman , and n. uraltsev , .i. bigi , m. shifman , n. uraltsev , and a. vainshtein , .bigi , n.g .uraltsev , and a.i .vainshtein , ; [ e : * b297 * ( 1993 ) 477 ] .b. guberina , s. nussinov , r. peccei , and r. rckl , .n. bili , b. guberina , and j. trampeti , .i. bigi , b. blok , m.a .shifman , and a. vainshtein , .g. altarelli and s. petrarca , .e. bagan , p. ball , v.m .braun , and p. gosdzinsky , .q. hokim and x.y .a. czarnecki and m. jezabek , .e. bagan , p. ball , v.m .braun and p. gosdzinsky , .voloshin , .voloshin , surv ., * 8 * ( 1995 ) 27 .p. ball and v. braun , .m. neubert , trieste summer school lectures , report clns 00/1660 , jan .2000 ; [ hep - ph/001334 ] .m. neubert and c.t .sachrajda , .m.a . shifman and m.b .voloshin , .voloshin , .b. guberina and b. meli , eur.phys.j .* c2 * ( 1998 ) 697 .particle data group ._ 1999 www update _ , http://pdg.lbl.gov/1999/bxxx.html ukqcd collaboration : m. di pierro , c.t .sachrajda , and c. michael , univ . of southampton report shep 99 - 07 , june 1999 ; [ hep - lat/9906031 ] .voloshin , .voloshin , phys.rev.*d61 * ( 2000 ) 074026 .voloshin , phys.lett . *b476 * ( 2000 ) 297 .

<S> a compact review of the theory , including some recent developments , of inclusive weak decay rates of charmed and @xmath0 hadrons with an emphasis on predictions that can be tested in the forthcoming experiments . </S><S> * theoretical physics institute + university of minnesota + * tpi - minn-00/17 + umn - th-1851 - 00 + april 2000 + * inclusive weak decay rates of heavy hadrons + * theoretical physics institute , university of minnesota , minneapolis , mn 55455 + and + institute of theoretical and experimental physics , moscow , 117259 + _ expanded version of a contribution to the final report book of the fermilab workshop on b physics at tevatron _ </S>

0706.1982

understanding the nature of the cosmological density perturbation is essential for the development of inflationary cosmology .recent years have seen a surge in the development of precision cosmology , with improved measurements of the temperature and polarization anisotropies of the cosmic microwave background ( cmb ) as well as increasingly accurate large scale structure ( lss ) surveys .the task of faithfully reconstructing the physics of inflation from this data is the central goal of inflationary phenomenology .the fundamental physics responsible for inflation remains elusive , however .models of inflation abound with motivations stemming from supergravity , the string landscape , d - branes , extensions to the standard model and others @xcite . a reliable reconstruction of the inflaton potential , and therefore the physics responsible for inflation , hinges on accurate measurement and subsequent analysis of cosmological data .there has been much progress in the development of methods of primordial power spectrum reconstruction @xcite , as well as methods for reconstructing the inflaton potential @xcite . with the recent release of the wmap3 data - set @xcite and with future cmb missions eminent @xcite ,it is both important and timely to investigate how well current and future experiments will be able to constrain the form of the primordial power spectrum .recent analyses report a nearly scale invariant spectrum of adiabatic , gaussian density perturbations , consistent with the simplest models of single field slow - roll inflation @xcite. however , experimental and theoretical inaccuracy , in addition to the limited detection of the polarization signal , limit parameter estimation efforts .the dominant source of error is that due to cosmic variance , a statistical uncertainty related to the fact that there is only one universe to observe .this effect is present in both temperature and polarization spectra and is most pronounced on large scales ( at low cmb multipoles ) .cosmic variance results in a fundamental limit on the resolution of cmb spectra , which restricts the accuracy with which we can reconstruct the physics of the early universe .since the primordial perturbation spectrum provides one of the few pieces of observational evidence for inflation , these uncertainties impose a limit on how well we can reconstruct the physics of the inflationary era .while single field slow - roll inflation is well supported by current data , it is of interest to determine whether more exotic inflationary scenarios might exist within the resolution limit of observational data . in this paper, we study our current ability to constrain the form of the power spectrum . using the latest wmap data - set @xcite, we generate a best - fit base cosmological model characterized by 7 parameters , ( @xmath3 ) .we are interested in identifying alternative models with more general power spectra that yield comparable fits to current data .such statistically degenerate models can be considered effectively indistinguishable , and differences in the underlying inflation models unresolvable . in order to test a wide range of spectral shapes ,we introduce a monte carlo reconstruction program that combines the inflationary flow technique @xcite with a numerical integration of the mode equations of quantum fluctuations . aside from being a novel reconstruction technique in its own right , we find it highly suitable for the present investigation .we use this method to stochastically generate an ensemble of power spectra to 6@xmath4-order in the flow space .for each spectrum it is possible reconstruct the exact inflaton potential . because we solve for the spectrum numerically , we can successfully reconstruct spectra arising from slow - roll as well as strongly non - slow - roll inflation models .additionally , the flow approach allows us to enforce that each model yield the appropriate amount of inflation , @xmath5 $ ] .for each spectrum , we fix the non - spectral parameters at the best - fit values and calculate the new model s likelihood .we establish a resolution criterion based on the relative likelihoods of the trial models and the best - fit model . by fixing the non - spectral parameters, we are not performing a bayesian analysis of the enlarged parameter space of flow parameters . by allowing all the parameters to vary as is done in bayesian parameter estimation ,it is possible that new parameter degeneracies would allow for a wider range of possible spectral shapes to be in agreement with the data ( see , for example , ref .our method therefore only samples a subset of possible spectra and should be viewed as a more conservative approach .this study is more akin to a model selection analysis : we attempt to identify models constructed from the inflationary flow space that are equally consistent with the data as models parameterized as simple power - laws .we utilize the p - value calculated from the model s chi - squared per degrees of freedom as a measure of statistical significance .this approach is frequentist in nature , yet we find general agreement with the bayesian study of @xcite . this paper is organized as follows : in section ii we review the inflationary flow formalism and in section iii we introduce the new method which combines the flow approach with a numerical integration of the mode equation . in sectioniv we investigate primordial power spectrum resolution with the wmap3 data - set .we find that with tt / te / ee data alone , our ability to reconstruct the inflaton potential is strongly affected by the limited spectral resolution .we find that strongly running spectra arising from non - slow - roll inflation models are currently indistinguishable from a best - fit power - law model typical of slow - roll inflation . however , since these non - slow - roll models are typically accompanied by large tensor components , any information regarding the b - mode polarization signal will significantly improve resolvability .furthermore , we find that the established confidence limits on the tensor - to - scalar ratio obtained from sdss and wmap3 when running is included are violated by the non - power - law spectra identified in this study .we find spectra with values for @xmath1 at @xmath6 that lie outside the 2-@xmath2 contour of the @xmath7-@xmath1 marginalized likelihood @xcite yield _equally good _ fits to the data as the best - fit power - law .section v includes discussion and conclusions .the evolution of a friedmann - robertson - walker ( frw ) universe dominated by a single minimally - coupled scalar field ( the _ inflaton _ ) with potential @xmath8 is given by the equations @xmath9 , \\\label{fe2 } \frac{\ddot{a}}{a } & = & \frac{8\pi}{3m_{\rm pl}^2}\left[v(\phi ) - \dot{\phi}^2\right].\end{aligned}\ ] ] we have assumed a flat universe described by the metric @xmath10 , where @xmath11 is the scale factor and dots denote derivatives with respect to coordinate time , @xmath12 .the evolution of the scalar field follows from stress - energy conservation and takes the form of a klein - gordon equation with a cosmological friction term , @xmath13 where primes denote derivatives with respect to the field , @xmath14 .equations ( [ fe1 ] ) and ( [ kg ] ) can be combined to yield the alternative equations of motion , @xmath15 where the hubble parameter , written as a function of @xmath14 , becomes the dynamical variable .the field @xmath14 serves as a convenient time variable so long as it is monotonic .the second of the above two equations is known as the _ hamilton - jacobi equation _ , and may be written more simply as @xmath16 = \left(\frac{8\pi}{3m_{\rm pl}^2}\right)v(\phi),\ ] ] where the parameter @xmath17 is defined as @xmath18 physically , @xmath17 is the equation - of - state parameter of the cosmological fluid , and from eq .( [ fe2 ] ) it follows that the condition for inflation , @xmath19 , requires that @xmath20 .starting with the equation - of - state parameter , it is possible to define an infinite hierarchy of parameters @xcite by taking successive derivatives of the hubble parameter , @xmath21 , @xmath22 in what follows , we will refer to eqs .( [ epsilon ] ) and ( [ flowpar ] ) collectively as the _ hubble flow parameters_. these are often referred to as _ slow - roll _ parameters in the literature , however , they are defined here without any assumption of slow - roll . for this analysisit is most convenient to use the number of e - folds before the end of inflation , @xmath23 , as our time variable .the scale factor may be written @xmath24 and from eq .( [ fe1 ] ) , @xmath25 making use of this relation , we take successive derivatives of the flow parameters with respect to @xmath23 , generating an infinite set of differential equations @xcite , @xmath26(^{\ell}\lambda_h ) + { ^{\ell + 1}}\lambda_h , \nonumber\end{aligned}\ ] ] where @xmath27 . in practice, this system is truncated at some finite order @xmath28 by requiring that @xmath29 .this system can then be solved numerically by specifying the initial conditions of the parameters @xmath17 , @xmath30 at some arbitrary time , @xmath31 . although the system is truncated at finite order , this results in an _ exact _ solution for the background evolution of an frw universe dominated by a single scalar field .this is due to the form of the flow equations , where it can be seen that the truncation @xmath29 ensures that all higher - order parameters vanish for all time . in ref .@xcite , the initial conditions were drawn randomly from the ranges @xmath32 \\ \sigma & \in & [ -0.5,0.5 ] \nonumber \\ \xi^2 & \in & [ -0.05,0.05 ] \nonumber \\ ^3\lambda_h & \in & [ -0.005,0.005 ] \nonumber \\ \vdots\nonumber \\ ^{m+1}\lambda_h & \in & 0 , \nonumber\end{aligned}\ ] ] although other choices are possible .the system is then evolved forward in time until either inflation ends ( @xmath33 ) , or the system reaches a late - time asymptote , @xmath34 .the latter possibility arises in models in which the field evolves to a point of nonzero vacuum energy , leading to eternal inflation . in such cases, inflation must end via the action of an auxiliary field , such as in hybrid models @xcite .such models generically predict scalar spectra with @xmath35 and negligible running , in conflict with the wmap3 data - set @xcite .we therefore focus solely on models for which inflation ends through a failure of slow - roll , @xmath33 , originally termed _ nontrivial _ models@xcite . once a nontrivial model is found , the flow equations are integrated backwards in time from @xmath36 to @xmath37 , where @xmath38 is drawn randomly from the range @xmath39 $ ] .the solution to the flow equations then comprise the full time evolution of the parameters @xmath17 , @xmath40 from @xmath41 .it is then possible to reconstruct the inflaton potential using the hamilton - jacobi equation , eq .( [ hje ] ) @xcite .the values of the flow parameters at @xmath38 can then be used to calculate observables at this point via the relations @xcite @xmath42 , \\n_s & = & 1 + \sigma - ( 5 - 3c)\epsilon^2 \nonumber \\ & & -\frac{1}{4}(3 - 5c)\sigma\epsilon+\frac{1}{2}(3-c)\xi^2 , \nonumber \\ \alpha & = & \frac{dn_s}{d{\rm ln}k } = -\left(\frac{1}{1-\epsilon}\right)\frac{dn_s}{dn } , \nonumber\end{aligned}\ ] ] where @xmath43 , @xmath44 , and @xmath45 denotes the running of the scalar spectral index , @xmath7 . by adopting a monte carlo approach , large numbers of models can be generated and their observable predictions compared with current observational data .this serves as a means for constructing classes of models that satisfy certain criteria , as well as gaining insight into the generic features of the inflationary parameter space .models generated stochastically using the above method are observed to cluster strongly in the @xmath46 and @xmath47 planes @xcite .it is difficult , however , to attribute any rigorous statistical meaning to such clustering , since there is no well - defined measure on the parameter space .the main drawback of the flow method is that while generating exact inflationary evolutions , it must rely on approximations when calculating the observable predictions of these models .these approximations are made as two separate series truncations .the first approximation has to do with the parameterization of the power spectrum . when quoting observables in terms of @xmath1 , @xmath7 and @xmath45 , one is considering a truncated taylor expansion of @xmath48 in @xmath49 , @xmath50 where @xmath51 is the comoving wavenumber .clearly , this parameterization is only useful if the higher - order terms are small across the range of scales for which it is expected to hold .the second approximation results from the use of eq .( [ obs ] ) to connect these spectral parameters to the inflationary flow parameters .these expressions are only accurate to order @xmath52 in the flow parameters . while the evolution of the lowest - order parameters appearing explicitly in eq .( [ obs ] ) are determined by the full set of flow parameters out to order @xmath28 , the use of these expressions still requires that these higher - order parameters be negligible on observable scales .while , in principle , eq . ( [ obs ] ) can be extended to arbitrary order in slow - roll , the expressions quickly become very algebraically complex @xcite . furthermore, this approach evaluates observables at ` horizon crossing ' , a technique that is not always applicable @xcite .these approximations are perfectly valid when considering inflation models that satisfy the slow - roll criteria , @xmath17 , @xmath53 . in this case( [ obs ] ) is highly accurate and higher - order spectral parameters are negligible , validating the use of eq .( [ spec ] ) .however , if one wishes to use flow methods to explore regions of the inflationary parameter not well approximated by slow - roll , this approach must be adapted . as pointed out by easther and peiris @xcite ,if one imposes an inflationary prior on the underlying model space , then there is no need to calculate observables in terms of the spectral parameters .the flow parameters themselves fully specify the physics of the inflationary epoch and one avoids introducing further , unnecessary errors .we adopt this philosophy in what follows , doing away completely with the parameterization eq .( [ spec ] ) and the specification of observables eq .( [ obs ] ) .we develop an improved method of reconstruction by combining the flow formalism with a numerical evaluation of the mode equation of inflationary perturbations . by solving the mode equation for multiple @xmath51-values ,it is possible to reconstruct the primordial power spectra without recourse to spectral parameters or the slow - roll approximation .this leads to a consistent match between the scalar field potential and associated perturbation spectra , to the same degree of accuracy .in addition , it allows for an investigation of regions of the inflationary parameter space that lie far from slow - roll , leading to a more robust reconstruction program .( for other efforts at numerical evaluation of inflationary perturbations , see refs . @xcite . )during inflation , vacuum fluctuations of the inflaton field are redshifted to superhorizon scales by the rapidly expanding spacetime where they become classical curvature perturbations .the inflaton field couples at linear order to the scalar metric perturbation , @xmath54 , which may be defined in terms of four scalar functions , @xmath55 if one works in _ comoving gauge _ , in which the spatial slices are orthogonal to the worldlines of comoving observers , the intrinsic curvature perturbation of the spacelike hypersurface is @xcite @xmath56 in comoving gauge , one has the additional freedom of requiring that the spatial part of the metric perturbation be isotropic , @xmath57 , so that @xmath58 .the coupling between inflaton and metric perturbations motivates the use of the gauge invariant mukhanov potential @xcite , @xmath59 where @xmath60 is the scale factor of the universe , @xmath61 is the hubble parameter and @xmath62 is the scalar field fluctuation .primes denote derivatives with respect to conformal time , @xmath63 . on comoving hypersurfaces , @xmath64 , and the mukhanov potentialis related to the comoving curvature perturbation , @xmath65 where @xmath66 .the two - point correlation function can be written in terms of the power spectrum , @xmath67 , @xmath68 the power spectrum of the curvature perturbation may then be written , @xmath69 where the @xmath70 are the fourier modes of the gauge - invariant potential satisfying the equation of motion , @xmath71 tensor metric perturbations , @xmath72 , are also excited during inflation , leading to a large - scale gravitational wave background .the metric perturbation is purely spatial and may be described by a single function , @xmath73 .the dynamics of these perturbations are determined by the linearized einstein equations , which follow from the action @xcite , @xmath74 because gravitational waves are both transverse and traceless ( _ i.e. _ @xmath75 ) , they may be decomposed into two independent polarization modes , denoted + and @xmath76 .the fourier decomposition may then be written @xmath77e^{i{\bf k \cdot x } } , \nonumber\end{aligned}\ ] ] where we have used the fact that @xmath78 which follows from the condition @xmath79 .the power spectrum of tensor fluctuations can then be be written @xmath80 following the field redefinition , @xmath81 the equation of motion that follows from eq .( [ gw ] ) for each polarization mode becomes that of a canonically normalized massless scalar field in an frw background , @xmath82 where the polarization indices have been suppressed . here@xmath83 is the quantum mode associated with positive - frequency excitations of the metric perturbation . in terms of the new field , the power spectrum eq .( [ spech ] ) becomes @xmath84 although written in terms of a single solution of eq .( [ tensoreq ] ) , this expression includes the contributions from both polarization modes .our goal is to combine eqs .( [ modeq ] ) and ( [ tensoreq ] ) with the flow equations , eq .( [ floweqs ] ) .then , for any initial point in the flow space , we will be able to determine the full evolution of the modes @xmath70 and @xmath83 together with the background evolution .since the most convenient time variable for use with the flow formalism is @xmath23 , the number of e - folds before the end of inflation , it is necessary to recast eqs .( [ modeq ] ) and ( [ tensoreq ] ) in terms of @xmath23 .these expressions become @xmath85u_k = 0,\ ] ] and @xmath86v_k = 0,\ ] ] where the function @xmath87 of eq .( [ modeqn ] ) is defined as @xmath88 the full system of differential equations is formed by eqs .( [ modeqn ] ) and ( [ tensoreqn ] ) together with the flow equations , eq .( [ floweqs ] ) .the standard choice of initial conditions for the mode function is that defined by the bunch - davies vacuum , @xmath89 when solving the mode equations numerically we can not use this exact condition for two reasons .first , we can not set this condition in the infinite past , but must impose it at a finite time . imposing the limit eq .( [ bd ] ) at a finite time , _i.e. _ finite length scale , results in modulations of the power spectrum akin to those arising from transplanckian effects @xcite .this effect can be minimized by initializing the mode functions at sufficiently early times ( small length scales ) .second , we must write eq .( [ bd ] ) in terms of @xmath90 for use with eqs ( [ modeqn ] ) and ( [ tensoreqn ] ) , and this function is not known in general . from the relation @xmath91, we obtain @xmath92 where @xmath93 is the ratio of the hubble radius to the proper wavelength of the fluctuation and is a function of @xmath23 . if @xmath94 is approximately constant , this equation can be integrated to give @xmath95 . from the equation@xmath96 we see that @xmath97 if @xmath98 is taken sufficiently large . by taking @xmath98 large, we are also ensuring that the modes are initialized in the short - wavelength limit .this motivates the use of the approximate initial conditions , @xmath99 for a choice of initial flow parameters @xmath100 , @xmath101 , ... , @xmath102 , eq . ( [ ics ] ), we set the initial condition for each @xmath51-mode at @xmath103 .this proves to be sufficiently large to ensure the accuracy of the conditions eq .( [ cond ] ) . since there are two complex solutions to each of eqs .( [ modeqn ] ) and ( [ tensoreqn ] ) , rather than work with complex coefficients , we define the orthogonal solution basis , @xmath104 each mode is evolved from @xmath105 to @xmath36 , the end of inflation .the amplitude of the power spectrum for each @xmath51-mode is then obtained by evaluating eqs .( [ specs ] ) and ( [ tspec ] ) at this time , when all modes have attained the long - wavelength limit .in this section we investigate the ability of the latest wmap data - set to accurately resolve the form of the power spectrum .we approach this problem by considering a 7-parameter fiducial best - fit model parameterized by inflationary and non - inflationary degrees of freedom .we suppose that this model accurately describes the universe and search for alternative models which are statistically indistinguishable from it .the non - inflationary parameters are taken to be the baryon and cdm densities , @xmath106 and @xmath107 , the hubble parameter , @xmath108 , the optical depth to reionization , @xmath63 , and the overall spectrum normalization , @xmath109 .the inflationary parameters are the scalar spectral index , @xmath7 , and the tensor contribution , @xmath1 .we consider purely adiabatic initial perturbations and assume spatial flatness . the tensor spectral index , @xmath110 ,is assumed to satisfy the inflationary consistency condition , @xmath111 , and does not represent an additional free parameter .since we only vary the form of the spectrum in this analysis , we fix the non - inflationary parameters at their best - fit values .we then replace the power - law parameterization ( @xmath7 , @xmath1 ) with tensor and scalar power spectra generated with the monte carlo method .we work to @xmath112-order in the flow space , _ i.e. _ @xmath113 for @xmath114 .we use a version of ` camb ` @xcite modified to accept arbitrary power spectra as input to generate the associated @xmath115-spectra .it is then possible to calculate the model s effective chi - square , @xmath116 , using the wmap3 likelihood software available at the ` lambda ` website .one can then collect models that satisfy specific likelihood criteria relative to the best - fit model .how does one compare the statistical significance of two different models ?if the two models have the same parameterization , then the confidence limits obtained from a maximum likelihood analysis of the parameter space suffice .however , in this study , we wish to compare a model comprising 2 spectral parameters with a model comprising 6 flow parameters .a simple statistic used in model selection analyses is the likelihood ratio test , @xmath117 , where the simple model is so named because it contains fewer free parameters than the more complex model .this statistic is approximately @xmath118 distributed with @xmath119 degrees of degrees of freedom equal to the difference in the number of free parameters between the complex and the simple model .the significance level associated with the value of the @xmath120 can then be used to quantify the relative goodness - of - fit between these models .use of this test requires that the two models be _ nested _ - that the complex model is formed by adding parameters to the base model .the space of flow parameters and spectral parameters are at best only approximately nested : there is no way to map the flow parameters @xmath121 to a finite number of spectral parameters .this is because there are infinitely many higher - order spectral parameters that are functions of these 6 flow parameters that in general will not vanish under the mapping .we say _ approximate _ because if these parameters are very small , they might be neglected .a further downside is that it is not clear how to tell from this statistic if two models are of comparable significance .another method often used in model selection analyses is the bayesian information criterion ( bic ) @xcite , @xmath122 where @xmath123 is the maximum likelihood of the model , @xmath51 the number of free parameters and @xmath23 the number of data points .the bic is an approximation of the bayesian evidence , which is the integral of the likelihood function over the full parameter space .the bic penalizes overparameterized models that do nt provide significantly better fits to the data .this approach is also ill - suited for our purposes , because it assumes that all @xmath51 parameters are well measured by the data .in fact , it is the opposite of this case that motivates this study : current data is _ not _ good enough to resolve higher order terms in the spectral decomposition , and we seek to determine what these resolution limits are .in what follows , we instead consider the p - value calculated from a model s chi - square per degrees of freedom as a measure of goodness - of - fit .this method is applicable so long as the likelihoods are approximately gaussian .the p - value of a proposed model with a given @xmath124 is @xmath125 where @xmath126 is the chi - square probability distribution function with @xmath119 degrees of freedom ( d.o.f . ) , @xmath127 for a given _ significance level _ , @xmath45 , if @xmath128 then the proposed model may be rejected at the @xmath129 confidence level .more precisely , the p - value is the probability that we obtain a particular @xmath124 due to chance alone .it can therefore be interpreted as the probability of falsely rejecting a correct model of the universe . to determine relative goodness - of - fit between two models ,one simply compares their p - values . in order to be as conservative as possible, we consider trial models lying within @xmath130 of the best - fit model to be effectively indistinguishable from it .the spectra generated by @xmath112-order monte carlo are described by 6 free parameters , corresponding to the flow parameters @xmath17 , @xmath2 , ... , @xmath131 .therefore , the trial spectra contain 4 more parameters than the best - fit power - law model . the difference in p - value between a trial model with @xmath132 d.o.f . and the best - fit model with @xmath119 d.o.f ., @xmath133 corresponds to the likelihood spread @xmath134 the chi - square per degrees of freedom of the best - fit model is @xmath135 . of the best - fit model .the spectrum of the best - fit model is represented by the heavy dashed line .the spectra are colored - coded according to their predicted values for @xmath1 at @xmath136 : red yield @xmath137 and black yield @xmath138.,width=360 ] of the best - fit model .the potentials are colored - coded according to their predicted values for @xmath1 at @xmath136 : red yield @xmath137 and black yield @xmath138 .the potentials have been given a common , arbitrary normalization at @xmath139 , when scales corresponding to the quadrupole exit the horizon.,width=360 ] we begin with a best - fit model with @xmath140 , @xmath141 and a tensor - to - scalar ratio , @xmath142 . in figure [ figure1 ]we present a sample of power spectra lying within @xmath130 of this model , and in figure [ figure5 ] some reconstructed potentials . in order to identify spectra exhibiting the strongest deviation from power - law, we initially imposed a non - slow - roll prior on the model space . in order for a model to be considered for full power spectrum integration , we require that the spectral index calculated to @xmath143-order in slow - roll eq .( [ obs ] ) differ from the @xmath144-order result @xcite by at least @xmath145 .this search resulted in the grouping of red - colored spectra exhibiting a large suppression of power on large scales in figure [ figure1 ] .we later relaxed this prior to obtain the black - colored spectra , allowing us to form a degeneracy envelope .remarkably , the red spectra remain equally good fits to the data as the fiducial power - law even when sdss data @xcite is included .for the case of power - law spectra , the latest analyses report @xmath146 at @xmath147-confidence when sdss is combined with wmap3 , while @xmath148 at @xmath147-confidence with wmap3 alone @xcite .the error bars open up considerably when running is allowed : @xmath149 at @xmath147-confidence sdss+wmap3 , versus @xmath150 with wmap3 alone .spectra with larger @xmath151 can be accommodated by the data if there is suppressed scalar power on these scales .as @xmath1 decreases on smaller scales , the scalar spectrum must run to larger amplitudes in order to maintain the correct amount of overall power .the sdss data provides an accurate measurement of the matter power spectrum on scales @xmath152 , and while not probing scales @xmath153 directly , if there is a strong negative _ constant _ running of the spectrum on these scales it may lead to a measurable change in power on the intermediate scales that are directly probed by sdss . in particular , if the running is too large at @xmath153 , there will be a loss of power on smaller scales as the spectrum dips blue again .this results in the much tightened bound on @xmath1 .while the red spectra in figure [ figure1 ] are strongly running on large scales , permitting a sizeable tensor - to - scalar ratio , this running is _ not constant _ and is unsubstantial on small scales . indeed ,when the red spectra are plotted in the @xmath7-@xmath1 plane , we find that they all lie outside the 1-@xmath2 contour of the sdss+wmap marginalized likelihood , figure [ figure4 ] .therefore , by considering more general spectra , the error bars inevitably get much larger .-@xmath1 plane from combined sdss and wmap data for spectra with constant running .the inner and outer contours mark the @xmath154- and @xmath147-confidence intervals , respectively .the red points denote the values of @xmath7 and @xmath1 of the red spectra in figure [ figure1 ] .the blue square is that of the fiducial power - law model.,width=360 ] the spectral index of the non - power - law spectra is calculated from the slope of the spectrum at @xmath155 , @xmath156 this is a measure of the local value of @xmath7 around @xmath155 and it should be emphasized that the large uncertainty is due to the generality of the spectral shapes .however , the uncertainty in the local value of @xmath7 is drastically reduced at smaller scales , as indicated in figure [ figure1 ] .recently , cortes _@xcite determined that the effect of the degeneracy in the @xmath7-@xmath1 plane can be minimized by quoting observables at @xmath157 .of this model .the potential energy of the alternative model is given by the red solid line , and the kinetic energy is given by the red dotted line .the vertical line denotes the time when scales corresponding to the quadrupole leave the horizon.,width=360 ] the statistical degeneracy of the power spectra exhibited in figure [ figure1 ] translates directly to a degeneracy in the space of inflation models , figure [ figure5 ] .the red - colored inflation models that give rise to the red - colored non - power - law spectra are characterized by an initially _ fast - rolling _ inflaton that has failed to reach the slow - roll attractor by the time observable scales exit the horizon , similar to models proposed by contaldi , _ et al ._ @xcite . in figure [ figure2 ]we plot one such potential .the kinetic energy is initially large and monotonically decreasing with respect to time . from the equation of motion( [ kg ] ) , it is evident that if the field velocity is initially large , the hubble drag term @xmath158 dominates the evolution , slowing the field .the field decelerates until the drag term becomes subdominant near the end of inflation .in contrast , the kinetic energy of the best - fit power - law model ( not included in the figure ) is @xmath159 and is a monotonically increasing function of time .we conclude that current cmb data does not allow for a precise reconstruction of the inflationary power spectrum and the corresponding inflaton dynamics .in particular , we find a statistical degeneracy amongst slow - roll and non - slow - roll models .the span of the degeneracy envelope in figure [ figure1 ] is largely shaped by our inability to constrain @xmath1 , with models exhibiting the strongest deviation from power - law also predicting values of @xmath1 an order of magnitude larger than the fiducial model .future cmb missions @xcite with the ability to detect the b - mode polarization signal characteristic of gravitational radiation will put tighter constraints on @xmath1 @xcite , reducing or breaking the @xmath1-@xmath7 degeneracy and significantly improving primordial power spectrum resolution .we have investigated the ability of current data to constrain the form of the primordial power spectrum .the ability of the data to resolve the power spectrum can be tested by how well it singles out a best - fit model . starting with the latest wmap release, we generate a best - fit power law model consistent with the simplest models of slow - roll inflation .we test a wide variety of more complicated power spectra , and identify those that provide equally good fits to the data relative to the best - fit model .if the alternative spectra arise in an inflationary context , then we can hope to identify both slow - roll and non - slow - roll models that are equally consistent with current data . in order to test a wide array of inflationary power spectra, we turn to monte carlo reconstruction .we combine the flow formalism , which is a method of stochastic model generation , with a numerical integration of the mode equations of quantum fluctuations .this allows us to handle inflation models that yield spectra that are not well described by the standard spectral parameterization . for each spectrum thus generated, we fix the non - spectral parameters at their best - fit values and calculate the likelihood of the model . by fixing the non - spectral parameters, we are only sampling a subset of possible spectra that might be degenerate with the best - fit power - law model , making this approach conservative .we determine the statistical significance of each model by obtaining the p - value calculated from the @xmath160 , eq . ( [ pval ] ) .we generate an ensemble of power spectra to @xmath112-order in the flow space and select only those lying within @xmath161 of the best - fit model .a sampling of power spectra meeting this criterion are shown in figure [ figure1 ] .the current cmb data provided by wmap3 @xcite only reliably constrains the form of the power spectrum on intermediate scales , @xmath162 , with much variation on larger spatial scales where cosmic - variance is the dominant source of error . by doing away with the spectral parameterization eq ( [ spec ] ), we also free ourselves from the constraints imposed on the tensor - to - scalar ratio by sdss data , the tightest bounds affecting models with a constant running of the spectral index .the flow method allows us to generate models with varied forms , including models with non - constant running .such models can support a relatively large value of @xmath151 by having a large running on these scales that turns off on the intermediate scales probed by sdss .for example , we find spectra with @xmath163 that yield equally good fits to the wmap+sdss data as the best - fit power - law , a value which is well outside the @xmath164 @xmath147-confidence limit for spectra with running . by considering more general power spectra ,one therefore opens up regions of parameter space excluded in simpler models .the inflation models responsible for generating the strongly running spectra in this study are characterized by an initially fast - rolling inflaton .observable scales exit the horizon before the field slows to the slow - roll attractor , yielding a large amplitude of gravitational waves and significant running .we are able to conclude that while slow - roll inflation models yield perhaps the simplest explanation for the origin of large scale structure , _ fast - rolling _ inflaton fields are equally suitable candidates from a strictly data - driven standpoint .another interesting approach is to expand the inflationary model space beyond the simplest class of canonical single field models .an example is the dbi inflation scenario @xcite analyzed using flow techniques by peiris _@xcite , who find that such models are strongly constrained by the existing data . finally , we draw attention to the work of lesgourgues and valkenburg , ref . @xcite . in that analysis ,the authors consider an enlarged inflationary parameter space consisting of the coefficients of a taylor expanded inflaton potential .they perform a bayesian parameter estimation analysis and obtain confidence intervals on an ensemble of power spectra and their associated inflaton potentials .this choice of parameterization excludes models that deviate strongly from slow - roll , in particular , models of the type found in this analysis . aside from this difference, we find good agreement between our results , which are based on inferential statistics that are frequentist in nature , and the results of ref .@xcite , derived using strictly bayesian methods .the bottom line is that current cmb data does not reliably constrain the form of the power spectrum , and this conclusion can be reached from either a bayesian or a frequentist approach .this research is supported in part by the national science foundation under grant nsf - phy-0456777 .99 d. h. lyth and a. riotto , phys .rept . * 314 * , 1 ( 1999 ) [ arxiv : hep - ph/9807278 ] .f. quevedo , class .* 19 * , 5721 ( 2002 ) [ arxiv : hep - th/0210292 ] .j. m. cline , arxiv : hep - th/0612129 .m. matsumiya , m. sasaki and j. yokoyama , phys .d * 65 * , 083007 ( 2002 ) [ arxiv : astro - ph/0111549 ] .s. l. bridle , a. m. lewis , j. weller and g. efstathiou , mon . not .soc . * 342 * , l72 ( 2003 ) [ arxiv : astro - ph/0302306 ] .p. mukherjee and y. wang , astrophys .j. * 599 * , 1 ( 2003 ) [ arxiv : astro - ph/0303211 ] . w. hu and t. okamoto , phys . rev .d * 69 * , 043004 ( 2004 ) [ arxiv : astro - ph/0308049 ] .s. hannestad , jcap * 0404 * , 002 ( 2004 ) [ arxiv : astro - ph/0311491 ]. s. habib , a. heinen , k. heitmann , g. jungman and c. molina - paris , phys .d * 70 * , 083507 ( 2004 ) [ arxiv : astro - ph/0406134 ] .d. tocchini - valentini , y. hoffman and j. silk , mon . not .soc . * 367 * , 1095 ( 2006 ) [ arxiv : astro - ph/0509478 ] .a. shafieloo and t. souradeep , phys .d * 70 * , 043523 ( 2004 ) [ arxiv : astro - ph/0312174 ] .a. shafieloo , t. souradeep , p. manimaran , p. k. panigrahi and r. rangarajan , phys .d * 75 * , 123502 ( 2007 ) [ arxiv : astro - ph/0611352 ] .m. bridges , a. n. lasenby and m. p. hobson , mon . not .soc . * 369 * , 1123 ( 2006 ) [ arxiv : astro - ph/0511573 ] .m. bridges , a. n. lasenby and m. p. hobson , arxiv : astro - ph/0607404 . c. ringeval , arxiv : astro - ph/0703486 .j. e. lidsey , a. r. liddle , e. w. kolb , e. j. copeland , t. barreiro and m. abney , rev .phys . * 69 * , 373 ( 1997 ) [ arxiv : astro - ph/9508078 ] .i. j. grivell and a. r. liddle , phys .d * 61 * , 081301 ( 2000 ) [ arxiv : astro - ph/9906327 ] .r. easther and w. h. kinney , phys .d * 67 * , 043511 ( 2003 ) [ arxiv : astro - ph/0210345 ] .s. habib , k. heitmann and g. jungman , phys .lett . * 94 * , 061303 ( 2005 ) [ arxiv : astro - ph/0409599 ] .m. joy , e. d. stewart , j. o. gong and h. c. lee , jcap * 0504 * , 012 ( 2005 ) [ arxiv : astro - ph/0501659 ] . k. kadota , s. dodelson , w. hu and e. d. stewart , phys . rev .d * 72 * , 023510 ( 2005 ) [ arxiv : astro - ph/0505158 ] .d. n. spergel _ et al ._ , arxiv : astro - ph/0603449 .see http://www.rssd.esa.int/index.php?project=planck .see http://universe.nasa.gov/program/inflation.htm l. l. alabidi and d. h. lyth , jcap * 0608 * , 013 ( 2006 ) [ arxiv : astro - ph/0603539 ] .h. j. de vega and n. g. sanchez , arxiv : astro - ph/0604136 .w. h. kinney , e. w. kolb , a. melchiorri and a. riotto , phys .d * 74 * , 023502 ( 2006 ) [ arxiv : astro - ph/0605338 ] .j. martin and c. ringeval , jcap * 0608 * , 009 ( 2006 ) [ arxiv : astro - ph/0605367 ] .a. r. liddle , p. mukherjee and d. parkinson , arxiv : astro - ph/0608184 . m. b. hoffman and m. s. turner , phys .d * 64 * , 023506 ( 2001 ) [ arxiv : astro - ph/0006321 ] .w. h. kinney , phys .d * 66 * , 083508 ( 2002 ) [ arxiv : astro - ph/0206032 ] .s. chongchitnan and g. efstathiou , phys .d * 72 * , 083520 ( 2005 ) [ arxiv : astro - ph/0508355 ] .w. h. kinney , phys .d * 63 * , 043001 ( 2001 ) [ arxiv : astro - ph/0005410 ] .j. lesgourgues and w. valkenburg , arxiv : astro - ph/0703625 .a. lewis , a. challinor and a. lasenby , astrophys .j. * 538 * , 473 ( 2000 ) [ arxiv : astro - ph/9911177 ] .e. d. stewart and d. h. lyth , phys .b * 302 * , 171 ( 1993 ) [ arxiv : gr - qc/9302019 ] .j. o. gong and e. d. stewart , phys .b * 510 * , 1 ( 2001 ) [ arxiv : astro - ph/0101225 ] .j. choe , j. o. gong and e. d. stewart , jcap * 0407 * , 012 ( 2004 ) [ arxiv : hep - ph/0405155 ] .l. m. wang , v. f. mukhanov and p. j. steinhardt , phys .b * 414 * , 18 ( 1997 ) [ arxiv : astro - ph/9709032 ] .s. m. leach , m. sasaki , d. wands and a. r. liddle , phys .d * 64 * , 023512 ( 2001 ) [ arxiv : astro - ph/0101406 ] .w. h. kinney , phys .d * 72 * , 023515 ( 2005 ) [ arxiv : gr - qc/0503017 ] .h. peiris and r. easther , jcap * 0607 * , 002 ( 2006 ) [ arxiv : astro - ph/0603587 ] .d. s. salopek , j. r. bond and j. m. bardeen , phys .d * 40 * , 1753 ( 1989 ) .j. a. adams , b. cresswell and r. easther , phys .d * 64 * , 123514 ( 2001 ) [ arxiv : astro - ph/0102236 ] . c. ringeval , arxiv : astro - ph/0703486 . g. calcagni , a. r. liddle and e. ramirez , phys .d * 72 * , 043513 ( 2005 ) [ arxiv : astro - ph/0506558 ] .a. r. liddle , p. parsons and j. d. barrow , phys .d * 50 * , 7222 ( 1994 ) [ arxiv : astro - ph/9408015 ] .a. d. linde , phys .d * 49 * , 748 ( 1994 ) [ arxiv : astro - ph/9307002 ] .j. m. bardeen , phys .d * 22 * ( 1980 ) 1882 .a. r. liddle and d. h. lyth , _ cambridge ,uk : univ .( 2000 ) 400 p _ v. f. mukhanov , h. a. feldman and r. h. brandenberger , phys .rept . * 215 * , 203 ( 1992 ) .v. f. mukhanov , sov .jetp * 67 * ( 1988 ) 1297 [ zh .eksp . teor .fiz . * 94n7 * ( 1988 ) 1 ] .v. f. mukhanov , jetp lett .* 41 * ( 1985 ) 493 [ pisma zh .* 41 * ( 1985 ) 402 ] .l. p. grishchuk , sov .jetp * 40 * ( 1975 ) 409 [ zh .* 67 * ( 1974 ) 825 ] .r. easther , b. r. greene , w. h. kinney and g. shiu , phys .d * 66 * , 023518 ( 2002 ) [ arxiv : hep - th/0204129 ] . m. tegmark , a. j. s. hamilton and y. z. s. xu , mon . not .astron . soc .* 335 * , 887 ( 2002 ) [ arxiv : astro - ph/0111575 ] .g. schwartz , ann .statist . * 5 * , 461 ( 1978 ) m. cortes , a. r. liddle and p. mukherjee , phys .d * 75 * , 083520 ( 2007 ) [ arxiv : astro - ph/0702170 ] . c. r. contaldi , m. peloso , l. kofman and a. linde , jcap * 0307 * , 002 ( 2003 ) [ arxiv : astro - ph/0303636 ]. w. h. kinney , phys .d * 58 * , 123506 ( 1998 ) [ arxiv : astro - ph/9806259 ] . l. verde , h. peiris and r. jimenez , jcap * 0601 * , 019 ( 2006 ) [ arxiv : astro - ph/0506036 ] .e. silverstein and d. tong , phys .d * 70 * , 103505 ( 2004 ) [ arxiv : hep - th/0310221 ] .m. alishahiha , e. silverstein and d. tong , phys .d * 70 * , 123505 ( 2004 ) [ arxiv : hep - th/0404084 ] .h. v. peiris , d. baumann , b. friedman and a. cooray , arxiv:0706.1240 [ astro - ph ] .

<S> we investigate the ability of current cmb data to reliably constrain the form of the primordial power spectrum generated during inflation . </S><S> we attempt to identify more exotic power spectra that yield equally good fits to the data as simple power - law spectra . in order to test a wide variety of spectral shapes , we combine the flow formalism , which is a method of stochastic model generation , with a numerical integration of the mode equations of quantum fluctuations . </S><S> this allows us to handle inflation models that yield spectra that are not well described by the standard spectral parameterization . using the latest wmap data - set </S><S> , we find a high degree of variation in possible spectral shapes . </S><S> in particular , we find strongly running spectra arising from fast - rolling inflaton fields providing equally good fits to the data as power - law spectra arising from slowly - rolling fields . </S><S> current data poorly constrains the spectrum on scales @xmath0 , where the error due to cosmic variance is large . among the statistically degenerate models , we identify spectra with strong running on these larger scales , but with reduced running at smaller scales . </S><S> these models predict values for the tensor - to - scalar ratio , @xmath1 , that lie outside the 2-@xmath2 confidence interval obtained from sdss+wmap data for spectra that are parametrized as power - laws or spectra with constant running . by considering more generalized power spectra , </S><S> we therefore open up regions of parameter space excluded for simpler models . </S>