article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
the basic variables of loop gravity are an @xmath0 connection @xmath1 and a densitized triad @xmath2 . these are related to the traditional adm phase - space variables the spatial metric and extrinsic curvature @xmath3 by @xmath4 where @xmath5 , @xmath6 is the inverse of @xmath7 , @xmath8 is the spin connection determined by the triad @xmath7 , @xmath9 , and @xmath10 is the barbero - immirzi parameter@xcite , which usually is fixed through considerations of black hole entropy@xcite .
we seek a condition on these variables equivalent to homogeneity and isotropy that is nevertheless _
diffeomorphism invariant_. an initial data set @xmath11 is homogeneous and isotropic if and only if @xmath12 is _ maximally symmetric _ that is , possesses six killing vector fields and @xmath13 is proportional to @xmath12 by a constant : @xmath14 it is well known that if @xmath12 is maximally symmetric , then its riemann curvature has the form @xmath15 where the scalar curvature @xmath16 is constant over space .
in fact , the converse is also true @xcite .
it follows that a spacetime admits a foliation by maximally symmetric spatial slices if and only if it admits initial data satisfying @xmath17 for some constants @xmath18 and @xmath19 , where @xmath20 is the curvature of @xmath8 and @xmath21 .
the two conditions ( [ triadsymm ] ) can be combined in a useful form for quantization using the complexifier method of thiemann @xcite . fixing a complexifier function @xmath22 on phase space
, one defines a 1-parameter family @xmath23 of complex functions @xmath24 on phase space deriving from a given ( generally real - valued ) function @xmath25 .
importantly , the corresponding family of operators in the quantum theory will have the form @xmath26 , so in principle it is straightforward to quantize such complexified observables .
choosing @xmath22 to be proportional to the total volume @xmath27 of space and setting @xmath28 yields a complexified connection @xmath29 where @xmath30 is a constant with units of inverse length .
the ( co-)triad @xmath31 is invariant under this complexification flow . calculating the curvature of this complexified connection shows that the real and imaginary parts of the single relation @xmath32 where we have set @xmath33 and @xmath34 is a complex constant , imply _ both _ of the real symmetry conditions ( [ triadsymm ] ) .
furthermore , taking a wedge product with @xmath35 , where @xmath36 is an arbitrary smearing function , and integrating over all space gives @xmath37 : = \int f_{ij}\ , \mathbf{f}^i \wedge e^j = \mathbf{b } \int f_{ij}\ , \delta^{ij}\ , \epsilon_{klm}\ , e^k\ , e^l\ , e^m = : \mathbf{b}\ , v[f].\ ] ] the right side here is proportional to the proper - volume integral of the trace of @xmath36 over all space .
cross - multiplying two such equations allows us to eliminate the constant @xmath34 and write the symmetry conditions in the form @xmath38 : = \mathbf{b}[f]\ , v[g ] - \mathbf{b}[g]\ , v[f ] = 0\ ] ] for all @xmath36 and @xmath39 .
the non - complexified version of @xmath40 $ ] is nearly identical to the euclidean self - dual hamiltonian constraint of general relativity , which arises when @xmath41 .
it can be quantized using the same basic method used for the euclidean hamiltonian constraint .
finally , we can complexify the result in the quantum theory by setting @xmath42 : = \mathrm{e}^{-\hat v / v_0}\ , \hat b[f]\ , \mathrm{e}^{\hat v / v_0},\ ] ] where @xmath43 is an fixed , but arbitrary , constant with units of volume . the functions in equation ( [ sdef ] ) are our constraints , whose vanishing is equivalent to homogeneity and isotropy of the gravitational field with respect to _ some _ action of a maximal symmetry group .
they are manifestly diffeomorphism and gauge covariant , so that their vanishing for _ all _ test functions is a diffeomorphism and gauge _ invariant _ condition .
a necessary condition for a given set @xmath44 of classical constraints to admit operator analogues @xmath45 that can be imposed strongly ( _ i.e. _ , @xmath46 ) and consistently ( _ i.e. _ , simultaneously ) at the quantum level is that their classical poisson algebra should close .
that is , there should exist phase - space functions @xmath47 such that @xmath48 we refer to such constraints as a _ first - class set_. we wish to impose the symmetry conditions @xmath49 $ ] discussed above strongly in quantum theory , as well as the gauss and diffeomorphism constraints @xmath50 $ ] and @xmath51 $ ] of general relativity .
remarkably , these functions _ do _ form a first class set classically .
the first thing to check is that the symmetry conditions form a first class set by themselves .
one finds @xmath52 , \mathbf{s}[h , k ] \bigr\ } & = \frac{\kappa\gamma}{2}\ , v[g]\ , \mathbf{s}[f { \mathop{\#}}h , k ] + \frac{\kappa\gamma}{2}\ , \mathbf{s}[g , k ] \int \operatorname{tr}f\ , e^k \wedge \mathbf{d } \bigl ( h_{kl}\ , e^l \bigr ) \\&\hspace{10em}\notag - ( f \leftrightarrow g ) - ( h \leftrightarrow k ) + ( fh \leftrightarrow gk ) , \end{aligned}\ ] ] where @xmath53 denotes the gauge - covariant derivative defined by @xmath54 , and where we have defined the `` hash product '' @xmath55 of smearing functions . the next thing to check is the algebra of the symmetry conditions and the constraints .
one finds immediately that @xmath56 , \mathbf{s}[f , g ] \bigr\ } = 0 \qquad\text{and}\qquad \bigl\ { c[\vec n ] , \mathbf{s}[f , g ] \bigr\ } = \mathbf{s}[\mathcal{l}_{\vec n } f , g ] + \mathbf{s}[f , \mathcal{l}_{\vec n } g]\ ] ] because @xmath57 $ ] is ( gauss ) gauge - invariant and diffeomorphism - covariant .
the total algebra of symmetry conditions and constraints therefore closes .
thus , it is reasonable to seek a sector of ( distributional ) states in loop quantum gravity lying simultaneously in the kernels of all the symmetry condition operators , as well as all the gauss and diffeomorphism constraint operators .
to construct an operator corresponding to the smeared constraint function @xmath49 $ ] ( [ sdef ] ) , one first constructs operators corresponding to @xmath58 $ ] and @xmath59 $ ] in ( [ bvdef ] ) .
this is a simple task in the framework of loop quantum gravity .
first , the standard quantization of the volume element given in ref . yields a canonical definition for @xmath60 $ ] .
second , as we mentioned above , @xmath58 $ ] has a form closely related to that of the euclidean self - dual hamiltonian constraint @xmath61 = \int n f^i \wedge e_i , \end{aligned}\ ] ] where @xmath62 is the curvature of @xmath1 . specifically ,
@xmath58 $ ] differs only in that ( a ) the two indices are no longer contracted with each other , but rather with those of a matrix - valued smearing function , and ( b ) one uses the complexified connection @xmath63 instead of the real connection @xmath1 .
neither of these differences affect thiemann s procedure @xcite for quantizing @xmath64 $ ] .
exactly the same methods can be used to define an operator @xmath65 $ ] .
one can either carry out the thiemann strategy using the quantum complexified connection directly , or first using the quantum real connection and then complexifying as in ( [ complexifyb ] )
. the resulting operator in either case will be the same due to the structure of quantum complexification .
there is a further subtlety to be addressed , even with @xmath65 $ ] and @xmath66 $ ] constructed . as in the case of thiemann
s hamiltonian constraint , the operator @xmath65 $ ] has a well - defined action only within the space of solutions @xmath67 to the quantum diffeomorphism constraint .
furthermore , as with thiemann s constraint ( when it is smeared with a non - constant lapse ) , @xmath65 $ ] maps diffeomorphism - invariant states out of @xmath67 .
that is , its domain and range are mutually exclusive .
in contrast , the smeared volume operator @xmath66 $ ] is well - defined on _ all _ states , diffeomorphism invariant or not , but generally maps diffeomorphism - invariant states out of @xmath67 .
these facts determine the operator ordering to use when building the operators @xmath68 $ ] analogous to ( [ sdef ] ) .
namely , if we choose the domain of @xmath68 $ ] to be @xmath67 , then we must order @xmath65 $ ] to act before @xmath69 $ ] , and likewise @xmath70 $ ] to act before @xmath66 $ ] .
the space of states annihilated by @xmath71 $ ] for all smearing functions @xmath72 and @xmath73 then defines our _ homogeneous and isotropic sector _
the gauss and diffeomorphism constraints hold identically in loop quantum cosmology , and also in the new homegeneous and isotropic sector of loop quantum gravity described above . the only remaining constraint is the ( lorentzian ) hamiltonian constraint @xmath75 $ ] .
furthermore , there is only one independent smearing of the hamiltonian constraint that needs to be imposed in the homogeneous and isotropic context , which can be taken to be the unit - lapse smearing @xmath76 $ ] .
there exists a quantum analogue of @xmath76 $ ] in both loop quantum cosmology and loop quantum gravity , which we denote @xmath77 and @xmath78 $ ] , respectively .
one of the key aims of our program is to give a direct relation between these two operators at the quantum level .
this part of our program remains incomplete , but the general strategy we intend to follow is to construct an embedding @xmath79 of lqc states into @xmath80 such that , at least in an approximate sense , the matrix elements of the two hamiltonian constraint operators match : @xmath81 \iota \phi \rangle } { \sqrt{\langle \iota \psi , \iota \psi \rangle \langle \iota \phi , \iota \phi \rangle } } = \frac{\langle \psi , \hat{c}_s \phi \rangle } { \sqrt{\langle \psi , \psi \rangle \langle \phi , \phi \rangle}}.\end{aligned}\ ] ] this condition should hold for all lqc states @xmath82 and @xmath83 .
the division by the norms here is needed because @xmath79 will in general not be norm - preserving , so that @xmath83 , @xmath82 , @xmath84 and @xmath85 will in general not be simultaneously normalizable . the extent to which the above condition can be satisfied for some @xmath79 is then a measure of the extent to which the two definitions of dynamics are consistent .
alternatively , @xmath79 can be determined by imposing a similar equality of matrix elements for one or more pairs of simpler operators .
then , given an operator @xmath78 $ ] in the full theory , equation ( [ matel ] ) will uniquely determine a corresponding operator in lqc , which can then be compared with the hamiltonian constraint operator used in the lqc literature .
in practice , the above prescriptions are more involved than they sound , as @xmath80 is expected to consist in states which are not normalizable in the diffeomorphism invariant inner product on @xmath67@xcite , so that the left hand side of the above equation will likely be of the form @xmath86 and so will need to be regularized .
we refer to the larger paper @xcite for such details .
even though the program of section [ dynamics ] has not been carried to completion in relating the full theory of lqg to lqc , it _ has _
been completed@xcite to impose isotropy at the quantum level in a bianchi i model @xcite .
this is a simpler , but still non - trivial , problem .
this application allows a detailed comparison of the isotropy - reduced quantum dynamics of bianchi i @xcite with that of the well - established isotropic lqc model @xcite .
the results were unexpected .
we found an exact match of kinematics and dynamics in the two theories , as well as an exact match of naturally defined operators in each model corresponding to spatial volume , average spatial curvature , and hubble rate .
furthermore , the embedding @xmath79 in this application to bianchi i turned out to be none other than the adjoint of the dynamical projector @xmath87 from bianchi i to isotropic lqc proposed by ashtekar and wilson - ewing in ref . .
that is , the image of the adjoint of their dynamical projector satisfies precisely the symmetry constraint operator equation proposed in this paper , specialized to bianchi i in a straightforward way .
if one includes scalar matter in the analysis , all of the above statements are again true , as long as the matter hamiltonian depends on the geometry only via the volume , as is the case for the matter considered in ref . .
in light of the successful application to bianchi i , and of the equivalence to the t is dynamical projector method of ref . in that case , one can view the present program as way to reformulate the dynamical projector method and generalize it to the full theory of lqg .
what makes this possible is the sharpened understanding of the role of homogeneity and isotropy .
a further benefit of the application to bianchi i is that , in matching operators in the two models , such as those corresponding to spatial curvature and hubble rate , the ambiguity in how the operators are quantized is necessarily restricted .
if one similarly tries to match operators corresponding to , for example , components of the four dimensional ricci tensor , one again expects a preferred class of quantizations to be selected .
it is worth checking whether the use of such quantizations of ricci might resolve the issue of apparent non - zero ricci components in vacuum discussed in ref . .
the authors are grateful to ted jacobson , atousa chaharsough shirazi , brajesh gupt , jorge pullin , parampreet singh , xuping wang , and shawn wilder for discussions .
this work was funded in part by the u.s .
national science foundation under grants phy-1205968 and phy-1505490 , and by nasa through the university of central florida s nasa - florida space grant consortium . | this paper summarizes a new proposal to define rigorously a sector of loop quantum gravity at the diffeomorphism invariant level corresponding to homogeneous and isotropic cosmologies , thereby enabling a detailed comparison of results in loop quantum gravity and loop quantum cosmology .
the key technical steps we have completed are ( a ) to formulate conditions for homogeneity and isotropy in a diffeomorphism covariant way on the classical phase space of general relativity , and ( b ) to translate these conditions consistently using well - understood techniques to loop quantum gravity . some additional steps , such as constructing a specific embedding of the hilbert space of loop quantum cosmology into a space of ( distributional )
states in the full theory , remain incomplete . however , we also describe , as a proof of concept , a complete analysis of an analogous embedding of homogeneous and isotropic loop quantum cosmology into the quantum bianchi i model of ashtekar and wilson - ewing .
details will appear in a pair of forthcoming papers .
[ [ section ] ] symmetry is a powerful tool for simplifying the physics of a system in order to extract predictions . in the case of quantum gravity ,
the most promising place to seek potentially observable predictions is in cosmology , where classical gravity breaks down at the big bang and quantum effects become important .
the universe appears spatially homogeneous and isotropic to a very good approximation at large scales , and cosmological models exploit this symmetry to enable exact solutions of the otherwise prohibitively complicated dynamics of einstein s theory . imposing homogeneity and isotropy , or indeed any spatial symmetry , in _ quantum _ gravity
, however , immediately runs into two obstacles : 1 .
symmetry must be imposed on the full set of initial data both configuration and momentum variables .
classically this is necessary for the future evolution to continue to be symmetric .
however , at the quantum level , this violates the heisenberg uncertainty principle .
2 .
usually symmetry is imposed by requiring invariance of the initial data under a fixed action of the symmetry group .
however , such a method of imposing symmetry necessarily breaks the diffeomorphism invariance of general relativity . as a consequence , in the quantum theory , such a method of imposing symmetry will never be consistent with imposition of the dynamical constraints . in this paper
, we introduce a way to impose symmetry in quantum gravity that overcomes these fundamental obstacles .
specifically , we formulate new _ diffeomorphism covariant _ constraints that impose homogeneity and isotropy in the phase space of general relativity .
these constraints involve complex functions on phase space , which have a _ closed _ poisson algebra .
the latter fact allows us to overcome the first obstacle mentioned above in a manner similar to that used in the gupta
bleuler quantization of the electromagnetic field@xcite .
furthermore , the complex constraint functions can be quantized using existing techniques in the framework of loop quantum gravity .
this yields for the first time a systematically motivated homogeneous and isotropic sector at the diffeomorphism - invariant level of the quantum theory .
our ultimate goal is to use this homogeneous and isotropic sector of loop quantum gravity to develop a clear and detailed connection to homogeneous and isotropic loop quantum cosmology .
the latter arises by imposing the symmetry classically , prior to quantization .
the resulting quantum model is much more tractable than the full theory , and allows explicit comparison between theoretical predictions and observations , but loses any _ a priori _ connection to the full quantum theory .
our aim is to restore that connection , so that observational constraints on a quantum cosmological model can also help dictate choices that must be made in quantizing the full theory . in particular
, one would like to be able to use constraints on the dynamics of the quantum cosmological model to similarly constrain the dynamics of the full theory .
some steps of our program to relate a symmetric sector of loop quantum gravity to loop quantum cosmology remain incomplete , however .
we will summarize the steps that have been completed below , and mention those that are the subject of ongoing work .
we have , however , completed the analogous task relating the loop quantum bianchi i model of ashtekar and wilson - ewing @xcite to homogeneous and isotropic loop quantum cosmology .
we were surprised to find that the homogeneous and isotropic model could be embedded in the bianchi i model using a known linear mapping .
specifically , the embedding is the adjoint of the projection from bianchi i to the fully symmetric model previously found by ashtekar and wilson - ewing .
we will discuss this simplified embedding scheme , both as a proof of concept for our larger goal and as an interesting result in its own right .
details of our results will be published in two forthcoming papers @xcite . |
the causal dynamical triangulation ( cdt ) model was first introduced in ref . as a non - perturbative theory of quantum gravity . the basic idea behind this model is to interpret the gravitational path - integral by discretizing spacetime manifolds through so - called _
causal triangulations_. thus , the problem of computing the gravitational path - integral is replaced by a statistical mechanical problem , that is the study of the partition function of an ensemble of finite random triangulations , whose weight is given by the boltzmann factor @xmath0}$ ] , where @xmath1 $ ] is the discrete einstein - hilbert action .
having described the partition function , one can study its `` thermodynamic '' limit , when the size of triangulations grows to infinity .
we refer the reader to ref . , where the subject is treated in detail .
even in two dimensions , the cdt model turns out to possess a rich mathematical structure .
in particular , it has been proved ( see refs . )
that it is critical at a certain ( known ) value @xmath2 of the cosmological constant .
this critical behaviour is reflected in the geometrical properties ( hausdorff and spectral dimensions ) of the infinite triangulations : they show 2-dimensional features at criticality , whereas in the subcritical regime they are essentially 1-dimensional .
matter is inserted in the theory by running a statistical mechanical model on the random triangulation .
technically , this can be implemented by weighting the boltzmann factor associated to each triangulation @xmath3 with the partition function of an ising model running on @xmath3 .
this is generally called _ annealed _ coupling ( see ref . for further details ) . as in the pure gravity case
, the coupled model is believed to become critical when the parameters of the model ( the cosmological constant @xmath4 and the inverse temperature @xmath5 of the spin system ) are suitably tuned .
however , the exact value of the critical parameters are still unknown and only bounds on the region where they lie have been provided ( see refs . and ) .
in this paper we review some recent results , first appeared in ref .
, on the annealed coupling between the ising model and the random causal triangulations of the plane .
in particular , we show the existence of a critical line in the positive @xmath6 quarter plane , give a description of some of its analytical properties and provide bounds on the region where the critical line is located .
furthermore , we add some new results on the relation between the critical line and its dual .
[ dm ] a _ causal triangulation _ @xmath3 is a rooted planar locally finite connected graph satisfying the following properties . 1 .
the set of vertices at graph distance @xmath7 from the root , together with the edges connecting them , form a cycle , denoted by @xmath8 and called @xmath7-th _ slice _ ( when there is only one vertex the corresponding cycle has only one edge , i.e. it is a loop ) .
2 . all faces of the graph are triangles , with the only exception of the external face .
one edge attached to the root vertex is marked , we call it _ root edge_. the last condition in the above definition is a technical requirement , needed to cancel out possible rotational symmetries around the root .
we denote by @xmath9 the set of causal triangulations with @xmath10 slices and fixed number @xmath11 of vertices on @xmath12 . in the following
, we shall denote by @xmath13 , @xmath14 and @xmath15 , the vertex set , the edge set and the triangles set of @xmath3 , respectively .
the cardinality of a set @xmath16 will be denote by @xmath17 . given a triangulation @xmath18 ,
a spin configuration @xmath19 on @xmath3 is an element of the set @xmath20 the partition function of the coupled system ( ising model and random causal triangulations ) is defined as @xmath21 where @xmath22 is the inverse temperature , @xmath23 is the cosmological constant and @xmath24 is the classical nearest - neighbour spin interaction hamiltonian .
since the set @xmath9 is countable , the series defined in ( [ z ] ) might diverge . on the other hand , for any fixed @xmath22 , each term of the series is a positive and strictly decreasing function of @xmath4 , therefore there exists a value @xmath25 , such that @xmath26 , for any @xmath27 , and @xmath28 , for any @xmath29 .
it is easy to see that the sequence @xmath25 is bounded from above .
indeed , from the inequality @xmath30 it follows that @xmath31 where the series on the right is finite ( see ref . ) for all @xmath10 and @xmath11 whenever @xmath32 .
therefore @xmath33 in the following theorem , we prove the existence of a critical value @xmath34 for each @xmath35 , above which the partition function is finite for any @xmath10 and @xmath11 .
[ te1 ] for each @xmath22 and @xmath36 , as @xmath37 @xmath38 where the limit exists and does not depend on @xmath11 .
furthermore , the function @xmath39 is non - decreasing in @xmath5 .
first we show that for each @xmath22 and @xmath36 function @xmath25 is non - decreasing in @xmath10 .
consider the partition function in ( [ z ] ) for the triangulations with a fixed last slice @xmath12 with @xmath40 vertices .
it is straightforward to derive the following bound @xmath41 hence , if @xmath42 for some @xmath11 then also @xmath43 .
this implies for any @xmath44 @xmath45 when @xmath46 this together with the uniform bound ( [ u ] ) yields existence of @xmath47 and then also by ( [ z3 ] ) for any fixed @xmath48 @xmath49 as well as @xmath50 we conclude that for any fixed @xmath40 there exists @xmath34 such that @xmath51 this proves statement ( [ t1 ] ) . to prove the monotonicity of @xmath34 consider again @xmath52 .
assume , @xmath53 then @xmath54 but for any @xmath18 and @xmath55 @xmath56 by the first griffiths inequality for ferromagnetic spin systems .
hence , @xmath57 i.e. , for any @xmath58 @xmath59 which in turn yields @xmath60 then monotonicity of @xmath34 follows from here by the statement ( [ t1 ] ) .
since the function @xmath34 is non - negative and bounded from above by a linear function ( see ( [ u ] ) ) , its monotonicity implies that it has at most a countable number of jumps .
therefore we shall call the graph of @xmath34 , @xmath22 , _ the critical line_. theorem [ te1 ] implies that @xmath61 , and therefore by a theorem of ref .
proved for @xmath62 , we immediately have the following result .
[ thm : cr_bounds ] the critical line @xmath34 is monotone non - decreasing , it has at most a countable number of jumps and it is contained in the region of the @xmath6-plane defined by @xmath63 where @xmath64 above the critical line , that is in the region @xmath65 the partition function is finite for all @xmath10 ( and @xmath11 ) .
we call this
_ subcritical region_. on the other hand , below the critical line , for each @xmath6 , @xmath66 for @xmath10 large enough . furthermore , from thm .
[ thm : cr_bounds ] it follows that the asymptotic behaviour of @xmath34 is given by @xmath67 in particular , eq .
( [ eq : crl_inf ] ) shows that the ground - state spin configurations are dominant at low temperature , as expected ( see ref . ) .
in this section we compare the partition function ( [ z ] ) , to the partition function of the system constructed by placing the ising spins on the faces of a causal triangulation , that is on the vertices of the dual graph @xmath68 of a causal triangulation @xmath3 .
hence , the partition function of the dual model is defined as @xmath69 applying a kramers - wannier duality argument , it can be shown ( see ref . ) that the partition functions of the two systems are related by the equation @xmath70 where @xmath71 and @xmath72 are such that @xmath73 equation ( [ eq : zduz ] ) implies that the subcritical region of the original model is bijectively mapped onto the subcritical region of the dual model by the transformations ( [ eq : tempdual])-([eq : cosmdual ] ) .
hence , in particular the critical lines of the two models are related by the equation @xmath74 by the relation ( [ eq : mucrdual_mucr ] ) , thm .
[ thm : cr_bounds ] implies the following
. the dual critical line @xmath75 is monotone non - decreasing , it has at most a countable number of jumps and it is contained in the region of the @xmath76-plane defined by @xmath77 where @xmath78 in particular , we have the following asymptotic results @xmath79
in this paper we presented some results on the annealed coupling of an ising model and the causal dynamical triangulation model .
we proved the existence of a critical line @xmath80 delimiting the region of the @xmath6-plane where the partition function @xmath81 of the system is finite for all @xmath10 and @xmath11 and showed that the critical line is a monotone non - decreasing function of @xmath82 and has at most a countable number of jumps .
moreover , the critical line is proved to lie in a region of the plane shrinking to a straight line when @xmath83 .
this gives in turn the exact asymptotic behaviour of the critical line and shows that ground - state spin configurations are dominant at low temperature .
in addition , we obtained exact relations between the partition functions of the model itself and its dual , as well as between their critical lines . | in this paper we study the annealed coupling of an ising model with 2-dimensional causal dynamical triangulation model . after a short review of previous results ,
we prove the existence of the so - called critical line and derive its analytical properties and asymptotics .
in addition , relations between the model and its dual are investigated . |
the cosmic star - formation history in the early universe has been studied in optical / near - infrared ( nir ) wavelengths , which trace ultraviolet ( uv ) radiation in rest - frame at high redshifts ( e.g.,@xcite ) . the uv star formation rate density ( sfrd )
does not account for all components of star - forming galaxies ( e.g.,@xcite ) .
recent studies suggest that far - infrared ( fir ) sfrd contributes more than half of the total at @xmath13 ( e.g.,@xcite ) .
millimeter / submillimeter ( mm / submm ) galaxy surveys would be , in principle , efficient to probe the dust - obscured component of sfrd at high - redshift .
the advantage of such observations in mm / submm is the well - known negative @xmath14-correction ; the continuum flux of a typical star - forming galaxy of fixed sfr remains approximately constant with increasing redshift @xcite .
however , it is often difficult to estimate redshifts for very faint and dusty sources ( e.g.,@xcite ) . strong emission lines such as [ cii]158@xmath15 m or [ oiii ] 88@xmath15 m lines can be used to study the sfr and gas properties of high-@xmath16 star - forming galaxies as well as to determine their spectroscopic redshifts ( e.g. , ) .
interestingly , @xcite report that lyman - break galaxies ( lbgs ) at @xmath17 show enhancement of [ cii ] emission relative to the fir continuum compared with mm / submm - selected galaxies .
they also serendipitously detected a [ cii ] emitter which is faint in both the rest - uv and fir continuum . combining observations in rest - uv
, fir and mm / submm emission lines appears to be essential to understand the physical properties of galaxies at high redshifts ( e.g.,@xcite ) .
one of the brightest submm emission lines is [ cii ] ( e.g. , ) .
carbon in the interstellar medium is largely in a singly ionised state in a variety of environments , from hii regions to molecular clouds , because the ionization potential of atomic carbon is 11.3 ev , lower than that of hydrogen and close to dissociation energy of co of 11.1 ev ( e.g.,@xcite ) .
the critical density of [ cii ] emission is about @xmath18 @xmath19 , and thus [ cii ] emission can arise even in a molecular cloud with temperature around 92 k @xcite
. therefore [ cii ] radiative cooling often dominates in regions with a wide range of densities ( e.g.,@xcite ) .
finally , [ cii ] emission is thought to be a potential tracer of sfr because of its main origin of photo - dissociated region associated with young , massive stars ( e.g.,@xcite ) .
an important observational advantage is that [ cii ] line emission at @xmath20 is redshifted to wavelengths with low atmospheric absorption and thus it is possible to detect [ cii ] line emission even from galaxies at @xmath21 ( e.g.,@xcite ) .
a number of high - redshift [ cii ] emitters are expected to be detected with forthcoming high sensitivity observations with the atacama large millimeter / submillimeter array ( alma ) ( e.g.,@xcite ) . in this paper
, we present a blind search for [ cii ] emitters using alma cycle 2 data @xcite .
we briefly introduce the observations in 2 .
the details of our data analysis is described in 3 .
then we show the results in 4 and discuss the implications for cosmic star formation history in 5 .
we summarize the results and discussions in 6 . throughout the paper ,
we adopt the standard @xmath22cdm cosmology with the matter density @xmath23 , the cosmological constant @xmath24 , the hubble constant @xmath25 in the unit of @xmath26 .
all magnitudes are given in the ab system , unless otherwise noted .
we calculate sfr assuming chabrier initial mass function ( imf ) @xcite , with an integration range from 0.08 @xmath27 to 100 @xmath27 .
when needed , we use the conversion factor of 1.8 from the chabrier imf to the equivalent salpeter imf @xcite and 1.1 from the chabrier imf to the kroupa imf @xcite .
we analyze data from the alma deep field survey of ssa22 ( adf22 ) observed in band 6 in alma cycle 2 in june 2014 and april 2015 ( proposal i d 2013.1.00162.s , pi : h. umehata ) .
the details of the observation are described in @xcite .
adf22 is a survey field with an area of @xmath28 centered on a @xmath29 proto - cluster ; ra ( j2000 ) = 22@xmath3017@xmath3134@xmath32 , dec ( j2000 ) = + 00@xmath3317@xmath3400@xmath1 consisting of 103 pointing fields .
the field was observed using four 1.875 ghz spectral windows ( spw ) with the central frequency of 263 ghz , which corresponds to the [ cii ] redshift of @xmath35 .
the typical angular resolution of a combined data is @xmath36 corresponding to @xmath37 kpc at @xmath35 .
the on - source time per pointing in the fields is 4.5 min .
the data observed in 2014 and 2015 have angular resolution of @xmath38 and @xmath39 , and on - source time per pointing in the fields of 2.5 min . and 2.0 min . for 2014 and 2015 , respectively .
the four spws have root - mean - square ( rms ) noise level of 0.7 , 0.7 , 0.8 or 0.9 mjy beam@xmath2 at a 36 km s@xmath2 velocity resolution .
the rms of each spw at 36 km s@xmath2 resolution of combined , 2014 and 2015 data as a function of the observed frequency are shown in figure 1 , where no significant atmosphere absorption is seen .
other properties of the data are listed in table 1 . in order to search faint emission line sources , we use high sensitivity data of 80 pointing fields ; field 1 - field 80 and
search in a rectangle area of @xmath40 5 arcmin@xmath0 ; ( ra ( j2000 ) , dec ( j2000 ) ) = ( 22@[email protected]@xmath32 , + 00@[email protected]@xmath1 ) to ( 22@[email protected]@xmath32 , + 00@[email protected]@xmath1 ) , and a frequency coverage of 253.1272.8 ghz ( table 1 ) .
the effective survey area corresponds to about 29 comoving mpc@xmath0 and the effective survey volume is @xmath40 2.2 @xmath41 @xmath42 comoving mpc@xmath43 at @xmath35 .
the flowchart of our source selection method is shown in figure 2 .
the data are analyzed with common astronomy software application ( casa ) ver.4.5.3 @xcite .
we make continuum - subtracted datacube by using uvcontsub and clean .
we first spectrally smooth the data to obtain high signal - to - noise ( s / n ) ratios .
the top - hat spectral smoothing window is set to be 0 , 2 , 4 , ... , 12 , 15 , 18 , ... , 21 slices , with a slice width corresponding to @xmath4018 km s@xmath2 .
we use the spectral smoothing function `` boxcar '' so that the velocity sampling of the output data is kept constant . as
each spectral data slice has a different rms value as shown in figure 1 , we normalise each slice by its rms .
we call a datacube thus - generated as `` s / n cube '' .
we use clumpfind @xcite to search emission line sources in the s / n cube .
we search for sources with a threshold value `` low '' of clumpfind of @xmath44 4.5 .
we then do ` matching ' of the clumps detected at the same position between the s / n cubes in the same spw with different resolutions and retain the clump that has maximum s / n ratio ( see also table 1 ) .
we select clumps that have the s / n ratio larger than @xmath45 and also larger than the maximum negative s / n ratio measured in the inverted s / n cube in each spw ( see also figure 2 ) , in order to avoid contamination by spurious sources ( e.g.,@xcite ) .
we also check line spectral features of the detected clumps ( sources ) in the datacube separately for those observed in 2014 and 2015 . for the detected sources , we search for their counterparts in @xmath46 band taken with the canada france hawaii telescope / megacam obtained by archival data @xcite , @xmath47 , @xmath48 , @xmath49 , @xmath50 , @xmath51 , nb912 , @xmath52 , @xmath53 , @xmath54 band taken with the subaru telescope @xcite , 3.6 @xmath55 m , 4.5 @xmath55 m , 5.8 @xmath55 m , 8.0 @xmath55 m , 24 @xmath55 m taken with the _ spitzer _ space telescope / irac and mips @xcite 0.5 kev , 2 kev and 8 kev taken with the _ chandra _ x - ray observatory @xcite .
[ cii ] line emitting galaxies at @xmath56 are likely to be detected only longward of @xmath51 band and/or in narrow - band nb912 if they are lbgs or ly@xmath57 emitters ( laes ) ( e.g.,@xcite ) , although the available @xmath51 band and nb912 data could be too shallow for high - redshift [ cii ] emitters in our blind search . for the sources with counterparts , we estimate either their photometric redshift by means of spectral energy distribution ( sed ) fitting or spectroscopic redshift by assuming their line species .
sed fitting is calculated by using hyperz software @xcite . in
4.3 , we also use the equivalent width and the source number density to consider if the detected [ cii ] emitter candidates are other line emitters . at each step is given on the right .
we select the targets by setting a s / n threshold @xmath45 in each spw , and then by imposing that their s / n are larger than the maximum negative s / n ratio ( see also table 1 ) .
finally , the selection leaves four clumps as line emitting galaxy candidates .
, width=321 ]
we detect four line emitter candidates .
hereafter , we call the two sources without optical , nir and fir counterparts adf22-linea and adf22-lineb . those with counterparts are dubbed as adf - linec and adf - lined .
the peak s / n ratio are @xmath58 , @xmath59 , @xmath60 and @xmath58 , for adf22-linea , b , c and d , respectively
. the first moment images of the candidates are shown in figure a , and their properties are shown in table 2 .
figure 3 shows the cumulative number of positive and negative clumps as a function of s / n ratio .
although the s / n ratios of adf22-linea , b and d are below @xmath3 at the original spectral sampling , the lines are detected at @xmath61 in the smoothed s / n cubes .
we compare the spectral line features of the emitter candidates in different observation epochs , 2014 and 2015 ( figure 4 ) .
overall , the contiguous positive signals over a velocity range of @xmath44 180 km s@xmath2 and the line features commonly seen suggest that the candidates are likely real sources .
we note here that we also detect one clump with @xmath62 in the inverted s / n cube , and thus we would naively be concerned that one candidate with @xmath59 could be a spurious source .
however , the most - negative clump is actually detected in spw 3 , where none of our four candidates is located .
we also find that datacubes with higher rms value have higher maximum negative detection ( table 1 ) . since the datacubes in different spws have different properties , the existence of the high-@xmath63 negative clump in spw 3 does not immediately impacts the confidence of our line emitter candidates .
adf22-lineb has a lower s / n ratio than adf22-linea , whereas it has non - negative @xmath51 band counterparts with @xmath64 ( see also figure 5 ) .
velocity - gradient is also seen around adf22-lineb ( see also figure a ) .
figure 5 shows the images of the four candidates in @xmath47 , @xmath48 , @xmath51 , 3.6 @xmath55 m and 1.1 mm wavebands .
we plot sed and model fit for adf22-lined in figure b , and the measured photometry in the detected bands are given in table 3 .
the photometric redshift is estimated by using hyperz @xcite .
we fit the sed templates by @xcite to the spectral coverage from uv to 8 @xmath65 , assuming a calzetti dust extinction law @xcite .
we also use sed templates from swire library @xcite .
* adf22-linea and b : * we do not find any secure counterpart nor close sources within 2@xmath1 of the sources .
therefore we regard linea and lineb as good [ cii ] emitter candidates .
* adf22-linec : * linec very likely arises from the galaxy adf22.4 reported in @xcite , whose redshift is determined to be @xmath66 from far - infrared spectroscopic follow up observations ( umehata et al . in prep . ) .
thus we identify adf22-linec as co(9 - 8 ) line emission at @xmath66 .
in addition , the optical component near adf22-linec is a known galaxy at @xmath67 @xcite , but we exclude possibility of adf22-linec to be at @xmath67 because there is no obvious line species observed at 1.1 mm .
linec is also detected in x - ray @xcite , which may indicate that adf22-linec is an agn - host galaxy .
further details of the galaxy will be discussed in umehata et al.(in prep . ) .
* adf22-lined : * lined is spatially consistent with the position of a tentatively detected continuum source adf22.21 reported in @xcite .
the result of sed fitting shows that reduced @xmath68 values becomes minimum at @xmath69 ( figure b left ) .
interestingly , the sed is well fitted by that of arp220 placed at @xmath70 ( figure b right ) . by searching for possible lines in this redshift range ,
we conclude that adf22-lined is likely a co(4 - 3 ) emitter at @xmath71 . besides the [ cii ]
line emission , there are also possibilities that adf22-linea and lineb are other emission line sources , such as @xmath72co line emission at @xmath73 , h@xmath74o at @xmath75 or @xmath76 , [ nii]205@xmath65 at @xmath77 , [ oi]145@xmath65 at @xmath78 , [ nii]122@xmath65 at @xmath79 or [ oiii]88@xmath65 at @xmath80 @xcite
. if adf22-linea and lineb are @xmath72co emitters , the number density is consistent with the result of aspecs survey @xcite and with semi - analytical / empirical predictions referred to the article @xcite .
thus we can not exclude the possibility of @xmath72co emitters by the discussion of detectability .
we compare the equivalent widths ( ews ) in observed frame of the four candidates .
the estimated ews are @xmath81 8.6 , @xmath81 14.6 , 1.1 and 7.3 @xmath65 for adf22-linea , b , c and d , respectively , assuming @xmath82 continuum flux limit .
adf22-linea and b have higher ew than the blindly detected @xmath72co emitters in our survey .
the left and middle panels of figure 6 also show the distribution of the ews in 0.91.3 mm observed frame of the four candidates , high - redshift [ cii ] emitting lbgs and laes @xcite , [ cii ] emitter candidates detected in aspecs @xcite , @xmath72co emitter candidates detected in band 6 in aspecs @xcite .
the ews of adf22-linea , b and other high - redshift [ cii ] emitter / candidates are comparable . given these information , we argue that adf22-linea and b are more likely [ cii ] emitters at @xmath83 and @xmath84 , rather than co emitters at @xmath73 .
ew values of adf22-linea and b are comparable to those of the blindly detected co emitters .
further consideration by using forthcoming follow up observation and theoretical study will be needed to yield any insight about the trends of ew distributions . as with adf22-linea and b ,
blindly detected line emitter candidates are expected to have often no counterpart @xcite .
thus it is important to study the ews of a large sample of co/[cii ] emitters .
we note that h@xmath74o molecular lines are expected to have similar line flux to co line emission in the submm band ( e.g,@xcite ) , and thus can be distinguished from high - redshift [ cii ] emitters by comparing their ews .
[ cii ] luminosity , @xmath85}$ ] of adf22-linea and b is calculated by using luminosity distance @xmath86 , observed frequency @xmath87 , velocity - integrated flux @xmath88 ( e.g.,@xcite ) , @xmath89}}{l_\odot } = 1.04 \times 10^{-3 } \left ( \frac{d_{\rm l}}{\rm mpc } \right ) ^2 \frac{\nu_{\rm o}}{\rm ghz } \frac{s^{\rm v}}{\rm jy~km~s^{-1}}.\ ] ] the estimated @xmath85}$ ] of @xmath90 is consistent with the values of normal star - forming galaxies in the local universe ( e.g.,@xcite ) , thus we do not consider the effect of [ cii ] line deficit ( e.g.,@xcite ) .
we then derive [ cii ] luminosity function by using sfr-[cii ] luminosity relation @xcite and sfr function at @xmath91 @xcite .
the right panel of figure 6 shows that the detection of one [ cii ] emitter candidate in the survey area is roughly consistent with the expected [ cii ] number counts , if we use the sfr-@xmath85}$ ] relation by @xcite that is calibrated from observations of nearby low - metallicity dwarf galaxies ( see also 5 ) : @xmath92}}{\rm m_\odot yr^{-1 } } = 10^{-5.73 \pm 0.32 } \left ( \frac{l_{\rm [ cii]}}{\rm l_\odot}\right ) ^{0.80 \pm 0.05}.\ ] ] we also plot the predicted number counts of [ nii ] 122@xmath55 m and [ nii ] 205@xmath65 from the model of orsi et al .
the predicted number count of [ oiii ] 88@xmath55 m emission at @xmath80 are lower than the [ nii ] 122@xmath55 m emission @xcite .
it is expected that such line emitters will not be found in our survey area . from the discussion above
, we assume adf22-linea and b to be [ cii ] emitter candidates .
in order to discuss the cosmic star formation history , we derive the sfrs of adf22-linea and adf22-lineb assuming that they are [ cii ] emitters at @xmath91 .
we calculate total sfr by summing up the dust - uncorrected sfr@xmath93 and sfr@xmath94 , sfr@xmath95(e.g.,@xcite ) , by using the following equations @xcite ; @xmath96 where l@xmath97 refers to the uv luminosity density in the wavelength range 1500 - 2800 , and l@xmath94 refers to the ir luminosity integrated over 8 - 1000 @xmath55 m .
we estimate @xmath98 of adf22-linea , b and d from the observed 1.1 mm continuum fluxes by using the sed fitting method of @xcite .
continuum upper limits of linea and b are assumed to be @xmath82 .
@xmath98 of adf22-linec is referred to estimation of @xcite .
upper limit of uv luminosity is estimated in @xcite .
the obtained sfr@xmath99 is @xmath100 30 @xmath101 , being consistent with the sfr-@xmath85}$ ] relation of @xcite that is calibrated by local low - metallicity dwarf galaxies ( figure 6 left ) .
we note that the relation calibrated by high-@xmath16 galaxies is considered to be applicable to bright [ cii ] emitters with @xmath102 @xcite and exceeds the sfr@xmath99 upper limit for the candidates in this survey .
the estimated sfr@xmath103}$ ] from the low - metal dwarf relation are 13@xmath104 and 20@xmath105 m@xmath106/yr for adf22-linea and lineb , respectively ( see also table 3 ) .
we estimate the [ cii ] luminosity function ( lf ) at @xmath107 from only one source , because one of the two [ cii ] candidates has a relatively low-@xmath63 and thus could possibly be a spurious source ( see 4.1 ) .
we show this result in figure 7 and compare it to [ cii ] lfs from previous studies .
the estimated [ cii ] lfs at @xmath108 @xcite are derived from the follow up observation of the _ iras _ sources @xcite or samples from the great observatories all - sky lirg survey @xcite .
we indicate the upper limit at @xmath109 derived by @xcite using alma cycle 0 archive data , and the lower limit at @xmath109 based on two serendipitous detection in aless survey @xcite . the estimation at @xmath17 is derived from follow up observation by @xcite .
we also indicate the estimation of over - dense region at @xmath91 by @xcite .
the constraint for [ cii ] lf at @xmath110 is provided by result of aspecs @xcite , which is based on an assumption that all [ cii ] candidates are real [ cii ] emitters . as discussed in 4.2 , we also derive the simple model of [ cii ] luminosity function at @xmath16 = 6 by using sfr-@xmath85}$ ] relation @xcite and sfr function at @xmath16 = 6 @xcite .
our [ cii ] lf model is close to our own observational result and the other studies , whereas the estimated lfs based on the empirical relations for high - redshift , and for all galaxies @xcite , do not match the observational result at @xmath111 .
we note that if completeness of the detection is lower than unity , the estimated [ cii ] lf represents the lower limit .
we calculate a conservative limit of [ cii ] sfrd from the mean of the sfr@xmath103}$ ] of the two sources divided by the survey volume ( figure 8) .
the derived [ cii ] sfrd is @xmath112 .
interestingly , this is close to the dust - uncorrected uv sfrd at @xmath35 .
the contribution of the only one [ cii ] emitter with faint uv and dust emission to the cosmic sfrd might already constitute a major contribution .
the result may imply the existence of the untraceable component of the sfrd by rest - uv . in order to confirm the truth of this
, the estimation of a faint end slope of the [ cii ] lf would be crucial . in figure 8
, we also derive upper limits of sfrds at @xmath113 from the non - detections of [ oi ] 145@xmath15 m , [ nii ] 122@xmath55 m [ oiii ] 88@xmath55 m lines in our search as discussed in 4.3 .
the sfrs are calculated from line luminosities by using observational relations estimated by @xcite .
this result demonstrates that line survey enables us to estimate sfrds at multiple redshifts at once .
there are a few possible mechanisms for the [ cii ] line emission to be particularly intense relative to fir and uv emission . for example
, it can be caused by high far - uv radiation from massive , young stars in the early universe ( e.g.,@xcite ) .
the environment of a low metallicity and a low dust - to - gas ratio can also cause enhancement of [ cii ] radiative cooling @xcite . in particular , the low dust - to - metal environment may not only enhance [ cii ] line emission but
also weaken dust continuum emission @xcite .
hot dust dominates the short - wavelength portion of the sed @xcite , making the dust continuum at long - wavelengths to be relatively suppressed .
the size distribution of dust grains also affect faint fir continuum @xcite .
altogether , observations in the submm - band can provide invaluable information on the physical properties of high - redshift galaxies .
future deep submm surveys will enable us to understand the formation of galaxies and to probe the early cosmic star - formation history .
we search millimeter line emitters by using 1.1 mm adf22 survey data taken in alma cycle2 . our newly constructed method for line search worked for detecting two co emitters at @xmath5 and @xmath114 and two [ cii ] emitter candidates at @xmath6 and @xmath7 with @xmath115 .
[ cii ] emitter candidates are faint in all counterparts .
the line species of the co emitters are identified by sed fitting or spectral follow up observation . for [ cii ] emitter candidates , the possibility of other line emissions are excluded by discussion about number counts , line ratio and ews . since one spurious source is possibly contaminated with the candidates , we assume at least one of the two candidates to be a real [ cii ] emitter .
we constrain @xmath91 [ cii ] lf for one source and found that the [ cii ] lfs at @xmath111 show good agreement with the predicted lf by using sfr-@xmath85}$ ] relation calibrated by local metal poor dwarfs .
we also found that estimated [ cii]-based sfr are consistent with upper limit of total sfr if we use the sfr-@xmath85}$ ] relation for local metal poor dwarfs .
we estimate a conservative limit of [ cii ] sfrd at @xmath35 for one source , which is close to the dust - uncorrected uv sfrd at @xmath35
. the results might be imply that mm / submm line survey is a powerful probe to estimate untraceable sfrd component from rest - uv observation at high - redshift .
the constrain for faint end slope of [ cii ] lf from further line survey and fir / uv follow - up observation will give us the truth of such implication and detailed picture of cosmic star - formation history .
.properties of datacube of four spectral windows . [ cols="^,^,^,^,^,^,^,^,^,^ " , ] note .
all unit of photometry flux is ab magnitude .
upper limits are given by 3@xmath63 .
the cfht megacam @xmath46 ( in archive , p.i .
cowie , see also matsuda et al .
subaru suprime - cam @xmath47 , @xmath48 , @xmath49 , @xmath50 , @xmath51 band @xcite , subaru moircs @xmath52 , @xmath53 , @xmath54s ( uchimoto et al .
2012 ) and the spitzer irac 3.6 , 4.5 , 5.8 , 8.0 @xmath15 m ( webb et al .
2009)-band photometry of the line emitters .
the psf difference in@xmath46 @xmath40 8.0 @xmath55 m are corrected following kubo et al .
the 24 @xmath55 m photometry flux is calculated over a 2@xmath1 diameter aperture .
this work was supported by the alma japan research grant of national astronomical observatory of japan ( naoj ) chile observatory , naoj - alma-0071 and naoj - alma-0160 .
nhh was supported by the grant of naoj visiting fellow program supported by the research coordination committee , national astronomical observatory of japan ( naoj ) and by funding from foundation for promotion of astronomy .
irs acknowledges support from stfc ( st / l00075x/1 ) , the erc advanced grant dustygal ( 321334 ) and a royal society wolfson research merit award .
we are grateful to the referee r. maiolino for his useful comments and suggestions .
the authors wish to thank a. sternberg , e. seaquist , l. yao , m. oguri , h. nagai , i. shimizu , k. mawatari , t. saito .
this paper makes use of the following alma data : ads / jao.alma#2013.1.00162.s .
data analysis ware carried out on common use data analysis computer system at the astronomy data centre , adc , of the naoj .
irac data was reduced and provided by j. huang .
alma is a partnership of eso ( representing its member states ) , nsf ( usa ) and nins ( japan ) , together with nrc ( canada ) and nsc and asiaa ( taiwan ) and kasi ( republic of korea ) , in cooperation with the republic of chile .
the joint alma observatory is operated by eso , aui / nrao and naoj . | we report the identification of four millimeter line emitting galaxies with the atacama large milli / submillimeter array ( alma ) in ssa22 field ( adf22 ) .
we analyze the alma 1.1 mm survey data , with an effective survey area of 5 arcmin@xmath0 , a frequency range of 253.1256.8 and 269.1272.8 ghz , angular resolution of [email protected] and rms noise of 0.8 mjy beam@xmath2 at 36 km s@xmath2 velocity resolution .
we detect four line emitter candidates with significance levels above @xmath3 .
we identify one of the four sources as a co(9 - 8 ) emitter at @xmath4 in a member of the proto - cluster known in this field .
another line emitter with an optical counterpart is likely a co(4 - 3 ) emitter at @xmath5 .
the other two sources without any millimeter continuum or optical / near - infrared counterpart are likely to be [ cii ] emitter candidates at @xmath6 and @xmath7 .
the equivalent widths of the [ cii ] candidates are consistent with those of confirmed high - redshift [ cii ] emitters and candidates , and are a factor of 10 times larger than that of the co(9 - 8 ) emitter detected in this search .
the [ cii ] luminosity of the candidates are @xmath8 . the star formation rates ( sfrs ) of these sources
are estimated to be @xmath9 if we adopt an empirical [ cii ] luminosity - sfr relation .
one of them has a relatively low - s / n ratio , but shows features characteristic of emission lines . assuming that at least one of the two candidates is a [ cii ] emitter
, we derive a lower limit of [ cii]-based star formation rate density ( sfrd ) at @xmath10 .
the resulting value of @xmath11 @xmath12 is consistent with the dust - uncorrected uv - based sfrd .
future millimeter / submillimeter surveys can be used to detect a number of high redshift line emitters , with which to study the star formation history in the early universe . |
the time evolution of the hot matter of quarks and gluons created in the relativistic heavy ion collider ( rhic ) experiments at brookhaven national laboratory is well described by the relativistic hydrodynamic equation with _ almost no _ dissipative effects @xcite .
this fact means that the quark - gluon plasma ( qgp ) slightly above the deconfinement phase transition temperature @xmath1 behaves as an almost ideal fluid , of which viscosity and heat conductivity , i.e. , transport coefficients , are negligiblely small .
this interesting non - equilibrium property of qgp as an almost ideal fluid is considered to be one of the properties of a strongly interacting system , and consistent with hadronic excitations surviving in qgp shown by several groups using the lattice qcd @xcite . in order to reach a full understanding of this non - equilibrium property of qgp
, we need to study the rhic experimental data using the relativistic hydrodynamic equation with _
finite _ dissipative effects . at present , a good candidate for the relativistic hydrodynamic equation with dissipative effects is considered to be the israel - steart equation @xcite instead of the standard equations by eckart @xcite and landau and lifshitz @xcite .
this is because the israel - stewart equation holds the causality in contrast with the eckart and landau - lifshitz equations . to assure the causality
, the israel - stewart equation contains higher derivative terms with relaxation times , i.e. , new parameters other than transport coefficients , which are dependent on materials we consider .
an important matter is to determine these relaxation times from microscopic dynamics governing the materials . in 1979 ,
israel and stewart @xcite calculated the relaxation times of a rarefied gas by using the relativistic boltzmann equation with the grad s fourteen moment approximation @xcite .
these quantities have been used by several groups @xcite to carry out ( 2 + 1)-dimensional hydrodynamic simulations based on the israel - stewart equation . in 2008 ,
natsuume and okamura @xcite derived the relaxation times of a strongly interacting supersymmetric gauge - theory plasma by utilizing the dispersion relations of the hydrodynamic modes obtained with use of the ads / cft duality .
however , a calculation of the relaxation times of qgp based on qcd is unfortunately not yet done .
the aim of this paper is ( 1 ) to construct a method which enables us to determine the relaxation times from the microscopic theory including qcd , and ( 2 ) to present the first lattice qcd calculation of the relaxation times of qgp on the basis of this method . to construct the method
, we utilize a _ relaxation function _ , which gives a linear but non - local ( non - markovian ) relation between an external field applied to the system and the current induced in the system @xcite .
the relaxation functions contain full information about non - equilibrium dissipative processes , e.g. , transport coefficients and relaxation times .
moreover a microscopic calculation of the relaxation functions is straightforward using the linear response theory . in this paper
, we first express the relaxation function in terms of a _ spectral function _ for a conserved current , which can be calculated from the microscopic dynamics by a quantum - field theoretic method .
this expression tells us that the relaxation function is determined by a whole behaviour of the spectral function in contrast to the transport coefficient defined as the low - frequency and long - wavelength limit of the spectral function .
we show that , only as long as the calculated relaxation function shows an exponentially damping behaviour , the damping time constant can be identified as the relaxation time in the israel - stewart equation .
taking a baryon - number dissipative current of a plasma consisting of charm quarks and gluons ( referred to as the charm quark - gluon plasma later on ) as a simple example , we calculate the relaxation function from the spectral function obtained using the lattice qcd together with the maximum entropy method ( mem ) @xcite .
the obtained relaxation function shows a strongly - oscillation damping behaviour instead of an exponentially damping one , due to the charmed vector hadron @xmath0 surviving above @xmath1 .
this disagreement suggests that the applicability of the israel - stewart equation to the baryon - number dissipative current of the charm quark - gluon plasma is doubtful .
we show an idea for the improvement of the israel - stewart equation by deriving the hydrodynamic equation consistent with the strongly - oscillation damping relaxation function .
the conclusion of this paper is that the israel - stewart equation is not unique as the causal and dissipative hydrodynamic equation , and the form of the causal and dissipative hydrodynamic equation applicable to qgp is determined by the shape of the spectral function reflecting the properties of hadronic excitations in qgp .
this indicates an important link between the non - equilibrium properties of qgp and the hadron spectroscopy at finite temperature .
this paper is organized as follows . in section [ sec:002 ] ,
we construct a method which enables a microscopic calculation of the relaxation times in the israel - stewart equation . for this purpose ,
the relaxation functions are utilized . in section [ sec:003 ] ,
we present the first calculation of the relaxation function for the baryon - number dissipative current in qgp , with use of the monte carlo simulation based on the lattice qcd .
we discuss whether the exponentially damping relaxation function of the israel - stewart equation can be justified from a microscopic point of view .
section [ sec:004 ] is devoted to the summary and the concluding remarks . in appendix
[ sec:005 ] , taking a shear viscous dissipative process as a typical example , we show the detailed derivation of the microscopic representation of the relaxation function .
in this section , we give a method which enables us to determine the relaxation time from the microscopic theory including qcd .
our discussion is based on the relaxation function established in the phenomenological relaxation theory @xcite .
the microscopic representation of the relaxation function is given in the linear response theory @xcite .
to explain the physical meaning of the relaxation functions , let us start from the israel - stewart equation for a simple system , where the conserved quantities are the energy and the momentum , and the dissipative effect comes solely from the shear viscosity .
we denote the energy - momentum tensor by @xmath2 where @xmath3 and @xmath4 are the internal energy and the pressure at the temperature @xmath5 , respectively , @xmath6 the flow velocity , @xmath7 the shear viscous stress and @xmath8 the projection operator with @xmath9 . here
@xmath6 is normalized as @xmath10 , and @xmath7 is satisfied with @xmath11 and @xmath12 , which are reproduced by @xmath13 with the symmetric traceless projection operator @xmath14 and the shear viscous stress before the projection @xmath15 .
it is noted that the israel - stewart equation is the system of the nine differential equations with respect to the nine independent variables , @xmath5 , @xmath6 and @xmath7 .
the equations consist of the continuity equation , @xmath16 and the relaxation equation @xcite , @xmath17 where @xmath18 and @xmath19 denote the shear viscosity and the relaxation time of the shear viscous dissipative process , respectively , and @xmath20 with the proper time @xmath21 .
the solution of eq.([eq:001 ] ) for the shear viscous stress reads @xmath22 \ , 2 \ , \partial_\rho u_\sigma(\tau^\prime ) .
\end{aligned}\ ] ] equation ( [ eq:002 ] ) tells us that @xmath23 at the proper time @xmath21 depends on the history of the external field @xmath24 from past to present , and the weighting function is given by @xmath25 .
notice that eq.([eq:002 ] ) is equivalent to that derived in the stewart s non - local thermodynamics @xcite , and has been recently rediscovered by koide _
the meaning of eq.([eq:002 ] ) becomes clearer from the viewpoint of the phenomenological relaxation theory @xcite , where the linear but non - markovian constituent equation , @xmath26 is well established and @xmath27 is identically the relaxation function .
therefore , we find that the israel - stewart causal hydrodynamics is equivalent to the special case of the phenomenological relaxation theory with @xmath28 i.e. , the exponentially damping relaxation function .
we first notice that the linear response theory @xcite enables us to calculate the relaxation function directly from the microscopic dynamics .
we now consider the relaxation function of a generic dissipative process , @xmath29 where @xmath30 denotes the external field , @xmath31 the current induced by @xmath30 , and @xmath32 the relaxation function .
the linear response theory allows that the microscopic derivation of the relaxation function is straightforward .
in fact , as shown in appendix [ sec:005 ] , we can derive the following relation , @xmath33 \
, \rangle , \end{aligned}\ ] ] where @xmath34 is a heisenberg operator and gives the microscopic representation for the induced macroscopic current @xmath31 , and the bracket indicates the thermal average .
notice that the integrand of the right hand side of eq.([eq:003 ] ) is the retarded green function , which has the following spectral representation , @xmath35 \
, \rangle \nonumber\\ { } = \int\!\!\frac{\mathrm{d}^4k}{(2\pi)^4 } \ , \mathrm{e}^{-ik^0 t+i \mbox{\boldmath$\scriptstyle k$}\cdot \mbox{\boldmath$\scriptstyle x$ } } \ , \int\!\!\mathrm{d}\omega \ , \frac{a(\omega,\,\vec{k})}{\omega - k^0 - i \ , \varepsilon } , \end{aligned}\ ] ] where @xmath36 is the spectral function and @xmath37 the infinitesimal positive constant .
equations ( [ eq:003 ] ) and ( [ eq:004 ] ) lead us to the relation between @xmath32 and @xmath38 given by @xmath39 this formula assures that the relaxation function can be obtained as the fourier transformation of @xmath40 , i.e. , the real part of the complex admittance @xcite , which is defined by @xmath41 here we notice the two important properties concerning the complex admittance @xmath42 : ( i ) the low - frequency and long - wavelength limit of the complex admittance , @xmath43 , gives the transport coefficients .
in fact , with use of this representation , many groups @xcite have calculated the transport coefficients of qgp employing the lattice qcd .
( ii ) using @xmath42 , we can write down the generic relaxation equation equivalent to the non - markovian constituent equation ( [ non - markov ] ) , @xmath44 by combining eqs.([eq : admittance ] ) and ( [ eq:2 - 004 ] ) , we can construct the relaxation equation from a given spectral function .
the above observation tells us the following three points : ( a ) the relaxation function in qgp can be calculated from the spectral function for the microscopic current of qcd .
( b ) the israel - stewart equation can be valid for the description of qgp only as long as the relaxation function calculated in the microscopic way is an exponentially damping function , and the relaxation time of qgp can be obtained as its damping time constant .
( c ) if the relaxation function calculated from qcd is different from that derived from the israel - stewart equation , we should carry out the causal hydrodynamic calculation by incorporating the relaxation equation ( [ eq:2 - 004 ] ) or the non - markovian constituent equation ( [ non - markov ] ) into the continuity equation .
it is interesting to investigate which spectral functions reproduce the exponentially damping relaxation function in the israel - stewart equation . taking the shear viscous dissipative process at the rest frame as a typical example
, we find it to be a lorentzian function , @xmath45 in fact , it can be checked that eqs.([eq:005 ] ) and ( [ eq:2 - 005 ] ) give the exponentially damping relaxation function , @xmath46 and the correspondent complex admittance reads @xmath47 it is noteworthy that eq.([eq:2 - 004 ] ) with this complex admittance reproduces the israel - stewart - type relaxation equation , @xmath48 here we replace @xmath49 with @xmath50 in order to obtain the equation at an arbitrary frame from that at the rest frame .
therefore it is found that the ansatz of the lorentzian - type spectral function shown in eq.([eq:2 - 005 ] ) is an essential point of the israel - stewart equation .
notice that there is no reason to believe that the spectral function for the current of qcd is approximated well by the lorentzian function in the low - frequency and long - wavelength region .
in this section , we demonstrate the first lattice qcd calculation of the relaxation function . from the viewpoint of availability of lattice data ,
we choose the relaxation function for the baryon - number dissipative process .
the continuity equation reads @xmath51 and the non - markovian constituent equation @xmath52 where @xmath53 and @xmath54 denote the baryon - number density and the baryon - number chemical potential , respectively .
this relaxation function can be calculated from the spectral function of the spatial components of the baryon - number current , @xmath55 , where @xmath56 denotes the @xmath57-flavour @xmath58-colour quark field , @xmath59 the gamma matrix , @xmath60 , @xmath61 the number of the flavour and @xmath62 the number of the colour , respectively . as a simple example , we consider the single - flavour deconfined system of which constituents are gluons and charm quarks ( @xmath63 ) , i.e. , the charm quark - gluon plasma . furthermore we treat a spatially integrated relaxation function , @xmath64 where @xmath65 is a zero - momentum projected spectral function .
this relation indicates that the calculation of @xmath66 requires an explicit form of @xmath67 .
therefore we need to reconstruct @xmath67 from the temporal correlation function , @xmath68 obtained using monte carlo simulation based on the lattice qcd . to get the full information about @xmath67
, we utilize mem @xcite . here
we shall give a brief account of mem .
the temporal correlation function @xmath69 and the associated @xmath67 are related by the laplace transformation , @xmath70 where @xmath71 / \sinh(\omega/2t)$ ] is a kernel function .
since the number of data points in the temporal direction is finite , the direct inverse laplace transformation from @xmath69 to @xmath67 is an ill - posed problem .
mem is a method to overcome this difficulty on the basis of the bayes theorem in the theory of statistical inference . using mem
, we can infer @xmath67 as the most probable one which achieves a balance between reproducing the lattice data @xmath69 and being kept not so far from the default model @xmath72 , defined by a plausible form of @xmath67 .
mem has proved to be a powerful tool to investigate @xmath67 of mesons and baryons at both of @xmath73 @xcite and @xmath74 @xcite .
recently , the author @xcite has improved mem to avoid a numerical instability present in the study of the shear viscosity of the gluon plasma .
an essential point of the improved mem analysis is that the laplace transformation combining @xmath69 and @xmath67 is converted as follows , @xmath75 \ ,
\bigg [ \frac{\pi}{\omega } \ , a(\omega ) \bigg ] .
\end{aligned}\ ] ] regarding @xmath76 as a new kernel function , we can infer the real part of the complex admittance @xmath77 directly from @xmath69 .
further , in the study of the scalar tetraquark @xcite , the extension of this method , such as @xmath78 \ , [ a(\omega)/\lambda(\omega)]$ ] with @xmath79 , has also been used to reduce the uncertainty of the inferred @xmath67 .
aarts _ et al .
_ @xcite independently have prepared this method , and have applied this method to calculate the electrical conductivity of qgp . in fig.[fig:000 ]
, we show the schematic overview which summarizes the way to obtain the relaxation function and the relaxation time of qgp on the basis of the lattice qcd and the improved mem analysis .
then we explain the default model used in the present analysis .
the default model @xmath72 is usually given as the form predicted by the perturbative qcd calculation valid in the high - energy region , @xmath80 .
due to the tree - level result , @xmath81 with @xmath82 , the default model reads @xmath83 such a monotonically increasing default model seems suitable to avoid the artificial peaks due to the choice of the default model , and using @xmath84 we can identify any peak structure of the inferred @xmath67 as that derived from the lattice data not the default model . in this paper , we are interested in the applicability of the israel - stewart equation to qgp , i.e. , the validity of the ansatz that the spectral function for the conserved currents is approximated by a lorentzian function shown in eq.([eq:2 - 005 ] ) . to test this ansatz
, we carry out the mem analysis based on not only the monotonically increasing default model @xmath84 but also the following lorentzian - type one especially in the low - energy region , @xmath85 where @xmath86 and @xmath87 denote the center value and width of the border between the strongly interacting low - energy region and the asymptotically free high - energy one , respectively , and @xmath88 and @xmath89 denote the transport coefficient and relaxation time of the baryon - number dissipative process , respectively .
if @xmath90 inferred by the mem analysis based on @xmath91 is quite different from the lorentzian function in the low - frequency region , the applicability of the israel - stewart equation to qgp is doubtful .
the system considered here is the charm quark - gluon plasma at @xmath92 .
we use the quenched approximation in calculating @xmath69 . a detailed set up of the lattice qcd numerical simulation
is identical to that in ref.@xcite .
we use the naive plaquette gauge action and the standard wilson quark action with the gauge coupling constant @xmath93 , the bare anisotropy @xmath94 and the fermion anisotropy @xmath95 with the spatial ( temporal ) hopping parameter @xmath96 ( @xmath97 ) .
these parameters reproduce the anisotropic lattice @xmath98 with the temporal ( spatial ) lattice spacing @xmath99 ( @xmath100 ) in the quenched - level simulation @xcite .
the adopted lattice size is @xmath101 , where the spatial size @xmath102 and the temperature @xmath103 .
this spatial size seems small . by varying the spatial boundary condition , however , we have checked that @xmath104 is large enough to simulate the charm quark - gluon plasma in lattice @xcite .
the number of the gauge configurations is 100 .
the improved mem analysis is done using the lattice data @xmath69 ( @xmath105 ) with @xmath106 and @xmath107 and the default models @xmath84 in eq.([eq:3 - 006 ] ) and @xmath108 in eq.([eq:007 ] ) . in fig.[fig:001 ] , we show the inferred @xmath90 s , which have been checked to be stable against the change of @xmath109 as @xmath110 , @xmath111 and @xmath112 . inferred using the improved mem analysis based on @xmath113 in eq.([eq:3 - 006 ] ) .
( b ) the result of the improved mem analysis based on @xmath91 in eq.([eq:007 ] ) with @xmath114 , @xmath115 , @xmath116 and @xmath117 .
the dash line denotes the default model @xmath118 and the solid line the inferred @xmath119 .
the horizontal position and length of the bars indicate the region over which @xmath120 is averaged , while the vertical height of the bars denotes the standard deviation in the averaged value of @xmath120 in the interval . ]
inferred using the improved mem analysis based on @xmath113 in eq.([eq:3 - 006 ] ) .
( b ) the result of the improved mem analysis based on @xmath91 in eq.([eq:007 ] ) with @xmath114 , @xmath115 , @xmath116 and @xmath117 .
the dash line denotes the default model @xmath118 and the solid line the inferred @xmath119 .
the horizontal position and length of the bars indicate the region over which @xmath120 is averaged , while the vertical height of the bars denotes the standard deviation in the averaged value of @xmath120 in the interval . ]
figure [ fig:001]-(a ) indicates that @xmath90 has no strength in the low - frequency region @xmath121 and the peak structure with large width around @xmath122 .
notice that due to @xmath84 having no strength in @xmath123 and taking a zero value at @xmath124 , we can not get any information about the applicability of the israel - stewart equation to the baryon - number dissipative process of the charm quark - gluon plasma and the smallness of the transport coefficient of the baryon - number dissipative process @xmath88 from this result . on the other hand , the peak structure with large width around @xmath122 can be considered to be a signal of the charmed vector hadron @xmath0 surviving above @xmath1 .
this result is in good agreement with those of the several groups @xcite .
figure [ fig:001]-(b ) shows that @xmath90 is quite different from @xmath108 , the lorentzian function in the low - frequency region .
this result suggests that the applicability of the israel - stewart equation to the baryon - number dissipative process of the charm quark - gluon plasma is questionable .
we have checked the stability of the inferred @xmath90 against the change of @xmath86 , @xmath125 , @xmath88 and @xmath89 .
we calculate the relaxation function expected to be quite different from the exponentially damping function . for this purpose ,
we carry out the fourier transformation of the inferred @xmath90 in fig.[fig:001]-(a ) .
since @xmath90 exhibits a linear divergence of @xmath126 for @xmath127 , the numerical fourier transformation is unstable . to avoid this numerical instability ,
we convert eq.([eq:3 - 003 ] ) into @xmath128 \bigg\ } , \end{aligned}\ ] ] where we have used @xmath129 and the following relation , @xmath130 in eq.([eq:008 ] ) , @xmath131 denotes the cut - off frequency defined as @xmath132 for @xmath133 . here
we use @xmath134 . in fig.[fig:002 ] , we show the calculated @xmath66 together with the error bars . with @xmath103 .
the meaning of the horizontal position and length of the bars is explained in the caption of fig.[fig:001 ] .
@xmath135 exhibits a strongly - oscillation damping behaviour within the error bars , instead of the exponentially damping behaviour derived from the israel - stewart equation . ]
the singularity of @xmath136 is inevitable because it is originated from the divergence of @xmath90 at @xmath137 due to the composite operator @xmath138 being ill - defined at the same coordinate .
therefore , we concentrate on @xmath66 for @xmath139 .
as expected , @xmath66 is not a exponential damping function but a strongly - oscillation damping function .
this oscillation damping behaviour of @xmath66 is attributed to the peak structure with the finite width in @xmath90 correspondent to the surviving @xmath0 at @xmath103 .
the relaxation time @xmath140 of the oscillation damping function is approximated by the inverse of the width of the low - lying peak structure .
accordingly , we can estimate @xmath141 from fig.[fig:001]-(a ) , which seems to be consistent with @xmath66 in fig .
[ fig:002 ] .
the disappearance of @xmath0 at the higher temperature would drive the drastic change of @xmath66 and @xmath140 .
finally we discuss the form of a relaxation equation corresponding to the spectral function with the peak structures in the finite - frequency region .
here we consider the spectral function for the shear viscous dissipative process at the rest frame as a typical example , @xmath142 where @xmath143 denotes the peak position .
notice that eq.([eq:3 - 008 ] ) agrees with eq.([eq:2 - 005 ] ) by the setting of @xmath144 .
using eqs.([eq : admittance ] ) and ( [ eq:3 - 008 ] ) , we have the correspondent complex admittance , @xmath145 by combining eqs.([eq:2 - 004 ] ) and ( [ eq:3 - 009 ] ) , we find the form of the relaxation equation corresponding to the spectral function with the two peak structures at @xmath146 to be @xmath147 \ , j(t,\,\vec{x})\nonumber\\ = \eta \ , ( \omega^2_\pi \ , \tau^2_\pi + 1 ) \ , ( 1 + \tau_\pi \ , \partial_t ) \ , x(t,\,\vec{x } ) .
\end{aligned}\ ] ] it is found that the temporal derivative @xmath49 appears also in the right hand side of eq.([eq:3 - 010 ] ) , in contrast with the israel - stewart - type relaxation equation ( [ eq:2 - 008 ] ) .
we identify eq.([eq:3 - 010 ] ) as a more probable form of the relaxation equation applicable to the strongly interacting system including qgp slightly above @xmath1 , of which spectral function has peak structures in the finite - frequency region .
as shown in subsection [ sec:003b ] , the baryon - number dissipative process of the charm quark - gluon plasma can not be described by the israel - steart equation .
we should carry out the causal hydrodynamic simulations by incorporating the relaxation equation based on eq.([eq:3 - 010 ] ) not eq.([eq:2 - 008 ] ) into the continuity equation ( [ eq:3 - 001 ] ) .
first , in this paper , we have pointed out that the israel - stewart causal hydrodynamic equation is acceptable only as long as the relaxation function is a exponentially damping function , and the relaxation time can be obtained as its damping time constant . using the linear response theory , we have derived the formula where the relaxation function is represented by the spectral function for the conserved current , which can be calculated from the microscopic dynamics .
furthermore , we have discussed the systematic way to construct a causal hydrodynamic equation corresponding to the calculated relaxation function .
then , taking the baryon - number dissipative process of the charm quark - gluon plasma for a simple example , we have calculated the relaxation function from the spectral function obtained using the quenched lattice qcd together with the maximum entropy method .
the obtained relaxation function shows the strongly - oscillation damping behaviour within the error bars , not the exponential damping .
the origin of the strongly - oscillation damping behaviour is the charmed vector hadron @xmath0 surviving at @xmath103 .
this fact suggests that the applicability of the israel - stewart equation to the baryon - number dissipative process of the charm quark - gluon plasma is questionable .
we have shown an idea for the improvement of the israel - stewart equation by deriving the relaxation equation consistent with the strongly - oscillation damping relaxation function .
we conclude that the israel - stewart equation is not unique as the causal and dissipative hydrodynamic equation . by calculating the relaxation function from the spectral function with use of the formula in eq.([eq:005 ] ) , we can investigate the applicability of the israel - stewart equation .
it is noteworthy that the form of the causal hydrodynamic equation with dissipative effects applicable to qgp is determined by the shape of the spectral function reflecting the hadronic excitations in qgp .
this indicates an important link between the non - equilibrium properties of qgp and the hadron spectroscopy at the finite temperature . in the next paper @xcite
, we will investigate the properties of the relaxation equation corresponding to the spectral function with the peak structures in the finite - frequency region .
it seems an interesting subject to derive the relativistic hydrodynamic equation by combining the relaxation equation shown in eq.([eq:3 - 010 ] ) and the continuity equation based on the energy - flow equation by landau @xcite or the particle - flow equation by tsumura _
@xcite , which have a stable equilibrium solution . to discuss the more realistic system than the charm quark - gluon plasma
, we must investigate the contribution of up , down and strange quarks to the spectral function for the baryon - number conserved current .
the light vector hadrons composed of these quarks slightly above @xmath1 seem to play a more essential role for the slow and long - wavelength fluid behaviour of qgp than @xmath0 . the answer to this conjecture
will be reported elsewhere .
in addition , the calculation of the relaxation function and/or the spectral function for the shear and bulk viscous dissipative process of qgp is an interesting future study using the lattice qcd with the maximum entropy method or another non - perturbative method , such as the qcd sum rule or the ads / cft duality .
the author is grateful to prof .
t. kunihiro , prof .
h. suganuma and dr .
t. kojo for their useful comments and discussions .
the lattice qcd calculations have been performed on nec - sx5 at osaka university .
in this section , we show the derivation of eq.([eq:003 ] ) based on the linear response theory @xcite together with the non - equilibrium statistical operator method proposed by zubarev @xcite . we consider a statistical operator @xmath148 , where @xmath149 denotes a state of the system .
the time evolution of @xmath150 is governed by the liouville equation , @xmath151 = 0 , \end{aligned}\ ] ] or the neumann equation , i.e. , the equation converted from eq.([eq : app01 ] ) , @xmath152 = 0 , \end{aligned}\ ] ] both of which are equivalent to the schrdinger equation , @xmath153 with @xmath154 being the hamiltonian .
it is noted that eqs.([eq : app01 ] ) and ( [ eq : app02 ] ) hold the time - reversal symmetry .
therefore , the entropy of the system defined by @xmath155 is normally conserved . to obtain the non - equilibrium statistical operator @xmath156 , which exhibits the increase of the entropy , the breaking of the time - reversal symmetry is needed . in ref.@xcite , zubarev has proposed to construct @xmath156 as the solution of the neumann equation ( [ eq : app02 ] ) under the time - reversal asymmetric boundary condition @xmath157 , where @xmath158 denotes the quasi - equilibrium statistical operator defined by @xmath159 , \end{aligned}\ ] ] with @xmath160 . here , @xmath161 denotes a energy - momentum tensor operator in the heisenberg picture , which agrees with that of the schrodinder picture at @xmath162 .
@xmath163 denotes a lagrange multiplier characterizing the system .
the physical meaning of @xmath163 becomes clear from the setting of @xmath164 , which reproduces the equilibrium statistical operator @xmath165 with @xmath166 .
it is noted that imposing the time - reversal asymmetric boundary condition @xmath157 to eq.([eq : app02 ] ) is equivalent to adding a time - reversal asymmetric source term to the right hand side of eq.([eq : app02 ] ) , @xmath167 = - \epsilon \
, \ln \big [ \hat{\rho}(t ) \big/ \hat{\rho}_\mathrm{qeq}(t|0 ) \big ] , \end{aligned}\ ] ] where @xmath168 denotes an infinitesimal positive constant and the limit of @xmath169 follows the thermodynamical limit .
@xmath156 can be obtained as @xmath170 , where @xmath171 denotes the solution of eq.([eq : app04 ] ) and @xmath172 is suppressed later on .
the solution of eq.([eq : app04 ] ) reads @xmath173 a partial integration of the right hand side of eq.([eq : app05 ] ) leads us to @xmath174 where @xmath175 here we have used @xmath176 .
notice that @xmath177 contains @xmath178 , i.e. , the gradient of the lagrange multiplier characterizing the system .
we shall consider the system not so far from the equilibrium state , where @xmath178 is small .
it is important that under this consideration we can regard @xmath177 as a perturbation term to @xmath179 and utilize the linear response theory , where @xmath178 is identified as an external field .
we expand @xmath156 up to the first order of @xmath177 as follows , @xmath180 where @xmath181 and @xmath182 with an arbitrary operator @xmath183 .
notice that @xmath184 .
using eq.([eq : app08 ] ) , we calculate the energy - momentum tensor realized in the non - equilibrium state , @xmath185 we decompose @xmath186 into the unperturbative term and the perturbative one as @xmath187 , where @xmath188 here @xmath189 represents the perfect - fluid part of the energy - momentum tensor , while @xmath190 the dissipative part .
@xmath191 in eq.([eq : app11 ] ) is a relaxation function defined by @xmath192 .
\end{aligned}\ ] ] equation ( [ eq : app12 ] ) is one of the main results in this section . as the thermal average used in eq.([eq : app12 ] ) is defined with use of @xmath184 not @xmath193 , eq.([eq : app12 ] ) seems inconvenient to calculate @xmath191 employing a technique established in the finite - temperature field theory . to derive @xmath191 written by @xmath193
, we shall make an approximation for eq.([eq : app11 ] ) .
we consider a situation where a fluctuation of @xmath163 is small .
notice that this approximation differs from that used in the linear response theory , and is implemented in a systematic way using the expansion , @xmath194 where @xmath37 is an infinitesimal constant different from @xmath168 .
we evaluate a leading order of @xmath195 with respect to @xmath37 . due to
@xmath196 being the first order of @xmath37 , the zeroth - order evaluation of @xmath191 is sufficient .
first , using @xmath197 , we have @xmath198 where the energy - momentum vector operator @xmath199 . with use of the unitary transformation for the temporal and spatial translation by @xmath200 , @xmath201 and the cluster property of the correlation function ,
@xmath202 we can convert the right hand side of eq.([eq : app12 ] ) into @xmath203 by replacing @xmath204 , integrating with respect to @xmath205 and using the translational invariance , we find eq.([eq : app17 ] ) to be @xmath206 \ , \rangle}_a .
\end{aligned}\ ] ] using eq.([eq : app18 ] ) , we can rewrite eq.([eq : app11 ] ) as a lorentz - covariant form , @xmath207 \ , \rangle}_a .
\end{aligned}\ ] ] then , using @xmath208 ( @xmath10 ) , we derive @xmath209 written by @xmath193 . for this purpose ,
we utilize a decomposition of @xmath210 based on the irreducible tensors @xmath6 , @xmath211 and @xmath212 , namely , @xmath213 where @xmath214 , @xmath215 , @xmath216 and @xmath217
. here we focus the shear viscous stress @xmath218 . using curie s principle @xcite and the spatial isotropy
, we have @xmath219 \ , \rangle}_a\nonumber\\ = \delta^{\mu\nu\rho\sigma } \ , \frac{1}{10 } \ , { \langle \ , [ \hat{\pi}^{ab}(x ) \,,\ , \hat{\pi}_{ab}(s \ , u / t ) ] \ , \rangle}_a , \end{aligned}\ ] ] where @xmath220 .
this relation leads us to @xmath221 \ , \rangle}_a .
\end{aligned}\ ] ] it is noted that , due to the integrand of @xmath222 in eq.([eq : app22 ] ) being a lorentz - scalar function , the integrand takes the same value for an arbitrary @xmath6 .
therefore , we can set @xmath223 in eq.([eq : app22 ] ) without loss of generality , and using the spatial isotropy we arrive at @xmath224 \ , \rangle , \end{aligned}\ ] ] where @xmath225 .
notice that the integrand of eq.([eq : app23 ] ) is identical to the retarded green function which has the spectral representation as shown in eq.([eq:004 ] ) .
equation ( [ eq : app23 ] ) is the other of the main results in this section .
a. k. chaudhuri and u. w. heinz , j. phys .
* 50 * ( 2006 ) 251 ; u. w. heinz , h. song and a. k. chaudhuri , phys .
c * 73 * ( 2006 ) 034904 ; a. k. chaudhuri , arxiv : nucl - th/0703027 ; arxiv : nucl - th/0703029 ; arxiv:0704.0134 [ nucl - th ] .
r. baier , p. romatschke and u. a. wiedemann , phys .
c * 73 * ( 2006 ) 064903 ; nucl .
phys . a * 782 * ( 2007 ) 313
; r. baier and p.romatschke , eur . phys .
j. c * 51 * ( 2007 ) 677 ; p. romatschke , eur .
j. c * 52 * ( 2007 ) 203 ; p. romatschke and u. romatschke , arxiv:0706.1522 [ nucl - th ] . | we propose a method which enables a qcd - based calculation of a relaxation time for a dissipative current in the causal and dissipative hydrodynamic equation derived by israel and stewart .
we point out that the israel - stewart equation is not unique as a causal and dissipative hydrodynamic equation , and the form of the causal and dissipative hydrodynamic equation is determined by the shape of a spectral function reflecting the properties of elementary excitations in the system we consider .
our method utilizes a relaxation function , which can be calculated from qcd using the linear response theory .
we show that the relaxation function can be derived from a spectral function for a microscopic representation of the dissipative current .
we also show that the israel - stewart equation is acceptable only as long as the calculated relaxation function is approximated well by a exponentially damping function , and the relaxation time can be obtained as its damping time constant . taking a baryon - number dissipative current of a plasma consisting of charm quarks and gluons as a simple example
, we present the first calculation of the relaxation function with use of the spectral function derived employing the quenched lattice qcd together with the maximum entropy method .
the calculated relaxation function shows a strongly - oscillation damping behaviour due to the charmed vector hadron @xmath0 surviving above the deconfinement phase transition temperature in qcd .
this result suggests that the applicability of the israel - stewart equation to the baryon - number dissipative current of the charm quark - gluon plasma is quite doubtful .
we present an idea for the improvement of the israel - stewart equation by deriving the hydrodynamic equation consistent with the strongly - oscillation damping relaxation function . |
active galaxies and quasars are thought to be powered by the presence of massive black holes ( bhs ) in their nuclei ( e.g. , rees 1997 ) .
such bhs are believed to exist in a large fraction of quiescent galaxies as well ( e.g. , tremaine 1996 ) .
the best dynamical bh detections in active galaxies come from studies of gas kinematics ; in quiescent galaxies only stars are generally available as a tracer of the gravitational potential .
ground - based stellar kinematical studies have yielded tentative detections for bhs in a handful of nearby quiescent galaxies ( kormendy & richstone 1995 ) , and this evidence is now being improved with the superior spatial resolution of the hubble space telescope ( hst ; kormendy 1996a , b ; van der marel 1997a ) .
however , the interpretation of the observed stellar motions is never straightforward ; only projected quantities are observed and the three - dimensional orbital structure is unknown .
construction of detailed dynamical models is therefore required .
axisymmetric models currently provide the state - of - the - art for this application . in such models ,
the phase - space distribution function ( df ) generally depends on three integrals of motion ( e.g. , binney & tremaine 1987 ) , the two classical integrals , @xmath10 and @xmath11 ( the binding energy and the angular momentum component along the symmetry axis , both per unit mass ) , and the third integral , @xmath12 .
the latter can not generally be expressed analytically in terms of the phase - space coordinates , which makes the construction of models with ` three - integral ' dfs , @xmath13 , complicated and time - consuming , although not impossible ( dehnen & gerhard 1993 ; cretton 1997 ; gebhardt 1997 ) . as a result
, significant attention has been given in recent years to a special class of axisymmetric models , those in which the df depends only on the two classical integrals of motion , @xmath1 .
the attraction of these ` two - integral ' models lies in the fact that their properties are tractable semi - analytically ( e.g. , hunter & qian 1993 ; dehnen & gerhard 1994 ) .
these models can be viewed as the axisymmetric generalization of spherical isotropic models : the velocity dispersions in the meridional plane are isotropic , i.e. , @xmath14 , and the flattening of the models ( when oblate ) is due to an excess of azimuthal motion . the present paper deals with the nearby e3 galaxy m32 , which has been a strong candidate for containing a bh for many years .
the highest spatial resolution ground - based data are well fit by an @xmath7 model with a 2@xmath15 bh ( van der marel 1994 ; qian 1995 ; dehnen 1995 ; bender , kormendy & dehnen 1996 ) .
such a model fits the observed line - of - sight velocity profile shapes without any adjustable parameters , suggesting that the df of m32 is indeed close to the @xmath7 form .
we recently completed the most detailed study to date of m32 ( van der marel 1997a ) , using new hst data ( van der marel , de zeeuw & rix 1997b ) and fully general three - integral modeling ( van der marel 1997c ; hereafter vdm97c ) .
the main results of these studies are that : ( i ) m32 _ must _ have a central dark mass in the range @xmath16 ( at the formal @xmath17% confidence level ) , enclosed within @xmath18 , most likely a bh ; and ( ii ) the best - fitting models have a dynamical structure that is indeed similar ( although not identical ) to that of an @xmath7 model .
although dynamical models with @xmath7 dfs can reproduce most features of the m32 data , they must also be stable if they are to be physically meaningful .
only limited research has been done on the stability of @xmath7 models .
bending modes and lopsided modes in extremely flattened systems ( axial ratio @xmath19 ) were discussed by merritt & fridman ( 1995 ) .
kuijken & dubinski ( 1994 ) performed n - body simulations of models with a king - type mass density profile and an axial ratio varying between @xmath20 in the center and @xmath21 in the outer parts .
models with little or no mean streaming were stable , but rapidly rotating models were found to be unstable to the formation of a bar .
preliminary results on the bar - stability in @xmath7 models with central density cusps , but without central dark objects , were presented by dehnen ( 1996 ) .
his n - body simulations indicate that : ( i ) non - rotating models are stable , even if they are as flat as @xmath8 ; and ( ii ) maximally rotating models tend to be bar - unstable if they are flatter than @xmath22 . the most detailed study to date of the stability of @xmath7 models has been that for a family of kuzmin - kutuzov models by sellwood & valluri ( 1997 ) .
they identify all the most important unstable modes as a function of shape and rotational support .
their results confirm the main conclusions on bar modes found by earlier authors , and they also present detailed studies of lopsided , bending and other unstable modes . the introduction of their paper provides a useful reference to other recent work on the stability of hot stellar systems ( see also the introductory chapter of robijn 1995 ) .
none of the previous studies included both a central density cusp and a central black hole , as is appropriate for m32 .
if m32 is being viewed edge - on , its intrinsic axial ratio is @xmath23 .
the galaxy must be intrinsically flatter if it is inclined with respect to the line of sight .
the observed rotation velocities imply that , over the radial range @xmath24 for which kinematical data is available , 8590% of the stars ( more - or - less independent of the assumed inclination ) must have the same sense of rotation around the symmetry axis ( van der marel 1994 ) .
the previous work on the stability of @xmath7 models suggests that the m32 models should certainly be stable to , e.g. , lopsided and bending modes ( unless for extreme inclinations ) , but that they might be unstable to bar modes .
we therefore tested the stability of the @xmath7 models for m32 through n - body simulations on a parallel supercomputer .
our results are described in what follows .
section 2 summarizes our choice of models , sections 3 describes the n - body technique , and section 4 discusses the results of the simulations .
conclusions are presented in section 5 .
an appendix outlines the construction of the initial conditions .
the dynamical models that we study in this paper are based on those presented in vdm97c .
the _ luminous _ mass density is assumed to be axisymmetric and of the form @xmath25^{\beta } \ > [ 1+(m / c)^2]^{\gamma } , \qquad m^2 \equiv r^2 + ( z / q)^2 . \label{massdens}\ ] ] the intrinsic axial ratio @xmath26 is related to the projected axial ratio @xmath27 and the inclination @xmath28 according to @xmath29 .
the parameters are set to @xmath30 , @xmath31 , @xmath32 , @xmath33 , @xmath34 , @xmath35 , @xmath36 , @xmath37 , and the distance to m32 is assumed to be @xmath38 .
the quantity @xmath39 is the luminosity density , and @xmath40 is the average v - band mass - to - light ratio of the stellar population . both @xmath40 and @xmath27 are assumed to be constant with radius .
this model provides an accurate fit to the observed surface brightness profile of m32 , including the available hst photometry in the central few arcsec ( lauer 1992 ) . the adopted constant
@xmath41 provides a good approximation in the central @xmath42 , although at larger radii the observed ellipticity increases slowly to @xmath43 at @xmath44 .
the model has finite mass , and two free parameters : the inclination @xmath28 and the mass - to - light ratio @xmath40 .
the results of this paper do not depend on the particular parametrization ( [ massdens ] ) that is used to characterize the m32 luminosity density .
the ( positive ) gravitational potential of m32 is assumed to be @xmath45 , where @xmath46 is the potential generated by the luminous matter with mass density ( [ massdens ] ) , and @xmath47 is the potential of a nuclear massive dark object .
the latter is taken to be the ` softened point mass ' potential @xmath48 this is the potential generated by a cluster with a plummer model mass density ( binney & tremaine 1987 ) .
for the case of a nuclear bh ( i.e. , a point mass ) the potential is keplerian and @xmath49 .
we do not include the potential of a dark halo around m32 ; there are no ( strong ) observational constraints on the possible presence and characteristics of such a dark halo .
also , if anything , adding a dark halo probably has a stabilizing effect ( ostriker & peebles 1973 ; stiavelli & sparke 1991 ) , and omitting a dark halo is therefore conservative in the present context . in vdm97c , models with two- and three - integral dfs
are constructed to interpret the available kinematic m32 data .
two - integral models do not reproduce all features of the data , but come close .
the more general three - integral modeling shows indeed that the best - fitting models do not have two - integral dfs .
however , they have a similar dynamical structure : @xmath50 , whereas for two - integral models @xmath51 . because the construction of @xmath52-body initial conditions is significantly more complicated for three - integral than for two - integral dfs
, we restrict ourselves here to testing only the stability of the two - integral models for m32 .
we do not expect the results to be much different for more general models . for a given gravitational potential
there is a unique _ even _
df @xmath53 that reproduces the mass density @xmath54 . for our models we calculated this df as described in vdm97c .
the _ total _
df is the sum of the part @xmath55 that is even in @xmath11 and the part @xmath56 that is odd in @xmath11 . in principle @xmath56
is determined completely by the mean streaming velocities @xmath57 , but these are not constrained well enough by the data to make an inversion feasible .
we therefore adopt a convenient parametrization for the odd part : @xmath58 , \label{fodddef}\ ] ] where @xmath59 is the maximum angular momentum at a given energy .
the function @xmath60 is a smoothed step function as defined in van der marel ( 1994 ) , with the parameter @xmath61 regulating the step - smoothness .
for any given potential , the free parameters @xmath62 and @xmath61 can be specified so as to best fit the observed rotation velocities .
our results should not depend sensitively on the particular choice of parametrization adopted here .
we chose to simulate models with @xmath63 and @xmath64 ( @xmath65 ) , using @xmath2512,000 particles .
the use of a softened dark object in this context ( rather than a point mass ) was motivated by numerical considerations .
however , the adopted value of @xmath66 is not inconsistent with the kinematic observations ( vdm97c ; the size of the smallest hst / fos aperture is @xmath67 ) .
there are no independent constraints on the inclination of m32 , and we therefore studied models with two representative values : @xmath3 ( edge - on ) and @xmath4 .
these models have intrinsic axial ratios @xmath5 and @xmath6 , respectively . for both inclinations we characterized the odd part of the df by @xmath68 and @xmath69 , which provides an adequate fit to the observed rotation velocities .
the mass - to - light ratio was chosen to be @xmath70 for @xmath3 , and @xmath71 for @xmath4 . a more complete study of the influence of the assumed inclination on the stability of the models would be interesting , but also expensive in terms of computing resources .
appendix a describes how initial conditions were generated from the df .
the most straightforward approach is to use equal - mass particles .
however , even with @xmath2512,000 equal - mass particles , the resolution near the center is poor .
the solid histogram in figure 1 shows the number of particles as function of radius @xmath72 for the @xmath4 model with equal - mass particles . to improve the resolution it is useful to adopt a spectrum of particle masses , such that particles near the center are on average less massive
this can be achieved by choosing the particle mass @xmath73 to be a decreasing function of the binding energy @xmath10 .
a convenient choice is @xmath74 \ > / \ > [ r_{\psi}(e)+r_1 ] \
> \rbrace^{\lambda } , \label{massspec}\ ] ] where @xmath75 is the gravitational potential , and @xmath76 is defined by @xmath77 .
after some experimentation , a mass spectrum was adopted with parameters @xmath78 , @xmath79 and @xmath80 .
this yields a particle distribution that well samples the density cusp around the dark nuclear object , as well as the main body of the galaxy ( figure 1 , dotted curve ) . for these multi - mass models
( with @xmath2512,000 ) , particles with @xmath81 have @xmath82 , while particles with @xmath83 have @xmath84 .
figure 2 shows a scatter plot of the particle mass @xmath73 as a function of radius for the @xmath4 multi - mass model .
we performed n - body simulations on the cray t3d parallel supercomputer at the pittsburgh supercomputing center .
the ` self - consistent field ' ( scf ) code developed by hernquist & ostriker ( 1992 ) was used , implemented for parallel architectures as described in hernquist , sigurdsson & bryan ( 1995 ) and sigurdsson ( 1997 ) . to improve the accuracy of the particle integration near the black hole , and to allow the smoothing length for the black hole to be small , the previously used standard time centered leap - frog integrator was replaced by a fourth order hermite integration scheme ( makino 1991 ; makino & aarseth 1992 ) . while the leapfrog has the virtue of being both simple and symplectic , a fixed time step leapfrog is vulnerable to spurious numerical scattering of orbits when potential gradients are large , e.g. , in steep density cusps and deep in keplerian potentials .
choosing a small enough fixed time step for the leapfrog integrator to be completely free of numerical scattering in the center is highly inefficient .
a hermite scheme provides a higher order , but relatively cheap integration scheme which excels at adapting to rapidly changing force terms . since the hermite scheme is not symplectic , global energy conservation is limited by the poisson noise in the global potential due to the discrete realization of the potential field ( see sigurdsson & mihos 1997 for a discussion ) .
a variable time step scheme was used , where the minimum time step was set by the minimum time step for any particle , using aarseth s condition on the force derivatives ( aarseth 1985 ; makino & aarseth 1992 ) . to improve accuracy near the center ( where time scales are short ) ,
while not slowing down the entire calculation too much , a modified two level ` multistep ' approach was used ( sigurdsson & mihos 1997 ) in which the accelerations due to the central massive object and the self - gravity of the innermost group of particles are updated more often than for the bulk of the particles at large radii .
this approach speeds up the calculations by an order of magnitude relative to a single step integration .
compared to the multistep approach that we used previously ( hernquist , sigurdsson & bryan 1995 ) this approach is about twice as fast at fixed integration precision .
the adaptive time step scheme is essential when the cusp is well resolved by the particle realization and the black hole smoothing length is small , to follow the particles near the center where forces and jerks are large .
the expansion uses a hernquist ( 1990 ) model as its lowest order term , for which @xmath85 at small radii . for the case of m32 the scale - length of this model
was chosen to be @xmath86 .
this minimizes the contribution to the potential from the higher - order terms in the expansion .
the calculations are done with units in which @xmath87 . in these dimensionless units ,
the smoothing length is @xmath88 , and the central dark mass is @xmath89 . in this system of units ,
time is measured in units of @xmath90 yrs , and the periods of circular orbits at radii of @xmath91 , @xmath92 and @xmath93 are @xmath94 , @xmath95 and @xmath96 , respectively .
we performed small simulations with 25,600 and 51,200 particles to test the dependence of the results on the number of radial and angular terms in the scf expansion .
based on these results we adopted @xmath97 and @xmath98 for the final simulations .
non - zero @xmath99 terms were included in the expansion to allow any non - axisymmetric instabilities to develop .
figure 3 illustrates the accuracy of the scf expansion , by comparing its circular velocity to that of the true model potential . at small radii the _ stellar _ contribution to the circular velocity is poorly represented , due to the finite number of radial terms in the scf expansion ; in fact , increasing @xmath100 by a small amount does not lead to a significantly improved fit .
however , the _ total _ circular velocity is well represented for all @xmath101 , because it is dominated by the central dark object at the small radii where the scf expansion fails . only at radii @xmath102
does the numerical approximation to the total circular velocity become poor , and only as a result of the numerically imposed softening ( which causes the circular velocity at @xmath103 to be dominated by the stellar density cusp , rather than by the dark object ) . in principle
, the potential could have been represented more accurately by using another set of basis functions .
a basis set that has @xmath104 at small radii for its lowest order term , with @xmath30 as in equation ( [ massdens ] ) , would probably provide a better description for m32 .
such basis sets can indeed be constructed ( saha 1993 ; dehnen 1996 ; zhao 1996 ) , but in practice are generally more difficult to implement and slower in use . the fact that in our application the potential at small radii is dominated by the dark object , justifies our use of the more conventional hernquist & ostriker ( 1992 ) basis set .
we also used smaller simulations to make a proper choice for the time integration parameters . in the final simulations the smallest step size ( near the center )
was typically @xmath105 in dimensionless units , with the step size at larger radii being @xmath10650 times larger .
global energy conservation was about 0.5% over @xmath107 .
one of the more useful diagnostics to monitor in the simulations is the evolution of the axial ratios of the mass distribution as a function of radius . for this we used an iterative scheme similar to that used by dubinski & carlberg ( 1991 ) .
a mass bin is adopted that contains a fixed fraction of the total mass ( e.g. , the central 010% ) .
the moment of inertia tensor is calculated for the particles that fall in the bin , and diagonalization then yields the axial ratios @xmath108 and @xmath109 ( the eigenvalues of the tensor ) and the angles of the principal axes with respect to the initial coordinate axes .
iteration is required because the sorting of the particles in the ellipsoidal radial coordinate depends on the angles and axial ratios .
we found that the algorithm has convergence problems when the number of particles is small , or at small radii .
to improve this , we made modifications to the basic algorithm to allow the mass fraction to vary , to ensure that the centroid of the particles in the bin was at all times well inside their mean ellipsoidal radius .
iteration was done by taking the geometric mean of the derived axial ratio and the previously derived axial ratio .
this provides for stabler convergence of the shape parameters , and allowed sensible axial ratio estimates to be obtained for a large range in radius ( see also sigurdsson & mihos 1997 ) .
we ran an equal - mass model and a multi - mass model for each inclination .
the equal - mass simulations were run to @xmath110 for the @xmath3 model , and to @xmath111 for the @xmath4 model , to explore global stability and stability at large radii .
the multi - mass models were run to @xmath112 to explore stability at small radii .
the main result is that _ the models for both inclinations are dynamically stable .
_ we illustrate this by discussing and displaying the time evolution for the @xmath4 model .
the results for the edge - on model are similar .
figure 4 shows the initial and final mass densities .
figure 5 shows the evolution of the axial ratios @xmath108 and @xmath109 for four radial bins . at small radii we display the results from the multi - mass models , and at large radii those from the equal - mass models .
there are no evolutionary trends in the mass distribution .
the axial ratios are constant at their values for the continuous limit ( @xmath113 , @xmath114 ) , to within the accuracy imposed by the finite number of particles .
the orientation of the principal axes of the mass distribution also showed no trends with time .
a more sensitive diagnostic of possible instabilities is provided by an analysis of the individual expansion terms of the mass distribution .
such an analysis is straightforward , because the scf technique makes a harmonic decomposition of the density and potential at each time step .
figure 6 shows the time evolution of some of the expansion terms , in particular those that could have been expected to show signatures of an instability . however , there is no sign of even a weak lopsided ( @xmath115 ) or bar ( @xmath116 ) instability .
other expansion terms showed the same result of constant amplitudes with time , to within the noise limits imposed by the discreteness of the model . not only the mass distribution is stable , but also the phase - space structure of the model .
figure 7 shows for the multi - mass model the initial and final mean azimuthal velocity @xmath117 , and the rms velocities @xmath118^{1/2}$ ] , @xmath119^{1/2}$ ] and @xmath120^{1/2}$ ] , all as a function of the spheroidal radius @xmath99 .
there is some evolution in the velocities at radii close to the softening scale @xmath66 , due to small number statistics and the finite resolution of the scf expansion .
overall , however , the velocity structure of the model is stable .
the equality of @xmath119^{1/2}$ ] and @xmath120^{1/2}$ ] is consistent with the @xmath7 form of the df .
figure 8 shows that the model remains in virial equilibrium throughout , as expected from the stability of both the mass distribution and the kinematic structure .
two integral @xmath7 models with a central bh provide a reasonable fit to the observed kinematics of m32 .
previous work indicated that @xmath7 models with as much rotation as m32 could be bar - unstable , but none of the previous studies took into account both a central cusp in the mass density and a central bh .
we therefore constructed particle realizations of @xmath7 models for m32 , and calculated their time evolution using a scf n - body code on a cray t3d parallel supercomputer .
models were studied for two representative inclinations , @xmath3 and @xmath4 , corresponding to intrinsic axial ratios of @xmath5 and @xmath6 , respectively . both these models were found to be dynamically stable .
this result has two implications .
first , the equilibrium models that best fit the kinematic data for m32 are stable , and therefore physically meaningful .
second , imposing the requirement of stability on the models does not put very stringent constraints on the inclination or the intrinsic axial ratio of the models . previous work on the stability of @xmath7 models has shown that the @xmath116 bar - mode is the only mode that might be unstable for systems rounder than @xmath8 , and that the susceptibility to this instability increases with flattening and rotation rate ( dehnen 1996 ; sellwood & valluri 1997 ) . here
we have shown that @xmath7 models for m32 as flat as @xmath6 are stable .
combined with the fact that m32 has a higher rotation rate @xmath121 than nearly all other elliptical galaxies ( m32 rotates more rapidly than an oblate isotropic rotator ; van der marel 1994 ) , this suggests that all @xmath7 models with @xmath9 that are constructed to fit real galaxies will generally be stable .
this might in fact be true even for flatter models , but that is not something that can be addressed with the results presented here .
nonetheless , the general applicability of @xmath7 models to real galaxies remains unclear .
two independent lines of observational evidence suggest that elliptical galaxies , especially those flatter than e2 , can not have @xmath7 dfs .
the models tend to possess too much motion on the major axis relative to the minor axis ( binney , davies & illingworth 1990 ; van der marel 1991 ) , and line - of - sight velocity profiles that are too flat - topped ( bender , saglia & gerhard 1994 ) .
either way , the relative simplicity of these models and their demonstrated use for the case of m32 , are likely to leave them as a popular tool for the interpretation of stellar kinematic data for some time to come .
a more detailed study of the general stability properties of @xmath7 models is outside the scope of the present paper .
however , it will certainly be of interest to study the influence of a central dark object on the stability of flattened rotating models , and , more generally , to search for evolution in the region close to the dark object as it grows with time .
parallel n - body integration algorithms such as those used here are well suited to address these issues .
the longest simulation we ran covered @xmath122 yrs .
this is many dynamical times in the central region of m32 , and therefore sufficient to test dynamical stability .
however , it is only 1% of the hubble time , and therefore insufficient for a study of very slow secular evolution .
although collisionless secular evolution need not necessarily be present , m32 will certainly suffer some secular evolution as a result of two - body relaxation ; the relaxation time in the inner regions is only @xmath123 yrs , independent of radius ( lauer 1992 ) .
simulations that address the effects of this relaxation using a hybrid direct / scf n - body code are in progress ( hemsendorf , spurzem & sigurdsson 1997 ) .
more complicated simulations are required to also include the effects of mass loss due to stellar evolution and stellar collisions , which operate on similar time scales .
we are grateful to tim de zeeuw , kathryn johnston , chris mihos , gerry quinlan , hans - walter rix and rainer spurzem , for helpful discussions , useful comments , and for collaboration in various related projects .
the calculations reported here were performed at the pittsburgh supercomputing center ; computing resources were also provided by the national center for supercomputing applications .
financial support was provided by : ( a ) nasa , through grant number # go-05847.01 - 94a , and through a hubble fellowship # hf-1065.01 - 94a awarded to rpvdm , both from the space telescope science institute which is operated by the association of universities for research in astronomy , incorporated , under nasa contract nas5 - 26555 ; ( b ) nsf , through grant asc 93 - 18185 ; ( c ) the european union , through a tmr cat .
30 marie curie fellowship awarded to ss ; ( d ) the presidential faculty fellows program ; ( e ) the aspen center for physics ( which also provided hospitality ) .
the n - body simulations require the generation of initial conditions from known @xmath7 dfs .
this appendix describes the techniques that were used for this .
the cases of equal - mass particles and of particles with a mass spectrum depending on energy are described separately .
the former is more straightforward , while the latter can be useful to increase the numerical resolution of a simulation near the center . for equal - mass particles one can first draw the particle positions .
the ellipsoidal radius @xmath99 and ellipsoidal angle @xmath127 in the meridional @xmath128 plane are defined through @xmath129 where @xmath26 is the axial ratio of the system .
the total mass of the system is @xmath130 where @xmath131 is the azimuthal angle and @xmath132 is the mass density defined in equation ( [ massdens ] ) .
the probability distributions of @xmath131 , @xmath127 and @xmath99 for a random mass element are therefore @xmath133 ) ; \qquad \cp(\tau ) = { 1\over2}\sin\tau , \quad ( \tau \in [ 0,\pi ] ) ; \qquad \nonumber \\ & & \cp(m ) = 4 \pi q \ > m^2 \rho(m ) \ > / \
> m , \quad ( m \in [ 0,\infty ) ) .\end{aligned}\ ] ] the total mass @xmath134 is easily calculated numerically .
the ` rejection method ' ( e.g. , press 1992 ) with comparison function @xmath135 can be used to draw a random @xmath99 from @xmath136 .
the constants @xmath61 and @xmath137 must be chosen such that @xmath138 for all @xmath139 . a random position from the mass density
is then obtained as follows : 1 .
draw a random number @xmath140 .
set @xmath141 $ ] .
2 . draw a random number @xmath142 .
reject the previous step and return to step 1 if @xmath143 .
3 . draw random numbers @xmath144 and @xmath145 . set @xmath146 and @xmath147 .
after the particle positions have been drawn from the mass density , the particle velocities must be drawn from the df .
the discussion can be restricted to @xmath7 models in which all stars have @xmath148 .
such a df , @xmath149 , is characterized by the choice of a maximally rotating odd part .
once initial conditions for this df are available , initial conditions can be obtained for any other df @xmath7 with the same even part ( and arbitrary odd part ) by simply reversing the velocity of each particle with probability @xmath150 .
the maximum angular momentum at a given energy , @xmath59 , is the angular momentum of a circular orbit in the equatorial plane .
define @xmath151 .
the maximum @xmath62 for a star with known position and energy is @xmath152^{1/2 } / l_{\rm max}(e ) , \label{etamax}\ ] ] where @xmath75 is the ( positive ) gravitational potential .
the mass - density is the integral of the df over velocity space , which can be written as ( e.g. , binney & tremaine 1987 ) @xmath153 the joint probability distribution of @xmath154 at a fixed position is therefore @xmath155,\quad \eta \in [ 0,\eta_{\rm max } ] ) .\ ] ] for flattened models , the df is a monotonically increasing function of @xmath62 ( e.g. , qian 1995 ) .
hence , one may use the rejection method with comparison function @xmath156 } \cp(e,\eta_{\rm max } ) , \ ] ] which can be shown to be finite .
it is convenient to calculate @xmath157 on a grid in the meridional plane , such that its value at any position can be obtained quickly through spline interpolation . a random velocity vector for a particle with known position
is then obtained as follows : 1 .
draw random numbers @xmath140 and @xmath142 . set @xmath158 and @xmath159 .
2 . calculate @xmath160 from equation ( [ etamax ] ) .
draw a random number @xmath144 .
reject the previous step and return to step 1 if either @xmath161 or @xmath162 .
3 . draw a random number @xmath145 .
set @xmath163 .
calculate the particle velocities from @xmath164 particle positions and velocities must be drawn simultaneously if the particle mass @xmath73 depends on the phase space coordinates , @xmath165 .
the phase space number density is the ratio of the df and the particle mass . with equations ( [ totmass ] ) , ( [ rhoasint ] ) and ( [ velocities ] ) one obtains formally for the total number of particles in the system @xmath166
the integrand is independent of the angles @xmath131 and @xmath167 , provided that @xmath73 does not depend on them . for efficient use of the rejection method
one must have phase space coordinates with a finite range , and an integrand that is as close as possible to constant .
this can be achieved with the coordinate transformations @xmath168 which yield @xmath169 the arbitrary constants @xmath170 and @xmath171 can be chosen to optimize the efficiency of the algorithm .
the integrand is a known function of @xmath172 , so one may use the rejection method with the comparison constant @xmath173 which is finite for most @xmath174 of interest .
its value can be found with , e.g. , the simplex algorithm of press ( 1992 ) .
it can be shown that @xmath175 for an axisymmetric system .
particles with an arbitrary mass spectrum @xmath174 can therefore be generated using the following algorithm : 1 .
draw random numbers @xmath176 and @xmath145 . set @xmath177^{-2 } \right\rbrace .\end{aligned}\ ] ] 2 .
calculate @xmath160 from equations ( [ mtau ] ) and ( [ etamax ] ) .
reject the previous step and return to step 1 if @xmath178 .
3 . draw a random number @xmath179 .
calculate the integrand in equation ( [ npart ] ) .
reject the previous steps and return to step 1 if the integrand is @xmath180 .
4 . draw random numbers @xmath181 and @xmath182 . set @xmath183 and @xmath184 .
calculate the velocities from equation ( [ velocities ] ) .
after the required number of particles ( @xmath52 ) has been drawn , all particle masses @xmath185 must be renormalized such that the total mass @xmath186 equals @xmath134 , as given by equation ( [ totmass ] ) . for a system of equal - mass particles
one simply sets @xmath187 for @xmath188 . | we study the stability of stellar dynamical equilibrium models for m32 .
kinematic observations show that m32 has a central dark mass of @xmath0 , most likely a black hole , and a phase - space distribution function that is close to the ` two - integral ' form @xmath1 .
m32 is also rapidly rotating ; 8590% of the stars have the same sense of rotation around the symmetry axis .
previous work has shown that flattened , rapidly rotating two - integral models can be bar - unstable .
we have performed n - body simulations to test whether this is the case for m32 .
this is the first stability analysis of two - integral models that have both a central density cusp and a nuclear black hole .
particle realizations with @xmath2512,000 were generated from distribution functions that fit the photometric and kinematic data of m32 .
we constructed equal - mass particle realizations , and also realizations with a mass spectrum to improve the central resolution .
models were studied for two representative inclinations , @xmath3 ( edge - on ) and @xmath4 , corresponding to intrinsic axial ratios of @xmath5 and @xmath6 , respectively .
the time evolution of the models was calculated with a ` self - consistent field ' code on a cray t3d parallel supercomputer .
we find both models to be dynamically stable .
this implies that they provide a physically meaningful description of m32 , and that the inclination of m32 ( and hence its intrinsic flattening ) can not be strongly constrained through stability arguments .
previous work on the stability of @xmath7 models has shown that the bar - mode is the only possibly unstable mode for systems rounder than @xmath8 ( i.e. , e7 ) , and that the likelihood for this mode to be unstable increases with flattening and rotation rate .
the @xmath7 models studied for m32 are stable , and m32 has a higher rotation rate than nearly all other elliptical galaxies .
this suggests that @xmath7 models constructed to fit data for real elliptical galaxies will generally be stable , at least for systems rounder than @xmath9 , and possibly for flatter systems as well . # |
precision measurements using systems with small dissipation was a pioneering field of research of vladimir braginsky@xcite .
as pointed out in his work , such systems are very valuable in many areas of physics since they preserve coherence for very long times and may serve as very sensitive probes of physical quantities , such as material property characterisation@xcite and tests of fundamental physics@xcite , where the ultimate limit is given by quantum mechanics@xcite .
one area gaining considerable recent attention is the development of low - temperature microwave cavities , with high - q factors and low noise readouts to search for the dark matter axion@xcite .
axion searches in the microwave frequency band ( greater than @xmath0ghz ) poses certain difficulties related to contradicting requirements . as the frequency space pushes towards higher frequencies , detector cavity sizes
shrink decreasing the mode volume and corresponding sensitivity .
an obvious solution to the problem is to increase the number of detecting cavities @xcite .
unfortunately , this immediately leads to an increase in system complexity as such a system requires the need for additional amplifiers , microwave lines , etc .
so , there is a need for solutions that can lead to an increase in the axion detecting mode volume while preserving the number of detecting systems and high resonant frequency . in particular ,
high mass axions yielding high frequency photons ( @xmath1ghz ) are motivated theoretically @xcite , and by some curious experimental results @xcite , but are as yet largely unprobed , although some recent proposals and experiments have begun to move in this direction @xcite . in this work ,
we propose new methods to scale the mode volume based on negative coupling between cavities .
this approach gives an advantage over simultaneous averaging over the same number of independent cavities ( giving the same scaling law ) , as the proposed approach requires only one measurement system .
axion electrodynamics may be considered as an extension to the classical electromagnetism with an additional lagrangian term coupling the axion scalar field @xmath2 and the electric @xmath3 and magnetic @xmath4 components of the electromagnetic field@xcite : @xmath5 where @xmath6 is the coupling strength .
the conventional approach @xcite to detect the presence of axions is to apply the strongest available external magnetic field , @xmath7 , and measure photons created by the three particle interaction ( [ ep001e ] ) , the so - called axion - two photon coupling .
the measurement is usually done with a single photonic cavity as a probe antenna , so the interaction reduces to direct coupling between axions and cavity photons : @xmath8 where @xmath9 ( @xmath10 ) and @xmath11 ( @xmath12 ) are creation ( annihilation ) operators for the cavity mode ( @xmath13=i$ ] ) and axions .
the coupling coefficient @xmath14 is proportional the the dc field and is a function of cavity geometry and fundamental parameters @xcite : @xmath15 where @xmath16 is the cavity resonance frequency , @xmath17 is the mode volume of the cavity , @xmath18 is the loaded quality factor and @xmath19 is the unity scaled electromagnetic form factor that is a scaled overlap integral between the external field and the cavity mode .
this parameter depends only on the cavity geometry but not on its volume , thus it stays constant when the cavity is scaled . on the other hand
, @xmath19 depends on the form of a mode used for the detection . for a typical haloscope with a uniform , static magnetic field , @xmath7 , along a single axis ,
the electric field generated by axion to photon conversion is parallel to this axis . since , for detection purposes , axions may be thought of as a space uniform field the maximally sensitive mode is the one having the most uniform structure and the lowest number of nodes , i.e. the lowest order mode .
thus , the aim of the detector design for the axion search is to maximise the mode volume @xmath17 and @xmath19 for the given frequency @xmath16 .
instead of one isolated cavity , it is possible to consider a set of a regular one dimensional chain of linearly coupled cavities .
each cavity supporting a particular mode may be considered as a harmonic oscillator .
a particularly instructive model is an @xmath20 circuit , characterised by an ideal capacitance @xmath21 and inductance @xmath22 .
the next - neighbouring individual elements of this chain are coupled via mutual inductance @xmath23 .
the hamiltonian of the chain may be written in the form : @xmath24 where @xmath25 , @xmath26 .
it is important to underline that the cavity - cavity coupling strength @xmath27 and thus the mutual inductance @xmath23 could be both negative or positive depending on the mutual winding of neighbouring inductors . more precisely , the mutual inductance can be designed to be in the region @xmath28 $ ] .
it is straightforward to obtain a dispersion relationship from equation ( [ lp001r ] ) by substituting the wave solution @xmath29 : @xmath30 where @xmath31 , @xmath32 is the wave number and @xmath33 is the angular frequency of a wave propagating along the chain . here parameter @xmath34 $ ] due to limits imposed on @xmath23 .
if , for example , @xmath35 , then we obtain a dispersion relationship of a simple vibration lattice of masses and springs where , for small wave numbers , @xmath36 .
although , it is more interesting to consider the parameter subspace where @xmath37 . in the limiting case when @xmath38 , @xmath33 spans between @xmath39 for @xmath40 and @xmath41 for @xmath42 .
what is important about @xmath37 is that the lower wave number waves correspond to higher frequencies and vice a versa .
this property can be exploited in an axion search in order to get higher mode volumes at high frequencies , without resorting to moving to higher order modes with decreased form factors .
indeed , the fundamental mode of the chain is the one that has the highest sensitivity to the detectable axion signal , so by reversing the sign of the coupling , one can shift this mode from the lowest to the highest frequency achievable with such a chain .
the inverse is true for the highest order mode that has the lowest axion sensitivity . under the considered experimental conditions , to quantify the sensitivity to axions of a cavity chain in terms of eigenvalues of system ( [ lp001r ] )
, one needs to calculate an overlap between a chain mode and a distribution of sensitivities of coupled cavities .
the product of overall chain form factor and volume may be viewed as a figure of merit for this , and maybe be expressed for its @xmath32th order mode in terms of the form factor and volume of each element of the chain @xmath43 and the value of normalised eigenvectors @xmath44 of the @xmath32th order mode : @xmath45 where @xmath46 is the number of cavities in the chain . from this relation
it follows that for identical cavities the sensitivity is maximised for the mode with the largest sum of elements of the corresponding eigenvector .
this is fulfilled for the uniform mode giving the sum equals to one : @xmath47 for non uniform modes the divergence from this relation may be expressed as a ratio : @xmath48 here , @xmath49 is a numeric value of @xmath50 , where the value of unity represents completely uniform modes and is the ideal value , in practice this value needs to be maximised . to demonstrate the effect of negative coupling on resonance frequencies of a chain of cavities
, we calculate the eigenfrequencies of such a system .
the distribution of resonance frequencies for a system with @xmath46 cavities and different coupling parameters @xmath51 as a function of mode order @xmath32 is shown in fig .
[ scale1 ] ( a ) .
the plot illustrates the fact that higher frequency modes correspond to lower order numbers @xmath32 and , thus , to larger mode volumes that can be observed in fig .
[ scale1 ] ( b ) .
the maximum achievable frequency for the chain is @xmath39 , this follows form the dispersion relationship ( [ lp002r ] ) . for higher dimensional structures with dimensionality @xmath52 , the frequency scales as @xmath53 .
( b ) as a function of mode order @xmath32 for a chain of @xmath46 cavities.,scaledwidth=45.0% ] fig .
[ scale1 ] ( b ) demonstrates divergence factor @xmath49 calculated as a sum over mode eigenvectors . here
, even order modes for all parameter systems give zero parameter values as positive and negative parts of eigenvectors cancel .
also , even the lowest order mode does not reach the limit of @xmath54 .
this result is explained by the fact that due to the open boundary conditions imposed on both ends of the chain , none of these modes are uniform showing decay of intensity from the chain centre to its ends .
another important conclusion is that the overall mode volume increases with @xmath46 , thus , it is possible to scale this parameter ( and all results that depend on @xmath55 ) with @xmath46 .
scaling of the form factor and volume product was calculated for the first mode , @xmath40 , with a varying number of cavities , @xmath46 , for two types of boundary conditions .
the results are shown in fig .
[ scale2 ] , which demonstrates the factor of @xmath56 scaling law .
the same result can be obtained by power combining @xmath46 independent cavities , but this introduces many complex technical challenges @xcite .
recent work suggest that post - processing the data acquired from @xmath46 independent cavities may yield a small improvement over the system proposed here , although , this would require @xmath46 independent measurement systems @xcite .
additionally , the figure shows that the chain with open boundary conditions has suboptimal form factors due to non - uniformity caused by edge effects .
a possible complication with the proposed system of a negatively coupled chain of cavities is high density of modes towards the zeroth order
. this problem can make it difficult to tune the most sensitive probe mode _ in situ _ due to high mode density .
however , in principle , if all cavity frequencies , @xmath57 , are tuned synchronously , the whole dispersive curve moves up or down with the constant separation between the chain modes .
this implies no avoided crossings or other mode interaction effects are present to limit the tunability of the most sensitive mode .
tuning of all cavities simultaneously can be achieved if cavities are made of individual electromagnetically resonant elements enclosed in a shared conducting movable surface .
an example of such structure is a multi - element re - entrant type structures described below where one controls post gaps and thus the equivalent capacitance @xmath21 and thus the @xmath57 for all elements .
a system of coupled cavities is often realised using 2d split ring resonators or similar structures making a base for implementation of metamaterials . recently ,
a 3d version of this resonator , a multi - post reentant cavity@xcite , has been proposed as a controllable base for metastructures @xcite .
such a system is made up of a number of conducting posts attached to one cavity wall and closely approaching another with their tips .
every post forms a separate resonant structure with equivalent inductance primarily due to the length and capacitance formed by the gap between the tip and the opposite surface .
the additional advantage of such system is its high tunability and possibility to control all post gaps simultaneously .
it is possible to demonstrate that such a system of parallel straight posts gives negative mutual inductance and , thus , a negative sign of @xmath51 .
so , a system constructed of a regular grid of such posts effectively composes a right - handed metamaterial with a normal dispersion curve .
the sign of coupling between resonators can be reversed by replacing straight posts with coils as shown in the insets of fig .
[ dispersion ] .
such a system can be still regarded as a re - entrant structure as the gap between the coils ends and the opposite surface forming capacitance is preserved .
thus , the system inherits one of the most important features of the re - entrant cavity , i.e. high tunability by changing post gap .
this feature is important for axion detection as it allows to scan a wide frequency range .
indeed , highly tuneable re - entrant cavities have previously been proposed as a resonator for axion searches @xcite . for different lattice parameters @xmath2 compared to the coil diameter @xmath52 .
color plots show the electric and magnetic field distribution for the lowest and highest order modes .
the insets show 3d views of the system .
( b ) filling factors for all re - entrant modes of @xmath58 and @xmath59 chains of coupled coils as a function of mode order @xmath32.,scaledwidth=47.0% ] the performance of the coil structure has been confirmed by full 3d finite element modelling ( fem ) in comsol multiphysics .
the resulting dispersion relationship is shown in fig .
[ dispersion ] ( a ) .
four solid curves represent three values of the lattice spacing @xmath2 relative to the coil outer diameter @xmath52 .
resonance frequencies are scaled to the frequency of a stand alone reentrant coil @xmath60 .
the color plots demonstrate the field distribution for the highest and lowest order modes .
all dispersion curves have a character specific to the negatively coupled cavity structure described above using the harmonic oscillator model .
in particular , the highest frequency mode of the chain is characterised by the most uniform distribution of electric and magnetic fields .
this fact will result in the strongest axion sensitivity at the highest frequency of the band as seen from calculations of the @xmath41th mode form factor @xmath61 in fig .
[ dispersion ] ( b ) demonstrating dependence of the filling factor on the mode order .
it is also observed that the coupling between the chain elements grows with decreasing spacing between the coils . in the limit of large @xmath2
, the dispersion curve is a horizontal line describing the case of uncoupled resonators having the same frequency @xmath57 .
the opposite limiting case is a situation when coils are placed without intermediate gaps , i.e. @xmath62 .
in fact , the situation with @xmath63 is possible , but leads to other effects that can not be explained by normal mode decomposition on the simple chain of cavities . to demonstrate the ability to scale the axion sensitivity
, we numerically compute the product of form factor and volume for the most sensitive ( zero order ) mode , which can be represented as : @xmath64 where @xmath65 is the chain volume and @xmath66 is the total electric field in the cavity .
this quantity characterises sensitivity of the whole structure .
the results for different numbers of coils @xmath46 in the open boundary condition case is shown in fig .
[ fig : cvvn ] , which shows that the sensitivity increases proportional to the number of cavities for larger chain lengths of @xmath67 . for the lower values of @xmath46 ,
the performance is suppressed by a constant value that becomes negligible for @xmath67 .
this suggests that this suppression is due to edge effects . as it is seen from fig . [ dispersion ] ( b ) , the re - entrant mode of the coil with gap exhibits a moderate sensitivity due to significant components of the e field in the @xmath68 plane .
although the proposed array structure improves the performance proportional to @xmath46 , the base sensitivity of a single coil requires some further fine tuning to improve the performance .
this may include optimisation of such parameters as the coil radius , thickness , length , axial pitch , etc or the coil structure itself , but such optimisation falls out of the scope of this work .
we also note that the production of many , nominally identical , small coils is well within the limits of standard manufacturing processes .
the main difficulty related to the physical realisation described in the previous section is manufacturability of the coil - like elements that are used to control the sign of cavity couplings . on the other hand , it is possible to mimic the coil - like ( chiral ) behaviour between array elements by grouping cavities into quasi - molecules . according to this idea , we consider closely situated inter - coupled groups of cavities as building blocks of sensing arrays and design internal structure of these groups in such a way that desired characteristics of couplings between adjacent groups are achieved . in this case , it is possible to choose the simplest possible structure for the cavity , i.e. simple straight re - entrant post , but make internal group structure more complex in order to mimic negative couplings in the array elements . in other words ,
the complexity of a cavity is transferred onto the complexity of the group structure . the idea of cavity groups or `` quasi - molecules '' has been demonstrated in multi - post reentrant cavity systems with straight posts , to realise a photonic topological insulator@xcite .
this work proposed to implement left - handed and right - handed quasi - molecules with straight ( non - chiral ) posts by decreasing the gap size under each post when going clock - wise or anti - clockwise across each molecule . with this chiral structure at hand , 1d and 2d arrays of these molecules demonstrated bulk / edge correspondence as well as one - way photon transport .
moreover , the system preserved such nontrivial topological effects when realised on a regular two dimensional grid of posts with accordingly adjusted gaps leading to possible reconfigurability of the overall topological properties . and decrease in resonance frequency @xmath69 of a corresponding harmonic oscillator.,scaledwidth=30.0% ] the idea of left handed and right handed groups of straight posts
may be applied in the attempt to avoid coil - like posts in the cavity array .
indeed , it is possible to replace one cavity with the coil - like structure with a group of plain cavities having certain sense of handedness and further implementing an array of such groups .
following the first demonstration of this idea@xcite , we consider groups of four posts with different gaps @xmath70 distributed clockwise or anti - clockwise ( see fig .
[ groups ] ) implemented on a regular grid . in this section , each post / cavity is modelled as a harmonic oscillator of a certain resonance frequency @xmath69 positively coupled to its nearest neighbours .
for the numerical experiment , we consider a chain of 40 molecules with the same handedness and linear change of the resonant frequency of cavities across a single molecule . as the first test , we compute the resonance frequencies @xmath71 and the uniformity of the mode @xmath55 ( see eqn .
( [ lp005r ] ) ) for every eigenmode of the array .
the described system exhibit multiple frequency bands , for this reason the following discussing will be given in terms of mode number @xmath72 counted from the lowest to the highest in frequency .
this is in contrast to the mode order @xmath32 counted from the most uniform to the mode with the maximum number of alternations .
result are shown in fig .
[ divergence ] depicting four frequency bands .
in particular , the uniformity , @xmath73}$ ] , increases towards the highest frequency mode of the second and fourth frequency bands , indicating the desired property of negative couplings between parts of the structure .
( a ) and mode uniformity factors @xmath55 ( b ) for a chain of post groups as a function of mode number @xmath72 . vertical lines separate the four frequency bands.,scaledwidth=50.0% ] in order to investigate the scaling laws for the considered chain , we calculate the form factor of the highest frequency mode as a function of total number of cavity quasi - molecules @xmath74 ( here @xmath75 ) .
[ scaling ] demonstrates the result of the form factor scaling for the cases of open and periodic boundary conditions .
these conditions are considered with respect to the cavity groups rather than individual cavities .
it is possible to optimise the system so that the periodic boundary condition case approaches the @xmath74 line by increasing the difference between resonance frequencies of cavities within a single molecule and adjusting inter - cavity coupling .
finally , a periodic lattice of groups with different handedness may lead to similar effects but its spectrum would be more complex with each of bands splitting into two .
[ scaling ] demonstrates that the chain of quasi molecules of cavities with the same handedness exhibits a @xmath76 scaling law similar for the case of coil - like elements .
however , this solution requires at least four times as many posts or cavities , but does not require a complex coil structure .
this may result in higher mode density and more complex mode structures but eliminates the necessity of manufacturing identical coils . under posts for one unit cell.,scaledwidth=50.0% ] to confirm the results of the analysis of the cavity system , we perform full 3d fem analysis for a straight post realisation of such a system .
the system is taken to have 16 quasi - molecules giving 64 as the total number of posts , with the post gaps changed linearly across each unit cell . the simulated resonance frequencies and filling factors are shown in fig .
[ postsfem ] where it is seen that the highest resonance frequency of the forth band exhibits the largest filling factor , as expected .
it should be mention that the system may be further parametrically optimised to achieve higher sensitivity by optimising the post radii , distances and gaps .
such optimisation falls out of the scope of this work .
in conclusion , we have considered metastructures of negatively coupled cavities that exhibits a dispersion relationship that can help to enhance axion haloscope sensitivity in higher frequency ranges .
the structure exhibits the most uniform and hence sensitive modes at the highest frequency of the dispersion curve , higher than the frequency of an uncoupled individual element .
thus , the whole metastructure works as an axion sensitive resonator whose mode volume may be infinitely expanded with the resonant frequency held constant .
this technique requires only one measurement system ( a set of amplifiers , signal lines , mixers and data acquisition channels ) , and does not require the synchronization and phase matching of a large array of independent cavities .
moreover , it is demonstrated that this behaviour is possible to achieve in a real microwave cavity employing coils as resonant cavity elements or re - entrant posts , grouped into left or right handed unit cells .
+ + this work was supported by the australian research council grant no . ce110001013 .
32ifxundefined [ 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 _ _ ( , ) link:\doibase 10.1126/science.209.4456.547 [ * * , ( ) ] http://link.aps.org/doi/10.1103/physrevlett.38.1440 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.50.925 [ * * , ( ) ] _ _ , http://wwwlib.umi.com/dissertations/fullcit?p3019020[ph.d . thesis ] , ( ) link:\doibase 10.1142/s0217751x14430040 [ * * , ( ) ] , in @noop _ _ ( ) link:\doibase 10.1103/physrevlett.118.071802 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.111.231801 [ * * , ( ) ] in https://inspirehep.net/record/1500189/files/arxiv:1611.08082.pdf[__ ] ( ) link:\doibase 10.3204/desy - proc-2015 - 02/woohyun_chung [ * * , ( ) ] link:\doibase 10.1103/physrevlett.118.061302 [ * * , ( ) ] , link:\doibase 10.1103/physrevlett.118.091801 [ * * , ( ) ] , in link:\doibase 10.3204/desy - proc-2015 - 02/simanovskaia_maria [ _ _ ] ( ) pp .
, http://link.aps.org/doi/10.1103/physrevlett.58.1799 [ * * , ( ) ] http://link.aps.org/doi/10.1103/physrevlett.51.1415 [ * * , ( ) ] link:\doibase 10.1103/physrevd.36.974 , 10.1103/physrevd.32.2988 [ * * , ( ) ] , http://link.aps.org/doi/10.1103/physrevlett.104.041301 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.104.041301 [ * * , ( ) ] in link:\doibase 10.1073/pnas.1308788112 [ _ _ ]
( ) http://link.aps.org/doi/10.1103/physrevlett.117.159901 [ * * , ( ) ] @noop ( ) ,
@noop * * , ( ) @noop * * , ( ) @noop * * , ( ) http://stacks.iop.org/1367-2630/17/i=2/a=023003 [ * * , ( ) ] http://scitation.aip.org/content/aip/journal/jap/118/20/10.1063/1.4936268 [ * * , ( ) ] http://link.aps.org/doi/10.1103/physrevapplied.6.064006 [ * * , ( ) ] link:\doibase 10.1103/physrevd.94.042001 [ * * , ( ) ] | using eigenmode analysis and full 3d fem modelling , we demonstrate that a closed cavity built of an array of elementary harmonic oscillators with negative mutual couplings exhibits a dispersion curve with lower order modes corresponding to higher frequencies .
such cavity arrays may help to achieve infinitely large mode volumes for boosting sensitivity of the axion searches , where the mode volume for the composed array scales proportional to the number of elements , but the frequency remains constant .
the negatively coupled cavity array is demonstrated with magnetically coupling coils , where the sign of next - neighbour coupling ( controlled with their chirality ) sets the dispersion curve properties of the resonator array medium .
furthermore , we show that similar effects can be achieved using only positively coupled cavities of different frequencies assembled in periodic cells .
this principle is demonstrated for the multi - post re - entrant system , which can be realised with an array of straight metallic rods organised in chiral structures . |
consider a multicast network , which is a finite directed acyclic multigraph with a unique source node @xmath8 and a set @xmath5 of receivers .
every edge in the network represents a noiseless transmission channel of unit capacity .
the source generates @xmath9 data symbols belonging to a fixed symbol alphabet and will transmit them to all receivers via the network . the maximum flow , which is equal to the number of edge - disjoint paths , from @xmath8 to every receiver is assumed to be no smaller than @xmath9 .
the network is said to be solvable if all receivers can recover all @xmath9 source symbols based on their respective received data symbols .
when the network has only one receiver , it is solvable by network routing .
when @xmath10 , the paradigm of network routing does not guarantee the network to be solvable .
the seminal paper @xcite introduced the concept of network coding ( nc ) and proved that the considered multicast network has a nc solution over some infinitely large symbol alphabet .
it was further shown in @xcite that _ linear _ nc suffices to yield a solution when the symbol alphabet is algebraically modeled as a sufficiently large finite field , and every intermediate node transmits a linear combination of its received data symbols over the symbol field . since then
, there have been extensive studies on the field size requirement of a linear solution for a multicast network . from an algebraic approach , reference @xcite first showed that a multicast network has a linear solution over gf(@xmath0 ) as long as the prime power @xmath0 is larger than @xmath9 times the number @xmath11 of receivers .
the requirement of @xmath0 for the existence of a linear solution over gf(@xmath0 ) is further relaxed by @xcite to be larger than @xmath11 , and such a solution can be efficiently constructed by the algorithm proposed in @xcite . meanwhile , the efficient algorithm in @xcite is able to construct a linear solution over gf(@xmath0 ) when @xmath0 is no smaller than @xmath11 , and hence this condition on @xmath0 is slightly relaxed compared with the ones in @xcite and @xcite .
this efficient algorithm requires one to initially identify , for each receiver in @xmath5 , @xmath9 edge - disjoint paths starting from the source and ending at it .
denote by @xmath12 the maximum number of paths among the @xmath13 paths that contain a common edge .
the parameter @xmath12 is always no larger than @xmath11 . by a more elaborate argument ,
the algorithm in @xcite is refined in @xcite such that it can construct a linear solution over gf(@xmath0 ) as long as @xmath0 is no smaller than @xmath12 .
denote by @xmath6 the minimum field size for the existence of a linear solution over gf(@xmath6 ) , and by @xmath14 the maximum field size for the non - existence of a linear solution over gf(@xmath14 ) ( if the network is linearly solvable over every finite field , then @xmath14 is not well defined and we set it to 1 as a convention . ) the algorithm in @xcite implies that @xmath11 is an upper bound on @xmath14 .
for the special case that the source dimension @xmath9 is equal to 2 , the upper bound on @xmath14 is reduced to @xmath15 in @xcite . in both cases , the upper bounds additionally guarantee the existence of a linear solution over every gf(@xmath0 ) with @xmath0 larger than the bounds . on the other hand ,
references @xcite and @xcite independently constructed a class of multicast networks with @xmath6 lower bounded by @xmath7 . on any network in this class ,
a linear solution exists over every gf(@xmath0 ) with @xmath0 no smaller than @xmath6 , that is , @xmath16 .
these results indicate that in many cases , _ a multicast network that is linearly solvable over a given finite field _
gf_(@xmath0 ) is also linearly solvable over every larger finite field .
_ to the best of our knowledge , there is no explicit proof or disproof of the above claim for a general multicast network and for the case @xmath17 ; furthermore , all known multicast networks studied in the network coding literature satisfy @xmath16 .
although it has been shown in @xcite that the @xmath18-combination ( multicast ) network depicted in fig .
[ fig:4 - 2-combination_networks ] has a nonlinear solution over a ternary symbol alphabet but has neither a linear nor a nonlinear solution over any symbol alphabet of size 6 , it does not shed light on disproving the claim @xmath19 because this combination network has a linear solution over every gf(@xmath0 ) with @xmath20 ( see , for example , @xcite . )
moreover , in the case @xmath21 , as revealed in @xcite and @xcite , there exists a linear solution over gf(@xmath0 ) if and only if there exists a @xmath22-vertex coloring in an appropriately defined associated graph . since a @xmath0-vertex coloring in a graph always guarantees a @xmath23-vertex coloring with @xmath24 in the same graph , a multicast network with @xmath21 is linearly solvable over every gf(@xmath23 ) with @xmath25
this evidence seemed to add more support for the correctness of the claim @xmath19 for an arbitrary multicast network . in the present paper
, we shall show by constructive proofs that the claim @xmath19 is _ not _ always true .
in particular ,
* we show that there is a multicast network that is linearly solvable over gf(@xmath6 ) but not over gf(@xmath14 ) for each of : ( i ) @xmath26 , @xmath27 ; ( ii ) @xmath28 , @xmath27 ; ( iii ) @xmath29 , @xmath30 ; * we show that for any positive integer @xmath31 with less than 17,425,170 ( base-10 ) digits , there is a multicast network with @xmath26 whereas @xmath32 . *
we construct a new class of multicast networks with @xmath6 lower bounded by @xmath7 and with an additional property that not every network in it has @xmath14 smaller than @xmath6 .
the insight of our results is that not only the field size but also the orders of the proper multiplicative subgroups in the symbol field affect the linear solvability over the finite field . as we shall see ,
if a finite field does not contain a large enough proper multiplicative subgroup , or the complement of a large multiplicative subgroup in the finite field is not large enough , it is possible to construct a multicast network that is not linearly solvable over this finite field but linearly solvable over a smaller finite field . in comparison , the characteristic of the symbol field does not appear as important in designing examples here as in the ones in @xcite which show the non - existence of a linear solution for a general multi - source network , because in our exemplifying networks , both @xmath6 and @xmath33 can be of either odd and even characteristic .
the classical solvable ( non - multicast ) network that is not linearly solvable , proposed in @xcite , makes use of two types of subnetworks : one is linearly solvable only over a field with even characteristic whereas the other is linearly solvable only over a field with odd characteristic .
consequently , the proposed network as a whole is not linearly solvable over any field , even or odd .
our results bring about a new facet on the connection between the symbol field structures and network coding solvability problems .
the remainder of this paper is organized as follows .
section [ sec : fundamental_results ] presents the fundamental results that there exist multicast networks such that @xmath34 .
section [ sec : gap ] discusses how large the gap @xmath35 can be . to unify the justification of the linear solvability of different networks presented
, section [ sec : general_network ] constructs a general multicast network and obtains an equivalent condition of its linear representability over a field gf(@xmath0 ) .
section [ sec : lower_bound ] further constructs a new class of multicast networks with @xmath6 lower bounded by @xmath7 and with the additional property that @xmath34 for some networks in it .
section [ sec : summary ] concludes the paper and discusses some interesting problems along this new research thread in network coding theory .
a ( single - source ) multicast network is a finite directed acyclic multigraph with a unique source node @xmath8 and a set @xmath5 of receivers . on a multicast network , for every node @xmath36 , denote by @xmath37 and @xmath38 , respectively , the set of its incoming and outgoing edges .
every edge in a multicast network represents a noiseless transmission channel of unit capacity .
the source generates @xmath9 data symbols belonging to a fixed symbol field and will transmit them to all receivers in @xmath5 . without loss of generality ( wlog ) , assume that @xmath39 , which is referred to as the _ source dimension _ ( otherwise a new source can be created , connected to the old source with @xmath9 edges . ) for an arbitrary set @xmath40 of non - source nodes , denote by @xmath41 the maximum number of edge - disjoint paths starting from @xmath8 and ending at nodes in @xmath40 .
each receiver @xmath42 in @xmath5 has @xmath43 .
a linear network code ( lnc ) over gf(@xmath0 ) is an assignment of a _ coding coefficient _
@xmath44 to every pair @xmath45 of edges such that @xmath46 when @xmath45 is not an _
adjacent pair _
, that is , when there is not a node @xmath36 such that @xmath47 and @xmath48 .
the lnc uniquely determines a _ coding vector _
@xmath49 , which is an @xmath9-dim column vector , for each edge @xmath50 in the network such that : * @xmath51 forms the natural basis of @xmath52 . *
@xmath53 when @xmath48 for some @xmath54 .
wlog , we assume throughout this paper that * all lncs on a given multicast network have coding coefficients @xmath55 for all those adjacent pairs @xmath45 where @xmath31 is the unique incoming edge to some node .
a multicast network is said to be _ linearly solvable _ over gf(@xmath0 ) if there is an lnc over gf(@xmath0 ) such that for each receiver @xmath56 , the @xmath57 matrix @xmath58_{e\in in(t)}$ ] over gf(@xmath0 ) is full rank .
such an lnc is called a _ linear solution _ over gf(@xmath0 ) for the multicast network .
denote by @xmath6 the minimum field size for the existence of a linear solution over gf(@xmath6 ) , and by @xmath14 the maximum field size for the nonexistence of a linear solution over gf(@xmath14 ) .
specific to a finite field gf(@xmath0 ) , let gf(@xmath0)@xmath59 represent the multiplicative group of nonzero elements in gf(@xmath0 ) .
@xmath60 when @xmath61 , the efficient algorithm in @xcite can be adopted to construct a linear solution over gf(@xmath0 ) . in the case
@xmath21 , it has been shown in @xcite that every linear solution over gf(@xmath0 ) can induce a @xmath22-vertex coloring in an appropriately associated graph of the multicast network and vice versa .
since every @xmath22-vertex coloring in a graph can also be regarded as a @xmath62-vertex coloring for @xmath24 , it in turn induces a linear solution over gf(@xmath23 ) when @xmath23 is a prime power .
thus , every linear solution over a finite field can induce a linear solution over a larger finite field .
all these are tempting facts for one to conjecture that when @xmath63 and @xmath17 , a linear solution over a finite field might also imply the existence of a linear solution over a larger field .
the central theme of the present paper is to _ refute _ this conjecture in several aspects .
[ theorem:1 ] a multicast network with @xmath64 that is linearly solvable over a finite field is _ not _ necessarily linearly solvable over all larger finite fields .
when @xmath65 , this theorem is a direct consequence of the lemma below .
assume @xmath66 .
expand the network @xmath67 depicted in fig .
[ fig : rank_3_networks](a ) to a new multicast network @xmath68 as follows .
create @xmath69 new nodes each of which has an incoming edge emanating from the source and an outgoing edge entering every receiver . in @xmath68 ,
every receiver has the maximum flow from the source equal to @xmath9 .
consider an lnc over a given gf(@xmath0 ) . by the topology of @xmath68 ,
the coding vector for every edge that is originally in @xmath67 is a linear combination of the coding vectors for those edges in @xmath70 that are also in @xmath67 .
since the network @xmath67 is linearly solvable over gf(@xmath0 ) with every @xmath71 except for @xmath72 by the lemma below , so is the network @xmath68 .
[ lemma : rank_3_network_a ] consider the multicast network depicted in fig .
[ fig : rank_3_networks](a ) .
denote by @xmath73 , @xmath74 @xmath75 @xmath76 @xmath75 @xmath77 the unique incoming edge to node @xmath78 . for every set @xmath40 of 3 grey nodes with @xmath79
, there is a receiver connected from it .
this network is linearly solvable over every finite field gf(@xmath0 ) with @xmath80 except for @xmath81 . a detailed technical proof for lemma [ lemma : rank_3_network_a ] is given in appendix [ appendix_proof_rank_3_network_a ] . in the network in fig .
[ fig : rank_3_networks](a ) , observe that every receiver is connected with three nodes @xmath82 where either @xmath83 , @xmath84 , @xmath85 are distinct ( such as @xmath86 ) or two among @xmath83 , @xmath84 , @xmath85 are same ( such as @xmath87 ) .
one can then check an lnc over gf(7 ) with @xmath88 = \left [ \begin{array}{ccccccccc } 1 & 1 & 1 & 0 & 0 & 0 & 1 & 1 & 1 \\ 1 & 2 & 4 & 1 & 1 & 1 & 0 & 0 & 0 \\ 0 & 0 & 0 & 1 & 2 & 4 & 1 & 2 & 4 \\ \end{array } \right],\ ] ] where @xmath89 represents the unique incoming edge to node @xmath90 , qualifies to be a linear solution for the network .
note that all nonzero entries in this matrix belong to a proper subgroup of order 3 in the multiplicative group @xmath91 .
the key reason that results in the network not linearly solvable over gf(8 ) is that there is not a proper subgroup of order at least 3 in the multiplicative group gf(8)@xmath59 . in general , the network in fig .
[ fig : rank_3_networks](a ) is linearly solvable over a finite field gf(@xmath0 ) if and only if there exist @xmath92 subject to @xmath93 this is a direct consequence of theorem [ thm : general_network_solvability ] in section [ sec : general_network ] when a more general multicast network depicted in fig .
[ fig : general_network ] is to be introduced which subsumes the network in fig .
[ fig : rank_3_networks](a ) as a special instance . on a solvable multi - source network , it is well - known that linear network coding ( with linearity in terms of a more general algebraic structure ) is not sufficient to yield a solution .
the classical example in @xcite to show this consists of two subnetworks , one is only linearly solvable over a field with even characteristic , whereas the other is only linearly solvable over a field with odd characteristic .
more generally , a procedure is introduced in @xcite to construct a ( matroidal ) multi - source network based on a matroid such that the constructed network is linearly solvable over gf(@xmath0 ) if and only if the matroid is representable over gf(@xmath0 ) .
this connection is powerful for designing a number of non - linearly solvable networks from a variety of interesting matroid structures ( see the appendix in @xcite for example , ) and the network in @xcite is a instance under this construction .
subsequently , the role of characteristics of finite fields on the linear solvability of a general multi - source network is further revealed in @xcite : for an arbitrary finite or co - finite set @xmath94 of prime numbers , a multi - source network can be constructed such that the network is linearly solvable over some field of characteristic @xmath95 for every prime number @xmath95 in @xmath94 , whereas it is not linearly solvable over any finite field whose characteristic is not in @xmath94 .
however , the examples presented in this paper can not be established by the procedures introduced in @xcite and @xcite , because they were not designed to construct ( single - source ) multicast networks . moreover , as a result of lemma [ lemma : rank_3_network_a ] and the next lemma , the role of the characteristic of a finite field in the examples designed in the present paper is not as important as in the example in @xcite . [
lemma : rank_3_network_b ] consider the multicast network depicted in fig .
[ fig : rank_3_networks](b ) .
there are in total 20 grey nodes . for every set @xmath40 of 3 grey nodes with @xmath79 ,
there is a receiver connected from it .
this network is linearly solvable over every finite field gf(@xmath0 ) with @xmath96 except for @xmath97 .
a detailed proof of this lemma can be found in appendix [ appendix_proof_rank_3_network_b ] .
similar to the network in fig .
[ fig : rank_3_networks](a ) , the network in fig .
[ fig : rank_3_networks](b ) can also be regarded as a special instance of the general network to be introduced in section [ sec : general_network ] .
then , theorem [ thm : general_network_solvability ] to be developed in section [ sec : general_network ] implies that the network in fig .
[ fig : rank_3_networks](b ) is linearly solvable over a finite field gf(@xmath0 ) if and only if there is an assignment of @xmath98 , @xmath99 from gf(@xmath0)@xmath59 subject to @xmath100 and @xmath101 for example , when @xmath102 , we can set @xmath103 for all @xmath104 , where @xmath105 is a primitive element in gf(16 ) , and set @xmath106 to be the 10 elements in @xmath107 . such an assignment satisfies the above condition and hence the network is linearly solvable over gf(16 ) .
the insight for the network in fig .
[ fig : rank_3_networks](b ) to be linearly solvable over gf(16 ) rather than gf(17 ) is that there is such a subgroup in the multiplicative group gf(16)@xmath59 but not in gf(17)@xmath59 that ( i ) the subgroup has order no smaller than 5 ; ( ii ) the complement of the subgroup in the multiplicative group contains at least 10 elements . [
corollary : characteristic ] given a multicast network with @xmath34 , both @xmath6 and @xmath14 can be of either even or odd characteristic .
this can be seen from lemma [ lemma : rank_3_network_a ] together with lemma [ lemma : rank_3_network_b ] .
to the best of our knowledge , all known multicast networks studied in the network coding literature have the property that @xmath16 .
the results in the previous section reveal that it is possible for @xmath34 . however , both examples which illustrate this fact have the special property @xmath108 . a natural question next is
how far away can @xmath14 be from @xmath6 .
the main result in this section is to show that for some multicast networks , the difference @xmath35 can be extremely large .
consider the _
swirl network _ with source dimension @xmath64 depicted in fig .
[ fig : swirl_network ] . for every set @xmath40 of @xmath9 grey nodes with @xmath109 ,
there is a non - depicted receiver connected from it .
corresponding to each node @xmath78 of in - degree 2 , where @xmath110 , let @xmath111 denote the two outgoing edges from it . by viewing the swirl network as a special instance of the general network to be introduced in the next section , the following necessary and sufficient condition for the swirl network to be linearly solvable over gf(@xmath0 ) can be readily obtained . a detailed proof is given in appendix [ appendix_proof_swirl_network ] .
[ lemma : swirl_network ] the swirl network is linearly solvable over a given gf(@xmath0 ) if and only if there exist @xmath112 subject to @xmath113 under which a linear solution over gf(@xmath0 ) can be subsequently given by prescribing the coding vectors for @xmath114 as @xmath115 \\ % & = \left [ % \begin{matrix } % 1 & 1 & 0 & 0 & \cdots & 0 & 0 & 1 & 1 \\ % 1 & \alpha_1 & 1 & 1 & \ddots & 0 & 0 & 0 & 0 \\ % 0 & 0 & 1 & \alpha_2 & \ddots & 0 & 0 & 0 & 0 \\ % \vdots & \vdots & \vdots & \vdots & \ddots & \vdots & \vdots & \vdots & \vdots \\ % 0 & 0 & 0 & 0 & \cdots & 1 & 1 & 0 & 0 \\ % 0 & 0 & 0 & 0 & \cdots & 1 & \alpha_{\omega-1 } & \delta_1 & \delta_2
\\ % \end{matrix } % \right ] , [ f_{e_{11}}~f_{e_{12}}~f_{e_{21}}~f_{e_{22 } } \cdots f_{e_{\omega1}}~f_{e_{\omega2 } } ] = \left [ \begin{matrix } 1 & 1 & 0 & 0 & \cdots & 0 & 0 & 1 & 1 \\ 1 & \alpha_1 & 1 & 1 & \ddots & 0 & 0 & 0 & 0 \\ 0 & 0 & 1 & \alpha_2 & \ddots & 0 & 0 & 0 & 0 \\ \vdots & \vdots & \vdots & \vdots & \ddots & \vdots & \vdots & \vdots & \vdots \\ 0 & 0 & 0 & 0 & \cdots & 1 & 1 & 0 & 0 \\ 0 & 0 & 0 & 0 & \cdots & 1 & \alpha_{\omega-1 } & \delta_1 & \delta_2 \\ \end{matrix } \right].\ ] ] consider the swirl network with @xmath116 , and an lnc over gf(5 ) with the coding vectors for @xmath117 prescribed by @xmath118 = \left [ \begin{array}{cccccccccccc } 1 & 1 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 1 & 1 \\ 1 & 4 & 1 & 1 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 1 & 4 & 1 & 1 & 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 1 & 4 & 1 & 1 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 & 1 & 4 & 1 & 1 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 1 & 4 & 2 & 3 \\ \end{array } \right].\ ] ] apparently condition ( [ eqn : swirl_1 ] ) is satisfied by this lnc . since @xmath119 is a subgroup of gf(5)@xmath59 , it is closed under multiplication by elements in it .
moreover , @xmath120 , and so condition ( [ eqn : swirl_2 ] ) is also satisfied .
thus , this lnc over gf(5 ) qualifies as a linear solution . on the other hand , since gf(8)@xmath59 does not have a proper subgroup other than @xmath121 , it is not difficult to check that for arbitrary @xmath122 , the set @xmath123 contains at least 6 elements .
thus , there are not enough distinct elements in gf(8)@xmath59 to assign for @xmath124 and @xmath125 subject to conditions ( [ eqn : swirl_1 ] ) and ( [ eqn : swirl_2 ] ) .
hence , the network is not linearly solvable over gf(8 ) .
the argument in the example above can simply be generalized to derive the linear solvability of the swirl network with general source dimension @xmath9 over a given gf(@xmath0 ) .
first , assume that the order of the multiplicative group gf(@xmath0)@xmath59 is not prime .
this implies @xmath126 .
then there is a subgroup @xmath127 in gf(@xmath0)@xmath59 with @xmath128 , and gf(@xmath0)@xmath129 contains no less than 2 elements .
let @xmath130 be an element in @xmath127 not equal to 1 and @xmath131 be two distinct elements in gf(@xmath0)@xmath129 .
we can assign @xmath132 for all @xmath133 and @xmath134 .
such an assignment obeys conditions ( [ eqn : swirl_1 ] ) and ( [ eqn : swirl_2 ] ) .
hence , the network is linearly solvable over gf(@xmath0 ) .
next , assume that the order of gf(@xmath0)@xmath59 is a prime , in which case @xmath0 can always be written in the form of @xmath135 for some prime @xmath95 .
we shall show that for any @xmath136 , the set @xmath137 contains at least @xmath138 elements .
this is obviously true when @xmath139 .
assume that it is true for @xmath140 and consider the case @xmath141 .
note that @xmath142 . if @xmath143 , then so is @xmath144 .
assume that @xmath145 .
since @xmath146 and @xmath147 is a prime , there is at least one element in @xmath148 that is not in @xmath149 .
thus , @xmath150 . as a consequence , we conclude that for any assignment of @xmath151 , where @xmath105 is a primitive element in gf(@xmath135 ) and @xmath152 , the set @xmath153 contains at least @xmath9 elements . in order to further successfully assign @xmath125 subject to ( [ eqn : swirl_1 ] ) and ( [ eqn : swirl_2 ] ) , the size of gf(@xmath135 ) has to be at least @xmath154 . when @xmath155 , we can set @xmath156 for all @xmath157 and @xmath158 , @xmath159 , where @xmath105 is a primitive element in gf(@xmath135 ) , so that conditions ( [ eqn : swirl_1 ] ) and ( [ eqn : swirl_2 ] ) obey . we have proved the following theorem .
[ theorem : swirl ] the swirl network with @xmath64 has @xmath26 and @xmath160 .
moreover , it is not linearly solvable over all those @xmath161 when @xmath162 is prime .
recall that a prime integer in the form @xmath162 is known to be a _
mersenne prime_. then @xmath14 for the swirl network with @xmath64 is equal to one plus the largest mersenne prime no larger than @xmath163 .
while whether there are infinitely many mersenne primes is still an open problem , the @xmath164 known mersenne prime ( which is also the largest known prime ) found under the gimps project ( see @xcite ) has a length of 17,425,170 digits under base 10 .
thus , when @xmath9 is sufficiently large , the difference @xmath35 for the swirl network is so enormous as to having tens of millions of digits . [
corollary : swirl ] if there are infinitely many mersenne primes , then there are infinitely many multicast networks with @xmath165 , and moreover , the difference @xmath35 can tend to infinity .
another interesting consequence of theorem [ theorem : swirl ] is that the swirl network is linearly solvable over gf(@xmath166 ) , but not over a larger gf(@xmath135 ) when @xmath167 is a prime .
this unveils that the linear solvability over a finite field does not even guarantee the linear solvability over all larger finite fields of the _ same _ characteristic .
moreover , it was pointed out in @xcite , without demonstrating an explicit example or rigorous proof , that a multicast network linearly solvable over gf(@xmath2 ) and over gf(@xmath3 ) may not be linearly solvable over gf(@xmath168 ) .
the swirl network is the first class of explicit networks that asserts this conjecture to be correct .
[ corollary : swirl_characteristic ] there exists a multicast network such that it is linearly solvable over gf(@xmath2 ) and over gf(@xmath3 ) but _ not _ over gf(@xmath4 ) .
assume @xmath169 .
since @xmath170 is a prime whereas @xmath171 are not , the swirl network is linearly solvable over gf(@xmath166 ) and over gf(@xmath172 ) but not over gf(@xmath173 ) . * remark . *
the inspiration of designing the class of swirl networks stems from a matroid structure referred to as the _ free swirl _ ( see @xcite or chapter 14 in @xcite ) . the matrix in ( [ eqn : swirl_matrix ] )
can be regarded as a ( linear ) representation of the free swirl of rank @xmath9 when conditions ( [ eqn : swirl_1 ] ) and ( [ eqn : swirl_2 ] ) hold . in a multicast network ,
the coding vectors of an lnc naturally induces a representable matroid on the edge set , and in the strongest sense , thus the induced representable matroid is referred to as a _ network matroid _
@xcite , in which case the linear independence of coding vectors coincides with the independence structure of edge - disjoint paths @xcite .
similar to a network matroid , a _ gammoid _ in matroid theory characterizes the independence of node - disjoint paths in a directed graph .
it turns out that if the independence structure of a matroid can be reflected on a multicast network , then the matroid should be not only representable , but ( isomorphic to ) a gammoid as well .
being representable and a gammoid is the essential structure of the free swirl that guides the construction of multicast networks we look for . in comparison , those representable matroids , such as fano and non - fano matroids that have well - known applications to construct a multi - source network @xcite , are not isomorphic to any gammoid , and hence they do not have corresponding multicast networks reflecting their independence structures .
one may observe that the networks presented in the previous two sections with @xmath34 share a similar layered structure . in order to unify the derivation of their respective linear solvability
, we next construct a general multicast network with parameters @xmath174 which subsumes the networks depicted in fig .
[ fig : rank_3_networks ] and fig .
[ fig : swirl_network ] as special cases , and then derive an equivalent condition for the linear solvability of this general network over a field gf(@xmath0 ) .
[ alg : general_construction ] given a 3-tuple of positive integers ( @xmath175 ) as input parameters , the procedure below constructs a multicast network consisting of nodes on five layers , which are labeled 1 - 5 from upstream to downstream , and all edges are between adjacent layers . .
create a source @xmath8 , which forms the unique node at layer 1 . .
create @xmath9 layer-2 nodes , each of which is connected with @xmath8 by an edge .
sequentially label these nodes as @xmath176 . .
create @xmath9 layer-3 nodes , labeled as @xmath177 .
each node @xmath178 , @xmath179 has 2 incoming edges , one leading from @xmath180 and the other from @xmath181 , where @xmath182 will represent @xmath183 . . for each layer-3
node @xmath178 , @xmath157 , create @xmath184 downstream layer-4 nodes @xmath185 , each of which has the unique incoming edge @xmath186 leading from @xmath187 .
corresponding to the layer-3 node @xmath188 , create @xmath189 downstream layer-4 nodes @xmath190 , each of which has the unique incoming edge @xmath191 leading from @xmath188 . .
for every set @xmath40 of @xmath9 layer-4 nodes with @xmath109 , create a layer-5 node connected from every node in @xmath40 by an edge . set all layer-5 nodes to be receivers . @xmath60 fig .
[ fig : general_network ] depicts the general multicast network constructed by algorithm [ alg : general_construction ] with parameters @xmath174 .
it can be easily checked that the networks in fig .
[ fig : rank_3_networks](a ) and fig .
[ fig : rank_3_networks](b ) , as well as the swirl network are all instances of the general network with respective parameters @xmath192 , @xmath193 , @xmath194 .
[ thm : general_network_solvability ] consider the general network constructed by algorithm 1 with parameters @xmath174 .
the network is linearly solvable over a finite field gf(@xmath0 ) if and only if there exist @xmath195 , @xmath196 , @xmath197 , @xmath198 such that @xmath199 under which a linear solution over gf(@xmath0 ) can be given by prescribing the coding vectors @xmath200 \nonumber \\ \label{eqn : general_network_matrix_reduced } & = \left [ \begin{array}{ccccccccccccc } 1 & \cdots & 1 & 0 & \cdots & 0 & & 0 & \cdots & 0 & 1 & \cdots & 1 \\ \alpha_{11 } & \cdots & \alpha_{1d_1 } & 1 & \cdots & 1 & & \vdots & \ddots & \vdots & 0 & \cdots & 0\\ 0 & \cdots & 0 & \alpha_{21 } & \cdots & \alpha_{2d_1 } & & 0 & \cdots & 0 & 0 & \cdots & 0\\ 0 & \cdots & 0 & 0 & \cdots & 0 & \ddots & 0 & \cdots & 0 & \vdots & \ddots & \vdots \\ \vdots & \ddots & \vdots & \vdots & \ddots & \vdots & & 1 & \cdots & 1 & 0 & \cdots & 0 \\ 0 & \cdots & 0 & 0 & \cdots & 0 & & \alpha_{(\omega-1)1 } & \cdots & \alpha_{(\omega-1)d_1 } & \delta_{1 } & \cdots & \delta_{d_2 } \\ \end{array } \right].\end{aligned}\ ] ] on the network constructed by algorithm [ alg : general_construction ] with parameters @xmath174 , consider an lnc with all coding coefficients being indeterminates . assume the coding vector for the unique incoming edge to node
@xmath201 is equal to the @xmath202 @xmath9-dim unit vector .
then , the juxtaposition of the coding vectors for edges @xmath186 , @xmath179 can be represented as the @xmath203 matrix @xmath200 \nonumber \\ \label{eqn : matrix_coding_vectors } & = \left [ \begin{array}{ccccccccccccc } \alpha_{11,1 } & \cdots & \alpha_{1d_1,1 } & 0 & \cdots & 0 & & 0 & \cdots & 0 & \delta_{1,1 } & \cdots & \alpha_{d_2,1 } \\
\alpha_{11,2 } & \cdots & \alpha_{1d_1,2 } & \alpha_{21,1 } & \cdots & \alpha_{2d_1,1 } & & \vdots & \ddots & \vdots & 0 & \cdots & 0\\ 0 & \cdots & 0 & \alpha_{21,2 } & \cdots & \alpha_{2d_1,2 } & & 0 & \cdots & 0 & 0 & \cdots & 0\\ 0 & \cdots & 0 & 0 & \cdots & 0 & \ddots & 0 & \cdots & 0 & \vdots & \ddots & \vdots \\
\vdots & \ddots & \vdots & \vdots & \ddots & \vdots & & \alpha_{(\omega-1)1,1 } & \cdots & \alpha_{(\omega-1)d_1,1 } & 0 & \cdots & 0 \\ 0 & \cdots & 0 & 0 & \cdots & 0 & & \alpha_{(\omega-1)1,2 } & \cdots & \alpha_{(\omega-1)d_1,2 } & \delta_{1,2 } & \cdots & \delta_{d_2,2 } \\ \end{array } \right]\end{aligned}\ ] ] where @xmath204 , @xmath205 respectively represent the coding coefficient indeterminates for adjacent pairs @xmath206 and @xmath207 , @xmath157 , @xmath197 , and @xmath208 , @xmath209 respectively represent the coding coefficient indeterminates for adjacent pairs @xmath210 and @xmath211 , @xmath212 .
denote by @xmath213 this juxtaposed matrix , and by @xmath214 the set of indeterminates @xmath215 , where @xmath196 , @xmath197 , @xmath198 , and @xmath216 . for every @xmath217 submatrix in @xmath213 ,
its determinant is a polynomial in indeterminates in @xmath214 .
then , for every set @xmath40 of @xmath9 layer-4 nodes , the maximum flow from @xmath8 to @xmath40 is equal to @xmath9 if and only if the juxtaposition of coding vectors for incoming edges to nodes in @xmath40 has a nonzero determinant polynomial .
consequently , there is a linear solution over gf(@xmath0 ) if and only if there is a _ matrix completion _ for matrix @xmath213 over gf(@xmath0 ) , that is , an assignment of values in gf(@xmath0 ) to indeterminates in @xmath214 such that for every @xmath217 full - rank submatrix in @xmath213 , the evaluation of its determinant polynomial is nonzero .
assume that the network is linearly solvable over gf(@xmath0 ) .
then , there is a matrix completion for @xmath213 over gf(@xmath0 ) .
consider such a matrix completion that the indeterminates in @xmath214 are assigned to values in gf(@xmath0 ) .
observe that the @xmath218 matrix @xmath219 = \left[\begin{smallmatrix } \alpha_{11 , 1 } & \alpha_{12,1 } & 0 & & 0
\\ \alpha_{11 , 2 } & \alpha_{12 , 2 } & \alpha_{21 , 1 } & & \cdots \\ 0 & 0 & \alpha_{21 , 2 } & \cdots & 0 \\ \cdots & \cdots & 0 & & \alpha_{(\omega-1)1 , 1 } \\ 0 & 0 & 0 & & \alpha_{(\omega-1)1 , 2 } \end{smallmatrix}\right]$ ] is full rank before matrix completion , and its determinant is equal to @xmath220 .
hence , @xmath221 and @xmath222 . by an analogous argument
to all those @xmath218 full - rank submatrices of @xmath213 which are in the form @xmath223 $ ] , where @xmath186 represents @xmath224 when @xmath225 , it can be deduced that * @xmath226 , @xmath227 ; * @xmath228 , @xmath229 ; * @xmath230 , @xmath231 .
thus , we can form another matrix @xmath232 by the multiplication of @xmath213 and a diagonal square matrix with diagonal entries equal to @xmath233 .
any @xmath9 columns in @xmath213 are linearly independent if and only if the @xmath9 columns in @xmath232 of same indices are linearly independent .
now , set @xmath234 , where @xmath157 , @xmath197 , and set @xmath235 , where @xmath198 .
then , @xmath232 is in the same form as expressed in ( [ eqn : general_network_matrix_reduced ] ) , and * @xmath236 , @xmath237 ; * @xmath238 , @xmath239 @xmath240 .
next , observe that the @xmath218 submatrix @xmath241 = \left[\begin{smallmatrix } \alpha_{11 , 1 } & 0 & & 0 & \delta_{1 , 1}\\ \alpha_{11 , 2 } & \alpha_{21 , 1 } & & \cdots & 0 \\ 0 & \alpha_{21 , 2 } & \cdots & 0 & \cdots \\ \cdots & 0 & & \alpha_{(\omega-1)1 , 1 } & 0 \\ 0 & 0 & & \alpha_{(\omega-1)1 , 2 } & \delta_{1 , 2 } \end{smallmatrix}\right]$ ] of @xmath213 is of full rank before matrix completion .
hence , the @xmath242 submatrix in @xmath232 with the same column indices @xmath243 $ ] is of full rank .
equivalently , its determinant @xmath244 is nonzero . by the same argument to the full - rank @xmath217 submatrices
@xmath245 $ ] , @xmath246 and @xmath247 , it can be deduced that @xmath248 thus , the values @xmath249 induced from the considered matrix completion for @xmath213 over gf(@xmath0 ) satisfies condition ( [ eqn : general_network_linear_solvability ] ) .
the proof of the necessity part is complete . to prove the sufficiency part ,
assume that there exist @xmath195 subject to ( [ eqn : general_network_linear_solvability ] ) .
then , it can be carefully checked that the assignment @xmath250 qualifies as a matrix completion for the matrix @xmath213 depicted in ( [ eqn : matrix_coding_vectors ] ) over gf(@xmath0 ) , and hence the matrix in ( [ eqn : general_network_matrix_reduced ] ) is the juxtaposition of coding vectors from a linear solution over gf(@xmath0 ) .
the network depicted in fig .
[ fig : general_network ] with parameters @xmath174 is linearly solvable over gf(@xmath0 ) if gf(@xmath0)@xmath59 has a proper subgroup @xmath127 such that @xmath251 and @xmath252 .
let @xmath127 be a subgroup in gf(@xmath0)@xmath59 such that @xmath251 and @xmath252 .
for each @xmath157 , assign @xmath253 to be arbitrary @xmath184 elements in @xmath127 , and @xmath254 to be arbitrary @xmath189 elements in @xmath255 . under such assignments , @xmath256
is always contained in @xmath127 and thus condition ( [ eqn : general_network_linear_solvability ] ) is satisfied and the general network is linearly solvable over gf(@xmath0 ) .
in the literature , it is well known that @xmath11 is an upper bound on @xmath6 for multicast networks . to the best of our knowledge ,
the highest lower bound on @xmath6 for multicast networks is of the order @xmath7 .
references @xcite and @xcite independently constructed a class of multicast networks whose @xmath6 is lower bounded by @xmath7 .
the two classes of networks are essentially the same , and they are based on the class of @xmath257-combination networks with @xmath21 ( the ( 4 , 2)-combination network is depicted in fig . [ fig:4 - 2-combination_networks ] . )
moreover , they share the same property that @xmath16 . inspired by the results in section [ sec : fundamental_results ]
, we shall establish a new class of multicast networks with @xmath6 also lower bounded by @xmath7 whereas _ not _ every network in it has @xmath14 smaller than @xmath6 .
let us first revisit the network depicted in fig .
[ fig : rank_3_networks](a ) .
as proved in lemma [ lemma : rank_3_network_a ] , it is linearly solvable over gf(@xmath6 ) where @xmath28 but not over gf(8 ) .
the network is assumed to have a receiver connected from every set @xmath40 of 3 grey nodes with @xmath79 . actually
, appropriate deletion of some receivers ( e.g. the ones respectively connected from @xmath258 and @xmath259 ) from the network will not affect the linearly solvability over any particular finite field .
stemming from this observation , we shall next construct a new class of multicast networks with @xmath6 lower bounded by @xmath7 by appropriately increasing grey nodes and redefining receivers in the network in fig .
[ fig : rank_3_networks](a ) . consider the network with @xmath65 depicted in fig .
[ fig : network_lowerbound ] .
it has a total of @xmath260 grey nodes , which are classified into three sets as @xmath261 , @xmath262 , and @xmath263 .
the set of receivers in this network is prescribed as follows : * type - i : corresponding to every ( unordered ) pair @xmath264 of grey nodes in @xmath265 , there is a receiver connected from @xmath266 , where @xmath267 when @xmath268 and @xmath269 otherwise ; * type - ii : corresponding to every ( unordered ) pair @xmath270 of grey nodes in @xmath271 , there is a type - ii receiver connected from @xmath272 , where @xmath267 when @xmath268 and @xmath269 otherwise ; * type - iii : there is a respective receiver connected from @xmath273 for every @xmath274 , from @xmath275 for every @xmath276 , and from @xmath277 .
there is also a respective receiver connected from @xmath278 and from @xmath279 for every @xmath280 ; * type - iv : there is a receiver connected from @xmath281 for every possible @xmath282 .
the total number @xmath11 of receivers in the network is then equal to @xmath283 .
denote by @xmath73 the incoming edge to node @xmath78 for all @xmath284 .
the network depicted in fig .
[ fig : network_lowerbound ] can be viewed as an instance of the general network constructed by algorithm 1 with parameters @xmath285 , with some particularly chosen receivers deleted . by noticing that such careful deletion of receivers will not affect the linear solvability of the network
, we can directly obtain the next lemma as a corollary of theorem [ thm : general_network_solvability ] .
[ lemma : lower_bound ] the network depicted in fig . [ fig : network_lowerbound ] is linearly solvable over a given gf(@xmath0 ) if and only if there is an lnc over gf(@xmath0 ) with coding vectors for @xmath286 prescribed by @xmath287 \\ & = \left [ \begin{matrix } 1 & \cdots & 1 & 0 & \cdots & 0 & 1 & 1 & 1 \\ \alpha_1 & \cdots & \alpha_{m } & 1 & \cdots & 1 & 0 & 0 & 0 \\ 0 & \cdots & 0 & \beta_1 & \cdots & \beta_m & \delta_1 & \delta_2 & \delta_3 \end{matrix } \right ] , \end{aligned}\ ] ] where @xmath288 and @xmath289 in this lemma , prescribed by condition ( [ eqn : lower_bound_1 ] ) , the set @xmath290 contains at least @xmath291 elements .
hence , condition ( [ eqn : lower_bound_2 ] ) obeys only if @xmath292 .
consider the special case that @xmath293 and @xmath294 is equal to a mersenne prime . since there is a subgroup @xmath127 of order @xmath291 in the multiplicative group gf(@xmath294)@xmath59 , and @xmath295 , we are able to assign @xmath296 and @xmath297 subject to ( [ eqn : lower_bound_1 ] ) and ( [ eqn : lower_bound_2 ] ) by setting @xmath298 and @xmath299 to be arbitrary distinct elements in @xmath300
. therefore , the network is linearly solvable over gf(@xmath294 ) .
we next show that the network is not linearly solvable over gf(@xmath301 ) .
recall the cauchy - davenport theorem ( see @xcite for example ) which asserts that for any two nonempty subsets @xmath302 and @xmath303 in the additive group @xmath304 , where @xmath95 is a prime , the cardinality of @xmath305 contains at least @xmath306 elements . since the multiplicative group gf(@xmath301)@xmath59 is isomorphic to the additive group @xmath307 , where @xmath294 is prime
, there are at least @xmath308 elements in @xmath290 as a direct consequence of the cauchy - davenport theorem .
hence , there are only at most 2 elements in @xmath309 , so that condition ( [ eqn : lower_bound_2 ] ) can not be satisfied .
the above results are summarized as follows .
[ theorem : lower_bound ] for the network depicted in fig .
[ fig : network_lowerbound ] , @xmath310 , and hence @xmath6 is lower bounded by @xmath7 . when @xmath293 and @xmath294 is equal to a mersenne prime , @xmath311 whereas the network is _ not _ linearly solvable over gf(@xmath301 )
on an acyclic multicast network , if there is a linear solution over gf(@xmath0 ) , could we claim that there is a linear solution over every gf(@xmath23 ) with @xmath312 ?
it would be tempting to answer it positively because by the result in @xcite , the claim is correct when @xmath0 is no smaller than the number of receivers and moreover , as a consequence of the result in @xcite , the positive answer is affirmed for the special case that the source dimension of the network is equal to 2 . in the present paper , however , we show the negative answer for general cases by constructing several classes of multicast networks with different emphasis .
these networks are the first ones discovered in the network coding literature with the property that @xmath14 , the maximum field size for the nonexistence of a linear solution over gf(@xmath14 ) , is larger than @xmath6 , the minimum field size for the existence of a linear solution over gf(@xmath6 ) .
the insight of various exemplifying networks established in the present paper is that not only the field size of gf(@xmath0 ) , but also the _ order of the proper multiplicative subgroup _ of gf(@xmath0)@xmath59 affects the networks linear solvability over gf(@xmath0 ) .
the results in this paper bring about a new thread on the fundamental study of linear network coding , specific to the case of multicast networks , as discussed below : * besides the field size and multiplicative subgroup orders , it is not clear whether there are some other undiscovered inherent structures in a finite field that affects the linear solvability of a multicast network
. it deserves further investigation . * all multicast networks presented in this paper that are linearly solvable over a field gf(@xmath0 ) but not over a larger field gf(@xmath23 ) share a common property that for some values @xmath313 , there is a proper subgroup @xmath127 in gf(@xmath0 ) subject to @xmath251 and @xmath252 , but there does not exist a proper subgroup @xmath314 in gf(@xmath23 ) subject to @xmath315 and @xmath316 .
thus , when @xmath317 does not contain any proper subgroup other than @xmath121 , _
i.e. _ , @xmath318 is a prime , we can simply set @xmath319 and @xmath320 , such that for any values @xmath313 subject to @xmath321 and @xmath322 , the conditions @xmath323 and @xmath324 hold as well .
this leads us to an ambitious conjecture that if a multicast network is linearly solvable over gf(@xmath0 ) where @xmath325 is a prime , then it is linearly solvable over all finite fields of sizes larger than @xmath0 . * as special cases of the previous conjecture ,
it is of particular interest to study : i ) whether a multicast network linearly solvable over gf(2 ) ( or over gf(3 ) ) is linearly solvable over all larger finite fields ; ii ) whether a multicast network linearly solvable over gf(2 ) and over gf(3 ) is linearly solvable over all finite fields .
these two cases are related to other interesting conjectures in the network coding literature : it was conjectured and partially proved in @xcite that every planar multicast network is linearly solvable over gf(3 ) , and certain special planar multicast networks ( including relay - coface networks and terminal - coface networks ) are always linearly solvable over both gf(2 ) and gf(3 ) .
the conjectures presented here will further suggest that these special multicast networks may be linearly solvable over all finite fields .
stemming from the above discussions , we end this paper by proposing a number of open problems , all of which , except for the first , we conjecture to have positive answers : * for a multicast network , what is the smallest prime power @xmath0 larger than @xmath14 ( such that the network is linearly solvable over all gf(@xmath23 ) with @xmath312 ) ? * can the gap @xmath326 tend to infinity ?
* are there infinitely many prime power pairs @xmath327 with @xmath328 such that each @xmath327 corresponds to @xmath329 of some multicast network ? * if a multicast network is linearly solvable over such a gf(@xmath0 ) that gf(@xmath0)@xmath59 does not contain any proper multiplicative subgroup other than @xmath121 , is it linearly solvable over all larger finite fields than gf(@xmath0 ) ? * if a multicast network is linearly solvable over gf(2 ) ( or over gf(3 ) ) , is it linearly solvable over all larger finite fields ? *
if a multicast network is linearly solvable over both gf(2 ) and gf(3 ) , is it linearly solvable over all finite fields ?
observe that the network in fig .
[ fig : rank_3_networks](a ) can be constructed by algorithm 1 introduced in section [ sec : general_network ] with parameters @xmath192 .
then theorem [ thm : general_network_solvability ] implies that the network in fig .
[ fig : rank_3_networks](a ) is linearly solvable over gf(@xmath0 ) if and only if there exist @xmath330 , @xmath331 such that @xmath332 under constraint ( [ eqn : gf_7_gf_8_1 ] ) , the cardinality of @xmath333 is no smaller than 3 and no larger than 9 .
thus , * when @xmath334 , there does not exist any assignment of @xmath335 from gf(@xmath0)@xmath59 satisfying ( [ eqn : gf_7_gf_8_2 ] ) , and hence the network is not linearly solvable over gf(@xmath0 ) ; * when @xmath336 , there are always ways to assign @xmath335 from gf(@xmath0)@xmath59 subject to ( [ eqn : gf_7_gf_8_1 ] ) and ( [ eqn : gf_7_gf_8_2 ] ) , and hence the network is linearly solvable over gf(@xmath0 ) .
it remains to consider the case @xmath0 = 7 , 8 , 9 , or 11 . when @xmath0 = 7 , 9 , or 11 , there is a proper subgroup @xmath127 of order at least 3 in the multiplicative group @xmath317 .
we can then assign arbitrary three distinct values in @xmath127 to @xmath337 , @xmath338 , @xmath339 and to @xmath340 , @xmath341 , @xmath342 . in this way ,
the cardinality of @xmath333 is upper bounded by @xmath343 and thus there are at least three values remained in @xmath344 , which @xmath345 , @xmath346 , @xmath347 can be assigned to . in the last case @xmath0
= 8 , since there is no proper subgroup in gf@xmath348 , the method depicted in the previous paragraph does not work any more and an exhaustive search will verify that there does not exist any assignment of @xmath335 from gf(@xmath0)@xmath59 satisfying ( [ eqn : gf_7_gf_8_1 ] ) and ( [ eqn : gf_7_gf_8_2 ] ) .
we can now affirm that the network depicted in fig .
[ fig : rank_3_networks](a ) is linearly solvable over every gf(@xmath0 ) with @xmath80 except for @xmath72 .
observer that the network in fig .
[ fig : rank_3_networks](b ) can be constructed by algorithm 1 introduced in section [ sec : general_network ] with parameters @xmath193 .
then theorem [ thm : general_network_solvability ] implies that the network in fig .
[ fig : rank_3_networks](b ) is linearly solvable over gf(@xmath0 ) if and only if there is an assignment of @xmath98 , @xmath99 from gf(@xmath0)@xmath59 subject to @xmath349 under condition ( [ eqn : gf_16_gf_17_1 ] ) , the cardinality of @xmath350 is lower bounded by 5 and upper bounded by 25 .
thus , * when @xmath351 , there does not exist any assignment of @xmath98 , @xmath99 from gf(@xmath0)@xmath59 satisfying ( [ eqn : gf_16_gf_17_1 ] ) , and hence the network is not linearly solvable over gf(@xmath0 ) ; * when @xmath352 , there are always ways to assign @xmath98 , @xmath99 subject to ( [ eqn : gf_16_gf_17_1 ] ) and ( [ eqn : gf_16_gf_17_2 ] ) , and hence the network is linearly solvable gf(@xmath0 ) . it remains to consider the case that @xmath353 .
note that when @xmath0 is not equal to 17 or 32 , there is also a proper subgroup @xmath127 in the multiplicative group gf(@xmath0)@xmath59 such that @xmath127 has order no smaller than 5 and the cardinality of gf(@xmath0)@xmath354 is at least 10 .
then , we can respectively assign any 5 distinct elements in @xmath127 to @xmath355 and to @xmath356 , and assign any 10 distinct elements in gf(@xmath0)@xmath357 to @xmath358 .
such an assignment obeys conditions ( [ eqn : gf_16_gf_17_1 ] ) and ( [ eqn : gf_16_gf_17_2 ] ) . when @xmath359 , denote by @xmath105 be a primitive element in gf(32)@xmath59 .
assign @xmath360 for all @xmath104 , and @xmath361 for all @xmath362 .
it is easy to check that such an assignment obeys conditions ( [ eqn : gf_16_gf_17_1 ] ) and ( [ eqn : gf_16_gf_17_2 ] ) . when @xmath363 , in order to make @xmath364 contains at least 10 elements , @xmath355 and @xmath356 should be such assigned that the cardinality of @xmath365 is no larger than 6 .
to minimize this cardinality , as many as @xmath366 and @xmath367 should be assigned to a same proper subgroup in gf(17)@xmath59 .
an exhaustive search will then verify that it is infeasible to assign @xmath355 and @xmath356 from @xmath368 so that @xmath365 contains no more than 6 elements .
we can now assert that the network depicted in fig . [
fig : rank_3_networks](b ) is linearly solvable over every gf(@xmath0 ) with @xmath96 except for @xmath363 .
observe that the swirl network can be constructed by algorithm 1 in section [ sec : general_network ] with parameters @xmath194 .
then according to theorem [ thm : general_network_solvability ] , the swirl network is linearly solvable over gf(@xmath0 ) if and only if there exist @xmath369 , @xmath370 subject to @xmath371 given arbitrary @xmath369 , @xmath370 , define @xmath372 and @xmath373 , @xmath374 .
it is straightforward to check that condition ( [ eqn : swirl_proof_1 ] ) holds if and only if @xmath375 moreover , because @xmath376 condition ( [ eqn : swirl_proof_2 ] ) holds if and only if @xmath377 we can now conclude that the swirl network is linearly solvable over gf(@xmath0 ) if and only if there exist @xmath112 satisfying conditions ( [ eqn : swirl_1 ] ) and ( [ eqn : swirl_2 ] ) .
the authors would like to thank sidharth jaggi for helping identify and motivate the problem studied in this work .
the authors would also appreciate the helpful suggestions by the associate editor as well as anonymous reviewers to help improve the presentation of the paper .
s. jaggi , p. sanders , p. a. chou , m. effros , s. egner , k. jain , and l. tolhuizen , `` polynomial time algorithms for multicast network code construction , '' _ ieee trans .
inf . theory _
1973 - 1982 , jun . 2005 .
t. xiahou , z. li , c. wu , `` information multicast in ( pseudo-)planar networks : efficient network coding over small finite fields , '' _ international symposium on network coding ( netcod ) _ , calgary , canada , jun . | in an acyclic multicast network , it is well known that a linear network coding solution over gf(@xmath0 ) exists when @xmath0 is sufficiently large . in particular , for each prime power @xmath0 no smaller than the number of receivers , a linear solution over gf(@xmath0 ) can be efficiently constructed . in this work
, we reveal that a linear solution over a given finite field does _ not _ necessarily imply the existence of a linear solution over all larger finite fields .
specifically , we prove by construction that : ( i ) for every source dimension no smaller than 3 , there is a multicast network linearly solvable over gf(7 ) but not over gf(8 ) , and another multicast network linearly solvable over gf(16 ) but not over gf(17 ) ; ( ii ) there is a multicast network linearly solvable over gf(5 ) but not over such gf(@xmath0 ) that @xmath1 is a mersenne prime plus 1 , which can be extremely large ; ( iii ) a multicast network linearly solvable over gf(@xmath2 ) and over gf(@xmath3 ) is _ not _ necessarily linearly solvable over gf(@xmath4 ) ; ( iv ) there exists a class of multicast networks with a set @xmath5 of receivers such that the minimum field size @xmath6 for a linear solution over gf(@xmath6 ) is lower bounded by @xmath7 , but not every larger field than gf(@xmath6 ) suffices to yield a linear solution .
the insight brought from this work is that not only the field size , but also the order of subgroups in the multiplicative group of a finite field affects the linear solvability of a multicast network .
linear network coding , multicast network , field size , lower bound , mersenne prime . |
quantum adiabatic processes are a powerful strategy to implement quantum state engineering , which aims at manipulating a quantum system to attain a target state at a designed time t. in the adiabatic scenario , the quantum system evolves under a sufficiently slowly - varying hamiltonian , which prevents changes in the populations of the energy eigenlevels .
in particular , if the system is prepared in an eigenstate @xmath0 of the hamiltonian @xmath1 at a time @xmath2 , it will evolve to the corresponding instantaneous eigenstate @xmath3 at later times .
this transitionless evolution is ensured by the adiabatic theorem , which is one of the oldest and most explored tools in quantum mechanics @xcite .
the huge amount of applications of the adiabatic behavior has motivated renewed interest in the adiabatic theorem , which has implied in its rigorous formulation @xcite as well as in new bounds for adiabaticity @xcite . in quantum information processing ,
the adiabatic theorem is the basis for the methodology of adiabatic quantum computation ( aqc ) @xcite , which has been originally proposed as an approach for the solution of hard combinatorial search problems .
more generally , aqc has been proved to be universal for quantum computing , being equivalent to the standard circuit model of quantum computation up to polynomial resource - overhead @xcite .
moreover , it is a physically appealing approach , with a number of experimental implementations in distinct architectures , e.g. , nuclear magnetic resonance @xcite , ion traps @xcite , and superconducting flux quantum bits ( qubits ) through the d - wave quantum annealer @xcite .
recently , the circuit model has been directly connected with aqc via hybrid approaches @xcite .
then , an adiabatic circuit can be designed based on the adiabatic realization of quantum gates , which allows for the translation of the quantum circuit to the aqc framework with no further resources required . in particular , it is possible to implement universal sets of quantum gates through controlled adiabatic evolutions ( cae ) @xcite . in turn , cae are used to perform one - qubit and two - qubit gates , allowing for universality through the set of one - qubit rotations joint with an entangling two - qubit gate @xcite . however , since these processes are ruled by the adiabatic approximation , it turns out that each gate of the adiabatic circuit will be implemented within some fixed probability ( for a finite evolution time ) .
moreover , the time for performing each individual gate will be bounded from below by the adiabatic time condition @xcite . for a recent analysis on adiabatic control of quantum gates and its corresponding non - adiabatic errors ,
see ref .
@xcite . in order to resolve the limitations of adiabaticity in the hybrid model ,
we propose here a general shortcut to cae through simple time - independent counter - diabatic assistant hamiltonians within the framework of the superadiabatic theory @xcite .
the physical resources spent by this strategy will be governed by the quantum circuit complexity , but no adiabatic constraint will be required in the individual implementation of the quantum gates . moreover ,
the gates will be deterministically implemented with probability one as long as decoherence effects can be avoided .
in particular , we discuss the realization of rotation gates and arbitrary n - qubit controlled gates , which can be used to design different sets of universal quantum gates .
this analog approach allows for fast implementation of individual gates , whose time consumption is only dictated by the quantum speed limit ( qsl ) ( for closed systems , see refs .
indeed , the time demanded for each gate will imply in an energy cost , which increases with the speed of the evolution . in this context , by analyzing the energy - time complementarity , we will show that the qsl provides an energy cost for superadiabatic evolutions that upper bounds the cost of adiabatic implementations .
let us begin by discussing the design of adiabatic quantum circuits as introduced by hen @xcite through the implementation of quantum gates via cae . in order to define quantum gates through cae
, we will introduce a discrete bipartite system @xmath4 associated with a hilbert space @xmath5 . the system @xmath4 is composed by a target subsystem @xmath6 and an auxiliary subsystem @xmath7 , whose individual hilbert spaces @xmath8 and @xmath9 have dimensions @xmath10 and @xmath11 , respectively .
the dynamics of @xmath4 will be governed by a hamiltonian in the form @xcite @xmath12 + g\left ( t\right ) \left [ \sum\nolimits_{k}p_{k}\otimes h_{k}^{\left ( f\right ) } \right ] , \label{sce.1.1}\]]where @xmath13 , @xmath14 , and @xmath15 denotes a complete set of orthogonal projectors over @xmath6 , so that they satisfy @xmath16 and @xmath17 .
alternatively , we can write eq .
( [ sce.1.1 ] ) as @xmath18with @xmath19 denoting a hamiltonian that acts on @xmath7 .
suppose now that we prepare the system @xmath4 in the initial state @xmath20 , where @xmath21 is an arbitrary state of @xmath6 and @xmath22 is the ( non - degenerate ) ground state of @xmath23 .
then @xmath24 is the ground state of the initial hamiltonian @xmath25 . by applying the adiabatic theorem @xcite ,
a sufficiently slowing - varying evolution of @xmath26 will drive the system ( up to a phase ) to the final state @xmath27 where @xmath28 is the ground state of @xmath29 @xcite .
we can perform a single - qubit unitary transformation through a general rotation of an angle @xmath30 around a direction @xmath31 on the bloch sphere . in this direction , we begin by preparing the system @xmath4 , taken here as two qubits , in the initial state @xmath32 , where @xmath33 are the computational states of the auxiliary system @xmath7 .
then , we let the system adiabatically evolve driven by the hamiltonian @xcite @xmath34where @xmath35 and @xmath36 are adiabatically - evolved hamiltonians , whose effect will be restricted to the respective subspaces of the projectors @xmath37 , where @xmath31 is a unitary vector on the bloch sphere associated with @xmath6 and @xmath38 , with @xmath39 denoting the set of pauli matrices .
the hamiltonians are taken as @xmath40 \right\}$ ] , where @xmath41 , @xmath42 sets the energy scale ( @xmath43 ) , @xmath44 is a constant parameter , and @xmath45 denotes a dimensionless ( parametrized ) time , with @xmath46 the total time of evolution .
note then that @xmath47 \right\}. \nonumber \\ \label{pw - h}\end{aligned}\ ] ] by writing the initial state of @xmath4 as @xmath48 , where @xmath49 is an arbitrary ( not necessarily known ) qubit state , the final state @xmath50 follows from eq .
( [ psi - final ] ) , i.e. @xmath51 .
note that @xmath52 is the ground state of @xmath53 , reading @xmath54 , with @xmath41 .
an equivalent form of writing @xmath55 is given by @xmath56 hence , we have a probabilistic implementation of the rotated target state @xmath57 for an arbitrary angle @xmath30 around an arbitrarily chosen axis @xmath31 , with probability @xmath58 .
in particular , this probability approximates to one by taking @xmath59 . in order to perform controlled rotations of a qubit by an angle @xmath30 around a direction @xmath31
, the starting point will be to take the subsystem @xmath6 as a two - qubit system and keeping @xmath7 as a single auxiliary qubit .
the hamiltonian is now chosen to be @xmath60which will govern the evolution of the initial composite state @xmath61 , with @xmath62 . from eq .
( [ psi - final ] ) , the final state of the subsystem @xmath6 in the limit @xmath63 is now the controlled rotated vector @xmath64 . by combining controlled rotations with the single - qubit unitaries discussed above ,
it is possible to design universal sets of quantum gates through an adiabatic implementation .
in this section we present the main results of this work .
we start by generalizing the adiabatic implementation of quantum gates proposed in ref .
@xcite for n - qubit controlled gates . even though n - qubit controlled gates can be decomposable into one and two - qubit gates ( see , e.g. refs .
@xcite ) , this implementation implies in an extended class of adiabatic universal gates , e.g. the set @xmath65toffoli
@xmath66 hadamard@xmath67 @xcite .
then , we will derive the main result of this work , which is a shortcut for general adiabatic circuits through constant counter - diabatic hamiltonians , which implies in the possibility of fast analog implementations of quantum circuits .
moreover , we will present an analysis of the quantum speed limit in the context of the energetic cost of the superadiabatic circuit . in order to implement @xmath68-controlled gates ,
we define the subsystem @xmath6 as an @xmath69-qubit system , with the first @xmath68 qubits used as the control register and the last qubit taken as the target register . for the auxiliary system @xmath7 ,
we keep it as a single qubit .
then , we take the initial state as @xmath70 , with the subsystem @xmath6 described by @xmath71where @xmath72 are complex amplitudes , @xmath73 , @xmath74 , and @xmath31 is an arbitrary axis in the bloch sphere . here
we have written the target qubit in the basis @xmath75 , leaving the remaining qubits of @xmath6 in the computational basis .
for simplicity , we will write the states in its decimal representation , i.e. @xmath76where @xmath77 . then
, we let the system evolve driven by the hamiltonian @xmath78 \otimes h_{0}\left ( s\right ) \nonumber \\ & & + \left\vert n-1,\hat{n}_{-}\right\rangle \left\langle n-1,\hat{n}_{-}\right\vert \otimes h_{\phi } \left ( s\right ) .
\label{adg.1.6}\end{aligned}\]]we note that the rotation of the target qubit is expected to be applied if the state of the control system is @xmath79 . then
, if the hamiltonian is sufficiently slowly - varying so that we can apply the adiabatic theorem , the system will achieve the final state@xmath80where @xmath81 is defined as @xmath82 ( @xmath83 ) .
an equivalent form of writing eq .
( [ adg.1.8 ] ) is @xmath84with @xmath85 thus , by performing a measurement over the auxiliary qubit , we find the rotated state @xmath86 with probability @xmath87 .
as in the previous case of a rotation controlled by one qubit , this probability can be enhanced to one in the limit @xmath88 .
indeed , this state implies in a rotation of the target qubit in @xmath6 conditioned by the state @xmath89 of the control register in @xmath6 .
an application of this scheme is the adiabatic implementation of the toffoli gate , which constitutes an unitary operation implementing an @xmath90 gate over the target qubit if all control qubits are in the state @xmath91 , with no effect if any qubit of the control register is in the state @xmath92 .
this can be easily achieved here by a rotation of an angle @xmath93 around of the direction x , therefore choosing @xmath94 and @xmath95 , with @xmath96 denoting the eigenstates of @xmath97 .
let us introduce now a shortcut to general cae through the superadiabatic approach .
this will allow for fast piecewise implementation of quantum gates , whose evolution time will not be constrained by the adiabatic theorem .
we begin by defining the evolution operator @xcite@xmath98 which leads an initial state @xmath99 into an evolved state @xmath100 given by @xmath101 where @xmath102 are the eigenvectors of the adiabatic hamiltonian .
note that this evolution mimics the adiabatic behavior .
the hamiltonian that guides the evolution of the system is the _ superadiabatic _
hamiltonian , which reads @xmath103where the additional term @xmath104 is the _ counter - diabatic _
hamiltonian @xmath105therefore , a superadiabatic implementation of a dynamical evolution involves the knowledge of the eigenstates of the adiabatic hamiltonian @xmath106 , which limits the direct application of the superadiabatic approach in quantum computation .
for instance , by adopting the original aqc approach @xcite , superadiabatic implementations seem prohibitive , since the whole set of eigenlevels of a many - body hamiltonian is required . in a similar context ,
counter - diabatic driving protocols based on realizable settings have been investigated for assisted evolutions in quantum critical phenomena @xcite . here ,
as we shall see , the superadiabatic implementation of universal quantum circuits in the hybrid approach can be promptly achieved , since we deal with the eigenspectrum of piecewise hamiltonians , which act over a few qubits .
it is then appealing to formulate a superadiabatic theory to cae and then to specify it to the implementation of universal sets of quantum gates .
let us begin by establishing the complete set of eigenstates of the hamiltonian @xmath106 provided by eq .
( [ sce.1.2 ] ) . to this end , consider the eigenvalue equation to each hamiltonian @xmath107 given by@xmath108with @xmath109 . by defining the projectors @xmath110 in eq .
( [ sce.1.2 ] ) as @xmath111 , with @xmath112 and @xmath113 , we can write the complete set of eigenstates of @xmath106 as @xmath114 such that @xmath115 . indeed , from eq .
( [ sce.1.2 ] ) , we have @xmath116 \left[\left\vert \lambda_{l}\right\rangle \otimes \left\vert \varepsilon_{l}^{i}\left ( t\right ) \right\rangle\right ] = e_{l}^{i}\left ( t\right ) \left\vert \gamma^{i}_{l}\left ( t\right)\right\rangle$ ] .
note that each projector @xmath117 is associated with a hamiltonian @xmath118 .
for instance , for the adiabatic implementation of @xmath68-controlled gates , we have defined the hamiltonian @xmath1 in eq .
( [ adg.1.6 ] ) by linking the set @xmath119 with @xmath120 and by linking the remaining projector @xmath121 with @xmath122 .
the next step is to obtain the counter - diabatic hamiltonian @xmath123 that implements the shortcut to the adiabatic evolution of @xmath106 . in this direction
, we use the eigenstates of @xmath106 as given by eq .
( [ sce1.5 ] ) .
then , we get @xmath124 , \label{sce1.5.a}\]]with @xmath125 . therefore @xmath126 \nonumber \\ & = & \sum\nolimits_{l}\left [ p_{l}\otimes h_{l}^{cd}\left ( t\right ) \right ] , \label{sce1.5.b}\end{aligned}\]]where @xmath127 and @xmath128 is the counter - diabatic hamiltonian to be associated with the piecewise adiabatic contribution @xmath129 acting over subsystem @xmath7 , which reads @xmath130 .
\label{sce1.6}\]]hence , from eq .
( [ sfa.1.1 ] ) , we can implement the shortcut dynamics through the superadiabatic hamiltonian @xmath131where @xmath132 is the piecewise superadiabatic hamiltonian .
note that the cost of performing superadiabatic evolutions requires the knowledge of the eigenvalues and eigenstates of @xmath129 .
for the implementation of general @xmath68-controlled gates , this is a hamiltonian acting over a single qubit , which is independent of the circuit complexity .
moreover , we can show that , for an arbitrary @xmath68-controlled quantum gate , the counter - diabatic hamiltonians @xmath133 ( @xmath41 ) associated with shortcuts to adiabatic evolutions driven by @xmath40 \right\}$ ] , with @xmath134 , are _ time - independent _ operators given by @xmath135 .
\label{sce1.7.ad}\]]eq .
( [ sce1.7.ad ] ) shows that the implementation of the shortcut can be achieved with a very simple assistant hamiltonian in the quantum dynamics .
its proof is provided in section _
methods_. it is expected that the shortcut via a counter - diabatic hamiltonian is faster than the evolution via an adiabatic hamiltonian , but how much faster can it be ? to answer this question , we shall take a lower bound to the time evolution in quantum dynamics as provided by the _ quantum speed limit _ ( qsl ) . we will consider a closed quantum system evolving between arbitrary pure states @xmath136 and @xmath137 . since the evolution is driven by a time - dependent superadiabatic hamiltonian @xmath138 , we will take the generalized margolus - levitin bound @xcite derived by deffner and lutz @xcite , which reads @xmath139where @xmath140 $ ] is the bures metric for pure states @xcite and@xmath141 for superadiabatic evolutions
, the initial state @xmath142 evolves to @xmath143 , where @xmath144 denotes the instantaneous ground state of the adiabatic hamiltonian @xmath26 . by using the parametrized time @xmath45
, we can show from eqs .
( [ qsl.1 ] ) and ( [ qsl.2 ] ) that the total time @xmath46 that mimics the adiabatic evolution within the superadiabatic approach can be reduced to an arbitrary small value .
more specifically , the addition of a counter - diabatic hamiltonian implies into the qsl bound @xmath145 with @xmath146 and @xmath147 , as shown in section _
methods_. therefore , the qsl bound reduces to @xmath148 with @xmath46 and @xmath149 defined by the superadiabatic hamiltonian @xmath138 .
this means that the superadiabatic implementation is compatible with an arbitrary reduction of the total time @xmath46 , which holds _ independently _ of the boundary states @xmath150 and @xmath151 .
naturally , a higher energetic cost is expected to be involved for a smaller evolution time @xmath46 .
in particular , saturation of eq .
( [ qsl.final2 ] ) is achieved for either @xmath152 or @xmath153 , with both cases implying in @xmath154 .
note that this limit is forbidden in the adiabatic regime for finite @xmath149 , since the energy gap is proportional to @xmath155 , which implies in an adiabatic time of the order @xmath156 , with @xmath157 @xcite .
hence , eq . ( [ qsl.final2 ] ) leads to a flexible running time in a superadiabatic implementation , only limited by the energy - time complementarity .
let us show now that time and energy are complementary resources in superadiabatic implementations of quantum evolutions .
we shall define the energetic cost associated with a superadiabatic hamiltonian through @xmath158with @xmath159 given by eq .
( [ sce1.7 ] ) and the norm provided by the hilbert - schmidt norm @xmath160 } $ ] . since @xmath159 is hermitian , we can write@xmath161 } dt \nonumber \\ & = & \frac{1}{\tau } \int_{0}^{\tau } \sqrt{\text{tr}\left
[ { h}^{2}\left ( t\right ) + { h}_{cd}^{2}\left ( t\right ) \right ] } dt . \label{cost1.2}\end{aligned}\ ] ] to derive eq . ( [ cost1.2 ] ) , we have used that @xmath162 .
this can be obtained by computing the trace in the eigenbasis of @xmath26 and noticing that the expectation value of @xmath163 taken in an eigenstate of @xmath26 vanishes , i.e. @xmath164 . in particular , let us define the energetic cost to the adiabatic hamiltonian as @xmath165 } dt .\]]then , it follows that the energetic cost @xmath166 in superadiabatic evolutions supersedes the energetic cost @xmath167 for a corresponding adiabatic physical process . in order to evaluate @xmath166
we adopt the basis of eigenstates of the adiabatic hamiltonian @xmath168 . by using eq .
( [ sce1.5 ] ) , this yields @xmath169 } dt , \label{cost1.4}\ ] ] where @xmath170 are the energies of the adiabatic hamiltonian @xmath171 and @xmath172 in order to analyze the energetic cost as provided by eq .
( [ cost1.4 ] ) for superadiabatic qubit rotation gates , we set @xmath173 and @xmath174 @xmath175 . moreover , by using eq .
( [ sce1.5 ] ) , we obtain @xmath176 , which leads to @xmath177 [ see eqs .
( [ cqa.2.5a ] ) and ( [ cqa.2.5b ] ) in section _ methods _ ] . hence @xmath178 as a function of @xmath179 for different values of @xmath44 . ]
we illustrate the behavior of @xmath180 in fig .
[ graph1 ] , where it is apparent that the energetic cost increases inversely proportional to the total time of evolution .
in particular , note also that , for a fixed energetic cost , the optimal choice @xmath181 requires a longer evolution .
this is because of the fact that , in this case , the final state associated with the auxiliary qubit is orthogonal to its initial state , so it is farther in the bloch sphere . in the more general case of controlled gates , the analysis is similar as in the case of single - qubit gates .
however , we must take into account the number of projectors composing the set @xmath182 . more specifically , the sum over @xmath183 in eq .
( [ cost1.4 ] ) shall run over @xmath91 to @xmath184 , which is the number of projectors over the subsystem @xmath6 .
thus we can show that energetic cost @xmath185 to implement controlled gates is @xmath186 .
let us explicitly design here the superadiabatic implementation of controlled evolutions for piecewise hamiltonians @xmath187 as provided by eqs .
( [ pw - h ] ) . to this end , consider the eigenvalue equation@xmath188with @xmath189 , where@xmath190 from eq .
( [ sce1.5 ] ) , it follows that the eigenstates for the adiabatic hamiltonian @xmath191 governing the composite system @xmath192 are given by the sets @xmath193 and @xmath194 associated with the set of eigenvalues @xmath195 and @xmath196 , respectively . by evaluating the eigenvalues of @xmath197 and @xmath198
, we obtain that their spectra are equal , being provided by @xmath199 .
thus , @xmath191 exhibits doubly degenerate levels , with @xmath200 and @xmath201 associated with levels @xmath202 and @xmath203 , respectively . by using now eqs .
( [ cqa.2.5a ] ) and ( [ cqa.2.5b ] ) , we obtain @xmath204 , for any @xmath205 and @xmath189 .
then , from eq .
( [ sce1.5.b ] ) , we obtain that the counter - diabatic hamiltonian is @xmath206 , which leads to the time - independent counter - diabatic hamiltonian given by eq .
( [ sce1.7.ad ] ) .
the extension to the case of @xmath68-controlled gates can be achieved as follows . from eq .
( [ sce1.5 ] ) , the eigenstates of @xmath191 read@xmath207 where @xmath208 , @xmath209 and @xmath210 . by computing the eigenvalues of @xmath191
, we obtain that the spectrum of @xmath191 is @xmath211-degenerate , with @xmath212 and @xmath213 associated with the levels @xmath202 and @xmath203 , respectively . by using these results into eq .
( [ sce1.5.b ] ) , we obtain that the counter - diabatic piecewise hamiltonian @xmath214 is given by eq .
( [ sce1.7.ad ] ) .
hence , the implementation any @xmath68-controlled gate is achieved through a time - independent counter - diabatic hamiltonian .
let us apply here the qsl bound to superadiabatic evolutions . by using the fact than the @xmath215 evolves in the ground state @xmath216 of @xmath217 and that @xmath218 is given by eq .
( [ sfa.1.1 ] ) , we have@xmath219 where @xmath220 is the instantaneous ground state energy of @xmath26 .
now we use eq .
( [ sfa.1.2 ] ) and the inequality @xmath221 , which yields @xmath222by using the parametrized time @xmath223 , we obtain @xmath224where the parameters @xmath225 ( @xmath226 ) are given by @xmath227 @xmath228 . since
the ground state energy for the adiabatic hamiltonian @xmath229 in the case of @xmath68-controlled gates is @xmath230 [ see eqs .
( [ pw - h ] ) and ( [ adg.1.6 ] ) ] , we write @xmath231 , with @xmath232 . moreover , we define @xmath233 . then @xmath234let us now analise the term @xmath235 . first , note that @xmath236 ( see proof in ref .
@xcite ) , which yields @xmath237where we have used the inequality @xmath238 . from the definition of the bures
metric , we have @xmath239 .
hence , @xmath240 , which implies into eq .
( [ qsl.final2 ] ) .
we have proposed a scheme for implementing universal sets of quantum gates within the superadiabatic approach . in particular , we have shown that this can be achieved by applying a _ time - independent _ counter - diabatic hamiltonian in the auxiliary qubit to induce fast controlled evolutions .
remarkably , this hamiltonian is universal , holding both for performing single - qubit and @xmath68-controlled qubit gates .
therefore , a shortcut to the adiabatic implementation of quantum gates can be achieved through a rather simple mechanism . in particular
, different sets of universal quantum gates can be designed by using essentially the same counter - diabatic hamiltonian .
moreover , we have shown that the flexibility of the evolution time in a superadiabatic dynamics can be directly traced back from the qsl bound . in this context , the running time is only constrained by the energetic cost of the superadiabatic implementation , within a time - energy complementarity relationship .
implications of the superadiabatic approach under decoherence and a fault - tolerance analysis of superadiabatic circuits are further challenges of immediate interest . in a quantum open - systems scenario ,
there is a compromise between the time required by adiabaticity and the decoherence time of the quantum device .
therefore , the superadiabatic implementation may provide a direction to obtain an optimal running time for the quantum algorithm while keeping an inherent protection against decoherence . in turn
, a basis for such development may be provided by the generalization of the superadiabatic theory for the context of open systems @xcite . concerning error - protection
, it may also be fruitful the comparison of our approach with non - adiabatic holonomic quantum computation , where non - adiabatic geometric phases are used to perform universal quantum gates ( see , e.g. recent proposals in refs .
@xcite ) . moreover
, the behavior of correlations such as entanglement may also be an additional relevant resource for superadiabaticity applied to quantum computation .
these investigations as well as experimental proposals for superadiabatic circuits are left for future research .
we are grateful to itay hen and adolfo del campo for useful discussions .
m. s. s. thanks daniel lidar for his hospitality at the university of southern california .
we acknowledge financial support from the brazilian agencies cnpq , capes , and faperj . this work has been performed as part of the brazilian national institute of science and technology for quantum information ( inct - iq ) .
richerme , p. , senko , c. , smith , j. , lee , a. , korenblit , s. & monroe , c. experimental performance of a quantum simulator : optimizing adiabatic evolution and identifying many - body ground states .
rev . a _ * 88 * , 012334 ( 2013 ) .
torrontegui , e. , ibanez , s. , martinez - garaot , m. , modugno , m. , del campo , a. , guery - odelin , d. , ruschhaupt , a. , chen , x. & muga , j. g. shortcuts to adiabaticity _ adv .
atom . mol .
phys . _ * 62 * , 117 ( 2013 ) . | adiabatic state engineering is a powerful technique in quantum information and quantum control .
however , its performance is limited by the adiabatic theorem of quantum mechanics . in this scenario ,
shortcuts to adiabaticity , such as provided by the superadiabatic theory , constitute a valuable tool to speed up the adiabatic quantum behavior . here , we propose a superadiabatic route to implement universal quantum computation .
our method is based on the realization of piecewise controlled superadiabatic evolutions .
remarkably , they can be obtained by simple time - independent counter - diabatic hamiltonians . in particular
, we discuss the implementation of fast rotation gates and arbitrary n - qubit controlled gates , which can be used to design different sets of universal quantum gates .
concerning the energy cost of the superadiabatic implementation , we show that it is dictated by the quantum speed limit , providing an upper bound for the corresponding adiabatic counterparts . |
low - mass star formation typically involves the collapse of a dense core inside a molecular cloud to form a condensed central object and an extended disk .
the protostar gains much of its mass by accretion of disk material , which can only proceed if there is some mechanism for removing angular momentum from the accreting material .
protostars that are actively accreting disk material frequently exhibit associated bipolar outflows in the form of collimated jets oriented perpendicular to the disk , stretching out over distances much larger than the radial scale of the disk ( e.g. , * ? ? ?
this association between accretion disks and large - scale , collimated jets is also seen in other systems , including x - ray binaries and active galactic nuclei . the strongest candidate mechanism for launching jets in these systems
is described by the magnetocentrifugal disk wind model ( * ? ? ?
* hereafter bp82 ) , which relies on the acceleration of a fraction of disk material out along a large - scale , ordered magnetic field threading the disk at a sufficiently small inclination angle
. this magnetic field could be the interstellar field in the parent molecular cloud that is dragged in during core collapse , or it could be generated by dynamo activity in the disk or the star .
a natural consequence of the disk wind model is that the wind efficiently removes angular momentum from the disk material , which could explain the strong inferred link between accretion and outflow phenomena in protostars ( e.g. , * ? ? ?
several previous semianalytic studies have matched solutions of the equations of non - ideal mhd that describe a weakly ionized , magnetized accretion disk , and solutions of the equations of cold , ideal mhd for a polytropic fluid that describe a bp82-type magnetocentrifugal wind that removes all the disk s excess angular momentum .
these models focused on a radially localized portion of the disk ( e.g. , @xcite , hereafter wk93 ; @xcite , hereafter skw07 ; @xcite , hereafter skw11 ) , or on a global disk model ( e.g. , @xcite ; @xcite , hereafter l96 ; @xcite , hereafter f97 ) based on the same assumption of self - similarity in spherical radius that underlies the bp82 wind model .
another approach is presented in @xcite , where the thinness of the disk is used to justify solutions for the disk structure that are separable in the radial and vertical coordinates ; these solutions are matched to an isothermal wind .
a key parameter in determining whether a disk can launch a magnetocentrifugal wind is the field line inclination angle at the disk surface @xmath0 , where @xmath1 and @xmath2 are the radial and vertical field components at the disk surface ( subscript ` s ' ) .
this inclination angle is determined by the flux distribution along the disk surface and can only be calculated in a self - consistent fashion in a global treatment ( * ? ? ?
* hereafter ol01 ) .
a second key parameter describing the interaction of the disk and wind is @xmath3 , the azimuthal field strength at the disk surface , which sets the torque exerted by the wind on the disk ( bp82 ) . as with the surface field inclination angle ,
the value of @xmath3 is determined by processes acting outside the disk ; specifically , the value of @xmath3 is determined by the constraint that the wind solution pass smoothly through the alfvn point .
accretion flows in a weakly ionized disk will drag in magnetic flux , opposing the natural tendency of the magnetic flux to diffuse outwards .
studies of accretion disks that shed angular momentum due to mhd turbulence developed from the magnetorotational instability ( mri , @xcite ) suggest that the magnetic diffusivity and the effective viscosity of such disks are comparable @xcite , agreeing with previous work by @xcite which suggested that the magnetic diffusivity of mri - active accretion disks may be too large to allow efficient inward dragging of magnetic flux .
this problem can be overcome in an mri - active disk threaded by a large - scale field that removes angular momentum from surface layers in which the mri is suppressed @xcite .
more generally , this problem disappears if the angular momentum transport mechanism is decoupled from the mechanism that provides the magnetic diffusivity so that the diffusivity is much smaller than the effective viscosity of the disk .
this occurs naturally in a disk from which angular momentum is removed through a magnetocentrifugal wind , allowing magnetic flux to be dragged in and distributed over a wind - driving disk .
as noted by ol01 , the evolution of the magnetic flux through the disk determines whether magnetic configurations capable of launching outflows can be maintained over the accretion timescale .
previous studies of wind - driving disks have not treated the effects of magnetic flux evolution in a self - consistent manner : in a local model , ol01 calculated the rate of flux migration given @xmath1 and @xmath3 , which were free parameters .
the local models of @xcite expressed @xmath3 and the surface field inclination angle in terms of unconstrained parameters that describe the structure of the disk and the processes operating within it .
wk93 also employed a local model in which they determined @xmath3 self - consistently using the alfvn point constraint , but left the surface field inclination angle unconstrained , describing the magnetic flux migration speed using the free parameter @xmath4 .
the global models of l96 and f97 also solved for @xmath3 using the alfvn point constraint but left the surface field inclination angle and magnetic flux migration unconstrained , essentially setting @xmath5 . for an overview of recent work on semianalytic models and complementary numerical simulations of protostellar jets ,
see the article by @xcite .
i present here a new , semianalytic , radially self - similar , global model that matches disk solutions with bp82-type wind solutions , adopting an approach similar to that of l96 but including for the first time a fully self - consistent calculation of the effects of magnetic flux migration and the magnetic field structure above the disk : @xmath4 , @xmath1 , @xmath6 and @xmath0 are no longer free parameters but are determined by physical constraints imposed at the disk surface and at critical points of the outflow . in addition , the equations describing the disk are formulated in terms of the conductivity tensor , allowing for fully general non - ideal disk models .
the plan of the paper is as follows . in section [ form ]
i review the equations describing a ( nearly ) steady - state , axisymmetric , weakly ionized disk threaded by a magnetic field , and i show how the system reduces to a set of ordinary differential equations under the radial self - similarity assumption . in section [ paramconstr
] i discuss constraints from the self - consistent treatment of the field - line inclination at the top of the disk and from matching disk and wind solutions . in section [ res ]
i present a representative matched disk / wind solution , and in section [ concs ] i summarize the study and outline plans to improve on the current model .
i consider a weakly ionized , geometrically thin , vertically isothermal , axisymmetric disk in near - keplerian rotation around a protostar of mass @xmath7 that is threaded by an ordered , large - scale magnetic field .
i examine the general , non - ideal mhd case ( section [ casegen ] ) , expressing the matter field interaction using the conductivity tensor formalism , and seek solutions that are steady on timescales shorter than the accretion time ( section [ goveqgen ] ) .
i follow l96 and f97 in making the simplifying assumption of radial self - similarity to produce a global disk / wind model ( section [ selfgen ] ) , and i derive the midplane values of various physical quantities ( section [ midgen ] ) . in section [ casead ]
i specialize to the case of the ambipolar conductivity regime .
i once again derive the relevant self - similar disk equations ( section [ selfad ] ) and the midplane values for the variables of integration ( section [ midad ] ) .
the interiors of protostellar disks have a low ionization fraction , necessitating the use of non - ideal mhd to describe the behavior of the disk material in the presence of a large - scale magnetic field that is assumed to thread the disk .
the non - ideal matter - field interaction can be described using a multi - fluid approach or by introducing the conductivity tensor ( e.g. , * ? ? ?
* ) , which encapsulates the response of the dominant neutral fluid to the presence of the magnetic field .
the effects of the magnetic field are communicated to the neutrals by the development of charge
neutral drift velocities due to the lorentz forces on charged species , and the consequent systematic charge
neutral drag forces . under the conductivity tensor formalism , this is expressed in the generalized version of ohm s law , @xmath8 where @xmath9@xmath10 is the electric field in a frame comoving with the dominant neutral fluid , @xmath9@xmath11 and @xmath9@xmath12 are the components of @xmath9@xmath10 parallel and perpendicular to the magnetic field , respectively , and @xmath13is the conductivity tensor . the components of @xmath13are the ohm conductivity @xmath14 , the hall conductivity @xmath15 , and the pedersen conductivity @xmath16 . for a given charged species ( denoted by subscript
` @xmath17 ' ) , the product of the gyrofrequency @xmath18 and the neutral - charge momentum exchange time @xmath19 gives a measure of the coupling of the charged species to the magnetic field in a sea of neutrals . in these expressions , @xmath20 , @xmath21 , and @xmath22 are the signed charge , mass , and charge
neutral collisional drag coefficient of species @xmath17 , and @xmath23 is the neutral mass density .
the signed value of the ratio of the gyrofrequency and neutral collision frequency is the hall parameter , @xmath24 the definition of the hall parameter uses the signed value of the magnetic field strength , @xmath25 , where @xmath26 is the @xmath27 component of the magnetic field in cylindrical coordinates @xmath28 , so that the hall conductivity retains its dependence on the magnetic field polarity ( see equation ( [ eq : sigmahall ] ) ) .
large values @xmath29 imply that charged species @xmath17 is tightly coupled to the magnetic field , whereas small values @xmath30 imply that species @xmath17 is weakly coupled to the field .
the conductivity tensor components can be expressed in terms of the hall parameters as @xmath31 @xmath32 and @xmath33 note that since @xmath34 is the signed value of the magnetic field strength , the hall conductivity can be positive or negative .
the factor of @xmath34 in @xmath35 cancels the pre - factor of @xmath36 in the expressions for the ohm and pedersen conductivities , which are positive regardless of the magnetic field orientation .
the coupling of the neutral matter to the magnetic field is described by the elsasser number @xmath37 , given by @xmath38 where @xmath39 is the alfvn speed , @xmath40 is the keplerian angular frequency , @xmath41 , and @xmath42 ( e.g. , * ? ? ? * ) . as with the hall parameter ,
large values @xmath43 imply that the neutral matter is strongly coupled to the magnetic field , whereas small values @xmath44 imply that the neutral matter is only weakly coupled to the magnetic field . launching a magnetocentrifugal wind typically requires strong neutral field coupling ( @xmath45 ) .
the governing equations for the disk are the continuity of mass , @xmath46 and the conservation of momentum , @xmath47 in these equations , @xmath48is the neutral gas velocity and @xmath49 is its pressure , where @xmath50 is the isothermal speed of sound .
the gravitational potential of the central object is given by @xmath51 , where @xmath52 is the spherical radius and @xmath53 is the keplerian rotation speed .
within the disk , the ionization fraction is very low , which implies that the equations of momentum conservation for the charged species are dominated by the charge
neutral drag force and the lorentz force ( cf .
equations 2.152.17 in wk93 ) .
we can thus replace the charge neutral drag force with the lorentz force ; this substitution has already been made in the conservation of momentum for the neutral disk material ( equation ( [ eq : moment ] ) ) .
the current density satisfies ampre s law ( neglecting the displacement current ) @xmath54 and ohm s law ( equation ( [ eq : ohm ] ) ) , with the comoving electric field @xmath9@xmath10 related to the inertial - frame electric field
@xmath9by @xmath55 the magnetic field satisfies the solenoidal condition @xmath56 and the electric field @xmath9satisfies the induction equation , @xmath57 or , in terms of the comoving electric field , @xmath58 in a strictly steady state , @xmath59 identically .
however , the magnetic flux threading the disk interior to a fixed radius @xmath60 can change on the accretion timescale due to advection of field lines by the accretion flow and diffusion of field lines through the non - ideal disk material .
i therefore consider disk solutions that are steady on timescales longer than the dynamic time @xmath61 but relax the steady - state assumption on timescales comparable to the accretion time @xmath62 . expanding equation ( [ eq : induct3 ] ) yields @xmath63 @xmath64 and @xmath65 where the solenoidal condition ( equation ( [ eq : solen ] ) ) was used to obtain equation ( [ eq : inductphi ] ) .
now , @xmath66 , @xmath67 through much of the disk and @xmath68 near the top of the disk ( see figure 2 ) , @xmath69 ( where @xmath70 is the tidal scale height ) , and @xmath71 .
these relations imply that the terms @xmath72 , @xmath73 , @xmath74 , @xmath75 , and @xmath76 in equation ( [ eq : inductr ] ) describe processes operating on the dynamic timescale .
similarly , the terms @xmath77 , @xmath78 , @xmath79 , @xmath80 , @xmath81 , @xmath82 , and @xmath83 in equation ( [ eq : inductphi ] ) are associated with processes operating on the dynamic timescale . however
, every term in equation ( [ eq : inductz ] ) describes a process that operates on the longer accretion timescale .
i therefore ignore @xmath84 and @xmath85 in the @xmath60 and @xmath86 components of the induction equation , since those terms represent changes in the magnetic field on the accretion timescale and are negligible compared with the terms representing changes in the magnetic field on the dynamic timescale , but i retain @xmath87 in the @xmath27 component of the induction equation ( cf . * ? ? ?
the components of equation ( [ eq : induct ] ) thus become @xmath88 @xmath89 and @xmath90 the @xmath60 and @xmath27 components of the induction equation together imply that @xmath91 is constant with height at a given @xmath60 but need not vanish identically .
the disk equations reduce from partial to ordinary differential equations under the assumption of radial self - similarity , which implies that all quantities are described by power laws in the spherical radius @xmath92 for a fixed inclination angle .
i normalize all quantities by midplane values on a reference magnetic field line ( subscript ` * ' ) , using a scheme similar to that found in @xcite and l96 .
magnetic field lines can be labeled by the magnetic flux function @xmath93 , which satisfies @xmath94 , where @xmath95 is the poloidal magnetic field .
note that the notation used here corresponds to that of @xcite and l96 but differs from that of @xcite , where the flux function is labeled @xmath96 and @xmath93 is used to denote the poloidal flux ; the two quantities are related by @xmath97 . under the self - similarity assumption ,
the midplane values of all physical quantities scale as power laws in @xmath93 , and @xmath98 the magnetic flux distribution parameter @xmath99 is denoted by @xmath100 in l96 .
the cold - wind model of bp82 has @xmath101 . in a disk that drives a magnetocentrifugal wind , @xmath102 , and
if the outflow rate in the wind is much smaller than the accretion rate , as is the case in protostellar disks , then @xmath103 ( l96 ) . in terms of the self - similarity dimensionless variables @xmath104 and @xmath105 ( denoted by @xmath106 in l96 ) , various physical quantities are normalized as follows : @xmath107 the variable @xmath108 raised to various powers converts midplane values on the reference field line ( subscript ` * ' ) , which is labeled by flux function value @xmath109 , to midplane values on a generic field line ( subscript ` 0 ' ) with flux function value @xmath110 .
the combination @xmath111 , for example , is equal to @xmath112 , the midplane cylindrical radius of the field line labeled by flux function value @xmath113 .
the exponents for the power laws in @xmath108 in equation ( [ eq : nondim ] ) come from dimensional analysis of the governing equations .
the variable @xmath114 expresses position along a field line .
the new coordinates @xmath115 are expressed in terms of the spatial coordinates @xmath116 as @xmath117 and @xmath118 .
this gives @xmath119 in these relations , @xmath120 . in terms of the new coordinates , the spatial derivatives of a generic quantity @xmath121
take the following forms @xcite : @xmath122 and @xmath123 where @xmath124 .
the partial differential equations describing the disk can now be recast as ordinary differential equations and algebraic relations in the normalized quantities .
the equation of mass continuity becomes @xmath125 and the conservation of momentum ( equation ( [ eq : moment ] ) ) becomes @xmath126 in the @xmath60 direction , @xmath127 in the @xmath86 direction , and @xmath128 in the @xmath27 direction , where @xmath129 and @xmath130 the quantity @xmath131 is denoted by @xmath13 in l96 . by definition , @xmath132 .
the relation @xmath133 therefore gives @xmath134 .
from the definition of the flux function , @xmath135 , so we have @xmath136 at a point @xmath137 on a field line , the unit tangent vector along the field line is @xmath138 this relation , combined with equation ( [ eq : xprime ] ) , gives @xmath139 equation ( [ eq : brbz ] ) , which expresses @xmath140 in terms of @xmath141 , @xmath142 , and @xmath114 , automatically satisfies the solenoidal condition ( equation ( [ eq : solen ] ) ) .
ampre s law ( equation ( [ eq : amp ] ) ) becomes @xmath143 in the @xmath60 direction , @xmath144 \label{eq : ampphi3}\ ] ] in the @xmath86 direction , and @xmath145 in the @xmath27 direction .
using the algebraic relation for @xmath140 , the @xmath86 component simplifies further to @xmath146}{1+s^{2 } } \ , .
\label{eq : ampphi2}\ ] ] the poloidal components of ampre s law ( equations ( [ eq : ampr2 ] ) and ( [ eq : ampz2 ] ) ) together imply @xmath147 ohm s law ( equation ( [ eq : ohm ] ) ) becomes @xmath148 where the comoving electric field is given by @xmath149 the three components of equation ( [ eq : ohm2 ] ) are @xmath150 @xmath151 @xmath152 where @xmath153 , and the magnetic field - parallel and -perpendicular components of the comoving electric field satisfy @xmath154 @xmath155 and @xmath156 the poloidal components of the induction equation ( equations ( [ eq : inductr2 ] ) and ( [ eq : inductz2 ] ) ) yield the relation @xmath157 , which implies @xmath158 the @xmath86 component of the induction equation ( equation ( [ eq : inductphi2 ] ) ) yields @xmath159 \ , .
\label{eq : induct2a}\ ] ] combining equations ( [ eq : mass2 ] ) , ( [ eq : mom2r ] ) , and ( [ eq : mom2z ] ) yields @xmath160 \left[\left(2-\frac{2}{\zeta}\right)v_{r}\left(v_{z}-v_{r}s \right)\right . \nonumber \\ & \left.+\left(\frac{2}{\zeta}-3\right)\frac{x^{\prime}}{x}\left(v_{z}-v_{r}s \right)^{2 } \right .
\nonumber \\ & \left .
+ \left(4-\frac{2}{\zeta}\right)\theta\left(s+\frac{x^{\prime}}{x}+\frac{x^{\prime}s ^{2}}{x}\right)+v_{\phi}^{2}s\right .
\nonumber \\ & \left .
+ \frac{\nu x}{\tilde{\rho}}\left(j_{\phi}\left[b_{r}+b_{z}s\right]-b_{\phi}\left[j_{r}+j_{z}s\right]\right)\right]\ , .
\label{eq : rhoprime}\end{aligned}\ ] ] this relation shows that the point where @xmath161 is a singular point of the system of equations , since the denominator of @xmath162 vanishes there . the quantity @xmath163 is the component of the poloidal velocity perpendicular to the local surface of self - similarity ( the surface of constant @xmath114 ) , normalized by the keplerian velocity .
the denominator of @xmath162 vanishes when @xmath164 .
this occurs when the unnormalized velocity component @xmath165 becomes equal to the local sound speed ; this defines the sonic point .
the alfvn point is similarly a critical point of the cold ideal wind equations , and occurs where the same poloidal velocity component @xmath165 reaches the local alfvn speed . combining equation ( [ eq : rhoprime ] ) with equations ( [ eq : mom2r ] ) and
( [ eq : mom2z ] ) gives expressions for @xmath166 and @xmath167 .
equation ( [ eq : xprime ] ) gives @xmath168 , equation ( [ eq : mom2phi ] ) gives @xmath169 , and equations ( [ eq : ampphi2 ] ) and ( [ eq : ampr2 ] ) give @xmath170 and @xmath171 , respectively .
the final ordinary differential equation is equation ( [ eq : induct2a ] ) .
solving these 8 ordinary differential equations requires knowledge of the conductivity profile @xmath172 , @xmath173 , and @xmath174 , which can be obtained using a known disk ionoization profile or some simplifying assumptions . in order to solve for @xmath175 in equation ( [ eq : induct2a ] ) ,
it is necessary to express @xmath176 as a function of the other variables and then differentiate the resulting relation .
this expression is derived in appendix [ appez ] .
all physical quantities are assumed to have either odd or even symmetry in @xmath27 about the midplane .
the magnetic field is vertical at the midplane , so the @xmath114 derivative coincides with a @xmath27 derivative at that location . by definition , @xmath177 and @xmath178 . since the magnetic field is vertical at the midplane , @xmath179 and @xmath180 , and since the vertical component of the flow velocity must be continuous across the midplane , @xmath181 .
the midplane radial velocity and azimuthal electric field are given by @xmath182 and @xmath183 in dimensionless form , these relations are @xmath184 and @xmath185 , which combines with equation ( [ eq : induct2b ] ) to give @xmath186 the quantities @xmath187 and @xmath4 are parameters of the model whose values are set by conditions that must be satisfied at special points in the flow ( see section [ paramconstr ] ) .
the definitions of @xmath187 and @xmath4 given here are the same as in wk93 ; a factor of @xmath188 appears in the expressions for @xmath189 and @xmath190 because velocities here are normalized by @xmath191 rather than by @xmath50 as in wk93 . at the midplane ,
equations ( [ eq : mom2r ] ) and ( [ eq : mom2phi ] ) become @xmath192 and @xmath193 respectively . since the magnetic field is vertical at the midplane , @xmath194 , @xmath195 , @xmath196 , and @xmath197 .
the radial and azimuthal midplane current densities are therefore @xmath198 and @xmath199 ( see equations ( [ eq : ohm2r ] ) and ( [ eq : ohm2phi ] ) )
. substituting these relations into equations ( [ eq : mom2r0 ] ) and ( [ eq : mom2phi0 ] ) gives @xmath200 + \tilde{\sigma}_{\rm p,0}\left[\epsilon-\epsilon_{\rm b}\right]\theta^{\frac{1}{2}}\right ) = 0 \ , , \label{eq : mom2r02}\end{aligned}\ ] ] and @xmath201\theta^{\frac{1}{2 } } + \tilde{\sigma}_{\rm p,0}\left[e_{r,0 } + v_{\phi,0}\right]\right ) = 0\ , .
\label{eq : mom2phi02}\end{aligned}\ ] ] combining these two equations in the unknowns @xmath202 and @xmath203 gives @xmath204 = 0 \ , , \label{eq : mom2r03}\end{aligned}\ ] ] which can be solved for @xmath202 .
either equation ( [ eq : mom2r02 ] ) or equation ( [ eq : mom2phi02 ] ) can then be used to calculate @xmath203 .
ccccc @xmath142&&1&&0 + @xmath205&&1&&0 + @xmath206&&@xmath207&&0 + @xmath208&&@xmath202&&0 + @xmath209&&0&&@xmath210 + @xmath141&&0&&@xmath211 + @xmath212&&0&&@xmath213 + @xmath140&&1&&0 + @xmath214&&@xmath215&&0 + @xmath216&&@xmath217&&0 + @xmath218&&@xmath219&&0 + @xmath220&&@xmath221$]&&0 + [ tab : bcs ] the values of several physical quantities and their @xmath114 derivatives at the midplane as functions of the model parameters and @xmath202 are summarized in table [ tab : bcs ] .
note that the expression for @xmath222 is calculated using the mass continuity equation ( [ eq : mass2 ] ) , rather than equation ( [ eq : mom2z ] ) , which is identically 0 at the midplane .
also , from equation ( [ eq : ampz2 ] ) we have @xmath223 .
the magnetic field is vertical at the midplane , so @xmath224 . from ohm s law ( [ eq : ohm2z ] ) we have @xmath225 ; since @xmath226 is non - zero , @xmath223 implies @xmath227 . the midplane is a singular point of the differential equations describing the disk , and special care
is needed in initiating the numerical integration from that location .
a common choice for dealing with this issue is to approximate the values of the dependent variables a small distance above the midplane using taylor series expansions , a task that is made much easier by the assumption that all physical quantities have either even or odd symmetry about the midplane , which implies that the taylor series for all quantities contain only even power terms or only odd power terms for instance ,
the fact that @xmath228 immediately implies that @xmath229 . looking at table [ tab : bcs ] , it is clear that second - order taylor series expansions for @xmath209 , @xmath141 and @xmath212 only have one term apiece , corresponding to the first derivatives of those quantities at the midplane , which can be calculated using the expressions in table [ tab : bcs ] and equation ( [ eq : mom2r03 ] ) for @xmath202 .
the second - order taylor series expansions for @xmath142 , @xmath205 , @xmath206 , @xmath208 , and @xmath214 include terms involving @xmath230 , @xmath231 , @xmath232 , @xmath233 , and @xmath234 ; the procedure for calculating these quantities is described in appendix [ taylorgen ] .
the ambipolar conductivity regime occurs when all charged species @xmath17 have @xmath235 . substituting into equations ( [ eq : sigmaohm])([eq : sigmapedersen ] )
gives @xmath236 in the ambipolar diffusion limit .
i assume that the charged species consist of ions ( subscript ` i ' ) and electrons ( subscript ` e ' ) .
charge neutrality and a comparison of the rate coefficients for electron neutral and ion neutral momentum exchange given by @xcite imply that the heavier ions dominate the charge
neutral drag force . combining this with the balancing of the charge
neutral drag and lorentz forces yields @xmath237 where @xmath48@xmath238 is the ion velocity and @xmath239 is the ion mass density . for a disk in the ambipolar diffusion limit where the only charged species are ions and electrons ,
the elsasser number is @xmath240 , the ratio of the dynamic timescale to the neutral ion momentum exchange time @xcite .
this ratio is denoted by @xmath241 in l96 .
the neutral field coupling parameter @xmath242 in this case is related to the pedersen conductivity by @xmath243 ( skw07 ) .
the tight coupling of the ions to the magnetic field in the ambipolar diffusion limit implies that the comoving electric field vanishes in the frame of the ions .
the electric field in the disk is therefore @xmath244 equations ( [ eq : induct ] ) and ( [ eq : efield ] ) yield the induction equation @xmath245 from equation ( [ eq : efield ] ) and the definition of @xmath4 , @xmath246 , or @xmath247 where all quantities [ except the midplane quantity @xmath248 are evaluated at a point @xmath249 on the field line of interest .
the @xmath86 component of the induction equation ( [ eq : induct ] ) is @xmath250 or @xmath251 the ordinary differential equations obtained under the radial self - similarity assumption are identical to equations ( 12)(14 ) and ( 16)(21 ) in l96 .
the equations used here differ from those in l96 only in the treatment of the induction equation ; i allow for a nonzero @xmath4 ( or , equivalently , a nonzero @xmath91 ) , which changes the usual flux - freezing condition for the ions . as a result ,
equation ( 10 ) in l96 is replaced by equations ( [ eq : virz ] ) and ( [ eq : inductphi3 ] ) .
equations ( 22)(24 ) in l96 ( the self - similar , nondimensionalized versions of equation ( 10 ) in l96 ) are replaced by the self - similar , nondimensionalized versions of equation ( [ eq : virz ] ) , @xmath252 and equation ( [ eq : inductphi3 ] ) , @xmath253 where the normalization scheme from section [ selfgen ] is used , with the addition of @xmath254 taking the dot product of @xmath34with both sides of equation ( [ eq : momentad ] ) gives @xmath34@xmath255 , which results in the following algebraic relation for @xmath256 : @xmath257 which can be differentiated to give @xmath258 for use in equation ( [ eq : virphiprime ] ) .
i now have a system of 8 ordinary differential equations in the variables @xmath142 , @xmath205 , @xmath206 , @xmath208 , @xmath209 , @xmath259 , @xmath141 and @xmath212 .
this system differs from that of l96 in the treatment of the induction equation .
a solution of this version of the disk equations and a matching wind solution are presented in section [ res ] .
that solution assumes a functional form for the matter
field coupling parameter @xmath260 that differs from the one used in l96while l96 assumes that @xmath242 increases monotonically with height , i take @xmath242 to be constant along a field line , as would be the case when the ion density is constant ( wk93 ) . in fact , the coupling parameter is expected to _ decrease _ with height in the upper reaches of real protostellar disks not too far from the origin @xcite , reflecting the decrease of the ion mass density @xmath261 with height ( which occurs even though the ionization fraction increases with height because the neutral mass density @xmath262 drops rapidly with @xmath27 ) .
on the other hand , the coupling parameter increases with radial position because the keplerian frequency @xmath40 decreases . taking @xmath242 constant is a compromise between these competing effects as the integration proceeds upwards and outwards along a field line .
finally , the solution presented here differs from the solutions presented in l96 in that i initiate the numerical integration of the disk equations using a taylor series expansion ( see appendix [ taylorad ] ) , whereas l96 uses the approximation @xmath263 in a small region near the midplane to obtain algebraic relations for @xmath206 and @xmath208 . from equation ( [ eq : virz ] ) , @xmath264 .
the @xmath60 component of momentum conservation ( equation ( [ eq : moment ] ) ) , with equation ( [ eq : momentad ] ) used to replace the lorentz force term with an ion - neutral drag force term , yields equation ( 13 ) of l96 . at the midplane
, this equation reduces to @xmath265 or @xmath266 which gives @xmath202 in terms of the parameters @xmath267 , @xmath99 , @xmath268 , @xmath187 , and @xmath4 .
continuity of @xmath269 implies @xmath270 , and the @xmath86 component of the momentum equation ( cf . equation ( 14 ) of l96 ) evaluated at the midplane gives @xmath271 which can be used to calculate @xmath272 . combining ampre s law ( equation ( [ eq : amp ] ) ) with equation ( [ eq : momentad ] ) yields equations for @xmath170 and @xmath171 ( cf .
equations ( 17)(19 ) of l96 ) ; evaluating at the midplane yields @xmath273 and @xmath274 the expression for @xmath222 ( see table [ tab : bcs ] ) remains unchanged .
the taylor expansion procedure used to initiate the integration is described in appendix [ taylorad ] .
the dimensionless parameters @xmath275 and @xmath276 are model inputs , as are the conductivity tensor component profiles @xmath277 , @xmath278 , and @xmath279 .
alternatively , values for @xmath267 and @xmath280 together with a profile @xmath281 for the neutral field coupling parameter suffices for the case of a disk in the ambipolar diffusion regime . in either case , a disk / wind solution involves constraining three parameters to satisfy certain conditions imposed by physical processes outside the disk . the first constrained parameter is @xmath99 , the power - law index of the magnetic flux distribution ( equation ( [ eq : psidef ] ) ) .
the second is @xmath187 , the normalized midplane value of @xmath282 ( equation ( [ eq : epsdef ] ) ) .
the third is @xmath4 , the normalized midplane value of @xmath91 ( equation ( [ eq : epsbdef ] ) ) .
two of the conditions to be satisfied are that the disk solution pass smoothly through the sonic point and match onto a wind solution that passes smoothly through the alfvn point .
the third condition relates the angle of the magnetic field at the disk surface to the distribution of the magnetic field exterior to the disk ( see section [ surfconstr ] ) .
the sonic point ( @xmath283 ) is a singular point of the system of equations describing the disk ( see equation ( [ eq : rhoprime ] ) and subsequent discussion ) , so it is not possible to integrate directly into the sonic point , but it is possible to get close .
i run a bisection routine on @xmath187 , holding @xmath99 and @xmath4 fixed , until i obtain a disk solution that accelerates smoothly almost all the way up to the sonic point .
i then extrapolate the solution to @xmath284 and attempt to match the disk solution to a bp82-type wind solution , using the procedure outlined in skw11 .
that is , i calculate the value of the wind parameter @xmath285 ( the normalized mass - to - magnetic flux ratio ) using the disk solution at the sonic point , and i calculate @xmath286 and @xmath287 , respectively the ratio of the radial and vertical components of the magnetic field ( @xmath288 in the notation of skw11 ) and the total specific angular momentum , using the disk solution at the disk surface . here
, the disk surface is taken to be the point where the azimuthal neutral gas velocity matches the local keplerian rotation rate .
i run a bisection routine on @xmath99 until the values of @xmath285 , @xmath289 and @xmath290 calculated from the disk solution match the values at the base of a wind solution .
the matching procedure used here guarantees continuity of the magnetic field components , the poloidal mass flux , and the total specific angular momentum .
however , it should be pointed out that the electric field is , in general , not continuous at the transition from the disk solution to the wind solution : in particular , there is a non - zero azimuthal comoving electric field @xmath291 in the disk solution because the parameter @xmath4 is generally non - zero , but the cold , ideal mhd equations describing the wind solution have @xmath292 identically .
in addition , the bp82 wind solution is derived for a razor - thin disk that has vanishing radial and vertical gas velocities at the base of the wind ( negligible compared with the azimuthal velocity , which is exactly keplerian at the base of the wind ) ; these are good approximations when the region of interest is the large - scale outflow , but break down on small scales comparable to the thickness of the disk . the wind solution is therefore expected to give physically uninteresting results very close to the disk surface .
in fact , the wind solution has a singular point at the midplane just as the disk solution does .
i follow bp82 in integrating the wind equations downwards starting from the alfvn point and stopping very close to the disk surface , where the solution is matched to a taylor expansion .
i then interpolate between the disk solution at the top of the disk and the wind solution at a distance @xmath293 above the disk surface , where @xmath294 , @xmath295 is the radial distance of the disk surface along the field line of interest , and @xmath296 is the value of @xmath297 at the alfvn point .
the wind solution extends past the alfvn point .
there is a second critical point of the equations for a self - similar wind corresponding to the modified fast magnetosonic point ( bp82 ) .
however , before a cold bp82-type wind solution reaches the modified fast magnetosonic point , the flow `` over - collimates '' as the field lines bend back towards the axis of symmetry .
this unphysical behavior results from ignoring heating of the jet material and thermal pressure support in the cold wind model .
`` warm '' self - similar jet models whose solutions do cross the modified fast magnetosonic have been found by @xcite and @xcite .
for this study , i terminate the wind solution when the field lines begin to bend back towards the jet axis .
ol01 pointed out that in a self - consistent , global model of a magnetized disk , the inclination of the poloidal field at the disk surface can be determined under the approximation that the external magnetic field is nearly potential ( @xmath298 ) .
an analytic expression relating @xmath1 to the distribution of @xmath299 along the disk surface was derived , under this approximation and assuming an infinitely thin disk , by @xcite , who modeled magnetically supported , self - gravitating molecular cloud cores , and by @xcite , who modeled flux advection in viscous accretion disks .
the poloidal field inclination at the disk s surface is directly related to the balance between the inward advection of magnetic flux by the accretion flow and the outward diffusion of magnetic flux through the non - ideal disk .
this connection follows from the fact that the radial component of the magnetic force ( per unit area ) at the disk surface , which is @xmath300 , is balanced in the ambipolar diffusion regime by the radial ion
neutral drag force , which is @xmath301 ( or , taking the midplane values , @xmath302 , which indicates the explicit relation of @xmath303 to the disk model parameters ) . determining @xmath1 self - consistently in this manner generally implies a nonzero value of @xmath4 .
the relevant expression for @xmath1 is @xmath304 j_{0}(kr^ { ' } ) \ , , \label{eq : disksurf}\end{aligned}\ ] ] where @xmath305 and @xmath306 are bessel functions of the first kind of order 0 and 1 , respectively , @xmath307 is the midplane vertical field component at radius @xmath60 , which is assumed to remain constant up to the surface [ so that @xmath308 , and @xmath309 is the external magnetic field at `` infinity , '' which is taken to be @xmath310 , so it may be safely ignored in the second integral . for a power - law profile @xmath311 , equation ( [ eq : disksurf ] ) yields @xmath312[\gamma({a}/{2 } ) \ , \gamma({3}/{2 } - { a}/{2})]$ ] for @xmath313 , where @xmath314 is the gamma function .
the power - law exponent is related to the disk model parameter @xmath99 by @xmath315 , so @xmath316\ , \gamma\left[3/2 - 1/\left(2\zeta\right)\right]}{\gamma\left[1 - 1/\left(2\zeta\right ) \right ] \ , \gamma\left[1/2 + 1/\left(2\zeta\right)\right ] } \;. \label{eq : disksurf2}\ ] ] this result holds for @xmath317 and is therefore appropriate for the values @xmath318 that are expected to characterize physically viable solutions ( e.g. , l96 ; f97 ) . for @xmath319 , equation ( [ eq : disksurf2 ] ) yields @xmath320 .
once a disk solution that passes smoothly through the sonic point has been matched to a wind solution , i bisect on @xmath4 until the value of @xmath321 matches the value imposed by the field distribution .
protostellar disks have typical inferred temperatures at 1 au of @xmath322 k , masses of @xmath323 , column densities at 5 au of @xmath324 g @xmath325 , and extend to radii @xmath326 au ( e.g. , * ? ? ?
disk / wind model solutions typically have @xmath327 ( e.g. , wk93 ) , and observations of protostellar systems with jets indicate that the ratio of the outflow and accretion rates is @xmath328 ( e.g. , * ? ? ?
finally , observations of remanent magnetization of meteorite samples suggest a magnetic field of @xmath329 g at 3 au in the early solar nebula @xcite .
motivated by these facts , i choose parameter values @xmath330 and @xmath331 .
these values together with the constant coupling profile @xmath332 yield a solution of the disk equations described in section [ casead ] .
this solution passes smoothly through the sonic point , matches to a wind solution that passes smoothly through the alfvn point , and satisfies the constraint on the magnetic field at the surface of the disk .
the corresponding disk and wind parameters are @xmath333 , @xmath334 , @xmath335 , @xmath336 , @xmath337 , and @xmath338 .
the small value of @xmath339 found here is consistent with the predictions of l96 and f97 that cold wind - driving disks should have field - line distributions very close to the limiting case of @xmath101 ( but see @xcite for arguments that larger values of @xmath339 can occur in wind - driving disks when thermal effects are considered is related to the quantity @xmath100 in @xcite as @xmath340 ; the ratio @xmath341 of the mass outflow in the wind between radii @xmath342 and @xmath343 to the mass accretion rate at @xmath342 is @xmath344 . for @xmath345 and an observed @xmath346 , @xmath100
is inferred to lie in the range @xmath347 .
of the two representative solutions in @xcite , the one with almost no entropy generation gives @xmath348 , while the one with strong entropy generation gives @xmath349 , in which case almost all of the disk mass flows out into the wind rather than accreting onto the central protostar .
] ) . a rough estimate of the ratio of the mass outflow rate from both sides of the disk , between an outermost radius @xmath343 and a given radius @xmath60 , to the mass accretion rate through the disk at @xmath60 , can be derived from equation ( 89 ) of skw11 as @xmath350 $ ] ; for the derived solution , and taking @xmath351 , this ratio is @xmath352 . for a central protostar of mass 0.5 @xmath353 , the model has a column density of @xmath354 g @xmath325 at 1 au and a column density of @xmath355 g @xmath325 at 5 au if the density is normalized so that the accretion rate is @xmath356 and the total disk mass out to 200 au is @xmath357 . in comparison ,
the minimum mass solar nebula model @xcite has a column density at 1 au of @xmath358 g @xmath325 and at 5 au of @xmath359 g @xmath325 .
the density normalization used here corresponds to a midplane magnetic field strength at 1 au of @xmath360 g and at 3 au of @xmath361 g. the disk temperature at 1 au is @xmath362 125 k. the gas density , velocity , and magnetic field structures of the disk solution as functions of position along a field line are shown in figures [ fig : disk1rho][fig : disk1b ] .
the plots extend from the midplane , @xmath363 , to the sonic point , @xmath364 .
the location of the sonic point is marked in figure [ fig : disk1fieldline ] , which shows the poloidal magnetic field line structure in the disk .
the magnetic field components and the density are normalized by the midplane values of @xmath34 and @xmath23 , respectively .
the velocity components , with the keplerian velocity subtracted from the azimuthal neutral and ion velocities , are normalized by the local value of the sound speed along the field line ( that is , taking into account the spatial variation of the sound speed ) . at the midplane ,
the field is vertical by symmetry ; above the midplane , the radial component grows rapidly , a small negative toroidal component develops , and the vertical component decreases slightly .
the vertical velocity @xmath365 starts at zero at the midplane , decreases to small negative values in the near - midplane region ( reflecting the radial convergence of the accretion flow ) , and then increases back through zero at @xmath366 , becoming comparable to @xmath50 at @xmath367 .
the radial velocity starts out at @xmath368 at the midplane and decreases in the near - midplane region , reaching a minimum at @xmath369 and then increasing , switching from an inflow to an outflow at @xmath370 .
the azimuthal velocity is sub - keplerian at the midplane ( @xmath371 ) but increases smoothly with height , becoming super - keplerian at the disk surface , @xmath372 , slightly above the point where @xmath282 becomes positive . in the hydrostatic approximation ( @xmath373 ) , @xmath374 and @xmath375 at the same point ( wk93 ) .
the effective density scale height @xmath376 ( defined so that @xmath377 ) is @xmath378 , indicating that magnetic squeezing has a significant effect on the disk structure ( wk93 ) .
the gas density , velocity , and magnetic field structures of the matched disk and wind solutions as functions of position along a field line are shown in figures [ fig : diskwind1rho][fig : diskwind1b ] .
the plots extend from @xmath363 to @xmath379 , where the field lines begin to bend back towards the jet axis and the wind solution terminates .
along the way , the wind solution passes through the alfvn point , located at @xmath380 .
figures [ fig : diskwind1rho][fig : diskwind1b ] are shown with the variable @xmath381 plotted on a logarithmic axis so that the structures of the disk and wind regions can be distinguished .
the location of the alfvn point is marked in figure [ fig : diskwind1fieldline ] , which shows the poloidal magnetic field line structure in the disk and wind .
in the region above the disk , the poloidal field quickly drops while the azimuthal field amplitude first grows slightly ( @xmath382 becomes more negative ) and then decreases , in accordance with the expected behavior of a wind - driving disk .
the vertical and azimuthal velocities increase smoothly throughout the outflow , while the radial velocity reaches a maximum and begins decreasing well below the alfvn point .
note that figure [ fig : diskwind1v ] differs from figure [ fig : disk1v ] in that the unmodified azimuthal velocity is plotted and the velocities are normalized by the local value of the keplerian speed ( that is , taking into account the spatial variation of the keplerian speed ) .
the solution presented here is similar to the wk93 model in that it considers a strongly coupled , ambipolar diffusive disk , but differs in that it is a global model with a non - zero value of @xmath4 determined self - consistently from a constraint on the field line inclination at the disk surface .
the equilibrium solution curves for the present model are undoubtedly modified from those of k04 , but on the assumption that the changes are not drastic , a comparison of the representative solution and the results given in k04 may be instructive .
the representative solution given here has @xmath336 , @xmath383 , and @xmath384 ;
the quantity @xmath241 in k04 corresponds to @xmath242 . comparing these values to figures 1 and 2 of k04
, it appears that the representative solution given here lies near the lowest point of the lower branch of the @xmath385 equilibrium solution curve in figure 1 , and to the left of the upper branch of the @xmath385 equilibrium solution curve in figure 2 . in both cases
, the comparison suggests that the representative solution given here is stable .
this illustrative solution assumes that the underlying disk is in the ambipolar conductivity regime .
ambipolar diffusion is likely a reasonable assumption in the outer regions of typical protostellar disks , but may not hold throughout the inner regions of protostellar disks , which are more likely in the hall or ohm conductivity regimes .
however , it is worth noting in this connection that models of wind - driving disks have column densities in the inner regions that are much smaller than in typical turbulent - viscosity accretion disks .
these smaller column densities suggest that the region of a wind - driving disk in which ambipolar diffusivity is a good approximation may be substantially larger than in other types of disk models .
the model for the global structure of a magnetized protostellar disk / wind system presented here improves on previous studies ( e.g. , wk93 ; l96 ; f97 ; ol01 ) by including , for the first time , a fully self - consistent treatment of flux migration on the accretion timescale .
this treatment employs the approximation that the magnetic field on scales @xmath386 above the disk surface is very nearly potential , leading to a simple constraint on the surface field inclination ( equation ( [ eq : disksurf ] ) ) .
the model also improves on previous studies ( e.g. , skw07 ) by extending the conductivity tensor formalism for describing non - ideal disk material from radially localized disk models to global , radially self - similar disk models .
the model presented here describes solutions for a disk and wind that are steady on timescales up to the flux migration timescale .
studies of simplified equilibrium models of disk winds by @xcite , @xcite , and @xcite have raised the possibility that magnetocentrifugal disk winds are inherently unstable .
* hereafter k04 ) showed that for strongly coupled wind - driving disk models of the sort described in wk93 there are two equilibrium solution branches , one stable and one unstable , and argued that protostellar systems correspond to the stable branch .
this line of argument has received some support from simulations of axisymmetric , diffusive accretion disks launching self - collimating , magnetocentrifugal disk winds that have been carried out by @xcite and @xcite .
these simulations confirm that the resulting large - scale jets are continuous and steady over many dynamic timescales , providing support for the present model s quasi - steady state assumption . in turn
, the semianalytic model presented here facilitates calculations that can not currently be carried out in full numerical simulations .
the formulation presented here allows construction of global disk / wind models with realistic conductivity profiles in the disk .
finding matched disk / wind solutions of the sort presented here is a difficult task due to the imposed restrictions of requiring all angular momentum transport to occur via the wind , and the strict self - similarity of the disk and wind solutions . in reality
, protostellar disk / wind systems may transport angular momentum simultaneously through multiple mechanisms , such as magnetic braking , gravitational instability , and turbulence ( see skw07 for a radially localized disk model that includes angular momentum transport by both a wind and turbulence driven by the magnetorotational instability ) , and are unlikely to be globally self - similar . however , simple models of the sort presented here are valuable tools for investigating global processes in disk / wind systems , such as dust transport and processing ( s. teitler , in preparation ) .
studies of dust evolution processes in wind - driving disks will complement similar studies carried out in simple models of turbulent disks ( e.g. , * ? ? ?
* ; * ? ? ?
* ) and in the x - wind model ( e.g. , * ? ? ?
* ) , and are of particular importance for wind - driving disks since small dust grains can have strong effects on the coupling of neutral disk material to magnetic fields ( e.g. , * ? ?
* ; * ? ? ?
i thank arieh knigl for his unwearying support and guidance over the past several years , and my thesis committee members fausto cattaneo , fred ciesla , joshua frieman , and don york for their patience and advice .
i also thank raquel salmeron , konstantinos tassis , and mark wardle for many helpful discussions .
in addition , raquel salmeron and zhi - yun li graciously made available unpublished material that proved very helpful .
finally , i thank the anonymous referee for many helpful comments . this work was supported in part by nasa astrophysics theory and fundamental physics program grant nnx09ah38 g and nsf grant ast-0908184 .
given the values of the independent variable @xmath114 and the dependent variables @xmath142 , @xmath206 , @xmath208 , @xmath209 , @xmath141 , @xmath212 , @xmath214 , and @xmath205 , the auxiliary variables @xmath140 and @xmath216 can be obtained directly from the algebraic relations ( [ eq : brbz ] ) and ( [ eq : induct2b ] ) .
their derivatives are @xmath387 and @xmath388 the auxiliary variables @xmath389 and @xmath390 are obtained from equation ( [ eq : ec2 ] ) . the next step is to combine the algebraic relations ( [ eq : jrjz ] ) , ( [ eq : ohm2r ] ) , and ( [ eq : ohm2z ] ) to obtain @xmath391 using equations ( [ eq : ecpar ] ) and ( [ eq : ecperp ] ) , we have @xmath392\right ) \
, , \label{eq : ecrels}\end{aligned}\ ] ] giving @xmath393@xmath394 and @xmath393@xmath395 in terms of @xmath396 and the known quantities @xmath389 , @xmath390 , and @xmath397 . substituting these expressions into equation ( [ eq : jrjz2 ] ) and solving ,
we obtain @xmath398 = & \nonumber \\
e_{\rm{c}\it{r}}\left[\tilde{\sigma}_{\rm o}\left(s-\frac{b_{z}}{b_{r}}\right)-\tilde{\sigma}_{\rm p}\frac{b_{\phi}^{2}}{b_{r}b_{z}}+\frac{\tilde{\sigma}_{\rm h}}{b}\left(b_{\phi}+s\left[\frac{b_{\phi}^{3}}{b_{r}b_{z}}+\frac{b_{\phi}b_{z}}{b_{r}}\right]\right)\right ] & \nonumber\\ + e_{\rm{c}\it{\phi}}\left[\tilde{\sigma}_{\rm p}\frac{b_{\phi}}{b_{z}}-\frac{\tilde{\sigma}_{\rm h}}{b}\left(s\left[\frac{b_{\phi}^{2}}{b_{z}}+b_{z}\right]+b_{r}\right)\right ] + \left(\frac{1}{\zeta}-1\right)\frac{b_{\phi}}{x } & \nonumber \,,\end{aligned}\ ] ] which simplifies to @xmath399 where @xmath400}{\left[\left(\tilde{\sigma}_{\rm o}-\tilde{\sigma}_{\rm p}\right)bb_{z}+\tilde{\sigma}_{\rm p}x^{2}b^{3}-s\tilde{\sigma}_{\rm h}x^{2}b^{2}b_{\phi}\right ] } \nonumber\ ] ] and @xmath401}{\left[\left(\tilde{\sigma}_{\rm o}-\tilde{\sigma}_{\rm p}\right)bb_{z}+\tilde{\sigma}_{\rm p}x^{2}b^{3}-s\tilde{\sigma}_{\rm h}x^{2}b^{2}b_{\phi}\right ] } \nonumber \,.\ ] ] equation ( [ eq : ecperpr2 ] ) for @xmath402 in terms of known quantities can be used to infer @xmath403 , @xmath404 , @xmath405 , @xmath406 , @xmath407 , @xmath408 , @xmath176 , @xmath218 , @xmath220 , and @xmath409 using equations ( [ eq : ecrels ] ) and ( [ eq : ohm2r])([eq : ohm2z ] ) .
these auxiliary quantities can in turn be used in equations ( [ eq : rhoprime ] ) , ( [ eq : mom2r])([eq : mom2z ] ) , ( [ eq : ampphi2 ] ) and ( [ eq : ampr2 ] ) to obtain @xmath162 , @xmath166 , @xmath169 , @xmath167 , @xmath170 , and @xmath171 .
the differential equation ( [ eq : induct2a ] ) for @xmath175 involves the derivative of the auxiliary quantity @xmath176 ; @xmath176 is given by @xmath410\right)+v_{\phi}b_{r}-v_{r}b_{\phi } \nonumber \,.\end{aligned}\ ] ] upon substituting for @xmath402 from equation ( [ eq : ecperpr2 ] ) , this becomes @xmath411-\frac{b_{r}}{b_{z}}a+\frac{b_{\phi}^{2}}{b_{r}b_{z}}\left[1-a\right]\right ) \nonumber \\ & & + \left(v_{\phi}b_{z}-v_{z}b_{\phi}\right)\left(\frac{b_{z}}{b_{r}}\left[1-a\right]-\frac{b_{r}}{b_{z}}a+\frac{b_{\phi}^{2}}{b_{r}b_{z}}\left[1-a\right]\right ) \nonumber \\ & & -b\frac{b^{2}}{b_{r}b_{z}}-\frac{b_{\phi}}{b_{z}}e_{\rm{c}\it{\phi}}+v_{\phi}b_{r}-v_{r}b_{\phi } \ , .
\label{eq : ezer}\end{aligned}\ ] ] differentiating equation ( [ eq : ezer ] ) for @xmath176 gives @xmath412 where @xmath413 @xmath414 @xmath415 @xmath416\bigg)\bigg/\left[\left(\tilde{\sigma}_{\rm o}-\tilde{\sigma}_{\rm p}\right)bb_{z}+\tilde{\sigma}_{\rm p}x^{2}b^{3}-s\tilde{\sigma}_{\rm h}x^{2}b^{2}b_{\phi}\right ] \nonumber\\ & & -\bigg(\left[\tilde{\sigma}_{\rm o}bb_{z}+\tilde{\sigma}_{\rm p}x^{2}bb_{\phi}^{2}-s\tilde{\sigma}_{\rm h}x^{2}b^{2}b_{\phi}-\tilde{\sigma}_{\rm h}b_{r}b_{\phi}\right ] \big[\left(\tilde{\sigma}_{\rm o}^{\prime}-\tilde{\sigma}_{\rm p}^{\prime}\right)bb_{z}+\left(\tilde{\sigma}_{\rm o}-\tilde{\sigma}_{\rm p}\right)b^{\prime}b_{z } \nonumber \\ & & + \left(\tilde{\sigma}_{\rm o}-\tilde{\sigma}_{\rm p}\right)bb_{z}^{\prime}+\tilde{\sigma}_{\rm p}x^{2}b^{3}+2\tilde{\sigma}_{\rm p}xx^{\prime}b^{3}+3\tilde{\sigma}_{\rm p}x^{2}b^{2}b^{\prime}-\tilde{\sigma}_{\rm h}x^{2}b^{2}b_{\phi}-s\tilde{\sigma}_{\rm h}^{\prime}x^{2}b^{2}b_{\phi } \nonumber \\ & & -2s\tilde{\sigma}_{\rm h}xx^{\prime}b^{2}b_{\phi}-2s\tilde{\sigma}_{\rm h}x^{2}bb^{\prime}b_{\phi}-s\tilde{\sigma}_{\rm h}x^{2}b^{2}b_{\phi}^{\prime}\big]\bigg)\bigg/\left[\left(\tilde{\sigma}_{\rm o}-\tilde{\sigma}_{\rm p}\right)bb_{z}+\tilde{\sigma}_{\rm p}x^{2}b^{3}-s\tilde{\sigma}_{\rm h}x^{2}b^{2}b_{\phi}\right]^{2 } \nonumber \,,\end{aligned}\ ] ] and @xmath417\left[-\tilde{\sigma}_{\rm p}x^{2}bb_{r}b_{\phi}+s\tilde{\sigma}_{\rm h}x^{2}b^{2}b_{r}+\tilde{\sigma}_{\rm h}b_{r}^{2}\right ] \nonumber \\ & & + e_{\rm{c}\it{\phi}}\left[-\tilde{\sigma}_{\rm p}^{\prime}x^{2}bb_{r}b_{\phi}-2\tilde{\sigma}_{\rm p}xx^{\prime}bb_{r}b_{\phi}-\tilde{\sigma}_{\rm p}x^{2}b^{\prime}b_{r}b_{\phi}-\tilde{\sigma}_{\rm p}x^{2}bb_{r}^{\prime}b_{\phi}-\tilde{\sigma}_{\rm p}x^{2}bb_{r}b_{\phi}^{\prime } \right .
\nonumber \\ & & \left .
+ \tilde{\sigma}_{\rm h}x^{2}b^{2}b_{r}+s\tilde{\sigma}_{\rm h}^{\prime}x^{2}b^{2}b_{r}+2s\tilde{\sigma}_{\rm h}xx^{\prime}b^{2}b_{r}+2s\tilde{\sigma}_{\rm h}x^{2}bb^{\prime}b_{r}+s\tilde{\sigma}_{\rm h}x^{2}b^{2}b_{r}^{\prime } \right .
\nonumber \\ & & \left .
+ \tilde{\sigma}_{\rm h}^{\prime}b_{r}^{2}+2\tilde{\sigma}_{\rm h}b_{r}b_{r}^{\prime}\right]-\left[\frac{1}{\zeta}-1\right]\big[x^{\prime}bb_{r}b_{\phi}b_{z}+xb^{\prime}b_{r}b_{\phi}b_{z}+xbb_{r}^{\prime}b_{\phi}b_{z } \nonumber\\ & & + xbb_{r}b_{\phi}^{\prime}b_{z}+xbb_{r}b_{\phi}b_{z}^{\prime}\big]\bigg)\bigg/\left[\left(\tilde{\sigma}_{\rm o}-\tilde{\sigma}_{\rm p}\right)bb_{z}+\tilde{\sigma}_{\rm p}x^{2}b^{3}-s\tilde{\sigma}_{\rm h}x^{2}b^{2}b_{\phi}\right ] \nonumber \\ & & -\bigg(\left[e_{\rm{c}\it{\phi}}\left(-\tilde{\sigma}_{\rm p}x^{2}bb_{r}b_{\phi}+s\tilde{\sigma}_{\rm h}x^{2}b^{2}b_{r}+\tilde{\sigma}_{\rm h}b_{r}^{2}\right)-\left(\frac{1}{\zeta}-1\right)xbb_{r}b_{\phi}b_{z}\right ] \big[\left(\tilde{\sigma}_{\rm o}^{\prime}-\tilde{\sigma}_{\rm p}^{\prime}\right)bb_{z } \nonumber \\ & & + \left(\tilde{\sigma}_{\rm o}-\tilde{\sigma}_{\rm p}\right)b^{\prime}b_{z}+\left(\tilde{\sigma}_{\rm o}-\tilde{\sigma}_{\rm p}\right)bb_{z}^{\prime}+\tilde{\sigma}_{\rm p}x^{2}b^{3}+2\tilde{\sigma}_{\rm p}xx^{\prime}b^{3}+3\tilde{\sigma}_{\rm p}x^{2}b^{2}b^{\prime}-\tilde{\sigma}_{\rm h}x^{2}b^{2}b_{\phi}-s\tilde{\sigma}_{\rm h}^{\prime}x^{2}b^{2}b_{\phi } \nonumber \\ & & -2s\tilde{\sigma}_{\rm h}xx^{\prime}b^{2}b_{\phi}-2s\tilde{\sigma}_{\rm h}x^{2}bb^{\prime}b_{\phi}-s\tilde{\sigma}_{\rm h}x^{2}b^{2}b_{\phi}^{\prime}\big]\bigg)\bigg/\left[\left(\tilde{\sigma}_{\rm o}-\tilde{\sigma}_{\rm p}\right)bb_{z}+\tilde{\sigma}_{\rm p}x^{2}b^{3}-s\tilde{\sigma}_{\rm h}x^{2}b^{2}b_{\phi}\right]^{2 } \nonumber \,.\end{aligned}\ ] ] substituting equation ( [ eq : ezerprime ] ) into equation ( [ eq : induct2a ] ) , we have @xmath418+e_{z}\right ) \nonumber \,,\ ] ] which gives @xmath419+e_{z}\right)-ds}{1+cs } \ , .
\label{eq : erprime}\ ] ]
as described in section [ midgen ] , the midplane is a singular point of the disk equations .
the procedure adopted here for dealing with this issue is to calculate taylor series expansions of the variables of integration from the midplane to a small distance @xmath420 above the midplane before beginning numerical integration of the disk equations .
the taylor series expansions for a general , non - ideal disk described using the conductivity tensor formalism are presented in section [ taylorgen ] .
the case of a disk in the ambipolar diffusion regime with constant neutral field coupling parameter @xmath242 is covered in section [ taylorad ] .
differentiating equation ( [ eq : rhoprime ] ) once with respect to @xmath114 and dropping terms that vanish at the midplane gives @xmath423\bigg[\left(2-\frac{2}{\zeta}\right)v_{r,0}\left(v_{z,0}^{\prime}-v_{r,0}\right)+\left(4-\frac{2}{\zeta}\right)\theta\left(1+x_{0}^{\prime\prime}\right ) \nonumber \\ & & + v_{\phi,0}^{2}+\nu\left(j_{\phi,0}\left[b_{r,0}^{\prime}+1\right]-b_{\phi,0}^{\prime}j_{r,0}\right)\bigg ] \ , .
\label{eq : pprho0}\end{aligned}\ ] ] differentiating the @xmath27 component of equation ( [ eq : ec2 ] ) gives @xmath425 .
differentiating equation ( [ eq : ecperp ] ) yields @xmath426 .
differentiating equation ( [ eq : ampz2 ] ) and equation ( [ eq : ohm2z ] ) gives @xmath427 and @xmath428-\tilde{\sigma}_{\rm p,0}e_{{\rm c}z\perp,0}^{\prime}\right)/\tilde{\sigma}_{\rm o,0}$ ] , respectively .
these relations can be used to solve for @xmath234 in equation ( [ eq : pper0 ] ) .
calculating @xmath232 and @xmath233 is somewhat more complex .
differentiating equations ( [ eq : mom2r ] ) and ( [ eq : mom2phi ] ) twice with respect to @xmath114 and dropping terms that vanish at the midplane gives @xmath429 expressions for the unknown quantities @xmath431 and @xmath432 can be derived by differentiating equations ( [ eq : ampr2 ] ) and ( [ eq : ampphi2 ] ) twice with respect to @xmath114 and dropping terms that vanish at the midplane , which gives we have @xmath435 , since @xmath436 .
differentiating equation ( [ eq : ecpar ] ) twice gives the quantities @xmath437 and @xmath438 as @xmath439 and @xmath440 .
the quantities @xmath441 and @xmath442 are obtained by differentiating equation ( [ eq : ecparperp ] ) twice to obtain @xmath443 and @xmath444 .
finally , differentiating equation ( [ eq : ec2 ] ) twice gives @xmath445 and @xmath446 .
plugging these relations back into equation ( [ eq : ppvr0 ] ) and equation ( [ eq : ppvphi0 ] ) gives two equations that can be solved for @xmath232 and @xmath233 in terms of known quantities .
differentiating the @xmath27 component of equation ( [ eq : moment ] ) ( cf .
equation ( 12 ) of l96 ) once and evaluating at the midplane gives @xmath450-\left[4-\frac{2}{\zeta}\right]\theta x_{0}^{\prime\prime}-1+\upsilon_{0}\left[v_{\rm{i}\mathit{z},0}^{\prime}-v_{z,0}^{\prime}\right]\right)/\theta \ , . \label{eq : pprho0ad}\ ] ] andrews , s. m. , & williams , j. p. 2007
, , 659 , 705 balbus , s. a. , & hawley , j. f. 1991 , , 376 , 214 bally , j. , reipurth , b. , & davis , c. j. 2007 , in protostars & planets v , ed .
b. reipurth , d. jewitt , & k. keil ( tucson : univ .
arizona press ) , 215 blandford , r. d. , & payne , d. g. 1982 , , 199 , 883 ( bp82 ) brauer , f. , dullemond , c. p. , & henning , th .
2008 , 480 , 859 campbell , c. g. 2003 , , 345 , 123 campbell , c. g. 2005 , , 361 , 396 campbell , c. g. 2009 , , 392 , 271 cao , x. , & spruit , h. c. 2002 , , 385 , 289 casse , f. , & ferreira , j. 2000 , , 361 , 1178 casse , f. , & keppens , r. 2002 , , 581 , 988 casse , f. , & keppens , r. 2004 , , 601 , 90 ciolek , g. e. , & mouschovias , t. ch .
1993 , , 418 , 774 draine , b. t. , roberge , w. g. , & dalgarno , a. 1983 , , 264 , 485 ferreira , j. 1997 , , 319 , 340 ( f97 ) ferreira , j. , & casse , f. 2004 , , 601 , l139 guan , x. , & gammie , c. f. 2009 , , 697 , 1901 hartigan , p. , edwards , s. , & ghandour , l. 1995 , , 452 , 736 hayashi , c. 1981 , prog theor phys supp , 70 , 35 heyvaerts , j. , & norman , c. 1989 , , 347 , 1055 keller , ch .
, & gail , h .- p .
2004 , , 415 , 1177 knigl , a. 1989 , , 342 , 208 knigl , a. 2004 , , 617 , 1267 ( k04 ) knigl , a. , & salmeron , r. 2011 , in physical processes in circumstellar disks around young stars , ed .
p. j. v. garcia ( chicago : univ .
chicago press ) , in press ( arxiv:1004:1875 ) knigl , a. , salmeron , r. , & wardle , m. 2010 , , 401 , 479 lesur , g. , & longaretti , p .- y .
2009 , , 504 , 309 levy , e. h. , & sonett , c. p. 1978 , in protostars & planets , ed .
t. gehrels ( tucson : univ .
arizona press ) , 516 li , z .- y .
1995 , , 444 , 848 li , z .- y .
1996 , , 465 , 855 ( l96 ) lovelace , r. v. e. , rothstein , d. m. , & bisnovatyi - kogan , g. s. 2009 , , 701 , 885 lubow , s. h. , papaloizou , j. c. b. , & pringle j. e. 1994 , , 267 , 235 lubow , s. h. , papaloizou , j. c. b. , & pringle j. e. 1994 , , 268 , 1010 nishi , r. , nakano , t. , & umebayashi , t. 1991 , , 368 , 181 ogilvie , g. i. & livio , m. 2001 , , 553 , 158 ( ol01 ) ouyed , r. , & pudritz , r. e. 1997 , , 482 , 712 rothstein , d. m. , & lovelace , r. v. e. 2008 , , 677 , 1221 salmeron , r. , knigl , a. , & wardle , m. 2007 , , 375 , 177 ( skw07 ) salmeron , r. , knigl , a. , & wardle , m. 2011 , mnras , in press ( arxiv:1006.0299 ) ( skw11 ) salmeron , r. , & wardle , m. 2005 , , 361 , 45 shu , f. h. , shang , h. , gounelle , m. , glassgold , a. e. , & lee , t. 2001 , , 548 , 1029 vlahakis , n. , tsinganos , k. , sauty , c. , & trussoni , e. 2000 , , 318 , 417 wardle , m. 2007 , , 311 , 35 wardle , m. , & knigl , a. 1993 , , 410 , 218 ( wk93 ) zanni , c. , ferrari , a. , rosner , r. , bodo , g. , & massaglia , s. 2007 , , 469 , 811 | the magnetocentrifugal disk wind mechanism is the leading candidate for producing the large - scale , bipolar jets commonly seen in protostellar systems .
i present a detailed formulation of a global , radially self - similar model for a non - ideal disk that launches a magnetocentrifugal wind .
this formulation generalizes the conductivity tensor formalism previously used in radially localized disk models .
the model involves matching a solution of the equations of non - ideal mhd describing matter in the disk to a solution of the equations of ideal mhd describing a `` cold '' wind .
the disk solution must pass smoothly through the sonic point , the wind solution must pass smoothly through the alfvn point , and the two solutions must match at the disk / wind interface .
this model includes for the first time a self - consistent treatment of the evolution of magnetic flux threading the disk , which can change on the disk accretion timescale .
the formulation presented here also allows a realistic conductivity profile for the disk to be used in a global disk / wind model for the first time .
the physical constraints on the model solutions fix the distribution of the magnetic field threading the disk , the midplane accretion speed , and the midplane migration speed of flux surfaces .
i present a representative solution that corresponds to a disk in the ambipolar conductivity regime with a nominal neutral - matter magnetic - field coupling parameter that is constant along field lines , matched to a wind solution .
i conclude with a brief discussion of the importance of self - similar disk / wind models in studying global processes such as dust evolution in protostellar systems . |
in this section we discuss our principal results ; the proofs are presented in sections [ s : proof - of - t1][s : more - proofs ] .
the central character of this paper is the function @xmath13,\ ] ] where @xmath14 runs over positive integers , and @xmath15 denotes the distance from @xmath16 to the nearest integer ; that is , @xmath1 for @xmath17 $ ] .
more explicitly , letting @xmath18 and @xmath19 for integer @xmath20 ( so that @xmath21 ) , we have @xmath22 whenever @xmath23 , and @xmath24 . the graphs of the function @xmath2 and the functions @xmath25 for @xmath26 are presented in figure [ f : fig ] . [ f : fig ] 0.08 recall that for @xmath27 , a real - valued function @xmath3 defined on a convex subset of a ( real ) normed vector space is called @xmath28-convex if it satisfies @xmath29 for all @xmath30 and @xmath31 in the domain of @xmath3 , and all @xmath32 $ ] .
we denote by @xmath33 the class of all @xmath34-convex functions on the interval @xmath4 $ ] ; that is , @xmath33 consists of all real - valued functions @xmath3 , defined on @xmath4 $ ] and satisfying @xmath35.\ ] ] also , let @xmath36 be the subclass of all those functions @xmath37 satisfying the boundary condition @xmath38 since @xmath33 is closed under translates by a linear function , any function @xmath37 can be forced into @xmath36 just by adding to it an appropriate linear summand .
hence , studying these two classes is essentially equivalent .
it is immediate from the definition that the classes @xmath33 and @xmath36 are `` symmetric around @xmath39 '' in the sense that a function @xmath3 belongs to the class @xmath33 ( @xmath36 ) if and only if so does the function @xmath40)$ ] .
our first principal result shows that the above - defined function @xmath2 lies in the class @xmath36 and indeed , is the ( pointwise ) largest function of this class .
[ t : max - f ] we have @xmath41 and @xmath42 for every function @xmath43 .
we remark that substituting @xmath44 and @xmath45 into shows that all functions from the class @xmath36 are uniformly bounded from above , whence the function @xmath46 is well defined .
it is not difficult to see that this function itself belongs to @xmath36 and is pointwise bounded from above by the function @xmath2 .
however , proving that @xmath41 is more delicate . for integer @xmath47 ,
let @xmath48 denote the class of all real - valued functions , defined on the interval @xmath4 $ ] and satisfying the boundary condition and the estimate @xmath49 for all @xmath50 $ ] with @xmath51 and @xmath52 .
these classes were introduced in @xcite , except that the functions from the class @xmath53 ( up to the normalization ) are known as @xmath34-_midconvex _ and under this name have been studied in a number of papers ; see , for instance , @xcite . as shown in ( * ? ? ?
* lemma 3 ) , every concave function from the class @xmath36 belongs to all classes @xmath48 .
thus , from theorem [ t : max - f ] we get [ c : finfm ] we have @xmath54 for all @xmath47 . as an application , we establish a result from the seemingly unrelated realm of edge isoperimetry . the edge - isoperimetric problem for a graph on the vertex set
@xmath55 is to find , for every non - negative integer @xmath56 , the smallest possible number of edges between an @xmath57-element set of vertices and its complement in @xmath55 .
we refer the reader to the survey of bezrukov @xcite and the monograph of harper @xcite for the history , general perspective , numerous results , variations , and further references on this and related problems . in the present paper we are concerned with the situation where the graph under consideration is a cayley graph on a finite abelian group .
we use the following notation . given two subsets @xmath58 of a finite abelian group @xmath9 , we consider the directed cayley graph , induced by @xmath8 on @xmath9 , and we write @xmath59 for the edge - boundary of @xmath7 ; that is , @xmath59 is number of edges in this graph from an element of @xmath7 to an element in its complement @xmath11 : @xmath60 equivalently , @xmath59 is the number of edges between @xmath7 and @xmath11 in the _ undirected _ cayley graph induced on @xmath9 by the set @xmath61 . as a consequence of corollary [ c : finfm ] ( thus , ultimately , of theorem [ t : max - f ] ) , we prove [ t : isoper ] let @xmath7 and @xmath8 be subsets of a finite abelian group @xmath9 , of which @xmath8 is generating .
if @xmath10 is a positive integer such that the order of every element of @xmath8 does not exceed @xmath10 , then @xmath62 our proof of theorem [ t : isoper ] is a variation of the argument used in ( * ? ? ?
* theorem 5 ) where a slightly weaker estimate is established under the stronger assumption that _ all elements of @xmath9 _ have order at most @xmath10 . in the special particular case where @xmath9 is a homocyclic group of exponent @xmath10 , and
@xmath63 is a standard generating subset , theorem [ t : isoper ] gives a result of bollobs and leader ( * ? ? ?
* theorem 8) .
the estimate of theorem [ t : isoper ] , in general , fails to hold for non - abelian groups .
for instance , if @xmath9 is the symmetric group of order @xmath64 , and @xmath63 consists of two involutions , then the cayley graph induced on @xmath9 by @xmath8 is a bi - directional cycle of length @xmath65 .
consequently , for a non - empty proper subset @xmath66 inducing a path in this cycle , one has @xmath67 ( as @xmath68 for @xmath69 and @xmath70 for @xmath71 ) .
it would be interesting to investigate the sharpness of the estimate of theorem [ t : isoper ] and to determine whether the function @xmath2 in its right - hand side can be replaced with a larger function .
back to the class @xmath33 , from theorem [ t : max - f ] we derive the following result showing that , somewhat surprisingly , any function satisfying must actually satisfy a stronger inequality .
[ t : strong ] for any function @xmath37 , and any @xmath72 $ ] , we have @xmath73 substituting @xmath44 , @xmath45 , and @xmath74 into the inequality of theorem [ t : strong ] ,
we conclude that the factor @xmath75 in the right - hand side is optimal , and the function @xmath2 can not be replaced with a larger function . as shown in (
* theorem 6 ) , for each @xmath47 , the functions @xmath76 are well - defined and belong themselves to the classes @xmath48 , and ( * ? ? ?
* theorem 5 ) gives a lower bound for the edge - boundary @xmath59 in terms of these functions . in this context
it is natural to investigate the infimum @xmath77 .
it is somewhat surprising that this infimum can be found explicitly , even though the individual functions @xmath78 are known for @xmath79 only ( see ( * ? ? ? * theorem 7 ) ) .
[ t : inf - fm ] we have @xmath80 as mentioned above , ( * ? ? ?
* lemma 3 ) says that every concave function from the class @xmath36 belongs to all the classes @xmath48 . for the proof of theorem [ t : inf - fm ]
we need the converse assertion , which turns out to be true even with the concavity assumption dropped .
[ l : convl3 ] if @xmath81 for all @xmath47 , then @xmath43 .
clearly , every convex function lies in the class @xmath33 .
the last result of our paper shows that all `` sufficiently flat '' _ concave _ functions also lie in @xmath33 ; this complements in a sense theorem [ t : max - f ] which implies that every concave function from the class @xmath33 is `` flat '' .
let @xmath82 be the class of all functions , defined and concave on the interval @xmath4 $ ] and vanishing at the endpoints of this interval .
[ t : flat ] if @xmath83 and @xmath84 for all @xmath17 $ ] , then @xmath43 .
moreover , the function @xmath85 is best possible here in the following sense : if @xmath86 has the property that for any function @xmath83 with @xmath87 we have @xmath43 , then @xmath88 for all @xmath17 $ ] .
we now turn to the proofs of the results discussed above .
we prove theorems [ t : max - f ] and [ t : isoper ] in sections [ s : proof - of - t1 ] and [ s : proof - of - t2 ] , respectively , and the proofs of theorems [ t : strong ] , [ t : inf - fm ] , and [ t : flat ] and lemma [ l : convl3 ] are presented in section [ s : more - proofs ] .
concluding remarks are gathered in section [ s : remarks ] .
first , we show that for any function @xmath43 , and any real @xmath17 $ ] and integer @xmath20 , one has @xmath89 ; this will prove the second assertion of the theorem . by symmetry
, @xmath90 can be assumed without loss of generality . applying with @xmath91 and @xmath45 ,
we get @xmath92 that is , @xmath93 iterating , we obtain @xmath94 whence , substituting @xmath95 , @xmath96 as wanted .
we now prove that @xmath2 satisfies , and hence @xmath41 . the proof is based on the following lemma showing that if a concave and continuous function satisfies whenever @xmath97 , then it actually satisfies for all @xmath98 $ ] .
[ l : redend ] suppose that the function @xmath3 is concave and continuous on the interval @xmath4 $ ] . in order for to hold for all @xmath99 $ ]
, it is sufficient that it holds for all @xmath100 $ ] and @xmath97 . to avoid interrupting the flow of exposition ,
we proceed with the proof of theorem [ t : max - f ] , postponing the proof of lemma [ l : redend ] to the end of this section .
as lemma [ l : redend ] shows , it suffices to establish with @xmath74 and @xmath97 .
indeed , the case @xmath101 reduces easily to that where @xmath45 using the symmetry of @xmath2 around the point @xmath102 .
thus , @xmath45 can be assumed , and in view of @xmath103 , renaming the remaining variable , we have to prove that @xmath104.\ ] ] the situation where @xmath105 is immediate , and we therefore assume that @xmath106 .
addressing first the case @xmath90 , we find @xmath20 such that @xmath22 , and notice that @xmath107 by the definition of the function @xmath2 ; consequently , it suffices to prove that @xmath108 dividing through by @xmath16 , substituting @xmath109 , and rearranging the terms gives this inequality the shape @xmath110 and to complete the proof it remains to notice that the left - hand side factors as @xmath111 the case @xmath112 reduces to that where @xmath113 as follows .
let @xmath114 , so that @xmath115 .
assuming that fails to hold , we get @xmath116 and also @xmath117 which , in view of @xmath118 , jointly yield @xmath119 since @xmath120 is a decreasing function of @xmath121 ( which is obvious for @xmath122 $ ] , and follows directly from the definition of the function @xmath2 for @xmath123 $ ] ) , we conclude that @xmath124 .
thus , @xmath125 , and recalling that @xmath126 , by what we have shown above , @xmath127 hence , from the assumption that is false , @xmath128 which can be written as @xmath129 contradicting .
it remains to prove lemma [ l : redend ] .
the proof uses the well - known fact that a strictly concave function is unimodal ; the specific version we need here is that if @xmath3 is continuous and strictly concave on the closed interval @xmath130 $ ] , then either it is strictly monotonic on @xmath130 $ ] , or there exists @xmath131 such that @xmath3 is strictly increasing on @xmath132 $ ] and strictly decreasing on @xmath133 $ ] . as a result , the minimum of a function , strictly concave on a closed interval , is attained at one of the endpoints of the interval .
we also need [ m : conv - inv ] suppose that the function @xmath3 is defined and strictly concave on a closed interval @xmath130 $ ] .
if @xmath3 is monotonically increasing on @xmath130 $ ] , then its inverse is strictly convex .
if @xmath3 is monotonically decreasing on @xmath130 $ ] , then its inverse is strictly concave .
suppose first that @xmath3 is _ strictly _ concave on @xmath4 $ ] , and assume that holds for all @xmath134 $ ] and @xmath97 ; hance , by symmetry , also for all @xmath135 $ ] and @xmath136 .
given @xmath98 $ ] with @xmath137 , let @xmath138 , and consider the auxiliary function @xmath139 ( depending on @xmath30 and @xmath31 ) .
furthermore , set @xmath140 and @xmath141\}$ ] .
notice , that @xmath142 , in conjunction with the strict concavity of @xmath143 , implies @xmath144 . for any @xmath32 $ ] we have then @xmath145 and therefore it suffices to show that @xmath146 for any @xmath147 and with @xmath148 and @xmath149 defined as above .
we now put the reasoning onto its head .
suppose that a real @xmath14 is fixed so that , if @xmath150 and @xmath151\}$ ] , then @xmath152 . let @xmath153 be defined by @xmath154 . by the intermediate value property and monotonicity of @xmath143 on each of the intervals
@xmath155 $ ] and @xmath156 $ ] , to any given @xmath157 with @xmath158 there corresponds then a unique pair @xmath159 with @xmath160 and @xmath161 . as the above argument shows , to complete the proof ( under the _ strict _ concavity assumption ) it suffices to establish with @xmath30 and @xmath31 understood as functions of the variable @xmath157 ranging from @xmath162 to @xmath163 .
since @xmath30 and @xmath31 are actually inverses of the function @xmath143 restricted to the appropriate intervals , by claim [ m : conv - inv ] , @xmath30 is convex and continuous , and @xmath31 is concave and continuous , so that @xmath164 is concave and continuous and consequently , will follow once we obtain it for @xmath165 and also for @xmath166 . the latter case ( with equality sign )
is immediate from @xmath167 . for the former case
, we notice that if @xmath165 , then @xmath168 , whence , having @xmath169 $ ] defined by @xmath170 , we get @xmath171 by the assumption of the lemma ( and the remark at the very beginning of the proof ) .
finally , suppose that @xmath3 is concave but , perhaps , not _
strictly _ concave on @xmath4 $ ] . for @xmath172
let @xmath173 and define @xmath174 a straightforward computations confirms that @xmath175 consequently , if @xmath176 when @xmath100 $ ] and @xmath97 , then also @xmath177 under the same assumptions .
since @xmath178 is _ strictly _ concave ( as it is easy to verify ) , we conclude that @xmath177 actually holds for all @xmath72 $ ] .
now shows that @xmath179 for all @xmath72 $ ] , and since @xmath180 can be chosen arbitrarily small , we have indeed @xmath176 .
we use induction on @xmath181 . without loss of generality , we assume that @xmath8 is a minimal ( under inclusion ) generating subset .
fix an element @xmath182 and write @xmath183 .
if @xmath184 , then @xmath9 is cyclic with @xmath181 being equal to the order of @xmath185 , whence @xmath186 and the assertion follows from @xmath187 .
assuming now that @xmath188 , let @xmath189 be the subgroup of @xmath9 , generated by @xmath190 ; thus , @xmath189 is proper and non - trivial .
since the quotient group @xmath191 is cyclic , generated by @xmath192 , its order @xmath193 does not exceed @xmath10 .
for @xmath194 set @xmath195 and @xmath196 .
fix @xmath197 $ ] . by the induction hypothesis ( as applied to the subset @xmath198 of the group @xmath189 with the generating subset @xmath190 ) , the number of edges from an element of @xmath199 to an element of @xmath200 is at least @xmath201 .
furthermore , the number of edges from @xmath199 to @xmath202 is at least @xmath203 ( where @xmath204 is to be replaced with @xmath30 for @xmath205 ) .
it follows that @xmath206 choose @xmath207 $ ] so that @xmath208 is the smallest , and @xmath209 the largest of the numbers @xmath210 . from the triangle inequality , @xmath211 whence @xmath212 as wanted .
fix @xmath98 $ ] with @xmath137 and consider the function @xmath213 defined by @xmath214.\ ] ] clearly , we have @xmath215 , and for any @xmath216 $ ] with @xmath217 , using the fact that @xmath37 , we get @xmath218 hence , @xmath219 , and therefore @xmath220 by theorem [ t : max - f ] ; that is , @xmath221 for all @xmath72 $ ] with @xmath137 . aiming at a contradiction , suppose that @xmath222 , but @xmath223 . by the latter assumption , there exist @xmath99 $ ] with @xmath224 such that @xmath225 clearly , we have @xmath226 .
this implies @xmath227 , whence @xmath3 is continuous at @xmath16 by ( * ? ? ?
* lemma 4 ) ( which says that all functions from the classes @xmath48 are continuous on @xmath228 ) .
it follows that there is a _ rational _ @xmath226 for which holds true ; say , @xmath229 with integer @xmath230 .
now can be written as @xmath231 contradicting the assumption @xmath81 .
let @xmath232 ; our goal , therefore , is to show that @xmath233 . to begin with , we prove that @xmath234 to this end , we fix @xmath235 and @xmath50 $ ] with @xmath51 and @xmath52 , and show that @xmath236 as shown in @xcite , if @xmath14 and @xmath237 are integers with @xmath238 , then @xmath239 , and hence @xmath240 .
it follows that @xmath241;\ ] ] thus , there is an integer @xmath242 such that @xmath243 for each @xmath244 $ ] . since @xmath245 in view of @xmath246
, we then get @xmath247 establishing , and therefore . to complete the proof we notice that and lemma [ l : convl3 ] yield @xmath248 , whence , by theorem [ t : max - f ] , @xmath249 on the other hand , since @xmath250 is concave , by ( * ?
? ? * lemma 3 ) we have @xmath54 for every @xmath47 .
it follows that @xmath251 for every @xmath252 , implying @xmath253 comparing and , we get the assertion . for the first assertion of the theorem
we show that holds true , provided that @xmath83 and @xmath254 $ ] . using symmetry and disposing of the trivial cases ,
we assume that @xmath137 and @xmath255 . furthermore ,
letting @xmath256 , we rewrite the inequality to prove as @xmath257 by concavity , we have @xmath258 ( as the point @xmath259 lies above the segment joining the points @xmath260 and @xmath261 , and @xmath262 ( as @xmath263 is above the segment joining @xmath261 and @xmath264 ) .
also , by the assumptions , @xmath265 comparing with , we see that it suffices to prove that @xmath266 assuming for a contradiction that this is wrong , after tedious , but routine algebraic manipulations we then derive @xmath267 multiplying the first of these inequalities by @xmath268 and the second by @xmath269 , and adding up the resulting estimates and the third inequality , we get @xmath270 it is easily verified that this simplifies to @xmath271 and further to @xmath272 which can not hold since the left - hand side is a square . to prove the second assertion ,
suppose that @xmath273 is a concave function on @xmath4 $ ] with @xmath274 and @xmath275 for some @xmath276 , and let @xmath277 thus , @xmath3 is a concave function on @xmath4 $ ] with @xmath278 , and @xmath87 by concavity of @xmath273 .
we show that , on the other hand , @xmath223 , and indeed , assuming for definiteness @xmath279 , that @xmath280 to this end we just plug in the definition of @xmath3 and rewrite this inequality as @xmath281 which is equivalent to the assumption @xmath275 .
it seems natural to extend the class @xmath33 by considering , for every finite closed interval @xmath282 and every constant @xmath283 , the class @xmath284 of those real - valued functions @xmath3 defined on @xmath282 and satisfying @xmath285 for all @xmath286 and @xmath32 $ ] .
this , however , does not lead to any principally new results , as one has @xmath287 if and only if @xmath288 , where @xmath213 is a linear bijection of @xmath4 $ ] onto @xmath282 ; that is , @xmath284 is obtained from @xmath33 by a simple linear scaling . in particular
, @xmath289 is the largest function of the subclass of all functions from @xmath284 , non - positive at the endpoints of @xmath282 . also , as a corollary of theorem [ t : strong ] , if @xmath287 , then @xmath290 for all @xmath286 and @xmath32 $ ] .
in contrast , it might be interesting to extend the results of this paper , and in particular theorem [ t : max - f ] , onto the class of all @xmath28-convex functions on a given closed interval , for every fixed @xmath291 .
normalization reduces this to studying the class @xmath292 of all real - valued functions on @xmath4 $ ] , satisfying the boundary condition and the inequality @xmath293 for all @xmath72 $ ] .
it is not difficult to see that the function @xmath294 is well defined and lies itself in the class @xmath295 , and that for any @xmath296 we have @xmath297 for all @xmath72 $ ] ; moreover , @xmath298 is the largest function with this property . in view of theorem [ t : max - f ] , one can expect that , perhaps , an explicit expression for the functions @xmath298 can be found .
the author is grateful to eugen ionescu for his interest and useful discussions . | we study extremal properties of the function @xmath0,\ ] ] where @xmath1 . in particular , we show that @xmath2 is the pointwise largest function of the class of all real - valued functions @xmath3 defined on the interval @xmath4 $ ] , and satisfying the relaxed convexity condition @xmath5\ ] ] and the boundary condition @xmath6 . as an application
, we prove that if @xmath7 and @xmath8 are subsets of a finite abelian group @xmath9 , such that @xmath8 is generating and all of its elements have order at most @xmath10 , then the number of edges from @xmath7 to its complement @xmath11 in the directed cayley graph induced by @xmath8 on @xmath9 is @xmath12 |
the conventional wisdom based on the standard theory of structure formation says that the universe was reionized sometime in the redshift range @xmath8 ( barkana & loeb 2001 ) .
the relatively large uncertain range in redshift reflects our imperfect knowledge of the density fluctuations on small scales , star formation processes ( e.g. , efficiency , imf , etc . ) and feedback processes at high redshift .
the latest observations of high redshift quasars ( e.g. , fan et al .
2001 ) are beginning to probe the lower bound of that window and suggestions have been made that we may be witnessing the end phase of the cosmological reionization process at @xmath9 , based solely on the appearance of a precipitous drop of transmitted flux at the rest - frame ly@xmath10 wavelength near @xmath9 in a single quasar spectrum ( becker et al .
2001 , hereafter b01 ; barkana 2001 ) . in this _
we present an analysis of the ionizing background radiation field in the redshift range @xmath11 , using the absorption measurements in b01 . combining with previous data at lower redshift
, we find that , coming from high redshift , the ionizing radiation intensity displays a sharp rise at @xmath9 peaking at @xmath12 , a significant downturn from @xmath13 to @xmath5 by a factor of @xmath14 , and subsequently a consistent ascent from @xmath15 to @xmath16 by a factor of @xmath17 .
all these three features are consistent with a picture that the cosmological reionization was near completion at @xmath9 . the first feature ( i.e. , the initial sharp rise ) has been predicted by several authors previously ( cen & ostriker 1993 ; gnedin 2000a ; miralda - escud , haehnelt , & rees 2001 ) . at present , the primary uncertainty for its determination observationally is the possibility that the high level of absorption in the single observed quasar at @xmath18 is some kind of anomaly ( we show that it can not be a simple statistical fluctuation in the absorption level ) .
additional observed quasars at @xmath19 will be critical in this regard .
here we focus our attention on the relatively more robust measurements of the ionizing radiation field at @xmath20 and show that the observed pause in the rise of the amplitude of the ionizing radiation field from @xmath13 to @xmath21 could be due to suppression of star formation following reionization at @xmath9 .
assuming the model in which the igm is almost completely ionized and the radiation background is uniform , we infer the evolution of the ionization rate by requiring that the mean flux decrement in simulated spectra matches the observed values in the upper panel of figure 2 of b01 .
this method was pioneered by rauch et al .
( 1997 ) and extended in mcdonald et al .
( 2000,2001 ) and mcdonald & miralda - escud ( 2001 ) .
we apply the standard procedure using a hydrodynamic simulation of a flat universe dominated by a cosmological constant and cold dark matter ( @xmath22cdm ) , with cdm density @xmath23 , hubble parameter @xmath24 ( @xmath25 ) , baryon density @xmath26 , power spectrum normalization @xmath27 , and large - scale primordial power spectrum slope @xmath28 . the simulation is eulerian , with box size @xmath29 divided into @xmath30 cells ( see cen et al .
2001 for a more complete description ) .
we use outputs from the simulation at @xmath31 and @xmath32 , interpolating linearly between them when necessary .
we start by assuming the model in the simulation correctly represents the universe , and then consider whether or not our results are self - consistent .
figure [ gev ] shows the results for the evolution of @xmath33 , the hydrogen ionization rate in units of @xmath34 .
the open squares are the old results from mcdonald & miralda - escud ( 2001 ) .
the filled squares are the new results , leaving out the highest redshift point , which we analyze differently than the others because the detected transmission is not significant and the detector noise needs to be taken into account carefully ( this last point is represented by the triangle with 1 and 2 @xmath35 upper limit bars ) .
the equally spaced bins in redshift are chosen to coincide with the majority of the points in figure 2 of b01 .
as discussed below , the results have been rescaled to apply for the value of the baryon density favored by big bang nucleosynthesis , @xmath36 ( burles & tytler 1998 ) , instead of the value in the simulation , @xmath37 . in detail
, we determine a value of @xmath38 and its error for each b01 point using the simulation as follows : first , we construct the optical depth field , @xmath39 ( i.e. , the spectrum is @xmath40 $ ] ) along a large number of lines of sight through the simulation cube ( @xmath41 for each face ) , normalized using an arbitrary value of @xmath38 ( i.e. , @xmath42 for low density gas in ionization equilibrium ) .
we then randomly choose many sets of five @xmath43-long segments of spectrum , so that each set forms a combined spectrum of the length of our redshift bins , @xmath44 ( note that this length depends slightly on cosmology and redshift , i.e. , the spectra should be shorter by 26% at @xmath32 in our model , corresponding to a 13% increase in the estimated error bars , which we account for in the plot ) . for each @xmath45 chunk ,
we determine the value of @xmath38 that produces the observed value of the mean transmitted flux fraction , @xmath46 [ @xmath47 in the notation of b01 ] .
the mean of the @xmath38 values for all the chunks is the inferred value of @xmath38 for this observed data point , and the dispersion is the estimated error .
most redshift bins in b01 have several measured values of @xmath48 in them , so we combine the inferred values of @xmath38 within the bin using an error weighted average which takes into account fractional overlap when necessary ( because of the overlapping points , our error bars are correlated , in the sense that point - to - point differences are more significant than the error bars seem to indicate ) . to test that the treatment of @xmath43 segments as independent does not underestimate the error by ignoring correlation along the lines of sight on larger scales , we repeat this procedure using @xmath49 segments , and find almost identical error bars .
we now turn to the highest redshift point in b01 . the given value and error are @xmath50 ; however , b01 state that the sky subtraction has an uncertainty of order 1@xmath35 , so we use @xmath51 when interpreting this point .
b01 also give the measured transmitted flux fraction in the region of the spectrum corresponding to ly@xmath52 absorption by gas at this redshift , @xmath53 , and point out that the implied constraint on the ly@xmath10 optical depth is actually stronger than the direct one .
we assume the same sky subtraction error for this point as for the ly@xmath10 region .
we start by analyzing the ly@xmath10 region .
we use the method described above to create chunks of spectra of the appropriate length . for each value of @xmath38
we compute @xmath48 for many different chunks , and compute the mean likelihood of producing @xmath54 given the error @xmath55 . the relative likelihood function peaks at @xmath56 , but does not decrease significantly for arbitrarily small @xmath38 .
the likelihood has dropped by factors of 0.61 and 0.14 at @xmath57 and 0.135 , respectively , so we call these the 1 and 2@xmath35 upper limits .
we analyze the ly@xmath52 region in two ways .
first we repeat the above procedure assuming a constant factor 0.13 suppression of the continuum by ly@xmath10 absorption corresponding to gas at @xmath58 ( see the lowest redshift point in b01 ) .
the inferred 1 and 2@xmath35 upper limits on @xmath38 are 0.032 and 0.056 .
these constraints are stronger than the ones from the ly@xmath10 region ; however , they are not as realistic as possible because they ignore fluctuations in the ly@xmath10 absorption in the ly@xmath52 region , which might either weaken the constraints , by completely obscuring substantial portions of the ly@xmath52 region , or strengthen the constraints , by providing regions of significantly less than average absorption where residual transmission through the ly@xmath52 forest could be observed .
we simulate this effect by creating ly@xmath10 absorption spectra to serve as the continua for the ly@xmath52 region , using the @xmath31 simulation output and our inferred value of @xmath38 from the lowest redshift b01 point .
this procedure results in essentially identical constraints to the assumption of a flat continuum : @xmath59 and 0.062 at 1 and 2@xmath35 , respectively . it may be possible to obtain even stronger constraints by using ly@xmath60 , as suggested by b01 ( x. fan , private communication ) , or by considering the limits imposed by individual pixel values ( d. weinberg , private communication ) .
however , if reionization is on - going or just completed at @xmath61 , the detailed constraints on @xmath38 obtained from any kind of analysis based on a simulation in which reionization happened much earlier probably are not very meaningful ( @xmath38 is not well defined if the radiation background is inhomogeneous ) . our formal constraints on @xmath38 near @xmath32
should be taken simply as an indication that something dramatic is happening at this redshift and we think it likely indicates the end of the cosmological reionization process .
the overall amplitude of our @xmath0 results ( but not the evolution over a short redshift interval ) is sensitive to the cosmological model and gas temperature - density relation in the simulation , as one can see from the equation for optical depth at a point in real space ( i.e. , ignoring peculiar velocities and thermal broadening ) with baryon overdensity @xmath62 and gas temperature @xmath63 : @xmath64 where @xmath65 is the hubble parameter at redshift @xmath66 [ @xmath67 at high @xmath66 in a @xmath22cdm universe ] .
sensitivity of the inferred value of @xmath0 to the power spectrum of density perturbations enters through the @xmath68 term , as well as the peculiar velocities , and additional sensitivity to temperature can arise from the thermal broadening .
the effect of changing @xmath69 or @xmath65 on the inferred value of @xmath0 is easy to correct for , but the effect of changing the temperature - density relation or the input power spectrum generally must be investigated by running additional simulations .
we do this using approximate hydro - pm simulations ( hereafter hpm , gnedin & hui 1998 ) , after checking that they give similar results to our fully hydrodynamic simulation ( to @xmath70% in @xmath0 ) when performed using the same model and initial conditions .
first , we analyze hpm simulations with a variety of different temperature - density relations and find that the thermal broadening is not very important to the derivation of @xmath0 , so the dependence @xmath71 is accurate ( to @xmath72 in the power law index ) , as long as the temperature is specified at the most relevant density for the transmission features that determine @xmath0 , which we find to be @xmath73 .
the temperature in our hydrodynamic simulation at this density is @xmath74 k , but in the aftermath of reionization the temperature of the gas would be higher than this ( e.g. , miralda - escud & rees 1994 ) , requiring a correction .
the amplitude of the initial density fluctuations in our hydrodynamic simulation is higher by a factor @xmath75 on the scale of the ly@xmath10 forest than the measurement of mcdonald et al .
( 2000 ) , which was independently confirmed by croft et al .
( 2001 ) . by varying the amplitude , @xmath76 , in hpm simulations ,
we determine that @xmath77 ( roughly ) .
the required ionizing background decreases when @xmath76 increases because the voids where transmission is observed become deeper ( i.e. , @xmath78 becomes smaller ) . in the end , the corrections for power spectrum amplitude and temperature may roughly cancel ( if we use @xmath79 k ) , so the points in figure [ gev ] should be reasonably accurate ; however , we emphasize that the temperature at this redshift is unknown , so significant uncertainty in @xmath0 remains .
the reader may be wondering at this point `` could the inferred decrease in @xmath38 from @xmath4 to @xmath80 possibly be real '' ?
it appears to be a 3@xmath35 effect and we think it could be real .
the same trend is clearly visible to the eye in the bottom panel of figure 2 of b01 , once we consider that the value of their @xmath81 is generally expected to increase with time .
the signal is almost equally strong in the spectra of two different quasars .
these two are at almost identical redshift so a continuum - related effect might be the same in both ; however , the observed @xmath82% increase in transmission fraction between @xmath80 and @xmath4 would require an increase @xmath83 in the quasar continuum .
finally , we have every reason to believe that our statistical error calculation is correct .
the scatter in the inferred values of @xmath38 for different b01 points in each redshift bin is in perfect agreement with our estimated error bars ( e.g. , @xmath84 for 10 degrees of freedom when we treat the three average values of @xmath38 for the bins in the range @xmath85 as fitting parameters to be estimated from the individual points with error bars ) .
note that , while it may look small to the eye , the fractional error on the @xmath86 point is actually larger than the fractional errors on the following two points .
purely adiabatic temperature evolution @xmath87 , which constitutes the steepest possible evolution in temperature ( by ignoring the subsequent photo - heating ) , instead of the essentially constant temperature in the simulation , would only cause a 7.5% relative change in the values of the @xmath0s .
we think that the sharp rise of the radiation intensity at @xmath7 may be a fairly strong case for the completion of cosmological reionization at that epoch .
however , note that a value of @xmath88 is not inconsistent in principle with a fully ionized igm , since the neutral fraction near the mean density is still only @xmath89 .
one is hampered by the high optical depth of an igm that is still mostly ionized .
it will be extremely valuable to look for additional signatures of cosmological reionization in its aftermath , where more flux is transmitted .
we will therefore focus our discussion on the radiation field at @xmath90 .
in the standard hierarchical structure formation theory smaller structures began to form earlier than larger structures , simply because the amplitude of density fluctuations is a decreasing function of scale and in linear theory density fluctuations on all scales grow at the same rate . under this general picture the universe is conventionally thought to be reionized mostly by photons from stellar systems more massive than @xmath91 , where atomic line cooling provides an efficient energy sink for gas that is collected and heated during gravitational collapse of halos ( haiman , rees & loeb 1997 ; gnedin & ostriker 1997 ) . less massive systems , after having produced an insufficient amount of stars ( pop iii ) to reionize the universe at an earlier epoch , can no longer form stars due to lack of cooling processes ( haiman , thoul , & loeb 1996 ; haiman , rees & loeb 1997 ; tegmark et al .
1997 ) .
after reionization , the temperature of the intergalactic gas is raised to @xmath92k ( cen & ostriker 1993 ; miralda - escud & rees 1994 ; gnedin & ostriker 1997 ) . the subsequent effect of suppression of gas accretion onto small halos has been noted by many authors ( efstathiou 1992 ; thoul & weinberg 1996 ; quinn , katz , & efstathiou 1996 ; kepner , babul , & spergel 1997 ; navarro & steinmetz 1997 ; kitayama & ikeuchi 2000 ; gnedin 2000b ) .
the extent of this effect somewhat varies among the studies and it is clear that our current treatment of the reionization process is far from ideal . obviously , a sufficiently adequate study would require simulations that include an accurate treatment of three - dimensional radiative transfer process and have high enough spatial and mass resolution .
we do not have simulations of this caliber at present and attempt to only make a semi - quantitative calculation to illustrate a consequence of reionization on star formation and the background radiation field , which we believe captures a primary feature of this process in a simple and intuitive fashion .
figure [ nl ] shows the nonlinear mass @xmath93 and linear jeans mass @xmath94 as a function of redshift . the nonlinear mass @xmath93 is the mass within a top - hat window at which the linear rms density fluctuation is unity , and @xmath95 ( peebles 1993 ) , where @xmath63 is the gas temperature and @xmath96 is the present matter density parameter .
the assumption of instantaneous reionization is just for the sake of plotting convenience and is not required to make our point . in a real universe the reionization phase is complicated and is thought to progress on a time scale of a hubble time , during which the mean ( volume weighted ) radiation field builds up slowly up to a value of approximately @xmath97erg/@xmath98/hz / sec / sr at lyman limit , followed by a brief phase , when the majority of the baryons are ionized and a sudden jump in the amplitude of the mean radiation field intensity at lyman limit to @xmath99erg/@xmath98/hz / sec / sr occurs within a redshift interval of a fraction of unity ( miralda - escud et al . 2000 ; gnedin 2000a ) .
we simply call this epoch of a sudden rise in radiation field the reionization epoch , marking the completion of the reionization process .
we see , in figure 2 , that gas is able to accrete onto halos with mass greater than the jeans mass @xmath100 prior to @xmath9 .
stars formed inside halos with @xmath101 are presumably primarily responsible for producing most of the ionizing photons immediately prior to @xmath9 , where atomic line cooling provides an efficient cooling process .
right after reionization , there is a dramatic inversion of the situation from @xmath9 to @xmath102 , where jeans mass surpasses the nonlinear mass .
such a switch has some interesting consequences : gas accretion onto the most abundant halos ( about the nonlinear mass ) is suddenly suppressed , simply because the ambient gas has a temperature that is higher than the virial temperature of the typical halos and is incapable of cooling to a lower temperature .
gas accretion and thus star formation may be brought to a halt in typical halos ( larger halos with higher virial temperature may still be able to form stars with a somewhat reduced rate due to the same effect ) .
below @xmath102 , nonlinear mass again becomes larger than the jeans mass and gas is once again capable of accreting onto typical halos with mass @xmath103 , and star formation rate bounces back .
we note that the cosmological model adopted may not be correct in fine details and is used only to illustrate the process .
therefore , the exact crossing redshift of the two curves at @xmath104 should not be taken at its face value , but rather as a generic feature .
let us now relate the star formation rate to the ionizing background radiation field .
our treatment is intended to be simple but we think it contains the essence of the relevant physical processes . for gas immersed in a meta - galactic ionizing radiation background ,
locally hydrogen photoionization is balanced by hydrogen recombination ( to levels @xmath105 or higher ) : @xmath106 where @xmath0 is the photoionization rate ( see figure 1 ) , @xmath107 , @xmath108 and @xmath109 are the neutral hydrogen , ionized hydrogen , and electron number densities , respectively , and @xmath110 is the hydrogen recombination rate .
averaging spatially both sides of equation ( 1 ) gives @xmath111 where @xmath112 is the clumping factor of all gas participating in receiving ionizing photons ( i.e. , excluding gas in optically shielded regions ) .
for simplicity we have assumed that the gas is composed entirely of hydrogen .
since the neutral hydrogen fraction is much less than unity , we have simply replaced both the ionized hydrogen and electron number densities by the total hydrogen number density @xmath113 .
the mean free path of an ionizing photon in comoving length units is then @xmath114 where @xmath115 is the hydrogen photoionization cross - section .
using @xmath116 , @xmath117sec@xmath118 ( see figure 1 ) and @xmath119@xmath98 we obtain @xmath120 in comoving megaparsecs : @xmath121 evidently , at @xmath122 , the mean free path of an ionizing photon is much smaller than the hubble radius and radiative processes may be treated locally " , by ignoring cosmological effects .
but the mean free path is much greater than the typical separation between ionizing sources ( @xmath123comoving mpc ) .
in this case we may relate the mean _ comoving _ specific emissivity of ionizing radiation @xmath124 ( in units of erg / sec / hz / comoving @xmath125 ) to the mean ionizing radiation intensity @xmath126 ( in units of erg / sec / hz / sr/@xmath98 ) by @xmath127 the evolution of @xmath128 is complicated and currently undetermined without detailed simulations that ionized the universe at @xmath9 .
however , simple estimates may be made to gauge the range of variation . in the linear regime
we have @xmath129 and in the extreme nonlinear regime @xmath130 ; the actual situation is likely to be between these two limiting cases , parameterized as @xmath131 with @xmath132 . inserting @xmath128 into equation ( 5 ) yields @xmath133 with @xmath132 .
clearly , in order for @xmath134 to increase with redshift , as indicated by observations in the redshift range @xmath135 as shown in figure 1 , the comoving specific emissivity or comoving star formation rate has to increase with redshift . taking the values shown in figure 1 in the range @xmath136 , we obtain @xmath137 .
( this fit used the 3rd through 5th filled squares in figure [ gev ] .
we obtain a flatter slope , @xmath138 , with an acceptable @xmath139 , if we add the 2nd point , at @xmath140 . )
this requires a decrease in star formation rate from @xmath12 to @xmath141 as @xmath142 .
our analysis is in qualitative agreement with barkana & loeb ( 2000 ) , who pointed out that the decrease in star formation subsequent to cosmological reionization may be detectable in the evolution of number counts of faint galaxies by ngst
. it may be that we have detected this signature in the evolution @xmath134 or @xmath0 .
in hierarchical structure formation theory the evolution of the meta - galactic ionizing radiation background may be characterized by five distinct phases : 1 ) from @xmath143 to @xmath144 ( which marks the end of reionization ) the radiation field builds up slowly before individual ionizing hii regions overlap ( pre - overlap period " ; gnedin 2000a ) 2 ) in a brief period up to @xmath144 the majority of baryons are ionized in a short time scale and the radiation field jumps up by about two to three orders of magnitude , completing the reionization process ( reionization epoch " ) ; 3 ) from @xmath144 to @xmath145 the radiation field pulls back significantly due to a decrease in star formation rate as a result of reionization ( we call this aftermath of reionization " ) ; 4 ) @xmath145 to @xmath146 the radiation field rises steadily with contributions from both stars in larger galaxies and quasars ; 5 ) @xmath146 to @xmath147 the radiation field drops off sharply due to the combined effect of decreasing star / quasar formation rate and cosmological effects .
consistent with theoretical calculations , observations have confirmed phases ( 4 ) ( e.g. , mcdonald & miralda - escud 2001 ; shown in figure 1 of this paper ) and ( 5 ) ( e.g. , shull et al .
1999 ) .
in this paper we have interpreted the absorption in the spectra of the latest high redshift quasars in the redshift range @xmath11 , searching for salient features pertaining to the reionization epoch " ( phase 2 ) and the aftermath of reionization " ( phase 3 ) .
quite intriguingly , we appear to be seeing the aftermath of reionization " and the end of the reionization epoch " in these spectra , both of which seem to point to a cosmological reionization redshift of @xmath148
. specifically , a predicted decrease of star formation rate in the aftermath of the reionization is indicated by the observations .
while a few more high redshift quasars are required to check that the complete absorption at @xmath61 is not some kind of anomaly ( e.g. , a rare leftover neutral patch in an almost completely ionized universe ) , it appears that our conclusion with regard to the aftermath of reionization " is more secure due to its independence of any single spectrum .
it is still very urgent to enlarge the observational data sets to greatly firm up the conclusion and sharpen up the rate of evolution .
if confirmed by future observations , this may be indication that our current understanding of galaxy formation may be approximately valid up to redshift @xmath9 .
furthermore , as suggested by barkana & loeb ( 2000 ) , it will be important to independently detect this aftermath of reionization " signature in the evolution of number counts of faint galaxies by ngst . since we appear to be observing the period directly after reionization
, it becomes more important to perform dedicated simulations that achieve reionization at the appropriate redshift @xmath9 .
this paper should be understood as an attempt to guide that effort by making some preliminary comparisons between data and an existing simulation in which reionization occurred somewhat earlier .
such future simulations are quite demanding , because not only high resolutions are required but also an accurate treatment of three - dimensional radiative transfer must be implemented .
barkana , r. , & loeb , a. 2000 , apj , 539 , 20 barkana , r. , & loeb , a. 2001 , phys .
, 349 , 125 barkana , r. , 2001 , astro - ph/0108431 becker , r.h . , et al .
2001 , astro - ph/0108097 burles , s. & tytler , d. 1998 , apj , 507 , 732 cen , r. , & ostriker , j.p .
1993 , apj , 417 , 404 cen , r. , ostriker , j.p .
, prochaska , j.x . , wolfe ,
a.m. 2001 , in preparation croft , r. a. c. , weinberg , d. h. , bolte , m. , burles , s. , hernquist , l. , katz , n. , kirkman , d. , & tytler , d. 2001 , apj , submitted ( astro - ph/0012324 ) djorgovski , s.g . ,
castro , s.m . , stern , d. , & mahabal , a.a .
2001 , astro - ph/0108069 efstathiou , g. 1992 , mnras , 256 , 43 fan , x. , et al .
2001 , astro - ph/0108063 gnedin , n.y .
2000a , apj , 535 , 530 gnedin , n.y . 2000b , apj , 542 , 535 gnedin , n. y. & hui , l. 1998 , mnras , 296 , 44 gnedin , n.y . , & ostriker , j.p .
1997 , apj , 486 , 581 haiman , z. , thoul , a.a .
, & loeb , a. 1996 , apj , 464 , 523 haiman , z. , rees , m.j . , & loeb , a. 1997 , apj , 476 , 458 kepner , j.v . , babul , a. , & spergel , d.n .
1997 , apj , 487 , 61 mcdonald , p. & miralda - escud , j. 2001 , apj , 549 , l11 mcdonald , p. , miralda - escud , j. , rauch , m. , sargent , w. l. w. , barlow , t. a. , cen , r. , & ostriker , j. p. 2000 , apj , 543 , 1 mcdonald , p. , miralda - escud , j. , rauch , m. , sargent , w. l. w. , barlow , t. a. , & cen , r. 2001 , apj , in press ( astro - ph/0005553 ) miralda - escud , j. , haehnelt , m. , & rees , m.j .
2000 , apj , 530 , 1 miralda - escud , j. & rees , m. j. 1994 , mnras , 266 , 343 kitayama , t. , & ikeuchi , s. 2000 , apj , 529 , 615 navarro , j.f . , & steinmetz , m. 1997 , apj , 478 , 13 rauch , m. , miralda - escud , j. , sargent , w. l. w. , barlow , t. a. , weinberg , d. h. , hernquist , l. , katz , n. , cen , r. , & ostriker , j. p. 1997
, apj , 489 , 7 tegmark , m. , silk , j. , rees , m.j . , blanchard , a. , abel , t. , & palla , f. 1997 , apj , 474 , 1 peebles , p.j.e .
1993 , principles of physical cosmology ( princeton : princeton university press ) shull , j.m . , et al . 1999 ,
aj , 118 , 1450 thoul , a.a . , & weinberg , d.h .
1996 , apj , 465 , 608 quinn , t. , katz , n. , & efstathiou , g. 1996 , mnras , 278 , l49 | the temporal evolution of the ionizing uv background radiation field at high redshift provides a probe of the evolution of the early star formation rate . by comparing the observed levels of absorption in the highest redshift quasar spectra to the predictions of a hydrodynamic simulation
, we determine the evolution of the photoionization rate ( @xmath0 ) for neutral hydrogen in the intergalactic medium , over the redshift range @xmath1 .
after accounting for sampling variance , we infer a sharp increase in @xmath0 from @xmath2 to @xmath3 , probably implying reionization at this redshift .
we find a _ decrease _ in @xmath0 from @xmath4 to @xmath5 , at @xmath6 significance .
this feature is a generic signature in the aftermath of reionization , entirely consistent with the cosmological reionization process being completed at @xmath7 .
this generic feature is a result of a significant change in the star formation rate subsequent to the cosmological reionization .
there is an abrupt rise of the temperature of the intergalactic medium due to photo - heating , when it is reionized .
this translates to a correspondingly sudden jump in the jeans mass and a dramatic suppression of gas accretion onto the most abundant ( sub - galactic ) halos at the epochs of interest .
the star formation rate suffers a temporary setback in the aftermath of reionization , resulting in a temporary decrease in the amplitude of the ionizing radiation field . |
the fractional diffusion equation has been treated by a number of authors ( see @xcite ) and , among the several applications that have been studied , mainardi @xcite studied the application to the theory of linear viscoelasticity .
+ the free boundary problems for the one - dimensional heat equation are problems linked to the processes of melting and freezing which have a latent heat - type condition at the interface connecting the velocity of the free boundary and the heat flux of the temperatures in both phases .
this kind of problems have been widely studied ( see @xcite ) . in this paper
, we deal with a two - phase lam - clapeyron - stefan problem for the time fractional diffusion equation , obtained from the standard diffusion equation by replacing the first order time - derivative by a fractional derivative of order @xmath2 in the caputo sense .
+ we use here the definition introduced by caputo in 1967 @xcite , and we will call it _ fractional derivative in the caputo sense _ , which is defined by @xmath3 where @xmath4 is the order of derivation , @xmath5 , @xmath6 is the gamma function defined by @xmath7 and @xmath8 is a differentiable function up to order @xmath9 in @xmath10 $ ] . + an interesting physical meaning of the fractional stefan problems
is discussed in @xcite and many authors were recently studying this kind of problems ( see .
some applications are linked to the behaviour in simulations of gas in polymer glasses ( @xcite ) or propagation in porous media ( @xcite ) . in @xcite
the classical lam - clapeyron - stefan problem was studied by using the fractional derivative of order @xmath11 .
+ in this paper we consider the following two - phase fractional lam - clapeyron - stefan problem @xmath12 where @xmath13 and @xmath14 , @xmath15(solid phase ) , @xmath16 ( liquid phase ) .
+ in this problem , the two governing diffusion equations ( [ st-2fases]-@xmath17 ) and ( [ st-2fases]-@xmath18 ) for @xmath19 and @xmath20 respectively , and the governing condition on the free boundary @xmath21 ( [ st-2fases]-@xmath22 ) include a fractional time derivative in the caputo sense of order @xmath0 .
the goal of this paper is to obtain an explicit solution of the free boundary problem ( [ st-2fases ] ) , called a generalized neumann solution with respect to the classical one given in @xcite , @xcite , @xcite .
this explicit solution is obtained through the wright and mainardi functions ( @xcite ) . in section
@xmath16 a summary of some properties related to these special functions are given which will be useful in the next section . in section @xmath23 the existence of a generalized neumann solution is given and an open problem for the uniqueness is posed .
moreover , the classical neumann solution for the two - phase lam - clapeyron - stefan problem for a semi - infinite material is well recovered by considering the limit when @xmath24 .
for every @xmath25 , @xmath26 and @xmath27 the wright function is defined by @xmath28 this function will play a fundamental role in this paper .
it is known that the wright function is an entire function if @xmath26 .
+ taking @xmath29 and @xmath30 , we get @xmath31 where @xmath32 is the mainardi function ( see @xcite ) , defined by @xmath33 which is a particular case of the wright function .
+ due to the uniform convergence of the series on compact sets , we have ( see @xcite ) @xmath34 then , for @xmath35 , and taking account that @xmath36 we have @xmath37 @xmath38 @xmath39 that is , @xmath40 where _ erf _ and _ erfc _ are the error and complementary error functions .
+ the next two propositions were proved in @xcite .
[ m pos y decrec ] if @xmath41 , then : 1 . @xmath42 is a positive and strictly decreasing positive function in @xmath43 such that @xmath44 ; 2 .
@xmath45 is a positive and strictly decreasing function in @xmath43 such that @xmath46 @xmath47 .
[ conv m al/2 cuando al tiende a 1]if @xmath35 and @xmath48 then : 1 .
@xmath49 ; 2 .
@xmath50=\frac{1}{\sqrt{\pi } } erf\left(\frac{x}{2}\right ) .$ ] due to the results in @xcite , the following assertions are true @xmath51 let us work on some problems in the first quadrant .
it is known that ( see @xcite ) @xmath52 is a solution for the fractional diffusion problem @xmath53 using this fact , it is easy to see that @xmath54 f_0 \ , d\xi\ ] ] is a solution for the fractional diffusion problem @xmath55 an equivalent expression of ( [ sol dfe 1er cuad ] ) is given by @xmath56 f_0 d\xi\ ] ] @xmath57\ ] ] @xmath58= f_0\left[1-w\left(-\frac{x}{\lambda t^{\frac{{\alpha}}{2}}},-\frac{{\alpha}}{2},1\right)\right].\ ] ] analogously we can check that @xmath59 is a solution for the fractional diffusion problem @xmath60
hereinafter we will call @xmath61 to the fractional derivative in the caputo sense of extreme @xmath62 , @xmath63 . + let us return to problem ( [ st-2fases ] ) . taking into account the previous section and the method developed in @xcite , the following explicit solution is obtained . an explicit solution for the two - phase lam - clapeyron - stefan problem ( [ st-2fases ] )
is given by @xmath64 where @xmath65 is a solution to the equation @xmath66 and the function @xmath67 is defined by @xmath68 with @xmath69 the following solution is proposed @xmath70\\ u_1(x , t)=c+d\left[1-w\left(-\frac{x}{\lambda_1 t^{{\alpha}/2}},-\frac{{\alpha}}{2},1\right)\right ] \\
s(t)=\xi \lambda_1 t^{{\alpha}/2}\end{array}\right.\ ] ] where @xmath71 and @xmath72 must be determined . + according with the results in the previous section and the linearity of the fractional derivative @xmath61 , functions @xmath20 and @xmath19 are solutions of the fractional diffusion equations ( [ st-2fases]-i ) and ( [ st-2fases]-ii ) respectively .
from conditions ( [ st-2fases]-@xmath73 ) and ( [ st-2fases]-@xmath74 ) we have , @xmath75=u_0\ ] ] @xmath76=u_m.\ ] ] and therefore we obtain : @xmath77 so , @xmath78 or equivalently @xmath79 taking into account the results in proposition [ m pos y decrec ] , ( [ u2a ] ) and ( [ u2b ] ) it is easy to see that @xmath80 from conditions ( [ st-2fases]-@xmath81 ) and ( [ st-2fases]-@xmath73 ) we have , @xmath82=c+d = u_i,\ ] ] @xmath83=u_m,\ ] ] and therefore we get : @xmath84 accordingly , @xmath85 or equivalently @xmath86.\ ] ] taking into account proposition [ m pos y decrec ] , ( [ u1a ] ) and ( [ u1b ] ) we obtain @xmath87 in order to determine @xmath88 , let us work with the `` fractional lam - clapeyronn - stefan condition '' ( [ st-2fases]-@xmath22 ) . from ( [ w ] ) and ( [ derivada de w ] ) we have @xmath89 which evaluated on @xmath90 , gives @xmath91 taking into account that ( @xcite ) @xmath92 it results that @xmath93 replacing ( [ cond st 1 ] ) and ( [ d^al s ] ) in the fractional condition ( [ st-2fases]-iii ) , we get for the unknown coefficient @xmath94 the following equation : @xmath95 @xmath96 @xmath97 @xmath98 @xmath99 that is , the equation ( [ eq xi ] ) holds , where @xmath100 , @xmath101 and @xmath102 where defined in ( [ f ] ) and ( [ f1 y f2 ] ) respectively .
+ in order to guarantee the existence of a solution of the equation ( [ eq xi ] ) , we will study the behavior of the functions @xmath100 , @xmath101 and @xmath102 . from proposition [ m pos
y decrec ] and ( [ lim w(-x ,- al/2,1)=0 y m=0 ] ) , it results that @xmath103 and @xmath104 let us prove that @xmath105 in @xcite the asymptotic expansion for @xmath106 of the wright function was studied , and an interesting summary of these results can be founded in @xcite , from where we can say that if @xmath48 we have @xmath107 , \quad a_0=\frac{1}{\sqrt{2\pi(1-{\alpha}/2)}}\ ] ] therefore @xmath108 where @xmath109 and @xmath110 .
+ on the other hand @xmath111 , \ ] ] therefore @xmath112 where @xmath113 .
+ from ( [ comp asint m ] ) and ( [ comp asint w ] ) , we have @xmath114 and then ( [ f_2(+infty)=+infty ] ) holds .
+ now , from proposition [ m pos y decrec ] and properties ( [ caract f_2 ] ) and ( [ f_2(+infty)=+infty ] ) , we can ensure that @xmath115 therefore , there exists at least one @xmath88 which is solution of the equation ( [ eq xi ] ) .
finally , we are able to state that ( [ sol st-2f ] ) is a solution to the free boundary problem ( [ st-2fases ] )
. we will denote ( [ sol st-2f])-([eq xi ] ) as the generalized neumann solution of the two - phase fractional lam - clapeyron - stefan problem ( [ st-2fases ] ) .
the limit when @xmath24 of the generalized neumann solution ( [ sol st-2f])-([eq xi ] ) is the classical neumann solution for the two - phase lam - clapeyron - stefan problem .
we denote @xmath116 , @xmath117 and @xmath118 as the functions defined in ( [ sol st-2f ] ) , and @xmath119 the solution of the equation ( [ eq xi ] ) for each @xmath120 .
now , we analyze the convergence of ( [ sol st-2f ] ) when @xmath24 .
applying proposition [ conv m al/2 cuando al tiende a 1 ] we obtain @xmath121 @xmath122 @xmath123 where @xmath124 is a solution to the equation @xmath125 the expressions ( [ conv 1])-([conv 4 ] ) give us the classical neumann solution , given in @xcite , for the two - phase lam - clapeyron - stefan problem defined by the following equations , and constant boundary and initial conditions : @xmath126 it is an open problem to prove that @xmath102 is an increasing function , which is a sufficient condition to could ensure the uniqueness of the solution to equation ( [ eq xi ] ) . by using maple
we show below some graphs for different values of @xmath120 , from which it can be seen that @xmath102 is an increasing function on @xmath43 .
by using the wright and mainardi functions and the fractional error function @xmath127 , a generalized neumann solution for the two - phase fractional lam - clapeyron - stefan problem is obtained for each @xmath120 . moreover
, the classical neumann solution is recovered through the limit when @xmath24 .
caputo , m. , linear model of dissipation whose q is almost frequency independent - ii , _ geophys .
j. r. astr .
* 13 * ( 1967 ) , 529 - 539 .
carslaw , h.s . , &
jaeger j.c . , _ conduction of heat in solids _ , clarendon press , oxford ( 1959 ) .
fellah , m. , fellah , z. e. a. , mitri , f. g. , & ogam , e. , transient ultrasound propagation in porous media using biot theory and fractional calculus : application to human cancellous bone , _ j. acoust .
* 133 * no 4 ( 2013 ) , 1867 - 1881 .
kholpanov , l. p. , zaklev , s. e. , & fedotov , v. a. , neumann - lam - clapeyron - stefan problem and its solution using fractional differential - integral calculus , _ theoretical fpundations of chemical engineering _ * 37 * , no 2 ( 2003 ) , 113 - 121 .
luchko , y. , some uniqueness and existence results for the initial - boundary - value problems for the generalized time - fractional diffusion equation , _ computer and mathematics with applications _ * 59 * , ( 2010 ) , 1766 - 1772 .
mainardi , f. , mura , a. , & pagnini , g. , the m - wright function in time - fractional diffusion processes : a tutorial survey , _ international journal of differential equations _ , vol .
2010 , article i d 104505 , 29 pages .
tarzia , d.a . , a bibliography on moving - free boundary problems for the heat diffusion equation .
the stefan and related problems , _ mat - serie a _ * 2 * ( 2000 ) , 1 - 297 .
available from : http://web.austral.edu.ar / descargas / facultad - cienciasempresariales / mat / tarzia - mat - seriea-2(2000).pdf tarzia , d.a .
, _ explicit and approximated solutions for heat and mass transfer problems with a moving interface _ , chapter 20 , in advanced topics in mass transfer , mohamed el - amin ( ed . ) ,
intech open access publisher , rijeka ( 2011 ) , 439 - 484 .
available from : http://www.intechopen.com/articles/show/title/explicit-and-approximated-solutions-for-heat-and-mass-transfer-problems-with-a-moving-interface | we obtain a generalized neumann solution for the two - phase fractional lam - clapeyron - stefan problem for a semi - infinite material with constant boundary and initial conditions . in this problem ,
the two governing equations and a governing condition for the free boundary include a fractional time derivative in the caputo sense of order @xmath0 . when @xmath1 1 we recover the classical neumann solution for the two - phase lam - clapeyron - stefan problem given through the error function .
_ 2010 mathematics subject classification : _
: 26a33 , 35r35 ; secondary : 35c05 , 35r11 , 80a22 . _
_ keywords:__ - clapeyron - stefan problem ; neumann solution ; fractional diffusion equation , caputo fractional derivative , explicit solution . |
gamma - ray bursts ( grbs ) have been an astronomical puzzle for 30 years .
details of the gamma - ray emission from bursts are still elusive , although there is an emerging agreement that persistent afterglow emission at other wavelengths from locations consistent with those of several recent grbs can be related to expanding fireballs at cosmological distances ( @xcite ) .
one promising mechanism that has been proposed for gamma - ray emission is the synchrotron shock model ( ssm ) : synchrotron emission from particles accelerated in a relativistic shock is lorentz - boosted into the gamma - ray band ( @xcite ) .
the ssm identifies the spectral break observed in most burst spectra with the characteristic synchrotron energy in the emitter s rest frame , boosted into the observer s frame .
thus , the fitted energy of the spectral break would contain information about both the bulk lorentz motion of the emitters and the equipartition magnetic field producing the synchrotron radiation ( @xcite ) .
in addition , the ssm makes the specific prediction that the low - energy power - law spectral index of the observed photon number spectra should not exceed @xmath0 , with the assumption that the optical depth of the shocked material is less than unity .
we use the convention that the power - law index @xmath2 takes a positive sign : @xmath3 .
the @xmath0 value is derived from the synchrotron single - particle emission spectrum : bulk relativistic motion ensures that the mean particle energy is large enough that there are few low - energy particles ( @xcite ) . below the cyclotron fundamental energy ,
the low - energy spectrum is approximately that produced by mono - energetic particles and is thus independent of their actual distribution and should be a constant @xmath0 spectral index power law . if that were the total story , the model would already be rejected , since it is well - known that bursts are observed to have a variety of spectral behaviors at low energies ( @xcite ) .
however , it also has been noted that the timescale for synchrotron cooling of the particles may be shorter than the duration of burst pulses ( @xcite ) .
a cooling distribution of particles is characterized by a power - law index of @xmath4 , which translates into a @xmath5 photon number index through the synchrotron power - law emission formula ( @xcite ) .
if one includes the effects of cooling of the particle distribution , the low - energy spectral index can encompass the range of @xmath5 to @xmath0 .
however , the spectral slope still can not be greater than the fundamental single - particle limit of @xmath0 . in this _
letter _ , we test the ssm limit on spectral behavior by examining how well the data support it .
we draw our results from a catalog of time - sequences of spectral fits to 137 bursts selected for their high flux and fluence , using ( mostly ) large - area detector ( lad ) data from the burst and transient source experiment ( batse ) on the _ compton gamma ray observatory _ ( see @xcite for details of the analysis ) . in the next section ,
we determine the best spectral indicators derivable from the data and show how they should bracket the ` true ' low - energy spectral behavior . in 3
, for each of the bursts in the catalog we compare the effective low - energy power - law spectral index with the ssm limit line .
finally , in 4 we discuss the implications these results have on further theoretical modeling .
the model typically chosen for spectral fitting of bursts is the empirical ` grb ' function ( @xcite ) : @xmath6}\nonumber\\ { \rm if } \quad e & < & \frac{(\alpha-\beta)e_{\rm peak}}{(2+\alpha ) } \equiv e_{\rm break } { \rm , } \\ { \rm and } \quad f(e ) & = & a \biggl [ % \ { \frac{(\alpha-\beta ) e_{\rm peak}}{100(2+\alpha)}\biggr]^{(\alpha-\beta ) } \exp{(\beta-\alpha ) } ( e/100)^{\beta}\nonumber\\ { \rm if } \quad e & \geq & \frac{(\alpha-\beta)e_{\rm peak}}{(2+\alpha)}{\rm , } \nonumber\end{aligned}\ ] ] where the four model parameters are : the amplitude a , a low - energy spectral index @xmath2 , a high - energy spectral index @xmath7 and an energy @xmath8 that corresponds to the peak of the spectrum in @xmath9 if @xmath7 is less than @xmath4 . in this expression , @xmath10 and @xmath2
are jointly determined by the low - energy continuum , while @xmath7 is solely determined by the spectrum above the break energy , @xmath11 .
if this break energy lies above the highest energy available to the detector , @xmath7 is ill - defined , so we must substitute a related form of the model with the high - energy power - law omitted . in cases where a sharp inherent curvature of the spectrum results in an unacceptable value of @xmath12 for a fit to the grb spectral form , a broken power - law ( bpl ) model often generates better results
. observed burst spectra are such that the data rarely approach the grb spectral low - energy power - law within the energy range of batse and most other burst experiments . at the low end of the spectrum , @xmath2 is approached only asymptotically , as seen by comparing the dotted line in figure [ multi_purp ] with the grb model fit to an example spectrum ( _ solid line _ ) .
the model - dependent photon ` data ' rates consist of photon model rates weighted by the ratios of the deconvolved model rates and the count rates in each data channel , allowing different photon models to be represented simply on a single plot .
a better measure of the actual lower - energy behavior is an effective spectral index , e. g. , the slope of the power - law tangent to the grb function at some chosen energy ( @xmath13 ) , which can be found analytically : @xmath14 for typical values of @xmath2 , i. e. , @xmath15 , @xmath16 will always be less than @xmath2 by an amount that depends upon the value chosen for @xmath13 .
we chose @xmath17 kev for the following reasons : the observed spectrum is consistent with the grb function above about 25 kev ; below this value , deviations from the standard grb function have been observed that may indicate the existence of an additional low - energy spectral component ( @xcite ) .
in addition , 25 kev is just greater than the low - energy cut - off for the lads ; it is above an electronic spectral distortion in the sds at typical gain settings ( @xcite ) and it is also below fitted values for @xmath8 in nearly all bursts . the effective spectral index at 25 kev will be denoted @xmath18 . once @xmath13 has been fixed , the correction gets smaller as the fitted value for @xmath8 gets larger , since @xmath19 is closer to the asymptotic value @xmath2 .
this is also seen in figure [ multi_purp ] , where @xmath20 kev , so @xmath21 ( the grb function is not appropriate for this spectrum and is shown for illustration only ) . in any practical case
, @xmath18 serves as an upper bound on the low - energy spectral index , in the sense that it is more positive than an average slope through the model .
we will use @xmath18 in this work , rather than @xmath2 , since it is more indicative of the slope actually seen in the data . to the extent that all burst spectra have continuous curvature rather than a sharp break , a bpl function fit ( figure [ multi_purp ] _ dashed line _ ) will generate a spectral index ( which we will call @xmath22 ) that is a lower bound on the ` true ' low - energy spectral behavior , in that it will be more negative than it would be for a model that takes into account some curvature , such as the grb function .
thus , to be conservative in our estimate for how well the batse data support the ssm , @xmath22 provides the best available comparison , since if it lies above the limiting @xmath0 line , we can be reasonably sure the ` true ' spectrum will violate the ssm . of course , this will do us no good if the bpl model is not an acceptable fit to the data .
thus , we use fits to two models to put bounds on the ` true ' low - energy spectral behavior : @xmath2 is a model parameter asymptotic to the data , but we can approximate the upper bound with the quantity @xmath18 defined above .
the bpl estimate , @xmath22 , is a lower bound .
in figure [ max_alpha_lod ] , we show results from our batse catalog of time - resolved spectroscopy of bright bursts ( peak flux @xmath23 ph @xmath24 s@xmath25 and fluence @xmath26 erg @xmath24 ) , using mostly lad data .
the value of @xmath18 is plotted against @xmath8 for the spectrum in each burst for which @xmath18 reaches its maximum value .
in addition to the @xmath18 @xmath8 parameters obtained by fitting the grb function , we have also included @xmath22 @xmath11 pairs ( _ diamonds _ ) for those bursts where use of the bpl model was more appropriate . overlaying this plot
is the synchrotron @xmath0 spectral index ` death line ' ( _ dashes _ ) , as well as the lower bound allowed for cooling distributions ( _ dotted line _ ) .
the ssm is quite safe within these boundaries ; however , 44% of the total have maximum low - energy spectral indices above @xmath0 , in a region that is totally excluded in the ssm .
in particular , 32% of the bpl bursts have points that lie above this line . with some correlation between the two displayed parameters ,
the most stringent limits are set by the spectral indices of spectra with the highest values of @xmath8 , as well as any bursts that have a bpl index greater than @xmath0 .
clearly , many such bursts exist ; one is the example shown in figure [ multi_purp ] , which is a bpl fit with @xmath27 and @xmath28 kev , the highest - energy diamond above the @xmath0 limiting line .
the error bars shown in figure [ max_alpha_lod ] represent the 1@xmath29 error for each parameter considered singly , as obtained from the covariance matrix of the fit .
if the errors in the determination of the spectral index were not normally distributed , perhaps having a broad tail instead , then values that are much greater than the ssm limit of @xmath0 might violate the ssm . to test this
, we created 4 sets each of 1000 simulated spectra , based on the grb model spectrum propagated through the detector response , with random poisson fluctuations determined for the counts in each data channel .
these consisted of bright , medium and dim spectra , with expected errors of 0.06 , 0.2 and 0.34 , respectively , for the fitted value of @xmath2 .
the parameter values assumed were e@xmath30 kev , @xmath31 and @xmath32 , to provide a worst case test .
a second set of medium - brightness spectra was also created , to test the effect of a small assumed value of e@xmath33 kev .
a histogram of the resulting fitted @xmath2 values is approximately consistent with a gaussian distribution for each set , however , distributions for the dimmer sets of spectra have slightly extended tails on the side of more positive values of @xmath2 .
these larger values all have large computed errors associated with them , so the resulting distributions of deviations of @xmath2 from the mean , weighted by their errors ( see below ) , do not have a tail .
rather , the average over the 4 simulations for the number of error - weighted deviations greater than 2 @xmath29 is 7.25 , compared with the 22.7 deviations expected for 1000 trials in a normal distribution .
interestingly , the effect where large values of @xmath2 have larger associated errors can be observed in figure [ max_alpha_lod ] , where the errors above the @xmath0 line are , on average , 1.5 times those below .
the ks probability we calculate from comparing the @xmath34 tails between our observations and the simulations is @xmath35 ( @xmath36 for @xmath37 and @xmath38 ) , indicating that it is quite unlikely that the two distributions are the same .
another way to view these results is as a histogram of @xmath39 ( here , we use @xmath2 to stand in as a generic low - energy power - law index ) ; that is , the deviation of @xmath2 from @xmath0 in units of the standard error for each fit .
figure [ sig_dev ] shows the distribution of @xmath40 for the entire ensemble of 3957 spectra from the 137 bursts fit , where the ssm - violating region consists of the positive values to the right of zero .
there are 312 spectra total in the bins containing @xmath41 and greater ( that is : @xmath42 ) .
this may be compared with our expected error distribution , based upon the 4000 total simulated spectral fits described above ( fig .
[ sig_dev ] _ dashed line _ ) , where the total over the same bins is 8 ( in fact , @xmath43 ) .
we may estimate the probability that 312 out of 3957 spectra would arise from @xmath44 fluctuations larger than @xmath0 , assuming the ssm prediction that no such spectra exist .
the simulations show that the chance probability of a @xmath0 slope spectra having a fit value more than @xmath45 away from @xmath0 is @xmath46 . using the binomial probability distribution , the chance that the observed 312 out of 3957 could result from random fluctuations is @xmath47 , showing the observations to be quite inconsistent with the ssm .
we have shown that there are a large number of bursts that violate the limit on low - energy spectral behavior imposed by the basic synchrotron emission mechanism acting in a relativistic shock .
it is worthwhile considering which spectral models can accommodate this observation .
first of all , it has been suggested ( @xcite ) that compton upscattering of soft photons by an energetic distribution of particles can significantly modify the basic synchrotron emission spectrum , with energetic particles boosting their own synchrotron emission into the observed gamma - ray band .
the details of the spectral shape thus depend upon the particle distribution , as well as the shape of the photon spectrum at the energies where it is being sampled for upscattering .
the observed steep low - energy spectral indices would arise for certain combinations of the source parameters and then would evolve to smaller values as the particles cool .
an important prediction from this model is that a low - energy component , independent of the observed gamma - ray emission , must be present with sufficient strength to serve as a pool of photons for upscattering .
it is indeed possible that this separate component may already have been observed ( @xcite ) .
if the low - energy portion is truly an independent component , it will have an independent time history as well .
the troubling part of this idea is that _ all _ bursts should have this component to some extent , since there is no evident bi - modality to the low - energy behavior ( cf .
figure [ max_alpha_lod ] ) that would indicate that the ssm - violating bursts are somehow different . synchrotron self - absorption ( ssa ) is another mechanism that would tend to increase the low - energy continuum spectral index .
the maximum photon spectral index that could be observed is @xmath48 , so all of the bursts presented here would be consistent .
the photon opacity must be close to unity in all these sources for some to be self - absorbed . for ssa to work the optical depth must be greater than one at energies below @xmath8 .
thus , we trade one mystery for another : the narrow distribution of @xmath8 arises in the fact that many bursts have optical depths close to unity , rather than from a narrow distribution of lorentz factors ( see next paragraph ) . also , if the photon density is high , it may be very difficult to overcome the opacity arising from photon - photon pair - production . since the observed value of @xmath8 should scale as the bulk lorentz factor of the emitting material to the fourth power , a narrow observed distribution of @xmath8 implies either that the rest - frame value is extraordinarily precisely determined for all bursts or that the lorentz factors of the entire ensemble lie in a very narrow distribution .
since neither alternative is satisfactory , @xcite proposed that the observed spectra may arise from much different spectra that have been attenuated by intervening material through relativistic compton scattering .
the low - energy behavior is directly related to conditions at the source , namely the optical depth of the scatterers .
large optical depths will result in steep fitted low - energy power - law indices , and the distribution of these should somehow relate to the distribution of densities of the material surrounding putative sources .
thanks to b. schaefer who suggested adding error bars to the @xmath2 @xmath8 figure from preece et al .
the anonymous referee has contributed suggestions that have lead to an improved error analysis . band , d. l. , et al .
1992 , exp .
astron . , 2 , 307 band , d. l. , et al .
1993 , , 413 , 281 brainerd , j. j. 1994 , , 428 , 21 cohen , e. , katz , j. i. , piran , t. , sari , r. , preece , r. d. , & band , d. l. 1997 , , 488 , 330 crider , a. , et al .
1997 , , 497 , l39 goodman , j. 1986 , , 308 , l47 katz , j. i. 1994a , , 432 , l107 katz , j. i. 1994b , , 422 , 248 liang , e. p. , kunose , m. , smith , i. a. , & crider , a. 1997 , , 497 , l35 meegan , c. a. , et al .
1996 , , 106 , 65 ( batse 3b catalog ) metzger et al .
1997 , nature , in press mszros , p. & rees , m. j. 1993 , , 405 , 278 preece , r. d. , et al .
1996 , , 473 , 310 preece , r. d. , et al .
1998 , , 496 , 849 rees , m. j. , & mszros , p. 1992
, , 258 , 41 rees , m. j. , & mszros , p. 1994
, , 430 , l93 rybicki , g. b. , & lightman , a. p. 1979
, radiative processes in astrophysics , new york : wiley , 221 sari , r. , narayan , r. , & piran , t. 1996 , , 473 , 204 sari , r. , & piran , t. 1997 , mnras , in press strohmayer , t. e. , fenimore , e. e. , murakami , t. , & yoshida , a. 1997 , , submitted tavani , m. 1995 , ap&ss , 231(1 ) , 181 tavani , m. 1996 , , 466 , 768 waxman , e. 1997 , , 489 , in press | the synchrotron shock model ( ssm ) for gamma - ray burst emission makes a testable prediction : that the observed low - energy power - law photon number spectral index can not exceed @xmath0 ( where the photon model is defined with a positive index : @xmath1 ) .
we have collected time - resolved spectral fit parameters for over 100 bright bursts observed by the burst and transient source experiment on board the _ compton gamma ray observatory_. using this database , we find 23 bursts in which the spectral index limit of the ssm is violated .
we discuss elements of the analysis methodology that affect the robustness of this result , as well as some of the escape hatches left for the ssm by theory . |
clock synchronization is a well studied problem with many practical and scientific applications . in the special theory of relativity
there are two standard methods for synchronizing a pair of spatially separated clocks , einstein synchronization @xcite and eddington s slow clock transport @xcite .
recently , two new quantum protocols have been proposed for synchronizing remote clocks .
the first one uses prior quantum bit entanglement between the two parties and was proposed by _
jozsa et al _ @xcite .
this protocol is based on the assumption that the entanglement can be achieved without any relative phase error .
however , the validity of this assumption has been discussed and questioned in a number of papers @xcite . once and if this entanglement can be obtained , their algorithm determines the time difference @xmath0 between the clocks by essentially monitoring the oscillation of a function @xmath4 , and thus requires @xmath5 shared singlets .
the second protocol was proposed by _
i. chuang _
@xcite , and obtains @xmath0 to @xmath1 bits of accuracy by communicating only @xmath6 qubits and using an @xmath2 range of frequencies . after communicating the bits according to his protocol , they are in the state corresponding to the fourier transform ( over @xmath7 ) of the state @xmath8 , for some fixed and known @xmath9 . as a result , one can apply an inverse fourier transform and subsequently measure the value of @xmath10 and hence @xmath0 . in this paper
, we improve significantly on chuang s result by presenting an algorithm that is able to calculate @xmath0 to @xmath1 bits of accuracy while communicating only one qubit in one direction and using an @xmath2 range of frequencies .
further , we prove that , under our computational model , the product of the frequency range and the number of transmitted qubits must be @xmath3 , and conclude that our algorithm is optimal in this model .
in our protocol alice sends a photon @xmath11 to bob with some tick rate @xmath9 .
the state of the received photon is @xmath12 , where @xmath13 is the time the photon spent in transit and @xmath14 is the pauli matrix @xmath15}.\ ] ] even though we only use one - way communication from alice to bob , for the purposes of proving computational lower bounds we assume an even stronger model , where the two parties can exchange photons back and forth .
the information they get about the time difference @xmath0 between the two clocks comes from a phase change in the state of the qubit , which depends on @xmath0 and the tick rate @xmath9 .
let s now define this procedure and see how we can actually implement it .
the input to this procedure will be a quantum register which holds the tick rate @xmath16 and a qubit @xmath11 .
the output is a state that has a phase which depends on @xmath0 and @xmath16 .
let tqh be a black box quantum procedure defined by the equation @xmath17 where @xmath0 is the time difference between the two parties and @xmath18 is a known base tick rate .
this is a very reasonable and powerful model , since we know that all the information one can get about the time difference via such photon communications is in the form of a relative phase change . here
the first register handles the tick rate of the photon to be transmitted and the second register is the photon that alice communicates to bob ( or bob to alice ) .
the implementation of this black box is based on the ticking qubit handshake protocol ( tqh ) described in i. chuang s paper @xcite .
suppose alice wants to create the state @xmath19 .
she first sends the qubit @xmath11 to bob with ticking rate @xmath20 . along a classical channel
she also tells him her time @xmath21 at the moment of the quantum communication .
bob receives at time @xmath22 ( according to bob s clock ) a quantum state @xmath23 , where @xmath24 is the time the qubit spent in transit .
finally , bob applies a phase change @xmath25 and thus the final state of the qubit is @xmath26 .
we are going to describe a protocol for synchronizing two remote clocks by communicating one photon . in this algorithm , alice starts by preparing a register @xmath27 of @xmath1 qubits in a certain superposition . then she sends a photon to bob with the superposition of tick rates specified in @xmath27 .
bob measures the received photon and alice obtains @xmath0 to @xmath1 bits of accuracy by processing a phase estimation on @xmath27 . in more detail , 1 .
alice starts with a register of @xmath1 qubits initialized to @xmath28 , and after applying a fourier transform to them she obtains @xmath29 she also prepares a photon with polarization state @xmath30 .
alice now transmits the prepared photon with the tick rate described by her first register .
if the photon had a definite tick rate @xmath16 and polarization state @xmath11 the final state would be @xmath31 . since the register described in step 1 is in a superposition of tick rates , the outcome will be in a superposition of states @xmath32 3 .
bob measures the received photon .
assuming without loss of generality that the outcome is @xmath28 , alice s register @xmath27 becomes @xmath33 4 .
alice then applies an inverse fourier transform , obtaining the state @xmath34 in @xmath27 .
it is easy to see that this algorithm is an application of the general procedure known as _ phase estimation_. in this procedure , we assume a unitary operator @xmath35 with an eigenvector @xmath36 and eigenvalue @xmath37 .
the goal is to estimate @xmath38 to @xmath1 bits of accuracy . to perform the estimation we start with two registers , the first one in a uniform superposition over all states in @xmath7 and the second one in the state @xmath36 .
then we apply the unitary operation @xmath35 to the second register @xmath39 times , where @xmath40 is the content of the first register . by analyzing the performance of this procedure
it can be seen that our algorithm obtains @xmath0 to @xmath1 bits of accuracy with constant probability @xmath41 .
we can boost the probability of success to @xmath42 by increasing the size of the first register to @xmath43 .
further analysis can be found in @xcite , page 221 .
in this section we will prove a lower bound on the product of the range of tick rates ( frequencies ) we use and the number of qubits we communicate .
if a quantum algorithm in the tqh model makes only queries to the black box with a single tick rate @xmath9 , then it must make a total number of @xmath3 queries in order to obtain @xmath0 to @xmath1 digits of accuracy . by making a query to the black box ,
the input @xmath48 will become @xmath49 ; after applying a hadamard transform we obtain the quantum state @xmath50 from this we see that the problem of determining @xmath0 is equivalent to estimating the amplitude of @xmath28 ( or @xmath51 ) . the problem of estimating the amplitude of a quantum state , which is equivalent to the problem of counting the number of solutions to a quantum problem , is well - studied @xcite . in @xcite
they prove that @xmath52 queries are required for a @xmath53-approximate count , where @xmath13 is the number of solutions , @xmath54 is the set of possible inputs and @xmath53 defines the closeness of the approximation .
if we use this lower bound for the case of amplitude estimation , we get a lower bound of @xmath55 , since the amplitude is @xmath56 . in our case , @xmath57 can take any value in @xmath58 , @xmath53 must be less than @xmath59 and @xmath60 , so we obtain the lower bound of @xmath3 qubits .
+ suppose we are able to query the black box with frequencies in the range @xmath61 $ ] .
we claim that this black box can be simulated by a black box with only one tick rate @xmath9 at the cost of replacing each query with at most @xmath45 queries .
this can be done since one query to the @xmath61 $ ] black box with tick rate @xmath62 is equivalent to @xmath16 consecutive queries with tick rate @xmath9 , using the output of one query as the input to the next .
notice that a superposition of queries to the @xmath61 $ ] black box does not pose any challenge to the simulation , since we can also query the one - tick rate black box in a superposition of times .
for such an input , the number of queries is defined to be the maximum over all states of the superpositions . since in all cases
we query the black box at most @xmath45 times , this means that the one - tick rate version will run with at most @xmath63 queries . now , in lemma 1 we have already proved that when we use only one tick rate , we need to communicate at least @xmath3 qubits , and therefore @xmath64 . | the clock synchronization problem is to determine the time difference @xmath0 between two spatially separated parties .
we improve on i. chuang s quantum clock synchronization algorithm and show that it is possible to obtain @xmath0 to @xmath1 bits of accuracy while communicating only one qubit in one direction and using an @xmath2 frequency range .
we also prove a quantum lower bound of @xmath3 for the product of the transmitted qubits and the range of frequencies , thus showing that our algorithm is optimal . |
we know that the black hole entropy is proportional to the horizon aria @xmath0 , known as entropy - area law of the black hole @xcite . on the other hand , thermal fluctuations will happen for any thermodynamical system like black holes .
specially , thermal fluctuations are more important for the small objects , because one can see that thermal fluctuations arise due to quantum fluctuations in the geometry of space - time .
so , one can neglect such fluctuations for the large black holes .
however , when the size of black hole decreased due to hawking radiation , the effect of quantum fluctuations on the geometry of the black holes will be important .
thermal fluctuations modify thermodynamic of system and they appear in the black hole entropy as logarithmic term , so logarithmic corrected entropy may be written as @xcite , @xmath1 where @xmath2 is a constant usually fixed as unity .
we can track effect of thermal fluctuations using @xmath2 and obtain ordinary thermodynamics by setting @xmath3 .
recently effect of thermal fluctuations considered for several kinds of black objects like charged ads black hole @xcite , black saturn @xcite and modified hayward black hole @xcite .
moreover , in the ref .
@xcite black hole entropy corrections due to thermal fluctuations in the black hole extensive parameters for various ads black holes have been studied for both canonical and grand canonical ensembles .
some asymptotic ads black holes like btz , d=4 kn - ads and r - charged black holes in various dimensions considered in that paper and found an universality in the logarithmic corrections to charged ads black holes entropy in various dimensions , which are expressed in terms of the black hole response coefficients via fluctuation moments .
+ in this paper , we would like to consider an interesting kind of black hole which is called dyonic charged ads black hole in four dimensions which has both electric and magnetic charges @xcite . in the ref .
@xcite using critical point investigation it is found that charged ads black holes are corresponding to van der waals fluid .
also , it has been argued that dyonic charged ads black hole may be holographic dual of van der waals fluid with chemical potential @xcite .
it is also found that van der waals fluid is holographic dual of rn ads black hole @xcite .
so , by using holographic principles one can study dyonic charged ads black hole via a van der waals .
we use this motivation to study logarithmic corrected thermodynamics of dyonic charged ads black hole to see such quantum gravity effect .
+ the paper is organized as follows . in the next section
we give brief review of dyonic charged ads black hole and recall necessary relations which need to study its thermodynamics under thermal fluctuation effect . in section 3 we obtain several thermodynamics relations and see the effect of logarithmic correction . in section 4
we use holographic picture which is a van der waals fluid and study about critical behavior . in section 5
we discuss about stability of black hole , and finally in section 6 we give conclusion .
dyonic charged ads black hole is solution of einstein - maxwell theory with negative cosmological constant in four dimensions which described by the following action @xcite , @xmath4 where @xmath5 is the ricci scalar , @xmath6 is the strength of electromagnetic field and @xmath7 is the curvature radius of ads space , also @xmath8 assumed .
solving einstein equations yields to the following metric , @xmath9 where , @xmath10 where @xmath11 , @xmath12 and @xmath13 are electric charge , magnetic charge and mass of the black hole respectively .
electric and magnetic potential @xmath14 defined by the following equation , @xmath15 where the black hole horizon @xmath16 is given by , @xmath17 which has at least two real and positive roots @xmath18 .
so , one can write the mass of the black hole as follow , @xmath19 @xmath20 in terms of @xmath21 , and black hole mass as a function of @xmath16 plotted by the fig .
. we will discuss about them in the next section when study the logarithmic corrected thermodynamics of the solution discussed above .
+ before end of this section it is useful to note that 4 dimensional dyonic charged ads black hole in asymptotically ads space - time is conjectured to be dual to a 3 dimensional cft living on the boundary of the ads space @xcite .
the bulk gauge field is dual to a global @xmath22 current operator @xmath23 , while conserved global charge of cft side given by , @xmath24 where @xmath25 is four dimensional newtonian constant related to degree of the gauge group @xmath26 of the cft using the holographic dictionary , @xmath27 where @xmath28 related to the strength of the magnetic field @xmath29 .
we know that a black hole may be considered as thermodynamic systems characterized by a hawking temperature @xmath30 and the corresponding logarithmic corrected entropy @xmath31 given by the equation ( [ s1 ] ) , with @xmath32 , hence we have , @xmath33 where @xmath34 where we used the equation ( [ s7 ] ) in the last equality .
+ using the entropy and temperature we can find the helmholtz function , @xmath35 where @xmath36\nonumber\\ & -&\frac{3\alpha r_{+}}{8\pi l^{2}}\left[4+\ln{\frac{8\pi l^{4}r_{+}^{4}}{(l^{2}(q_{e}^{2}+q_{m}^{2})-l^{2}r_{+}^{2}-3r_{+}^{4})^{2}}}\right]\nonumber\\ & -&\frac{\alpha}{8\pi r_{+}}\left[\ln{\frac{8\pi l^{4}r_{+}^{4}}{(l^{2}(q_{e}^{2}+q_{m}^{2})-l^{2}r_{+}^{2}-3r_{+}^{4})^{2}}}\right].\end{aligned}\ ] ] neglecting thermal fluctuation ( @xmath3 ) equation ( [ s11 ] ) vanishes .
also , the first term of r.h.s of the equation ( [ s10 ] ) is corresponding to ordinary ads black hole with thermodynamics pressure related to the cosmological constant @xcite , @xmath37 and thermodynamic volume , @xmath38 hence , in presence of electric and magnetic charge and thermal fluctuations , at least on of the above equations should be modified including charge and @xmath2 dependent terms .
we will assume thermodynamic volume given by the equation ( [ s13 ] ) and obtain modified pressure .
+ let us now discuss about helmholtz free energy given by the equation ( [ s10 ] ) . we will fix parameters as @xmath39 and @xmath40 . in that case from the fig .
[ fig1 ] ( a ) we can see that if @xmath41 then @xmath42 .
so , from the fig .
[ fig1 ] ( b ) one can see extremal limit given by @xmath43 , while @xmath44 is corresponding to @xmath45 or @xmath46 is corresponding to @xmath47 .
+ @xmath48{1 - 1.eps}\includegraphics[width=50 mm]{1 - 2.eps } \end{array}$ ] @xmath48{2 - 1.eps}\includegraphics[width=50 mm]{2 - 2.eps } \end{array}$ ] then , from the fig .
[ fig2 ] we can see four different situations for the helmholtz function and internal energy , all have similar behavior for the large black hole with @xmath49 where @xmath50 has negative large value and @xmath51 has positive large value .
effect of thermal fluctuations on the black hole with small size is obvious and important .
solid ( red ) and dashed ( yellow ) lines represent logarithmic corrected cases ( charged and uncharged respectively ) where helmholtz function goes to negative infinity at @xmath52 limit , while for the case of @xmath3 we have @xmath53 ( charged ) and @xmath54 ( zero - charge ) at @xmath52 limit .
the solid red line may be corresponding to the black hole with mass @xmath46 where @xmath55 near the outer horizon ( @xmath56 ) and @xmath57 near the inner horizon ( see fig . [ fig1 ] ( b ) ) .
outside of the black hole , free energy is negative for the mentioned black hole mass .
we can see critical radius @xmath58 , before it internal energy vanishes ( fig .
[ fig1 ] ( b ) ) and helmholtz energy have different behavior .
as we will see later it relate to stability of black hole , ie .
black hole is stable for @xmath59 .
it means that the logarithmic effect is more important for small black hole .
+ in order to obtain internal energy we used well - known thermodynamics relation , @xmath60 effect of thermal fluctuation on the internal energy is decreasing its value for the large black holes . on the other hand for
the infinitesimal black hole ( @xmath61 ) internal energy will be infinite ( see dashed yellow line of right plot of the fig .
[ fig2 ] ) . in the case of @xmath62
the internal energy will be imaginary for @xmath63 , which may be interpreted as critical value of horizon radius where black hole can exist .
there is also a minimum for the internal energy corresponding to minimum mass shown in the fig .
[ fig1 ] ( a ) .
+ as we told already , modified pressure due to thermal fluctuation can be obtained using the derivative of helmholtz function with respect to the volume , @xmath64 by using the equations ( [ s11 ] ) and ( [ s13 ] ) one can obtain pressure of dyonic charged ads black holes as follow , @xmath65 the first term of r.h.s is corresponding to ordinary ads black hole pressure given by the equation ( [ s12 ] ) .
+ then , we can calculate enthalpy using the equations ( [ s13 ] ) , ( [ s14 ] ) , ( [ s16 ] ) and the following relation , @xmath66 where @xmath67 is contribution of thermal fluctuations .
enthalpy of the black hole may be considered as black hole mass @xcite , from the left plot of the fig .
[ fig3 ] we can see as the @xmath3 limit ( dotted line ) enthalpy behaves as black hole mass which illustrated by the fig . [ fig1 ] ( a ) .
as expected we can see that enthalpy of large black hole has similar behavior for @xmath3 , @xmath68 , @xmath40 and @xmath69 . for the uncharged black hole we can see that effect of thermal fluctuation
is some negative enthalpy ( negative mass ) which is forbidden .
hence , as before , there is a minimum value for the horizon radius of the black hole .
there is a critical value for the horizon radius @xmath70 , effect of thermal fluctuations is decreasing @xmath71 for @xmath72 and is increasing @xmath71 for @xmath73 .
+ @xmath48{3 - 1.eps}\includegraphics[width=50 mm]{3 - 2.eps } \end{array}$ ] then we can obtain gibbs free energy using the following relation , @xmath74 where @xmath75 is contribution of thermal fluctuations . in the right plot of the fig .
[ fig3 ] we can see behavior of gibbs free energy with variation of @xmath2 and black hole charge . we can see similar behavior with helmhotz free energy .
apparent kink in the right plot and initial value of left plot is about @xmath76 as explained after the fig .
+ the modified first law of black hole thermodynamics of this case given by @xcite , @xmath77 it is clear that the case of @xmath3 satisfied first law of black hole thermodynamics . it would be interesting to investigate validity of above equation in presence of logarithmic correction and obtain appropriate condition to satisfy the equation ( [ first ] ) .
after a bit calculation we find that the following condition @xmath78 is necessary to have valid first law of black hole thermodynamics .
dyonic charged ads black hole in absence of thermal fluctuations is holographic dual of a fluid with van der waals equation state given by @xcite , @xmath79 where @xmath80 is the boltzmann constant .
the constant @xmath81 denotes nonzero size of the molecules of a given fluid , while the constant @xmath82 is a measure of the interaction between them .
we will show numerically that the thermal fluctuations have not so important effect on p - v diagram , and discuss about critical points where , @xmath83 at @xmath84 direction of p - v line is zero but it is not minimum or maximum . below it
we have some region with negative compressibility corresponding to dual van der waals fluid , which means instability of black hole .
@xmath48{4 - 1.eps}\includegraphics[width=50 mm]{4 - 2.eps } \end{array}$ ] the typical behavior of p - v diagram corresponding to van der waals fluid plotted in fig .
[ fig4 ] ( a ) .
+ we can use equation ( [ s13 ] ) and ( [ s16 ] ) to investigate p - v diagram of dyonic charged ads black hole in presence of thermal fluctuations .
our numerical analysis illustrated by fig .
[ fig4 ] ( b ) .
we can see solid red lines in both ( a ) and ( b ) plots of fig .
[ fig4 ] to find that dyonic charged ads black hole in presence of thermal fluctuations with suitable value of electric and magnetic charges is also dual of van der waals fluid .
we find that value of charge is many important to have dual van der waals fluid .
therefore , thermodynamics quantities obtained in the previous section use to see thermal fluctuation and hence quantum gravity effects . + we will discuss about critical points ( corresponding to black lines ( dash dot ) of both fig .
[ fig4 ] ( a ) and ( b ) ) and stability of the model in the next section . +
critical points of the model are corresponding to the dash dotted black lines of fig . [ fig4 ] ( a ) and ( b ) . at @xmath3 limit and using both conditions given by ( [ s20 ] ) one can obtain the following equation , @xmath85 it gives us critical horizon radius as follow , @xmath86 where we defined , @xmath87^{\frac{1}{3}},\ ] ] which means that @xmath88 .
however we expect that effect of thermal fluctuations change charge interval .
+ specific heat is an important measurable physical quantity which can determine stability of system .
it has the following relation with internal energy @xmath51 , entropy @xmath31 and temperature @xmath30 , @xmath89 if @xmath90 black hole is stable and if @xmath91 black hole is in unstable phase .
so , @xmath92 is corresponding to phase transition of van der waals fluid , similar to the critical point discussed above . by using the equation ( [ s14 ] )
we can obtain specific heat as following , @xmath93 where @xmath94 is logarithmic corrected terms . in the fig .
[ fig5 ] we can see effect of thermal fluctuations on the stability of black hole .
we can see that there is a minimum size for the black hole as expected .
below minimum size @xmath95 , black hole in unstable and it means that van der waals fluid changes phase .
fortunately , it is possible to see thermal fluctuation effects for @xmath96 where black hole is stable and we have van der waals fluid .
reducing electric and magnetic charges help to obtain smaller minimum size , therefore we have more chance to see effects of thermal fluctuations . for the case of @xmath97 one
may have @xmath58 as discussed after the fig .
[ fig2 ] and fig .
[ fig3 ] . without thermal fluctuations
it is clear that the black hole can exist at smaller size which illustrated by the fig .
[ fig5 ] ( compare solid and dashed red lines for example ) . we see that while thermal fluctuations reduce stability of black hole but it has enough stable region to reflect quantum gravity effects .
also , green lines represent uncharged case which is completely unstable without logarithmic correction and we can see opposite behavior comparing with charged black hole . @xmath48{5.eps }
in present work we considered special solutions of einstein - maxwell theory with negative cosmological constant in four dimensions which is dyonic charged ads black hole .
our main work is finding the effect of thermal fluctuations on the thermodynamics quantities .
+ thermal fluctuations exist for any black object , but they are important for small black hole and negligible for the large black hole .
advantage of dyonic charged ads black hole background is its holographic picture which is a van der waals fluid .
we have shown that , in presence of thermal fluctuations there is still a van der waals fluid as dual picture .
only we should fix black hole charge which is corresponding to the electric charges of a van der waals fluid .
we obtained some thermodynamics quantities like gibbs and helmholts free energies and shown that thermal fluctuations have not important effects for the large black hole . on the other hand for the small black holes they are important and
have crucial role .
we found that thermal fluctuations reduced stable regions of the black hole , however there are enough stable regions to see quantum gravity effects before phase transition of a van der waals fluid .
it means that there is a minimum radius , we called critical radius , where black hole is stable in presence of thermal fluctuations , and in this region dyonic charged ads black hole in presence of thermal fluctuations is dual of van der waals fluid .
+ + + * acknowledgments * it is pleasure to thanks prof .
robert b. mann for introducing origin of holographic dual of charged ads black hole .
also we would like to thanks s. mahapatra and yu tian for useful comments .
45 s.w . hawking ,
nature 248 , 30 ( 1974 ) s. das , p. majumdar and r. k. bhaduri , class .
19 , 2355 ( 2002 ) j. sadeghi , b. pourhassan , and f. rahimi , can .
92 , 1638 ( 2014 ) b. pourhassan and m. faizal , europhys .
111 , 40006 ( 2015 ) m. faizal and b. pourhassan , phys .
b 751 , 487 ( 2015 ) b. pourhassan , m. faizal , phys . lett .
b 755 , 444 ( 2016 ) b. pourhassan , m. faizal , u. debnath , eur . phys
. j. c 76 , 145 ( 2016 ) s. mahapatra , p. phukon , t. sarkar , phys . rev .
d.84 , 044041 ( 2011 ) h. l , yi pang , and c.n .
pope , jhep 1311 , 033 ( 2013 ) p. chaturvedi , a. das , g. sengupta , [ arxiv:1412.3880 ] d. kubiznak , r.b .
mann , jhep 1207 , 033 ( 2012 ) s. dutta , a. jain and r. soni , jhep 2013 , 60 ( 2013 ) c. niu , y. tian , x. wu , phys .
d85 , 024017 ( 2012 ) d. kastor , s. ray , j. traschen , class .
26 , 195011 ( 2009 ) b.p .
dolan , class .
28 , 125020 ( 2011 ) j. sadeghi , k. jafarzade , b. pourhassan , int . j. theor
51 , 3891 ( 2012 ) | in this paper , we consider dyonic charged ads black hole which is holographic dual of a van der waals fluid .
we use logarithmic corrected entropy and study thermodynamics of the black hole and show that holographic picture is still valid .
critical behaviors and stability also discussed .
logarithmic corrections arises due to thermal fluctuations which are important when size of black hole be small .
so , thermal fluctuations interpreted as quantum effect .
it means that we can see quantum effect of a black hole which is a gravitational system .
+ + * keywords : * dyonic charged ads black hole ; quantum gravity ; thermodynamics ; holography .
+ + * pacs number : * 04.60.-m 04.70.-s 05.70.-a |
anisotropy has been shown to be a relevant perturbation for gauge theories@xcite .
the coupling to fermions allows us to put chiral fermions on the lattice in either the overlap or the domain wall variants@xcite .
the continuum limit may be taken , since a new phase of `` layers '' , or 3-branes , appears@xcite and the transition from this phase to the 5d confining phase turns out to be second order@xcite . in the past year
we have studied what happens , when scalar fields are included @xcite .
we have found that an additional `` layered '' phase appears .
it consists of a stack of 3-branes , one lattice spacing apart , that are in the higgs phase and the fields are confined on each layer .
the phase diagram comprises thus of five phases , three bulk and two layered : a confining phase ( @xmath0 ) , a bulk higgs phase ( @xmath1 ) a bulk coulomb phase ( @xmath2 ) , a layered coulomb phase ( @xmath3 ) and a layered higgs phase ( @xmath4 ) .
we obtained the phase diagram by monte carlo simulations and mean field theory calculations . indeed the phase of `` layers '' is found also as a solution of the mean field ( i.e. classical ) equations of motion .
we use the standard , compact , abelian higgs action , with provision made for different couplings along a single ( @xmath5 ) direction from the other four .
@xmath6\\ + \beta_{h}^{\prime } \sum _ { x } { \rm re } [ \varphi^{*}(x)\varphi ( x ) - \varphi^{*}(x)u_{\hat 5}(x ) \varphi ( x+\hat 5)]\\ + \sum _ { x}[(1 - 2\beta_{r}-4 \beta_{h}- \beta_{h}^{\prime})\varphi^{*}(x)\varphi ( x)\\ + \beta_{r}(\varphi ^*(x)\varphi ( x))^2]\\ \end{array } \label{compactaction}\ ] ] the order parameters we will use are the expectation values of the plaquette in the bulk @xmath7 and the plaquette in the transverse direction , @xmath8 defined by @xmath9 and the susceptibility of the link in the bulk @xmath10 where @xmath11 is the phase of the higgs field , @xmath12 and @xmath13 .
the phase diagram of the 5d , anisotropic , compact @xmath14 theory is , displayed , for reference purposes , in fig .
[ u1phasediag]@xcite . including the higgs adds two more dimensions .
we fix @xmath15 and @xmath16 and vary @xmath17 and @xmath18 . for @xmath19 (
weak 4d gauge coupling)@xcite we find the following `` snapshot '' cf .
[ u1higgsphasediagw ] , while for @xmath20 ( strong 4d gauge coupling)@xcite we find the snapshot in fig .
[ u1higgsphasediags ] .
as expected , at strong coupling the @xmath3 phase is no longer there . for generic values of the higgs parameters , @xmath17 and @xmath21we
find that the phase transitions are 1st order : we display typical hysteresis loops for the `` bulk '' and `` transverse '' plaquettes , @xmath7 and @xmath8 in figs .
[ plaquettes ] for the transitions from the bulk confining phase @xmath0 to the higgs phases @xmath4 and @xmath1 .
however , we also find hints of continuous transitions , for some subsets of parameter values
the susceptibility in fig .
[ suscept ] . theory . ] , @xmath15 and @xmath22 . here
@xmath23 . ] , @xmath15 . ] , vs. @xmath24 for @xmath20 , @xmath25 , @xmath15 and @xmath22 . ]
we have found 3-brane configurations in the 5d , anisotropic , abelian higgs model .
they may be in either the coulomb or the higgs phases .
the fields in these configurations are _ confined _ on these layers .
this confinement is not put in by hand it is the defining characteristic of the layered phase(s ) .
the transitions between these layered phases and the usual bulk phases are , generically , 1st order ; however it is also possible to find subsets in the space of parameter values , that lead to second order transitions .
this points to the possibility of new , strongly coupled , continuum theories@xcite , whose elucidation is of major interest .
it is to be noted that yang - mills theories should exhibit @xmath26-brane configurations with @xmath27 , since they have a coulomb phase in more than four dimensions .
another alternative , relevant for four dimensions , would be the partial breaking of the yang - mills gauge group , that leaves a @xmath14 factor in the residual gauge group .
9 y. k. fu and h. b. nielsen , _ nucl .
_ * b236 * ( 1984 ) 167 ; _ nucl .
* b254 * ( 1985 ) 127 .
p. hernndez , these proceedings ; y. kikukawa , these proceedings . c. p. korthals - altes , s. nicolis and j. prades , _ phys .
lett . _ * b316 * ( 1993 ) 339 , [ hep - lat/9306017 ]
. a. hulsebos , c. p. korthals - altes and s. nicolis , _ nucl .
phys . _ * b450 * ( 1995 ) 437 , [ hep - th/9406003 ] .
a. hulsebos , _ nucl .
* 42 * ( 1995 ) 618 , [ hep - lat/9412031 ] ; p. dimopoulos , k. farakos , a. kehagias and g. koutsoumbas , [ hep - th/0007079 ] , _ nucl . phys . _
* b * _ in press _
p. dimopoulos , k. farakos , c. p. korthals - altes , g. koutsoumbas and s. nicolis , _ j. high energy phys .
_ * 02(2001)005 * [ hep - lat/0012028 ] .
p. dimopoulos , k. farakos and s. nicolis , _ multi - layer structure in the strongly coupled 5d abelian higgs model _ , [ hep - lat/0105014 ]
also p. dimopoulos , k. farakos and s. nicolis , these proceedings . | we find 3-brane higgs and coulomb phases in the 5d abelian higgs model and determine the transition surfaces that separate them from the usual bulk phases . |
decays of charmed baryons , unlike charmed mesons , are not colour or helicity suppressed , allowing us to investigate the contribution of w - exchange diagrams .
there are also possible interference effects due to the presence of identical quarks .
this makes the study of these decays a useful tool to test theoretical models that predict exclusive decay rates @xcite . during the past several years
there has been significant progress in the experimental study of hadronic decays of charmed baryons .
new results on masses , widths , lifetimes and decay asymmetry parameters have been published by various experiments @xcite .
however the accuracy of branching ratio measurements does not exceed 30% for many cabibbo - favoured modes : for cabibbo - suppressed and w - exchange dominated decays , the experimental accuracy is even worse . as a result , we are not yet able to conclusively distinguish between the decay rate predictions made by different theoretical models .
in this paper we present a study of @xmath10 baryons produced in the @xmath67 continuum at belle , relying on the excellent particle identification capability of the detector to measure decays with kaons in the final state .
we report the first observation of the cabibbo - suppressed decays @xmath68 and @xmath69 , and the first observation of @xmath70 with large statistics .
( here and throughout this paper , the inclusion of charge - conjugate states is implied . )
we present improved measurements of the cabibbo - suppressed decays @xmath3 and @xmath4 , and the w - exchange decays @xmath5 and @xmath6 ; we also report the first evidence for @xmath7 , and set an upper limit on non - resonant @xmath5 decay .
the data used for this analysis were taken on the @xmath71 resonance and in the nearby continuum using the belle detector at the asymmetric @xmath9 collider kekb .
the integrated luminosity of the data sample is equal to 32.6 fb@xmath8 .
belle is a general purpose detector based on a 1.5 t superconducting solenoid ; a detailed description can be found elsewhere @xcite .
tracking is performed with a silicon vertex detector ( svd ) composed of three concentric layers of double - sided silicon strip detectors , and a 50 layer drift chamber .
particle identification for charged hadrons , important for the measurement of final states with kaons and/or protons , is based on the combination of energy loss measurements @xmath72 in the drift chamber , time of flight measurements and aerogel erenkov counter information . for each charged track
, measurements from these three subdetectors are combined to form @xmath73 and @xmath74 likelihood ratios in the range from 0 to 1 , @xmath75 where @xmath76 , @xmath77 and @xmath78 are the likelihood values assigned to each identification hypothesis for a given track . for the analyses presented here , we require p(@xmath79 for pions , p(@xmath80 for kaons , and p(@xmath81 for protons , unless stated otherwise . candidate @xmath82
s are reconstructed from pairs of photons detected in the csi calorimeter , with a minimum energy of 50 mev per photon .
the interaction point ( ip ) coordinates in the @xmath83 plane are determined from beam profile measurements .
other particles are identified as follows : * @xmath84 are reconstructed in the decay mode @xmath85 , fitting the @xmath86 and @xmath87 tracks to a common vertex and requiring an invariant mass in a @xmath88 mev@xmath89 ( @xmath90 ) interval around the nominal value .
the likelihood ratio cut on the proton is relaxed to p@xmath91 .
we then make the following cuts on the @xmath84 decay vertex : * * the closest distance of approach along the beam direction between the proton and pion tracks must be less than 1 cm ; * * the distance between the decay vertex and the interaction point in the @xmath83 plane must be greater than 1 mm ; * * the cosine of the angle in the @xmath83 plane between the @xmath84 momentum vector and the vector pointing from the ip to the decay vertex must be greater than 0.995 . *
@xmath92 are reconstructed in the decay mode @xmath93 , fitting the @xmath94 and @xmath87 tracks to a common vertex and requiring an invariant mass in a @xmath95 mev@xmath89 ( @xmath90 ) interval around the nominal value .
we then make the same vertex cuts as in the @xmath84 case .
* @xmath96 are reconstructed in the decay mode @xmath97 , requiring an invariant mass within @xmath98 ( @xmath99 ) of the nominal value .
we require the proton to have at least one hit in the svd , to improve its impact parameter resolution with respect to the ip ; we then require the impact parameter in the @xmath83 plane to be greater than @xmath100 , to make sure the @xmath96 vertex is displaced from the ip .
* @xmath101 decays are formed using identified @xmath84 and photons with calorimeter cluster energies above @xmath102 gev ; we accept candidates with invariant masses within @xmath103 mev@xmath89 ( @xmath104 ) of the nominal value .
to suppress combinatorial and @xmath105 backgrounds , we require @xmath10 candidates to have scaled momentum @xmath106 ; here @xmath107 is the reconstructed momentum of the @xmath108 candidate in the @xmath109 center of mass , and @xmath110 , where @xmath111 is the total center of mass energy and @xmath112 is the reconstructed mass of the @xmath10 candidate . in modes where there are two or more charged tracks at the @xmath10 vertex
, we perform a vertex fit and require @xmath113 . in the various mass fits described below , the central value and width of the signal peaks are always allowed to float , unless stated otherwise .
wherever the final state includes a hyperon , we improve the invariant mass resolution by plotting the corrected mass difference , _ e.g. _
@xmath114 instead of @xmath115 .
the cabibbo - suppressed decay @xmath118 has not been previously observed . reconstructing @xmath119 combinations as described in section [ section - selection ]
, we see a clear signal at the @xmath10 mass , as shown in fig .
[ fig_lamc_lam0k_1 ] . to study backgrounds due to cabibbo - allowed decays , we select a second sample with a reversed identification requirement p@xmath120 applied to the `` kaon '' . in the mass spectrum of this sample , where the kaon mass hypothesis is still used ,
we see a broad structure centered around @xmath121 , produced by @xmath122 and @xmath123 decays .
( 550,200 ) ( 80,40 ) ( 145,0)@xmath124 $ ] ( 110,10 ) : invariant mass spectrum of the selected @xmath119 combinations .
the broad structure to the right of the signal peak , due to @xmath125 and @xmath126 decays , is included in the fit .
, title="fig:",scaledwidth=60.0% ] we fit this distribution using two gaussians ( to model the @xmath122 and @xmath123 contributions ) , and a second order polynomial ( to model the broad reflections and the remaining background ) .
the shape of this function is then used to model the @xmath127 background in the main sample . the remaining combinatorial background in fig . [ fig_lamc_lam0k_1 ]
is represented using a second order polynomial , and the @xmath128 signal is described by a gaussian with width @xmath129 ( fixed from monte carlo ) ; the result of the fit is shown by the superimposed curve .
we find a yield of @xmath130 @xmath128 decays , the first observation of this decay mode . for normalization
, we use the decay @xmath122 .
the @xmath122 mass distribution is fitted with a gaussian for the signal and a second order polynomial for the background .
we find @xmath131 events .
the relative reconstruction efficiency was determined using , monte carlo simulation ( mc ) , to be @xmath132 . using this value , we extract @xmath133 the first error is statistical , and
the second is systematic .
we provide a detailed description of the sources of systematic error for this and other measured decay modes in section [ errors ] . the cabibbo - suppressed decay @xmath134 is reconstructed in a similar way , with the scaled momentum cut tightened to @xmath135 to suppress the large background due to soft photons . the invariant mass distribution of the selected @xmath136 candidates is shown in fig . [ fig_lamc_sig0k_1 ] : a peak is seen at the @xmath10 mass , and a reflection due to misidentified two - body cabibbo - allowed @xmath10 decays is seen at higher masses .
the superimposed curve shows the result of a fit following the method described for @xmath119 , with the exception that in this case the width of the signal gaussian is fixed from the mc to @xmath137 mev@xmath89 .
we find @xmath138 @xmath1 events , the first observation of this decay mode .
for normalization , we use the decay @xmath139 .
( 550,200 ) ( 80,40 ) ( 145,0)@xmath140 $ ] ( 110,10 ) : invariant mass spectrum of the selected @xmath141 combinations .
the broad structure to the right of the signal peak , due to @xmath125 and @xmath126 decays , is included in the fit .
, title="fig:",scaledwidth=60.0% ] we fit the distribution with a gaussian for the signal , a second gaussian to describe the broad enhancement due to @xmath122 ( with the addition of a random @xmath142 ) , and a second order polynomial for the remaining background .
the fit gives @xmath143 @xmath139 decays .
the relative reconstruction efficiency found to be @xmath144 in the mc : we then calculate @xmath145
the first evidence for the cabibbo - suppressed decay @xmath70 was published by the na32 collaboration in 1992 @xcite : they found 2 events in the signal region . reconstructing @xmath147 combinations with the cuts of section [ section - selection ] tightened to require @xmath148
, we see a clear signal peak at the @xmath10 mass , as shown in fig .
[ fig_lamc_sigkpi ] .
the tighter cut is used to suppress the large combinatorial background .
we also form @xmath147 combinations using `` @xmath96 '' candidates from mass sidebands ( two 10 mev/@xmath149 intervals centered @xmath150 mev/@xmath149 below and above the nominal @xmath96 mass @xcite ) , shown with the shaded histogram : no enhancement is seen near the @xmath10 mass .
the mass distribution is fitted with a gaussian for the signal ( with width fixed to @xmath151 mev@xmath89 from the mc ) and a second order polynomial for the background : we find @xmath152 @xmath70 events . for normalization
we reconstruct @xmath153 decays with the same cuts , finding @xmath154 events .
the relative efficiency of the @xmath70 channel reconstruction with respect to @xmath153 is found to be @xmath155 in the mc . using this value ,
we obtain @xmath156 ( 550,200 ) ( 80,40 ) ( 135,0)@xmath157 $ ] ( 110,10 ) : invariant mass spectrum of the selected @xmath147 combinations .
the shaded histogram shows the equivalent spectrum for the @xmath96 sidebands .
the mass difference for the sidebands is corrected using the central value of the corresponding sideband interval.,title="fig:",scaledwidth=60.0% ]
the decays @xmath5 and @xmath160 proceed dominantly via w - exchange diagrams , and were observed by cleo in 1993 @xcite .
here we measure these decay channels with improved accuracy and provide the first evidence for the @xmath7 decay .
figure [ fig_lamc_sigkk ] shows the invariant mass spectrum for @xmath5 combinations selected according to section [ section - selection ] .
( 550,200 ) ( 80,40 ) ( 135,0)@xmath161 $ ] ( 110,10 ) : invariant mass spectrum of the selected @xmath162 combinations.,title="fig:",scaledwidth=60.0% ] a clear peak is seen at the @xmath10 mass , over a low background .
we fit the distribution using a gaussian ( with width fixed to 2.2 mev@xmath89 from the mc ) plus a second order polynomial : the fit yields @xmath163 @xmath5 decays . for normalization
we reconstruct the @xmath153 decay mode with equivalent cuts , and fit the distribution with a gaussian and a second order polynomial : we find @xmath164 @xmath153 events .
the relative efficiency of the @xmath5 decay reconstruction with respect to the @xmath153 decay is calculated by mc simulation and is found to be @xmath165 .
we thus extract @xmath166 in order to obtain the @xmath160 signal , we take @xmath162 from a @xmath167 mev@xmath89 window around the fitted @xmath10 mass ( 2286 mev@xmath89 ) , and plot the invariant mass of the @xmath168 combination , as shown in fig . [ fig_lamc_sigkk_mkk ]
( points with error bars ) ; the equivalent distribution is also shown for @xmath162 in 5 mev/@xmath149 sidebands ( 550,200 ) ( 80,40 ) ( 195,0)@xmath169 $ ] ( 110,10 ) component : the invariant mass spectra of @xmath168 combinations from the @xmath5 signal area ( points with error bars ) and @xmath10 sidebands ( shaded histogram ) are shown.,title="fig:",scaledwidth=60.0% ] centered 12.5 mev@xmath89 below and above the fitted @xmath10 mass ( shaded histogram ) .
the distributions are fitted with a breit - wigner function ( describing the @xmath170 signal ) convolved with a gaussian of fixed width ( representing the detector mass resolution ) plus a second order polynomial multiplied by a square root threshold factor .
the intrinsic width of the @xmath170 breit - wigner function is fixed to its nominal value @xcite , and the width of the gaussian resolution is fixed to 1.0 mev@xmath89 based on the mc simulation .
the fit yields @xmath171 events for the @xmath170 signal in the @xmath10 region and @xmath172 in the @xmath10 sidebands . to extract the @xmath160 contribution
we subtract the @xmath170 yield in the sidebands from the yield in the @xmath10 signal region , correcting for the phase space factor obtained from the @xmath162 background fitting function . after making a further correction for the missing signal outside the @xmath10 mass interval
, we obtain @xmath173 @xmath160 decays .
the relative efficiency of the @xmath160 reconstruction with respect to @xmath153 is calculated using the mc and found to be @xmath174 .
taking into account the @xmath170 branching fraction @xmath175 @xcite , we calculate @xmath176 we also search for resonant structure in the @xmath177 system in these decays .
figure [ fig_lamc_sigkk_sigk ] shows the @xmath177 invariant mass spectrum for @xmath162 combinations in a @xmath167 mev@xmath89 interval around the fitted @xmath10 mass ( data points ) : we also require @xmath178 to suppress @xmath179 .
( 550,200 ) ( 80,40 ) ( 145,0)@xmath180 $ ] ( 110,10 ) component : the invariant mass spectrum of @xmath177 combinations from the @xmath5 signal area ( points with error bars ) and @xmath10 sidebands ( shaded histogram ) are shown , with the @xmath179 signal region excluded in both cases.,title="fig:",scaledwidth=60.0% ] also shown is the @xmath177 invariant mass spectrum from @xmath162 combinations selected inside @xmath181 sideband intervals centered 12.5 mev@xmath89 below and above the fitted @xmath10 mass ( shaded histogram ) .
the @xmath177 mass distribution shows evidence for the @xmath182 resonant state . in order to extract this resonant contribution
the histograms are fitted with a relativistic breit - wigner function ( describing the @xmath182 signal ) plus a @xmath183 function multiplied by a square root threshold factor ( here @xmath184 is the maximal allowed value of the @xmath177 invariant mass ) .
the fit yields @xmath185 events for the @xmath182 signal in the @xmath10 region , with a fitted mass ( @xmath186 ) mev@xmath89 and width ( @xmath187 ) mev in good agreement with previous measurements of the @xmath182 parameters @xcite . to fit the sidebands ,
the function parameters are fixed to the central values obtained from the signal fit , and both the signal and background normalizations are floated .
a yield of @xmath188 events is found .
the @xmath7 contribution is obtained by subtracting the @xmath182 yield in the sidebands from the yield in the @xmath10 signal region , correcting the sideband contribution using the phase space factor obtained from the @xmath162 background fitting function .
after a further correction for the missing signal outside the @xmath10 mass interval , we obtain @xmath189 @xmath7 decays .
we then find @xmath190 the possible effects due to interference with @xmath160 are included in the systematic error ( see the discussion in section [ errors ] ) .
finally , the non - resonant @xmath5 contribution is estimated by making invariant mass cuts @xmath191 and @xmath192 to suppress the @xmath170 and @xmath182 contributions ( here , @xmath193 is the fitted @xmath182 mass ) .
the resulting @xmath162 mass spectrum is fitted with a gaussian ( with width fixed to 2.2 mev@xmath89 from the mc ) plus a second order polynomial .
the fit yields @xmath194 events . integrating the @xmath170 breit - wigner function over the allowed @xmath195 region , we find that @xmath196 of the total @xmath160 signal contributes to this sample : @xmath197 events .
the contribution of the @xmath182 mass tails is estimated to be approximately 12@xmath198 of the fitted @xmath182 signal : @xmath199 events .
subtracting these contributions , @xmath200 non - resonant events remain . the phase space correction factor to account for the missing region around the @xmath170 and @xmath182 masses
is found to be 1.63 by mc simulation of the non - resonant @xmath195 spectrum . applying this correction we obtain @xmath201 @xmath5 non - resonant decays .
taking into account the systematic error , we obtain an upper limit @xmath202 at the 90% confidence level , including the possible effects due to interference with @xmath160 in the systematic errors ( see section [ errors ] ) .
another possible decay mode of the @xmath182 resonant state is @xmath204 .
hence we have searched for the decay @xmath7 by reconstructing @xmath205 decays and looking at the @xmath206 invariant mass distribution . reconstructing @xmath207 combinations with the cuts of section [ section - selection ] we obtain an invariant mass spectrum , which is fitted with a gaussian for the signal and a second order polynomial for the background : we find @xmath208 @xmath205 events . in order to obtain the @xmath7 signal , we take @xmath209 from a @xmath210 mev@xmath89 window ( @xmath211 ) around the fitted @xmath10 mass ( 2287 mev@xmath89 ) , and plot the invariant mass of the @xmath206 combination , as shown in fig . [ fig_lamc_lamk_lamkk ]
( points with error bars ) ; the equivalent distribution is also shown for @xmath209 from @xmath212 mev@xmath89 sideband intervals centered @xmath150 mev@xmath89 below and above the fitted @xmath10 mass ( shaded histogram ) .
a peak at the expected position is clearly seen .
( 550,200 ) ( 80,40 ) ( 145,0)@xmath213 $ ] ( 110,10 ) component : the invariant mass spectrum of @xmath214 combinations from the @xmath215 signal area ( points with error bars ) and @xmath10 sidebands ( shaded histogram ) are shown . the dashed curve represents the background function.,title="fig:",scaledwidth=60.0% ] we use a fitting procedure similar to that described in section [ section - sigkk ] for the @xmath160 analysis .
after subtraction of the sideband contribution and corrections we obtain @xmath216 @xmath7 decays .
this confirms our observation of the @xmath7 decay : significant signals are seen for both @xmath217 and @xmath204 . using the normalization to the inclusive decay mode @xmath218 and the measured values for the @xmath219 and @xmath220 @xcite we find @xmath221 using the value of the @xmath222 combined branching ratio , obtained in section [ section - sigkk ] , and the ratio of the normalization decay rates @xcite
, we find the following ratio of @xmath182 decay rates : @xmath223 the corresponding ratio of the @xmath182 decay rates quoted by @xcite ( @xmath224 after isospin correction ) is based on a single measurement reported in @xcite . in order to check for possible interference effects
, we studied the corresponding @xmath119 invariant mass distribution and did not find any structure above a smooth background .
we have also searched for the @xmath7 decay in the @xmath225 decay mode , but did not find any @xmath182 signal in the @xmath226 invariant mass spectrum , in agreement with the @xmath227 upper limit value from @xcite .
the first evidence for the @xmath230 decay was reported by na32 in 1990 , who claimed a signal of @xmath231 events @xcite .
the decay @xmath3 was observed for the first time by e687 in 1993 , who also obtained an upper limit for the branching ratio of @xmath230 @xcite .
the most recent statistically significant resonant analysis was published by cleo in 1996 , who found the following branching ratios : @xmath232 and @xmath233 @xcite .
( 550,200 ) ( 80,40 ) ( 190,0)@xmath234 $ ] ( 110,10 ) : invariant mass spectrum of the selected @xmath235 combinations.,title="fig:",scaledwidth=60.0% ] reconstructing @xmath3 candidates according to the procedure described in section [ section - selection ] , we see a clear peak at the @xmath10 mass , as shown in fig .
[ fig_lamc_pkk ] .
we fit the distribution with a gaussian ( with width fixed to @xmath236 from the mc ) plus a second order polynomial , and find @xmath237 @xmath3 events . for normalization
we reconstruct the @xmath238 decay mode with equivalent cuts and fit the distribution with a double gaussian for the large signal peak , and a second order polynomial , finding @xmath239 events .
the relative efficiency of the @xmath240 decay reconstruction with respect to @xmath238 is found to be @xmath241 in the mc : using this value , we extract @xmath242 in order to obtain the @xmath230 signal we take @xmath235 from a @xmath243 window around the fitted @xmath10 mass ( 2286 mev@xmath89 ) , and plot the invariant mass of the @xmath244 combination , as shown in fig .
[ fig_lamc_pkk_mkk ] ( points with error bars ) ; the equivalent distribution is also shown for @xmath235 from @xmath245 sideband intervals centered @xmath246 below and above the fitted @xmath10 mass ( shaded histogram ) .
( 550,200 ) ( 80,40 ) ( 195,0)@xmath169 $ ] ( 110,10 ) component : the invariant mass spectra of @xmath168 combinations from the @xmath3 signal area ( points with error bars ) and sidebands ( shaded histogram).,title="fig:",scaledwidth=60.0% ] the distributions are fitted using a method similar to that used for the @xmath160 analysis ( section [ section - sigkk ] ) . after making a sideband subtraction and correction for the signal outside the @xmath10 mass interval we obtain @xmath247 @xmath230 decays . the reconstruction efficiency of the @xmath230 decay relative to @xmath238 was calculated using the mc and found to be @xmath165 . using this value ,
we extract @xmath248 the non-@xmath170 @xmath3 signal is estimated by making an invariant mass cut @xmath178 to suppress the @xmath179 contribution . after fitting the resulting @xmath235 mass spectrum and applying corrections accounting for the @xmath170 tails and the missing phase space region around the @xmath170 mass we obtain @xmath249 @xmath3 non-@xmath170 decays .
this corresponds to @xmath250
we have considered several possible sources for the systematic errors in our measurements .
the most important is the uncertainty in the pion and kaon identification efficiencies , which affects all ratios of signal and reference branchings .
based on a study of kaons and pions from @xmath251-tagged @xmath252 decays , we assign a systematic uncertainty of @xmath253 per @xmath73 ratio ( _ e.g. _ @xmath253 for @xmath254 , @xmath255 for @xmath256 ) .
possible biases due to fitting procedure have also been studied . in each fit , the shape of the background function has been varied by changing the order of the polynomial function , with any change in the signal yield being taken as a systematic uncertainty . for each fit where the width of the signal gaussian was fixed to the mc prediction , we have redone the fit with a floating width , and taken the resulting change in the yield as a systematic uncertainty . for the @xmath257 and @xmath119 analyses ,
we have assigned additional uncertainties of @xmath253 and @xmath258 respectively on the signal yields , based on the fractions of signal events found in non - gaussian tails for the normalization modes @xmath259 and @xmath260 ( the @xmath235 and @xmath119 samples are too small to fit for the presence of non - gaussian tails ) . for the breit - wigner fit of the @xmath170 signal we have varied the function by letting the width of the convolved gaussian float and varying the shape of the background parameterization .
we have also included the @xmath261 uncertainty of @xmath262 and varied the @xmath170 nominal width within its error @xcite . in the case of @xmath160 and @xmath263
decays there is an additional source of systematic error due to the difference in kinematics between the signal and normalization modes .
this has been estimated to be @xmath253 for @xmath159 and @xmath264 for @xmath263 , based on the difference between the mc predictions for the efficiency in resonant and non - resonant cases . in the @xmath222 resonant analysis we neglected the possible interference between @xmath182 and @xmath170 contributions .
mc studies show that this leads to an uncertainty of less than @xmath265 , due to phase space limitations in the interference region .
in summary , we report the first observation of the cabibbo - suppressed decays @xmath128 and @xmath1 , and the first observation of @xmath2 with large statistics .
the decays @xmath3 , @xmath4 and @xmath266 , and the w - exchange decays @xmath5 and @xmath160 have been measured with the best accuracy to date .
we have also observed evidence for the decay @xmath7 and set an upper limit on the non - resonant decay mode @xmath5 .
the results for these decay modes are listed in table 1 .
we wish to thank the kekb accelerator group for the excellent operation of the kekb accelerator .
we acknowledge support from the ministry of education , culture , sports , science , and technology of japan and the japan society for the promotion of science ; the australian research council and the australian department of industry , science and resources ; the department of science and technology of india ; the bk21 program of the ministry of education of korea and the chep src program of the korea science and engineering foundation ; the polish state committee for scientific research under contract no.2p03b 17017 ; the ministry of science and technology of russian federation ; the national science council and the ministry of education of taiwan ; and the u.s .
department of energy .
y. kohara , nuovo cim . * a * 111 ( 1998 ) 67 ; + m. a. ivanov et al .
, phys . rev . *
d * 57 ( 1998 ) 5632 ; + k. k. sharma and r. c. verma , phys . rev . *
d * 55 ( 1997 ) 7067 ; + l. chau et al .
, phys . rev .
* d * 54 ( 1996 ) 2132 ; + a. datta , hep - ph/9504428 ; + t. uppal et al .
, phys . rev .
* d * 49 ( 1994 ) 3417 ; + p. zenczykowski , phys .
* d * 50 ( 1994 ) 402 ; + j. krner and m. krmer , z. phys .
* c * 55 ( 1992 ) 659 . | we present measurements of the cabibbo - suppressed decays @xmath0 and @xmath1 ( both first observations ) , @xmath2 ( seen with large statistics for the first time ) , @xmath3 and @xmath4 ( measured with improved accuracy ) .
improved branching ratio measurements for the decays @xmath5 and @xmath6 , which are attributed to w - exchange diagrams , are shown .
we also present the first evidence for @xmath7 and set an upper limit on the non - resonant decay @xmath5 .
this analysis was performed using 32.6 fb@xmath8 of data collected by the belle detector at the asymmetric @xmath9 collider kekb .
-3 cm belle preprint 2001 - 17 + kek preprint 2001 - 137 + * observation of cabibbo - suppressed and w - exchange + @xmath10 baryon decays * the belle collaboration + k. abe@xmath11 , k. abe@xmath12 , r. abe@xmath13 , t. abe@xmath14 , i. adachi@xmath11 , byoung sup ahn@xmath15 , h. aihara@xmath16 , m. akatsu@xmath17 , y. asano@xmath18 , t. aso@xmath19 , v. aulchenko@xmath20 , t. aushev@xmath21 , a. m. bakich@xmath22 , y. ban@xmath23 , e. banas@xmath24 , s. behari@xmath11 , p. k. behera@xmath25 , a. bondar@xmath20 , a. bozek@xmath24 , t. e. browder@xmath26 , b. c. k. casey@xmath26 , p. chang@xmath27 , y. chao@xmath27 , b. g. cheon@xmath28 , r. chistov@xmath21 , s .- k .
choi@xmath29 , y. choi@xmath28 , l. y. dong@xmath30 , a. drutskoy@xmath21 , s. eidelman@xmath20 , v. eiges@xmath21 , y. enari@xmath17 , f. fang@xmath26 , h. fujii@xmath11 , c. fukunaga@xmath31 , m. fukushima@xmath32 , n. gabyshev@xmath11 , a. garmash@xmath33 , t. gershon@xmath11 , a. gordon@xmath34 , r. guo@xmath35 , j. haba@xmath11 , h. hamasaki@xmath11 , f. handa@xmath14 , k. hara@xmath36 , t. hara@xmath36 , n. c. hastings@xmath34 , h. hayashii@xmath37 , m. hazumi@xmath26 , e. m. heenan@xmath34 , i. higuchi@xmath14 , t. higuchi@xmath16 , t. hojo@xmath36 , t. hokuue@xmath17 , y. hoshi@xmath12 , k. hoshina@xmath38 , s. r. hou@xmath27 , w .- s .
hou@xmath27 , s .- c .
hsu@xmath27 , h .- c .
huang@xmath27 , y. igarashi@xmath11 , t. iijima@xmath11 , h. ikeda@xmath11 , k. inami@xmath17 , a. ishikawa@xmath17 , h. ishino@xmath39 , r. itoh@xmath11 , h. iwasaki@xmath11 , y. iwasaki@xmath11 , d. j. jackson@xmath36 , p. jalocha@xmath24 , h. k. jang@xmath40 , r. kagan@xmath21 , j. h. kang@xmath41 , j. s. kang@xmath15 , p. kapusta@xmath24 , n. katayama@xmath11 , h. kawai@xmath42 , h. kawai@xmath16 , n. kawamura@xmath43 , t. kawasaki@xmath13 , h. kichimi@xmath11 , d. w. kim@xmath28 , heejong kim@xmath41 , h. j. kim@xmath41 , h. o. kim@xmath28 , hyunwoo kim@xmath15 , s. k. kim@xmath40 , t. h. kim@xmath41 , k. kinoshita@xmath44 , h. konishi@xmath38 , s. korpar@xmath45 , p. krian@xmath46 , p. krokovny@xmath20 , r. kulasiri@xmath44 , s. kumar@xmath47 , a. kuzmin@xmath20 , y .- j .
kwon@xmath41 , j. s. lange@xmath48 , g. leder@xmath49 , s. h. lee@xmath40 , d. liventsev@xmath21 , r .- s .
lu@xmath27 , j. macnaughton@xmath49 , t. matsubara@xmath16 , s. matsumoto@xmath50 , t. matsumoto@xmath17 , y. mikami@xmath14 , k. miyabayashi@xmath37 , h. miyake@xmath36 , h. miyata@xmath13 , g. r. moloney@xmath34 , s. mori@xmath18 , t. mori@xmath50 , t. nagamine@xmath14 , y. nagasaka@xmath51 , y. nagashima@xmath36 , t. nakadaira@xmath16 , e. nakano@xmath52 , m. nakao@xmath11 , j. w. nam@xmath28 , z. natkaniec@xmath24 , k. neichi@xmath12 , s. nishida@xmath53 , o. nitoh@xmath38 , s. noguchi@xmath37 , s. ogawa@xmath54 , t. ohshima@xmath17 , t. okabe@xmath17 , s. okuno@xmath55 , s. l. olsen@xmath26 , w. ostrowicz@xmath24 , h. ozaki@xmath11 , p. pakhlov@xmath21 , h. palka@xmath24 , c. s. park@xmath40 , c. w. park@xmath15 , h. park@xmath56 , k. s. park@xmath28 , l. s. peak@xmath22 , j .- p .
perroud@xmath57 , m. peters@xmath26 , l. e. piilonen@xmath58 , n. root@xmath20 , m. rozanska@xmath24 , k. rybicki@xmath24 , j. ryuko@xmath36 , h. sagawa@xmath11 , y. sakai@xmath11 , h. sakamoto@xmath53 , m. satapathy@xmath25 , o. schneider@xmath57 , s. schrenk@xmath44 , s. semenov@xmath21 , k. senyo@xmath17 , m. e. sevior@xmath34 , h. shibuya@xmath54 , b. shwartz@xmath20 , j. b. singh@xmath47 , s. stani@xmath18 , a. sugi@xmath17 , a. sugiyama@xmath17 , k. sumisawa@xmath11 , t. sumiyoshi@xmath11 , k. suzuki@xmath42 , s. suzuki@xmath59 , s. y. suzuki@xmath11 , s. k. swain@xmath26 , t. takahashi@xmath52 , f. takasaki@xmath11 , m. takita@xmath36 , k. tamai@xmath11 , n. tamura@xmath13 , j. tanaka@xmath16 , m. tanaka@xmath11 , y. tanaka@xmath60 , g. n. taylor@xmath34 , y. teramoto@xmath52 , m. tomoto@xmath11 , t. tomura@xmath16 , s. n. tovey@xmath34 , t. tsuboyama@xmath11 , t. tsukamoto@xmath11 , s. uehara@xmath11 , k. ueno@xmath27 , y. unno@xmath42 , s. uno@xmath11 , y. ushiroda@xmath11 , s. e. vahsen@xmath61 , k. e. varvell@xmath22 , c. c. wang@xmath27 , c. h. wang@xmath62 , j. g. wang@xmath58 , m .- z .
wang@xmath27 , y. watanabe@xmath39 , e. won@xmath40 , b. d. yabsley@xmath11 , y. yamada@xmath11 , m. yamaga@xmath14 , a. yamaguchi@xmath14 , y. yamashita@xmath63 , m. yamauchi@xmath11 , s. yanaka@xmath39 , j. yashima@xmath11 , m. yokoyama@xmath16 , k. yoshida@xmath17 , y. yuan@xmath30 , y. yusa@xmath14 , c. c. zhang@xmath30 , j. zhang@xmath18 , h. w. zhao@xmath11 , y. zheng@xmath26 , v. zhilich@xmath20 , and d. ontar@xmath18 @xmath43aomori university , aomori + @xmath20budker institute of nuclear physics , novosibirsk + @xmath42chiba university , chiba + @xmath50chuo university , tokyo + @xmath44university of cincinnati , cincinnati oh + @xmath48university of frankfurt , frankfurt + @xmath29gyeongsang national university , chinju + @xmath26university of hawaii , honolulu hi + @xmath11high energy accelerator research organization ( kek ) , tsukuba + @xmath51hiroshima institute of technology , hiroshima + @xmath32institute for cosmic ray research , university of tokyo , tokyo + @xmath30institute of high energy physics , chinese academy of sciences , beijing + @xmath49institute of high energy physics , vienna + @xmath21institute for theoretical and experimental physics , moscow + @xmath64j .
stefan institute , ljubljana + @xmath55kanagawa university , yokohama + @xmath15korea university , seoul + @xmath53kyoto university , kyoto + @xmath56kyungpook national university , taegu + @xmath57iphe , university of lausanne , lausanne + @xmath65university of ljubljana , ljubljana + @xmath66university of maribor , maribor + @xmath34university of melbourne , victoria + @xmath60nagasaki institute of applied science , nagasaki + @xmath17nagoya university , nagoya + @xmath37nara women s university , nara + @xmath35national kaohsiung normal university , kaohsiung + @xmath62national lien - ho institute of technology , miao li + @xmath27national taiwan university , taipei + @xmath24h .
niewodniczanski institute of nuclear physics , krakow + @xmath63nihon dental college , niigata + @xmath13niigata university , niigata + @xmath52osaka city university , osaka + @xmath36osaka university , osaka + @xmath47panjab university , chandigarh + @xmath23peking university , beijing + @xmath61princeton university , princeton nj + @xmath40seoul national university , seoul + @xmath28sungkyunkwan university , suwon + @xmath22university of sydney , sydney nsw
+ @xmath54toho university , funabashi + @xmath12tohoku gakuin university , tagajo + @xmath14tohoku university , sendai + @xmath16university of tokyo , tokyo + @xmath39tokyo institute of technology , tokyo + @xmath31tokyo metropolitan university , tokyo + @xmath38tokyo university of agriculture and technology , tokyo + @xmath19toyama national college of maritime technology , toyama + @xmath18university of tsukuba , tsukuba + @xmath25utkal university , bhubaneswer + @xmath58virginia polytechnic institute and state university , blacksburg va + @xmath59yokkaichi university , yokkaichi + @xmath41yonsei university , seoul + |
@xmath1-cellular maps can be viewed as drawings on a topological surface , they represent a cell - complex of the latter and inherit the topological genus of the surface as their geometric realization . in a seminal paper harer and zagier
@xcite computed the virtual euler characteristic of the moduli space of curves , independently derived by penner @xcite and still lack a combinatorial interpretation .
key object here play unicellular maps @xcite of genus @xmath2 with @xmath3 edges , @xmath4 , i.e. fatgraphs@xcite with a unique boundary component .
most prominantly here is the recursion @xmath5 in @xcite the generating function of unicellular maps is obtained as @xmath6 where @xmath7 is polynomial defined over the integers of degree at most @xmath8 that is divisible by @xmath9 with @xmath10 , @xmath11p_g(z ) \neq 0 $ ] and @xmath12p_g(z ) = 0 $ ] for @xmath13 .
matrix - theory @xcite , via the schwinger - dyson equation or representation theorey @xcite , connects the generating functions of unicellular , @xmath14 , and bicellular maps , @xmath15 .
the latter counts fatgraphs having two boundary components that are connected as combinatorial graphs .
the relation can also be proved using the representation theoretic framework of zagier @xcite and is given by @xmath16 recently @xcite the authors presented a bijective proof of the corresponding coefficient equation @xmath17 which revealed a simple construction mechanism .
the bijective proof can for instance be applied , to significantly speed up the folding of rna interaction structures @xcite .
an analogous relation between unicellular , bicellular and tricellular maps can also be obtained via matrix - theory . in this paper
we give a bijective proof of this relation which reads @xmath18 where @xmath19 is explicitly expressed via numbers of unicellular and bicellular maps .
our strategy is to derive a partition of the set of unicellular maps of genus @xmath20 with @xmath21 edges , see fig .
[ f : visualpar ] for a first step of how to decompose the latter . .
] it is interesting to note that matrix - theory does not provide any insight w.r.t .
for instance quadricellular maps .
it seems in fact unlikely that such relations can be derived using this formal framework .
the bijective proof presented here however is rather straightforward once the correct partitioning is identified .
we believe that it is very well possible to prove similar relations for cellular maps with more than three boundary components .
let @xmath22 denote the permutation group over @xmath23 elements .
let @xmath24 be positive integers .
a @xmath1-cellular map is a triple @xmath25 , where @xmath26 is a set of cardinality @xmath23 , @xmath27 a fixed - point free involution and @xmath28 are cycles such that @xmath29 .
the elements of @xmath26 are half - edges , the cycles of @xmath27 are edges .
the cycles of the permutation @xmath30 are the vertices @xmath31 , @xmath32 .
the length of @xmath31 is its degree .
the cycle @xmath28 is the @xmath33-th face .
the combinatorial graph @xmath34 of a @xmath1-cellular map is the graph whose edges and vertices are the cycles of @xmath27 and @xmath35 .
we can regard a @xmath34-edge as a ribbon whose two sides are labeled by the half - edges as follows : each side of the ribbon represents one half - edge , we decide which half - edge corresponds to which side of the ribbon by the convention that , if a half - edge @xmath36 belongs to a cycle @xmath37 of @xmath27 and a certain @xmath38 of @xmath35 , then @xmath36 is the right - hand side of the ribbon corresponding to @xmath37 , when entering @xmath38 .
furthermore , around each vertex @xmath38 , the counterclockwise ordering of the half - edges belonging to the cycle @xmath38 is given by that cycle , we obtain a graphical object called the fatgraph , @xmath39 , tantamount to @xmath40 and the graph @xmath34 is the corresponding combinatorial graph of @xmath39 .
a planted @xmath1-cellular map is a @xmath1-cellular map in which each @xmath28 contains a distinguished half - edge @xmath41 , such that @xmath42 is a @xmath35-cycle .
@xmath42 is called the plant of the face @xmath28 and @xmath35-cycles , except of the plants are called @xmath43-vertices . in the following , we refer to edges not incident to plants as @xmath43-edges .
let @xmath44 denote the set of planted @xmath1-cellular maps that contain @xmath3 @xmath43-edges . in planted maps we shall label the half - edges of @xmath26 such that @xmath45 , that is @xmath46 given @xmath47 we define the linear order @xmath48 on @xmath26 for each face @xmath28 via : @xmath49 let furthermore @xmath50 denote the set consisting the half - edges in one of these @xmath28 . in particular
, @xmath51 is the set of half - edges contained in the face @xmath28 .
there is a natural equivalence relation over half - edges , @xmath52 and in particular , @xmath53 .
if @xmath54 , then @xmath55 is called a one - sided edge and @xmath55 is called a two - sided edge , otherwise . for each vertex @xmath56 ,
let @xmath57 denote the first half - edge via which @xmath28 enters @xmath56 .
this gives a canonical way of writing the cycle , starting at @xmath58 namely @xmath59 .
in particular , the vertex containing the half - edge @xmath60 is @xmath61 , the `` first '' vertex .
@xmath62-cellular maps are also called unicellular maps @xcite .
let @xmath63 denote the set of planted , unicellular maps of genus @xmath2 , having @xmath3 @xmath43-edges . in particular ,
let @xmath64 denote the unicellular map of genus zero , containing no @xmath43-edge .
this map contains only one edge , the plant , and one additional @xmath43-vertex .
let @xmath65 with face @xmath66 $ ] . then @xmath67 where @xmath68 .
thus @xmath69 and @xmath70 $ ] . in the following we shall identify a partition of @xmath71 that will facilitate our main bijection in theorem [ t : result ] . to begin
, we consider for @xmath72 the four half edges @xmath73 , @xmath74 , @xmath75 and @xmath76 .
clearly , @xmath77 , whence @xmath78 .
furthermore , by construction , @xmath79 see also fig .
[ f : branch ] .
accordingly , there are the two scenarios @xmath80 the case @xmath81 belongs to scenario @xmath82 , which then reduces to @xmath83 this generates the bipartition of @xmath71 , @xmath84 in @xmath85-elements the half - edges @xmath74 and @xmath73 belong to two different vertices , @xmath61 and @xmath86 .
we have @xmath87 and @xmath88 .
suppose now @xmath74 and @xmath73 belong to @xmath61 .
then there exists a half - edge @xmath89 satisfying @xmath90 such that @xmath91 or @xmath92 , but this implies @xmath93 , a contradiction .
we next refine @xmath94 : for @xmath95 , we consider the cycle @xmath96 and we use @xmath97 to split the @xmath98 into @xmath99 see fig .
[ f : branch ] . and @xmath100 .
] suppose the restriction @xmath101 is a welldefined fixed - point free involution , then we call @xmath102 closed .
similarly , the sets @xmath50 and @xmath103 , @xmath104 are called closed , if @xmath105 and @xmath106 are fixed - point free involutions .
let @xmath107 denote the subset of @xmath94-elements in which no @xmath103 is closed and let @xmath108 denote its complement .
then @xmath109 we refine @xmath110 further : * @xmath111 : the set of @xmath108-elements in which exactly two @xmath112 are empty , * @xmath113 : the set of @xmath108-elements in which exactly one @xmath114 is empty , * @xmath115 : the complement of @xmath111 and @xmath113 , that is , the set of @xmath108 in which no @xmath114 is empty . thus @xmath116 we refine @xmath117 a bit more , for this purpose let * @xmath118 : be the subset of @xmath119-elements in which @xmath120 . * @xmath121 : be the subset of @xmath119-elements in which @xmath122 , @xmath123 and @xmath124 .
* @xmath125 : the complement of @xmath118 and @xmath121 , that is subset of @xmath126-elements in which @xmath122 and @xmath127 , @xmath128 .
accordingly , @xmath129 furthermore we present @xmath130 : @xmath131 where * @xmath132 denotes the subset of @xmath108-elements in which @xmath120 , * @xmath133 denotes the subset of @xmath113-elements in which @xmath120 .
furthermore we present @xmath134 as @xmath135 where * @xmath136 is the subset of @xmath108-elements with @xmath124 , * @xmath137 is the subset of @xmath113-elements with @xmath124 , * @xmath138 is the subset of @xmath126-elements with @xmath120 and @xmath124 .
in this section we state three procedures that are employed repeatedly in our bijection .
they are `` cutting '' , `` contraction '' and `` deletion '' .
these procedures constitute the key three operations that , applied in various contexts , facilitate the bijection .
[ l : alpha - cutting]*(cutting ) * suppose we are given a planted , unicellular map @xmath139 with @xmath140 then @xmath141 can be mapped to a planted , @xmath142-cellular map , @xmath143 , with the three faces @xmath144 via @xmath145 where @xmath146 furthermore , the mapping @xmath147 has the following inverse : @xmath148 by assumption we have @xmath149 whence the face of @xmath150 can be written as in eq . ( [ e : face ] ) .
we use @xmath97 and @xmath151 which are given by [ e:3ghs ] , then concatenate the sequence of half - edges of @xmath152 , @xmath153 and @xmath154 to form @xmath155 and relabel the cycles as in eq .
( [ e : gamma123 ] ) .
this produces the plants @xmath154 , @xmath156 and @xmath157 . since @xmath158
, @xmath159 is a @xmath142-cellular map , @xmath147 is well - defined , see fig .
[ f : a - cutting ] . and @xmath160 .
] we next construct an explicit inverse of @xmath147 .
suppose we are given a @xmath142-cellular map @xmath161 , in which the @xmath28 are as in eq .
( [ e : h - bc ] ) . then we concatenate the sequences of half - edges of the three @xmath28-cycles and relabel as in eq .
( [ e : face ] ) , i.e. @xmath162 , @xmath163 and @xmath164 .
we derive , by construction , @xmath165 accordingly , @xmath166 is a unicellular map of genus @xmath20 with property @xmath82 .
[ l : contract2 ] * ( contraction ) * suppose @xmath167 has a one - sided edge @xmath168 , @xmath169 , such that @xmath170 and @xmath171 are incident to two different vertices @xmath172 . relabeling the two half - edges we can write the face @xmath173 here either @xmath174 or @xmath175 , @xmath176 or @xmath177 and either @xmath178 or @xmath179
. then @xmath141 corresponds to a unicellular map @xmath180 together with two distinguished half - edges via mapping @xmath181 where @xmath182 , @xmath183 , @xmath184 and @xmath185 , if @xmath186 , @xmath187 , if @xmath188 and @xmath189 , @xmath190 , @xmath191 , if @xmath192 and @xmath193 , and finally @xmath194 , if @xmath195 .
+ furthermore the mapping @xmath196 @xmath197 has the property @xmath198 .
@xmath199 is by construction unicellular and retains the genus of @xmath200 .
we describe the contraction in fig .
[ f : contract ] . and
@xmath196 . the edge @xmath168 ( green ) is one - sided edge . ]
[ l : beta - cutting]*(deletion ) * given a unicellular map @xmath201 with face @xmath202 where @xmath174 or @xmath175 , @xmath176 or @xmath177 and @xmath203 or @xmath179 .
+ then @xmath141 corresponds to a unicellular map @xmath204 together with two half - edges @xmath205 and @xmath206 , where @xmath207 , via the mapping @xmath208 where @xmath209 , @xmath210 and @xmath211 @xmath212 can be reversed by mapping a unicellular map @xmath213 , together with two arbitrary half - edges @xmath205 and @xmath206 ( @xmath214 ) as follows : @xmath215 by construction @xmath216 , @xmath217 is a fixed - point free involution and @xmath218 has cardinality @xmath219 , whence @xmath220 is unicellular .
euler characteristic implies that the genus of @xmath220 is @xmath221 .
moreover , we have in case of @xmath186 , @xmath222 , in case of @xmath188 @xmath189 , @xmath223 , in case of @xmath192 and @xmath193 , @xmath224 , @xmath225 , and in case of @xmath226 we have @xmath227 , see fig . [
f : typeb ] . and @xmath228 .
the deleting edges are @xmath229 ( blue ) and @xmath100 ( red ) .
the gray dotted lines denote a sequence of half - edges connecting two vertices . ]
given a unicellular map @xmath230 , there are @xmath231 ways to choose @xmath232 such that @xmath214 .
we now select two half - edges @xmath232 such that @xmath233 and insert the pairs of half - edges @xmath234 , @xmath235 into the face @xmath236 .
this produces the face @xmath237 , with @xmath238 , @xmath239 , @xmath240 and @xmath162 , @xmath241 and @xmath242 .
consequently we have @xmath243 we then relabel @xmath237 as in eq .
( [ e : delete - face1 ] ) .
since @xmath27 is a fixed - point free involution and @xmath26 is a set of cardinality @xmath244 , @xmath245 is a unicellular map with property @xmath246 .
euler characteristic implies @xmath247 has genus @xmath20 . by construction
, we have @xmath248 .
in this section we state some auxiliary bijections and our main result .
we furthermore give in fig .
[ f : algo ] an modular description of how our bijection works .
we call a planted @xmath249-cellular map , whose combinatorial graph is connected , a planted , bicellular map .
let @xmath250 denote the set of planted , bicellular maps of genus @xmath2 with @xmath3 @xmath43-edges .
let @xmath251 denote the subset of @xmath252 in which only a single @xmath103 , @xmath253 is closed and let @xmath254 denote the set of @xmath252-elements in which all @xmath103 , @xmath104 are closed , i.e. @xmath255 .
[ l : disconnect ] we have the bijections @xmath256 we prove lemma [ l : disconnect ] in section [ s : proofs ] .
[ l : add-1edge ] there are fours bijections , @xmath257 for @xmath258 , @xmath259 @xmath260 @xmath261 @xmath262 we prove lemma [ l : add-1edge ] in section [ s : proofs ] .
[ l : add-2edge ] we have the three bijections : @xmath263 @xmath264 @xmath265 we prove lemma [ l : add-2edge ] in section [ s : proofs ] . a planted @xmath142-cellular map that is connected as a combinatorial graph is called a planted tri - cellular map .
let @xmath266 denote the set of planted , tricellular maps of genus @xmath2 with @xmath3 @xmath43-edges .
[ p : tricellular ] there is a bijection @xmath267 we prove proposition [ p : tricellular ] in section [ s : proofs ] .
[ p : insert2edges ] there is a bijection @xmath268 we prove proposition [ p : tricellular ] in section [ s : proofs ] . in figure
[ f : algo ] we give an overview of how the above bijections are applied . for a set @xmath269
we denote its cardinality by @xmath270 .
[ t : result ] @xmath271 where @xmath272 with @xmath273
* proof of lemma [ l : disconnect ] * * claim @xmath62 : * the mapping @xmath274 is a bijection .
we first prove that @xmath275 is welldefined . for a planted unicellular map @xmath276 with face @xmath277 we employ the mapping @xmath147 of the cutting - lemma ( lemma [ l : alpha - cutting ] ) in order to decompose @xmath150 into a planted @xmath142-cellular map , @xmath278 , where @xmath279 where @xmath280 is obtained by concatenating the sequence of half - edges of @xmath152 , @xmath153 and @xmath154 . for any @xmath276 , @xmath281 is closed . since @xmath282 ,
the restriction @xmath283 is a fixed - point free involution .
accordingly , @xmath284 is a planted unicellular map . since @xmath285 is closed and @xmath28 , @xmath235 is given in eq .
( [ e : bc - p1 ] ) , the restriction @xmath286 is a welldefined fixed - point free involution .
furthermore , since neither @xmath287 nor @xmath288 are closed , @xmath289 and @xmath290 are not closed either .
therefore @xmath291 is a planted bicellular map with the plants @xmath154 and @xmath292 .
let @xmath293 and @xmath294 .
suppose @xmath150 , @xmath295 and @xmath296 have @xmath297 , @xmath298 and @xmath299 vertices , respectively .
then @xmath300 and @xmath301 , whence @xmath302 since the edges incident to plants and plants do not contribute to the number of edges and vertices , we have @xmath303 , @xmath304 . as a result
@xmath296 has genus @xmath305 , where @xmath306 , whence @xmath307 is welldefined .
we next show that @xmath275 is injective . in order to apply the mapping @xmath147 of the cutting - lemma
, we introduce @xmath308 where @xmath28 are given by eq .
( [ e : bc - p1 ] ) , @xmath309 and @xmath310 . for any @xmath311 where @xmath293 , and @xmath312
, we apply @xmath313 .
this generates the @xmath142-cellular map @xmath314 .
since @xmath295 has @xmath315 edges and @xmath316 has @xmath317 edges , and the process generates the edges @xmath229 and @xmath100 , we have @xmath318
. we can now apply @xmath147 of lemma [ l : alpha - cutting ] , which induces the mapping @xmath319 .
lemma [ l : alpha - cutting ] now implies furthermore @xmath320 whence the mapping @xmath275 is injective .
it thus remains to prove that @xmath275 is surjective .
this follows again from close inspection of the proof of the lemma [ l : alpha - cutting ] , which implies @xmath321 therefore , @xmath275 is surjective and claim @xmath62 is completed .
analogously we prove that @xmath322 and @xmath323 are injective .
* claim @xmath249 * : the mapping @xmath324 with @xmath325 and @xmath326 is a bijection .
we first show that @xmath327 is well - defined . as in the proof of claim @xmath62
, we employ the cutting - lemma which produces a @xmath142-cellular map with the boundary components @xmath328 . for any @xmath329 , each of the @xmath103 is closed .
thus the restrictions @xmath330 , for @xmath104 are welldefined and fixed - point free involutions . as a result , @xmath331 ,
@xmath332 and @xmath333 are unicellular maps , respectively .
let @xmath334 , @xmath335 and @xmath336 .
suppose that @xmath150 , @xmath337 , @xmath338 and @xmath339 have @xmath297 , @xmath298 , @xmath340 and @xmath341 vertices , respectively .
then @xmath342 furthermore , we have @xmath343 after applying the cutting - lemma , @xmath344 and @xmath292 become plants , similarly @xmath229 and @xmath100 become edges incident to plants .
thus , we have @xmath345 and @xmath346 and accordingly obtain @xmath347 consequently , @xmath348 has genus @xmath349 , where @xmath350 and @xmath351 is well - defined .
we next prove @xmath351 is injective .
we establish this as in claim @xmath62 , introducing @xmath352 where @xmath28 is given in eq .
( [ e : bc - p1 ] ) , @xmath309 and @xmath310 .
analogously , @xmath147 of lemma [ l : alpha - cutting ] induces the mapping @xmath353 and @xmath354 whence the mapping @xmath327 is injective .
subjectivity of @xmath351 is implied by the cutting - lemma which guarantees @xmath355 whence claim @xmath249 and the proof of the lemma is complete .
* proof of proposition [ p : tricellular]*. we prove that the mapping @xmath356 is a bijection . as for welldefinedness , suppose @xmath357 where @xmath358
we use mapping @xmath147 of the cutting - lemma and derive the planted @xmath142-cellular map , @xmath359 , where @xmath360 here @xmath280 is obtained by concatenating the sequence of half - edges contained in @xmath152 , @xmath153 and @xmath154 . for @xmath361 none of the @xmath103 , @xmath253
is closed , whence the associated combinatorial graph of @xmath362 is connected .
accordingly , @xmath363 is a planted tricellular map with plants @xmath344 , @xmath292 and @xmath154 .
euler s characteristic formula implies @xmath364 has genus @xmath2 and @xmath3 edges , whence @xmath365 is well - defined .
injectivity and surjectivity of @xmath365 are implied by the cutting lemma .
* proof of lemma [ l : add-1edge ] . * * claim @xmath62 : * the mapping @xmath366 is a bijection .
the contraction lemma implies that @xmath367 is welldefined .
injectivity of @xmath367 follows by considering the mapping @xmath368 induced by the mapping @xmath196 of lemma [ l : contract2 ] , where @xmath369 .
lemma [ l : contract2 ] guarantees @xmath370 , whence @xmath367 is injective .
surjectivity of @xmath367 is a consequence of @xmath371 , implied by lemma [ l : contract2 ] , see fig .
[ f : one ] . and the one - sided edge @xmath372 ( green ) .
] the proof that @xmath373 is a bijection for @xmath374 follows analogously , see fig .
[ f:2 - 3 - 4 ] . , @xmath375 and @xmath376 . ] * the proof of lemma [ l : add-2edge ] . * * claim @xmath62 : * the mapping @xmath377 is a bijection .
we first prove that @xmath378 is welldefined .
consider @xmath379 together with two one - sided edges , @xmath229 , @xmath100 , such that @xmath380 and @xmath381 are incident to @xmath61 , @xmath382 and @xmath383 are incident to @xmath31 and @xmath56 .
we then apply lemma [
l : contract2 ] to a @xmath384 together with the one - side edge @xmath229 .
we iterate applying lemma [ l : contract2 ] w.r.t . the edge @xmath100 . by definition of lemma [ l : contract2
] this generates the unicellular map
@xmath385 of genus @xmath20 having @xmath3 edges with distinguished four half - edges @xmath386 , @xmath387 , @xmath388 and @xmath389 .
since lemma [ l : contract2 ] preserves genus , @xmath390 has genus @xmath20 and @xmath3 edges , whence @xmath378 is well - defined .
we next prove @xmath378 is injective .
suppose we have a unicellular map @xmath391 with four distinguished half - edges @xmath386 , @xmath387 , @xmath388 and @xmath389 .
we observe that the mapping @xmath392 constructed in lemma [ l : contract2 ] allows us to obtain a mapping @xmath393 such that @xmath394 whence injectivity .
surjectivity follows by computing @xmath395 .
the proof that @xmath396 are bijections is analogous , see fig .
[ f:5 - 6 - 7 ] . , @xmath397 and @xmath398 .
] * the proof of proposition [ p : insert2edges]*. proposition [ p : insert2edges ] follows directly from lemma [ l : beta - cutting ] . * the proof of theorem [ t : result ] . *
according to lemma [ l : disconnect ] , lemma [ l : add-1edge ] , lemma [ l : add-2edge ] , proposition [ p : tricellular ] and proposition [ p : insert2edges ] , we have * @xmath399 , for @xmath400 , * @xmath401 , for @xmath400 , * @xmath402 , for @xmath403 , * @xmath404 , * @xmath405 . according to eq .
( [ e : ab ] ) , eq .
( [ e : i - ii ] ) and eq .
( [ f : disu ] ) we have @xmath406 furthermore , according to eq .
( [ e : geq2 ] ) , eq .
( [ e : u - g2 - 3 ] ) and eq .
( [ e : u - g3 - 4 ] ) , we have @xmath407 which establishes eq .
( [ e : result ] ) .
many thanks to our group at sdu for discussions and suggestions .
we furthermore acknowledge the financial support of the future and emerging technologies ( fet ) programme within the seventh framework programme ( fp7 ) for research of the european commission , under the fet - proactive grant agreement topdrim , number fp7-ict-318121 . | in this paper we give a bijective proof for a relation between uni- bi- and tricellular maps of certain topological genus .
while this relation can formally be obtained using matrix - theory as a result of the schwinger - dyson equation , we here present a bijection for the corresponding coefficient equation .
our construction is facilitated by repeated application of a certain cutting , the contraction of edges , incident to two vertices and the deletion of certain edges . * a bijection for tri - cellular maps * + hillary s. w. han and christian m. reidys@xmath0 department of mathematics and computer science + university of southern denmark , campusvej 55 , + dk-5230 , odense m , denmark + phone@xmath0 : 45 - 24409251 + fax@xmath0 : 45 - 65502325 + email@xmath0 : [email protected] |
when taxicab geometry is mentioned , the context is often a contrast with two - dimensional euclidean geometry . whatever the emphasis , the depth of the discussion frequently does not venture far beyond how the measurement of length is affected by the usual euclidean metric being replaced with the taxicab metric where the distance between two points @xmath0 and @xmath1 is given by @xmath2 papers have appeared that explore length in taxicab geometry in greater detail @xcite , and a few papers have derived area and volume formulae for certain figures and solids using exclusively taxicab measurements @xcite .
however , this well - built foundation has left a number of open questions . what is the taxicab length of a curve in two dimensions ? in three dimensions ?
what does area actually mean in two - dimensional taxicab geometry ?
is the taxicab area of a `` flat '' surface in three dimensions comparable to the taxicab area of the `` same '' surface ( in a euclidean sense ) in two dimensions ?
these are all fundamental questions that do not appear to have been answered by the current body of research in taxicab geometry . in this paper
we wish to provide a comprehensive , unified view of length , area , and volume in taxicab geometry .
where suitable and enlightening , we will use the value @xmath3 as the value for @xmath4 in taxicab geometry @xcite .
the first dimensional measure we will examine is the simplest of the measures : the measurement of length .
this is also , for line segments at least , the most well understood measure in taxicab geometry .
the simplest measurement of length is the length of a line segment in one dimension . on this point ,
euclidean and taxicab geometry are in complete agreement .
the length of a line segment from a point @xmath5 to another point @xmath6 is simply the number of unit lengths covered by the line segment , @xmath7 in two dimensions , however , the euclidean and taxicab metrics are not always in agreement on the length of a line segment . for line segments parallel to a coordinate axis , such as the line segment with endpoints @xmath8 and @xmath9 ,
there is agreement since both metrics reduce to one - dimensional measurement : @xmath10 . only when the line segment is not parallel to one of the coordinate axes do we finally see disagreement between the euclidean and taxicab metrics .
the taxicab length of such a line segment can be viewed as the sum of the euclidean lengths of the projections of the line segment onto the coordinate axes ( figure [ taxicab_length ] ) , @xmath11 the pythagorean theorem tells us the euclidean and taxicab lengths will generally not agree for line segments that are not parallel to one of the coordinate axes .
line segments of the same euclidean length will have various taxicab lengths as their position relative to the coordinate axes changes .
if one were to place a scale on a diagonal line , the euclidean and taxicab markings would differ with the largest discrepancy being at a @xmath12 angle to the coordinate axes ( figure [ scale_diff ] ) .
these cases and types of length measurement are well known and are well understood to those familiar with taxicab geometry .
but , even in two dimensions , there is at least one other type of length measurement in euclidean geometry : the length of a curve .
how is the length of a ( functional ) curve in two dimensions measured in taxicab geometry ? in euclidean geometry , the arc length of a curve described by a function @xmath13 with a continuous derivative over an interval @xmath14 $ ] is given by @xmath15 ^ 2 } \
, dx\ ] ] to find a corresponding formula in taxicab geometry , we may follow a traditional method for deriving arc length .
first , split the interval @xmath14 $ ] into @xmath16 subintervals with widths @xmath17 by defining points @xmath18 between @xmath19 and @xmath20 ( figure [ arc_length_2d ] ) .
this allows for definition of points @xmath21 on the curve whose @xmath22-coordinates are @xmath23 . by connecting these points
, we obtain a polygonal path approximating the curve . at this crucial point of the derivation
we apply the taxicab metric instead of the euclidean metric to obtain the taxicab length @xmath24 of the @xmath25th line segment , @xmath26 by the mean value theorem , there is a point @xmath27 between @xmath28 and @xmath29 such that @xmath30 .
thus , the taxicab length of the @xmath25th segment becomes @xmath31 and the taxicab length of the entire polygonal path is @xmath32 by increasing the number of subintervals while forcing @xmath33 , the length of the polygonal path will approach the arc length of the curve .
so , @xmath34 the right side of this equation is a definite integral , so the taxicab arc length of the curve described by the function @xmath13 over the interval @xmath14 $ ] is given by @xmath35 as an example , the northeast quadrant of a taxicab circle of radius @xmath36 centered at the origin is described by @xmath37 .
the arc length of this curve over the interval @xmath38 $ ] is given by @xmath39 which is precisely one - fourth of the circumference of a taxicab circle of radius @xmath36 .
a more interesting application involves the northeast quadrant of the euclidean circle of radius @xmath36 at the origin described by @xmath40 .
@xmath41 the distance is the same as if we had traveled along the taxicab circle between the two endpoints !
while this is a shocking result to the euclidean observer who is accustomed to distinct paths between two points generally having different lengths , the taxicab observer merely shrugs his shoulders .
a curve such as a euclidean circle can be approximated with increasingly small horizontal and vertical steps near its path ( figure [ different_paths]b ) . as any good introduction to taxicab geometry teaches us , such as krause @xcite ,
multiple straight - line paths between two points have the same length in taxicab geometry ( see figure [ different_paths]a ) .
so , each of these approximations of the euclidean circle will have the same taxicab length .
therefore , we should expect the limiting case to also have the same length . to further make the point
, we can follow the euclidean parabola @xmath42 in the first quadrant and arrive at the same result ( figure [ multiple_curves ] ) .
@xmath43 to formalize these observations , we have the following theorem .
[ thm_multiple_paths ] if a function @xmath13 is monotone increasing or decreasing and differentiable with a continuous derivative over an interval @xmath14 $ ] , then the arc length of @xmath13 over @xmath14 $ ] is @xmath44 ( i.e. the path from @xmath45 to @xmath46 is independent of the function @xmath13 under the stated conditions ) .
the taxicab arc length of @xmath13 over @xmath14 $ ] is @xmath47 if f is monotone increasing , @xmath48 . and ,
since f has a continuous derivative , we may apply the first fundamental theorem of calculus to get @xmath49 if f is monotone decreasing , we get the similar result @xmath50 in either case , the latter difference must be positive so we have @xmath44 for functions that are not monotone increasing or decreasing , the arc length can be found using this theorem by taking the sum of the arc lengths over the subintervals where the function is monotone increasing or decreasing .
the taxicab length of a line segment in three dimensions is a natural extension of the formula in two dimensions . for a line segment with endpoints @xmath51 and @xmath52 , @xmath53 a nice discussion of the three - dimensional metric and other properties
is given in @xcite .
there are no surprises here , and the taxicab length can be viewed as the sum of the euclidean lengths of the projections of the line segment onto the three coordinate axes .
the simplest extension of the arc length of a curve to three dimensions is for parametric curves . for a curve in two dimensions
, we can parameterize the function @xmath54 and modify equation ( [ eq_arc_length_2d ] ) to get @xmath55 generalizing this to three dimensions , we have @xmath56 intuitively , theorem [ thm_multiple_paths ] generalizes to three - dimensions saying that the length of a three - dimensional `` monotonic '' curve depends only on its endpoints .
in prior research regarding area in taxicab geometry @xcite , the underlying assumption has been that the area of a figure should agree in euclidean and taxicab geometry , although the formulae to compute the area could be vastly different .
additionally , it appears area has only been investigated in two dimensions . in this section
we wish to examine this assumption about area and explore surface area in three dimensions .
figure [ fig_square_area ] gets right to the question of what area means in taxicab geometry .
the figures shown are each squares in both geometries : all sides and angles have the same measure . if we cling to the notion that the area of a square is the square of the length of its sides , these figures have equal euclidean area ( 2 ) but very different taxicab areas ( 4 and 2 ) .
the position of the sides of the square relative to the coordinate axes in taxicab geometry affects their length which in turn affects the area calculation .
we have a choice before us .
we can cling to the area of a square being the square of its sides and live with the position of a figure affecting its taxicab area just as the position of a line segment affects its taxicab length . or
, we can maintain consistency with the euclidean area of the figure and , as in this example , seek a new formula for the area of a square .
several points here and later in this paper will hopefully help us to make a firm decision here .
reviewing the nature of length in taxicab geometry , we see agreement with euclidean geometry on the length of a line segment ( a one - dimensional `` figure '' ) in one dimension . only for line segments in two dimensions , a dimension higher than the figure , do we see discrepancies between the geometries .
if we apply this same logic to area , the geometries should agree on area ( a two - dimensional figure ) in two dimensions . only for the area of a surface in three dimensions
would we expect to see differences because the figure would have a different position relative to the coordinate planes .
in addition , the area of a figure has traditionally been viewed as the number of square units of the plane enclosed by the figure . since euclidean and taxicab geometry
are built on the cartesian coordinate system , it seems logical the computation of area in two dimensions should agree in these two geometries . with this reasoning
, we proceed with the traditional euclidean concept of area in two dimensions while looking for the proper extension of the concept to three dimensions .
therefore , for our example , we would need a new formula for the taxicab area of a square .
( for the interested reader , it can be shown that the area of a square in terms of the taxicab length @xmath57 of its sides is @xmath58 where @xmath59 is the taxicab angle between one of the sides and the @xmath22-axis and the trigonometric functions are taxicab @xcite . )
the standard approach to the computation of two - dimensional area in undergraduate calculus courses is integration .
estimates of area under a functional curve involve increasing numbers of rectangles that each approximate a portion of the area under the curve .
since euclidean and taxicab geometry agree on length in the horizontal and vertical directions , these area approximations in the two geometries will agree .
therefore , we should expect the calculation of area under a curve by integration to transfer seamlessly to taxicab geometry .
this is a further argument to view area in two dimensions consistently between the two geometries . as stated in a previous section
, we should expect the taxicab area of a `` flat '' surface in three dimensions to differ from the area of the `` same '' figure ( in a euclidean sense ) in one of the coordinate planes .
this follows from the pattern seen with line segments in one and two dimensions . to discover how the area will differ , consider a plane rotated upward from the @xmath60-plane along the @xmath61-axis ( figure [ fig_rotated_plane ] ) . in a manner similar to figure [ scale_diff ] , the taxicab scale in the @xmath22-direction
will be compressed in the rotated plane compared to the euclidean scale .
the angle @xmath59 formed in a plane parallel to the @xmath62-plane will determine the scaling effect seen in the one dimension of the rotated plane with the maximal difference occurring at @xmath12 .
the result is the `` same '' figure ( in a euclidean sense ) in the @xmath60-plane and in the rotated plane will have different taxicab areas .
( interestingly , the taxicab coordinate grid in the rotated plane is not even composed of euclidean squares but of euclidean rectangles since the taxicab scale in the @xmath61-direction is unaffected by the specific rotation performed . ) as a specific example , a unit square at the origin in the @xmath60-plane will have a euclidean and taxicab area of 1 .
if we rotate the square upward @xmath12 along the @xmath61-axis , the euclidean area will be unchanged but the taxicab area will now be @xmath63 since one side will now have taxicab length @xmath64 .
if the plane is instead rotated along the @xmath22-axis , the taxicab area is again found by scaling the euclidean area .
when rotation occurs in both manners , the scaling factors compound each other .
so , the taxicab area of a figure in the rotated plane can be found from its euclidean area using equation ( [ eq_taxicab_length ] ) as a guide for each direction of rotation : the scaling factors are sums of the ( euclidean ) cosine and sine of the angles @xmath65 and @xmath66 formed by lines in cross - sectional planes parallel to the @xmath62-plane and the @xmath67-plane similar to the single rotation illustrated in figure [ fig_rotated_plane ] , @xmath68 in @xcite janssen makes several attempts to revolve half a taxicab circle about an axis to obtain a taxicab sphere .
the first approach revolves the upper half of a taxicab circle in a _
euclidean _ manner about the horizontal axis resulting in a `` sphere '' with some straight edges but also with some curved surfaces ( figure [ fig_spheres]a ) .
this does not seem to quite fit the `` squarish '' nature of seemingly everything in taxicab geometry and does not appear to satisfy the usual definition of a sphere ( all points equidistant from the center ) .
the author s line of thought is quite instructive , however .
if we consider a euclidean solid of revolution about the @xmath22-axis , the cross - section of the solid in the @xmath67-plane is a euclidean circle .
therefore , for a taxicab revolution about the @xmath22-axis , the cross - section should be a _
taxicab _ circle in that plane .
a taxicab sphere obtained by revolving a taxicab circle about the horizontal axis in a taxicab manner would yield figure [ fig_spheres]b .
this completely `` squarish '' object seems to fit the geometry better and indeed does , by construction , satisfy the usual definition of a sphere .
so , revolution about the horizontal axis should follow the path of a taxicab circle in the @xmath67-plane and not the path of a euclidean circle . as an application of our approach to surface area in three dimensions
, we would like to find a formula for the surface area of a solid of revolution .
for a differentiable function @xmath13 over an interval @xmath14 $ ] divide the interval into @xmath16 subintervals . using a linear approximation of the curve over each subinterval
, we revolve each line segment about the @xmath22-axis . in the euclidean derivation
, this creates the frustrum of a cone . in taxicab geometry ,
the discussion above shows us this will create the frustrum of a taxicab cone , which is equivalent to the frustrum an euclidean right square pyramid ( partially shown in figure [ fig_revolve ] ) . to find the taxicab surface area of the frustrum
, we will compute the euclidean area and then scale the area by examining the rotations of the frustrum sides to the @xmath60-plane .
the euclidean surface area of the frustrum of a pyramid is @xmath69 where @xmath70 and @xmath71 are the perimeters of the `` top '' and `` bottom '' edges of the frustrum and @xmath57 is the slant height . for the @xmath25th subinterval ,
the euclidean perimeters are @xmath72 and @xmath73 . to compute the slant height , we project the shaded triangle in figure [ fig_revolve ] onto a plane parallel to the @xmath67-plane .
the slant height @xmath57 is the hypotenuse of the right triangle @xmath74 .
the projected triangle is an isoceles right triangle with legs of length @xmath75 ( labeled @xmath76 in the figure ) , so the altitude of this triangle is @xmath77 .
the other leg of triangle @xmath74 has length @xmath78 ( labeled @xmath79 in the figure ) .
therefore , the slant height is @xmath80 ^ 2}\ ] ]
so , we have the euclidean surface area of @xmath25th frustrum of the approximation : @xmath81 ^ 2}\\ & = & 2\sqrt{2}(f(x_{k-1})+f(x_k))\sqrt{(\triangle x_k)^2+\frac{1}{2}[f(x_{k})-f(x_{k-1})]^2}\end{aligned}\ ] ] the sides of this frustrum form euclidean angles of @xmath12 with the @xmath60-plane using a cross - sectional plane parallel to the @xmath62-plane .
this gives one scaling factor for the taxicab surface area of @xmath64 .
the other scaling factor is dependent on the sum of the cosine and sine of the angle of the linear approximation of the function and the @xmath22-axis .
this gives another scaling factor of @xmath82 ^ 2}}\ ] ] therefore , using equation ( [ eq_scaling ] ) the taxicab surface area of the @xmath25th frustrum of the approximation is @xmath83 ^ 2}}{\sqrt{(\triangle x_k)^2+[f(x_k)-f(x_{k-1})]^2 } } } \ ] ] as the derivation now continues along the same lines as the euclidean version , the intermediate value theorem and the mean value theorem give us values @xmath27 and @xmath84 such that @xmath85 and @xmath86 .
therefore , @xmath87
^ 2}}{\sqrt{(\triangle x_k)^2+[f'(x_k^{**})\triangle x_k]^2}}\\ & = & \frac{8f(x_k^*)(1+|f'(x_k^{**})|)\sqrt{1+\frac{1}{2}[f'(x_k^{**})]^2}}{\sqrt{1+[f'(x_k^{**})]^2 } } \triangle x_k\\ & = & 8f(x_k^*)(1+|f'(x_k^{**})|)\sqrt{1-\frac{[f'(x_k^{**})]^2}{2(1+[f'(x_k^{**})]^2 ) } } \triangle x_k\end{aligned}\ ] ] accumulating the frustrums and taking the limit yields the formula for the taxicab surface area of a solid of revolution : @xmath88 ^ 2}{2(1+[f'(x)]^2 ) } } \ , dx\ ] ] this is a very interesting formula .
the portion outside the radical is very reminiscent of the euclidean formula with the expected changes in the circle circumference factor ( @xmath89 ) and the arc length factor ( @xmath90 ) due to the taxicab metric .
the extra radical represents a scaling factor based on the ratio of the slant height to the linear approximation of the curve .
the greater the difference between these two quantities the more the frustrum sides are rotated with respect to the @xmath60-plane thus requiring more scaling .
if these two quantities are close , the frustrum sides are rotated very little and therefore require little scaling . to complete the analysis inspired by janssen , half of a taxicab sphere of radius @xmath36 is obtained by revolving the function @xmath91 over the interval @xmath38 $ ] .
doubling this result yields the surface area of a whole taxicab sphere : @xmath92 such a taxicab sphere is composed of two euclidean right square pyramids of height @xmath93 and base side length @xmath94 .
the euclidean surface area is therefore @xmath95 . since the sides are at @xmath12 angles to the @xmath60-plane in both directions , the taxicab area scaling factors are both @xmath63 .
this gives a taxicab surface area of @xmath96 in perfect agreement with the formula obtained from the solid of revolution .
if we follow the same pattern used for length and area measures , we expect the euclidean and taxicab volume of a solid ( a three - dimensional object ) in three dimensions to be equal . only in four dimensions
would we expect orientation of the solid to impact the taxicab volume when compared to the euclidean volume . to compute the volume of a euclidean solid of revolution ,
the traditional `` slicing '' method involves computing the volume of an infinitessimal cylinder as @xmath79 multipled by the area of the euclidean circle of revolution .
this method can be adapted directly into taxicab geometry by replacing the euclidean circle area with the taxicab circle area _ if _ the volume of a taxicab cylinder is found in the same manner .
a taxicab cylinder resembles an euclidean cylinder except that the cross section is a taxicab circle ( figure [ fig_cylinder ] ) .
for such a cylinder lying along the @xmath22-axis , the cross - sectional area will lie in a plane parallel to the @xmath67-plane and therefore the euclidean and taxicab areas will agree .
in addition , the height of the cylinder will agree in the two geometries as it lies along a coordinate axis .
based on our assumption that volume will agree in three dimensions between the geometries , the taxicab volume of a cylinder is the product of the area of the cross - sectional taxicab circle and the height . a taxicab circle of radius @xmath36 is composed of four isoceles triangles with area @xmath97 ( proposition 1 of @xcite ) .
therefore , a taxicab cylinder of radius @xmath36 and height @xmath98 will have volume @xmath99 with this information we can now compute the volume of a solid of revolution obtained by revolving a continuous function @xmath13 about the @xmath22-axis over an interval @xmath14 $ ] .
@xmath100 ^ 2 \ , dx\ ] ] continuing our primary example , the upper half of a taxicab circle of radius @xmath36 centered at the origin is described by @xmath101 using equation ( [ eq_solid_volume ] ) the volume of a taxicab sphere obtained by revolving the upper half of the circle about the @xmath22-axis is @xmath102^{2 } \
, dx\\ & = & \frac{1}{2}\pi_t \int_{-r}^{0 } \ ! ( x+r)^{2 } \ , dx + \frac{1}{2}\pi_t \int_{0}^{r } \ ! ( -x+r)^{2 } \ , dx\\ & = & \left.\frac{1}{6}\pi_t ( x+r)^{3 } \ , \right|_{-r}^{0 } - \left.\frac{1}{6}\pi_t ( -x+r)^{3 } \ , \right|_{0}^{r}\\ & = & \frac{1}{3}\pi_t r^{3}\end{aligned}\ ] ] this result agrees precisely with the volume of a taxicab sphere as a special case of a tetrahedron @xcite providing some assurance of our concept of volume and computational approach .
it should also be noted that the surface area of a sphere is not the derivative of the volume with respect to the radius .
this is a consequence of the radius of the sphere not being everywhere perpendicular to the surface .
to conclude our discussion , we can derive surface area and volume formulae for the taxicab equivalents of a few common solids . using the standard definition of a taxicab parabola described in @xcite , half of a ( horizontally ) parallel case of the parabola with focus @xmath103 and directrix @xmath104
is given by @xmath105 restricting the total `` height '' of the parabola to @xmath98 and revolving the curve about the @xmath61-axis yields an open - top taxicab paraboloid ( figure [ fig_paraboloid ] ) surface area @xmath106 as we would expect based on figure [ fig_paraboloid ] , this is the sum of the surface area of a cylinder with radius @xmath19 and height @xmath107 and half a sphere of radius @xmath19 . for the volume of a paraboloid we have @xmath108 again , as expected , this is the sum of the volume of a cylinder of radius @xmath19 and height @xmath107 and half a sphere of radius @xmath19 . in the defining paper concerning conics in taxicab geometry @xcite , nondegenerate ( or `` true '' ) two - foci
taxicab ellipses are described as taxicab circles , hexagons , and octagons .
if we revolve half of one of these figures about the @xmath22-axis we obtain a taxicab ellipsoid ( figure [ fig_ellipsoid ] ) .
if we consider a taxicab ellipse with major axis length @xmath109 , minor axis length @xmath110 , and @xmath57 the sum of the distances from a point on the ellipse to the foci , the function @xmath111 describes the upper half of an ellipse centered at the origin .
this function will generally cover the sphere ( @xmath112 and @xmath113 ) , hexagon ( @xmath112 and @xmath114 ) , and octagon ( @xmath115 ) cases for a taxicab ellipse .
the ellipsoid solid of revolution will have volume @xmath116 for the case of a spherical taxicab ellipsoid ( figure [ fig_ellipsoid]c ) , this formula reduces to @xmath117 in agreement with our previous result . for the case of a hexagon ( figure [ fig_ellipsoid]b )
, this formula reduces to @xmath118 which is the sum of the volume of a taxicab sphere of radius @xmath20 and a taxicab cylinder of radius @xmath20 and height @xmath119 .
this is to be expected since a hexagonal taxicab ellipsoid is composed of a taxicab cylinder capped by two taxicab half - spheres . in euclidean geometry ,
there is not a closed form for the surface area an ellipsoid .
since a taxicab ellipse is composed of straight lines , this problem is avoided in taxicab geometry .
the surface area of an taxicab ellipsoidal solid of revolution is @xmath120 for a hexagonal ellipsoid , the first term is the surface area of the ends which combine to equal a taxicab sphere of radius @xmath20 ; the second and last terms are zero ; and , the third term is the area of a taxicab cylinder of radius @xmath20 and length @xmath121 composing the middle of the ellipsoid . for the octogonal ellipsoid ( figure [ fig_ellipsoid]a ) , the first term is an overestimate since the top of the sphere is not present on either end .
this is corrected by the subtraction of the second term amounting to a taxicab sphere of radius @xmath122 .
( a similar correcion term is seen in the volume formula above . )
the last term accounts for the area of the ends of the ellipsoid which are taxicab circles of radius @xmath122 .
generalizing the patterns we have set forth in this paper , it appears that the taxicab measure of a @xmath16-dimensional figure in @xmath16-dimensional space will agree with the euclidean measure of the figure .
however , in @xmath123-dimensional space , the taxicab measure of the figure will in general depend on its position in the space when compared with the euclidean measure . along
the way we generalized the observation that multiple paths between two points can have the same taxicab length and described a strategy for dealing with figures living in a dimension higher than themselves .
we also developed in taxicab geometry the strategy of revolving a function about the @xmath22-axis axis by observing that the shape of the cross - section of the solid in the @xmath67-plane should be a taxicab circle .
as in euclidean geometry , this has yielded a simple method for deriving surface area and volume formulas for some standard taxicab solids created as solids of revolution . | while the concept of straight - line length is well understood in taxicab geometry , little research has been done into the length of curves or the nature of area and volume in this geometry .
this paper sets forth a comprehensive view of the basic dimensional measures in taxicab geometry . |
unless otherwise specified , all videos are filmed from above , under the following circumstances : droplet is @xmath66 @xmath14l of eutectic gallium indium ( 75.5% ga and 24.5% in by weight ) immersion solution is 1 m naoh , and the counter electrode is a copper ring concentric to the droplet and not visible in the movie . except for movie 6
, the voltage listed is measured vs. a saturated ag / agcl reference electrode , and the reported value of @xmath4 includes the @xmath10 v reference voltage .
movie 1 : : : regime a : applied @xmath25 v movie 2 : : : regime b : applied @xmath67 v movie 3 : : : regime c : applied @xmath68 v movie 4 : : : regime d : applied @xmath69 v movie 5 : : : applied @xmath70 v , 100 @xmath14l droplet wetted to a strip of copper affixed to the underlying substrate movie 6 : : : applied @xmath71 v between 100 @xmath14l droplet and ring electrode ( no reference electrode ) .
immersion solution is 1 m naf , preventing oxide dissolution movie 7 : : : applied @xmath67 v , but with the counter electrode placed in the solution directly above the droplet , instead of a ring electrode
direct measurement of the thickness of the oxide layer as a function of time and potential is hindered by a number of challenges : the dynamic nature of the droplet shape makes measurements via optical techniques ( e.g. ellipsometry ) impractical , and electrical measurements ( impedance ) are complicated by the presence of faradaic reactions . here , we instead use electrochemical measurements to estimate both the average growth and dissolution rates of the oxide layer on the surface with a simple mathematical model . to model the thickness
, we first estimate the dissolution rate of the oxide ( which we assume to be constant for all @xmath4 ) in 1 m naoh . in regime
a , where the droplets are spheroids , we employ a quasi - steady - state model , where the growth rate and dissolution rate of the oxide are assumed to be of equal magnitude .
these rates are determined by the two variables measured during the experiment : the faradaic current and the 2d projection of the droplet surface area ( @xmath72 ) .
the faradaic current allows calculation of the amount of gallium oxidized as a function of time by @xmath73 where @xmath11 represents the current as a function of time , and @xmath74 represents charge transferred .
the charge can be converted to moles of @xmath75 generated by @xmath76 where @xmath77 is the number of electrons transferred in the oxidation reaction and @xmath78 is the faraday constant ( 96485 c / mol ) .
we normalize this growth rate by the surface area of the droplet ; however , @xmath72 is not an accurate representation of the 3d surface area of the droplet @xmath79 in regime a. to estimate the actual surface area , we assume that the droplet is an oblate spheroid when the potential is applied . by measuring the radius of the droplet ( and with the volume already known ) , @xmath79 is determined by the surface area equation for an oblate spheroid @xmath80 where @xmath81 is @xmath82 from these values , the growth rate of the oxide in regime a ( and therefore , the dissolution rate across all regimes ) can be estimated @xmath83 the dissolution rate is assumed constant across regimes .
we verified the dissolution rate by measuring the concentration of gallium ions in solution using atomic absorption spectroscopy ( perkin elmer icp - optical emission spectrometer ) .
the measurements indicate that the calculated and measured value of gallium ions in solution differs by no more than a factor of 2 .
we correlate the current measured in regimes b - d to ascertain the growth rate of oxide for each regime . because the droplets are flat in these regimes
, @xmath72 is a good estimate of the true surface area of the drop .
the thickness of the oxide @xmath84 depends on the relative rates of growth and dissolution , as predicted by @xmath85 where @xmath86 and @xmath87 represent the molecular weight and density of the oxide , respectively .
we note that this model depends on several assumptions : we assume gallium oxide is the dominant species being grown on the surface of the droplet ; however , previous electrochemical studies indicate that gallium hydroxide and gallate salts form on the surface in alkaline solutions .
we also acknowledge that this approach assumes a single value for the average growth ( across the whole surface ) , while visual inspection suggests that the layer is thicker in areas closer to the counter - electrode .
further studies will be necessary to obtain a more quantitative measurement of oxide thickness gradients and their temporal dynamics .
figure [ fig : oxide ] shows the change in the growth rate as a function of time for the case of regimes b , c , and d. for each regime , the initial increase in the surface area causes a drastic decrease in the growth rate , which ultimately levels to a steady - state value .
figures [ fig : oxide]b , c ) show that the oxide thickness grows to a maximum of 1 - 4 @xmath14 m .
the maximum oxide layer thickness increases monotonically with electrical potential , and is approximately three orders of magnitude larger than the oxide thickness in air .
we perform a number of control experiments to isolate the key forces ( gravity , surface tension , oxidative stress ) and rule out other possible forces . here
, we provide a summary of these experiments , along with calculations to quantify these forces where possible .
to confirm the presence of oxidative stresses , we wet a liquid metal droplet onto a piece of copper tape , such that lateral spreading can no longer occur . applying the conditions of regime b causes visible buckling of the oxide to appear on the surface of the metal , which is indicative of compressive forces from oxidation ( see movie 5 ) . to further support the importance of these forces
, we apply up to 3 v potential difference to egain droplets placed in salt water ( 1 m naf ) with a two - electrode system . unlike naoh solutions , the naf does not dissolve the oxide , which allows the metal to form a thick oxide that prevents instabilities .
however , gradually increasing the potential causes the formation of localized protrusions arising from the oxidative stress .
furthermore , these protrusions occur at a lower potential than if the potential is set immediately ( rather than gradually ) to its highest value ( see movie 6 ) .
electrostatic forces have been shown to actuate liquid metal toward a counter electrode . despite these forces , we find that positioning the counter electrode in solution above the droplet still results in outward - spreading even though any electrical forces would point inward and upward ( movie 7 ) .
we determine the magnitude of the electrostatic force relative to gravity by tilting the apparatus .
this introduces a lateral gradient , and the angle at which the gravitational force exceeds the electrostatic force is an upper estimate of the value of the electrostatic forces . as shown in fig .
[ fig : theta ] , the average displacement of the droplet s leading edge changes sharply above @xmath91 . at this angle ,
@xmath92 which is lower than all calculated gravitational pressures and is therefore a negligible effect . as an additional test , we find that decreasing the inner radius of the counter - electrode from 7.6 cm to 3.2 cm ( increasing the electric field ) not alter the spreading behavior of the droplet .
another possible explanation of the behavior of regime b is lateral stresses caused by electrostriction . in the case of electrostriction
, the oxide would act as a dielectric in a capacitor , with charges on either side acting to compress the oxide , which could create lateral forces to cause spreading .
this force can be estimated by @xmath93 where @xmath94 is the dielectric permittivity of the oxide layer ( which we assume to be @xmath95 m@xmath96 kg@xmath97 s@xmath98 a@xmath99 ) . even assuming an oxide layer thickness of 3 nm and an applied potential of 1 v , this force remains lower than the yield stress of the oxide layer ( 5 mpa vs. 30 mpa , according to parallel plate rheological measurements that quantify the yield stress of the native oxide @xcite ) ; additionally , the oxide thickness grows more quickly than the potential , indicating that this force is smaller at higher potentials .
these calculations indicate that electrostriction is , by itself , unikely to be the driving force of regime b. l droplet .
the solid line is for a droplet brought immediately to @xmath100 v , while the dashed line is for a droplet brought to this same value at a rate of @xmath101 mv / s from the open circuit potential . ] in all regimes , the droplet starts as a sphere and spreads upon application of an electric potential @xmath4 .
however , this spreading also takes place under the influence of a small amount of inertia , which provides an effective pressure of @xmath102 pa where @xmath103 cp @xmath104 is the measured velocity at the leading edge . as shown in fig .
[ fig : inertia ] , these inertial forces are not the primary driver of spreading , but do influence the initial dynamics , including a small overshoot effect .
these overshoot effects are likely also influenced by the dynamics of oxide deposition , which depends on the thickness of the oxide and therefore the way potential is applied to the metal ( e.g. step change or sweep ) . | we identify and characterize a new class of fingering instabilities in liquid metals ; these instabilities are unexpected due to the large interfacial tension of metals .
electrochemical oxidation lowers the effective interfacial tension of a gallium - based liquid metal alloy to values approaching zero , thereby inducing drastic shape changes , including the formation of fractals . the measured fractal dimension ( @xmath0 )
places the instability in a different universality class than other fingering instabilities . by characterizing changes in morphology and dynamics as a function of droplet volume and applied electric potential , we identify the three main forces involved in this process : interfacial tension , gravity , and oxidative stress .
importantly , we find that electrochemical oxidation can generate compressive interfacial forces that oppose the tensile forces at a liquid interface .
thus , the surface oxide layer not only induces instabilities , but ultimately provides a physical barrier that halts the instabilities at larger positive potentials .
controlling the competition between surface tension and oxidative ( compressive ) stresses at the interface is important for the development of reconfigurable electronic , electromagnetic , and optical devices that take advantage of the metallic properties of liquid metals . fingering patterns
arise via a number of different spreading mechanisms : viscous fingering and diffusion limited aggregation @xcite , directional solidification @xcite , marangoni - driven spreading @xcite , zero - surface - tension granular @xcite and ordinary @xcite fluids , bacterial colony growth @xcite , and lichtenberg figures created by dielectric breakdown @xcite .
liquid metals have many potential applications @xcite due to their thermal , optical , and electrical properties but , with the largest surface tension of any known room - temperature fluid , they are both difficult to spread and an unlikely candidate to undergo fingering instabilities . here , we demonstrate a novel fingering mechanism by which liquid metals form branched , lobed structures via electrochemical surface oxidation .
this study focuses on eutectic gallium indium ( egain , from indium corp . ) , a room temperature liquid metal that offers a low - toxicity alternative to mercury .
although a native oxide forms spontaneously and rapidly on egain in the presence of air , acidic or alkaline electrolytes remove the oxide and leave the bare metal in a state of large interfacial tension ( @xmath1 mn / m ) . applying potentials
@xmath2 v to the metal in these solutions causes electrochemical deposition of a surface - active oxide that lowers @xmath3 @xcite .
the dissolution of the oxide by the electrolyte competes with the electrochemical deposition of the oxide , thereby allowing the metal to maintain fluidity at low potentials , despite the presence of the oxide .
consequently , droplets of metal assume a shape arising from the balance between interfacial tension and gravity .
increasing @xmath4 lowers the effective interfacial tension sufficiently to permit fingering instabilities .
however , at sufficiently high @xmath4 , the oxide thickens and provides both a resistive and mechanical barrier that suppresses the growth of instabilities . thus , the same oxide which destabilizes the droplet ultimately suppresses these instabilities
. figure [ fig : exp]a shows a sample fractal morphology observed at intermediate potentials .
the initially spherical drop becomes increasingly branched as it spreads outwards , with length scales spanning several orders of magnitude .
the droplet reaches a maximum surface area as the branches of the liquid metal become so thin that they pinch off from the main droplet .
( movies are available in the supp . mat . )
although this spreading process appears superficially similar to viscous fingering in a hele - shaw geometry @xcite , there are numerous distinguishing characteristics .
first , the measured fractal dimension ( @xmath5 ) is smaller than the value observed in viscous fingering ( @xmath6 ) .
second , whereas viscous fingering typically occurs when injecting fluid at constant flux or pressure , the instabilities here occur for an unconfined , constant - volume droplet .
finally , viscous fingering occurs in the less viscous of the two fluids ; here , egain is approximately twice as viscous as the surrounding fluid . in this paper , we identify the novel mechanisms behind this new instability as arising due to interfacial tension , gravity , and oxidative stress .
the last effect is the most noteworthy because interfacial forces are typically in tension , but here we show that electrochemical oxidation offers a way to create opposing forces ( compression ) that can overcome interfacial tension and help drive instabilities .
we explore the different morphologies observed , explain the phase boundaries between them , and identify a scaling collapse for droplets of different sizes that shows larger droplets require larger potentials to suppress instabilities .
[ [ experiments ] ] experiments : + + + + + + + + + + + + the apparatus consists of a shallow dish filled with 1 m sodium hydroxide ( naoh ) solution to provide active dissolution of the oxide layer and increase the ionic strength relative to de - ionized water .
the outer rim of the apparatus contains a copper ring electrode , with an inner radius of 7.6 cm and outer radius of 8.6 cm , centered around the egain droplet placed on the central working electrode ( see fig . [
fig : exp]a ) .
we vary two parameters : the volume @xmath7 of the droplet and applied potential @xmath4 .
all measurements utilize a saturated silver / silver - chloride ( ag / agcl ) reference electrode placed within the naoh solution . for this system ,
the potential at open circuit is @xmath8 @xmath9 v vs. ag / agcl .
therefore , all reported potentials are calculated relative to this open circuit ( i.e. adding @xmath10 v to the measured value ) . during each run
, we fix @xmath4 and monitor the current @xmath11 with a gamry potentiostat . leveling
the apparatus eliminates gravitational gradients , and an overhead video camera records a cross - sectional image of the droplet at 30 hz .
we measure the droplet area @xmath12 by thresholding the image at half the maximum grayscale value .
we perform a total of 365 trials spanning 4 droplet volumes ( @xmath13 @xmath14l ) and a range of electric potentials @xmath15 . to prevent undesired changes in droplet composition due to gallium dissolution into the fluid
, we typically use a new droplet for each @xmath4-series of measurements , and new solution for each @xmath7-series .
[ [ fractal - dimension ] ] fractal dimension : + + + + + + + + + + + + + + + + + + to quantify the morphology of the spreading droplets , we measure the hausdorff dimension @xmath16 using a box - counting algorithm @xcite , where @xmath17 is the box size , and @xmath18 is the number of boxes necessary to cover the area . as the droplet spreads , @xmath19 asymptotically approaches a value of @xmath20 at its maximum area ( immediately before droplet brfeakup ) , as shown by the solid line in fig .
[ fig : exp]b .
the same value is observed for droplet volumes ( see fig . [
fig : exp]c ) , where the value of @xmath4 is selected to provide the largest possible @xmath21 for that droplet volume .
this value of @xmath19 indicates that this new fingering phenomenon belongs to a different universality class than diffusion - limited aggregation @xcite , radial viscous fingering in hele - shaw cells @xcite , electrolytic metal deposition @xcite , directional solidification @xcite , and dielectric breakdown @xcite , all of which have fractal dimensions close to @xmath22 .
and ( c ) the electrical resistance measured from @xmath23 . the star indicates the time at maximum area .
all panels show data from the same 4 trials with @xmath24l and @xmath25 v ( regime a ) ; @xmath26 v ( regime b ) ; @xmath27 v ( regime c ) ; and @xmath28 v ( regime d ) .
[ fig : sampledata ] ] [ [ four - regimes ] ] four regimes : + + + + + + + + + + + + + across the range of @xmath29 values tested , we observe four distinct regimes , which we designate as a - d , in an order that corresponds with increasing @xmath4 .
this order also corresponds to increasing oxide thickness ( * ? ? ? * and supp . mat . ) .
fig .
[ fig : sampledata ] shows representative images and dynamics of each regime ; sample movies are available online ( movies 1 - 4 ) .
the photographs in fig .
[ fig : sampledata]a show the main distinguishing features of the droplet shape defining the 4 regimes : smooth ( a , d ) , branched ( b ) , or undulated ( c ) . whether spreading and/or subsequent contraction occur ( see fig .
[ fig : sampledata]b ) also distinguishes the regimes .
note that the droplet volume @xmath7 is conserved : any increases in area @xmath21 are accompanied by a decrease in mean droplet thickness ( @xmath30 ) , with actual thickness exhibiting a gradient with @xmath31 larger in the center .
regime a ( @xmath32 ) has been previously studied @xcite , and is characterized by smooth sessile droplets that remain shiny in appearance .
when mechanically perturbed , a droplet in this regime returns to its equilibrium shape .
the sessile shape represents a balance between spreading forces ( gravity ) and restoring forces ( interfacial tension ) . the gravitational force per unit area on the outer rim region is @xmath33 and the laplace pressure ( restoring force ) is @xmath34 where @xmath3 is the interfacial tension and @xmath35 is the radius of curvature at the leading edge . within regime
a , the droplet reaches an equilibrium shape determined by @xmath36 ( balancing eq . [ eq : pg ] against [ eq : pl ] ) . using values of @xmath31 estimated from @xmath21 , we calculate that typical pressures decrease from @xmath37 to @xmath38 pa as @xmath4 increases .
since the associated decrease in @xmath3 exceeds what can be achieved by electrocapillarity , we attribute it to surface oxidation @xcite .
surface oxides are known to lower the interfacial tension of molten metals , even outside of electrochemical environments @xcite . as @xmath4 increases above approximately @xmath39 v @xcite , the oxidation lowers @xmath3 ( and therefore @xmath40 ) to the point where an equilibrium shape is no longer possible , and the droplet spreads from a spherical cap into a flatter disk . as the disk spreads it becomes unstable to undulations along its perimeter ; this is regime b. these undulations simultaneously become deeper ( extending back to the center electrode ) and develop secondary , tertiary , etc .
undulations of their own .
the resulting fractal morphology characterized in fig .
[ fig : exp ] shows spreading of approximately @xmath41 the initial area of the droplet .
the droplet surface becomes less shiny during this process , particularly around the droplet perimeter , indicating the presence of a rough surface oxide . a substantial decrease in interfacial tension , as is known to occur in the presence of oxides @xcite , is sufficient to explain the observed fingering instabilities .
from the estimated values of @xmath31 and eq .
[ eq : pg ] , we find @xmath42 pa once the fractal is well - developed . balanced against eq .
[ eq : pl ] , this implies a conservative upper bound @xmath43 mn / m .
the low interfacial tension is also evidenced by the shape of the branches and undulations , which exhibit both positive and negative radii of curvature . in its final stages of spreading
, the droplet thins to the point where one or more branches pinch off into small satellite droplets .
if these satellite droplets lose their electrical contact , then their oxide layer dissolves , they become shiny in appearance , and their ( now high ) interfacial tension causes the liquid to bead up into a sphere .
if the satellite droplets happen to reattach to the electrode , the metal can begin spreading again from this new site . at larger values of @xmath4
, the droplet eventually grows a thicker oxide layer , which halts both the spreading and the development of the undulations ; this is regime c. instead of branching into a fractal , the droplet in fact _ contracts _ after about 10 seconds of initial spreading and undulating ( see fig . [
fig : sampledata]b ) . at still higher values of @xmath4 ,
the oxide growth is sufficiently rapid that the metal stops spreading before any undulations develop ; this is regime d and also exhibits the surprising contraction . the areal contraction of the metal in regimes c and d provides a key insight into the fractal fingering observed in regime b. the instabilities in regime b are only possible for @xmath44 .
yet , in regimes c and d the retraction is only possible for non - zero @xmath3 ; based on the height of the retracted droplets ( @xmath45 mm ) , we estimate a lower bound of @xmath46 mn / m .
thus , surface activity of the oxide indeed lowers the tension of the metal ( from @xmath1 mn / m in the case of bare metal to as low as @xmath46 mn / m ) , but this drop is not sufficient to enable the instabilities in regime b. after ruling out additional forces such as elasticity , electrostatics , electrostriction , and inertia based on either calculations or experimental observations ( see supp . mat . ) , we determined that oxidative forces help drive instabilities .
previous studies of anodic growth of oxides in aluminum @xcite have shown compressive oxidative stresses on the order of 0.1 gpa .
based on the size of the droplets ( @xmath47 mm ) , this would correspond to a force / length of approximately @xmath48 n / m ; this force is more than enough to exceed the force provided by the interfacial tension of the metal at this scale ( 0.1 n / m ) .
thus , the _ effective _ interfacial tension is the sum of the surface activity of the fluid ( tensile ) and the opposing ( compressive ) stresses from the oxidation process .
qualitatively , such stresses are consistent with the fan - shaped protrusions observed on the surface .
in addition , we find that the oxide buckles when the liquid metal ( under regime b conditions ) is laterally - confined , consistent with the presence of compressive forces .
second , protrusions form when the droplet is placed in a naf solution where oxide dissolution does not occur and the oxide should be mechanical barrier to spreading .
see the supp . mat . for details and movies .
subject to an applied electric potential @xmath4 .
the solid black lines correspond to the phase boundaries : a represents the area of surface tension reduction caused by the oxide , b and c represent oxidative stresses , and c and d represent the oxide as a mechanical shell .
the inset represents the electric potential @xmath49 at which the maximum in fig .
[ fig : shapevolt]a occurs , as a function of droplet volume ( on a logarithmic scale ) .
error bars represent the width @xmath50 of fig .
[ fig : shapevolt]a at half @xmath51 .
[ fig : phasediagram ] ] the electrical measurements ( see fig .
[ fig : sampledata]c ) provide additional insight into the role of the oxide . in both regime c and d
, the contraction of the droplet corresponds to a rapid increase in electrical resistance ( @xmath52 ) .
we associate this increase primarily with the thickening oxide layer , consistent with previous impedance @xcite and resistance measurements @xcite .
consequently , the contribution of the ( compressive ) oxidative stress to the effective interfacial tension diminishes when the oxide gets too thick . as a result , the droplets contract in regime c and d due to the non - zero tension of the metal within the oxide .
fig .
[ fig : sampledata]c shows that spreading only occurs at low resistance ( i.e. when ions can easily pass through the oxide ) and that it does not for higher resistance .
note that the largest values of @xmath53 occur in regime a. while this might be surprising since the oxide is thinnest for low @xmath4 , these are non - ohmic resistances .
we reason that the oxidative stress that drives regime b is possible since the potential is larger than regime a , yet the oxide is thinner than in regimes c and d ( see supp mat ) . as a point of clarification , liquid metals
have also been observed to migrate towards a counter electrode @xcite . while this might suggest the importance of electrostatic forces , our experiments suggest otherwise . positioning the counter electrode in solution directly above the droplet
still results in outward - spreading even though the field lines are largely vertical .
in addition , repeating the experiment with a counter electrode ring of half the diameter did not change the behavior of the regimes . however , replacing the circular counter electrode with a point - source placed to the side of the liquid metal caused it to migrate toward the counter electrode .
thus , we conclude that this translational motion is driven by marangoni forces arising from gradients in tension rather than electrostatic forces .
marangoni forces likely contribute to the growth of the branched structures in regime b , as evidenced by the eventual breakup of the thin filaments .
[ [ phase - diagram ] ] phase diagram : + + + + + + + + + + + + + + fig .
[ fig : phasediagram ] shows a phase diagram of all experimental runs as a function of @xmath29 , with each symbol representing the characterization of a single run . for each value of @xmath7
, there is a progression from regime a @xmath54 b @xmath54 c @xmath54 d as @xmath4 increases .
the phase boundary between regimes a and b is nearly vertical , as it arises solely from the formation of oxides for @xmath55 v , independent of @xmath7 @xcite .
however , the b / c and c / d boundaries are both tilted in such a way that droplets with a smaller surface - area to volume ratio ( larger @xmath7 ) require additional @xmath4 to achieve the same transition .
this effect is quantified in the inset to fig .
[ fig : phasediagram ] , which shows the potential @xmath49 at which a droplet reaches its maximum area .
this observation is consistent with two possible causes : that the laplace pressure due to the droplet radius @xmath56 is important for smaller droplets , or that the dynamics of oxide growth and dissolution depend on the surface area . as a function of applied electric potential @xmath4 .
( b ) the data from ( a ) with a rescaled horizontal axis : @xmath57 to show similarity of results , with @xmath58 measured from ( a ) for each droplet volume . ] it is helpful to consider the dimensionless ( scaled ) area of the metal @xmath59 , where droplets with low values of @xmath60 are more spherical , and high values are more branched . as shown in fig .
[ fig : shapevolt]a , the maximum value of @xmath60 is approximately 35 for all droplet volumes .
we find a scaling collapse for @xmath61 by re - plotting the data as a function of @xmath62 where @xmath50 is the full width of @xmath63 at its half - maximum .
this scaling collapses the data , as shown in fig .
[ fig : shapevolt]b . for all four droplet volumes , regimes a and d both have low values of @xmath60 .
regime b appears on the left side of the peak ( @xmath64 ) , and regime c on the right ( @xmath65 ) .
the peak roughly corresponds to the boundary between regimes b and c. ( note that in regime b , the metal continues to spread even after satellite droplets break off ; this process is not reflected here . )
[ [ conclusion ] ] conclusion : + + + + + + + + + + + our experiments indicate that surface oxidation lowers the effective interfacial tension and drives instabilities in a gallium - based liquid metal alloy , until the oxide grows too thick and retards further oxidation .
this behavior is interesting for a variety of reasons : it shows that an oxide layer can be tuned to both create fingering instabilities and to ultimately halt them ; it provides a method to localize and control compressive stresses at a liquid interface using modest potentials ; it demonstrates a new class of self - similar dynamics ; and it has the potential to be used in devices that necessitate shape - reconfigurable conductors . the ability to alter the shape of a metal in a simple
, low - power , and scalable way could create new types of electronics and actuators , including those that are extremely soft .
future work will focus on quantifying the role of ion - insertion in controlling compressive stresses , including the temporal dynamics and the role of the applied potential .
[ [ acknowledgments ] ] acknowledgments : + + + + + + + + + + + + + + + + ked is grateful for support under nsf ( dmr-1608097 ) and mdd is grateful for support by nsf ( cmmi-0954321 ) and air force reserach lab . ked and mdd acknowledge the nsf research triangle mrsec on programmable soft matter ( dmr-1121107 ) .
we are grateful to peter fedkiw for the loan of a potentiostat , and the ncsu environmental and agricultural testing service laboratory for performing the spectrscopy measurements . 32 natexlab#1#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 , * * , ( ) . , * * , ( ) .
, , , * * , ( ) . , , , * * , ( ) .
, , , , * * , ( ) , . , * * , ( ) . , , , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , , , , , * * , ( ) . , , , * * , ( ) .
, , , , , , * * , ( ) . , , , * * , ( ) . , , , * * , ( ) . ,
* * , ( ) . ,
* * , ( ) . , , , , , , , , * * , ( ) . , , , , , , , , , * * , ( ) . , * * , ( ) . , p. ( ) .
, , , , * * , ( ) . , , , * * , ( ) .
, , , , , * * , ( ) . , _ _ , https://www.mathworks.com/matlabcentral/fileexchange/30329-hausdorff--box-counting--fractal-dimension . , , , * * , ( ) . , , , * * , ( )
.
, , , * * , ( ) . , , , ,
* * , ( ) . , , , * * , ( ) .
, , , , * * , ( ) . , , , , , , , * * , ( ) . , , , , , , , , , , , * * , ( ) . , , , , , , pp . ( ) . * supplemental material * _ oxidation - mediated fingering instabilities in liquid metals
_ collin b. eaker , david hight , john oregan , michael d. dickey , karen e. daniels |
[ [ background ] ] background + + + + + + + + + + the power of quantum computing in various settings has been gradually clarified by many researchers : some problems can be solved on quantum computers much more efficiently than on classical computers , while others can not .
one computational model that has been extensively studied in this light is the communication complexity model . in particular , _ one - way communication _ is one of the simplest settings but it has rich connections to areas such as information theory , coding theory , on - line computing , and learning theory .
therefore , its quantum version has then been the target of intensive research @xcite .
let @xmath11 be a boolean function , where @xmath12 and @xmath13 are arbitrary sets .
the one - way communication task associated to @xmath14 is the following : alice has an input @xmath15 , bob has an input @xmath16 and the goal is for bob to output @xmath17 .
the assumption here is that only one message can be sent , from alice to bob , and the communication cost of a protocol is the number of bits of this message on the worst - case input .
we say that a protocol for @xmath14 has completeness error @xmath18 if it outputs @xmath19 with probability at least @xmath20 whenever @xmath21 , and soundness error @xmath22 if it outputs @xmath23 with probability at least @xmath24 whenever @xmath25 .
the one - way classical bounded - error communication complexity of @xmath14 , denoted by @xmath26 , is the minimum , over all protocols @xmath27 for @xmath14 with completeness and soundness error @xmath28 , of the communication cost of @xmath27 .
the one - way quantum bounded - error communication complexity of @xmath14 , denoted by @xmath29 , is defined similarly , but a quantum message can be used this time from alice to bob , and the number of qubits of the message is considered ( in this paper we suppose that there is no prior entanglement and no shared randomness between alice and bob ) .
obviously for any function @xmath14 , the relation @xmath30 holds .
one of the main open problems in quantum communication complexity is to understand how large the gap between @xmath31 and @xmath32 can be . for partial functions ( functions restricted to some domain @xmath33 or , equivalently , functions with a promise on their inputs ) ,
an exponential separation between these two quantities has been shown recently in @xcite .
however the situation for total functions is far less clear : the largest gap known is an asymptotic factor of @xmath34 @xcite . in the exact setting , i.e. , the setting where no error and no giving up are allowed , the quantum and classical one - way communication complexities are known to be the same for any total function @xcite . in the unbounded - error setting , i.e. , the setting where any error probability less than 1/2 is allowed , it is known that the gap is exactly a factor @xmath34 for both partial and total functions @xcite .
although bounded - error is a notion between the exact and unbounded - error , we stress that the bounded - error setting usually behaves quite differently from the other two in the case of total functions , e.g. , for two - way communication there is a quadratic gap in the bounded - error setting @xcite whereas in the exact setting no gap is known and , in the unbounded - error setting , the gap is again exactly a factor 2 @xcite .
note also that for total functions in the bounded - error setting , quadratic gaps are known in the two - way model @xcite and exponential gaps are known in the simultaneous message - passing model @xcite ; and these models are respectively stronger and weaker than the one - way model . thus , whether a superlinear gap between @xmath31 and @xmath32 can be achieved for some total function is an intriguing question . [ [ the - subgroup - membership - function ] ] the subgroup membership function + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + many of the problems for which quantum computation is more powerful than classical computation have group - theoretic structure .
in particular , watrous @xcite has used the subgroup membership problem ( as a computational problem ) to separate the complexity classes @xmath35 and @xmath36 relative to an oracle .
inspired by watrous s work @xcite , we propose the subgroup membership function as a candidate to show a superlinear gap between @xmath31 and @xmath32 .
let @xmath1 be any finite group , and let @xmath37 denote the set of subgroups of @xmath1 .
then the subgroup membership function for @xmath1 , denoted by @xmath38 , is the function with domain @xmath39 such that @xmath40 for any group @xmath1 , the upper bound @xmath41 follows easily from the fact that any subgroup of @xmath1 is generated by at most @xmath42 elements .. ] furthermore , there exist families of groups @xmath1 such that @xmath43 : for example , the abelian groups @xmath44 with @xmath45 . thus there
exist groups @xmath1 for which the `` trivial protocol , '' wherein alice simply sends bob the name of her subgroup , requires @xmath46 communication .
the one - way classical communication complexity of the function @xmath38 was previously considered by miltersen et al .
@xcite , who showed that for the family of groups @xmath44 , any one - way protocol with perfect soundness and completeness error @xmath47 requires @xmath48-bit communication . for certain groups
@xmath1 , we conjecture that @xmath48-bit communication is needed even if the completeness and soundness errors are both @xmath28 . on the other hand
, there is a simple _ quantum _ one - way protocol , using @xmath49-bit communication , by which bob can compute @xmath38 with perfect completeness and constant soundness for any group @xmath1 . in this protocol
inspired by @xcite
alice sends the quantum state @xmath50 .
bob then creates the state @xmath51 where @xmath52 , applies a hadamard gate on the last register , and measures it in the basis @xmath53 to decide which of @xmath54 and @xmath55 holds .
thus , proving that there exists a family of groups @xmath1 such that @xmath56 would lead to a quadratic separation between @xmath31 and @xmath32 for a total function .
in other words , a major objective has been to prove a 2-sided - error version of the lower bound by miltersen et al .
@xcite mentioned above .
apart from the goal of proving a separation between @xmath31 and @xmath32 , we believe that the communication complexity of deciding subgroup membership is interesting in itself , since the latter is a key task in most group - theoretic computational problems . [
[ overview - of - our - results ] ] overview of our results + + + + + + + + + + + + + + + + + + + + + + + in this paper we present three upper bounds on the one - way classical communication complexity of the subgroup membership function : * we give a classical protocol using @xmath57-bit communication , with perfect completeness and constant soundness , for the subgroup membership function in the case where alice s subgroup @xmath0 is normal .
this suggests that in order to obtain a separation between @xmath31 and @xmath32 using the subgroup membership problem , one must consider groups with many nonnormal subgroups .
we also present a lower bound which is tight for some families of groups .
notice that this situation appears to be similar to the status of the hidden subgroup problem : there exists an efficient quantum algorithm solving the problem in the case where the hidden subgroup is normal @xcite ; without the normality condition , however , very little is known .
our results rely on the theory of characters of finite groups and especially on the connection between kernels of irreducible characters and normal subgroups , as did the algorithms of @xcite .
* let @xmath7 be a prime not dividing @xmath58 .
then we show that @xmath59 , where @xmath60 is the maximum dimension of an irreducible @xmath10-representation of @xmath1 .
this result uses a beautiful characterization of the subspaces of the group algebra @xmath61 $ ] stabilized by @xmath0 .
we remark that for any group @xmath1 of exponent @xmath62 ( which is to say that @xmath63 for all @xmath64 ) , we have @xmath65 , where @xmath66 is the maximum dimension of a complex irreducible representation of @xmath1 and @xmath67 is the order of @xmath7 in @xmath68 , the multiplicative group of the integers relatively prime to @xmath62 . in particular , as there is always a prime @xmath7 of size @xmath4 relatively prime to @xmath8 , this protocol has complexity no more than @xmath69 . * finally , we show that @xmath70 , where @xmath71 is the maximum dimension of an irreducible complex representation of @xmath1 .
this upper bound is obtained by a protocol that mirrors the technique utilized in the modular case by suitably discretizing the vector space @xmath72 and controlling `` geometric expansion '' around invariant spaces .
one corollary is that any family of groups with an abelian subgroup of constant index has a protocol with complexity @xmath4 .
in particular , for groups such as @xmath73 , the action of @xmath74 on @xmath75 being to reverse the order of the coordinates , we have @xmath76 .
these results suggest a nontrivial connection between the representation theory of the group @xmath1 and the subgroup membership problem , and provide natural candidates for which a superlinear separation between @xmath77 and @xmath78 may be obtained : groups with large irreducible representations and many nonnormal subgroups , e.g. , the symmetric group .
we assume the reader is familiar with basic concepts of group theory . here
we introduce some notions from representation theory that we will need . in this paper
, @xmath1 always denotes a finite group and @xmath19 denotes its identity element . [
[ group - representations ] ] group representations + + + + + + + + + + + + + + + + + + + + + let @xmath79 be a field whose characteristic does not divide the order of @xmath1 ( so the characteristic of @xmath79 could be zero ) . an @xmath79-_representation _ @xmath80 of @xmath1 is a homomorphism from @xmath1 to @xmath81 , the group of invertible linear transformations over a vector space @xmath82 ( over the field @xmath79 ) .
the _ dimension _ of @xmath80 is the dimension of @xmath82 .
we say that a representation @xmath83 is _ irreducible _ if the only subspaces of @xmath84 simultaneously fixed by the entire family of linear operators @xmath85 are the trivial ones : @xmath86 and @xmath84 .
the _ group algebra _
$ ] is the @xmath79-algebra of formal sums @xmath88 with coordinatewise addition and multiplication defined by linearly extending the rule @xmath89 .
note that @xmath87 $ ] has dimension @xmath8 as a vector space over @xmath79 .
the natural action of @xmath1 on the group algebra defines the _ regular representation _ : the action of @xmath90 on a vector @xmath91 in @xmath87 $ ] is denoted by @xmath92 and defined as @xmath93 now , if @xmath0 is a subgroup of @xmath1 , let @xmath94 \mid \text{$h{\mathbf{v } } = { \mathbf{v}}$ for all $ h \in h$ } \}\ ] ] be the subspace of @xmath0-invariant vectors of @xmath87 $ ] .
it is easy to check that a vector @xmath95 lies in @xmath96 if and only if @xmath95 is constant on each left coset of @xmath0 in @xmath1 .
let @xmath97 be the set of right cosets of @xmath0 in @xmath1 .
the vectors @xmath98 for @xmath99 form a basis of @xmath96 and thus @xmath100\,,\ ] ] where @xmath101={\vert \mathcal s_{g : h } \vert}={\vert g \vert}/{\vert h \vert}$ ] denotes the index of @xmath0 in @xmath1 .
a theorem of maschke s ( see , e.g. , @xcite ) asserts that @xmath87 $ ] is semi - simple , i.e. , @xmath87 $ ] can be written as the direct sum of a family of irreducible representations . in this case , a theorem of wedderburn s
@xcite asserts that each irreducible representation appears with multiplicity equal to its dimension : @xmath102=\bigoplus_{\rho\in { \textrm{irr}(g,{\mathbb{f}})}}v_\rho^{\oplus d_\rho}\,,\ ] ] where @xmath103 denotes the set of ( representatives of ) all the irreducible @xmath79-representations @xmath104 and @xmath105 denotes the dimension of @xmath80 .
if @xmath106 is the subspace of @xmath84 pointwise fixed by @xmath0 , we see that @xmath107^{\oplus d_\rho}\,,\ ] ] and conclude that @xmath108\,.\ ] ] [ [ complex - characters ] ] complex characters + + + + + + + + + + + + + + + + + + let @xmath79 be the complex field @xmath109 . for any @xmath109-representation @xmath80 of @xmath1 , the character of @xmath80 is the function @xmath110 such that @xmath111 for any @xmath112 , where @xmath113 denotes the trace .
characters are conjugacy class functions : the relation @xmath114 holds for any two elements @xmath115 of @xmath1 .
moreover , the value @xmath116 is the dimension of the representation @xmath80 .
the kernel of @xmath117 , denoted by @xmath118 , is defined as @xmath119 .
it is easy to see that @xmath118 is a subgroup of @xmath1 .
a character is said to be irreducible if it is the character of an irreducible representation .
denote by @xmath120 the set of irreducible ( complex ) characters of @xmath1 . the relation @xmath121 ^
2={\vert g \vert}$ ] is well - known and implies the inequality @xmath122 .
let @xmath0 be a normal subgroup of @xmath1 .
denote @xmath123 then the relation @xmath124 ^ 2=[g : h]\ ] ] holds ( see , e.g. , @xcite ) .
in this section we give an efficient classical protocol computing the subgroup membership function in the special case where alice s subgroup @xmath0 is normal .
our protocol is actually more general : we show that one can decide efficiently membership in the normal closure of @xmath0 , denoted by @xmath125 ( the smallest normal subgroup of @xmath1 containing @xmath0 ) .
the protocol testing normal closure membership , denoted by @xmath126 , is as follows .
a subgroup @xmath127 . an element @xmath128 .
alice chooses a random element @xmath130 of @xmath131 with probability @xmath132 ^ 2{\vert { \overline{h } } \vert}/{\vert g \vert}$ ] ; alice sends the name of @xmath130 to bob ; bob outputs @xmath19 if @xmath133 and outputs @xmath23 otherwise .
observe that by equation , the weights of step 1 do indeed determine a probability distribution .
notice that @xmath134 since @xmath135 and @xmath122 .
thus protocol @xmath126 can be implemented using @xmath57 bits of communication .
we now show the correctness of this protocol . on any input @xmath136 , protocol @xmath126 outputs @xmath19 with probability @xmath19 if @xmath137 , and outputs @xmath23 with probability at least @xmath47 if @xmath138 .
if @xmath139 , then for any element @xmath130 in @xmath131 the equality @xmath133 holds . then bob always outputs 1 .
protocol @xmath126 has thus perfect completeness .
now suppose that @xmath140 .
denote @xmath141 .
the error probability of the protocol is @xmath142 ^ 2.\ ] ] to conclude our proof , we now prove that @xmath143 ^ 2\le \frac{{\vert g \vert}}{2{\vert { \overline{h } } \vert}}.\ ] ] let @xmath144 denote the normal closure of the set @xmath145 in @xmath1 .
remember that the normal closure of a set @xmath146 is the smallest normal subgroup of @xmath1 including @xmath147 , and can be defined explicitly as the subgroup of @xmath1 generated by all the elements @xmath148 for @xmath112 and @xmath149 . since @xmath140 the subgroup @xmath125 is a proper subgroup of @xmath144 . in particular @xmath150 .
we now claim that @xmath151
. then equation ( [ equation : char ] ) implies that @xmath143 ^ 2= \sum_{\chi\in \lambda_{k}}[\chi(1)]^2=[g : k]\le \frac{{\vert g \vert}}{2{\vert { \overline{h } } \vert}}.\ ] ] the proof of the claim follows .
first suppose that @xmath117 is an element of @xmath152 .
then @xmath153 and thus @xmath154 .
now suppose that @xmath117 is an element of @xmath155
. then @xmath156 . from the basic properties of characters ,
we conclude that @xmath157 and thus @xmath158 . given a finite group @xmath1 , let @xmath159 be the set of normal subgroups of @xmath1 .
since for a normal subgroup @xmath0 of @xmath1 we have @xmath160 , we conclude that protocol @xmath126 solves the restriction of @xmath38 to the domain @xmath161 ( notice that this is still a total function ) .
[ theorem_normal ] for any finite group @xmath1 , the restriction of @xmath38 to the domain @xmath161 can be computed with perfect completeness and soundness error @xmath47 by communicating at most @xmath57 bits .
we now show a simple lower bound on the communication complexity of @xmath38 .
we first recall the definition of the vc - dimension of a set of functions @xcite .
let @xmath162 be a set of boolean functions over a finite domain @xmath13 .
we say that a set @xmath163 is shattered by @xmath162 if for every subset @xmath164 there exists a function @xmath165 such that @xmath166 if and only if @xmath167 .
the largest size of set @xmath147 over all @xmath147 shattered by @xmath162 is the vc - dimension of @xmath162 , denoted by @xmath168 .
we say that a subset @xmath147 of a finite group @xmath1 is an independent subset of @xmath1 if , for each @xmath169 , element @xmath170 can not be written as any product of elements of @xmath171 .
we denote by @xmath172 the maximal size of an independent subset of @xmath1 . notice that , for any finite group @xmath1 , the inequality @xmath173 holds .
we now state our lower bound .
[ proposition : lower ] @xmath174 .
in particular , the family of groups @xmath44 for @xmath175 satisfies @xmath176 . for each subgroup @xmath177 , define the function @xmath178 as @xmath179 for every @xmath128 . denote @xmath180 .
a result by klauck @xcite shows that @xmath181 , where @xmath182 is the binary entropy function .
let @xmath183 be distinct elements of @xmath1 such that @xmath184 is a subset of independent elements of @xmath1 .
the subset @xmath146 is shattered by @xmath162 since it is easy to show that , for any subset @xmath164 , the function @xmath185 is such that @xmath186 if and only if @xmath187 ( here @xmath188 denotes the subgroup generated by the elements in r )
. then @xmath189 and @xmath190 .
the second part of the proposition follows from the observation that each group @xmath191 is also a vector space of dimension @xmath192 over the finite field @xmath193 and , thus , @xmath194 . proposition [ proposition : lower ] shows that , for the family of groups @xmath44 , protocol @xmath126 is optimal up to a constant factor .
in this section we present a protocol computing the group membership function for groups with small modular representations .
let @xmath195 be a finite field with characteristic @xmath7 not dividing @xmath8 .
our protocol , denoted by @xmath196 , is the following .
a subgroup @xmath127 . an element @xmath128 .
@xmath197 alice chooses a representation @xmath198 in @xmath199 with probability @xmath200 alice chooses a random vector @xmath201 ; alice sends the name of @xmath80 and the vector @xmath95 to bob ; bob outputs 1 if @xmath202 and outputs 0 otherwise .
observe that by equation , the weights of step 1 do indeed determine a probability distribution .
we now show the correctness of this protocol .
let @xmath1 be a finite group and @xmath195 be a finite field of characteristic @xmath7 not dividing @xmath8 .
protocol @xmath196 computes @xmath38 with perfect completeness and constant soundness error .
its communication complexity is at most @xmath203 bits , where @xmath204 is the maximum dimension of an irreducible @xmath195-representation of @xmath1 .
note that the protocol is clearly complete : if @xmath205 , then bob always accepts . to establish soundness ,
let @xmath206 and define @xmath207 , the smallest subgroup containing both @xmath0 and @xmath208 .
remember that @xmath209 denotes the subspace of @xmath84 pointwise fixed by @xmath144 .
we see that @xmath210 = \sum_\rho \frac{|h| d_\rho \dim i_k(\rho)}{|g| } = \frac{|h|}{|k| } = \frac{1}{[k : h ] } \leq \frac{1}{2}\ , , \ ] ] again by equation .
observe , then , that @xmath211 and so @xmath212\ge \pr[i_k(\rho ) = i_h(\rho)].\ ] ] then @xmath213 = 1-\pr[i_k(\rho ) = i_h(\rho)]\ge 1/2\,.\ ] ] when @xmath214 , the vector @xmath95 chosen by alice has probability no more than @xmath215 to be in @xmath209
. then @xmath216 with constant probability in her choices of @xmath80 and @xmath95 .
since @xmath217 , the communication complexity of the protocol is at most @xmath218 . in light of the complexity guarantee of the protocol above , it is natural to ask how the dimensions of the irreducible representations of a finite group @xmath1 compare over various fields and , especially , how the modular case compares to the complex case .
when the group algebras involved are semi - simple ( as they are in this paper due to our insistence that @xmath219 ) , there is a tight connection expressed in the following proposition .
let @xmath1 be a finite group of exponent @xmath62 and @xmath7 be any prime not dividing @xmath8 .
then the relation @xmath65 holds , where @xmath66 is the maximum dimension of a complex irreducible representation of @xmath1 , @xmath220 is the maximum dimension of an irreducible @xmath10-representation of @xmath1 , and @xmath67 is the order of @xmath7 in @xmath68 , the multiplicative group of the integers relatively prime to @xmath62 .
this is a consequence of the `` @xmath221-@xmath222-@xmath223 triangle '' ( see @xcite ) .
see appendix [ sec : cde ] for a brief discussion . as there always exists a prime @xmath7 of size @xmath224 that does not divide @xmath8 , we obtain the following corollary .
@xmath225 , where @xmath62 denotes the exponent of @xmath1 and @xmath66 is the maximum dimension of a complex irreducible representation of @xmath1 .
we now focus on the case where the dimensions of the irreducible @xmath109-representations of @xmath1 is under control .
the key idea is to discretize the protocol given in the previous section . to achieve this goal we use the concept of an @xmath227-net of a sphere .
( as our nets will lie in the vector spaces acted upon by the irreps of @xmath1 , we define them as subsets of complex hilbert spaces . )
let @xmath82 be a finite - dimensional complex hilbert space .
an _ @xmath227-net _ of @xmath82 is a finite family of unit - vectors @xmath228 so that for every unit - length vector @xmath229 , there is a vector @xmath230 so that @xmath231 . [ prop : net ] for any @xmath232 and for any complex hilbert space @xmath82 of dimension @xmath222 , there exists an @xmath227-net of size at most @xmath233 . for any dimension @xmath222 and distance @xmath234 ,
there is a set of points @xmath235 of cardinality no more than @xmath236 with the property that every point of @xmath237 has distance no more than @xmath227 from some point of @xmath238 ( see , e.g. , @xcite ) .
this yields a set with analogous properties of size no more than @xmath239 for the complex @xmath222-sphere , which has the same metric as the real @xmath240 sphere .
note that if @xmath95 and @xmath241 are two unit vectors of @xmath82 , we may write @xmath242 with @xmath243 in which case , @xmath244 . the statement of the proposition follows .
our protocol requires the choice of a sufficiently dense @xmath227-net for each irreducible representation in @xmath245 .
this choice is independent of the inputs of the protocol and so can be done by alice and bob without communication .
the protocol is as follows .
a subgroup @xmath127 an element @xmath128 @xmath197 alice and bob agree on an @xmath227-net @xmath246 of @xmath84 for each @xmath198 in @xmath245 ; alice chooses a representation @xmath247 in @xmath245 with probability @xmath200 alice chooses a random ( according to haar measure ) unit length vector @xmath201 ; alice sends bob the name of @xmath80 and the closest vector @xmath248 in @xmath246 to the vector @xmath95 ; if @xmath249 , then bob outputs 1 ; otherwise @xmath250 , and bob outputs 0 . observe that by equation , the weights at step 2 do indeed determine a probability distribution on @xmath245 .
ideally , at step 3 , alice would communicate @xmath95 to bob : bob could then check if @xmath251 and , if so , would figure that @xmath205 . if @xmath252 , bob would be sure that @xmath253 , since @xmath106 is precisely the fixed space of @xmath0 .
the proof below shows that by sending a sufficiently close approximation to @xmath95 , bob can still answer confidently .
the following theorem states the correctness and the communication complexity of this protocol .
[ th : complex ] there exists a choice of @xmath254 such that protocol @xmath255 computes @xmath38 with perfect completeness and constant soundness error by communicating @xmath256 bits , where @xmath66 denotes the maximum dimension of an irreducible @xmath109-representation of @xmath1 . as the name of the representation @xmath80 can be encoded using @xmath257 bits , the communication complexity of the protocol will be dominated by the number of bits necessary to encode the vector @xmath248 .
we will show that a choice @xmath258 suffices to achieve perfect completeness and constant soundness . according to proposition [ prop :
net ] , such an @xmath227-net can be indexed with @xmath259 bits .
this gives our upper bound .
we proceed with the analysis of the completeness and soundness of the protocol . *
completeness * observe that if @xmath205 , then the vector @xmath95 chosen by alice in the protocol is fixed by @xmath260 . recall that alice sends bob a vector @xmath248 for which @xmath261 ; writing @xmath262 ( where @xmath263 ) we have @xmath264 and @xmath265 .
considering that @xmath266 we conclude that @xmath267 recall that @xmath260 is unitary , so that @xmath268
. then , by the cauchy - schwarz inequality , @xmath269 as @xmath270 , we have @xmath271 and it follows that the protocol has perfect completeness . *
soundness * we wish to show that for sufficiently small @xmath227 @xmath272 , the protocol has constant soundness .
assume that @xmath253 and let @xmath207 , the smallest subgroup containing @xmath0 and @xmath208 .
our goal will be to show that with constant probability @xmath273 is far from @xmath19 , in which case the same can be said of @xmath248 so long as @xmath227 is sufficiently small . from equation ,
@xmath210 = \sum_\rho \frac{|h| d_\rho \dim i_k(\rho)}{|g| } = \frac{|h|}{|k| } = \frac{1}{[k : h ] } \leq \frac{1}{2}\ , .\ ] ] then , with constant probability , the subspace of @xmath106 fixed by @xmath208 has dimension no more than @xmath274 .
we may write the vector @xmath275 as @xmath276 , where @xmath277 and @xmath278^\perp$ ] , the space perpendicular to @xmath209 .
we then have @xmath279 and @xmath280 .
now , as @xmath95 is chosen uniformly on the unit sphere in @xmath84 , we have @xmath281 = \dim i_k(\rho ) / \dim i_h(\rho)$ ] and the probability @xmath282 $ ] is lower bounded by a constant .
, the random variable @xmath283 possesses much stronger concentration around the expected value than this .
] we wish to conclude that , conditioned on the event @xmath284 , the value @xmath285 can not be too close to 1 .
we will show , in fact , that the real part is appropriately bounded below 1 .
consider the restriction of the representation @xmath286 chosen by alice to the subgroup @xmath144 : specifically , we may decompose @xmath84 as an orthogonal direct sum of @xmath144-invariant subspaces : @xmath287 where each @xmath288 is in @xmath289 ( but copies of the same irrep may appear several times in the direct sum ) . in this decomposition , @xmath290 is precisely the projection of @xmath95 into the subspace @xmath291 corresponding to the copies of the trivial representation ; @xmath292 , on the other hand , lies solely in @xmath293 .
as both @xmath95 and @xmath290 lie in @xmath106 , the difference @xmath292 does as well and the projection of @xmath292 into each @xmath294 is @xmath0-invariant ( that is , lies in @xmath295 ) . with this in mind
, we shall upper bound @xmath296 by controlling @xmath297 taken over all nontrivial irreps @xmath298 of @xmath144 and all @xmath0-invariant vectors @xmath241 in @xmath299 . in particular , writing @xmath300 ( with each @xmath301 lying in @xmath294 ) , we have @xmath302 and @xmath303 observe that if @xmath238 is a set of generators for @xmath0 and @xmath241 is an @xmath0-invariant vector of @xmath299 , @xmath304 where @xmath305 . then @xmath306 ( note that the vector @xmath241 is not constrained to be @xmath0-invariant in this expression . ) if we choose @xmath238 to be a symmetric generating set ( so that @xmath307 ) then @xmath308 is self - adjoint and @xmath309 is unitary so that @xmath310.\ ] ] as the operator @xmath311 is hermitian , we have @xmath312 where @xmath313 denotes the operator norm . in order to control this operator norm , observe that the linear operator @xmath314 $ ] is precisely given by the left action of the group algebra element @xmath315 } \triangleq \frac{1}{2 |a| } \left[\sum_{a \in a } e_{ya}+ \sum_{a \in a } e_{ay^{-1 } } \right ] \in { \mathbb{c}}[k]\ ] ] on the invariant subspace @xmath299 of @xmath316 $ ] corresponding to the representation @xmath298 .
alternatively , we may consider the cayley graph on the group @xmath144 given by the symmetric generating ( multi-)set @xmath317 .
the ( normalized ) adjacency matrix of this cayley graph is identical to the regular representation evaluated at the group algebra element above . as @xmath317 is a ( symmetric ) generating set for @xmath144 , the operator norm of @xmath318})$ ]
is bounded below 1 for each nontrivial @xmath298 ( see , e.g. , @xcite ) . in order to conclude the proof
, we require explicit bounds on this spectral gap . a result of erds and renyi @xcite asserts that we may select a set of generators @xmath238 for @xmath0 of size @xmath319 so that the diameter of the resulting cayley graph ( generated by @xmath238 over @xmath0 ) is @xmath319 .
considering that the diameter of @xmath238 ( as generators for @xmath0 ) is @xmath319 , it is easy to see that the set @xmath317 induces a cayley graph on @xmath144 of diameter no more than @xmath320 \log |h|)$ ]
. now we may invoke a theorem of babai @xcite asserting that the second eigenvalue of any ( undirected ) cayley graph with degree @xmath222 and diameter @xmath321 is no more than @xmath322 .
( if we normalize the adjacency matrix by degree , the second eigenvalue is no more than @xmath323 . )
we conclude that @xmath324 ^ 2 \log^3 |h|}\right)\ ] ] and , considering that @xmath284 , that @xmath325 ^ 2 \log^3 |h|}\right)\,.\ ] ] finally , alice s @xmath248 can be written @xmath326 with @xmath327 , in which case @xmath328 ^ 2 \log^3 |h|}\right ) + 3\epsilon \leq 1 - 2\epsilon\,,\ ] ] for @xmath329 ^ 2 \log^3 |h|)$ ] ; thus the protocol is sound .
in particular , theorem [ th : complex ] shows that , over groups for which @xmath66 is constant , the subgroup membership problem can be solved using @xmath49-bit communication .
there is a very beautiful characterization of such groups : a family of groups has representations of bounded degree if and only each group of the family has an abelian subgroups of constant index @xcite .
we thus obtain the following corollary .
let @xmath1 be a family of groups each possessing an abelian subgroup of constant index
. then @xmath330
the authors are grateful to keith conrad , iordanis kerenidis , and troy lee for helpful discussions on this subject .
bcwdw01 scott aaronson and andris ambainis . quantum search of spatial regions .
, 1:4779 , 2005 .
. limitations of quantum advice and one - way communication . , 1:128 , 2005 .
. local expansion of vertex - transitive graphs and random generation in finite groups . in _ proceedings of the 23rd annual acm symposium on symposium of theory of computing _ , pages 164174 , 1991 .
harry buhrman , richard cleve , john watrous , and ronald de wolf .
quantum fingerprinting .
, 87(16):167902 , 2001 .
alexander borovik , laszlo pyber , and aner shalev .
maximal subgroups in finite and profinite groups . , 348(9):37343761 , 1996 .
charles w. curtis and irving reiner . .
american mathematical society , 2006 .
p. erds and a. rnyi .
probabilistic methods in group theory . , 14:127138 , 1965 .
dmitry gavinsky , julia kempe , iordanis kerenidis , ran raz , and ronald de wolf .
exponential separations for one - way quantum communication complexity , with applications to cryptography . in _ proceedings of the 39th annual acm symposium on theory of computing _ , pages 516525 , 2007 .
david gluck .
the largest irreducible degree of a finite group .
, 37(4):442451 , 1985 .
s. hoory , n. linial , and a. wigderson .
expander graphs and their applications . , 43(4):439561 , 2006 .
sean hallgren , alexander russell , and amnon ta - shma .
the hidden subgroup problem and quantum computation using group representations .
, 32(4):916934 , 2003 .
kazuo iwama , harumichi nishimura , rudy raymond , and shigeru yamashita .
unbounded - error one - way classical and quantum communication complexity . in _ proceedings of the 34th international colloqium on automata , languages and programming _ , pages 110121 , 2007
. i. martin isaacs . .
dover , 1976 .
hartmut klauck .
one - way communication complexity and the neiporuk lower bound on formula size .
, 37(2):552583 , 2007 .
bala kalyanasundaram and georg schnitger .
the probabilistic communication complexity of set intersection .
, 5(4):545557 , 1992 .
jiri matousek . .
springer , 2002 .
peter bro miltersen , noam nisan , shmuel safra , and avi wigderson . on data structures and asymmetric communication complexity .
, 57(1):3749 , 1998 . ilan newman and mario szegedy .
public vs. private coin flips in one round communication games ( extended abstract ) . in _ proceedings of the 28th annual acm symposium on theory of computing _ ,
pages 561570 , 1996 .
jean - pierre serre . .
springer , 1977 .
vladimir vapnik and alexey chervonenkis . on the uniform convergence of relative frequencies of events to their probabilities . , 16:264280 , 1971 .
john watrous .
succinct quantum proofs for properties of finite groups . in _ proceedings of the 41st annual symposium on foundations of computer science _ , pages 537546 , 2000 .
andreas winter .
quantum and classical message protect identification via quantum channels .
, 4(6&7):564578 , 2004 .
let @xmath1 be a finite group of exponent @xmath62 ( so @xmath62 is the smallest integer for which @xmath63 for all @xmath64 ) .
we outline a technique for reducing @xmath226-representations of @xmath1 to @xmath10-representations in a manner that preserves irreducibility . for a complete account , see @xcite . by a difficult theorem of brauer ( see , e.g. , @xcite ) , one may always realize a @xmath226-irrep over the field @xmath331 $ ] where @xmath332 is a principal @xmath62th root of unity .
( it is natural to guess that this might be true , as all eigenvalues of a representation of @xmath1 are @xmath62th roots of unity . )
let @xmath333 $ ] be the ring of algebraic integers in @xmath331 $ ] ( it so happens that in this cyclotomic case @xmath333 $ ] is indeed the ring of algebraic integers ) .
let @xmath334 be a prime , and let @xmath335(p))$ ] ; this is a prime ideal of @xmath333 $ ] lying over @xmath7 in the sense that @xmath336 . now ,
if only the representation could be realized over @xmath333 $ ] , we could simply reduce mod @xmath337 and obtain a representation over an extension of @xmath10 . however , this is either not always true or just not known to be true by the authors . to fix the problem , one first localizes at @xmath337 ; that is
, we consider the ring @xmath333_\mathfrak{p}$ ] of all fractions with the property that the denominator lies outside @xmath337 ; this is a principal ideal domain with a single prime ( and maximal ) ideal @xmath337 . in this case , the representation can be realized over @xmath333_\mathfrak{p}$ ] , as this pid generates the whole field as its field of fractions ( see @xcite ) .
now we can reduce mod @xmath337 ; the result is a matrix realization over the field @xmath333_\mathfrak{p}/\mathfrak{p}$ ] ; it is easy to check that this is an extension of the field @xmath338 .
furthermore , the dimension of this extension field is the multiplicative order of @xmath7 modulo @xmath62 ( the same as the extension of the splitting field of the polynomial @xmath339 over @xmath10 ) .
this immediately gives rise to a representation over the field @xmath195 with @xmath340 .
we remark that this process preserves irreducibility , and induces a complete decomposition of @xmath61 $ ] into irreducible representations . | this paper studies the one - way communication complexity of the _ subgroup membership problem _ , a classical problem closely related to basic questions in quantum computing . here
alice receives , as input , a subgroup @xmath0 of a finite group @xmath1 ; bob receives an element @xmath2 .
alice is permitted to send a single message to bob , after which he must decide if his input @xmath3 is an element of @xmath0 .
we prove the following upper bounds on the classical communication complexity of this problem in the bounded - error setting : 1 .
the problem can be solved with @xmath4 communication , provided the subgroup @xmath0 is normal .
2 .
the problem can be solved with @xmath5 communication , where @xmath6 is the maximum of the dimensions of the irreducible complex representations of @xmath1 .
3 . for any prime @xmath7
not dividing @xmath8 , the problem can be solved with @xmath9 communication , where @xmath6 is the maximum of the dimensions of the irreducible @xmath10-representations of @xmath1 . |
we want to study the following _ assembly maps _ : & ^g_n & & h_n^g(e_(g);k_r ) & & & & h_n^g(\{};k_r ) = k_n(rg ) ; [ ass_k_n ] + & ^g_n & & h_n^g(e_(g);kh_r ) & & & & h_n^g(\{};kh_r ) = kh_n(rg ) ; [ ass_kh_n ] + & ^g_n & & h_n^g(e_(g);l_r^- ) & & & & h_n^g(\{};l_r^- ) = l_n^- ( r g ) ; [ ass_l_n ] + & ^g_n & & h_n^g(e_(g);k^_a , l^1 ) & & & & h_n^g(\{};k^_a , l^1 ) = k_n(a _
l^1 g ) ; [ ass_k_ntopo_l1 ] + & ^g_n & & h_n^g(e_(g);k^_a , r ) & & & & h_n^g(\{};k^_a , r ) = k_n(a _ r g ) ; [ ass_k_ntopo_r ] + & ^g_n & & h_n^g(e_(g);k^_a , m ) & & & & h_n^g(\{};k^_a , m ) = k_n(a _ m g ) .
[ ass_k_ntopo_m ] some explanations are in order .
family of subgroups of @xmath1 _ is a collection of subgroups of @xmath1 which is closed under conjugation and taking subgroups .
examples are the family @xmath2 of finite subgroups and the family @xmath3 of virtually cyclic subgroups .
let @xmath4 be the _ classifying space associated to @xmath5_. it is uniquely characterized up to @xmath1-homotopy by the properties that it is a @xmath1-@xmath6-complex and that @xmath7 is contractible if @xmath8 and is empty if @xmath9 . for more information about these spaces
@xmath4 we refer for instance to the survey article @xcite .
given a group @xmath1 acting on a ring ( with involution ) by structure preserving maps , let @xmath10 be the twisted group ring ( with involution ) and denote by @xmath11 , @xmath12 and @xmath13 its _ algebraic @xmath0-theory _ in the non - connective sense ( see gersten @xcite or pedersen - weibel @xcite ) , _ its homotopy @xmath0-theory _ in the sense of weibel @xcite , and its _
@xmath14-theory with decoration @xmath15 _ in the sense of ranicki ( * ? ? ?
* chapter 17 ) . given a group @xmath1 acting on a @xmath16-algebra @xmath17 by automorphisms of @xmath16-algebras ,
let @xmath18 be the banach algebra obtained from @xmath19 by completion with respect to the @xmath20-norm , let @xmath21 be the reduced crossed product @xmath16-algebra , and let @xmath22 be the maximal crossed product @xmath16-algebra and denote by @xmath23 , @xmath24 and @xmath25 their _ topological
@xmath0-theory_. the source and target of the assembly maps are given by @xmath1-homology theories ( see definition [ def : g - homology - theory ] and theorem [ the : construction_of_equivariant_homology_theories ] ) with the property that for every subgroup @xmath26 and @xmath27 @xmath28 all the assembly maps are induced by the projection from @xmath29 or @xmath30 respectively to the one - point - space @xmath31 .
it might be surprising to the reader that we restrict to c*-algebra coefficients @xmath17 in the assembly map .
indeed , our main results rely heavily on the validity of the conjecture for hyperbolic groups , which , so far , is only known for c*-algebra coefficients .
moreover we also want to study the passage from the @xmath20-setting to the @xmath16-setting .
hence we decided to restrict ourselves to the case of @xmath16-coefficients throughout .
we mention that on the other hand the assembly map can also be defined for banach algebra coefficients @xcite . before we go on ,
let us fix some conventions .
a group @xmath1 is always discrete .
hyperbolic group is to be understood in the sense of gromov ( see for instance @xcite , @xcite , @xcite , @xcite ) .
ring means associative ring with unit and ring homomorphisms preserve units .
homomorphisms of banach algebras are assumed to be norm decreasing .
the _ farrell - jones conjecture _ for algebraic @xmath0-theory for a group @xmath1 and a ring @xmath32 with @xmath1-action by ring automorphisms says that the assembly map is bijective for all @xmath33 .
its version for homotopy @xmath0-theory says that the assembly map is bijective for all @xmath27 .
if @xmath32 is a ring with involution and @xmath1 acts on @xmath32 by automorphism of rings with involutions , the @xmath14-theoretic version of the farrell - jones conjecture predicts that the assembly map is bijective for all @xmath34 .
the farrell - jones conjecture for algebraic @xmath0- and @xmath14-theory was originally formulated in the paper by farrell - jones ( * ? ? ? * on page 257 ) for the trivial @xmath1-action on @xmath32 .
its homotopy @xmath0-theoretic version can be found in ( * ? ? ? * conjecture 7.3 ) , again for trivial @xmath1-action on @xmath32 .
let @xmath1 be a group acting on the @xmath16-algebra @xmath17 by automorphisms of @xmath16-algebras .
the _ bost conjecture with coefficients _ and the _ baum - connes conjecture with coefficients _ respectively predict that the assembly map and respectively are bijective for all @xmath27 .
the original statement of the baum - connes conjecture with trivial coefficients can be found in ( * ? ? ?
* conjecture 3.15 on page 254 ) .
our formulation of these conjectures follows the homotopy theoretic approach in @xcite .
the original assembly maps are defined differently .
we do not give the proof that our maps agree with the original ones but at least refer to @xcite , where the farrell - jones conjecture is treated and to hambleton - pedersen @xcite , where such identification is given for the baum - connes conjecture with trivial coefficients .
the main purpose of this paper is to prove that these conjectures are inherited under colimits over directed systems of groups ( with not necessarily injective structure maps ) .
we want to show : [ the : inheritance_under_colimits ] let @xmath35 be a directed system of groups with ( not necessarily injective ) structure maps @xmath36 .
let @xmath37 be its colimit with structure maps @xmath38 .
let @xmath32 be a ring ( with involution ) and let @xmath17 be a @xmath16-algebra with structure preserving @xmath1-action .
given @xmath39 and a subgroup @xmath40 , we let @xmath41 act on @xmath32 and @xmath17 by restriction with the group homomorphism @xmath42 .
fix @xmath27 . then
[ the : inheritance_under_colimits : arbitrary ] if the assembly map @xmath43 of is bijective for all @xmath27 , all @xmath39 and all subgroups @xmath40 , then for every subgroup @xmath44 of @xmath1 the assembly map @xmath45 of is bijective for all @xmath27 .
+ the corresponding version is true for the assembly maps given in , , , and ; 2 .
[ the : inheritance_under_colimits : injective ] suppose that all structure maps @xmath46 are injective and that the assembly map @xmath47 of is bijective for all @xmath27 and @xmath39 .
then the assembly map @xmath48 of is bijective for all @xmath27 ; + the corresponding statement is true for the assembly maps given in , , , , and .
theorem [ the : inheritance_under_colimits ] will follow from theorem [ the : isomorphism_conjecture_is_stable_under_colim ] and lemma [ lem : homology_theories_are_(strongly)_continuous ] as soon as we have proved theorem [ the : construction_of_equivariant_homology_theories ] .
notice that the version does not appear in assertion [ the : inheritance_under_colimits : arbitrary ] .
a counterexample will be discussed below . the ( fibered ) version of theorem [ the : inheritance_under_colimits ] [ the : inheritance_under_colimits : arbitrary ] in the case of algebraic @xmath0-theory and @xmath14-theory with coefficients in @xmath49 with trivial @xmath1-action has been proved by farrell - linnell ( * ? ? ?
* theorem 7.1 ) . in ( * ? ? ?
* section 7 ) higson , lafforgue and skandalis construct counterexamples to the _ baum - connes - conjecture with coefficients _ , actually with a commutative @xmath16-algebra as coefficients .
they formulate precise properties for a group @xmath1 which imply that it does _ not _ satisfy the baum - connes conjecture with coefficients .
gromov @xcite describes the construction of such a group @xmath1 as a colimit over a directed system of groups @xmath35 , where each @xmath50 is hyperbolic .
this construction did raise the hope that these groups @xmath1 may also be counterexamples to the baum - connes conjecture with trivial coefficients . but to the authors knowledge this has not been proved and no counterexample to the baum - connes conjecture with trivial coefficients is known . of course one
may wonder whether such counterexamples to the baum - connes conjecture with coefficients or with trivial coefficients respectively may also be counterexamples to the farrell - jones conjecture or the bost conjecture with coefficients or with trivial coefficients respectively .
however , this can be excluded by the following result .
[ the : conjectures_for_colimits_of_hyperbolic_groups ] let @xmath1 be the colimit of the directed system @xmath35 of groups ( with not necessarily injective structure maps ) .
suppose that each @xmath50 is hyperbolic .
let @xmath44 be a subgroup .
then : 1 . [
the : conjectures_for_colimits_of_hyperbolic_groups : k ] the group @xmath0 satisfies for every ring @xmath32 on which @xmath0 acts by ring automorphisms the farrell - jones conjecture for algebraic @xmath0-theory with coefficients in @xmath32 , i.e. , the assembly map is bijective for all @xmath27 ; 2 .
[ the : conjectures_for_colimits_of_hyperbolic_groups : kh ] the group @xmath0 satisfies for every ring @xmath32 on which @xmath0 acts by ring automorphisms the farrell - jones conjecture for homotopy @xmath0-theory with coefficients in @xmath32 , i.e. , the assembly map is bijective for all @xmath27 ; 3 .
[ the : conjectures_for_colimits_of_hyperbolic_groups : l1 ] the group @xmath0 satisfies for every @xmath16-algebra @xmath17 on which @xmath0 acts by @xmath16-automorphisms the bost conjecture with coefficients in @xmath17 , i.e. , the assembly map is bijective for all @xmath27 . if @xmath1 is the colimit of the directed system @xmath51 , then the subgroup @xmath44 is the colimit of the directed system @xmath52 , where @xmath53 is the structure map .
hence it suffices to prove theorem [ the : conjectures_for_colimits_of_hyperbolic_groups ] in the case @xmath54 .
this case follows from theorem [ the : inheritance_under_colimits ] [ the : inheritance_under_colimits : arbitrary ] as soon as one can show that the farrell - jones conjecture for algebraic @xmath0-theory , the farrell - jones conjecture for homotopy @xmath0-theory , or the bost conjecture respectively holds for every subgroup @xmath41 of a hyperbolic group @xmath1 with arbitrary coefficients @xmath32 and @xmath17 respectively .
firstly we prove this for the bost conjecture .
mineyev and yu ( * ? ? ?
* theorem 17 ) show that every hyperbolic group g admits a g - invariant metric @xmath55 which is weakly geodesic and strongly bolic .
since every subgroup @xmath41 of @xmath1 clearly acts properly on g with respect to any discrete metric , it follows that @xmath41 belongs to the class @xmath56 as described by lafforgue in @xcite ( see also the remarks at the top of page 6 in @xcite ) .
now the claim is a direct consequence of ( * ? ? ?
* theorem 0.0.2 ) .
the claim for the farrell - jones conjecture is proved for algebraic @xmath0-theory and homotopy @xmath0-theory in bartels - lck - reich @xcite which is based on the results of @xcite .
there are further groups with unusual properties that can be obtained as colimits of hyperbolic groups .
this class contains for instance a torsion - free non - cyclic group all whose proper subgroups are cyclic constructed by olshanskii @xcite .
further examples are mentioned in @xcite and ( * ? ? ?
* section 4 ) .
we mention that if one can prove the @xmath14-theoretic version of the farrell - jones conjecture for subgroups of hyperbolic groups with arbitrary coefficients , then it is also true for subgroups of colimits of hyperbolic groups by the argument above .
if @xmath1 is an infinite group which satisfies kazhdan s property ( t ) , then the assembly map for the maximal group @xmath16-algebra fails to be an isomorphism if the assembly map for the reduced group @xmath16-algebra is injective ( which is true for a very large class of groups and in particular for all hyperbolic groups by @xcite ) .
the reason is that a group has property ( t ) if and only if the trivial representation @xmath57 is isolated in the dual @xmath58 of @xmath1 .
this implies that @xmath59 has a splitting @xmath60 , where we regard @xmath57 as a representation of @xmath59 .
if @xmath1 is infinite , then the first summand is in the kernel of the regular representation @xmath61 ( see for instance @xcite ) , hence the @xmath0-theory map @xmath62 is not injective . for a finite group @xmath41 we have @xmath63 and
hence we can apply ( * ? ? ?
* lemma 4.6 ) to identify the domains of and . under this identification
the composition of the assembly map with @xmath64 is the assembly map and the claim follows . hence the baum - connes conjecture for the maximal group @xmath16-algebras is not true in general since the baum - connes conjecture for the reduced group @xmath16-algebras has been proved for some groups with property ( t ) by lafforgue @xcite ( see also @xcite ) .
so in the sequel our discussion refers always to the baum - connes conjecture for the reduced group @xmath16-algebra .
one may speculate that the baum - connes conjecture with trivial coefficients is less likely to be true for a given group @xmath1 than the farrell - jones conjecture or the bost conjecture .
some evidence for this speculation comes from lack of functoriality of the reduced group @xmath16-algebra .
a group homomorphism @xmath65 induces in general not a @xmath16-homomorphism @xmath66 , one has to require that its kernel is amenable .
here is a counterexample , namely , if @xmath67 is a non - abelian free group , then @xmath68 is a simple algebra @xcite and hence there is no unital algebra homomorphism @xmath69 . on the other hand ,
any group homomorphism @xmath65 induces a homomorphism @xmath70 where @xmath1 acts trivially on @xmath71 and @xmath72 is the up to @xmath1-homotopy unique @xmath1-map .
notice that the induction map @xmath73 exists since the isotropy groups of @xmath74 are finite .
moreover , this map is compatible under the assembly maps for @xmath41 and @xmath1 with the map @xmath75 provided that @xmath76 has amenable kernel and hence @xmath77 is defined .
so the baum - connes conjecture implies that every group homomorphism @xmath65 induces a group homomorphism @xmath78 , although there may be no @xmath16-homomorphism @xmath66 induced by @xmath76 .
no such direct construction of @xmath79 is known in general .
here is another failure of the reduced group @xmath16-algebra .
let @xmath1 be the colimit of the directed system @xmath35 of groups ( with not necessarily injective structure maps ) .
suppose that for every @xmath39 and preimage @xmath41 of a finite group under the canonical map @xmath38 the baum - connes conjecture for the maximal group @xmath16-algebra holds ( this is for instance true by @xcite if @xmath80 has the haagerup property ) . then @xmath81 is a composition of two isomorphisms .
the first map is bijective by the transitivity principle [ the : transitivity ] , the second by lemma [ lem : directed_limits_and_calh?_ast ] and lemma [ lem : homology_theories_are_(strongly)_continuous ] .
this implies that the following composition is an isomorphism @xmath82 namely , these two compositions are compatible with the passage from the maximal to the reduced setting .
this passage induces on the source and on the target isomorphisms since @xmath83 and @xmath29 have finite isotropy groups , for a finite group @xmath41 we have @xmath84 and hence we can apply ( * ? ? ?
* lemma 4.6 ) . now assume furthermore that the baum - connes conjecture for the reduced group @xmath16-algebra holds for @xmath50 for each @xmath39 and for @xmath1 .
then we obtain an isomorphism @xmath85 again it is in general not at all clear whether there exists such a map in the case , where the structure maps @xmath86 do not have amenable kernels and hence do not induce maps @xmath87 .
these arguments do not apply to the farrell - jones conjecture or the bost conjecture .
namely any group homomorphism @xmath65 induces maps @xmath88 , @xmath89 , and @xmath90 for a ring @xmath32 or a @xmath16-algebra @xmath17 with structure preserving @xmath1-action , where we equip @xmath32 and @xmath17 with the @xmath41-action coming from @xmath76 .
moreover we will show for a directed system @xmath35 of groups ( with not necessarily injective structure maps ) and @xmath37 that there are canonical isomorphisms ( see lemma [ lem : homology_theories_are_(strongly)_continuous ] ) @xmath91 let @xmath17 be a @xmath16-algebra with @xmath1-action by @xmath16-automorphisms .
we can consider @xmath17 as a ring only .
notice that we get a commutative diagram @xmath92 \ar[r ] & { { k\!h}}_n(a\rtimes g ) \ar[d ] \\
k}{\mathbf h}}_a ) \ar[r ] & { { k\!h}}_n(a\rtimes g ) \\ h_n^g({e_{{{\mathcal{f}\text{in}}}}(g)};{{\mathbf k}{\mathbf h}}_a ) \ar[d ] \ar[u]_{\cong } \ar[r ] & { { k\!h}}_n(a\rtimes g ) \ar[d ] \ar[u]_{\operatorname{id } } \\ h_n^g({e_{{{\mathcal{f}\text{in}}}}(g)};{{\mathbf k}}^{\operatorname{top}}_{a , l^1 } ) \ar[d]_{\cong } \ar[r ] & k_n(a \rtimes_{l^1 } g ) \ar[d ] \\ h_n^g({e_{{{\mathcal{f}\text{in}}}}(g)};{{\mathbf k}}^{\operatorname{top}}_{a , m } ) \ar[d]_{\cong } \ar[r ] & k_n(a \rtimes_m g)\ar[d ] \\
k}}^{\operatorname{top}}_{a , r } ) \ar[r ] & k_n(a \rtimes_r g ) } \ ] ] where the horizontal maps are assembly maps and the vertical maps are change of theory and rings maps or induced by the up to @xmath1-homotopy unique @xmath1-map @xmath93 .
the second left vertical map , which is marked with @xmath94 , is bijective .
this is shown in ( * ? ? ?
* remark 7.4 ) in the case , where @xmath1 acts trivially on @xmath32 , the proof carries directly over to the general case .
the fourth and fifth vertical left arrow , which are marked with @xmath94 , are bijective , since for a finite group @xmath41 we have @xmath95 and hence we can apply ( * ? ? ?
* lemma 4.6 ) .
in particular the bost conjecture and the baum - connes conjecture together imply that the map @xmath96 is bijective , the map @xmath97 is split injective and the map @xmath98 is split surjective .
the upshot of this discussions is : * the counterexamples of higson , lafforgue and skandalis ( * ? ? ?
* section 7 ) to the baum - connes conjecture with coefficients are not counterexamples to the farrell - jones conjecture or the bost conjecture ; * the counterexamples of higson , lafforgue and skandalis ( * ? ? ?
* section 7 ) show that the map @xmath99 is in general not bijective ; * the passage from the topological @xmath0-theory of the banach algebra @xmath100 to the reduced group @xmath16-algebra is problematic and may cause failures of the baum - connes conjecture ; * the bost conjecture and the farrell - jones conjecture are more likely to be true than the baum - connes conjecture ; * there is to the authors knowledge no promising candidate of a group @xmath1 for which a strategy is in sight to show that the farrell - jones conjecture or the bost conjecture are false .
( whether it is reasonable to believe that these conjectures are true for all groups is a different question . )
the general strategy of this paper is to present most of the arguments in terms of equivariant homology theories .
many of the arguments for the farrell - jones conjecture , the bost conjecture or the baum - connes conjecture become the same , the only difference lies in the homology theory we apply them to .
this is convenient for a reader who is not so familiar with spectra and prefers to think of @xmath0-groups and not of @xmath0-spectra .
the construction of these equivariant homology theories is a second step and done in terms of spectra .
spectra can not be avoided in algebraic @xmath0-theory by definition and since we want to compare also algebraic and topological @xmath0-theory , we need spectra descriptions here as well
. another nice feature of the approach to equivariant topological @xmath0-theory via spectra is that it yields a theory which can be applied to all @xmath1-@xmath6-complexes .
this will allow us to consider in the case @xmath37 the equivariant @xmath0-homology of the @xmath50-@xmath6-complex @xmath101 although @xmath102 has infinite isotropy groups if the structure map @xmath38 has infinite kernel .
details of the constructions of the relevant spectra , namely , the proof of theorem [ the : necessary_functors_from_groupoids_to_spectra ] , will be deferred to @xcite .
we will use the existence of these spectra as a black box .
these constructions require some work and technical skills , but their details are not at all relevant for the results and ideas of this paper and their existence is not at all surprising .
in the @xmath14-theory case one encounters also non - orientable manifolds . in this case twisting with the first stiefel - whitney class is required . in a more general setup one
is given a group @xmath1 , a ring @xmath32 with involution and a group homomorphism @xmath103 to the center of the multiplicative group of units in @xmath32 .
so far we have used the standard involution on the group ring @xmath104 , which is given by @xmath105 .
one may also consider the @xmath106-twisted involution given by @xmath107 .
all the results in this paper generalize directly to this case since one can construct a modified @xmath14-theory spectrum functor ( over @xmath1 ) using the @xmath106-twisted involution and then the homology arguments are just applied to the equivariant homology theory associated to this @xmath106-twisted @xmath14-theory spectrum . the work was financially supported by the sonderforschungsbereich 478 geometrische strukturen in der mathematik and the max - planck - forschungspreis of the third author .
in this section we briefly explain basic axioms , notions and facts about equivariant homology theories as needed for the purposes of this article . the main examples which will play a role in connection with the bost , the baum - connes and the farrell - jones conjecture will be presented later in theorem [ the : construction_of_equivariant_homology_theories ] .
fix a group @xmath1 and a ring @xmath108 . in most cases
@xmath108 will be @xmath49 .
the following definition is taken from ( * ? ? ?
* section 1 ) .
[ def : g - homology - theory ] a _ @xmath1-homology theory @xmath109 with values in @xmath108-modules _ is a collection of covariant functors @xmath110 from the category of @xmath1-@xmath6-pairs to the category of @xmath108-modules indexed by @xmath27 together with natural transformations @xmath111 for @xmath27 such that the following axioms are satisfied : * @xmath1-homotopy invariance + if @xmath112 and @xmath113 are @xmath1-homotopic maps @xmath114 of @xmath1-@xmath6-pairs , then @xmath115 for @xmath27 ; * long exact sequence of a pair + given a pair @xmath116 of @xmath1-@xmath6-complexes , there is a long exact sequence @xmath117 where @xmath118 and @xmath119 are the inclusions ;
* excision + let @xmath116 be a @xmath1-@xmath6-pair and let @xmath120 be a cellular @xmath1-map of @xmath1-@xmath6-complexes .
equip @xmath121 with the induced structure of a @xmath1-@xmath6-pair .
then the canonical map @xmath122 induces an isomorphism @xmath123 * disjoint union axiom + let @xmath124 be a family of @xmath1-@xmath6-complexes .
denote by @xmath125 the canonical inclusion .
then the map @xmath126 is bijective .
let @xmath127 and @xmath128 be @xmath1-homology theories .
a _ natural transformation @xmath129 of @xmath1-homology theories _ is a sequence of natural transformations @xmath130 of functors from the category of @xmath1-@xmath6-pairs to the category of @xmath108-modules which are compatible with the boundary homomorphisms .
[ lem : transf_between_g - homologies ] let @xmath129 be a natural transformation of @xmath1-homology theories .
suppose that @xmath131 is bijective for every homogeneous space @xmath132 and @xmath27 .
then @xmath133 is bijective for every @xmath1-@xmath6-pair @xmath116 and @xmath27 .
the disjoint union axiom implies that both @xmath1-homology theories are compatible with colimits over directed systems indexed by the natural numbers ( such as the system given by the skeletal filtration @xmath134 ) .
the argument for this claim is analogous to the one in @xcite .
hence it suffices to prove the bijectivity for finite - dimensional pairs .
using the axioms of a @xmath1-homology theory , the five lemma and induction over the dimension one reduces the proof to the special case @xmath135 .
next we present a slight variation of the notion of an equivariant homology theory introduced in ( * ? ? ?
* section 1 ) .
we have to treat this variation since we later want to study coefficients over a fixed group @xmath136 which we will then pullback via group homomorphisms with @xmath136 as target .
namely , fix a group @xmath136 .
a group @xmath137 over @xmath136 is a group @xmath1 together with a group homomorphism @xmath138 .
a map @xmath139 of groups over @xmath136 is a group homomorphisms @xmath140 satisfying @xmath141 .
let @xmath142 be a group homomorphism .
given an @xmath41-space @xmath143 , define the _ induction of @xmath143 with @xmath76 _ to be the @xmath1-space denoted by @xmath144 which is the quotient of @xmath145 by the right @xmath41-action @xmath146 for @xmath147 and @xmath148 .
if @xmath142 is an inclusion , we also write @xmath149 instead of @xmath79 .
if @xmath116 is an @xmath41-@xmath6-pair , then @xmath150 is a @xmath1-@xmath6-pair .
[ def : equivariant_homology_theory_over_a_group_gamma ] an _ equivariant homology theory @xmath151 with values in @xmath108-modules over a group @xmath136 _ assigns to every group @xmath137 over @xmath136 a @xmath1-homology theory @xmath127 with values in @xmath108-modules and comes with the following so called _ induction structure _ : given a homomorphism @xmath152 of groups over @xmath136 and an @xmath41-@xmath6-pair @xmath116 , there are for each @xmath27 natural homomorphisms @xmath153 satisfying * compatibility with the boundary homomorphisms + @xmath154 ; * functoriality + let @xmath155 be another morphism of groups over @xmath136 .
then we have for @xmath27 @xmath156 where @xmath157 is the natural @xmath0-homeomorphism ; * compatibility with conjugation + let @xmath137 be a group over @xmath136 and let @xmath158 be an element with @xmath159 .
then the conjugation homomorphisms @xmath160 defines a morphism @xmath161 of groups over @xmath136 .
let @xmath162 be the @xmath1-homeomorphism which sends @xmath163 to @xmath164 in @xmath165 . +
then for every @xmath27 and every @xmath1-@xmath6-pair @xmath116 the homomorphism @xmath166 agrees with @xmath167 .
* bijectivity + if @xmath152 is a morphism of groups over @xmath136 such that the underlying group homomorphism @xmath65 is an inclusion of groups , then @xmath168 is bijective for all @xmath27 .
definition [ def : equivariant_homology_theory_over_a_group_gamma ] reduces to the one of an equivariant homology in ( * ? ? ?
* section 1 ) if one puts @xmath169 .
[ lem : induction_bijective ] let @xmath152 be a morphism of groups over @xmath136 .
let @xmath116 be an @xmath41-@xmath6-pair such that @xmath170 acts freely on @xmath171 .
then @xmath172 is bijective for all @xmath27 .
let @xmath5 be the set of all subgroups of @xmath41 whose intersection with @xmath170 is trivial .
obviously , this is a family , i.e. , closed under conjugation and taking subgroups .
a @xmath41-@xmath6-pair @xmath116 is called a @xmath5-@xmath41-@xmath6-pair if the isotropy group of any point in @xmath171 belongs to @xmath5 .
a @xmath41-@xmath6-pair @xmath116 is a @xmath5-@xmath41-@xmath6-pair if and only if @xmath170 acts freely on @xmath171 .
the @xmath173-skeleton of @xmath150 is @xmath79 applied to the @xmath173-skeleton of @xmath116 .
let @xmath116 be an @xmath41-@xmath6-pair and let @xmath120 be a cellular @xmath41-map of @xmath41-@xmath6-complexes .
equip @xmath121 with the induced structure of a @xmath41-@xmath6-pair .
then there is an obvious natural isomorphism of @xmath1-@xmath6-pairs @xmath174 now we proceed as in the proof of lemma [ lem : transf_between_g - homologies ] but now considering the transformations @xmath175 only for @xmath5-@xmath41-@xmath6-pairs @xmath116 .
thus we can reduce the claim to the special case @xmath176 for some subgroup @xmath177 with @xmath178 .
this special case follows from the following commutative diagram whose vertical arrows are bijective by the axioms and whose upper horizontal arrow is bijective since @xmath76 induces an isomorphism @xmath179 . @xmath180 ^ -{\operatorname{ind}_{\alpha|_l \colon l \to \alpha(l ) } } \ar[d]^{\operatorname{ind}_l^h } & { \mathcal{h}}_n^{\alpha(l)}({\{\bullet\ } } ) \ar[d]^{\operatorname{ind}_{\alpha(l)}^g } \\ { \mathcal{h}}^h_n(h / l)\ar[r]^-{\operatorname{ind}_{\alpha } } & { \mathcal{h}}_n^g(\alpha_*h / l ) = { \mathcal{h}}^g_n(g/\alpha(l ) ) } } \ ] ]
fix a group @xmath136 and an equivariant homology theory @xmath151 with values in @xmath108-modules over @xmath136 .
let @xmath143 be a @xmath1-@xmath6-complex .
let @xmath65 be a group homomorphism .
denote by @xmath181 the @xmath41-@xmath6-complex obtained from @xmath143 by _ restriction with @xmath76_. we have already introduced the induction @xmath182 of an @xmath41-@xmath6-complex @xmath183 .
the functors @xmath184 and @xmath185 are adjoint to one another . in particular the adjoint of the identity on @xmath186
is a natural @xmath1-map @xmath187 it sends an element in @xmath188 given by @xmath189 to @xmath190 . consider a map @xmath152 of groups over @xmath136 .
define the @xmath108-map @xmath191 if @xmath155 is another morphism of groups over @xmath136 , then by the axioms of an induction structure the composite @xmath192 agrees with @xmath193 for a @xmath0-@xmath6-complex @xmath143 .
consider a directed system of groups @xmath35 with @xmath194 and structure maps @xmath38 for @xmath39 and @xmath195 for @xmath196 .
we obtain for every @xmath1-@xmath6-complex @xmath143 a system of @xmath108-modules @xmath197 with structure maps @xmath198 .
we get a map of @xmath108-modules @xmath199 the next definition is an extension of ( * ? ? ?
* definition 3.1 ) .
[ def : strongly_continuous_equivariant_homology_theory ] an equivariant homology theory @xmath151 over @xmath136 is called _ continuous _ if for every group @xmath137 over @xmath136 and every directed system of subgroups @xmath200 of @xmath1 with @xmath201 the @xmath108-map ( see ) @xmath202 is an isomorphism for every @xmath27 .
an equivariant homology theory @xmath151 over @xmath136 is called _ strongly continuous _ if for every group @xmath137 over @xmath136 and every directed system of groups @xmath35 with @xmath37 and structure maps @xmath53 for @xmath39 the @xmath108-map @xmath203 is an isomorphism for every @xmath27
. here and in the sequel we view @xmath50 as a group over @xmath136 by @xmath204 and @xmath38 as a morphism of groups over @xmath136 .
[ lem : directed_limits_and_calh?_ast ] let @xmath137 be a group over @xmath136 . consider a directed system of groups @xmath35 with @xmath37 and structure maps @xmath38 for @xmath39 .
let @xmath116 be a @xmath1-@xmath6-pair .
suppose that @xmath151 is strongly continuous .
then the @xmath108-homomorphism ( see ) @xmath205 is bijective for every @xmath27 .
the functor sending a directed systems of @xmath108-modules to its colimit is an exact functor and compatible with direct sums over arbitrary index maps .
if @xmath116 is a pair of @xmath1-@xmath6-complexes , then @xmath206 is a pair of @xmath50-@xmath6-complexes .
hence the collection of maps @xmath207 is a natural transformation of @xmath1-homology theories of pairs of @xmath1-@xmath6-complexes which satisfy the disjoint union axiom .
hence in order to show that @xmath208 is bijective for all @xmath27 and all pairs of @xmath1-@xmath6-complexes @xmath116 , it suffices by lemma [ lem : transf_between_g - homologies ] to prove this in the special case @xmath209 . for @xmath39 let @xmath210 be the @xmath50-map sending @xmath211 to @xmath212 .
consider a directed system of @xmath108-modules @xmath213 whose structure maps for @xmath214 are given by the composite @xmath215 for the @xmath216-map @xmath217 sending @xmath218 to @xmath219
. then the following diagram commutes @xmath220 ^ -{\operatorname{colim}_{i \in i } \operatorname{ind}_{\psi_i^{-1}(h)}^{g_i}}_{\cong } \ar[dd]_{t^h_n({\{\bullet\}})}^{\cong } & \operatorname{colim}_{i \in i } { \mathcal{h}}_n^{g_i}(g_i/\psi_i^{-1}(h ) ) \ar[d]^{\operatorname{colim}_{i \in i } { \mathcal{h}}^{g_i}_n(k_i ) } \\ & \operatorname{colim}_{i \in i } { \mathcal{h}}_n^{g_i}(\psi_i^*(g / h ) ) \ar[d]^{t^g_n(g / h ) } \\ { \mathcal{h}}_n^h({\{\bullet\ } } ) \ar[r]^-{\operatorname{ind}_h^g}_{\cong } & { \mathcal{h}}_n^g(g / h ) } \ ] ] where the horizontal maps are the isomorphism given by induction . for the directed system @xmath221 with structure maps @xmath222 , the group homomorphism @xmath223 is an isomorphism .
this follows by inspecting the standard model for the colimit over a directed system of groups .
hence the left vertical arrow is bijective since @xmath151 is strongly continuous by assumption .
therefore it remains to show that the map @xmath224 is surjective .
notice that the map given by the direct sum of the structure maps @xmath225 is surjective . hence it remains to show for a fixed @xmath39 that the image of the structure map @xmath226 is contained in the image of the map .
we have the decomposition of the @xmath50-set @xmath227 into its @xmath50-orbits @xmath228 it induces an identification of @xmath108-modules @xmath229 hence it remains to show for fixed elements @xmath39 and @xmath230 that the obvious composition @xmath231 is contained in the image of the map .
choose an index @xmath232 with @xmath233 and @xmath234 .
then the structure map for @xmath235 is a map @xmath236 which sends the summand corresponding to @xmath237 to the summand corresponding to @xmath238 which is by definition the image of @xmath239 obviously the image of composite of the last map with the structure map @xmath240 is contained in the image of the map .
hence the map is surjective .
this finishes the proof of lemma [ lem : directed_limits_and_calh?_ast ] .
a _ family @xmath5 of subgroups of @xmath1 _ is a collection of subgroups of @xmath1 which is closed under conjugation and taking subgroups .
let @xmath4 be the _ classifying space associated to @xmath5_. it is uniquely characterized up to @xmath1-homotopy by the properties that it is a @xmath1-@xmath6-complex and that @xmath7 is contractible if @xmath8 and is empty if @xmath241 . for more information about these spaces
@xmath4 we refer for instance to the survey article @xcite . given a group homomorphism @xmath242 and a family @xmath5 of subgroups of @xmath1 ,
define the family @xmath243 of subgroups of @xmath0 by @xmath244 if @xmath245 is an inclusion of subgroups , we also write @xmath246 instead of @xmath243 .
[ def : isomorphism_conjectures_for_calh?_ast ] fix a group @xmath136 and an equivariant homology theory @xmath151 with values in @xmath108-modules over @xmath136 .
a group @xmath137 over @xmath136 together with a family of subgroups @xmath5 of @xmath1 satisfies the _ isomorphism conjecture ( for @xmath151 ) _ if the projection @xmath247 to the one - point - space @xmath31 induces an isomorphism @xmath248 for all @xmath27 . from now on fix a group @xmath136 and an equivariant homology theory @xmath151 over @xmath136 .
[ the : transitivity ] let @xmath137 be a group over @xmath136 .
let @xmath249 be families of subgroups of @xmath1 .
assume that for every element @xmath250 the group @xmath251 over @xmath136 satisfies the isomorphism conjecture for @xmath252 .
then the up to @xmath1-homotopy unique map @xmath253 induces an isomorphism @xmath254 for all @xmath27 .
in particular , @xmath137 satisfies the isomorphism conjecture for @xmath255 if and only if @xmath137 satisfies the isomorphism conjecture for @xmath5 .
the proof is completely analogous to the one in ( * ? ? ?
* theorem 2.4 , lemma 2.2 ) , where only the case @xmath169 is treated .
[ the : ic_and_limits ] let @xmath137 be a group over @xmath136 .
let @xmath5 be a family of subgroups of @xmath1 . 1 . [
the : ic_and_limits : ic ] let @xmath1 be the directed union of subgroups @xmath35 .
suppose that @xmath151 is continuous and for every @xmath39 the isomorphism conjecture holds for @xmath256 and @xmath257 .
+ then the isomorphism conjecture holds for @xmath137 and @xmath5 ; 2 . [ the : ic_and_limits : fibered_ic ] let @xmath35 be a directed system of groups with @xmath37 and structure maps @xmath38 .
suppose that @xmath151 is strongly continuous and for every @xmath39 the isomorphism conjecture holds for @xmath258 and @xmath259 .
+ then the isomorphism conjecture holds for @xmath137 and @xmath5 .
[ the : ic_and_limits : ic ] the proof is analogous to the one in ( * ? ? ?
* proposition 3.4 ) .
+ [ the : ic_and_limits : fibered_ic ] this follows from the following commutative square whose horizontal arrows are bijective because of lemma [ lem : directed_limits_and_calh?_ast ] and the identification @xmath260
@xmath261 ^ -{t_n^g({e_{{\mathcal{f}}}(g)})}_{\cong } \ar[d ] & { \mathcal{h}}_n^{g}({e_{{\mathcal{f}}}(g ) } ) \ar[d ] \\
\operatorname{colim}_{i \in i } { \mathcal{h}}_n^{g_i}({\{\bullet\ } } ) \ar[r]^-{t_n^g({\{\bullet\}})}_{\cong } & { \mathcal{h}}_n^g({\{\bullet\ } } ) } \ ] ] fix a class of groups @xmath262 closed under isomorphisms , taking subgroups and taking quotients , e.g. , the class of finite groups or the class of virtually cyclic groups . for a group @xmath1
let @xmath263 be the family of subgroups of @xmath1 which belong to @xmath262 .
[ the : isomorphism_conjecture_is_stable_under_colim ] let @xmath137 be a group over @xmath136 . 1 .
[ the : isomorphism_conjecture_is_stable_under_colim : injective ] let @xmath1 be the directed union @xmath201 of subgroups @xmath50 suppose that @xmath151 is continuous and that the isomorphism conjecture is true for @xmath256 and @xmath264 for all @xmath39 .
+ then the isomorphism conjecture is true for @xmath137 and @xmath263 ; 2 .
[ the : isomorphism_conjecture_is_stable_under_colim : general ] let @xmath35 be a directed system of groups with @xmath265 and structure maps @xmath38 .
suppose that @xmath151 is strongly continuous and that the isomorphism conjecture is true for @xmath266 for every @xmath39 and every subgroup @xmath267 . +
then for every subgroup @xmath44 the isomorphism conjecture is true for @xmath268 and @xmath269 .
[ the : isomorphism_conjecture_is_stable_under_colim : injective ] this follows from theorem [ the : ic_and_limits ] [ the : ic_and_limits : ic ] since @xmath270 holds for @xmath39 . + [ the : isomorphism_conjecture_is_stable_under_colim : general ] if @xmath1 is the colimit of the directed system @xmath271 , then the subgroup @xmath44 is the colimit of the directed system @xmath52 .
hence we can assume @xmath54 without loss of generality .
since @xmath262 is closed under quotients by assumption , we have @xmath272 for every @xmath39
. hence we can consider for any @xmath39 the composition @xmath273 because of theorem [ the : ic_and_limits ] [ the : ic_and_limits : fibered_ic ] it suffices to show that the second map is bijective . by assumption
the composition of the two maps is bijective .
hence it remains to show that the first map is bijective . by theorem [ the : transitivity ]
this follows from the assumption that the isomorphism conjecture holds for every subgroup @xmath274 and in particular for any @xmath275 for @xmath276 .
in this section we also deal with the fibered version of the isomorphism conjectures .
( this is not directly needed for the purpose of this paper and the reader may skip this section . )
this is a stronger version of the farrell - jones conjecture .
the fibered farrell - jones conjecture does imply the farrell - jones conjecture and has better inheritance properties than the farrell - jones conjecture .
we generalize ( and shorten the proof of ) the result of farrell - linnell ( * ? ? ?
* theorem 7.1 ) to a more general setting about equivariant homology theories as developed in bartels - lck @xcite .
[ def:(fibered)_isomorphism_conjectures_for_calh?_ast ] fix a group @xmath136 and an equivariant homology theory @xmath151 with values in @xmath108-modules over @xmath136 .
a group @xmath137 over @xmath136 together with a family of subgroups @xmath5 of @xmath1 satisfies the _ fibered isomorphism conjecture ( for @xmath151 ) _
if for each group homomorphism @xmath277 the group @xmath278 over @xmath136 satisfies the isomorphism conjecture with respect to the family @xmath243 .
[ the : fic_and_limits ] let @xmath137 be a group over @xmath136 . let @xmath5 be a family of subgroups of @xmath1 . let @xmath35 be a directed system of groups with @xmath37 and structure maps @xmath38 .
suppose that @xmath151 is strongly continuous and for every @xmath39 the fibered isomorphism conjecture holds for @xmath279 and @xmath259
. then the fibered isomorphism conjecture holds for @xmath137 and @xmath5 .
let @xmath280 be a group homomorphism .
consider the pullback of groups @xmath281 explicitly @xmath282 .
let @xmath283 be the map induced by @xmath195 , @xmath284 and @xmath285 and the pullback property .
one easily checks by inspecting the standard model for the colimit over a directed set that we obtain a directed system @xmath283 of groups indexed by the directed set @xmath286 and the system of maps @xmath287 yields an isomorphism @xmath288 .
the following diagram commutes @xmath289 where the vertical arrows are induced by the obvious projections onto @xmath31 and the horizontal maps are the isomorphisms from lemma [ lem : directed_limits_and_calh?_ast ] .
notice that @xmath290 is a model for @xmath291 .
hence each map @xmath292 is bijective since @xmath258 satisfies the fibered isomorphism conjecture for @xmath259 and hence @xmath293 satisfies the isomorphism conjecture for @xmath294 .
this implies that the left vertical arrow is bijective .
hence the right vertical arrow is an isomorphism .
since @xmath295 is a model for @xmath296 , this means that @xmath297 satisfies the isomorphism conjecture for @xmath298 .
since @xmath280 is any group homomorphism , @xmath137 satisfies the fibered isomorphism conjecture for @xmath5 .
the proof of the following results are analogous to the one in ( * ? ? ?
* lemma 1.6 ) and ( * ? ? ?
* lemma 1.2 ) , where only the case @xmath169 is treated .
[ lem : enlarging_the_family_for_the_fibered_isomorphism_conjecture ] let @xmath137 be a group over @xmath136 and let @xmath299 be families of subgroups of @xmath1 .
suppose that @xmath137 satisfies the fibered isomorphism conjecture for the family @xmath5
. then @xmath137 satisfies the fibered isomorphism conjecture for the family @xmath255 .
[ lem : basic_inheritance_property_of_fibered_conjecture ] let @xmath137 be a group over @xmath136 .
let @xmath277 be a group homomorphism and let @xmath5 be a family of subgroups of @xmath1 .
if @xmath137 satisfies the fibered isomorphism conjecture for the family @xmath5 , then @xmath300 satisfies the fibered isomorphism conjecture for the family @xmath243 . for the remainder of this section fix a class of groups @xmath262 closed under isomorphisms , taking subgroups and taking quotients , e.g. , the families @xmath2 or @xmath3 .
[ lem : fibered_conjecture_passes_to_subgroups ] let @xmath137 be a group over @xmath136 .
suppose that the fibered isomorphism conjecture holds for @xmath137 and @xmath263 .
let @xmath26 be a subgroup .
then the fibered isomorphism conjecture holds for @xmath251 and @xmath301 .
this follows from lemma [ lem : basic_inheritance_property_of_fibered_conjecture ] applied to the inclusion @xmath302 since @xmath303 .
[ the : fibered_isomorphism_conjecture_is_stable_under_colim ] let @xmath137 be a group over @xmath136
. 1 . [
the : fibered_isomorphism_conjecture_is_stable_under_colim : injective ] let @xmath1 be the directed union @xmath201 of subgroups @xmath50 .
suppose that @xmath151 is continuous and that the fibered isomorphism conjecture is true for @xmath256 and @xmath264 for all @xmath39 .
+ then the fibered isomorphism conjecture is true for @xmath137 and @xmath263 ; 2 . [ the : fibered_isomorphism_conjecture_is_stable_under_colim : general ] let @xmath35 be a directed system of groups with @xmath37 and structure maps @xmath38 .
suppose that @xmath151 is strongly continuous and that the fibered isomorphism conjecture is true for @xmath304 and @xmath305 for all @xmath39 .
+ then the fibered isomorphism conjecture is true for @xmath137 and @xmath263 .
[ the : isomorphism_conjecture_is_stable_under_colim : injective ] the proof is analogous to the one in ( * ? ? ?
* proposition 3.4 ) , where the case @xmath169 is considered .
+ [ the : isomorphism_conjecture_is_stable_under_colim : general ] because @xmath262 is closed under taking quotients we conclude @xmath272 .
now the claim follows from theorem [ the : fic_and_limits ] and lemma [ lem : enlarging_the_family_for_the_fibered_isomorphism_conjecture ] .
[ cor : reducing_the_ic ] 1 .
[ cor : reducing_the_ic : unfibered ] suppose that @xmath151 is continuous .
then the ( fibered ) isomorphism conjecture for @xmath137 and @xmath263 is true for all groups @xmath137 over @xmath136 if and only if it is true for all such groups where @xmath1 is a finitely generated group ; 2 .
[ cor : reducing_the_ic : fibered ] suppose that @xmath151 is strongly continuous .
then the fibered isomorphism conjecture for @xmath137 and @xmath263 is true for all groups @xmath137 over @xmath136 if and only if it is true for all such groups where @xmath1 is finitely presented .
let @xmath137 be a group over @xmath136 where @xmath1 is finitely generated . choose a finitely generated free group @xmath67 together with an epimorphism @xmath306 .
let @xmath0 be the kernel of @xmath307 .
consider the directed system of finitely generated subgroups @xmath308 of @xmath0 .
let @xmath309 be the smallest normal subgroup of @xmath0 containing @xmath310 .
explicitly @xmath309 is given by elements which can be written as finite products of elements of the shape @xmath311 for @xmath312 and @xmath313 .
we obtain a directed system of groups @xmath314 , where for @xmath235 the structure map @xmath315 is the canonical projection . if @xmath316 is the canonical projection , then the collection of maps @xmath317 induces an isomorphism @xmath318 . by construction for each @xmath39 the group @xmath319 is finitely presented and the fibered isomorphism conjecture holds for @xmath320 and @xmath321 by assumption .
theorem [ the : fibered_isomorphism_conjecture_is_stable_under_colim ] [ the : fibered_isomorphism_conjecture_is_stable_under_colim : general ] implies that the fibered farrell - jones conjecture for @xmath137 and @xmath263 is true .
in this section we will describe the relevant homology theories over a group @xmath136 and show that they are ( strongly ) continuous .
( we have defined the notion of an equivariant homology theory over a group in definition [ def : equivariant_homology_theory_over_a_group_gamma ] . )
we will need the following [ the : construction_of_equivariant_homology_theories ] suppose that we are given a group @xmath136 and a ring @xmath32 ( with involution ) or a @xmath16-algebra @xmath17 respectively on which @xmath136 acts by structure preserving automorphisms . then : 1 .
[ the : construction_of_equivariant_homology_theories : list ] associated to these data there are equivariant homology theories with values in @xmath49-modules over the group @xmath136 @xmath322 where in the case @xmath323 we will have to impose the restriction to the induction structure that a homomorphisms @xmath152 over @xmath136 induces a transformation @xmath324 only if the kernel of the underlying group homomorphism @xmath65 acts with amenable isotropy on @xmath325 ; 2 .
[ the : construction_of_equivariant_homology_theories : coefficients ] + if @xmath326 is a group over @xmath136 and @xmath26 is a subgroup , then there are for every @xmath27 identifications @xmath327 here @xmath41 and @xmath1 act on @xmath32 and @xmath17 respectively via the given @xmath136-action , @xmath328 and the inclusion @xmath26 , @xmath329 is the algebraic @xmath0-theory of the twisted group ring @xmath330 , @xmath331 is the homotopy @xmath0-theory of the twisted group ring @xmath330 , @xmath332 is the algebraic @xmath14-theory with decoration @xmath333 of the twisted group ring with involution @xmath334 , @xmath335 is the topological @xmath0-theory of the crossed product banach algebra @xmath336 , @xmath337 is the topological @xmath0-theory of the reduced crossed product @xmath16-algebra @xmath338 , and @xmath339 is the topological @xmath0-theory of the maximal crossed product @xmath16-algebra @xmath340 ; 3 .
[ the : construction_of_equivariant_homology_theories : naturality_in_gamma ] let @xmath341 be a group homomorphism . let @xmath32 be a ring ( with involution ) and @xmath17 be a @xmath16-algebra on which @xmath342 acts by structure preserving automorphisms .
let @xmath326 be a group over @xmath343 .
then in all cases the evaluation at @xmath326 of the equivariant homology theory over @xmath343 associated to @xmath344 or @xmath345 respectively agrees with the evaluation at @xmath346 of the equivariant homology theory over @xmath342 associated to @xmath32 or @xmath17 respectively .
[ the : construction_of_equivariant_homology_theories : naturality_in_r_a ] suppose the group @xmath136 acts on the rings ( with involution ) @xmath32 and @xmath347 or on the @xmath16-algebras @xmath17 and @xmath348 respectively by structure preserving automorphisms .
let @xmath349 or @xmath350 be a @xmath136-equivariant homomorphism of rings ( with involution ) or @xmath16-algebras respectively .
then @xmath351 induces natural transformations of homology theories over @xmath136 @xmath352 they are compatible with the identifications appearing in assertion [ the : construction_of_equivariant_homology_theories : coefficients ] ; 5 .
[ the : construction_of_equivariant_homology_theories_naturality_in_k ] let @xmath136 act on the @xmath16-algebra @xmath17 by structure preserving automorphisms .
we can consider @xmath17 also as a ring with structure preserving @xmath1-action .
then there are natural transformations of equivariant homology theories with values in @xmath49-modules over @xmath136 @xmath353 they are compatible with the identifications appearing in assertion [ the : construction_of_equivariant_homology_theories : coefficients ] .
next we want to show [ lem : homology_theories_are_(strongly)_continuous ] suppose that we are given a group @xmath136 and a ring @xmath32 ( with involution ) or a @xmath16-algebra @xmath17 respectively on which @xmath1 acts by structure preserving automorphisms . then the homology theories with values in @xmath49-modules over @xmath136 @xmath354 ( see theorem [ the : construction_of_equivariant_homology_theories ] ) are strongly continuous in the sense of definition [ def : strongly_continuous_equivariant_homology_theory ] , whereas @xmath355 is only continuous .
we begin with @xmath356 and @xmath357 .
we have to show for every directed systems of groups @xmath358 with @xmath37 together with a map @xmath359 that the canonical maps @xmath360 are bijective for all @xmath27 . obviously @xmath361 is the colimit of rings @xmath362 .
now the claim follows for @xmath11 for @xmath363 from ( * ? ? ? * ( 12 ) on page 20 ) .
using the bass - heller - swan decomposition one gets the results for @xmath11 for all @xmath27 and that the map @xmath364 is bijective for all @xmath27 and all @xmath365 for the nil - groups @xmath366 defined by bass ( * ? ? ?
now the claim for homotopy @xmath0-theory follows from the spectral sequence due to weibel ( * ? ? ? * theorem 1.3 ) .
next we treat @xmath367 .
we have to show for every directed systems of groups @xmath35 with @xmath37 together with a map @xmath368 that the canonical map @xmath369 is bijective for all @xmath27 .
recall from ( * ? ? ?
* definition 17.1 and definition 17.7 ) that @xmath370)\right ) \ ; \text { for } m \ge 0 . \end{aligned}\ ] ] since @xmath371 is @xmath372 , it suffices to show that @xmath373 is bijective for all @xmath27 .
we give the proof of surjectivity for @xmath374 only , the proofs of injectivity for @xmath374 and of bijectivity for the other values of @xmath173 are similar .
the ring @xmath361 is the colimit of rings @xmath375 .
let @xmath376 and @xmath377 for @xmath378 be the structure maps .
one can define @xmath361 as the quotient of @xmath379 , where @xmath380 and @xmath381 satisfy @xmath382 if and only if @xmath383 holds for some @xmath384 with @xmath385 .
the addition and multiplication is given by adding and multiplying representatives belonging to the same @xmath386 .
let @xmath387 be the set of @xmath388-matrices with entries in @xmath361 .
given @xmath389 , define @xmath390 and @xmath391 by applying @xmath46 and @xmath392 to each entry of the matrix @xmath393 .
we need the following key properties which follow directly from inspecting the model for the colimit above : 1 . given @xmath394 , there exists @xmath39 and @xmath389 with @xmath395 ; 2 .
given @xmath389 and @xmath396 with @xmath397 , there exists @xmath384 with @xmath385 and @xmath398 .
an element @xmath399 $ ] in @xmath400 is represented by a quadratic form on a finitely generated free @xmath361-module , i.e. , a matrix @xmath401 for which there exists a matrix @xmath402 with @xmath403 , where @xmath404 is given by transposing the matrix @xmath348 and applying the involution of @xmath32 elementwise .
fix such a choice of a matrix @xmath348 .
choose @xmath39 and @xmath405 with @xmath406 .
then @xmath407 is invertible .
hence we can find @xmath408 with @xmath235 such that @xmath409 is invertible .
put @xmath410 . then @xmath411 and @xmath412 .
hence @xmath413 defines an element @xmath414 \in l_n^h(r\rtimes g_j)$ ] which is mapped to @xmath399 $ ] under the homomorphism @xmath415 induced by @xmath416 .
hence the map @xmath417 of is surjective .
next we deal with @xmath418 .
we have to show for every directed systems of groups @xmath35 with @xmath37 together with a map @xmath419 that the canonical map @xmath420 is bijective for all @xmath27 .
since topological @xmath0-theory is a continuous functor , it suffices to show that the colimit ( or sometimes also called inductive limit ) of the system of banach algebras @xmath421 in the category of banach algebras with norm decreasing homomorphisms is @xmath422 .
so we have to show that for any banach algebra @xmath348 and any system of ( norm deceasing ) homomorphisms of banach algebras @xmath423 compatible with the structure maps @xmath424 there exists precisely one homomorphism of banach algebras @xmath425 with the property that its composition with the structure map @xmath426 is @xmath427 for @xmath39 .
it is easy to see that in the category of @xmath71-algebras the colimit of the system @xmath428 is @xmath429 with structure maps @xmath430 .
hence the restrictions of the homomorphisms @xmath427 to the subalgebras @xmath431 yields a homomorphism of central @xmath71-algebras @xmath432 uniquely determined by the property that the composition of @xmath433 with the structure map @xmath434 is @xmath435 for @xmath436 .
if @xmath76 exists , its restriction to the dense subalgebra @xmath19 has to be @xmath433 .
hence @xmath76 is unique if it exists .
of course we want to define @xmath76 to be the extension of @xmath433 to the completion @xmath437 of @xmath19 with respect to the @xmath20-norm .
so it remains to show that @xmath438 is norm decreasing .
consider an element @xmath439 which is given by a finite formal sum @xmath440 , where @xmath441 is some finite subset of @xmath1 and @xmath442 for @xmath443 .
we can choose an index @xmath408 and a finite set @xmath444 such that @xmath445 is one - to - one .
for @xmath443 let @xmath446 denote the inverse image of @xmath447 under this map .
consider the element @xmath448 in @xmath449 . by construction
we have @xmath450 and @xmath451 .
we conclude @xmath452 the proof for @xmath453 follows similarly , using the fact that by definition of the norm on @xmath454 every @xmath455-homomorphism of @xmath456 into a @xmath16-algebra @xmath348 extends uiniquely to @xmath454 .
the proof for the continuity of @xmath323 follows from ( * ? ? ?
* theorem 4.1 ) .
notice that we have proved all promised results of the introduction as soon as we have completed the proof of theorem [ the : construction_of_equivariant_homology_theories ] which we have used as a black box so far .
in this section we explain how one can construct equivariant homology theories from spectra over groupoids . a _ spectrum _
@xmath457 is a sequence of pointed spaces @xmath458 together with pointed maps called _ structure maps _
a _ ( strong ) map _ of spectra ( sometimes also called function in the literature ) @xmath460 is a sequence of maps @xmath461 which are compatible with the structure maps @xmath462 , i.e. , we have @xmath463 for all @xmath27 .
this should not be confused with the notion of a map of spectra in the stable category ( see ( * ? ? ?
* iii.2 . ) ) . recall that the homotopy groups of a spectrum are defined by @xmath464 where the system @xmath465 is given by the composition @xmath466 of the suspension homomorphism and the homomorphism induced by the structure map .
we denote by @xmath467 the category of spectra .
a _ weak equivalence _ of spectra is a map @xmath468 of spectra inducing an isomorphism on all homotopy groups .
given a small groupoid @xmath255 , denote by @xmath469 the category of small groupoids over @xmath255 , i.e. , an object is a functor @xmath470 with a small groupoid as source and a morphism from @xmath470 to @xmath471 is a functor @xmath472 satisfying @xmath473 . we will consider a group @xmath136 as a groupoid with one object and @xmath136 as set of morphisms .
an _ equivalence _ @xmath472 of groupoids is a functor of groupoids @xmath67 for which there exists a functor of groupoids @xmath474 such that @xmath475 and @xmath476 are naturally equivalent to the identity functor .
a functor @xmath477 of small groupoids is an equivalence of groupoids if and only if it induces a bijection between the isomorphism classes of objects and for any object @xmath478 the map @xmath479 induced by @xmath67 is an isomorphism of groups .
[ lem : from_spectra_over_groupoids_equiv_homology_theories ] let @xmath136 be a group . consider a covariant functor @xmath480 which sends equivalences of groupoids to weak equivalences of spectra .
then we can associate to it an equivariant homology theory @xmath481 ( with values in @xmath49-modules ) over @xmath136 such that for every group @xmath326 over @xmath136 and subgroup @xmath26 we have a natural identification @xmath482 if @xmath483 is a natural transformation of such functors @xmath484 , then it induces a transformation of equivariant homology theories over @xmath136 @xmath485 such that for every group @xmath326 over @xmath136 and subgroup @xmath26 the homomorphism @xmath486 agrees under the identification above with @xmath487 .
we begin with explaining how we can associate to a group @xmath326 over @xmath136 a @xmath1-homology theory @xmath488 with the property that for every subgroup @xmath489 we have an identification @xmath490 we just follow the construction in ( * ? ? ?
* section 4 ) .
let @xmath491 be the _ orbit category _ of @xmath1 , i.e. , objects are homogenous spaces @xmath132 and morphisms are @xmath1-maps . given a @xmath1-set @xmath347 , the associated _ transport groupoid _
@xmath492 has @xmath347 as set of objects and the set of morphisms from @xmath493 to @xmath494 consists of the subset @xmath495 of @xmath1 .
composition is given by the group multiplication .
a @xmath1-map of sets induces a functor between the associated transport groupoids in the obvious way .
in particular the projection @xmath496 induces a functor of groupoids @xmath497 .
thus @xmath492 becomes an object in @xmath498 by the composite @xmath499 .
we obtain a covariant functor @xmath500 .
its composition with the given functor @xmath501 yields a covariant functor @xmath502 now define @xmath503 where @xmath504 is the @xmath1-homology theory which is associated to @xmath505 and defined in ( * ? ? ?
* section 4 and 7 ) .
namely , if @xmath143 is a @xmath1-@xmath6-complex , we can assign to it a contravariant functor @xmath506 sending @xmath132 to @xmath507 and put @xmath508 for the spectrum @xmath509 ( which is denoted in @xcite by @xmath510 ) .
next we have to explain the induction structure .
consider a group homomorphism @xmath152 of groups over @xmath136 and an @xmath41-@xmath6-complex @xmath143 .
we have to construct a homomorphism @xmath511 this will be done by constructing a map of spectra @xmath512 we follow the constructions in ( * ? ? ?
* section 1 ) .
the homomorphism @xmath76 induces a covariant functor @xmath513 by sending @xmath514 to @xmath515 .
given a contravariant functor @xmath516 , we can assign to it its induction with @xmath517 which is a contravariant functor @xmath518 . given a contravariant functor @xmath519
, we can assign to it its restriction which is the contravariant functor @xmath520 .
induction @xmath79 and @xmath185 form an adjoint pair . given an @xmath41-@xmath6-complex @xmath143
, there is a natural identification @xmath521 . using ( * ? ? ?
* lemma 1.9 ) we get for an @xmath41-@xmath6-complex @xmath143 a natural map of spectra @xmath522 given an @xmath41-set @xmath347 , we obtain a functor of groupoids @xmath523 sending @xmath524 to @xmath525 and a morphism in @xmath526 given by a group element @xmath527 to the one in @xmath528 given by @xmath529 .
this yields a natural transformation of covariant functors @xmath530 from @xmath531 .
composing with the functor @xmath501 gives a natural transformation of covariant functors @xmath532 from @xmath533 to @xmath534 .
it induces a map of spectra @xmath535 its composition with the maps of spectra constructed beforehand yields the desired map of spectra @xmath536 .
we omit the straightforward proof that the axioms of an induction structure are satisfied .
this finishes the proof of theorem [ the : construction_of_equivariant_homology_theories ] .
the statement about the natural transformation @xmath483 is obvious .
the last step in completing the proof of theorem [ the : construction_of_equivariant_homology_theories ] is to prove the following theorem [ the : necessary_functors_from_groupoids_to_spectra ] because then we can apply it in combination with lemma [ lem : from_spectra_over_groupoids_equiv_homology_theories ] .
( actually we only need the version of theorem [ the : necessary_functors_from_groupoids_to_spectra ] , where @xmath255 is given by a group @xmath136 . )
let @xmath537 be the subcategory of @xmath538 which has the same objects and for which a morphism from @xmath470 to @xmath539 given by a functor @xmath472 satisfying @xmath473 has the property that for every object @xmath478 the group homomorphism @xmath540 induced by @xmath67 has a finite kernel .
denote by @xmath541 , @xmath542 , and @xmath543 the categories of rings , rings with involution and @xmath16-algebras .
[ the : necessary_functors_from_groupoids_to_spectra ] let @xmath255 be a fixed groupoid .
let @xmath544 , @xmath545 , or @xmath546 respectively be a covariant functor .
then there exist covariant functors @xmath547 together with natural transformations @xmath548 of functors from @xmath469 or @xmath537 respectively to @xmath467 such that the following holds : 1 .
[ the : k - theory_spectra_of_groupoids : weak_equivalences ] let @xmath549 be objects for @xmath550 and @xmath551 be a morphism between them in @xmath469 or @xmath537 respectively such that the underlying functor of groupoids @xmath552 is an equivalence of groupoids .
then the functors send @xmath67 to a weak equivalences of spectra ; 2 .
[ the : k - theory_spectra_of_groupoids : one_object ] let @xmath470 be an object in @xmath469 or @xmath537 respectively such that the underlying groupoid @xmath553 has only one object @xmath163 .
let @xmath554 be its automorphisms group .
we obtain a ring @xmath555 , a ring @xmath555 with involution , or a @xmath16-algebra @xmath556 with @xmath1-operation by structure preserving maps from the evaluation of the functor @xmath32 or @xmath17 respectively at @xmath557 .
then : @xmath558 where @xmath559 is the algebraic @xmath0-theory of the twisted group ring @xmath560 , @xmath561 is the homotopy @xmath0-theory of the twisted group ring @xmath560 , @xmath562 is the algebraic @xmath14-theory with decoration @xmath333 of the twisted group ring with involution @xmath560 , @xmath563 is the topological @xmath0-theory of the crossed product banach algebra @xmath564 , @xmath565 is the topological @xmath0-theory of the reduced crossed product @xmath16-algebra @xmath566 , and @xmath567 is the topological @xmath0-theory of the maximal crossed product @xmath16-algebra @xmath568 . + the natural transformations @xmath569 , @xmath570 , @xmath571 and @xmath572 become under this identifications the obvious change of rings and theory homomorphisms 3 . [
the : k - theory_spectra_of_groupoids : naturality ] these constructions are in the obvious sense natural in @xmath32 and @xmath17 respectively and in @xmath255 .
we defer the details of the proof of theorem [ the : necessary_functors_from_groupoids_to_spectra ] in @xcite .
its proof requires some work but there are many special cases which have already been taken care of .
if we would not insist on groupoids but only on groups as input , these are the standard algebraic @xmath0- and @xmath14-theory spectra or topological @xmath0-theory spectra associated to group rings , group banach algebras and group @xmath16-algebras .
the construction for the algebraic @xmath0- and @xmath14-theory and the topological @xmath0-theory in the case , where @xmath1 acts trivially on a ring @xmath32 or a @xmath16-algebra are already carried out or can easily be derived from @xcite , @xcite , and @xcite except for the case of a banach algebra .
the case of the @xmath0-theory spectrum associated to an additive category with @xmath1-action has already been carried out in @xcite .
the main work which remains to do is to treat the banach case and to construct the relevant natural transformation from @xmath573 to @xmath574 a. bartels and w. lck .
induction theorems and isomorphism conjectures for @xmath0- and @xmath14-theory .
preprintreihe sfb 478 geometrische strukturen in der mathematik , heft 331 , mnster , arxiv : math.kt/0404486 , to appear in forum math . ,
2004 .
a. bartels , w. lck , and h. reich .
the @xmath575-theoretic farrell - jones conjecture for hyperbolic groups .
preprintreihe sfb 478 geometrische strukturen in der mathematik , heft 434 , mnster , arxiv : math.gt/0609685 , 2007 .
a. bartels , w. lck , and h. reich . on the farrell - jones conjecture and its applications .
preprintreihe sfb 478 geometrische strukturen in der mathematik , heft 457 , mnster , arxiv : math.kt/0703548 , 2007 .
p. baum , a. connes , and n. higson . classifying space for proper actions and @xmath575-theory of group @xmath576-algebras . in _
@xmath577-algebras : 19431993 ( san antonio , tx , 1993 ) _ , pages 240291 .
soc . , providence , ri , 1994 .
b. h. bowditch .
notes on gromov s hyperbolicity criterion for path - metric spaces . in _ group theory from a geometrical viewpoint ( trieste , 1990 ) _ , pages 64167 .
world sci .
publishing , river edge , nj , 1991 .
v. lafforgue .
une dmonstration de la conjecture de baum - connes pour les groupes rductifs sur un corps @xmath582-adique et pour certains groupes discrets possdant la proprit ( t ) . , 327(5):439444 , 1998 .
w. lck .
survey on classifying spaces for families of subgroups . in _
infinite groups : geometric , combinatorial and dynamical aspects _ , volume 248 of _ progr .
_ , pages 269322 .
birkhuser , basel , 2005 .
e. k. pedersen and c. a. weibel .
a non - connective delooping of algebraic @xmath575-theory . in _ algebraic and geometric topology ; proc .
rutgers uni .
, new brunswick 1983 _ , volume 1126 of _ lecture notes in mathematics _ , pages 166181 .
springer , 1985 .
d. quillen .
higher algebraic @xmath575-theory .
algebraic @xmath0-theory , i : higher @xmath0-theories ( proc .
battelle memorial inst . ,
seattle , wash . , 1972 )
_ , pages 85147 .
lecture notes in math .
, vol . 341 .
springer - verlag , berlin , 1973 .
c. a. weibel .
homotopy algebraic @xmath0-theory . in _
algebraic @xmath0-theory and algebraic number theory ( honolulu , hi , 1987 ) _ , volume 83 of _ contemp .
_ , pages 461488 .
soc . , providence , ri , 1989 . | we investigate when isomorphism conjectures , such as the ones due to baum - connes , bost and farrell - jones , are stable under colimits of groups over directed sets ( with not necessarily injective structure maps ) .
we show in particular that both the @xmath0-theoretic farrell - jones conjecture and the bost conjecture with coefficients hold for those groups for which higson , lafforgue and skandalis have disproved the baum - connes conjecture with coefficients .
+ keywords : bost conjecture , inheritance under colimits .
+ ams - classification : 19k99 , 18f25 , 55n91 . |
the randall - sundrum ( rs ) model @xcite is realized in a slice of the ads@xmath0 space - time with the background warped metric : @xmath1 where @xmath2 ( @xmath3 ) , @xmath4 being the `` radius '' of extra dimension , and @xmath5 is the minkowski metric .
the points @xmath6 and @xmath7 are identified .
the parameter @xmath8 defines a 5-dimensional scalar curvature of the ads@xmath0 space .
namely , the ricci curvature invariant @xmath9 .
that is why , in what follows we will call @xmath8 `` curvature '' .
there are two 3d branes in the model with equal and opposite tensions located at the point @xmath10 ( called the _ tev brane _ ) and point @xmath11 ( referred to as the _ plank brane _ ) . if @xmath12 , the tension on the tev brane is negative , whereas the tension on the planck brane is positive .
all the sm fields are confined to the tev brane , while the gravity propagates in all five dimensions .
it is necessary to note that the metric is chosen in such a way that 4-dimensional coordinates @xmath13 are galilean on the tev brane where all the sm field live , since the warp factor is equal to unity at @xmath10 .
( for details , see ref .
@xcite ) . ] by integrating a 5-dimensional action over variable @xmath14 , one gets an effective 4-dimensional action , that results in the `` hierarchy relation '' between the reduced planck scale @xmath15 and 5-dimensional gravity scale @xmath16 : @xmath17 the reduced 5-dimensional planck mass @xmath16 is related to the planck mass by @xmath18 . from the point of view of a 4-dimensional observer located on the tev brane , there exists an infinite number of graviton kk excitations with masses @xmath19 where @xmath20 are zeros of the bessel function @xmath21 .
the interaction lagrangian on the visible brane looks like the following : @xmath22 here @xmath23 is the energy - momentum tensor of the matter on this brane , @xmath24 is a graviton field with the kk - number @xmath25 , and @xmath26 is a scalar field called radion .
the parameter @xmath27 in eq .
, @xmath28 is a physical scale on the tev brane . in most of the papers which treat the rs model , the hierarchy relation is of the form @xmath29 while the curvature and 5-dimensional gravity scale are chosen to be of the order of the planck mass : @xmath30 in such a scheme , one obtains a series of massive graviton resonances in the tev region which interact rather strongly with the sm fields , since @xmath31 tev on the tev brane .
the lightest modes of the kk graviton have masses around 1 tev .
thus , an experimental signature of the `` _ _ large curvature option _ _ '' of the rs model is the real or virtual production of the massive kk graviton resonances .
there are lower bounds on the mass of the lightest kk mode , coming from the measurements of @xmath32 and @xmath33 final states on the tevatron @xcite : @xmath34 the lhc search limit on @xmath35 is @xcite @xmath36 for the di - jet mode and integrated luminosity @xmath37 . for the di - photon mode and @xmath38 , the limit looks like @xcite @xmath39 in the case of large curvature , the size of the @xmath40 slice is extremely small : @xmath41 where @xmath42 is a planck length .
thus , in order to explain the huge value of @xmath15 , one has to introduce large mass scales @xmath16 , @xmath8 , as well as @xmath43 . in other words ,
_ hierarchy problem is not solved , but reformulated _ in terms of the new parameter related to the size of the bulk along the extra dimension . in the present paper
we will consider the `` _ _ small curvature option _ _ '' of the rs model @xcite . in what follows , the 5-dimensional reduced planck mass @xmath16 is taken to be one or few tev . following ref .
@xcite-@xcite , we chose the parameter @xmath8 to be @xmath44 .
these values of @xmath8 obey the bounds derived in ref .
@xcite : @xmath45 then the mass of the lightest kk excitation @xmath46 gev . on the contrary ,
the value of the mass scale @xmath27 is rather large : @xmath47 it immediately follows from eqs .
, that there is no problem with the radion field @xmath26 in our scheme , since its coupling to the sm fields ( @xmath48 ) is strongly suppressed . as for the massive gravitons ,
their couplings are also defined by the scale @xmath27 .
however , the smallness of the coupling is compensated by the large number of the gravitons that can be produced in any inclusive process . as a result
, magnitudes of cross sections will be defined by the 5-dimensional gravity scale @xmath16 and collision energy @xmath49 ( see below ) .
thus , we have an infinite number of low - mass kk resonances with the small mass splitting , in contrast with the standard rs scenario .
nevertheless , due to the warp geometry of the @xmath40 space - time , the rs model with the small curvature differs significantly from the add model @xcite ( at least , at @xmath50 ev ) , as was demonstrated in ref .
@xcite .
let us stress that the hierarchy relation explains the value of the 4-dimensional planck mass without introducing large mass scales , if one put @xmath51 .
recently , the small curvature option of the rs model has been checked by the delphi collaboration .
the gravity effects were searched for by studying photon energy spectrum in the process @xmath52 .
no deviations from the sm prediction were seen . as a result , the following bound has been obtained @xcite : @xmath53 that corresponds to the reduced 5-dimensional scale @xmath54 tev ( see the relation between @xmath55 and @xmath16 after eq . ) .
note that this limit could not be inferred from the limits already given for larger than two flat dimensions owing to the totally different spectrum of the photon @xcite .
following ref .
@xcite , we will consider now the bhabha process at the ilc collider @xcite mediated by massive graviton exchanges : , @xmath56 are also promising reactions . ]
@xmath57 the collision energy @xmath49 is taken to be 1 tev .
it means that we are working in the following region : @xmath58 the matrix element of the process looks like @xmath59 the fist factor in eq . is the contraction of the the tensor part of the graviton propagator , @xmath60 , and the energy - momentum tensor of the electron , @xmath61 : @xmath62 the second factor in eq .
is _ universal _ for all types of processes mediated by @xmath63 or @xmath64-channel exchange of the massive kk excitations .
let us consider first the contribution to @xmath65 from the @xmath63-channel gravitons : @xmath66 here @xmath67 denotes the total width of the graviton with the kk number @xmath25 and mass @xmath68 @xcite : @xmath69 where @xmath70 .
note that the main contribution to sum comes from the region @xmath71 . for such large values of the kk number @xmath25 , _ nonzero widths of the gravitons should be taken into account_. the sum in eq .
can be calculated analytically by the use of the formula @xcite @xmath72 where @xmath73 ( @xmath74 ) are zeros of the function @xmath75 . as a result
, the following explicit expression can be obtained from ( for details , see ref .
@xcite ) : @xmath76 with the notations @xmath77 as one can see from , the magnitude of @xmath78 is defined by @xmath16 and @xmath49 , _ not by _ @xmath27 , although the latter describes the graviton coupling with the matter in the effective lagrangian
. the following inequalities result from : @xmath79 @xmath80 where the notation @xmath81 \ , \mathcal{s}(s)$ ] was introduced .
note that the ratio @xmath82 decreases with energy and becomes small at @xmath83 , while @xmath84 tends to @xmath85 .
should one ignores the widths of the massive gravitons , and replace a summation in kk number by integration over graviton masses , @xmath86 he gets @xcite : @xmath87 in contrast to the exact formula .
however , the series of low - massive resonances in the rs model with the small curvature can be replace by the continuous spectrum _ only in a trans - planckian energy region _ @xmath88 .
it can be understood as follows .
one may regard the set of narrow graviton resonances to be the continuous mass spectrum ( within a relevant interval of @xmath25 ) , if only @xmath89 is satisfied , where @xmath90 is the mass splitting . as was shown in ref .
@xcite , the inequality is equivalent to the above mentioned inequality @xmath91 we are working in another kinematical region , @xmath92 , since the 5-dimensional scale @xmath93 is assumed to be equal to ( or larger than ) 1 tev , while the collision energy is fixed to be 1 tev .
that is why , for our calculations we will use the analytical expression which takes into account both the discrete character of the graviton spectrum and finite widths of the kk gravitons .
our main formula can be also applied to the scattering of the brane particles , induced by exchanges of @xmath64-channel gravitons @xcite . in the kinematical region @xmath94
, the contribution from the @xmath64-channel gravitons looks like @xcite : @xmath95 let us now apply our theoretical formulae , for estimating virtual graviton contributions to the process .
the relations of cross sections with the quantities @xmath78 and @xmath96 can be found in ref .
@xcite .
note first that at the lep2 energy ( @xmath97 gev ) the gravity effects are very small with respect to the sm cross section @xcite .
thus , the above mentioned bound on @xmath16 from lep should be also applied to the 5-dimensional scale in our scheme . in fig .
[ fig:2.5tev_1gev ] the graviton contribution is presented as a function of the scattering angle of the final electrons at the ilc energy @xmath98 tev ( solid line ) .
another prediction is also shown ( dashed line ) which was calculated under the assumption that the dense spectrum of the kk gravitons can be approximated by a continuum @xcite .
the next figure shows the dependence of the cross section ratios on the gravity scale @xmath16 . figs .
[ fig:1.8tev_0.9gev ] , [ fig:1.8tev_0.994gev ] demonstrate us that these ratios can be very sensitive to small variations of the curvature @xmath8 within a narrow region ( 0.9 gev
1 gev , in our case ) .
let us note that the interference of gravity with the sm forces is constructive in the bhabha scattering , and the ratio @xmath99 is lager than 1 . up to now
, we considered the fixed collision energy .
however , the effects related with non - zero graviton widths remain significant after energy smearing within some interval around average value of @xmath49 , as one can see in fig .
[ fig : bhabha_average ] .
note that the energy smearing has a small influence on the cross section ratios in the kinematical region @xmath100 @xcite .
in the present paper the contributions of the virtual @xmath63- and @xmath64-channel kk gravitons to the bhabha scattering at the ilc energy @xmath98 tev were estimated .
we have considered the small curvature option of the rs model with two branes ( @xmath101 ) .
in such a scheme , the kk graviton spectrum is a series of the narrow low - mass resonances .
the 5-dimensional planck scale @xmath55 was taken to be 1.8 tev and 2.5 tev , with the curvature parameter being restricted to the small region @xmath102 , that means @xmath103 .
for numerical estimates , we used formula , which describes the gravity contribution to the process - independent part of the scattering amplitude @xmath65 . for comparison
, we have made calculations by using eqs . which treat the spectrum of the kk graviton as a continuum .
it was demonstrated that the sum in the kk number can not be approximated by integration over graviton mass .
moreover , the graviton widths should be properly taken into account , as formula does .
our predictions are presented in figs .
[ fig:2.5tev_1gev]-[fig : bhabha_average ] @xcite .
everywhere @xmath104 means the sm differential cross section with respect to @xmath105 , while @xmath106 means the differential cross section which takes into account both the sm and gravity interactions .
one should conclude that at fixed value of the fundamental gravity scale @xmath55 , the cross section ratio is very sensitive even to slight variations of the curvature @xmath8 ( compare figs .
[ fig:1.8tev_1gev ] , [ fig:1.8tev_0.9gev ] and [ fig:1.8tev_0.994gev ] ) .
the ratio @xmath99 can reach tens at some values of the parameters ( see figs .
[ fig:1.8tev_0.994gev ] ) .
such a behavior of the cross section ratios comes from the explicit form of the function @xmath78 . indeed , at @xmath107 the parameter @xmath108 in eq .
is much less than 1 , and the function @xmath78 has a significant real part . in the trans - planckian region @xmath109 . ] the interference with the sm contributions results in a nontrivial dependence of @xmath99 on @xmath105 . given the approximation with zero real part is used , no interference terms exist , and @xmath110-dependence of the cross section ratio becomes similar for all sets of the parameters ( dashed curves in figs .
[ fig:2.5tev_1gev]-[fig:1.8tev_0.994gev ] ) .
if the parameter @xmath111 in formula obeys the equation @xmath112 , the denominator in becomes small , and , correspondingly , the function @xmath78 becomes large .
it results in the rapid variation of the gravity contribution near corresponding values of @xmath8 .
it is worth to note that both a discrete character of the mass spectrum and nonzero widths of the kk gravitons are also important in a model with a _ flat _ compact extra dimension @xcite .
i am grateful to g.f . giudice and v.a .
petrov for fruitful discussions .
99 l. randall and r. sundrum , phys .
rev . lett .
* 83 * ( 1999 ) 3370 e.e .
boos , yu.a .
kubyshin , , m.n .
smolyakov and i.p .
volobuev , class .
* 19 * , 4591 ( 2002 ) ; e.e .
boos , yu.s .
mikhailov , m.n .
smolyakov and i.p .
volobuev , nucl .
b * 717 * ( 2005 ) 19 h. davoudiasl , j.l .
hewett and t.g .
rizzo , phys .
d * 63 * ( 2001 ) 075004 e. gallo , plenary talk at the ichep 2006 , moscow , russia , july 26 - august 2 , 2006 s. shmatov , talk at the ichep 2006 , moscow , russia , july 26 - august 2 , 2006 a.v . kisselev and v.a .
petrov , phys .
d * 71 * ( 2005 ) 124032 a.v .
kisselev , phys . rev .
d * 73 * ( 2006 ) 024007 g.f .
giudice , t. plehn and a. strumia , nucl . phys .
b * 706 * ( 2005 ) 455 n. arkani - hamed , s. dimopoulos and g. dvali , phys .
b * 429 * ( 1998 ) 263 ; i. antoniadis , n. arkani - hamed , s. dimopoulos and g. dvali , phys .
b * 436 * ( 1998 ) 257 ; n. arkani - hamed , s. dimopoulos and g. dvali , phys .
d * 59 * ( 1999 ) 086004 a.v .
kisselev and v.a .
petrov , eur .
j. c * 36 * ( 2004 ) 103 s. ask , v. hedberg , and f .-
navarria , talk at the 23-rd international symposium on lepton - photon interactions at high energy ( lp07 ) , daegu , korea , august 13 - 18 , 2007 ( hep - ex/0707.2102 ) g. laow _ et al .
_ , ilc technical review committee , second report , 2003 , slac - r-606 a.v .
kisselev , eur .
j. c * 42 * ( 2005 ) 217 a.v .
kisselev , jetp * 03 * ( 2007 ) 006 g.n .
watrson , _ treatise on the theory of bessel functions _
( mcmillan , new york , 1922 ) | the randall - sundrum ( rs ) model with the small curvature is studied . in such a scheme the mass spectrum of kaluza - klein ( kk ) gravitons is similar to that in a model with one extra flat dimension .
the gravity effects in the bhabha scattering at the energy 1 tev are estimated .
the calculations are based on the analytical formula which describes virtual graviton contributions .
it takes into account both a discrete character of the mass spectrum and nonzero widths of the kk gravitons . |
a @xmath0 surface @xmath3 is characterized by the condition @xmath4 and simply connectedness .
it means that @xmath3 is a 2-dimensional calabi - yau manifold . v. v. batyrev @xcite
introduced the notion of the reflexive polytope for the study of calabi - yau manifolds . in this article
we use the 3-dimensional reflexive polytopes with at most terminal singularities .
such a polytope @xmath5 is defined by the intersection of several half spaces @xmath6 in @xmath7 with the conditions \(i ) every vertex is a lattice point , \(ii ) the origin is the unique inner lattice point , \(iii ) only the vertices are the lattice points on the boundary .
+ moreover , if a reflexive polytope satisfies the condition \(iv ) every face is triangle and its 3 vertices generate the lattice , + it is called a fano polytope . all 3-dimensional 5-verticed reflexive polytopes with at most terminal singularity
are listed up ( see @xcite or @xcite ) : @xmath8 among them , @xmath9 and @xmath10 are fano polytopes .
we can find a family of @xmath0 surfaces for each polytope by a natural method .
in this article we study the polytopes @xmath11 and @xmath12 .
namely , we determine the lattice structure , the period differential equation and the projective monodromy group for each of them .
t. ishige @xcite has made a detailed research on the family of @xmath0 surfaces coming from the polytope @xmath13 .
especially he noticed characterization of the corresponding monodromy group by a numerical approach .
inspired by ishige s discovery , we have studied families of @xmath0 surfaces derived from the other polytopes @xmath14 and @xmath10 . we have made an intensive study on the polytope @xmath10 in our previous article @xcite .
there , we have studied the period map for a family , saying @xmath15 , of @xmath0 surfaces , where @xmath16 namely , we have determined the lattice structure of a generic member of the family @xmath17 , the period differential equation and the projective monodromy group using the torelli type theorem for polarized @xmath0 surfaces and the lattice theory .
furthermore , we have shown that our differential equation coincides with the uniformizing differential equation of the hilbert modular orbifold for @xmath18 studied by sasaki - yoshida @xcite and t. sato @xcite . here
, we study the remaining cases @xmath11 and @xmath12 . namely we investigate corresponding families @xmath19 of @xmath0 surfaces . in section 1
, we show explicit defining equations for the families @xmath20 ( see ( [ eq : p1 ] ) , ( [ eq : p2 ] ) and ( [ eq : p3 ] ) ) and we introduce elliptic fibrations for these families .
the singular fibres of each elliptic fibration are described in table 1 . in section 2
, we determine the lattice structure for a generic member of each family @xmath19 .
namely , we obtain the nron - severi lattice @xmath21 as in table 2 .
note that in the case @xmath10 we could determine @xmath22 for @xmath23 by a naive method ( see @xcite ) . in this article
we need more advanced theory of the mordell - weil lattice due to t. shioda @xcite . for 95 weighted projective @xmath0 surfaces ,
there is a result of s. m. belcastro @xcite . and for @xmath0 surfaces with non - symplectic involution , there is a result of v. v. nikulin @xcite .
our case is not contained in these results .
furthermore , we note that the result of k. koike @xcite and our result in this article support the mirror symmetry conjecture ( see remark [ mirrorrem ] ) . in section 3 , we determine the period differential equations ( theorem [ periodde ] )
furthermore , we obtain their monodromy groups ( theorem [ monodromytheorem ] ) .
we obtain a family of algebraic surfaces by the following canonical procedure from @xmath24 : \(i ) make a toric 3-fold @xmath25 from the reflexive polytope @xmath26 .
this is a rational variety .
\(ii ) take a divisor @xmath27 on @xmath25 that is linearly equivalent to @xmath28 .
\(iii ) generically @xmath27 is represented by a @xmath0 surface .
we obtain the corresponding families of @xmath0 surfaces @xmath29 for @xmath30 given by @xmath31 we can find an elliptic fibration for every surface of our family @xmath32 . moreover we can describe these surfaces in the form @xmath33 where @xmath34 ( @xmath35 , resp . ) is a polynomial of @xmath36 with @xmath37 ( @xmath38 , resp . ) . in this paper
we call it the kodaira normal form . from the kodaira normal form
we can obtain singular fibres of elliptic fibration .
corresponding singular fibres of our elliptic fibration of @xmath32 are shown in table 1 . [ cols="<,^,^,^",options="header " , ] * table 3 . * by comparing table 3 and table 2 , we can assure the mirror symmetry conjecture for the reflexive polytopes @xmath39 and @xmath12 .
let us recall the theory of mordell - weil lattices due to t. shioda .
for detail , see @xcite and @xcite .
let @xmath3 be a compact complex surface and @xmath40 be a algebraic curve .
let @xmath41 be an elliptic fibration with sections .
for generic @xmath42 , the fibre @xmath43 is an elliptic curve . in the following
we assume that the elliptic fibre @xmath44 has singular fibres .
@xmath45 denotes the algebraic function field on @xmath40 . if @xmath46 , the field @xmath45 is isomorphic to the rational function field @xmath47 . in this article
, @xmath48 denotes the intersection number and @xmath49 denotes the mordell - weil group of sections of @xmath44 . for all @xmath50 and @xmath42
, we have @xmath51 .
note that the section @xmath5 intersects an irreducible component with multiplicity @xmath52 of every fibre @xmath43 .
let @xmath53 be the zero of the group @xmath49 .
the section @xmath53 is given by the set of the points at infinity on every generic fibre . set @xmath54 for all @xmath55 we have @xmath56 where @xmath57 is the number of irreducible components of @xmath43 , @xmath58 are irreducible components with multiplicity @xmath59 of @xmath43 , and @xmath60 is the component with @xmath61 .
let @xmath62 be a generic fibre of @xmath63 .
set @xmath64 we call @xmath65 the trivial lattice for @xmath63 . for @xmath50 , @xmath66 denotes the corresponding element .
[ shiodatheorem ] ( t. shioda @xcite ( see also @xcite theorem ( 3@xmath6710 ) ) ) ( 1 ) the mordell - weil group @xmath49 is a finitely generated abelian group . + ( 2 )
the nron - severi group @xmath68 is a finitely generated abelian group and torsion free .
+ ( 3 ) we have the isomorphism of groups @xmath69 given by @xmath70 we set @xmath71 for the trivial lattice @xmath65 .
[ torsioncor ] ( @xcite , see also @xcite proposition ( 3@xmath7211 ) ) ( 1 ) we have @xmath73 + ( 2 ) we have @xmath74 where @xmath75 is the torsion part of @xmath76 .
set @xmath77 we have @xmath78 ( see @xcite , see also @xcite section 5 ) .
let @xmath55 . under the notation ( [ singcomp ] ) , we set @xmath79 where @xmath80
. set @xmath81 .
( @xcite , @xcite , see also @xcite section 7 ) let @xmath55 .
the set @xmath82 has a canonical group structure . especially , for the singular fibre @xmath43 of type @xmath83 , @xmath84 for the singular fibre @xmath85 of type @xmath86 , @xmath87 for each @xmath42
, we introduce the map @xmath88 note that @xmath89 ( see @xcite section 7 ) .
we call @xmath90 the specialization map .
( @xcite section 7 ) for all @xmath42 , the specialization map @xmath91 is a homomorphism of groups .
[ sprem ] especially for the singular fibre @xmath43 of type @xmath92 ( @xmath93 , resp . ) , the projection of @xmath90 @xmath94 is a homomorphism of groups . the elliptic fibration given by ( [ p1prekodaira ] ) is described in figure 1 .
+ * figure 1 .
* the trivial lattice for this fibration is @xmath96 let @xmath5 be the section in ( [ p1section ] ) .
@xmath97 at @xmath98 and @xmath99 at @xmath100 .
set @xmath101 this is a subgroup of @xmath102 .
we have @xmath103 . according to theorem [ 18theorem ] and theorem [ shiodatheorem ] ( 3 )
, we obtain @xmath104 and we obtain also @xmath105 for generic @xmath106 .
we have @xmath107=1 \quad{\rm or } \quad[{\rm ns}(s_1(\lambda,\mu)):l_1]=3.\end{aligned}\ ] ] in the following , we prove @xmath108=1.\ ] ] [ t1lemma ] for generic @xmath106 , we have @xmath109 from ( [ nsspan ] ) and ( [ p1 = 3 ] ) it is necessary that @xmath110 or @xmath111=3 $ ] .
we assume @xmath111=3 $ ] .
then , according to corollary [ torsioncor ] ( 2 ) , @xmath112 therefore there exists @xmath113 such that @xmath114 . by remark [ sprem ] and ( [ narrow ] )
, we can assume that @xmath115 at @xmath116 and @xmath117 at @xmath118 .
put @xmath119 .
set @xmath120 . by calculating the intersection matrix
, we have @xmath121 on the other hand , due to ( [ torsionisomorph ] ) , we have @xmath122 .
so it follows @xmath123 this contradicts ( [ rank18hatp1 ] ) . by the above lamma , we have @xmath124 [ l1lemma ] for generic @xmath106 , we have @xmath125 .
it is sufficient to prove @xmath126=1 $ ] .
we assume @xmath126=3 $ ] . by ( [ trivialp1 ] )
there exists @xmath127 such that @xmath128 . according to remark [ sprem ] , @xmath129 and @xmath130 we can assume @xmath131 , for @xmath5 in ( [ p1section ] ) does not intersect @xmath53 . by the addition theorem for elliptic curves , we have @xmath132 and we can check @xmath132 does not intersect @xmath53 .
so , we assume @xmath133 also .
set @xmath134 . by calculating the intersection matrix
, we have @xmath135 on the other hand , we have @xmath136 from theorem [ 18theorem ] .
hence , we obtain @xmath137 this contradicts ( [ rankp1tilde ] ) .
therefore , we have @xmath138=1 $ ] .
[ thm : intersectionp_1 ] the lattice @xmath139 is isomorphic to the lattice given by the intersection matrix @xmath140 and its orthogonal complement is given by the intersection matrix @xmath141 we obtain the corresponding intersection matrix @xmath142 for the lattice @xmath139 : @xmath143 let @xmath144 be the unimodular matrix @xmath145 we have @xmath146 therefore we obtain theorem [ latticethm ] for @xmath95 .
the elliptic fibration given by ( [ p2prekodaira ] ) is described in figure 2 .
+ * figure 2 .
* the trivial lattice for this fibration is @xmath148 let @xmath5 be the section in ( [ p2section ] ) .
note @xmath149 and @xmath150 .
set @xmath151 this is a subgroup of @xmath152 .
we have @xmath153 . as in the case @xmath95 , so we obtain @xmath154 for generic @xmath155 .
we have @xmath156=1 { \rm \enspace or \enspace } [ { \rm ns}(s_2(\lambda,\mu)):l_2]=3.\end{aligned}\ ] ] in the following , we prove @xmath157=1 $ ] .
[ t2lemma ] for generic @xmath155 , we have @xmath158 . by a direct calculation
, we have @xmath159 . from ( [ p2 = 3 ] )
, we have @xmath158 . therefore we obtain @xmath160 for generic @xmath155 , we have @xmath161 .
we assume @xmath157=3 $ ] . from ( [ trivialp2 ] )
there exists @xmath162 such that @xmath128 .
according to remark [ sprem ] , we obtain @xmath163 and @xmath164 .
because the section @xmath5 in ( [ p2section ] ) and the section @xmath132 do not intersect @xmath53 , we have @xmath131 and @xmath133 . set @xmath165 .
calculating its intersection matrix , we have @xmath166 . as in the proof of lemma [ l1lemma ] , this contradicts to theorem [ 18theorem ] .
[ thm : intersectionp_2 ] the lattice @xmath167 is isomorphic to the lattice given by the following intersection matrix @xmath168 and its orthogonal complement is given by the intersection matrix @xmath169 we obtain the corresponding intersection matrix @xmath170 for @xmath167 : @xmath171 let @xmath172 be the unimodular matrix @xmath173 we have @xmath174 therefore , we obtain theorem [ latticethm ] for @xmath147 . the elliptic fibration given by ( [ p3prekodaira ] ) is described in figure 3 .
+ * figure 3 .
* the trivial lattice for this fibration is @xmath176 let @xmath5 be the section in ( [ p3section ] ) .
set @xmath177 this is a subgroup of @xmath178 and we have @xmath179 .
moreover , the section @xmath180 in ( [ p3section ] ) is a @xmath181-torsion section for this elliptic fibretion . due to corollary [ torsioncor ] ,
@xmath182 $ ] is divided by @xmath181 .
hence , we have @xmath183=2 { \rm \enspace or \enspace } [ { \rm ns}(s_3(\lambda,\mu)):l_3']=6.\end{aligned}\ ] ] [ trivialp3 ] for generic @xmath184 , we have @xmath182=2 $ ] .
we have @xmath185 . from ( [ p3 = 6 ] )
, we obtain @xmath182=2 $ ] . for generic @xmath186
, we have @xmath187=2 $ ]
. we shall show that @xmath187=2 $ ] .
we assume @xmath187=6 $ ] . from lemma [ trivialp3 ] ,
there exists @xmath188 such that @xmath128 . according to remark [ sprem ] ,
it is necessary that @xmath189 and @xmath163 .
also we have @xmath131 , for p in ( [ p3section ] ) does not intersect @xmath53 .
moreover we can assume that @xmath133 or @xmath52 , for the section @xmath132 does not intersect @xmath53 at @xmath190 .
set @xmath191 . calculating the intersection matrix
, we have @xmath192 on the other hand , theorem [ 18theorem ] implies that @xmath193 and @xmath194 this is a contradiction to ( [ dettildel3 ] ) . due to the above lemma , we have @xmath195 for generic @xmath184 . to determine the explicit lattice structure for @xmath175 we use another elliptic fibration defined by ( [ p3prekodairaanother ] ) .
this fibration is described in figure 4 .
+ * figure 4 .
* let @xmath196 and @xmath197 be the sections in ( [ p3sectionanother ] ) for this elliptic fibration .
set @xmath198 we have @xmath199 for generic @xmath186 and @xmath200 .
therefore we have @xmath201 for generic @xmath186 .
[ thm : intersectionp_3 ] the lattice @xmath202 is isomorphic to the lattice given by the intersection matrix @xmath203 and its orthogonal complement is given by the intersection matrix @xmath204 we obtain the corresponding intersection matrix @xmath205 for the lattice @xmath202 : @xmath206 let @xmath207 be the unimodular matrix @xmath208 we have @xmath209 therefore , we obtain theorem [ latticethm ] for @xmath175 .
in this section , we determine the system of period differential equations and its projective monodromy group for the family @xmath32
. set @xmath210 the unique holomorphic @xmath181-form on the @xmath0 surface @xmath211 is given by @xmath212 up to a constant factor .
first , we consider a period of @xmath213 .
[ periodthm ] we can find a @xmath181-cycle @xmath214 @xmath215 so that we have the following power series expansion of the period @xmath216 which is valid in a sufficiently small neighborhood of @xmath217 .
\(1 ) ( a period for @xmath95 ) we have a period of @xmath218 : @xmath219 \(2 ) ( a period for @xmath147 ) we have a period of @xmath220 : @xmath221 \(3 ) ( a period for @xmath175 ) we have a period of @xmath222 : @xmath223 let @xmath224 by the same argument in the proof of theorem 3.1 of the article @xcite , we can choose a certain @xmath181-cycle @xmath214 on @xmath225 so that the period integral @xmath226 is given by a power series of @xmath227 . in the case
@xmath228 , our period is reduced to the appell @xmath229(see @xcite ): @xmath230 where @xmath62 is the gauss hypergeometric function and @xmath231 .
secondary , we apply the theory of the gkz hypergeometric functions to obtain the system of differential equations whose solution is the period integral in theorem 3.1 . in the following ,
set @xmath232 \(1 ) ( the gkz system of equations for @xmath95 ) set @xmath233 it holds @xmath234 \(2 ) ( the gkz system of equations for @xmath147 ) set @xmath235 it holds @xmath236 \(3 ) ( the gkz system of equations for @xmath175 ) set @xmath237 it holds @xmath238 set @xmath239 and @xmath240 let @xmath241 . from the matrix @xmath242 and the vector @xmath243 , we have the system of the gkz system of equations concerned with the period @xmath244 in theorem [ periodthm ] . for detail , see the proof of proposition 3.1 in @xcite . each system in the above proposition has the @xmath245-dimensional space of solutions . on the other hand , theorem [ 18theorem ]
says that the rank of transcendental lattice for @xmath246 is @xmath247 .
it implies that there are the system of period differential equations for the family @xmath248 with the @xmath247-dimensional space of solutions .
[ periodde ] ( 1 ) ( the period differential equation for @xmath95 ) set @xmath249 it holds @xmath250 the space of solutions of the system @xmath251 is 4-dimensional .
\(2 ) ( the period differential equation for @xmath147 ) set @xmath252 it holds @xmath253 the space of solutions of the system @xmath254 is 4-dimensional .
\(3 ) ( the period differential equation for @xmath175 ) set @xmath255 it holds @xmath256 the space of solutions of the system @xmath257 is 4-dimensional .
we determine these systems by the method of indeterminate coefficients .
for detail , see the proof of theorem 3.2 in @xcite . in the following we prove that those spaces of solutions is 4-dimensional .
\(1 ) set @xmath258 .
we obtain the corresponding pfaffian system @xmath259 with @xmath260 by the following way . setting @xmath261
we have @xmath262 with @xmath263 and @xmath264 with @xmath265 we have @xmath266 therefore the system @xmath267 has the @xmath247-dimensional space of solutions .
\(2 ) set @xmath258 .
we obtain the corresponding pfaffian system @xmath268 with @xmath269 as the following way . setting @xmath270
we have @xmath271 with @xmath272 and @xmath273 with @xmath274 we see @xmath275 therefore the system @xmath276 has the @xmath247-dimensional solution space .
\(3 ) set @xmath258 .
we obtain the corresponding pfaffian system @xmath277 with @xmath278 as the following way . setting @xmath279 we have @xmath280 with @xmath281 and @xmath282 with @xmath283 we have @xmath284 .
so the system @xmath285 has the @xmath247-dimensional space of solutions . from the puffian systems in the above proof , we obtain the singular locus of the system ( [ eq : systemp1 ] ) : @xmath286 the singular locus of the system ( [ eq : systemp2 ] ) : @xmath287 and the singular locus of the system ( [ eq : systemp3 ] ) : @xmath288 omitting these locus from @xmath289 we have the domain @xmath290 in ( [ lambda_1 ] ) , ( [ lambda_2 ] ) and ( [ lambda_3 ] ) . finally , we determine the projective monodromy groups .
let @xmath241 .
for generic @xmath291 , we can take a basis @xmath292 of @xmath293 such that the intersection matrix @xmath294 is equal to the matrix in theorem [ latticethm ] .
this basis is extended to a basis @xmath295 of @xmath296 .
let @xmath297 be its dual basis ( namely @xmath298 ) . by theorem [ latticethm ]
, we have @xmath299 . using this basis @xmath300
, we define the local period map as in the beginning of section 2 . moreover , we define the multivalued period map @xmath301 by the analytic continuation of the local period map along any arc in @xmath302
. set @xmath303 by the riemann - hodge relation , we have @xmath304 .
the fundamental group @xmath305 acts on @xmath306 by the analytic continuation of the local period map .
this action induces a group of projective linear transformations which is a subgroup of @xmath307 .
we call it the projective monodromy group of the multivalued period map @xmath308 . note that @xmath309 is composed of two connected components : @xmath310
. set @xmath311 it acts on @xmath309 by @xmath312 @xmath313 .
let @xmath314 be the subgroup of @xmath315 given by @xmath316 .
[ monodromytheorem ] let @xmath224 the projective monodromy group of the period differential equation for the family @xmath246 is equal to @xmath314 . because the projective monodromy group @xmath317 of the multivalued period map @xmath318 is equal to that of the period differential equation for @xmath246 , we determine @xmath317 .
it is obvious @xmath319 . however , we need a delicate observation to prove the converse @xmath320 . for precise argument ,
see section 4 in @xcite .
the author would like to express his gratitude to professor hironori shiga for various advices and suggestions .
he is also thankful to professor kimio ueno for encouragements and useful comments .
atsuhira nagano , + c.o .
kimio ueno , department of mathematics , waseda university , + okubo 3 - 4 - 1 , shinjuku - ku , tokyo 169 - 8555 , japan + tel : 03 - 5286 - 3015 + fax : 03 - 5286 - 3483 + e - mail : [email protected] | in this article we study the families of @xmath0 surfaces derived from @xmath1 dimensional @xmath2 verticed reflexive polytopes with at most terminal singularity .
we determine the lattice structures , the period differential equations and the projective monodromy groups for these families . |
the lattice boltzmann model for flat space reported in ref .
@xcite needs to be extended in order to simulate relativistic hydrodynamics in curved spaces .
this numerical model solves the following conservation equations : [ eqn : conservationflat ] @xmath102 @xmath103 where @xmath104 is the energy momentum tensor for flat space , @xmath105 in curved spaces , the energy momentum tensor has to be conserved as well
. however , conservation of quantities in a curved manifold needs to be expressed through covariant derivatives , instead of partial derivatives .
the conservation equations for the number of particles and energy momentum tensor are given by [ eqn : conservationcurved ] @xmath106 @xmath107 where the covariant derivatives @xmath31 of the 3-particle flow and energy - momentum tensor relate to the partial derivatives @xmath108 as follows : [ eqn : covariantderivtrafos ] @xmath109 @xmath110 where @xmath34 denote the christoffel symbols which are computed with the metric tensor using @xmath111 additionally , the energy momentum tensor in curved manifolds reads @xmath112 where @xmath113 is the metric tensor and @xmath23 the shear - stress tensor , which can be approximated by the equation @xmath114 . using the property that @xmath115 , we can rewrite the covariant derivative of the energy - momentum tensor as @xmath116 thus the conservation equations expressed with the energy - momentum tensor for the flat space @xmath117 reads @xmath118 where we have used the fact that partial derivatives equal covariant derivatives when acting on scalars . at this stage
we also include a temperature independent coupling constant @xmath36 which accounts for the strength of the inertial corrections .
thus , this set of equations can be written as [ eqn : conseqcnimplemented ] @xmath37 @xmath38 with [ eqn : forcingterms ] [ eqn : forcesfrommetric ] @xmath39 @xmath40 here @xmath119 and @xmath120 can be introduced in the numerical model as external forces using the technique described in ref . @xcite .
in order to obtain general results we transform the hydrodynamic equations to dimensionless form and determine the dimensionless quantities that characterise our systems . we also include a force density @xmath121 describing an external electric field @xmath122 to the energy - momentum conservation equation .
to write the equations in a dimensionless form , we first express all relevant quantities in dimensionless form : @xmath123 , @xmath124 @xmath125 , @xmath126 and @xmath127 where all primed variables are dimensionless .
furthermore , the relation @xmath128 holds , which allows us to write the temporal derivative @xmath129 , and thus for @xmath130 , the relation @xmath131 is also satisfied .
we divide the energy - momentum tensor in an equilibrium part and a dissipative part @xmath132 .
the equilibrium part can be written in its dimensionless form @xmath133 .
assuming that the shear - stress tensor is linearly dependent on the shear viscosity @xmath25 @xcite and taking into account that the shear stress tensor has the same units as the energy - momentum tensor ( @xmath134 ) , we get @xmath135 .
the force densities transform as @xmath136 and @xmath137 , respectively .
this can be derived from the fact that , in our case , the metric is dimensionless and thus the christoffel symbols ( as derivative of the metric ) transform as @xmath138 . with the considerations above ,
equations read [ eqn : conseqdimensionlessderiv ] @xmath139 @xmath140 while the first equation remains unchanged , we observe after multiplying by @xmath141 that the second equation only depends on two dimensionless numbers @xmath142 and @xmath143
as a function of temperature for different @xmath144 .
inset : temperature averaged standard deviation @xmath145 for different distances @xmath11 . ] all simulations are performed with the same number of carbon atoms ( @xmath146 ) . the atoms at the left and right boundary are fixed at a distance @xmath11 from each other and are not allowed to move nor change their effective bond lengths .
the simulations are preformed for different @xmath147 which correspond to initial effective bond lengths @xmath148 . the distance @xmath11 is related to the tension imposed on the graphene membrane .
one expects that increasing @xmath11 results in a more stretched membrane and the height fluctuations are reduced .
figure [ fig : heightstd ] confirms that the height fluctuations @xmath149 depend on @xmath11 and therefore also on the strain imposed on the graphene membrane .
interestingly , the height fluctuations are independent on the temperature . in contrast
, the width of the effective bond length distribution presents a strong temperature dependence as shown in figs . [
fig : distrbl ] a - c .
this finding is related to the ricci scalar analysis ( see fig .
[ fig : timeaveragedricciscalarvstmp ] ) , where it has been shown that the absolute value of the ricci scalar has a much stronger temperature dependence than a dependence on @xmath11 . a higher temperature results in a broader width of the effective bond length distribution .
this leads to a higher average ricci scalar and consequently to more curvature in the system .
the curvature introduces shear which is responsible for saturation of the current . | we introduce a different perspective describing electron - phonon interactions in graphene based on curved space hydrodynamics .
interactions of phonons with charge carriers increase the electrical resistivity of the material .
our approach captures the lattice vibrations as curvature changes in the space through which electrons move following hydrodynamic equations . in this picture ,
inertial corrections to the electronic flow arise naturally effectively producing electron - phonon interactions .
the strength of the interaction is controlled by a coupling constant , which is temperature independent .
we apply this model to graphene and recover satisfactorily the linear scaling law for the resistivity that is expected at high temperatures .
our findings open up a new perspective of treating electron - phonon interactions in graphene , and also in other materials where electrons can be described by the fermi liquid theory . at finite temperatures ,
phonons interact with charge carriers , and therefore , contribute to the electrical resistivity of the respective material @xcite . in suspended graphene , the temperature dependence of the resistivity follows a linear increase due to electron - phonon interactions above 100 k. graphene membrane at 300 k after the electronic flow has reached a steady state .
streamlines indicate the velocity field of the electronic flow that moves from left to right .
the colours represent the height . ]
k and @xmath0 k for planar and out - of - plane phonons , respectively @xcite . however , for low electron densities , the fermi energy can be substantially smaller than the debye energy , and only phonons with energy smaller than two times the fermi energy , corresponding to a full backscattering of electrons , can scatter with the electrons .
this defines the bloch - gruneisen temperature @xmath1 @xcite , which for doped suspended graphene is around @xmath2 k @xcite .
the linear relation also holds when external strain is applied to the graphene samples @xcite .
] recently it has been shown that electronic flow in graphene can be modelled using relativistic hydrodynamic equations @xcite .
usually , in this hydrodynamic formalism , electron - phonon interactions are included into the equation for the electrical current density @xmath3 as a damping term , @xmath4 , with a characteristic relaxation time @xmath5 @xcite .
however , a proper inclusion of lattice - fluid interactions are missing . here
, we present a novel approach where we account for the electron - phonon interactions by including inertial corrections due to the deformations of the graphene sheet .
we address the question whether these inertial corrections can recover the linear temperature relation of the electrical resistivity in the high temperature regime . for this purpose ,
we use molecular dynamics simulations to simulate different graphene membranes at different temperatures and imposed strains . from the position of the atoms , we build the coordinate system
in which the electrons flow , and extract the inertial corrections to include them into the two - dimensional hydrodynamic equations . in fig.[fig : sheetwithstreamlines ] , we can observe how the presence of thermal fluctuations produces curved streamlines of the electron velocity field . to simulate the graphene sheets with molecular dynamics we use the adaptive intermolecular reactive bond - order ( airebo ) potential @xcite .
this many - body potential has been developed to simulate molecules made of carbon and hydrogen atoms , and has been very successful to reproduce the phonon dispersion curves of graphene @xcite . in our simulations
we consider suspended graphene sheets of @xmath6 , ( see fig . [
fig : lblattice ] ) , but also two stretched cases with @xmath7 and @xmath8 . as shown in fig .
[ fig : lblattice ] the zigzag edge is located along the @xmath9-direction ( top and bottom ) and accordingly the armchair edge along the @xmath10-direction ( left and right ) .
the atoms can freely move in all directions , except the ones at the left and right boundary , representing the electrical contacts ( grey in fig .
[ fig : lblattice ] ) .
the distance between contacts will be denoted by @xmath11 .
note that since the left and right boundaries are fixed , we are effectively applying strain to the graphene samples .
we simulate several samples at different temperatures within the range of @xmath2-@xmath12 k controlling the temperature of the system with a nos - hoover thermostat .
all molecular dynamics simulations run ( during the thermalisation procedure ) with a time step @xmath13 fs , which is small enough to capture the dynamics of the carbon - carbon interactions accurately .
randomised velocities are attributed to the atoms at the beginning of each simulation . after reaching thermal equilibrium we are ready to couple the graphene sheet to the fluid solver and introduce electrical currents .
discrete mapping ( dashed arrow ) that describes the embedding of the two - dimensional triangular lattice ( bottom figure ) in the three - dimensional cartesian space ( top figure ) .
the circles represent the atoms .
the electrical contacts are shown in grey at both ends of the graphene sheet . ] the electronic flow in flat graphene can be modelled using relativistic hydrodynamic equations , which are responsible for the conservation of particles , @xmath14 , and energy and momentum , @xmath15 , where @xmath16 and @xmath17 are the @xmath18-particle flow and the energy - momentum tensor , respectively . here , @xmath19 is the number of particle density and @xmath20 the @xmath18-velocity of the electronic flow .
the energy momentum tensor reads @xmath21 , where @xmath22 are the components of the minkowski metric and @xmath23 is the shear - stress tensor , which can be approximated by the equation @xmath24 , with @xmath25 being the shear viscosity . @xmath26 and @xmath27 are energy density and pressure , respectively .
the equation of state completes this set of equations , which for the case of graphene is given by @xmath28 . here
, the fermi velocity is denoted by @xmath29 . to take into account the curvature of suspended graphene sheets , we consider inertial corrections in these equations . for this purpose ,
we first use the covariant formulation of the hydrodynamic equations in curved space to determine the terms responsible for the inertial corrections , and afterwards , we introduce them as a forcing term into the respective equations for flat graphene . the conservation equations in curved manifolds are derived by replacing the partial derivative by the covariant derivative and extending the energy - momentum tensor to curved space , i.e. @xmath30 .
the covariant derivative @xmath31 in the curvilinear coordinate system can be expressed through the respective partial derivative by @xmath32 , and @xmath33 , with @xmath34 being the christoffel symbols computed as follows : @xmath35 at this stage
we include a temperature independent coupling constant @xmath36 in front of the inertial corrections which accounts for the strength of the electron - phonon interaction .
thus , the set of equations can be written as [ eqn : conseqcnimplemented ] @xmath37 @xmath38 with [ eqn : forcingterms ] [ eqn : forcesfrommetric ] @xmath39 @xmath40 to solve the hydrodynamic equations , we use the lattice boltzmann solver described in ref . @xcite , and add the inertial contributions as external forces ( see supplementary information ) .
the model discretises space as regular triangular lattice , which we couple to graphene s hexagonal lattice . as shown in fig .
[ fig : lblattice ] , the atoms form a two - dimensional manifold which can be described by a discrete mapping @xmath41 from the curved space to the three - dimensional flat space ( reference frame of the laboratory , where the metric is given by the minkowski - metric ) . the metric tensor can be computed by @xmath42 similarly , the christoffel symbols are obtained from eq . .
further details on this calculations can be found in the supplementary information .
the graphene sheet possesses zigzag boundaries at the left and right end . for the fluid solver we impose periodic boundary conditions at these boundaries , which correspond to the in- and outlet .
the free sides of the graphene sheet possesses armchair geometry and , therefore , we impose free slip boundary conditions . in order to obtain general results and
allow for more insight , we make the hydrodynamic equations dimensionless ( see supplementary information for the detailed analysis ) . considering that the shear - stress tensor depends linearly on the shear viscosity @xcite we find that the physics of the electronic flow only depends on two dimensionless numbers : @xmath43 , and @xmath44 , where @xmath45 , @xmath46 , @xmath47 , @xmath48 , @xmath49 , and @xmath50 are characteristic values for the energy density , velocity , length , shear viscosity , electric charge , and electric field , respectively .
note that @xmath51 is related to the reynolds number @xcite .
the parameter range of our simulations is @xmath52 and @xmath53 . to produce an electrical current
, we apply an external electric field in @xmath9 direction , @xmath54 v / m . in our simulations
, we set @xmath55 /@xmath56 , chemical potential @xmath57 kgm@xmath58/s@xmath59 ev , and @xmath60 kg / s@xmath58 .
we use the shear viscosity for doped graphene given by @xcite @xmath61 with @xmath62 .
we couple the electronic fluid to the atomistic simulation using the same length - scale @xmath63 and time scale @xmath64 fs . after each iteration
we compute the metric tensor and the christoffel symbols and simulate the electronic flow until we obtain the electrical current in steady state .
the momentum change in the fluid is imposed on the atoms to ensure momentum conservation of the system .
temperature dependence of the electrical resistivity of graphene due to inertial corrections for different coupling constants @xmath36 .
the dashed lines correspond to the best fit for the function @xmath65 for @xmath66 with @xmath67/k,@xmath68/k and @xmath69/k .
the colors represent graphene membranes with different strains where red circles stand for a graphene sheet with size @xmath70 , green triangles for @xmath71 and blue squares for @xmath72 .
inset : log - log plot of @xmath73 dependence on @xmath36 . in both figures ,
the errorbars are smaller than the symbols . ]
we perform simulations for different strains and different values of the coupling constant and measure the resistivity of the graphene sheets using ohm s law , @xmath74 , where the electrical current is given by @xmath75 in fig .
[ fig : resistivitylogdifferentalpha ] , we observe that all our calculations of the electrical resistivity of graphene exhibit a linear dependence with temperature , @xmath76 .
this linear dependence is in agreement with experimental measurements and theoretical predictions @xcite .
additionally , the slope @xmath77 only depends on the strength of the coupling constant @xmath36 but not on the applied strain , which is also in agreement with previous works @xcite
. we can also observe from the inset of fig .
[ fig : resistivitylogdifferentalpha ] that @xmath78 .
the factor @xmath73 can be compared with the values from the boltzmann transport theory @xcite , where @xmath79 with @xmath80 kg / m@xmath58 being the mass density of graphene , @xmath81 the phonon velocity and @xmath82 the deformation - potential coupling constant .
experimental and theoretical results suggest that the deformation - potential varies within the range of @xmath83 ev @xcite , and phonon velocities within @xmath84 m / s .
thus , one expects @xmath85 to be in the range @xmath86/k . from our results
we deduce that @xmath87 to obtain results compatible with @xmath85 .
the residual resistivity @xmath88 is @xmath89 for @xmath90 and is a consequence of the static corrugations ( ripples ) present at any temperature .
these ripples stabilize the two - dimensional crystal circumventing the mermin wagner theorem , which states that crystalline order can not exist in two dimensions @xcite .
maximal absolute value of the ricci scalar @xmath91 for graphene membranes with distances @xmath11 .
main panel : time averaged ricci scalar @xmath92 for different temperatures and distances @xmath11 .
red circles correspond to @xmath93 , with an exponent @xmath94 .
green triangles correspond to @xmath95 , with an exponent @xmath96 .
blue squares correspond to @xmath97 , with an exponent @xmath98 .
inset panel : time evolution of @xmath91 for graphene membranes at different temperatures for @xmath97 . ]
the existence of a resistivity points to the presence of dissipation .
recently , it has been shown that energy dissipation in flows through curved manifolds arises from the curvature @xcite .
the ricci scalar ( or curvature scalar ) is a measure for the curvature and is given by @xmath99 .
we have analysed the ricci scalar for different strains and temperatures and found that its maximal absolute value remains constant during time evolution and depends only on the temperature of the system ( see inset of fig .
[ fig : timeaveragedricciscalarvstmp ] ) . as function of temperature
the ricci scalar follows a power - law with an exponent that increases with the applied strain ( see main panel in fig . [
fig : timeaveragedricciscalarvstmp ] ) .
we have also studied the standard deviation of the heights of our graphene sheets , and found that they are virtually not influenced by temperature , within the range of @xmath100 k ( see fig . [
fig : heightstd ] of supplementary information ) .
therefore , we conclude that , an increase of temperature mainly induces in - plane atomistic motion , increasing the average effective bond length ( see figs .
[ fig : distrbl ] a - c in the supplementary information ) , and consequently , increasing the local curvature .
since curvature induces energy dissipation @xcite , this explains why larger temperatures result in larger electrical resistivities .
to summarise , we have shown that inertial corrections lead to a linear dependence of the electrical resistivity with temperature in suspended graphene , which is in agreement with experimental measurements of the electrical resistivity due to electron - phonon interactions at high temperatures . to characterize the strength of the electron - phonon interaction in our model , we have introduced a coupling constant @xmath36 , and for values of
@xmath101 one recovers the same values of @xmath85 known from theoretical predictions and experimental measurements .
additionally , the resistivity does not depend on the strain imposed to the sample , which is also in agreement with previous work @xcite .
we also analysed the maximal absolute value of the curvature as function of temperature and found a power - law behaviour with an exponent that changes slightly with the applied strain .
finally , by studying the average height fluctuations of the graphene sheets , we also discovered that by increasing the temperature , the height fluctuations remain almost unchanged , and consequently , including larger local values of curvature .
in our model , larger curvature implies more energy dissipation , and consequently , larger electrical resistivity .
our finding opens up many interesting questions , as for instance , if one can discern the influence of different types of vibrational phonons , e.g. flexural , acoustic , and optical , by studying how they introduce curvature into the hydrodynamic system . by extending our molecular dynamics simulations , one could also explore the electrical resistivity due to the electron - phonon interactions when the graphene sample is placed on a substrate @xcite . finally , our approach can also be applied to other two- and three - dimensional crystals , where electrons are well described as fermi liquid . in those structures , ion displacements due to phonons
can induce intrinsic curvature , and consequently , energy dissipation in the electronic flow . applying our model to other materials
will be an interesting subject for future research .
we thank the european research council ( erc ) advanced grant no .
319968-flowccs for financial support .
25ifxundefined [ 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 _
_
( , ) link:\doibase 10.1103/physrevb.79.125416 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevlett.105.256805 [ * * , ( ) ] link:\doibase 10.1103/physrevb.77.115449 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.105.266601 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.103.025301 [ * * , ( ) ] link:\doibase 10.1103/physrevb.93.125410 [ * * , ( ) ] @noop ( ) @noop * * ( ) link:\doibase 10.1103/physrevlett.106.156601 [ * * , ( ) ] link:\doibase 10.1103/physrevb.91.085401 [ * * , ( ) ] link:\doibase 10.1103/physrevb.80.085109 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * ( ) link:\doibase 10.1103/physreve.88.013302 [ * * , ( ) ] @noop _ _ ( , ) @noop _ _ ( , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrev.176.250 [ * * , ( ) ] @noop ( ) link:\doibase 10.1103/physreve.93.043316 [ * * , ( ) ] |
here , we show that the definitions ( [ eq : corrs ] ) and ( [ eq : svetq ] ) of biseparable correlations are equivalent . for the sake of generality , we prove this equivalence in the @xmath0-partite scenario .
let @xmath109 denote a subset of @xmath110 containing from one to @xmath111 elements , and let @xmath112 denote the complementary subset .
the pair @xmath113 thus represent a partition of the @xmath0 parties into two non - empty groups .
let @xmath114 be an ( unormalized ) @xmath0-partite state that is product across the partition @xmath113 .
an @xmath0-partite state @xmath115 is then biseparable if it can be written as a mixture @xmath116 a state that can not be written in the above form is genuinely @xmath0-partite entangled .
let @xmath117 be an @xmath0-partite probability distribution characterizing a bell experiment with measurement settings @xmath118 and measurement outcomes @xmath119 .
we say that the correlations described by @xmath120 are biseparable quantum correlations iff @xmath121\ ] ] for some biseparable state @xmath115 and measurement operators @xmath122 .
equivalently , as we will see , biseparable quantum correlations can be defined as those that can be written in the form [ aeq : svetq ] p(|a||x ) = _ t_k_t
p^k_t_q(|a_t||x_t)p_q^k_t(|a_t||x_t ) , where we write @xmath123 and @xmath124 for the outcome and measurement settings of the parties belonging to the subset @xmath109 and where @xmath125 $ ] are arbitrary ( unnormalized ) quantum correlations for the parties in @xmath109 . that any correlations of the form ( [ acorr ] ) is of the form ( [ aeq : svetq ] )
is immediate using the definition ( [ abs ] ) of biseparable states .
conversely , ( [ aeq : svetq ] ) can be rewritten in the form [ aeq : svetq2 ] p(|a||x ) = _ t_k_t .
this gives a representation that is almost of the form ( [ acorr ] ) , except that a different set of measurement operators corresponds to each term in the decomposition ( [ abs ] ) of the biseparable state @xmath126 .
this can be fixed by introducing local ancillas on each system acting as labels that indicate which term in the decomposition is considered : defining @xmath127 , @xmath128 we finally get @xmath129 $ ] , and thus any correlations of the form can be written as in eq . .
let @xmath130 denote the set of quantum correlations defined for the parties belonging to @xmath109 and let @xmath131 denote the set of biseparable quantum correlations , which is a subset of the @xmath0-partite quantum correlations . from ( [ aeq : svetq ] ) , it is clear that @xmath131 is convex and that its extreme points are of the form @xmath132 where @xmath133 is an extremal point of the set @xmath130 .
here we derive the maximum value that the expression @xmath75 , see eq . , can take for measurements made on a biseparable quantum state ( the biseparable bound ) , and for svetlichny models ( the svetlichny bound ) . from ( [ aeq : svetq ] ) , it follows that the biseparable bound is obtained by maximizing @xmath75 over all correlations of the form [ beq : svetq ] p(|a||x ) = _ t_k_t
p^k_t_q(|a_t||x_t)p_q^k_t(|a_t||x_t ) . in a svetlichny model ,
on the other hand , one considers correlations of the form [ beq : svets ] p(|a||x ) = _ t_k_t
p^k_t(|a_t||x_t)p^k_t(|a_t||x_t ) , i.e. no constraints ( apart from positivity and normalization ) are imposed on the joint terms @xmath134 . but more refined models la svetlichny can be introduced , see e.g. @xcite , the more constraining one being the one where the joint terms @xmath134 are assumed to be no - signalling .
we will compute the bound on @xmath75 assuming these no - signalling constraints and see that even in this case there exists a gap between the quantum biseparable and svetlichny bounds ( note also that the no - signalling bound that we will derive actually coincides with the more general unconstrained svetlichny bound because @xmath75 is an inequality involving only full @xmath0-partite correlators ) .
we will now prove the biseparable bound given in eq .
for @xmath135 using induction on the number of parties @xmath0 .
let us start with the first step of the induction , which consists of showing that the inequality holds for @xmath23 .
by linearity of @xmath59 in the probabilities , convexity of the decomposition for tripartite biseparable quantum correlations and the fact that @xmath59 is invariant under any permutation of the parties , it is sufficient to prove the bound for correlations of the form @xmath136 .
moreover , it is sufficient to consider the case where @xmath137 is extremal , i.e. , one where every @xmath138 is determined unambiguously as a function of @xmath139 . to this end ,
let us label the eight distinct deterministic assignments of outcome @xmath138 for given input @xmath139 by @xmath140 , where the deterministic ( quantum ) probability distributions are such that @xmath141 if @xmath142 and @xmath143 otherwise .
substituting these eight possible strategies into @xmath59 [ see ] gives eight different bipartite bell expressions for the system @xmath144 : @xmath145 where @xmath146 .
determining the biseparable bound of @xmath59 therefore amounts to finding the maximum of the above bell expressions over arbitrary bipartite quantum correlations @xmath147 $ ] , i.e. , finding the corresponding tsirelson bounds .
it can be easily verified that the eight expressions @xmath148 are either vanishing or equivalent , under relabelling of inputs and outputs , to two times the 3-input chained bell inequality @xcite .
the biseparable bound on @xmath59 is thus equal to the two times the tsirelson bound of the 3-input chained bell inequality @xcite , i.e. , @xmath149 , which is in agreement with eq . .
next , we need to show that whenever inequality holds for @xmath0 parties , it must also hold for @xmath150 parties . by linearity of @xmath75 in the probabilities and convexity of the decomposition ( [ beq : svetq ] ) ,
it is sufficient to compute the bound over all product correlations of the form @xmath151 .
moreover , since @xmath152 is symmetric under any permutation of parties , it is sufficient to consider the biseparations @xmath153 , @xmath154 where @xmath155 . with obvious notation , we write the corresponding correlations as @xmath156 .
bayes rule and the fact that quantum correlations satisfy the no - signaling principle allow us to write the quantum correlations for parties @xmath157 to @xmath158 as @xmath159 . with straightforward algebra
, it can be seen that the quantity @xmath160 for this type of distribution can be rewritten as @xmath161 where the @xmath0-partite quantity @xmath162 depends on the values of @xmath163 and @xmath164 since it is evaluated on the distribution @xmath165 . inserting this expression in the definition @xmath166
, we obtain @xmath167 where we have defined @xmath168 ( note that @xmath169 is different from zero only if @xmath170 ) and write @xmath171 to remind that @xmath172 is evaluated on the distribution @xmath165 which depends on @xmath163 and @xmath164 .
using the fact that @xmath173 and that the probabilities @xmath174 are bounded between @xmath175 and @xmath176 , it follows from ( [ b5 ] ) that @xmath177 let @xmath178 denote the maximal value of the quantity @xmath179 taken by any biseparable correlations .
since the distribution @xmath180 is biseparable , we clearly have that @xmath181 , independently of the values of @xmath163 and @xmath164 .
we thus find @xmath182 an important point to note now is that @xmath179 for all @xmath183 are equivalent to @xmath75 , i.e. , one can obtain the expression for @xmath75 by starting from any of @xmath179 and applying a different labeling for the inputs and/or outputs . to see that this is the case , we first note from the definition of @xmath71 that @xmath184 . in other words , we can obtain @xmath185 by starting from @xmath75 and applying the mapping @xmath186 to each of the @xmath187 .
clearly , the same argument also demonstrates the equivalence between @xmath188 and @xmath189 .
next , note that if the following mappings are applied to the definition of @xmath187 , namely @xmath190 and @xmath191 except for @xmath192 in which case @xmath193 , then we can also obtain the expression for @xmath172 from @xmath194 , thus showing their equivalence .
we thus have that @xmath195 and thus @xmath196 now , by the induction hypothesis we know that @xmath197 for all biseparable correlations .
it then follows from eq .
( [ b8 ] ) that @xmath198 , which completes the proof .
note that , following the above reasoning , any witness for @xmath0 parties and @xmath24 inputs per party which can be written in the form @xmath199 , where @xmath200 is a function satisfying @xmath201 , can be generalized to more parties .
note also that the above biseparable bound for the witness is tight , as it can always be achieved by performing the following local measurements @xmath202 with @xmath203 on the biseparable state @xmath204 finally , let us note that the same procedure as the one detailed above can be followed to obtain the ( no - signalling ) svetlichny bound of @xmath152 .
the only difference is that in the first step of the induction proof , we must compute the no - signalling bound of the @xmath59 inequality instead of the quantum biseparable bound . as in the quantum biseparable case
, this reduces to computing twice the ( bipartite ) no - signalling bound of the 3-input chained bell inequality , which gives @xmath205 and thus @xmath206 . to show that this bound is tight consider a strategy where the @xmath0-th party is separated from the rest and always outputs 1 . using an analysis similar to the one leading to eq .
, we can then see that the remaining @xmath207 parties are playing an effective game defined by @xmath208 which can be shown to be equivalent to twice the expression @xmath209 .
it is a simple exercise to show that the algebraic maximum of @xmath209 is @xmath210 and that this is always achievable by @xmath207 players that are constrained only by the no - signalling principle .
therefore , with this particular strategy , one achieves @xmath211 , which saturates the bound derived above .
here we present a biseparable model that simulates von neumann measurements on noisy tripartite ghz states [ eq : state ] = v2+(1-v)8 of visibility @xmath212 .
clearly this allows to simulate states with @xmath213 as well , by mixing this model with another one in which all parties produce uniformly random outcomes .
the model is presented as a protocol in which a source distributes quantum states and random variables to the parties .
all parties can share the random variables , but since only biseparable states may be used in the model , not more than two parties at a time can share a quantum state . after distribution ,
the parties receive their respective measurement directions @xmath214 , @xmath215 or @xmath216 belonging to the bloch sphere , and measure their quantum system and/or process the information they received accordingly in order to produce binary outcomes @xmath217 .
we check that the outcomes produced by the model are identical to the ones found when measuring state in the given bases .
before the parties receive their inputs , a common source chooses a party @xmath218 at random , say charlie ( @xmath219 ) , and sends him the vector = ( , , ) , uniformly chosen on the sphere @xmath220 .
the source then also provides the two other parties , alice and bob in this case , with the quantum state [ eq : phiab ] |_ab=2|00 + 2 e^-i|11 .
moreover , the source sends to all parties the signs @xmath221 , which are independently and identically distributed with @xmath222 . at the time of measurement , alice and
bob measure their system according to @xmath214 and @xmath215 and get outcomes @xmath223 , while charlie calculates @xmath224 .
the parties then output respectively @xmath225 , @xmath226 and @xmath227 .
here we compare the correlations that are created when alice , bob and charlie apply the biseparable model above to the ones that they would get by measuring state .
since the model treats equally each party , its correlations are symmetric under exchange of the parties and we only need to check the following three relations : @xmath228 here we parametrized the parties measurements in terms of spherical coordinates : x=(_x_x,_x_x,_x ) and similarly for @xmath215 and @xmath216 .
for definiteness , in the following we use the brackets @xmath229 to express the expectation value of a quantity @xmath230 with respect to a quantum state , and the bar = 14_0 ^ 2 d_0^df ( , ) for the average of a function @xmath231 over the random variable @xmath232 and @xmath233 .
we always start by considering that charlie is the special party chosen by the source , to whom the hidden vector @xmath234 is sent , and perform the symetrization afterwards .
symmetrized quantities @xmath235 are indexed by the symbol @xmath236 . for simplicity we average over the depolarization signs @xmath237 only at the end of the calculation .
[ [ single - party - expectation - value ] ] single - party expectation value + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + one can easily verify that the outcomes of each party is locally random according to the model , in agreement with equation .
[ [ two - party - correlators ] ] two - party correlators + + + + + + + + + + + + + + + + + + + + + let us first calculate correlations @xmath238 , @xmath239 and @xmath240 for the case in which charlie receives the vector @xmath234 , and average over the choice of the party @xmath218 being alone afterwards . the first term is easily found from equation , which gives [ eq : ab ] ab = _ x _ y + _
x_y(+_x+_y ) and thus @xmath241 . to compute the correlation @xmath242
, it is useful to write alice s state as _ a=_b(|_ab-2.8pt_ab|)=^22|0 - 2.8pt0|+^22|1 - 2.8pt1| .
the expectation value for alice s outcome is then @xmath243 .
concerning charlie , his outcome @xmath138 is totally determined by @xmath244 and @xmath216 . for simplicity
we assume that @xmath245 , but the other situation can be treated similarly . in this case , for @xmath246 , charles always has @xmath247 , and for @xmath248 , he always has @xmath249 .
now for @xmath250 $ ] , we can write @xmath251 $ ] the interval of @xmath233 for which the product @xmath252 , and @xmath253\backslash \mathcal{i}_+$ ] its complement , with @xmath254 . we thus have the three following cases : 1 . @xmath255 . in this case
@xmath247 , which gives @xmath256 .
2 . @xmath257 . in this case
@xmath249 , which gives @xmath258 .
3 . @xmath259 . in this case one
has @xmath260 where @xmath261 is the indicator function . in total
, after integration over @xmath232 and @xmath233 , this gives @xmath262 .
similarly one can check that @xmath263 , and so once averaged over the choice of the party being alone , the bipartite correlations are given by @xmath264 finally , we need to apply the signs @xmath265 , @xmath266 , @xmath267 .
since the expectation value of the product of two signs is @xmath268 , it follows that overall the correlation between two outcomes produced by the model is ab = 12_x_y , in agreement with equation .
[ [ three - party - correlators ] ] three - party correlators + + + + + + + + + + + + + + + + + + + + + + + we now proceed to calculate the tripartite correlations that are created by the model .
let us consider the three preceding cases separately again : 1 .
@xmath255 . in this case @xmath247 , so @xmath269 as given by equation .
2 . @xmath257 . in this case @xmath249
, so @xmath270 .
3 . @xmath259 . in this case one
has @xmath271 . in total , the tripartite correlation
are found after integration over @xmath232 and @xmath233 to be = 12_x_y_z(_x+_y+_z ) , which does not depend on which party @xmath218 is alone .
one can check that the application of the signs @xmath272 has no influence on the tripartite correlations since they cancel out , and so @xmath273 .
the model thus reproduces the expected correlations .
it has been shown in @xcite how to define a hierarchy of sdp that characterizes the set @xmath51 of bipartite quantum correlations .
note that biseparable quantum correlations @xmath28 can be written as a finite sum of such bipartite quantum correlations using the fact that each @xmath274 in ( [ eq : svetq ] ) can be taken to be extremal and using the fact that there are a finite number of such extremal points corresponding to classical , deterministic strategies .
it therefore follows that biseparable quantum correlations can be characterized using a finite number of the sdp hierarchies introduced in @xcite .
note however that the number of single - party deterministic strategies , and thus the number of terms in eq . , grows exponentially with the number of measurement settings , making such an approach impractical even for small problems .
here we introduce an alternative sdp approach that has better scaling properties .
our approach is based on the observation that the tensor product separation @xmath91 at the level of states can be replaced by a commutation relation at the level of operators .
specifically , let @xmath92 denote the three possible partitions of the parties into two groups .
then , @xmath28 are biseparable quantum correlations if and only if there exist three arbitrary ( not necessarily biseparable ) states @xmath93 and three sets of measurement operators @xmath94 such that ( [ eq : decomp ] ) holds , i.e. let us start by showing that any correlations of the form is of the form .
note first the well - known fact that if the dimension of the hilbert space is not fixed , any single - party correlations @xmath274 admits a quantum representation @xmath275 $ ] where the operators @xmath45 commute among themselves .
indeed , let @xmath276 and @xmath277 , then a quantum representation of @xmath274 is achieved by defining @xmath278 and @xmath279 .
we can thus write in eq .
( [ eq : svetq ] ) , e.g. , @xmath280 $ ] with the operators @xmath34 commuting between themselves .
the decomposition ( [ eq : svetq ] ) is thus clearly a particular case of ( [ aeq : decomp ] ) .
let us now show the converse .
let us suppose first that @xmath281 and let @xmath276 .
since the projectors @xmath282 commute , the projectors @xmath283 defines a valid measurement and thus @xmath284 $ ] is a proper ( unormalized ) state of system @xmath144 .
define @xmath285 $ ] and @xmath286 as the deterministic point satisfying @xmath287 .
it is then easy to see using the properties of the operators @xmath282 that @xmath288=\sum_{\overline{c } } \delta(c_z , c)\text{tr}[m_{a|x}\otimes m_{b|y}\otimes m_{\overline{c } } \,\rho^s]=\sum_{\overline{c}}\delta(c_z , c)\text{tr}[m_{a|x}\otimes m_{b|y } \rho_{\overline{c}}]=\sum_{\overline{c}}p^{\overline{c}}_q(ab|xy)p^{\overline{c}}(c|z)$ ] .
this last expression is of the same form as the first series of terms in ( [ eq : svetq ] ) ( since any deterministic point @xmath286 can be seen as a single - party quantum point @xmath137 ) .
a similar argument for the other values of @xmath289 implies that any correlations of the form ( [ aeq : decomp ] ) are of the form ( [ eq : svetq ] ) .
we thus have reduced the problem of determining whether given correlations @xmath28 are biseparable to the problem of finding a set of operators satisfying a finite number of algebraic relations ( the projection defining relations of the type @xmath96 , @xmath97 and the commutation relations ) such that ( [ aeq : decomp1 ] ) holds .
such a problem is a typical instance of the sdp approach introduced in @xcite , which generalizes the results of @xcite .
specifically , it follows from the decomposition ( [ aeq : decomp ] ) and the results of @xcite that it is possible to define an infinite hierarchy of sets @xmath290 with the following properties : @xmath18 deciding whether given correlations belongs to @xmath291 can be determined using sdp ( though the size of the sdp increases with @xmath6 ) ; @xmath19 the sets @xmath291 approximate better and better @xmath48 from the outside , i.e. , @xmath292 and @xmath293 for all @xmath6 . if given correlations are found not to belong to some @xmath294
, we can thus conclude that they do not belong to @xmath48 and that they reveal genuinely tripartite entanglement .
further , it is possible in this case to derive an associated diew from the solution of the dual sdp .
note that modulo the assumption that the tensor product between the three measurement operators in ( [ aeq : decomp1 ] ) is equivalent to the commutation of these operators an assumption that is true in finite - dimensional hilbert space , but which still represents an open question in an infinite - dimensional hilbert space it can be shown that the hierarchy of relaxations converges to the set of biseparable states , i.e. , @xmath295 . thus if given correlations are not biseparable this will necessarily show - up at some finite step in the hierarchy . | we consider the problem of determining whether genuine multipartite entanglement was produced in an experiment , without relying on a characterization of the systems observed or of the measurements performed .
we present an @xmath0-partite inequality that is satisfied by all correlations produced by measurements on biseparable quantum states , but which can be violated by @xmath0-partite entangled states , such as ghz states .
in contrast to traditional entanglement witnesses , the violation of this inequality implies that the state is not biseparable independently of the hilbert space dimension and of the measured operators .
violation of this inequality does not imply , however , genuine multipartite non - locality .
we show more generically how the problem of identifying genuine tripartite entanglement in a device - independent way can be addressed through semidefinite programming .
the generation of multipartite entanglement is a central objective in experimental quantum physics .
for instance , entangled states of fourteen ions and six photons have recently been produced @xcite . in any such experiment
, a typical question arises : how can we be sure that _ genuine @xmath0-partite entanglement _ was present ?
a state is said to be genuinely @xmath0-partite entangled if it is not _
biseparable _ , that is , if it can not be prepared by mixing states that are separable with respect to some partition .
consider for instance the tripartite case : a state @xmath1 is said to be biseparable if it admits a decomposition [ stbis ] _
= _ k _ ab^k_c^k+_k _ ac^k_b^k+_k _ bc^k_a^k , where the weight of each individual state in the mixture has been included in its normalization ; a state that can not be written as above is genuinely tripartite entangled . determining whether genuine @xmath0-partite entanglement was produced in an experiment represents a difficult problem that has attracted much attention recently ( see e.g. @xcite ) .
the usual approach consists of measuring a witness of genuine multipartite entanglement , or of doing the full tomography of the state followed by a direct analysis of the reconstructed density matrix .
such approaches , however , not only rely on the observed statistics to conclude about the presence of entanglement , but also require a detailed characterization of the systems observed and of the measurements performed . consider for instance the following witness of genuine tripartite entanglement : [ mermin ] m = x_1x_2x_3 - x_1y_2y_3- y_1x_2y_3 - y_1y_2x_3 , where @xmath2 and @xmath3 are the pauli spin observables in the @xmath4 and @xmath5 direction for particle @xmath6 .
for any biseparable three - qubit state @xmath7 @xcite .
thus if we measure three spin-@xmath8 particles in the @xmath4 and @xmath5 direction and find an average value @xmath9 , we can conclude that the state exhibits genuine tripartite entanglement .
suppose , however , that the measurement @xmath10 carries a slight ( possibly unnoticed ) bias towards the @xmath4 direction .
that is , instead of measuring @xmath11 , we actually measure @xmath12 .
then it is not difficult to see , all other measurements being ideal , that the biseparable state @xmath13 , where @xmath14 , yields @xmath15 which is strictly larger than 2 for any @xmath16 .
thus , unless we measure all particles _ exactly _ along the @xmath4 and @xmath5 directions , we can no longer conclude that observing @xmath17 implies genuine tripartite entanglement .
importantly , this is not a unique feature of the above witness , but rather all conventional witnesses are , to some extent , susceptible to such systematic errors that are seldom taken into account .
furthermore , tomography and usual entanglement witnesses typically assume that the dimension of the hilbert space is known .
for instance , in a typical experiment demonstrating , say , entanglement between four ions , we usually view each ion as a two - level system .
but an ion is a complex object with many degrees of freedom ( position , vibrational modes , internal energy levels , etc ) .
how do we know , given the inevitable imperfections of the experiment , that it is justified to treat the relevant hilbert space of each ion as two - dimensional and how does this simplification affects our conclusions about the entanglement present in the system @xcite ? even if it is justified to view each ion as a qubit ,
is entanglement between four systems really necessary to reproduce the measurement data , or could they be reproduced with fewer entangled systems if qutrits were manipulated instead ?
these remarks motivate the introduction of entanglement witnesses that are able to guarantee that a quantum system exhibits ( @xmath0-partite ) entanglement , without relying on the types of measurements performed , the precision involved in their implementation , or on assumptions about the relevant hilbert space dimension .
we call such witnesses , _ device - independent _ entanglement witnesses ( diew ) .
this type of approach was already considered in refs .
@xcite .
note that other solutions to the above problems are possible , such as entanglement witnesses tolerating a certain misalignement in the measurement apparatuses @xcite or the characterization of realistic measurement apparatuses through squashing maps @xcite .
these types of more specific approaches , however , still require some partial characterization of the system and measurement apparatuses , which is not necessary when using diews .
any diew is a bell inequality ( i.e. a witness of nonlocality ) .
indeed , @xmath18 the violation of a bell inequality implies the presence of entanglement , and @xmath19 any measurement data that does not violate any bell inequality can be reproduced using quantum states that are fully separable @xcite .
the violation of a bell inequality is thus a necessary and sufficient condition for the detection of entanglement in a device - independent ( di ) setting .
this observation is the main insight behind di quantum cryptography @xcite , where the presence of entanglement is the basis of security .
the relation between diew for _ genuine @xmath0-partite _ entanglement and witnesses of multipartite nonlocality is more subtle . while there exist bell inequalities that detect genuine @xmath0-partite nonlocality @xcite , not every diew for @xmath0-partite entanglement corresponds to such a bell inequality .
consider for instance , the expression ( [ mermin ] ) .
if no assumptions are made on the type of systems observed and measurements performed , the inequality @xmath20 corresponds to mermin s bell - type inequality @xcite , i.e. , a value @xmath9 necessarily reveal non - locality , hence entanglement . moreover , a value @xmath21 guarantees genuine tripartite entanglement @xcite . the mermin expression ( [ mermin ] ) can thus be used as a tripartite diew .
yet , it can not be used as a bell inequality for genuine tripartite non - locality , since a simple model involving communication between two parties only already achieves the algebraic maximum @xmath22 @xcite .
the objectives of this paper are to formalize the concept of diew for genuine multipartite entanglement and initiate a systematic study that goes beyond the early examples given in @xcite . following this line ,
we start by introducing the notion of quantum biseparable correlations .
we then present a simple diew for @xmath0-partite entanglement which is stronger for ghz ( greenberger - horne - zeilinger ) states than all the inequalities introduced in @xcite . in the case
@xmath23 , we also provide a general method for determining whether given correlations reveal genuine tripartite entanglement and apply it to ghz and w states . apart from yielding practical criteria for the characterization of entanglement in a multipartite setting
, our results also clarify the relation between device - independent multipartite entanglement and mulitpartite nonlocality . *
1 . biseparable quantum correlations . * for simplicity of exposition ,
let us consider an arbitrary tripartite system ( the following discussion easily generalizes to the @xmath0-party case ) . to characterize in a di way its entanglement properties
, we consider a bell - type experiment : on each subsystem , one of @xmath24 possible measurements is performed , yielding one of @xmath25 possible outcomes .
we adopt a black - box description of the experiment and represent the measurements on each of the three subsystems by classical labels @xmath26 ( corresponding , e.g. , to the values of macroscopic knobs on the measurement apparatuses ) and denote the corresponding classical outcomes @xmath27 .
the correlations obtained in the experiment are characterized by the joint probabilities @xmath28 of finding the triple of outcomes @xmath29 given the measurement settings @xmath30 .
we say that the correlations @xmath28 are _ biseparable quantum correlations _ if they can be reproduced through local measurements on a biseparable state @xmath1 .
that is , if there exist a biseparable quantum state in some hilbert space @xmath31 , measurement operators @xmath32 , @xmath33 , and @xmath34 ( which without loss of generality we can take to be projections satisfying @xmath35 and @xmath36 ) , such that [ eq : corrs ] p(abc|xyz ) = .
if given quantum correlations @xmath28 are not biseparable , they necessarily arise from measurements on a genuinely tripartite entangled state , and this conclusion is independent of any assumptions on the type of measurements performed or on the hilbert space dimension .
equivalently , biseparable quantum correlations can be defined as those that can be written in the form [ eq : svetq ] p&(abc|xyz ) = _ k p^k_q(ab|xy)p_q^k(c|z ) + & + _ k p^k_q(ac|xz)p_q^k(b|y)+_k p^k_q(bc|yz)p_q^k(a|x ) , where @xmath37 and @xmath38 correspond , respectively , to arbitrary two - party and one - party quantum correlations , i.e. , they are of the form @xmath39 $ ] and @xmath40 $ ] for some unormalized quantum states @xmath41 , @xmath42 and measurement operators @xmath43 , @xmath44 , @xmath45 ( and similarly for the other terms in ( [ eq : svetq ] ) ) .
note that here the measurement operators for different bi - partitions do not need to be the same ( though this can always be achieved as shown in appendix [ app : sdp ] ) .
that any correlations of the form ( [ eq : corrs ] ) is of the form ( [ eq : svetq ] ) is immediate using the definition ( [ stbis ] ) of biseparable states .
conversely , it is easy to see that any correlations of the form ( [ eq : svetq ] ) is of the form ( [ eq : corrs ] ) , see appendix [ app : def ] .
let @xmath46 denote the set of tripartite quantum correlations and @xmath47 the set of biseparable quantum correlations . from ( [ eq
: svetq ] ) , it is clear that @xmath48 is convex and that its extremal points are of the form @xmath49 where @xmath50 is an extremal point of the set @xmath51 of bipartite quantum correlations and @xmath52 an extremal point of the set @xmath53 of single - party correlations ( the extreme points of @xmath53 are actually classical , deterministic points ) .
since the set @xmath48 is convex , it can be entirely characterized by linear inequalities . those linear inequalities separating @xmath48 from @xmath46
correspond to diews for genuine tripartite entanglement .
since @xmath48 has an infinite number of extremal points , there exist an infinite number of such inequalities .
note that the set of local correlations @xmath54 is contained in @xmath48 .
this implies that any diew for genuine tripartite entanglement is a bell inequality ( though not necessarily a tight bell inequality ) .
note also that the decomposition ( [ eq : svetq ] ) corresponds to a svetlichny - type decomposition @xcite where all bipartite factors are restricted to be quantum , whereas less restrictive constraints ( or even none in svetlichny original definition @xmath55 ) are imposed on these bipartite terms in the definitions of multipartite non - locality @xcite .
it follows that the set of genuinely bipartite non - local correlations is larger than the set of biseparable quantum correlations as illustrated in fig .
[ fig : fig ] .
thus while any bell inequality detecting genuine tripartite non - locality is a diew for genuine tripartite entanglement , the converse is not necessarily true .
all these observations extend to the @xmath0-party case . ) , svetlichny correlations ( @xmath56 ) and biseparable quantum correlations ( @xmath48 ) .
the point @xmath57 corresponds to random correlations and @xmath58 to the ghz correlations maximally violating the diew ( [ eq : witness ] ) , which is represented by the straight line @xmath59 ; note that a diew can be violated by svetlichny - local correlations .
( the svetlichny polytope @xmath56 can be determined exactly using linear programming , while @xmath46 and @xmath48 can be approximated efficiently using sdp techniques , see main text . ) ] * 2 .
a diew for @xmath0-partite entanglement . *
we now present a diew for @xmath0 parties , where each party @xmath60 performs a measurement @xmath61 and obtains an outcome @xmath62 .
we denote @xmath63 the correlator associated to the measurement settings @xmath64 , i.e. the expectation value @xmath65 , where @xmath66 denotes an @xmath0-tuple of outcomes .
let @xmath67 be the sum of correlators @xmath68 for which the measurement settings @xmath69 of the @xmath0 parties sum up to @xmath70 .
let @xmath71 be a function such that @xmath72 and taking successively the values @xmath73 $ ] on the integers @xmath74 .
then the inequality [ eq : witness ] i_n = _ k = n^3n f_k - ne_n^k2 3^n-3/2 is satisfied by all biseparable quantum correlations , and is thus a diew for genuine @xmath0-partite entanglement .
the proof of this statement is based on the decomposition ( [ eq : svetq ] ) and is given in appendix [ app : diew ] .
the svetlichny bound associated to the expression @xmath75 , on the other hand , is easily found to be @xmath76 ( see appendix [ app : diew ] ) @xcite . we now illustrate how this diew can be used to detect genuine multipartite entanglement . for this ,
let us consider the noisy ghz state @xmath77 characterized by the visibility @xmath78 .
carrying out the measurements @xmath79 on all parties , we obtain @xmath80 , which violates ( [ eq : witness ] ) provided that @xmath81 .
the diew can thus detect in a di way genuine @xmath0-partite entanglement in a noisy ghz state for visibilities as low as @xmath82 .
this significantly improves over the threshold visibility @xmath83 required to violate the diew based on the mermin expression or the different inequalities introduced in @xcite .
note that in a di setting it is not possible to detect the genuine multipartite entanglement in tripartite ghz states below @xmath84 using projective measurements .
indeed , we have shown in this case that there exists a biseparable model reproducing all ghz correlations ( see appendix [ app : model ] ) . in the case
@xmath23 , the diew ( [ eq : witness ] ) takes the form @xmath85 .
it therefore involves only 18 expectation values , compared to @xmath86 for a full tomography of a three - qubit system .
let us stress , however , that contrary to usual entanglement witnesses @xmath59 is not restricted to two - dimensional hilbert spaces , even though it uses observables with binary outcomes .
for instance , if all parties perform the measurements @xmath87 with @xmath88 on the three - qutrit state @xmath89 , then @xmath90 , showing that the state is genuinely tripartite - entangled . *
3 .
general characterization of biseparable quantum correlations in the case @xmath23 .
* though the diew ( [ eq : witness ] ) seems particularly well adapted to ghz states , we can not expect a single nor a finite set of diew to completely characterize the biseparable region , as illustrated in fig .
1 .
it is thus desirable to derive a general method to decide whether arbitrary correlations are biseparable .
here we show how the semi - definite programming ( sdp ) techniques introduced in @xcite can be used to certify that the correlations observed in an experiment are genuinely tripartite entangled .
our approach is based on the observation that the tensor product separation @xmath91 at the level of states in the definition ( [ stbis ] ) of biseparable states can be replaced by a commutation relation at the level of operators .
specifically , let @xmath92 denotes the three possible partitions of the parties into two groups .
then , @xmath28 are biseparable quantum correlations if and only if there exist three arbitrary ( not necessarily biseparable ) states @xmath93 and three sets of measurement operators @xmath94 such that [ eq : decomp ] p(abc|xyz)=_s [ m^s_a|xm^s_b|ym^s_c|z^s ] , where measurement operators corresponding to an isolated party commute , i.e. , @xmath95=0 $ ] , and similarly for the other partitions .
the equivalence between ( [ eq : corrs ] ) and ( [ eq : decomp ] ) is established in appendix [ app : sdp ] .
the problem of determining whether given correlations @xmath28 are biseparable thus amounts to finding a set of operators satisfying a finite number of algebraic relations ( the projection defining relations of the type @xmath96 , @xmath97 and the commutation relations mentioned above ) such that ( [ eq : decomp ] ) holds .
such a problem is a typical instance of the sdp approach introduced in @xcite ( see details in appendix [ app : sdp ] ) .
specifically , it follows from the results of @xcite that one can define an infinite hierarchy of criteria that are necessarily satisfied by any correlations of the form ( [ eq : decomp ] ) and which can be tested using sdp .
if given correlations do not satisfy one of these criteria , we can conclude that they reveal genuinely tripartite entanglement .
further , it is possible in this case to derive an associated diew from the solution of the dual sdp .
modulo a technical assumption , it can be shown that the hierarchy of sdp criteria is complete , that is , if given correlations are not biseparable this will necessarily show - up at some finite step in the hierarchy .
.summary of numerical investigations . [ cols="^,^,^",options="header " , ] *
4 . application to ghz and w states .
* using finite levels of this hierarchy and optimizing over the possible measurements , we investigated the minimal visibilities above which the ghz state @xmath98 and the w state @xmath99 exhibit correlations that are not biseparable ( and thus reveal genuine tripartite entanglement ) in the case of two and three measurement settings per party .
our results are summarized in table [ tab : tab ] . for ghz states ,
the reported visibility @xmath100 for three measurements per party correspond to the threshold required to violate the diew ( [ eq : witness ] ) , suggesting that this diew is optimal in this case . in the case of two measurements per party ,
we could not lower the visibility below the threshold @xmath101 , which corresponds to the visibility required to violate the diew based on mermin expression ( [ mermin ] ) and the diews introduced in @xcite . note , however , that for @xmath102 the ghz state violates svetlichny s inequality @xcite and thus exhibits genuine tripartite nonlocality .
thus for ghz state the diews introduced in @xcite do not improve over what can already be concluded using the standard notion of tripartite non - locality .
on the other hand , our numerical explorations suggest that the visibilities @xmath100 for ghz states with three measurements and @xmath103 for w states with two measurements can not be attained using the notion of genuine tripartite non - locality , illustrating the interest of the weaker notion of diew .
* dicussion . *
to conclude , we comment on some possible directions for future research .
first of all , note that by identifying the measurement settings `` @xmath104 '' with @xmath105 and `` @xmath106 '' with @xmath107 , the two - setting diew based on mermin expression ( [ mermin ] ) can be written as @xmath108 , which is of the same general form as the three - setting diew ( [ eq : witness ] ) .
this suggests that the diews based on ( [ mermin ] ) and ( [ eq : witness ] ) actually form part of a larger family of @xmath24-settings diews .
this question deserves further investigation .
a second problem is to derive simple diews that are adapted to w states and that can in particular reproduce the threshold visibilities obtained in table 1 .
finally , we have shown a practical method to characterize three - partite biseparable correlations using sdp .
it would be interesting to understand how to generalize these results to the @xmath0-partite case .
a possibility would be to combine the approach of @xcite with the symmetric extensions introduced in @xcite .
this question will be investigated elsewhere .
[ [ acknowledgments . ] ] acknowledgments .
+ + + + + + + + + + + + + + + + this work was supported by the swiss nccrs qp and qsit , the european erc - ag qore , and the brussels - capital region through a bb2b grant .
99 t. monz _
et al .
_ , arxiv:1009.6126 w. wieczorek et al , phys .
rev .
lett . * 103 * , 020504 ( 2009 ) . o. ghne , g. tth , phys .
rep . * 474 * , 1 ( 2009 ) .
o. ghne , m. seevinck , new j. phys .
* 12 , * 053002 ( 2010 ) ; b. jungnitsch , t. moroder , o. ghne , arxiv:1010.6049 ; m. huber et al . ,
arxiv:1011.4087 .
g. tth , o. ghne , m. seevinck , j. uffink , phys .
rev .
a * 72 * , 014101 ( 2005 ) .
t. moroder _
et al . _ ,
phys .
rev .
a * 81 * , 052342 ( 2010 ) .
m. seevinck , j. uffink , phys .
rev .
a * 65 * , 012107 ( 2001 ) .
k. nagata , m. koashi , n. imoto phys .
rev .
lett . * 89 * , 260401 ( 2002 ) ; j. uffink , phys . rev . lett . * 88 * , 230406 ( 2002 ) ; m. seevinck , g. svetlichny , phys . rev . lett .
* 89 * , 060401 ( 2002 ) .
m. seevinck , j. uffink phys .
rev .
a * 76 * , 042105 ( 2007 ) .
r. f. werner , , 4277 ( 1989 ) .
a. acn et al , phys .
rev .
lett . * 98 * , 230501 ( 2007 ) .
s. pironio _
et al . _ ,
nature * 464 * , 1021 ( 2010 ) .
g. svetlichny , phys .
rev .
d * 35 * , 3066 ( 1987 ) .
d. collins _ et al .
_ , phys .
rev .
lett . * 88 * , 170405 ( 2002 ) .
m. ukowski , d. kaszlikowski , * 56 * , r1682 ( 1997 ) .
j .-
d .
bancal , n. brunner , n. gisin , y .- c .
liang , phys .
rev .
lett * 106 * , 020406 ( 2011 ) .
n. d. mermin , phys .
rev .
lett .
* 65 * , 1838 ( 1990 ) .
m. navascus , s. pironio , a. acn , phys .
rev .
lett . * 98 * , 010401 ( 2007 ) ; new j. phys .
* 10 * , 073013 ( 2008 ) .
s. pironio , m. navascus , a. acn , siam j. optim .
vol * 20 * , issue 5 , 2157 ( 2010 ) .
b. m. terhal , a. c. doherty , d. schwab , phys .
rev .
lett . * 90 * , 157903 ( 2003 ) ; a. c. doherty , p. a. parrilo , f. m. spedalieri , phys . rev
. lett . * 88 * , 187904 ( 2002 ) .
s. pironio , j .- d .
bancal , v. scarani , j. phys . a : math .
theor .
_ 44 _ , 065303 ( 2011 )
. s. braunstein , c. caves , annals of phys .
* 202 * , 22 ( 1990 ) .
s. wehner , phys .
rev .
a * 73 * , 022110 ( 2006 ) .
see @xcite for the local bound of @xmath75 . |
the study of neutral pion production in proton - proton collision near threshold , @xmath0 , has been inspired by precise measurements of the near - threshold cross section @xcite .
surprisingly , the measured cross section turned out to be @xmath1 5 times larger than the early theoretical predictions @xcite .
subsequently , some mechanisms to account for the threshold experimental data have been suggested in model calculations @xcite .
heavy - baryon chiral perturbation theory ( hb@xmath2pt ) is a low - energy effective field theory ( eft ) of qcd and provides us a systematic perturbation scheme in terms of @xmath3 where @xmath4 denotes small external momentum and/or symmetry breaking term @xmath5 and @xmath6 denotes the chiral scale @xmath7 gev : @xmath8 is the pion decay constant .
though many works on the @xmath0 reaction near threshold in hb@xmath2pt have been done @xcite ( for a recent review , see ref .
@xcite and references therein ) , some issues in theoretically describing the process have not been fully clarified . in the next - to - leading order ( nlo )
hb@xmath2pt calculations @xcite , a significant enhancement of the off - shell @xmath9 vertex function obtained from the nlo hb@xmath2pt lagrangian is found .
however , the two - body ( one - pion - exchange ) matrix element with the off - shell @xmath9 vertex is almost exactly canceled with the one - body matrix element . thus the experimental data can not be reproduced in the nlo calculations . in the next - to - next - to leading order ( nnlo ) hb@xmath2pt calculations @xcite
, a significant contribution comes out of the nnlo corrections and a moderate agreement with the experimental data is obtained @xcite .
however , the chiral series based on the standard weinberg s counting rules @xcite shows poor convergence . a modification of the original weinberg s counting rules to account for the large momentum transfer , @xmath10 where @xmath5 and @xmath11 are the pion and nucleon masses , respectively , is discussed in ref .
the production operators at nlo using the modified counting rules are estimated , and it was reported that the nlo contributions exactly cancel among themselves @xcite . recently , some detailed issues for the loop calculations , such as a concept of reducibility @xcite , a representation invariance of the chiral fields among the loop diagrams , and a proper choice of the heavy - nucleon propagator @xcite , were also studied .
meanwhile , it is known that the energy dependence of the experimental data can be well reproduced in terms of the final state interaction and the phase space @xcite . a `` minimal '' formalism to take account of these two features would be a _
theory , in which virtual pions exchanged between the two nucleons are integrated out ; in this pionless theory , the one - pion exchange , two - pion exchange and contact terms in hb@xmath2pt are subsumed in a contact term .
furthermore , after taking these two features into account , the difference between the theory and experiment appears in the overall factor and the experimental data can be easily reproduced by fitting an unknown constant that appears in a contact vertex . in this work
@xcite we employ a pionless eft with a di - baryon @xcite cross section at bbn energies@xcite and neutron - neutron fusion process@xcite employing this formalism . ] and external pions @xcite to calculate the total cross section of the @xmath0 process .
an effective lagrangian without virtual pions and with a di - baryon and external pions for describing the @xmath0 reaction may read = l_n + l_s + l_ns + l_nn^p , where @xmath12 is the standard one - nucleon lagrangian in heavy - baryon formalism where the `` external '' pions are nonlinearly realized .
@xmath13 is for the @xmath14 channel di - baryon field , @xmath15 represents the contact interaction of an external pion , a di - baryon and two nucleons , and @xmath16 is for the two - nucleon @xmath17 channel .
the effective lagrangian for the two - nucleon part may read @xcite _ s & = & _ s s_a^s_a + & & -y_s , + l_ns & = & \ { i_abc s_a^ . + & & .
( iv_c ) + } , + l_nn^p & = & c_2 ^ 0 _ ij_kl14 ( n^to_ij , a^1,pn)^(n^to_kl , a^1,pn ) + , with @xmath18 and @xmath19 .
@xmath20 is the @xmath14 channel di - baryon field and @xmath21 is the sign factor @xmath22 which we fix below .
@xmath23 is a velocity vector @xmath24 and @xmath25 is the covariant derivative .
@xmath26 is the mass difference between the di - baryon mass @xmath27 and two - nucleon mass , @xmath28 .
@xmath29 is the coupling constant of the di - baryon and two - nucleon interaction .
@xmath31 is an unknown low energy constant ( lec ) of the ( external ) pion-(spin singlet ) dibaryon - nucleon - nucleon ( @xmath32 ) interaction .
@xmath33 is the effective range in the @xmath14 ( @xmath34 ) channel , and @xmath35 .
@xmath36 is the lec for the @xmath37 scattering in the @xmath17 channel .
now we calculate the @xmath38- and @xmath39-wave @xmath37 scattering amplitudes to fix the lecs in the two - nucleon part . in fig .
[ fig;dibaryon - propagator ] , diagrams for the dressed @xmath14 channel di - baryon propagator are shown where the two - nucleon bubble diagrams including the coulomb interaction are summed up to the infinite order . , width=340 ] -wave @xmath34 scattering amplitude with the coulomb interaction .
[ fig;nnamplitudes],width=139 ] in fig .
[ fig;nnamplitudes ] , a diagram of the @xmath38-wave @xmath34 scattering amplitude with the coulomb interaction is shown and thus we have the @xmath38-wave scattering amplitude as + & & = i , where @xmath40 and @xmath41 is the @xmath38-wave coulomb phase shift @xmath42 .
@xmath43 is the dressed di - baryon propagator and @xmath44 is the renormalized mass difference .
@xmath45 , @xmath46 , @xmath47 , and c_^2 = , = . the @xmath38-wave amplitude @xmath48 is given in terms of the effective range parameters as ia_s = i , where @xmath49 is the scattering length
, @xmath33 is the effective range , and the ellipsis represents the higher order corrections .
now it is easy to match the lecs with the effective range parameters .
thus we have @xmath50 and y_s & = & , [ eq;ys ] + d_s(p ) & = & .
we note that the sign of the lec @xmath51 can not be determined by the effective range parameters . in fig .
[ fig;pwavescattering ] , diagrams for the @xmath39-wave @xmath37 scattering are shown .
-wave @xmath37 scattering .
[ fig;pwavescattering],width=302 ] because the momenta of the two protons are quite large for the pion production reaction , the two - proton bubble diagrams are summed up to the infinite order .
the scattering amplitude for the @xmath52 channel is obtained as ia_p & = & .
the lec @xmath36 is fixed by the phase shift of the @xmath52 channel at pion production threshold , @xmath53 at @xmath54 .
thus we have p_th^3_p(p_th ) .
near threshold with the strong and coulomb final state interactions and without the initial state interaction .
[ fig;diagrams - non - fsi],width=226 ] with the strong initial and the strong and coulomb final state interactions .
[ fig;diagrams - fsi],width=302 ] in figs .
[ fig;diagrams - non - fsi ] and [ fig;diagrams - fsi ] , we show diagrams for @xmath0 near threshold . in diagram ( a ) in fig .
[ fig;diagrams - non - fsi ] and ( c ) in fig .
[ fig;diagrams - fsi ] , the pion is emitted from the one - body @xmath55 vertex . in the diagram ( b ) in fig .
[ fig;diagrams - non - fsi ] and ( d ) in fig .
[ fig;diagrams - fsi ] , the pion is emitted from the @xmath32 contact vertex which is proportional to the lec @xmath56 .
the one - body amplitude from the ( a ) and ( c ) diagrams and the two - body ( contact ) amplitude from the ( b ) and ( d ) diagrams are obtained as + & & , [ eq;aac ] + + & & [ eq;abd ] where @xmath57 and @xmath58 are the relative three momenta between incoming and outgoing two protons , respectively ; @xmath59 and @xmath60 . @xmath61 and @xmath62 is the energy of outgoing pion , @xmath63 : @xmath64 is the outgoing pion momentum .
we note that there remain no unknown parameters in the amplitudes except for the lec @xmath56 in the two - body ( contact ) amplitude in eq .
( [ eq;abd ] ) .
now we estimate an order of magnitude of the lec @xmath56 from hb@xmath2pt .
we here consider a one - pion - exchange ( ope ) diagram shown in fig .
[ fig;diagrams - loops ] .
process for estimating the lec @xmath56 in the contact vertex .
[ fig;diagrams - loops],width=139 ] this diagram is the lowest order ope contribution in the standard weinberg counting rules .
we include a higher order ( relativistic ) correction to the @xmath9 vertex which is found to be important @xcite and is , in the modified counting rules , of the same order as the lowest order diagram .
the effective chiral lagrangian to calculate the isoscalar @xmath9 interaction in the diagram in fig .
[ fig;diagrams - loops ] , reads @xcite @xmath65 where _
n^(2 ) & = & n^n + , [ eq;l2 ] and @xmath66 is the relativistic correction to the term proportional to @xmath67 in eq .
( [ eq;l2 ] ) .
the values of the lecs @xmath68 , @xmath69 and @xmath70 are fixed in the tree - level calculations @xcite as c_1 = -0.64 , c_2 = 1.79 , c_3 = -3.90 [ ] .
[ eq;c123 ] the value of the lec @xmath56 from the loop diagram in fig .
[ fig;diagrams - loops ] is obtained as _
^(2 ) & & ( -4c_1 + 2c_2 - + c_3 ) + & & 0.140 , [ eq;dpi2 ] where the different signs for @xmath56 have been obtained because of the lec @xmath51 in eq . ( [ eq;ys ] ) .
the total cross section of @xmath0 near threshold is calculated using the formula = 12 ^q^max_0 dq , = _ spin|a|^2 , [ eq;sigma ] with @xmath71 , and @xmath72 .
@xmath73 is the initial total energy @xmath74 and @xmath75 .
we have expanded the proton energies in the phase factor in terms of @xmath76 and kept up to the @xmath76 order .
@xmath77 is the amplitude @xmath78 where @xmath79 and @xmath80 are obtained in eqs .
( [ eq;aac ] ) and ( [ eq;abd ] ) , respectively . in fig .
[ fig;cs ] we plot our results for the total cross section as a function of @xmath81 .
the solid curve and long - dashed curve have been obtained by using @xmath82 @xmath83 fixed from the one - pion exchange diagram in fig .
[ fig;diagrams - loops ] in the previous section .
the lec @xmath56 is also fixed by using the experimental data as
_ ^(2)fitted = - 0.12 , + 0.55 ^5/2 , where we have two values of @xmath56 with different signs .
the short - dashed curve is obtained by using @xmath85 @xmath83 .
the dotted line corresponds to the case where only the contribution from the one - body amplitude @xmath79 is considered .
the experimental data are also included in the figure .
we find that the experimental data are reproduced reasonably well with the value of @xmath86 @xmath83 .
by contrast , we obtain almost vanishing total cross sections with the value @xmath87 @xmath83 because the two - body amplitude with @xmath87 @xmath83 is almost canceled with the amplitude from the one - body contribution .
on the other hand , for the whole energy range the experimental near - threshold cross section data are well reproduced with the use of the fitted parameter @xmath88 @xmath83 .
we also find that approximately a half of the observed cross section comes from the one - body ( ia ) amplitude in the pionless theory . in this work we calculated the total cross section for @xmath0 near threshold in pionless eft with the di - baryon and external pion fields .
the leading one - body amplitude and subleading contact amplitude were obtained including the strong initial state interaction and the strong and coulomb final - state interactions .
after we fix the lecs for the @xmath37 scatterings , there remains only one unknown constant , @xmath56 , in the amplitude .
we estimated it from the one - pion exchange diagram in the pionful theory .
although this method does not allow us to fix the sign of @xmath56 , we have found that one of the two choices for @xmath56 leads to the cross sections that agree with the experimental data reasonably well . on the other hand
, the whole range of the experimental data near threshold can be reproduced by adjusting the only unknown lec in the theory , @xmath56 . as discussed in introduction ,
this is an expected result because the energy dependence of the experimental total cross section is known to be well described by the phase factor and the final - state interaction @xcite , which have been taken into account in this work , and the overall strength of the cross section can be adjusted by the value of @xmath56
. this feature would be the same in the nnlo hb@xmath2pt calculations because an unknown constant appears in the contact @xmath89 vertex and can be adjusted so as to reproduce the experimental data though there are many other corrections coming out of the pion loop diagrams .
the author would like to thank the organizers for the conference appc10 , august 21 - 24 , 2007 , pohang , korea for hospitality .
this work is supported by korean research foundation and the korean federation of science and technology societies grant founded by korean government ( moehrd , basic research promotion fund ) : the brain pool program ( 052 - 1 - 6 ) and krf-2006 - 311-c00271 .
h. o. meyer _
et al . _ ,
* 23 * ( 1990 ) 2846 .
h. o. meyer _
et al._. nucl
. phys . * a 539 * ( 1992 ) 633 .
d. s. koltum and a. reitan , phys . rev .
* 141 * ( 1966 ) 1413 : g. a. miller and p. u. sauer , phys .
c * 44 * ( 1991 ) r1725 .
h. lee , d. o. riska , phys .
lett . * 70 * ( 1993 ) 2237 : e. hernndez , e. oset , phys . lett .
* b 350 * ( 1995 ) 158 .
y . park , , f. myhrer , j. r. morones , t. meissner , and k. kubodera , phys . rev .
c * 53 * ( 1996 ) 1519 .
t. d. cohen , j. l. friar , g. a. miller , and u. van kolck , phys .
c * 53 * ( 1996 ) 2661 .
t. sato , t .- s .
h. lee , f. myhrer , and k. kubodera , phys .
c * 56 * ( 1997 ) 1246 .
v. dmitrainovi , k. kubodera , f. myhrer , and t. sato , phys . lett .
* b 465 * ( 1999 ) 43 .
s. ando , t .- s .
park , and d .-
min , phys .
* b 509 * ( 2001 ) 253 .
y. kim , t. sato , f. myhrer , and k. kubodera , arxiv:0704.1342 .
u. van kolck , g. a. miller , and d. o. riska , phys . lett .
* b 388 * ( 1996 ) 679 .
v. bernard , n. kaiser , u .- g .
meiner , eur .
j. a * 4 * ( 1999 ) 259 . c. hanhart , phys
* 397 * ( 2004 ) 155 .
s. weinberg , phys .
* b 251 * ( 1990 ) 288 ; nucl . phys .
* b 363 * ( 1991 ) 3 . c. hanhart and n. kaiser , phys .
c * 66 * ( 2002 ) 054005 .
v. lensky , v. baru , j. haidenbauer , c. hanhart , a. e. kudrayavtsev , and u .- g .
meiner , eur .
j. * a 27 * ( 2006 ) 37 .
c. hanhart and a. wirzba , phys .
* b 650 * ( 2007 ) 354 .
s. ando , arxiv:0707.2157 , to appear in eur .
j. * a*. s. r. beane , m. j. savage , nucl
. phys . * a 694 * ( 2001 ) 511 .
s. ando and c. h. hyun , phys .
c * 72 * ( 2005 ) 014008 .
s. ando , r. h. cyburt , s. w. hong , c. h. hyun , phys .
c * 74 * ( 2006 ) 025809 .
s. ando and k. kubodera , phys .
* b 633 * ( 2006 ) 253 .
s. r. beane , m. j. savage , nucl . phys . * a 717 * ( 2003 ) 104 .
chen , g. rupak , m. j. savage , nucl .
phys . * a 653 * ( 1999 ) 386 .
s. fleming , t. mehen , i. w. stewart , nucl .
phys . * a 677 * ( 2000 ) 313 .
x. kong and f. ravndal , phys . lett .
* b 450 * ( 1999 ) 320 .
s. ando , j. w. shin , c. h. hyun , s. w. hong , arxiv:0704.2312 .
n. fettes , u .- g .
meiner , m. moji , s. steininger , ann . of phys .
* 283 * ( 2000 ) 273 ; ibid * 288 * ( 2001 ) 249 .
v. bernard , n. kaiser , u .- g .
meiner , nucl . phys .
* b 457 * ( 1995 ) 147 . | in this talk , we review our recent calculation for the @xmath0 reaction near threshold in pionless effective field theory with a di - baryon and external pions . |
powder samples of tb@xmath4ti@xmath5o@xmath6 with @xmath7= 0 , 0.01 and -0.01 were synthesized as described in ref . .
the value of @xmath7 was adjusted by changing the mass of the starting constituents , tb@xmath24o@xmath3 and tio@xmath2 , which were heated in air at 1350@xmath25c for several days with periodic grindings to ensure complete reactions .
the obtained samples were ground into powder then annealed at 800@xmath25c in air for one day .
another stoechiometric sample , synthesized in 2004 using a similar procedure @xcite and previously studied with neutrons @xcite was also measured for comparison .
all samples were characterized by x - ray diffraction and high resolution neutron diffraction patterns were also recorded for the stoechiometric samples on the spectrometer 3t2 at the laboratoire lon brillouin ( llb ) , yielding a crystal structure in agreement with previous results .
slight colour changes between the samples with @xmath7=-0.01 ( beige ) , 0 ( white ) , + 0.01 ( light brown ) are attributed to small variations in the concentration of ti@xmath9 and oxygen ions induced by the off - stoechiometry .
powder neutron diffraction patterns were recorded on the diffractometer d1b at the institut laue langevin ( ill ) with a neutron wavelength @xmath26 = 2.52 , down to 0.07k .
as shown previously @xcite a typical powder pattern of tb@xmath2ti@xmath2o@xmath3 consists of nuclear bragg peaks coexisting with a large diffuse signal from the environmental background , incoherent scattering , and short range magnetic correlations between first neighbors tb@xmath9 moments .
these @xmath19liquid - like@xmath19 correlations start to appear @xcite below about 50k .
the mesoscopic magnetic order studied here consists of small lorentzian peaks which appear on the top of this diffuse signal on the pattern measured at 0.07k ( fig .
[ figunsubtracted ] ) . to isolate them from the other contributions , a neutron pattern measured at 1.2(1)k
was subtracted .
the amplitude of the ordered magnetic moment was calibrated to that of the nuclear bragg peaks by refining the crystal structure on the pattern at 0.07k ( inset fig .
[ figunsubtracted ] ) .
specific heat data @xmath21 were recorded below 20 k and down to 0.4 k using the relaxation method with a physical property measurement system ( quantum design , inc . ) . the phonon contribution to the total specific heat
can be safely neglected below 8 k. in the temperature range of the measurements , i.e. above 0.4 k , the nuclear specific heat varies with the temperature as @xmath27 . following an earlier estimate @xcite
, its contribution corresponds to 0.35 and 0.155 j / k.mol at 0.4 and 0.6 k , respectively .
the influence of the nuclear part in the upturn of the specific heat below 0.5 k for the @xmath28 and @xmath29 samples is therefore marginal .
ti@xmath5o@xmath6 : neutron magnetic patterns , focusing on the region of the most intense magnetic peaks .
a pattern at 1.2(1)k was subtracted to extract the @xmath0= ( @xmath1 , @xmath1 , @xmath1 ) mesoscopic order .
( a ) neutron patterns at 0.07k for different tb contents
intensities have been scaled to the intensity of a nuclear bragg peak to allow a comparison between the samples .
( b ) neutron patterns versus temperature for the sample @xmath7=0 ( 2004 ) .
short and long dashed lines shows the positions of ( 111 ) and ( 200 ) bragg peaks respectively .
, width=8 ]
the lattice constant deduced from x - rays increases linearly with @xmath7 .
the two stoechiometric samples have close values ( a=10.15214(5 ) and 10.15229(5) at 300k for the x=0 ( 2004 ) and x=0 sample respectively ) .
the values for x=0 and 0.01 are slightly lower than reported in ref . , by less than 5 10@xmath30 ( inset fig . [ figcp]a ) .
this difference could be connected with the initial characteristics of the starting constituents , such as the particle sizes of the starting powders , or with some differences in the thermal treatment procedures such as different cooling rates at the end of the thermal treatments and annealing .
specific heat data provide a direct evidence of the exotic transition @xcite . as shown in fig .
[ figcp ] , the @xmath21 curves of all samples overlap at high temperature .
they strongly differ below 1k , showing either a decrease ( @xmath7=-0.01 ) or an upturn followed by a sharp peak ( @xmath7=0 , 0.01 ) .
these systematic changes with @xmath7 reflect the phase diagram of ref . .
the @xmath7=-0.01 sample situates in the @xmath19spin liquid@xmath19 region of ref . .
for @xmath7=0 and @xmath7=+0.01 , a phase transition occurs at 0.4k .
the sample with nominal value @xmath7=0 synthesized in 2004 does not show such an upturn , possibly due to different characteristics of the starting constituents or slight differences in the thermal treatment .
it confirms that a fine tuning of the off - stoechiometry requires a fully reproducible procedure in all the steps of the synthesis .
ti@xmath5o@xmath6 : neutron magnetic patterns at 0.07k for different tb contents
a pattern at 1.2(1)k was subtracted .
scales have been enlarged and error bars omitted for clarity . blue and red tick marks show the positions of the magnetic bragg peaks for the mesoscopic and long range order respectively .
short and long dashed lines show narrow signals at the position of the ( 111 ) and ( 200 ) bragg peaks respectively .
solid lines are the best refinements corresponding to spin ice - like and monopole - like structures , both yielding the same powder averaged intensity ( see text ) . , width=302 ] the magnetic patterns of all samples are shown in fig . [ figpatterns](a ) at the lowest temperature 0.07k , focusing on the low angle region .
intensities were scaled to the intensity of a nuclear bragg reflection to allow a direct comparison between the samples .
the patterns are quite similar , showing broad magnetic peaks with lorenzian lineshape . in the ( @xmath7=0 , 0.01 )
samples , a small bragg - like reflection of width limited by the experimental resolution can be detected on top of the strongest peaks .
this bragg - like contribution is clearly absent in the @xmath7=-0.01 sample .
the lorenzian peaks are indexed in the space group ( sg ) @xmath31 with a @xmath0= ( @xmath1 , @xmath1 , @xmath1 ) propagation vector .
the strongest ( @xmath1 @xmath1 @xmath1 ) and ( @xmath1 @xmath1 @xmath32 ) peaks are clearly seen at low angles .
the ( @xmath1 @xmath1 @xmath33 ) peaks and weaker peaks with higher miller indices can be detected at higher angles in the enlarged patterns ( figure 4 ) .
the temperature dependence of the magnetic pattern is shown in figure 3(b ) for the sample x=0 ( 2004 ) .
the mesoscopic magnetic order appears below about 0.3k .
small narrow contributions are also detected at the positions of the ( 111 ) intense nuclear bragg peak and at the ( 200 ) peak which corresponds to a nuclear extinction .
there are respectively attributed to an imperfect data correction and to a minority phase with @xmath0=0 propagation vector similar to that of tb@xmath2sn@xmath2o@xmath3 , and not considered in the analysis given below .
we first performed a symmetry analysis in the sg @xmath31 , with @xmath0= ( @xmath1 , @xmath1 , @xmath1 ) using the program basireps of the fullprof suite @xcite and the program sarah @xcite .
as already discussed for gd@xmath2ti@xmath2o@xmath3@xcite , the symmetry operations which keep @xmath0 invariant ( symmetry operations of the small group g@xmath34 ) split the four rare earth sites of the pyrochlore lattice in two orbits : the site tb@xmath35 with local anisotropy axis along @xmath0 remains invariant ( orbit 1 ) , whereas the three other sites tb@xmath36 are exchanged ( orbit 2 ) .
the decomposition of the magnetic representation @xmath37 for orbits 1 and 2 are : @xmath38 @xmath39 we show in table [ t : irrrep ] the irreducible representations ( ir ) and associated basis vectors ( bv ) from sarah output , which provides real basis vectors .
the four sites of a tetrahedron are tb@xmath35 ( 0,0,0 ) for the orbit 1 and tb@xmath2 @xmath40 , tb@xmath41 @xmath42 and tb@xmath24 @xmath43 , for the orbit 2 , which are exchanged by the symmetry operations ( xyz ) , ( zxy ) and ( yzx ) .
this table is similar to that given in ref . .
systematic refinements using the fullprof suite were performed to the data , for all possible combinations of the representations within the two orbits .
a general state was constructed by combining the basis vectors for a given representation .
good refinements were obtained with @xmath44 and @xmath45 for the orbit 1 and 2 respectively .
taking into account the strong ising anisotropy of the crystal field , we implemented additional constraints by constraining the tb moments to lie along their local @xmath46111@xmath14 anisotropy axes and to have equal magnitude for the two orbits . with these constraints ,
the best refinement found from the symmetry analysis corresponds to the magnetic structure called `` monopole - like '' in the following .
.[t : irrrep ] irreducible representations @xmath47 and associated basis vectors @xmath48 for the space group @xmath31 with @xmath0= ( @xmath1 , @xmath1 , @xmath1 ) , calculated using the program sarah @xcite .
the magnetic tb atoms are at the sites tb@xmath35 : @xmath49 , tb@xmath2 : @xmath40 , tb@xmath41 : @xmath42 and tb@xmath24 : @xmath43 . [ cols="<,^,^,^,^ " , ] by fitting the neutron patterns to the above structures , we determined the ordered magnetic moment amplitude and the correlation length relative to the mesoscopic magnetic order for all samples ( table 2 ) .
these two quantities are the only free parameters of the magnetic patterns , and they can be determined separately .
the amplitude of the ordered moment determines the peak intensities relative to the nuclear bragg peaks , whereas the domain size controls the peak width . in all samples ,
the fitted correlation length of the ( @xmath1 , @xmath1 , @xmath1 ) correlations is around 22(2 ) , close to the size of the cubic magnetic cell ( 20.3 ) .
the ordered moment is strongly enhanced for @xmath7=0 , in the region of the critical point @xmath50 , then decreases both in the spin liquid and ordered regions .
in addition , for the samples showing a peak of @xmath21 , a minute long range ordered ( lro ) moment of about 0.15 @xmath8 is detected . this lro spin component orders over a length scale of about 500 close to the resolution limit with the same magnetic structure . in all cases ,
the ordered moments remain well below the local tb moment of 5 @xmath8 estimated from the crystal field scheme @xcite at low temperature .
this means that in the ground state , a large component of the tb moment is correlated only over first neighbor distances , both in the spin liquid and in the ordered region of the phase diagram .
ti@xmath51o@xmath52 single crystal : elastic diffuse scattering maps in a ( hh0 , 00l ) plane .
the subtracted maps show the temperature evolution of the correlations .
a ) : between 4k and 500 mk , the diffuse scattering starts to be structured .
b ) : between 300 and 50mk , the correlations do not evolve.,width=302 ]
the above study focuses on mesoscopic ( @xmath1 , @xmath1 , @xmath1 ) correlations which emerge in tb@xmath2ti@xmath2o@xmath3 powder patterns below the exotic transition at t@xmath53 , situated in the range 0.3 - 0.4k , which presents some similarities with a spin glass transition @xcite .
our analysis provides a phenomenological quantitative description of these correlations , which involve only a fraction of the tb moment , amounting to 0.1 to 2 @xmath8 depending on the length scale , namely well below the local tb moment of 5 @xmath8 .
these correlations have been separated by subtracting a pattern above the transition .
these findings should now be confronted to recent studies in single crystals @xcite . in single crystals ,
the temperature variation of the diffuse scattering shows a gradual evolution from a fluctuating spin liquid state characterized by isotropic first neighbor correlations at high temperatures , to a precursor regime ( between t@xmath53 and 4k typically ) where the diffuse scattering starts to be structured and the elastic component increases steadily , then to a regime ( t@xmath46 t@xmath53 ) where the elastic signal is dominant , well structured and t - independent .
this temperature evolution is shown on the subtracted maps of fig .
[ figmapsubtracted ] .
we speculate that a similar evolution with temperature occurs in powders , so that the mesoscopic correlations can emerge .
the precursor regime is not obviously seen here on the powder data due to the low signal background ratio , but it was actually seen in tb@xmath2sn@xmath2o@xmath3 powder @xcite , where the mesoscopic order occurs on a much larger length scale and involves a much higher fraction of the tb moment . it has been previously argued @xcite that there are actually two transitions below 500mk , the magnetic transition at t@xmath53 occurring about 0.15 k below the @xmath21 anomaly .
this suggests two successive phenomena with decreasing temperature , related to quadrupolar and magnetic order respectively . in
the `` ordered '' samples above x@xmath54 , the onset of quadrupolar order seen by @xmath21 peak occurs at a higher temperature than the long range magnetic order , as frequently observed in rare earth intermetallics @xcite . in the `` spin liquid '' samples below x@xmath54 which show no anomaly of @xmath21 , the magnetic order would remain limited to short length scales .
ti@xmath51o@xmath52 : diffuse scattering maps in a ( hh0 , 00l ) plane at 0.07k .
( a ) : experimental map from ref . ; calculated maps assuming a single magnetic domain of volume 8@xmath55 with either spin - ice like ( b ) or monopole - like ( c ) structure ; ( d ) : average over all single magnetic domains with equal probability .
monopole and spin ice -like structures yield the same average map.,width=302 ] below t@xmath53 the elastic diffuse scattering measured on single crystals @xcite shows arm - like features along @xmath15111@xmath16 directions and butterfly shaped features centered at ( 002 ) and ( 220 ) .
these features result in diffuse maxima around ( @xmath1 , @xmath1 , @xmath1 ) positions and equivalent ( see fig . [ figmap]a ) .
moreover , the signal sharpens about certain wave vectors , defining pinch points in reciprocal space . in pyrochlores
, these pinch points typically result from the propagation of a local constraint governing the mutual orientations of the magnetic moments @xcite such as the `` two in / two out '' rule in spin ices .
the emergent gauge structure characterizes a coulomb phase @xcite .
et al _ have proposed that a similar ice rule should hold for tb@xmath2ti@xmath2o@xmath3 , spin components along the local @xmath15111@xmath16 axes being controlled by a `` two in / two out '' condition , and in the transverse direction by a `` 2-up/2-down '' condition @xcite . to compare our phenomenological description with these features , we have calculated diffuse scattering maps assuming cubic magnetic domains of size 2@xmath23 and the spin orientations determined above , for different domain orientations ( see the appendix for details ) .
we compare these maps to those measured in the ( hh0 , 00l ) plane by elastic neutron scattering @xcite at 0.06(1)k shown in figure [ figmap]a .
first , we notice that the butterfly features around ( 002 ) or ( 220 ) as well as the pinch points which appear between the diffuse maxima in the experimental map of fig [ figmap]a are not reproduced by these calculations .
next , for the spin - ice - like structure , a few calculated maps , such as the map of figure [ figmap]b or the maps k1s2 and k1s3 in the appendix , reproduce rather well the shape , extension and relative orientations of the diffuse maxima at ( @xmath1 , @xmath1 , @xmath1 ) and related positions . for the monopole - like structure ,
the agreement is not as good , whatever the domain orientation ( see an example in figure [ figmap]c ) .
averaging randomly over all domains orientations yields the same map for the two structures , shown in figure [ figmap]d , which is also in worse agreement with experiment .
this suggests that specific spin - ice like textures , involving magnetic domains of correlated orientations are preferred at the base temperature .
they may be generated through nanomagnetic twins , induced by the lowering of symmetry from the @xmath31 sg to the tetragonal or monoclinic sg s discussed above .
such nano twins exist in crystal structures @xcite but their extension to magnetism is beyond the purpose of this study . to situate this picture in the context of a local constraint , we notice that a local constraint supposes the existence of a manifold of degenerate configurations in the ground state . as temperature decreases ,
the system could select peculiar spin configurations out of this manifold .
the spin - ice - like and monopole - like structures determined above would belong to this manifold .
this selection could be favored by small off - stoichiometry such as minute tb substitution at the ti sites that would slightly perturb the energy scheme by introducing further neighbor interactions @xcite , local changes in the crystal field , random magneto - elastic couplings or local distortions @xcite .
these perturbations could reveal the basic interactions by inducing correlated regions around impurity sites , with the percolation of small long range ordered moment above a certain level of substitution .
similarly , a low fe or co substitution of a pd matrix reveals the high pd band susceptibility by generating giant moments around the fe / co impurity , with the onset of long range ferromagnetic order below one % substitution@xcite .
one could wonder if such energetic selection supports low energy propagative excitations .
this is the case in most fields of physics , where the stabilization of a mesoscopic order with static character allows exotic excitations to emerge .
for instance , the slowing down of thermal relaxations allows dispersive excitations in superparamagnetic particles @xcite , quantum tunneling in molecular magnets @xcite , reversible magnetization in spin glasses @xcite , or more trivially spin waves in classical magnets .
here we would expect excitations to stem from the @xmath0= ( @xmath1 , @xmath1 , @xmath1 ) positions .
propagative excitations have been actually observed in tb@xmath2ti@xmath2o@xmath3but they stem from the pinch points at @xmath0=(111 ) .
this qualitative difference suggests that these excitations are an intrinsic feature of the coulomb phase . in summary
, we have studied the mesoscopic order at play in tb@xmath4ti@xmath5o@xmath6 , which is tuned by small off - stoichiometry .
we find the same type of correlations in all samples , with two possible structures , monopole - like and spin ice - like , describing the powder averaged magnetic order .
magnetic moments of about 2 @xmath8 are correlated over micro domains of 20 size comparable to the cubic magnetic unit cell .
samples in the ordered region of the phase diagram involve in addition a very small long range ordered moment of less that 4% of the local tb moment ( @xmath56 5 @xmath8 ) .
comparison with single crystal maps suggests that in the ground state , the spin - ice like structures are preferred and that they stabilized through twinned magnetic nano domains .
our observations provide evidence that this mesoscopic order coexists in the ground state with a disordered coulomb phase , which involves the major part of the local tb moment .
their links remain however to be elucidated .
we would like to acknowledge fruitful discussions and useful advices from j. rodriguez - carvajal , a. wills , j. robert , a. gukasov , f. damay , and a. sazonov .
we thank m. j. p. gingras for a critical reading of the manuscript .
we are grateful to the ill for the use of neutron beam time and of the high - flux powder diffractometer d1b operated by the centre national de la recherche scientifique ( cnrs ) .
in order to generate the magnetic domains for a given magnetic structure ( either spin ice like or monopole like ) , we used the following procedure . for simplicity , we start from the cubic unit cell of lattice constant @xmath23 containing 16 tb ions . the monopole and spin ice structure are defined within this cubic cell .
we generate the k and s - domains within the sg fd-3 m . among the 48 symmetry operations of the space group fd-3 m , the first 24
are associated with rotation matrices r with det(r)=1 , and the 24 others deduced from them by an inversion center ( det r=-1 ) do not need to be considered . applying the 24 symmetry operations to the propagation vector @xmath0= ( @xmath1 , @xmath1 , @xmath1 ) splits them in 4 sub groups , which either leave @xmath0 invariant or transform @xmath0 in the @xmath0 vectors ( -@xmath1 , @xmath1 , @xmath1 ) , ( @xmath1 , -@xmath1 , @xmath1 ) , ( @xmath1 , @xmath1 , -@xmath1 ) of the star \{@xmath0}. these four subgroups are related to the @xmath0-domains .
for the little group g@xmath34 which leaves @xmath0 invariant , the symmetry operations apply both to atomic positions and atomic moments , using the usual rules : i ) classical spins transform as axial vectors under the symmetry operations ; ii ) for the symmetry operation which generate atomic positions out of the initial crystal cubic unit cell , equivalent atomic positions / magnetic moments in the cubic cell are obtained by appropriate translations @xmath57 , such as @xmath58 = @xmath59 exp ( 2i@xmath60 @xmath0 @xmath57 ) where @xmath59 and @xmath58 are the j moment of the initial cubic unit cell ( l=0 ) and at the distance r@xmath61 , respectively .
this procedure generates for each @xmath0-domain , 6 s - domains , each one corresponding to specific spin orientations of the tb moments within the cubic unit cell . for each domain ,
the structure factor and neutron intensity are calculated in the ( hh0 , 00l ) plane , by taking the fourier transform over a single magnetic cubic cell of size 2@xmath23 , which corresponds to the typical correlation length deduced from the powder data .
we then obtained 24 intensity maps , both for the spin ice - like and monopole - like structure .
figure [ figdomain ] shows these maps for the six s domains generated with g@xmath34 , where @xmath0= ( @xmath1 , @xmath1 , @xmath1 ) . in the monopole structure none of the 24 maps corresponds to the single crystal data at low temperature , whereas in the spin ice structure some of them such as the k1s2 and k1s3 shown in figure [ figdomain ] indeed bear some resemblance with the data .
performing an average over the intensities of the 24 domains with the same weight for each domain , yields the same intensity distribution for the spin ice and monopole structures ( figure [ figmap]d ) which does not well reproduce the experimental data .
it suggests that specific spin textures , associated with magnetic domains of correlated spin orientations , are actually stabilized .
this ground state mesoscopic texture could consist of antiferromagnetically ordered spin ice blocks , with correlated orientations of the staggered magnetization .
ti@xmath51o@xmath52 : diffuse scattering maps in the ( hh0 , 00l ) plane at 0.07k , calculated for the six s - domains corresponding to the propagation vector @xmath0= ( @xmath1 , @xmath1 , @xmath1 ) .
single magnetic domain of volume 8@xmath23@xmath62 with spin - ice like ( a ) and monopole - like ( b ) structures.,width=302 ] 99 c. castelnovo , r. moessner , s. l. sondhi , nature * 451 * , 42 , ( 2008 ) i. a. ryzhkin and m. a. ryzhkin jetp lett . * 93 * , 384 ( 2011 ) .
chang , s. onoda , y. su , y .- j .
tsuei , y. yasui , k. kakurai , and m. r. lees , nat .
commun . * 3 * , 992 ( 2012 ) .
s. guitteny , j. robert , p. bonville , j. ollivier , c. decorse , p. steffens , m. boehm , h. mutka , i. mirebeau , and s. petit phys . rev
. lett . * 111 * , 087201 ( 2013 ) . t. fennell , m. kenzelmann , b. roessli , h. mutka , j. ollivier , m. ruminy , u. stuhr , o. zaharko , l. bovo , a. cervellino , m.k .
haas , and r.j .. cava , phys .
lett . * 112 * , 017203 ( 2014 ) .
j. villain , z. phys .
b * 33 * , 31 ( 1979 ) . g. sala , m. j. gutmann , d. prabhakaran , d. pomaranski , c. mitchelitis , j. b. kycia , d. g. porter , c. castelnovo and j. p. goff , nature mat . * 13 * , 488 ( 2014 ) .
d. pomaranski , l. r. yaraskavitch , s. meng , k. a. ross , h. m. l. noad , h. a. dabkowska , b. d. gaulin and j. b. kycia , nature phys.*9 * , 353 ( 2013 ) .
c. paulsen , m. j. jackson , e. lhotel , b. canals , d. prabhakaran , k. matsuhira , s. r. giblin and s. t. bramwell , nature phys .
* 10 * , 135 ( 2014 ) .
r. m. dortenzio , h. a. dabkowska , s. r. dunsiger , b. d. gaulin , m. j. p. gingras , t. goko , j. b. kycia , l. liu , t. medina , t. j. munsie , d. pomaranski , k. a. ross , y. j. uemura , t. j. williams , and g. m. luke , phys rev b * 88 * , 134428 ( 2013 ) .
j. s. gardner , s. r. dunsiger , b. d. gaulin , m. j. p. gingras , j. e. greedan , r. f. kiefl , m. d. lumsden , w. a. macfarlane , n. p. raju , j. e. sonier , i. swainson , and z. tun , phys .
lett . * 82 * , 1012 ( 1999 ) .
t. taniguchi , h. kadowaki , h. takatsu , b. fak , j. ollivier , t. yamazaki , t. j. sato , h. yoshizawa , y. shimura , t. sakakibara , t. hong , k. goto , l. r. yaraskavitch , j. b. kycia , phys .
b * 87 * , 060408(r ) ( 2013 ) .
i. mirebeau , a. apetrei , j. rodriguez - carvajal , p. bonville , a. forget , d. colson , v. glazkov , j. p. sanchez , o. isnard , and e. suard , phys .
* 94 * , 246402 ( 2005 ) .
a. p. sazonov , a. gukasov , i. mirebeau , and p. bonville , phys .
b * 85 * , 214420 ( 2012 ) .
i. mirebeau , p. bonville , m. hennion , phys .
b * 76 * , 184436 ( 2007 ) .
fullprof , http://www.ill.eu/sites/fullprof .
a. s. wills , physica b * 276 - 278 * 680 , ( 2000 ) ; phys . rev .
b * 63 * , 064430 , ( 2001 ) ; program available from : http://www;ill.eu / sites / sarah .
j. d. m. champion , a. s. wills , t. fennell , s. t. bramwell , j. s. gardner , m. a. green , phys .
b * 64 * , 140407(r ) ( 2001 ) .
bilbao crystallographic server ; http://www.cryst.ehu.es/cgi-bin/cryst/programs/nph-wp-list d. b. litvin , magnetic space groups(2011 ) ; http://www.bk.psu.edu/faculty/litvin/download/the-magnetic-group-tables-j11.pdf ( 18.8.2012 ) .
r. alonard and p. morin j. mag . mag . mat . * 84 * , 255 ( 1990 ) .
l. kovarik _
et al _ , progress in material science * 54 * , 839 ( 2009 ) ; b. litvin and v. kopsky , on magnetic domain twin symmetry , ferroelectrics * 204 * , 217 ( 1997 ) .
l. thomas , f. lionti , r. ballou , d. gatteschi , r. sessoli and b. barbara nature ( london ) * 383 * 145 ( 1996 ) ; i. mirebeau , m. hennion , h. casalta , h. andres , h. u. gudel , a. v. irodova , and a. caneschi , phys .
lett * 83 * , 628 ( 1999 ) .
h. alloul , p. mendels , p. beauvillain , p. and c. chappert , europhys .
* 1 * 595 ( 1986 ) ; h. alloul , b. hennion , p. mendels .
j. de phys .
c8 * 49 * c8 - 1019 , ( 1988 ) . | we have studied the spin correlations with @xmath0= ( @xmath1 , @xmath1 , @xmath1 ) propagation vector which appear below 0.4 k in tb@xmath2ti@xmath2o@xmath3 spin liquid by combining powder neutron diffraction and specific heat on tb@xmath4ti@xmath5o@xmath6 samples with @xmath7=0 , 0.01 , -0.01 .
the @xmath0= ( @xmath1 , @xmath1 , @xmath1 ) order clearly appears on all neutron patterns by subtracting a pattern at 1.2(1)k .
refining the subtracted patterns at 0.07k yields two possible spin structures , with spin - ice - like and monopole - like correlations respectively .
mesoscopic correlations involve tb moments of 1 to 2 @xmath8 ordered on a length scale of about 20 .
in addition , long range order involving a small spin component of 0.1 to 0.2 @xmath8 is detected for the @xmath7= 0 and 0.01 samples showing a peak in the specific heat .
comparison with previous single crystals data suggests that the ( @xmath1 , @xmath1 , @xmath1 ) order settles in through nanometric spin textures with dominant spin ice character and correlated orientations , analogous to nanomagnetic twins .
rare earth pyrochlores with geometrical magnetic frustration are model compounds to investigate classical spin ice as well as quantum spin ice physics .
the short range ordered ground states with large degeneracy give rise to a large variety of interesting phenomena such as magnetic monopoles@xcite , first order transition @xcite , or unconventional magnetoelastic excitations @xcite .
these frustrated ground states are highly sensitive to small perturbations and to the presence of defects .
already in the 80s , villain @xcite predicted that for an arbitrary small concentration of defects the spinel lattice with pyrochlore structure could switch its ground state from spin liquid to spin glass .
more recently many studies focus on the role of a minute amount of defects in pyrochlores .
the crystal structure of r@xmath2ti@xmath2o@xmath3 pyrochlores ( where r@xmath9 is a rare earth ion , ti@xmath10 is non magnetic ) accommodates oxygen vacancies , small off - stoechiometries in the r@xmath9or ti@xmath10 lattice , `` stuffing '' or site inversion .
oxygen vacancies introduce ti@xmath9 defects instead of ti@xmath10 ions , modifying the r@xmath9 crystal field and inducing local distortions , with various consequences on the low temperature magnetic properties and spin ice physics . in classical spin ice dy@xmath2ti@xmath2o@xmath3 , oxygen depletion or a low level
dy stuffing in the ti sites strongly slows down the monopole dynamics at very low temperature @xcite and affects the non - equilibrium macroscopic properties @xcite . in the quantum spin ice yb@xmath2ti@xmath2o@xmath3 ,
small off - stoechiometry clearly affects the magnetic ground state @xcite .
it appears that precise studies of the magnetic defects are needed to resolve controversies concerning these unconventional ground states . among the pyrochlores , tb@xmath2ti@xmath2o@xmath3
is one of the most complex , the subject of numerous investigations since the discovery of its spin liquid character in 1999 .
the tb@xmath9 short range correlated moments fluctuate @xcite down to 20mk at the time scale of the muon probe ( 10@xmath11s ) .
the origin of these fluctuations has been highly debated @xcite .
several mechanisms have been proposed taking into account the non - kramers nature of the tb ion and the small energy gap between the ground and first excited crystal field doublets , which open the possibility of quantum spin ice behavior @xcite .
recent studies of the crystal and magnetic excitations using polarized neutrons show that the magneto - elastic coupling plays a crucial role , yielding an interaction between the first excited crystal field level and an acoustic phonon branch , resulting in a hybrid phonon - magnon mode @xcite .
the magnetoelastic coupling also induces original properties in the pyrochlore family , such as giant magnetostriction @xcite and very low thermal conductivity @xcite .
all these features suggest that the tb@xmath2ti@xmath2o@xmath3 ground state is determined by an interaction between quadrupolar moments , for which a dynamical jahn - teller distortion could be the first ( mean - field ) approximation @xcite . in tb@xmath2ti@xmath2o@xmath3
an exotic transition has been observed around 0.4k , as shown initially by anomalies of the specific heat @xcite .
below this temperature , spin fluctuations coexist with static spin correlations and spin glass - like irreversibilities @xcite . in single crystals ,
these correlations have been detected by an energy analysis of the neutron cross section @xcite , yielding an elastic contribution with diffuse maxima at the positions of half integer bragg peaks .
this intensity vanishes under a small magnetic field of about 0.1 t and shows some sample dependence , suggesting the influence of a minute amount of defects . in powders , a systematic study of tb@xmath4ti@xmath5o@xmath6 samples @xcite showed that the system could be tuned by minute changes of the tb content around a critical concentration @xmath12=-0.005 from a purely spin liquid state for @xmath7 @xmath13 @xmath12 to a state with partial antiferromagnetic order with @xmath0= ( @xmath1 , @xmath1 , @xmath1 ) propagation vector for @xmath7 @xmath14 @xmath12 .
the detailed nature of this @xmath0= ( @xmath1 , @xmath1 , @xmath1 ) order was not determined in ref . , but it seemed to extend over mesoscopic length scales , rather than the infinite length scale associated with symmetry breaking and bragg scattering . to investigate this magnetic order , we studied powder samples of tb@xmath4ti@xmath5o@xmath6 with a precise control of the stoechiometry @xmath7= -0.01 , 0 and 0.01 , by combining neutron diffraction and specific heat measurements . in all samples , by subtracting a pattern at 1.2(1)k , we observe magnetic peaks of lorentzian line shape , which can be indexed with the @xmath0= ( @xmath1 , @xmath1 , @xmath1 ) propagation vector .
these peaks correspond to the mesoscopic order discussed in the following .
its length scale deduced from the lorentzian line shape is around 20 , which corresponds to one magnetic cubic unit cell .
moreover , in the samples ( @xmath7=0 , 0.01 ) showing a transition in the specific heat , we also observe a bragg - like contribution , involving a minute tb moment ordered at a length scale of about 500 , close to the resolution limit .
both bragg - like and lorentzian contributions vanish around 0.4k . by refining the magnetic patterns at the lowest temperature ( 0.07k ) ,
we find two possible spin arrangements , fitting the data equally well . in both cases ,
the tb moments remain aligned along their local @xmath15111@xmath16 axes , therefore respect the ising character of the crystal field .
the first arrangement corresponds to magnetic moments in an ordered spin ice configuration @xcite , antiferromagnetically coupled along the @xmath15100@xmath16 cubic axes @xcite without violating the ice rules .
the second one is described by a piling along a @xmath17111@xmath18 $ ] axis of monopole and antimonopole layers @xcite separated by spin ice layers . comparing the calculated single crystal maps to
the experimental one suggests that spin ice - like configurations with non - random mutual orientations , describing a @xmath19nanomagnetic twin texture@xmath19 , are actually favored at @xmath20 .
the monopole - like structure could be also stabilized when heating .
we discuss the relationship between the mesoscopic order detected in the powder data and the diffuse scattering features ( anisotropic excitations , pinch points ) shown by single crystals .
ti@xmath2o@xmath3 .
neutron diffraction patterns measured at 0.07k and 1.2k , focusing on the region of the magnetic peaks , as shown by arrows . the full pattern at 0.07k is shown in the inset , with the refinement of the crystal structure yielding the scaling of the ordered magnetic moment .
, width=302 ] ti@xmath5o@xmath6 .
( a ) total specific heat @xmath21 versus temperature @xmath22 for several @xmath7 contents . in inset
the lattice constant of the cubic cell @xmath23 deduced from x - ray data at 300k ( red dots ) , in comparison with values from ref . revised in 2015 ( blue dots ) .
( b ) @xmath21 curves focusing on the low temperature region . in inset ,
the phase diagram of ref . where the arrows show the sample concentrations studied here .
, width=302 ] |
there has been a growing interest in photonic crystals , i. e. , structures with a periodically modulated dielectric constant .
the latter open new ways of manipulating electromagnetic wave emission and propagation processes @xcite .
in fact , there is a common belief that , in the near future , photonic crystals systems will allow us to perform many functions with light that ordinary crystals do with electrons @xcite .
they also promise to become a laboratory for testing fundamental processes involving interactions of radiation with matter under novel conditions . the basic principle behind all that is an analogy to an electron moving in a periodic potential which suggests that certain photon frequencies in a photonic crystal can also become forbidden , independent of photon polarization and the direction of propagation - a complete photonic bandgap ( cpbg ) .
however , unlike in an ideal single - electron picture , a photon always faces the possibility of being absorbed and it is , _ a priori _ not clear if theoretical proposals for manufacturing a cpbg and which are obtained in the absence of absorption @xcite can ever be realized experimentally .
naively , one expects that moderate absorption will only cause a slight perturbation of the band structure as confirmed in ref .
therefore , the neglect of absorption in band structure calculations seem to be justified for purely dielectric photonic structures @xcite . unfortunately , such structures exhibit only a limited dielectric contrast which is insufficient to open a cpbg below infrared wavelengths @xcite . a recent proposal @xcite to obtain such a cpbg involves metallo - dielectric structures .
since metals are usually rather absorbing , this again raises the question about the effect of absorption on band structure .
here we mention in passing that systems with gain , i.e. , where the imaginary part of the dielectric function @xmath1 has the opposite sign , are important for understanding laser action in new types of lasers involving photonic crystals @xcite .
although the study of absorption in photonic structures is of utmost importance , so far no rigorous results exist about fundamental matters such as the meaning of band structure and band gaps for absorptive systems . only recently
, using analytic continuation techniques , it has been shown that absorption turns bands into resonances in the lower - half complex plane @xcite .
literature on absorption in photonic crystals is rather sparse and , except for a recent article by yannopapas , modinos , and stefanou @xcite , only covers the cases of small absorption @xcite and small filling fraction ( @xmath2 ) of the absorptive component @xcite .
in order to investigate numerically the effect of considerable absorption on the band structure of photonic crystals , we started with studying a one - dimensional model consisting of a periodically layered ( or stratified ) medium , described by the dielectric function @xmath3 here @xmath4 is the lattice constant , i.e. , the length of the primitive cell , @xmath5 is an integer , and @xmath6 .
propagation of light at normal incidence in such a stratified medium is described by the one - dimensional periodic helmholtz equation @xmath7\,\psi({\bf x})=0 , \label{1dhe}\ ] ] where @xmath8 is the laplacian , @xmath9 the angular frequency , @xmath10 the speed of light in vacuum and @xmath11 an eigenfunction to be calculated .
another physical situation to which the one - dimensional periodic helmholtz equation applies ( [ 1dhe ] ) are grating - like structures @xcite , the so called one - dimensional ( 1d ) photonic crystals @xcite , and a number of different physical situations involving propagation of acoustic and elastic waves @xcite .
below a scatterer will be a region with @xmath12 neighboured by regions with @xmath13 . for notational simplicity we also set @xmath14 .
in the absence of absorption , the dispersion relation and the band structure are determined by the kkr equation , which in one dimension is identical to the kronig - penney equation @xcite @xmath15 where @xmath16 is the bloch momentum and @xmath17 is a phase shift @xcite .
the latter is determined by relations @xmath18 , \nonumber \\ \sin\eta_0 & = & c_0\left[-\sin \sigma r_s\cos\rho r_s + p \,\cos\sigma r_s
\sin\rho r_s \right ] , \nonumber \\ \cos\eta_1 & = & c_1\left[\sin \sigma r_s\sin\rho r_s + p \,\cos\sigma r_s
\cos\rho r_s \right ] , \nonumber \\ \sin\eta_1 & = & c_1\left[\cos \sigma r_s\sin\rho r_s - p\,\sin\sigma r_s \cos\rho r_s \right ] , \nonumber \\ c_l & = & \left\ { \begin{array}{cc } \left[1 + ( p^2 - 1)\,\sin^2\rho r_s\right]^{-1/2 } , & l=0,\\ \left [ 1 + ( p^2 - 1)\,\cos^2\rho r_s\right]^{-1/2 } , & l=1 , \end{array}\right . \nonumber \end{aligned}\ ] ] where @xmath19 , @xmath20 , and @xmath21 @xcite . equation ( [ kkreq ] ) still holds for absorbing scatterers but the ( frequency - dependent ) phase shifts @xmath17 are now complex . consequently , @xmath22 is complex and eq .
( [ kkreq ] ) has no real eigenvalues . in accordance with @xcite ,
the eigenvalues turn into resonances in the lower complex half - plane
. some characteristic features of the complex bands of resonances can already be seen in the long - wavelength limit ( @xmath23 ) , when eq .
( [ kkreq ] ) reduces to @xmath24 here @xmath25 is the _ effective dielectric constant_. if we denote by @xmath26 the absorptive material filling fraction , then @xmath27 i.e. , @xmath25 is nothing but the volume averaged dielectric constant . in the long - wavelength limit
, the first complex band contains an angle @xmath28 with the real positive axis which is determined by the relation @xmath29 in the limit @xmath30 , @xmath31 and hence @xmath32 . for @xmath33 and @xmath34 one
has @xmath35 down to such @xmath26 that @xmath36 and @xmath37 . in this range of @xmath26 ,
@xmath28 exhibits only a very weak dependence on @xmath26 .
surprisingly enough , as shown in fig .
1 , exact numerical calculations demonstrate that the angle @xmath38 remains very close to its value established in the long - wavelength limit often till the upper edge of the second complex band of resonances .
we determined the resonances numerically for an ensemble of 1d systems with i m @xmath39 varied between @xmath40 and @xmath41 for fixed @xmath42 and re @xmath43 .
we investigated how absorption alters the relative gap width @xmath44 , where @xmath45 and @xmath46 are the gap width and midgap frequency of the real part of @xmath9 , respectively .
the following lessons can be learned from an investigation of the first few bands of complex resonances for a representative sample of such 1d absorptive systems : 1 .
if ( i m @xmath39)/(re @xmath47 , the effect of absorption on the band gap is negligible ( both in shifting the gap edges and in changing gap width ) ( see figs . 2 and 3 ) 2 .
re @xmath48 remains a monotonically increasing ( decreasing ) function of the bloch momentum for an odd ( even ) band ( when labelled from the lowest one ) 3 .
absorption pushes band edges down ( see fig .
2 ) 4 . if i m @xmath39 increases , the corresponding @xmath49 tend to saturate rapidly beyond i m @xmath50 re @xmath39 ( see fig .
the dependence of the relative gap width @xmath49 on i m @xmath39 is monotonic up to i m @xmath51 re @xmath39 ( see fig .
absorption mostly increases the relative gap width @xmath49 , although in some rare cases it can also lead to a decrease of @xmath49 7 .
the largest effect was an increase of the @xmath49 by @xmath0 ( see fig .
3 ) and closing of the 2nd gap which , however , reappeared again as i m @xmath39 increased sufficiently enough 8 . even if the absorptive material filling fraction @xmath26 within the unit cell
is @xmath52 and i m @xmath39 is comparable to re @xmath39 , the imaginary part of the bloch frequency eigenvalues is @xmath53 .
actually , it has been quite surprising to find out that the imaginary part of the frequency eigenvalues can increase when the absorptive material filling fraction within the unit cell is lowered ! it is noteworthy to say that a band gap in re @xmath9 does not necessarily imply a very high reflectivity .
a study of finite layered structures reveals that in the presence of a strong absorption the reflectivity at the gap center can , for instance , saturate at @xmath54 or less , depending on a single bilayer absorption .
the latter yields a lower bound on the absorption of a composite structure .
the reflectivity at the gap center is usually more than twice that for frequencies within a band , almost exclusively as a result of decreased absorption within the gap which approaches that of a single bilayer . in both cases transmissions
are small .
if @xmath55 denotes the transmission for frequencies within a band , then the transmission within a gap is typically @xmath56 .
a typical system to which our calculation should apply is a polymer - semiconductor or polymer - insulator layered structure where the polymer absorption dominates .
our model calculations can also apply to a semiconductor - semiconductor or semiconductor - insulator heterostructure in which one component is doped by an increasing amount of an absorptive dye .
our choice of @xmath39 can also be regarded as that of a metal , or an artificial metal @xcite , for frequencies above the plasma wavelength .
to our best knowledge , our work is the first one to deal with the case of the photonic band structure in the presence of a strong absorption , when both @xmath57 and @xmath26 are large ( @xmath58 and @xmath59 ) ( cf .
@xcite ) . such a study of strong absorption in photonic structures is of utmost importance , both for understanding the behaviour of recently proposed metallo - dielectric structures with cpbg s below infrared wavelengths @xcite and for new types of lasers involving photonic crystals @xcite .
more realistic calculations involving dispersion of the dielectric constant and including transmission and reflection properties of finite layered structures will appear elsewhere .
this work is part of the research program by the stichting voor fundamenteel onderzoek der materie ( foundation for fundamental research on matter ) which was made possible by financial support from the nederlandse organisatie voor wetenschappelijk onderzoek ( netherlands organization for scientific research ) .
a. moroz , phys .
. lett . * 83 * , 5274 ( 1999 ) ; photonic crystals of coated metallic spheres , to appear in europhys .
h. van der lem and a. moroz , towards two - dimensional complete photonic - band - gap structures below infrared wavelengths , to appear in j. opt .
a : pure appl .
opt .
m. m. sigalas , c. m. soukoulis , c. t. chan , and k. m. ho , phys .
b * 49 * , 11 080 ( 1994 ) ; v. kuzmiak and a. a. maradudin , phys . rev .
b * 55 * , 7427 ( 1997 ) ; l .- m .
li , z .- q .
zhang , and x. zhang , phys .
b * 58 * , 15589 ( 1998 ) .
h. s. szer , j. w. haus , and r. inguva , phys .
b * 45 * , 13962 ( 1992 ) ; a. barra , d. cassagne , and c. jouanin , appl .
72 , 627 ( 1998 ) ; r. biswas , m. m. sigalas , g. subramania , and k .-
ho , phys . rev .
b * 57 * , 3701 ( 1998 ) ; a. moroz and c. sommers , j. phys . : condens .
matter * 11 * , 997 ( 1999 ) . | photonic band structure of metal - dielectric and semiconductor - dielectric layered structures are studied in the presence of a strong absorption .
it is shown that absorption can enlarge some gaps by as much as @xmath0 .
optical absorption and emission spectroscopy , superlattices , greens function method |
two of the fundamental parameters driving the evolution of galaxies across cosmic time is their total gas mass , and gas mass surface density .
the total gas mass of galaxies limits the total amount of stars that can form in a galaxy at any time , while its surface density is directly linked with the rate of star formation via the observed kennicutt - schmidt relation ( e.g. * ? ? ?
* ; * ? ? ?
the increase of the cosmic star formation rate ( sfr ) density up to redshifts 3 - 4 ( e.g. * ? ? ?
* ) is theorized to be caused by the increase in the cosmic gas mass density ( and therefore the average total gas mass of galaxies ) with increasing redshift ( as suggested by e.g. * ? ? ?
however , directly measuring the gas mass density out to these redshifts is a difficult task .
existing large area surveys of the 21 cm hi fine - structure line that trace the atomic gas have only measured the local universe ( @xmath2 , e.g. hipass , @xcite , and alfalfa , @xcite ) . in the near future , deeper surveys with the precursors for the square kilometer array ( ska ) such as dingo @xcite on the australian ska pathfinder ( askap ) will reach at most @xmath3 @xcite . however , it will not be until ska is available that hi will be observed in emission at @xmath4 , and even then it is expected to reach at most @xmath5 @xcite .
this leaves only hi absorption line studies , such as damped ly@xmath6 systems ( e.g. * ? ? ?
* ) , as the only determination of the evolution of the hi mass function over cosmic time . for molecular gas ,
surveys of large numbers of galaxies have been limited due to the difficulty of observing the transitions from the co molecule in the ( sub)mm .
the two largest efforts have been the fcrao extragalactic co survey @xcite , which measured the co@xmath7 line in 300 nearby galaxies , and the cold gass survey @xcite which measured fluxes in the co@xmath8 line for a purely mass - selected sample of @xmath9 galaxies at @xmath10 , matched in galex , arecibo , and sdss imaging .
the fcrao sample has recently been extended with co@xmath8 observations of a further 59 nearby galaxies as part of the herschel reference survey @xcite . at higher redshift
the co lines are still observable , but samples are smaller in number . while molecular gas measures were originally limited to very luminous , rare objects ( e.g. * ? ? ?
* ) , more typical galaxies are now being observed out to @xmath11 ( e.g. , egnog ; * ? ? ?
* ) , and farther out to @xmath12 ( e.g. , phibss ; * ? ? ?
* ) , with a full review of these surveys available in @xcite . with alma now operating , the samples of high - redshift galaxy molecular gas masses will only increase .
however , given its small instantaneous field - of - view , alma is a relatively slow survey instrument , and , while already providing impressive targeted observations , will not be suitable for very wide area blind co surveys at @xmath12 @xcite , though smaller blind searches have already been carried out in smaller regions ( e.g. , * ? ? ?
* ; * ? ? ?
another potential tracer of the total gas mass is the infrared emission arising from dust .
the amount of dust has long been known to be associated with gas column within the milky way ( e.g. * ? ? ?
* ) , and dust extinction has been used to map gas columns at higher dynamic ranges and spatial resolution than available from gas emission lines ( e.g. * ? ? ?
* ; * ? ? ?
. however , in extragalactic studies the lower physical resolution mixes lines of sight , making internal extinction a poor tracer of the gas column ( e.g. @xcite , and see also @xcite ) , and relying on the rare occultation of relatively bright background sources is not feasible for large extragalactic studies .
however , the dust continuum emission in the infrared
submillimeter range is a direct tracer of the total dust opacity and , if a mean grain emissivity is assumed , the dust column .
the idea of using the sub - mm continuum emission to measure the mass of the ism has existed for a while , having been suggested by @xcite , and used by @xcite in nearby galaxies .
as an independent tracer of the gas mass , the dust continuum has also been directly compared with co observations in nearby galaxies to constrain the conversion factor between co and gas mass ( i.e. the `` x '' factor , @xcite , @xcite , @xcite , @xcite , and the review by @xcite ) .
direct comparisons of the sub - mm continuum and gas content in galaxies have also been explored in @xcite , @xcite , and @xcite , and this correlation has already been used to estimate the total gas masses in higher redshift galaxies by @xcite and @xcite . yet
the dust continuum emission is not a simple direct tracer of the gas mass .
the dust emission is dependent upon the temperature of the dust grains , the conversion to dust column depends upon knowledge of the grain emissivity , and finally , the conversion of dust column to gas mass depends upon the dust to gas ratio . modelling
the ir spectral energy distribution typically involves the use of both physically - based models for the dust grains and reasonable assumptions for the heating of the grains to determine the total dust column directly from the ir emission ( see e.g. * ? ? ?
this approach has obtained reasonable results in many nearby galaxies ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . however , systematic uncertainties lie in both the dust models and the assumptions on the heating radiation field ( and the parameters in each ) , as well as the fitting procedure itself , introducing another level of complexity in determining the interstellar gas mass .
the recent work of @xcite demonstrated that using a simple modified - blackbody model for the dust emission and a constant conversion factor from dust to gas are not unreasonable assumptions for the determination of total gas mass .
they found a good correlation of the herschel - based dust masses with total gas masses determined from hi and co data in ten nearby galaxies , with an estimated 25% error on the dust method .
however , systematic uncertainties were found to dominate this method as well , with @xcite demonstrating the sensitivity of the determined dust mass to the fitted dust temperature , and the sensitivity of this dust temperature to both the assumed emissivity ( as demonstrated in detail in * ? ? ?
* ) and the observed wavelengths fitted . we can bypass these systematic issues by comparing directly the far - ir to submm continuum emission to the various tracers of the gas phases directly .
we can then determine calibrations linking the monochromatic ir emission and gas tracers by exploring a wide range of objects , which also enable the exploration of dependencies on other physical parameters like gas - phase metallicity .
a direct comparison of the sub - mm emission with gas tracers has already been explored somewhat by @xcite , who used a sample of 20 galaxies within the _ herschel_-atlas survey with existing hi and new co data .
they found that the sub - millimeter fluxes appear to be strongly associated with the diffuse atomic and molecular gas phases , and that the fir / co luminosity ratio decreases with increasing luminosity .
however , even though several ir fluxes were explored , the range in the sub - mm luminosities of the galaxies was small .
the most detailed examination to date has been by @xcite , who used both local luminous ir galaxies and higher redshift submillimeter galaxies with 850@xmath13 fluxes to calibrate the ratio of sub - millimeter luminosity to the ism mass assuming a constant emissivity to mass ratio .
they found that the ism mass and 850@xmath13 luminosity correlated within a factor of two for their galaxy sample , and then used this correlation to determine the ism masses in stacked submillimeter alma observations of high redshift galaxies .
however their analysis was focussed on more luminous sources and may not hold for all galaxies . to improve upon these surveys we use the most comprehensive nearby galaxy surveys available : the kingfish @xcite , things @xcite , and heracles @xcite surveys for the ir , hi , and co data , respectively .
the galaxies in these surveys cover a wide range in metallicity , stellar mass , and star formation rates , and with all these parameters measured in a consistent manner from the associated ancillary datasets .
it thus represents one of the largest homogenous datasets for co and hi in nearby galaxies , and covers the full ir spectral energy distribution ( sed ) from 70 to 500@xmath13 .
in addition , the galaxies in the sample are near enough that they can be resolved and radial trends can be examined as well .
this enables us to determine the correlation of sub - mm emission with gas mass over a wide range of galaxy parameters , and explore the limiting conditions where the sub - mm emission can be used . in the following section we introduce the surveys and the galaxy sample , followed by a description of the theory in section [ sec : irgaseq ] , and explore a direct comparison of the integrated ir emission to the total gas masses of galaxies in section [ sec : irgascorr ] .
we explore the issues of internal variations and radial trends in section [ sec : radial ] , and summarize our findings in section [ sec : summary ] .
our galaxy sample arises from the intersection of three complementary large programs surveying nearby galaxies ; kingfish@xcite , things @xcite , and heracles @xcite . these programs provide us with tracers of both the integrated gas and dust masses of galaxies , and resolved maps down to @xmath14kpc scales extending well beyond the optical radii for the hi and ir data .
the matched datasets result in 36 galaxies , ranging from dwarf galaxies to massive spirals , with the names and galaxy properties listed in table [ tab : galaxies ] .
all values were taken from @xcite and references therein except for the metallicity which is from @xcite when available .
the sample is dominated by spiral galaxies and irregulars .
the stellar masses and star formation rates both cover approximately 4 orders of magnitude , and approximately 2 orders of magnitude in specific star formation rates ( ssfr@xmath15 ) .
lccccccccc ddo053 & 08:34:07 & @xmath1666:10:54 & 3.61 & 132 & 1.5 & 1.3 & 7.98 & 6.35 & 0.0060 + ddo154 & 12:54:05 & @xmath1627:08:59 & 4.3 & 50 & 3.0 & 2.2 & 8.02 & 6.63 & 0.0020 + hoi & 09:40:32 & @xmath1671:10:56 & 3.9 & 0 & 3.6 & 3.0 & 8.04 & 6.87 & 0.0040 + hoii & 08:19:05 & @xmath1670:43:12 & 3.05 & 15 & 7.9 & 6.3 & 8.13 & 7.59 & 0.036 + ic2574 & 10:28:23 & @xmath1668:24:44 & 3.79 & 56 & 13.2 & 5.4 & 8.23 & 8.2 & 0.057 + m81dwb & 10:05:31 & @xmath1670:21:52 & 3.6 & 140 & 0.9 & 0.6 & 8.2 & 6.36 & 0.0010 + ngc337 & 00:59:50 & @xmath1707:34:41 & 19.3 & 130 & 2.9 & 1.8 & 8.84 & 9.32 & 1.3 + ngc628 & 01:36:42 & @xmath1615:47:00 & 7.2 & 25 & 10.5 & 9.5 & 8.88 & 9.56 & 0.68 + ngc925 & 02:27:17 & @xmath1633:34:45 & 9.12 & 107 & 10.5 & 5.9 & 8.73 & 9.49 & 0.54 + ngc2146 & 06:18:38 & @xmath1678:21:25 & 17.2 & 123 & 6.0 & 3.4 & @xmath18 & 10.3 & 7.94 + ngc2798 & 09:17:23 & @xmath1641:59:59 & 25.8 & 160 & 2.6 & 1.0 & 9.04 & 10.04 & 3.38 + ngc2841 & 09:22:03 & @xmath1650:58:35 & 14.1 & 153 & 8.1 & 3.5 & 9.19 & 10.17 & 2.45 + ngc2976 & 09:47:15 & @xmath1667:54:59 & 3.55 & 155 & 5.9 & 2.7 & 8.98 & 8.96 & 0.082 + ngc3077 & 10:03:19 & @xmath1668:44:02 & 3.83 & 45 & 5.4 & 4.5 & @xmath18 & 9.34 & 0.094 + ngc3184 & 10:18:17 & @xmath1641:25:28 & 11.7 & 135 & 7.4 & 6.9 & 9.07 & 9.5 & 0.66 + ngc3198 & 10:19:55 & @xmath1645:32:59 & 14.1 & 35 & 8.5 & 3.3 & 8.78 & 9.83 & 1.01 + ngc3351 & 10:43:58 & @xmath1611:42:14 & 9.33 & 12 & 7.4 & 5.0 & 9.21 & 10.24 & 0.58 + ngc3521 & 11:05:49 & @xmath1700:02:09 & 11.2 & 160 & 11.0 & 5.1 & 9.06 & 10.69 & 1.95 + ngc3627 & 11:20:15 & @xmath1612:59:30 & 9.38 & 173 & 9.1 & 4.2 & 8.99 & 10.49 & 1.7 + ngc3938 & 11:52:49 & @xmath1644:07:15 & 17.9 & 28 & 5.4 & 4.9 & 9.06 & 9.46 & 1.77 + ngc4236 & 12:16:42 & @xmath1669:27:45 & 4.45 & 162 & 21.9 & 7.2 & 8.74 & 8.36 & 0.13 + ngc4254 & 12:18:50 & @xmath1614:24:59 & 14.4 & 23 & 5.4 & 4.7 & 9.08 & 9.56 & 3.92 + ngc4321 & 12:22:55 & @xmath1615:49:21 & 14.3 & 30 & 7.4 & 6.3 & 9.12 & 10.3 & 2.61 + ngc4536 & 12:34:27 & @xmath1602:11:17 & 14.5 & 130 & 7.6 & 3.2 & 9.0 & 9.44 & 2.17 + ngc4569 & 12:36:50 & @xmath1613:09:46 & 9.86 & 23 & 9.5 & 4.4 & 9.26 & 10.0 & 0.29 + ngc4579 & 12:37:44 & @xmath1611:49:05 & 16.4 & 95 & 5.9 & 4.7 & 9.22 & 10.02 & 1.1 + ngc4625 & 12:41:53 & @xmath1641:16:26 & 9.3 & 27 & 2.2 & 1.9 & 9.05 & 8.72 & 0.052 + ngc4631 & 12:42:08 & @xmath1632:32:29 & 7.62 & 86 & 15.5 & 2.7 & 8.75 & 9.76 & 1.7 + ngc4725 & 12:50:27 & @xmath1625:30:03 & 11.9 & 35 & 10.7 & 7.6 & 9.1 & 10.52 & 0.44 + ngc4736 & 12:50:53 & @xmath1641:07:14 & 4.66 & 116 & 11.2 & 9.1 & 9.04 & 10.34 & 0.38 + ngc5055 & 13:15:49 & @xmath1642:01:45 & 7.94 & 102 & 12.6 & 7.2 & 9.11 & 10.55 & 1.04 + ngc5457 & 14:03:13 & @xmath1654:20:57 & 6.7 & 39 & 28.8 & 26.0 & @xmath18 & 9.98 & 2.33 + ngc5474 & 14:05:02 & @xmath1653:39:44 & 6.8 & 97 & 4.8 & 4.3 & 8.83 & 8.7 & 0.091 + ngc5713 & 14:40:12 & @xmath1700:17:20 & 21.4 & 11 & 2.8 & 2.5 & 9.03 & 10.07 & 2.52 + ngc6946 & 20:34:52 & @xmath1660:09:14 & 6.8 & 63 & 11.5 & 9.8 & 8.99 & 9.96 & 7.12 + ngc7331 & 22:37:04 & @xmath1634:24:56 & 14.5 & 168 & 10.5 & 3.7 & 9.05 & 10.56 & 2.74
+ the infrared data we use in this work are all from the kingfish sample ( key insights into nearby galaxies ; a far infrared survey with herschel ; * ? ? ?
* ) , an imaging and spectroscopic survey of 61 nearby ( @xmath19mpc ) galaxies with the _ herschel space observatory _ @xcite .
this survey provides imaging in 6 ir bands ; 70 , 100 , and 160@xmath13 with pacs @xcite , and 250 , 350 , and 500@xmath13 with spire @xcite , with the observing strategy described in detail in @xcite .
all bands ( pacs and spire ) are processed from level 1 using scanamorphos v17.0 @xcite .
the pacs data are calibrated using flight model 6 , however the 160@xmath13 band has a uniform correction factor of 0.925 applied to it to correct for the pacs distortion flatfield not included in the original scanamorphos reduction . as the effect is of the order of 1% for the other pacs bands , well below the photometric uncertainties , no correction is applied to these bands .
the spire bands assume factors of 97.7 , 55.0 , and 26.0 to convert from jy beam@xmath20 to mjy sr@xmath20 for the spire 250@xmath13 , 350@xmath13 , and 500@xmath13 bands , respectively .
lcccccc ddo053 & @xmath21 & @xmath22 & @xmath23 & @xmath24 & @xmath25 & @xmath26 + ddo154 & @xmath27 & @xmath28 & @xmath29 & @xmath30 & @xmath31 & @xmath32 + hoi & @xmath33 & @xmath34 & @xmath35 & @xmath36 & @xmath37 & @xmath38 + hoii & @xmath39 & @xmath40 & @xmath41 & @xmath42 & @xmath43 & @xmath44 + ic2574 & @xmath45 & @xmath46 & @xmath47 & @xmath48 & @xmath49 & @xmath50 + m81dwb & @xmath51 & @xmath52 & @xmath53 & @xmath54 & @xmath55 & @xmath56 + ngc337 & @xmath57 & @xmath58 & @xmath59 & @xmath60 & @xmath61 & @xmath62 + ngc628 & @xmath63 & @xmath64 & @xmath65 & @xmath66 & @xmath67 & @xmath68 + ngc925 & @xmath69 & @xmath70 & @xmath71 & @xmath72 & @xmath73 & @xmath74 + ngc2146 & @xmath75 & @xmath76 & @xmath77 & @xmath78 & @xmath79 & @xmath80 + ngc2798 & @xmath81 & @xmath82 & @xmath83 & @xmath84 & @xmath85 & @xmath86 + ngc2841 & @xmath87 & @xmath88 & @xmath89 & @xmath90 & @xmath91 & @xmath92 + ngc2976 & @xmath93 & @xmath94 & @xmath95 & @xmath96 & @xmath97 & @xmath98 + ngc3077 & @xmath99 & @xmath100 & @xmath101 & @xmath102 & @xmath103 & @xmath104 + ngc3184 & @xmath105 & @xmath106 & @xmath107 & @xmath108 & @xmath109 & @xmath110 + ngc3198 & @xmath111 & @xmath112 & @xmath113 & @xmath114 & @xmath115 & @xmath116 + ngc3351 & @xmath117 & @xmath118 & @xmath119 & @xmath120 & @xmath121 & @xmath122 + ngc3521 & @xmath123 & @xmath124 & @xmath125 & @xmath126 & @xmath127 & @xmath128 + ngc3627 & @xmath129 & @xmath130 & @xmath131 & @xmath132 & @xmath133 & @xmath134 + ngc3938 & @xmath135 & @xmath136 & @xmath137 & @xmath138 & @xmath139 & @xmath140 + ngc4236 & @xmath141 & @xmath142 & @xmath143 & @xmath144 & @xmath145 & @xmath146 + ngc4254 & @xmath147 & @xmath148 & @xmath149 & @xmath150 & @xmath151 & @xmath152 + ngc4321 & @xmath153 & @xmath154 & @xmath155 & @xmath156 & @xmath157 & @xmath158 + ngc4536 & @xmath159 & @xmath160 & @xmath161 & @xmath162 & @xmath163 & @xmath164 + ngc4569 & @xmath165 & @xmath166 & @xmath167 & @xmath168 & @xmath169 & @xmath170 + ngc4579 & @xmath171 & @xmath172 & @xmath173 & @xmath174 & @xmath175 & @xmath176 + ngc4625 & @xmath177 & @xmath178 & @xmath179 & @xmath180 & @xmath181 & @xmath182 + ngc4631 & @xmath183 & @xmath184 & @xmath185 & @xmath186 & @xmath187 & @xmath188 + ngc4725 & @xmath189 & @xmath190 & @xmath191 & @xmath192 & @xmath193 & @xmath194 + ngc4736 & @xmath195 & @xmath196 & @xmath197 & @xmath198 & @xmath199 & @xmath200 + ngc5055 & @xmath201 & @xmath202 & @xmath203 & @xmath204 & @xmath205 & @xmath206 + ngc5457 & @xmath207 & @xmath208 & @xmath209 & @xmath210 & @xmath211 & @xmath212 + ngc5474 & @xmath213 & @xmath214 & @xmath215 & @xmath216 & @xmath217 & @xmath218 + ngc5713 & @xmath219 & @xmath220 & @xmath221 & @xmath222 & @xmath223 & @xmath224 + ngc6946 & @xmath225 & @xmath226 & @xmath227 & @xmath228 & @xmath229 & @xmath230 + ngc7331 & @xmath231 & @xmath232 & @xmath233 & @xmath234 & @xmath235 & @xmath236 + to measure the fluxes of each galaxy , we first convolve all bands to a beam size of @xmath237 , the resolution of the spire 500@xmath13 beam ( the lowest resolution band ) , using the convolution kernels of @xcite .
we then use the galaxy centres listed in table [ tab : galaxies ] and elliptical apertures as used by @xcite to determine the total integrated galaxy flux .
these elliptical apertures have the same position angles and axis ratios as listed in table [ tab : galaxies ] , and encompass the full optical and ir light , and typically extend to @xmath238 optical radii .
the background for all images was assumed to be flat and was determined as in @xcite , by sampling empty regions around our elliptical apertures and calculating the background mean and uncertainty , as well as the pixel noise . to determine the luminosities we assume the distances given in table [ tab : galaxies ] .
the total fluxes and uncertainties for all herschel bands for each galaxy are given in table [ tab : fluxes ] .
these fluxes differ slightly from those in @xcite as updated beam sizes were assumed for all bands leading to different correction factors , and the convolution to the spire 500@xmath13 beam size also affected results ( though predominantly with the resolved investigations in section [ sec : radial ] ) .
in addition to the ir photometry , as the successor to the sings survey @xcite , a large ancillary dataset exists for the kingfish galaxies , including determinations of their metallicity , stellar masses and star formation rates , as shown in table [ tab : galaxies ] .
the metallicities of our sample are determined in @xcite , using the calibration of @xcite , and cover a range of 2 dex .
different metallicity calibrations , such as ( * ? ? ?
* pt05 ) , will introduce systematic offsets from the calibration used here ( as demonstrated for several calibrations by @xcite , and as can be seen for the kingfish galaxies , specifically in table 1 in @xcite ) . to determine the neutral gas mass of our galaxy sample we use the things hi sample of nearby galaxies @xcite , with the addition of the leroy et al .
( in prep . )
sample that extends the things sample to a larger number of nearby galaxies .
lccc ddo053 & @xmath239 & @xmath240 & @xmath241 + ddo154 & @xmath242 & @xmath243 & @xmath244 + hoi & @xmath245 & @xmath246 & @xmath247 + hoii & @xmath248 & @xmath249 & @xmath241 + ic2574 & @xmath250 & @xmath251 & @xmath252 + m81dwb & @xmath253 & @xmath254 & @xmath255 + ngc337 & @xmath256 & @xmath257 & @xmath258 + ngc628 & @xmath259 & @xmath260 & @xmath261 + ngc925 & @xmath262 & @xmath263 & @xmath264 + ngc2146 & @xmath265 & @xmath266 & @xmath267 + ngc2798 & @xmath268 & @xmath269 & @xmath270 + ngc2841 & @xmath271 & @xmath272 & @xmath273 + ngc2976 & @xmath274 & @xmath275 & @xmath276 + ngc3077 & @xmath277 & @xmath278 & @xmath279 + ngc3184 & @xmath280 & @xmath281 & @xmath282 + ngc3198 & @xmath283 & @xmath284 & @xmath285 + ngc3351 & @xmath286 & @xmath287 & @xmath288 + ngc3521 & @xmath289 & @xmath290 & @xmath291 + ngc3627 & @xmath292 & @xmath293 & @xmath294 + ngc3938 & @xmath295 & @xmath296 & @xmath297 + ngc4236 & @xmath298 & @xmath299 & @xmath300 + ngc4254 & @xmath301 & @xmath302 & @xmath303 + ngc4321 & @xmath304 & @xmath305 & @xmath306 + ngc4536 & @xmath307 & @xmath308 & @xmath309 + ngc4569 & @xmath310 & @xmath311 & @xmath312 + ngc4579 & @xmath313 & @xmath314 & @xmath315 + ngc4625 & @xmath316 & @xmath317 & @xmath318 + ngc4631 & @xmath319 & @xmath320 & @xmath321 + ngc4725 & @xmath322 & @xmath323 & @xmath324 + ngc4736 & @xmath325 & @xmath326 & @xmath327 + ngc5055 & @xmath328 & @xmath329 & @xmath330 + ngc5457 & @xmath331 & @xmath332 & @xmath333 + ngc5474 & @xmath334 & @xmath335 & @xmath336 + ngc5713 & @xmath337 & @xmath338 & @xmath339 + ngc6946 & @xmath340 & @xmath341 & @xmath342 + ngc7331 & @xmath343 & @xmath344 & @xmath345 + to determine the hi masses of the galaxies , we use the integrated hi moment 0 maps with robust weighting ( see * ? ? ?
* for details ) . to first convert the hi maps from their units of jybeam@xmath346 to m@xmath347
, we multiply by ( based on equations 1 and 5 from * ? ? ?
* ) , @xmath348/(b_{\rm maj}\times b_{\rm min}),\ ] ] where @xmath349 and @xmath350 are the beam major and minor axes in arcseconds , respectively .
@xmath351 is the galaxy distance in mpc ( from table [ tab : galaxies ] ) .
we then convolve the maps using first gaussians to circularize the beam shapes and from there to the spire 500@xmath13 resolution using the convolution kernels of @xcite .
the integrated hi masses are then determined using the same apertures as for the infrared fluxes and are given in table [ tab : masses ] with uncertainties and spire 500@xmath13 luminosities .
note that for some galaxies the hi emission extends beyond the chosen aperture that encompasses the optical and ir emission , which we discuss further on .
for the molecular gas masses of our galaxy sample we use the heracles co(2 - 1 ) survey of nearby galaxies @xcite , using the integrated moment 0 maps . to determine the total co luminosity we first convolve the moment 0 maps using the convolution kernels of @xcite , and then integrate using the same apertures as for the infrared and hi maps . if these apertures extended further than the available moment 0 maps , we assumed the flux in these regions was negligible and therefore 0 . to convert the co luminosity to molecular gas mass we used equation 3 from @xcite , @xmath352= \alpha_{\rm
co}r_{21}l_{\rm co}[{\rm k\,km\,s^{-1 } pc^{2}}].\ ] ] we assume a fixed conversion factor from the co(2 - 1 ) luminosity to h@xmath0 mass for all galaxies , based on the milky way conversion factor , @xmath353 m@xmath354 @xcite , and a fixed line ratio of co(2 - 1 ) to co(1 - 0 ) of @xmath355 ( the typical value found for the heracles galaxies in * ? ? ?
* though correctly this ratio is dependent upon the mean gas properties such as temperature ) .
the final h@xmath0 mass for each galaxy is given in table [ tab : masses ] .
the total gas mass for all galaxies is then taken simply as the sum of the hi and molecular gas mass within our chosen apertures .
given that our galaxies cover over 2 orders of magnitude in stellar mass and gas - phase metallicity , an assumption of a constant co luminosity to molecular gas mass conversion factor may lead to biases , especially as different conversion factors have already been determined for several of the galaxies in our sample ( see e.g. * ? ? ?
* ; * ? ? ?
we discuss this point further in the text , but this assumption must be considered when using any of our determined relationships .
the use of the ir emission to trace the total amount of gas is premised on the simple assumption that gas and dust are always associated . assuming that the dust has a single temperature , @xmath356 , the ir luminosity of a galaxy at a frequency , @xmath357 , is linked to the total gas mass of a galaxy m@xmath358 via the equation , @xmath359 where @xmath360 is the dust to gas mass ratio , @xmath361 is the planck blackbody function , and @xmath362 is the dust emissivity at the frequency @xmath357 .
a broad range of ir colors is seen within and between galaxies indicating variation of the mean dust temperature ( see , e.g. * ? ? ?
* ) , driven by the variation of radiation field strengths experienced by dust in galaxies . however , we can limit the impact of these temperature distributions on the above equation in two ways ; firstly , for most of the kingfish galaxies , it is possible to reproduce their ir sed by having the bulk of the dust heated by a single diffuse radiation field , with only a small fraction being heated by radiation fields stronger than this ( see , e.g. * ? ? ?
* ; * ? ? ?
* ) , thus assuming a single temperature within galaxies for the longer wavelengths is a reasonable assumption .
secondly , by concentrating on longer wavelengths that are well past the peak of the ir sed ( at @xmath363@xmath13 , see e.g. * ? ? ?
* ) , we move closer to the rayleigh - jeans tail of the black body radiation which is only linearly sensitive to temperature .
* particularly their figure 7 ) and ( * ? ? ?
* particularly their figure 12 ) have both shown that the assumption of a single temperature is reasonable , finding the temperatures of single modified blackbody fits to the integrated 100 - 500@xmath13 ir seds of galaxies in the herschel reference survey and herschel virgo cluster survey , respectively , lying within a factor of 2 ( @xmath364k and @xmath365k , respectively , with early type galaxies having higher temperatures ) .
such a small range of temperatures result in only a factor of @xmath142 uncertainty in the dust mass estimates and hence a factor of 2 in the gas mass determined from the ir in the linear regime .
the dust - to - gas mass ratio and dust emissivity at a given wavelength are somewhat degenerate , with the determination of one relying on an estimate of the other . while estimates exist for these quantities in the local group , it is uncertain how they will vary across different galaxies .
the emissivity is dependent on the composition of dust @xcite , and there have been suggestions that the emissivity will vary in molecular environments . the emissivity has also been suggested to vary with metallicity , such as seen in the lmc by @xcite , however @xcite found no strong trends when comparing more massive galaxies with dwarf galaxies . however , if it is assumed that emissivity at longer wavelengths does not change between galaxies , a clear trend of a decreasing dust - to - gas ratio with the metallicity of the galaxy is seen across the local group @xcite and nearby galaxies @xcite , and even if the emissivity is left free , a lower dust - to - gas ratio is seen in low metallicity dwarf galaxies @xcite .
this suggests that the dust to gas ratio must be accounted for in any empirical calibration , as we discuss below .
although equation [ eqn : dustlum ] indicates the direct link of gas mass and dust emission via the association of dust and gas , ir emission will also be indirectly associated with gas mass , or more precisely gas mass surface density via the kennicutt - schmidt relation ( ks relation , see e.g. * ? ? ?
an increase in the gas mass surface density has been empirically demonstrated to correlate on average with an increase in the star formation rate surface density on kpc scales ( see e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . given this increase in the sfr surface density , the interstellar radiation field that heats the dust
is also expected to be higher , leading in turn to higher average dust temperatures and correspondingly higher ir continuum emission .
this secondary correlation may actually limit the dispersive effect that dust temperature has on the gas mass ir emission connection , and may even lead to a tighter correlation where the ir luminosity is more responsive to gas mass ( i.e. where an increase in gas mass leads to both an increase in dust mass and increase in dust temperature due to the increased sfr , and thus a greater increase in ir luminosity ) .
however , unlike the correlation of dust mass and gas mass , the ks relation will act non - linearly on the ir emission . the shortest dust wavelengths that measure the hot dust and peak of the ir emission will see the greatest impact of the increase in dust temperature , and thus also be more sensitive to any variation in the ks relation . as a first comparison , we determine the simple correlations between the ir luminosities and total gas mass in galaxies . in table [ tab : pearson ] , we show the pearson correlation coefficient ( & spearman rank coefficient ) for all _ herschel _ bands against the total gas mass .
a strong correlation is observed with all bands , with the strongest correlation existing between the longest wavelength ( i.e. spire 500@xmath13 ) and the total gas mass .
this correlation is clearly seen in figure [ fig:500_mgas ] , where we plot the spire500 luminosity ( @xmath366 ) against the total gas mass .
.correlation coefficients for @xmath367 against individual herschel bands .
[ cols= " < , > , > " , ] the strong correlations seen with each band suggest that it is entirely reasonable to use broad - band ir photometry to estimate the total gas mass of galaxies from the ir emission .
based on the high pearson correlation coefficients , we performed robust linear fits to each _ herschel _ band luminosity versus the total gas mass , of the form , @xmath368band , where band is the log of the luminosity in the given band ( in log@xmath369 ) .
the coefficients @xmath370 and @xmath371 are given in table [ tab : linearfitgas ] for each band .
the fit for the spire500m@xmath358 relation is shown by the solid line on figure [ fig:500_mgas ] . in table
[ tab : linearfitgas ] , we also give the dispersion , @xmath372 , of the galaxies around each relation , which , as suggested by the correlation coefficients , decreases with increasing wavelength .
to compare with previous comparisons of dust and gas correlations , we have also limited the sample to the more massive objects , with log(m@xmath373 .
the fits to these galaxies show both a smaller dispersion and steeper slopes than the full sample .
the spire 500 fit is the tightest , with only a 30% dispersion , similar to that found by @xcite in their comparison of gas masses with dust masses determined from fits to the ir sed in 10 nearby galaxies . given that the full extent of a galaxy in the ir may not always be available , we have also repeated the fits using smaller apertures that are truncated at an optical radius , @xmath374 , using the values for the major and minor axes presented in table [ tab : galaxies ] .
these apertures are typically 30% smaller than the ir - encompassing apertures from @xcite that we use . however , as the ir light is generally concentrated to the optical disk , these smaller apertures contain a significant fraction ( @xmath375% ) of the ir light for most galaxies . only a few galaxies with clearly extended ir emission , mostly the smaller galaxies such as ngc3077 ,
have significantly reduced ir fluxes due to the smaller aperture .
similarly , most of the molecular gas is totally contained within an optical radius for all galaxies .
yet the atomic gas can extend well beyond an optical radius , and is observed at greater than @xmath376 in some galaxies .
truncating at @xmath374 reduces the total gas mass because of this extended emission ( this is discussed further in [ sec : radial ] ) .
we have performed these fits only for the three spire bands , as these show the strongest correlations in table [ tab : pearson ] , and are the most sensitive to extended emission ( and thus show the greatest change with reduced apertures ) .
all fits to the reduced apertures , both to the full and non - dwarf galaxy sub - sample , find slightly flatter slopes than the original apertures , leading also to a larger intercept .
this trend is driven by the dwarf galaxies , which have relatively larger hi gas disks , such that the smaller aperture preferentially reduces the gas mass relative to the ir , flattening the overall slope . thus biasing
an aperture to measure the ir luminosity of galaxy will tend to underestimate the total gas mass , approximately by 15% based on our galaxy sample .
lrrr|rrr & & all & & & log(m@xmath377 & + band ( log@xmath369 & @xmath370 & @xmath371 & @xmath372 & @xmath370 & @xmath371 & @xmath372 + pacs70 & 4.15 & 0.55 & 0.30 & 4.6 & 0.50 & 0.27 + pacs100 & 3.62 & 0.60 & 0.27 & 3.27 & 0.63 & 0.23 + pacs160 & 3.52 & 0.61 & 0.26 & 1.91 & 0.78 & 0.18 + spire250 & 3.17 & 0.69 & 0.25 & 1.17 & 0.90 & 0.15 + spire350 & 3.08 & 0.74 & 0.24 & 1.17 & 0.90 & 0.14 + spire500 & 3.19 & 0.78 & 0.23 & 1.44 & 0.99 & 0.10 + lrrr|rrr & & all & & & log(m@xmath377 & + band ( log@xmath369 & @xmath370 & @xmath371 & @xmath372 & @xmath370 & @xmath371 & @xmath372 + spire250 & 3.73 & 0.64 & 0.28 & 1.57 & 0.86 & 0.14 + spire350 & 3.63 & 0.68 & 0.25 & 1.49 & 0.92 & 0.10 + spire500 & 3.74 & 0.72 & 0.23 & 1.72 & 0.96 & 0.09 + we find that none of the fits are linear , neither the total gas masses and the sample limited within @xmath374 , nor the massive galaxy sub - samples in both , with all slopes less than 1 .
as discussed in the previous section , three factors may be playing a role in the non - linear slopes : the systematic variation of the dust temperature , the dust - to - gas ratio , and the variation of the co luminosity to molecular gas mass ( @xmath378 ) .
as the slope becomes flatter at shorter wavelengths , dust temperature must play at least some role . the sub - linear trend observed in the correlation at all wavelengths suggests that the kennicutt - schmidt relation may also be contributing to the correlation , that is higher gas masses lead to a higher sfrs , which in turn heats the dust to higher average temperatures .
the dependence upon the co conversion factor is a major source of uncertainty in these fits , and remains a general uncertainty in determining the total gas masses in galaxies .
our assumed value of @xmath353 m@xmath354 is the suggested value of the milky way conversion factor given in the extensive review of @xcite .
however , the analysis of @xmath378 in some of the same galaxies as our sample by @xcite found a slightly lower value ( @xmath379 ) , and a large scatter between , and within , the galaxies .
changing the value of @xmath378 we assume significantly affects only the more massive spirals ( m@xmath380 ) , as the dwarf galaxies are all hi dominated , as can be seen from table [ tab : masses ] .
increasing / decreasing @xmath378 by 0.7 dex ( larger than the dispersion between galaxies found by * ? ? ?
* ) leads to a steepening / flattening of the determined linear fits .
however , when only the galaxies with m@xmath380 are examined , the slope remains approximately constant , with a systematic shift up and down in the relation , scaling with our scaling of @xmath378 .
however , given the work of @xcite and review by @xcite , there is no a priori reason why our normal spiral galaxy sample should have a different @xmath378 to that assumed in our work .
however , there are two further issues with the simple linear fits revealed by figure [ fig:500_mgas ] ; a large part of the correlation of gas mass with ir luminosity is simply galaxy mass scaling , with galaxies with the highest stellar masses lying towards the top right , as indicated by the color of the points .
secondly , the lower mass galaxies tend to lie above the relation , leading to a much greater scatter , and limit the determination of the gas mass . in figure
[ fig : l500mgas_l500 ] , we remove the mass weighting intrinsic to a luminosity - luminosity plot by plotting the ratio m@xmath381 against @xmath382 , which also emphasizes the scatter of the low - mass galaxies .
the color scale in this figure reveals the average metallicity of the galaxy , as measured by @xcite ( using the @xcite conversion ) .
the similarity in colors between the two plots demonstrates the mass - metallicity relation . in this figure , it is clear that the more massive normal galaxies show a much more linear relation , with the offset caused by the low - mass , low - metallicity galaxies in the sample . in table
[ tab : linearfitgas ] we also include a fit to only the galaxies with log(m@xmath383 ( non - dwarf galaxies , hereafter `` normal '' galaxies ) , meaning a @xmath384 using the @xcite mass - metallicity relation ) .
this is shown as the dashed line in figures [ fig:500_mgas ] and [ fig : l500mgas_l500 ] which emphasizes the difference to the full - sample fit .
as clear from these figures and table [ tab : linearfitgas ] , the dispersion of the normal galaxies around the normal galaxy - only fit is reduced relative to that of that seen around the full - galaxy relation . the largest decrease in dispersion appears to be for the longest wavelengths , with the spire500 relation decreasing by a factor of 2 , while the dispersion at the shortest wavelengths only decreases slightly .
as the relation for the spire500 and m@xmath358 is almost linear for normal galaxies , we can naively assume a 1:1 relation .
in this case , we find for the normal kingfish galaxies ( m@xmath385m@xmath386 ) ; @xmath387 = 28.5 l_{500 } [ l_{\odot } ] , \ ] ] where @xmath388 is the spire 500@xmath13 luminosity ( i.e. @xmath366 ) . assuming a linear correlation at 500@xmath13 only increases the dispersion slightly to 0.118 dex ( i.e. most gas masses within 30% ) .
one issue with the linear relations we present is that they are only monochromatic , and that these rest - frame wavelengths may not be available at higher redshifts when observed using for example herschel or alma . in these cases
we suggest a simple linear interpolation of the slope , @xmath371 , and constant , @xmath370 in table [ tab : linearfitgas ] .
the slope is monotonic , at least within the range of wavelengths explored here , and so a simple linear interpolation should suffice . for the constant
, there appears to be a minimum at 350@xmath13 , however , again a linear interpolation between the bounding wavelengths should be sufficient and lead to only a small uncertainty ( as seen by the range of @xmath370 ) , except at the shortest wavelengths . for continuum measurements outside the range explored by herschel , an extrapolation based upon a simple modified blackbody can be applied ( as done for equation 10 in * ? ? ?
the offset observed at low stellar masses in figure [ fig : l500mgas_l500 ] is clearly associated with a decline in the gas phase metallicity .
given this , the most likely cause for this offset is the variation of the dust - to - gas ratio with metallicity .
naively , we would expect the dust - to - gas ratio ( dgr ) to at least linearly decrease with metallicity due to the lack of metals from which to form , with recent works suggesting that a linear dgr to metallicity relation is a reasonable assumption ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* though the latter work suggests this is true only for @xmath389log(o / h)@xmath390 ) . in figure
[ fig : l500mgas_corrected ] , we have used the measured metallicity of our galaxies from @xcite to `` correct '' for a linear variation of the dgr by scaling the m@xmath391 ratio by o / h . with this correction included , the relationship between m@xmath391 and @xmath388 is basically flat ( a fitted slope of @xmath392 ) , and a dispersion of only @xmath393 , much less than the original fits to all galaxies .
thus , over 4 orders of magnitude in ir luminosity , the total gas mass can be determined from the 500@xmath13 luminosity within @xmath1440% , as long as the effects of metallicity on the dgr can be corrected for .
one issue in correcting for the variable dgr is the difficulty in obtaining the gas - phase metallicity of a galaxy at higher redshift .
when such information is not available , using the mass - metallicity relation , presuming a stellar mass for the galaxy has been determined , provides one possible solution ( see e.g. , * ? ? ?
* ; * ? ? ?
the metallicity of the ism also affects the relation of dust and gas through the co luminosity to molecular gas mass conversion factor @xmath378 . for the lowest metallicity galaxies
, @xmath378 may increase by an order of magnitude or more from the mw value assumed here because of the presence of `` co dark '' molecular gas ( see , e.g. * ? ? ?
* ; * ? ? ?
* for the theory and for examples and discussion in the local universe )
. however , as mentioned in the previous section , the ` low - mass galaxies ' in our sample all have molecular gas fractions less than 0.01 ( given our assumed mw conversion factor , as seen in figure [ fig : fmol_irratio ] in the following section ) , meaning that the contribution of molecular to the total gas mass in these objects will always be relatively small .
this effect will act to increase the offset we observe in figure [ fig : l500mgas_l500 ] , but will contribute less than a factor of two ( 0.3 dex ) to any relation . while the correction for metallicity clearly works for the m@xmath394 relation at long wavelengths , at shorter wavelengths
the effect of dust temperature is still an issue , as suggested by the flattening of the slopes in table [ tab : linearfitgas ] .
figure [ fig : l250mgas_corrected ] shows the relation of the spire250 luminosity with m@xmath358 corrected for the variable dust - to - gas ratio , as in figure [ fig : l500mgas_corrected ] , with both the negative slope and increased scatter compared to the l@xmath395 relation are noticeable .
the color of the points indicates the pacs160/spire500 color , which we use as a proxy for the dust temperature , where a lower value indicates warmer average dust temperatures as the peak of the ir shifts to wavelengths shorter than 160@xmath13 .
note how at high masses ( high l@xmath396 ) , a higher m@xmath397 is seen with decreasing 160/250 ratio , indicative of warmer temperatures .
however , with the m@xmath398 plot ( figure [ fig : l500mgas_corrected ] ) no such gradient is seen with the pacs160/spire500 ratio , demonstrating the lack of sensitivity of the longer wavelength emission to dust temperature .
this suggests , not unexpectedly , that to obtain a more accurate gas mass at shorter wavelengths , at least two bands must be observed to correct for dust temperature .
however , given that the ir continuum at or around 500@xmath13 ( 600 ghz ) is observable at essentially all redshifts up to @xmath399 using existing alma bands ( and longer using bands 1 & 2 ) , and no sensitivity to the ir colors is seen in the scatter at these wavelengths , we suggest using longer rest - frame wavelengths when possible to determine the total gas mass . given that the star formation rate surface density in nearby galaxies appears to correlate better with the molecular gas surface density than the total gas density ( e.g. * ? ? ?
* ) , determining the molecular gas mass at higher redshift without the necessity of measuring a possibly weak line would be useful .
when we compare the molecular gas mass ( or , more precisely , co(2 - 1 ) luminosity ) to the infrared luminosities for our full sample of galaxies we find the wavelengths at the peak of the ir sed ( 100 and 160@xmath13 ) show the tightest correlation , with table [ tab : h2pearson ] listing the pearson correlation coefficient and spearman rank coefficient for each _ herschel _ band for the correlation .
we find this is true even when we limit the sample to the normal galaxies .
this is somewhat contrary to @xcite , who found the highest correlation of the molecular gas mass with the longest wavelengths .
lrr|rr & & + band ( log@xmath369 & pearson & spearman & pearson & spearman + pacs70 & 0.957 & 0.929 & 0.905 & 0.890 + pacs100 & 0.979 & 0.955 & 0.946 & 0.925 + pacs160 & 0.977 & 0.969 & 0.959 & 0.949 + spire250 & 0.980 & 0.958 & 0.951 & 0.932 + spire350 & 0.974 & 0.936 & 0.933 & 0.895 + spire500 & 0.964 & 0.901 & 0.908 & 0.836 + this is surprising given that molecular gas would be expected to be cold , and therefore mostly associated with cold dust .
however a closer examination reveals that this is exactly the case , as demonstrated in figure [ fig : fmol_irratio ] . in this figure
we plot the molecular gas mass fraction , m@xmath400 , as a function of the dust temperature as represented by the pacs160/spire500 flux ratio ( a higher ratio indicates colder dust as the ir peak shifts to 160@xmath13 ) . given that the denominators correlate , as shown in figure [ fig:500_mgas ]
, this removes any mass scaling , and acts to highlight the correlation between the 160@xmath13 luminosity and molecular gas mass indicated in table [ tab : h2pearson ] .
this figure reveals that as the dust gets colder ( higher 160@xmath13/500@xmath13 ) the molecular gas fraction increases , with both these quantities correlated with the galaxy stellar mass ( as shown by the color scale in figure [ fig : fmol_irratio ] ) , or similarly the average gas metallicity .
the evolving shape of the ir sed with the metallicity of the galaxy can be seen clearly in figure 8 in @xcite , demonstrating the warmer average dust temperatures in the lower metallicity objects .
the warmer temperatures in these galaxies are likely caused by the harder and stronger interstellar radiation fields ( isrf ) .
the isrf is harder due to the lower stellar metallicity in these galaxies , while the isrf tends to be stronger due to the lower dust density ( given the lower dust to gas ratio ) leading to decreased dust shielding and a longer mean free path of uv photons @xcite .
this lower dust shielding will lead to a lower molecular gas fraction due to increased h@xmath0 dissociation , and a relatively higher dissociation of the co molecule , and thus an increased @xmath378 factor . as we assume a constant @xmath378 factor , an increasing @xmath378 with metallicity ( due to the presence of co dark molecular gas , * ? ? ?
* ) , will also cause the decrease in the molecular gas fraction that we observe ( if a variable @xmath378 is assumed , it is uncertain how much the molecular gas fraction will change with metallicity , see e.g. * ? ? ? * specifically their figure 3 ) .
once the correlation of these ratios with stellar mass is accounted for , no association with other galaxy parameters such as sfr or ssfr is seen ( note however that these parameters correlate with stellar mass in our galaxy sample , as seen in sdss galaxies , e.g. * ? ? ?
this is surprising , given that the 160/500 ratio traces dust temperature , and that molecular gas is associated with star formation .
however , it may be that smaller scale effects not seen in the integrated quantities , such as the spatial distribution of gas , dust and stars , is affecting the correlation to an extent that 2nd order effects are not seen . in general , we find the best power - law fit to be : @xmath401 = -2.86 + 1.22 \log ( \nu{\rm l}_{\nu}(160\mu m)/{\rm l}_{\odot})\ ] ] with a dispersion of @xmath402 around this relation .
however once a galaxy is molecular rich , it is uncertain how well the 160@xmath13 luminosity will trace m@xmath403 . as can be seen in figure [ fig : fmol_irratio ] , at high molecular fractions a broad range of 160@xmath13/500@xmath13 colors are possible .
thus for the most massive objects , the 500@xmath13 luminosity is a better tracer of the molecular gas simply because m@xmath404 .
similarly , given the uncertain effects of dust heating on this relation , extrapolating this to higher redshifts may be problematic .
the sample of galaxies we explore here are typical local galaxies , and cover a range of luminosities and stellar masses ( as seen in table [ tab : galaxies ] ) , however these galaxies may not be representative of more luminous ir galaxies typically seen in sub - mm surveys .
@xcite explore a slightly higher mass range than our sample , though with a significant overlap . for our full sample we find much flatter slopes than @xcite found for the ir versus hi , co(2 - 1 ) and co(3 - 2 ) luminosities .
however for our more massive `` normal '' galaxy sample that overlaps the range observed in @xcite , our slopes appear very similar to those determined by ( * ? ? ? * specifically their figure 4 ) , at least for hi and co(2 - 1 ) , the lines we use for our total gas mass .
however , while they do see trends with stellar mass and metallicity ( such as the decrease in l@xmath405/m@xmath406 with mass and metallicity ) , their dynamic range is not large enough to determine the relations we see here .
@xcite explored the correlation of sub - mm luminosity with ism mass in a sample of sub - mm galaxies at @xmath407 , in addition to a local sample that overlaps our sample with 850@xmath13 detections , and a planck observation of our galaxy . assuming that all the gas was molecular and a similar conversion factor ( @xmath408 )
, @xcite find a ratio of l@xmath409/m@xmath410 ergs@xmath20hz@xmath20m@xmath411 for the sub - mm galaxies , and similar values for the local galaxies ( for which the hi mass was also included ) .
rearranging equation [ eqn : mgasl500 ] , we find a ratio at 500@xmath13 of @xmath412 ergs@xmath20hz@xmath20m@xmath411 for our normal galaxy sample , which , assuming a dust emissivity with a power - law of 1.8 , corresponds to @xmath413 ergs@xmath20hz@xmath20m@xmath411 , in good agreement with the sub - mm galaxy sample of @xcite .
however , applying a single sub - mm emission to gas mass conversion factor , @xmath414 , to all galaxies removes any dependence on other galaxy properties .
we find that for normal galaxies in the local universe , an assumption of a linear relation between sub - mm luminosity and gas mass is reasonable .
given the range of ir luminosities explored in our work and in @xcite , it suggests that , while dust temperatures will affect the sub - mm luminosity to gas mass relationship , the variation in temperature is small enough that it has only a small impact in the rayleigh - jeans regime and thus the relationship .
conversely , there appears to be a significant metallicity evolution over cosmic time ( see e.g. * ? ? ?
therefore to apply the observed l@xmath415m@xmath358 relation to all galaxies at higher redshifts , it is crucial that this evolution of the metallicity be accounted for .
this can be done through a direct correction for an inferred metallicity , either directly measured or determined from a mass - metallicity relation , or through allowing the relation ( or @xmath6 ) to evolve with redshift using the metallicity evolution as a basis for the evolving dgr .
an evolving dgr was allowed for in the work of @xcite , who calculated the total gas mass from ir observations in a sample of `` normal '' high - redshift galaxies ( @xmath416 ) . instead of using a direct correlation of sub - mm luminosity and gas mass
, they first calculated the dust mass using model fits to their observed ir seds ( such as the * ? ? ?
* model ) , and then converted to the total gas content using an empirical relationship between the dgr and metallicity determined locally @xcite . while this two - step approach may introduce systematic errors , @xcite find their determined gas masses consistent with the co luminosities ( assuming milky way conversion factors that also vary with metallicity , and assuming the gas is mostly molecular ) .
an example of the application of using ir fluxes to determine the evolution of gas mass is the work of @xcite , who stacked the ir fluxes of galaxies with similar stellar masses , star formation rates and redshifts , to create representative ir seds .
they then fit these seds using dust models , including simple modified blackbodies , to determine the dust masses of these objects , which they converted to a gas mass allowing for a variable dust - to - gas ratio dependent upon stellar mass and sfr . using these
they find a strong correlation in the gas fractions of galaxies with their stellar mass and sfrs , and find strong evolution in both the gas fraction and star formation efficiency ( sfr / m@xmath358 ) with redshift .
in the previous section , we only considered the galaxies as integrated objects , with a single gas mass and luminosity .
however , the kingfish sample provides much more detail than this , and the sub - arcsecond capabilities of alma will mean , even at high - redshift , the continuum emission from galaxies can potentially be resolved ( see , e.g. * ? ? ?
thus to explore the gas
l@xmath395 relation further , we break the galaxies into elliptical annuli using the position angles and axis ratios of the galaxies given in table [ tab : galaxies ] .
the annuli are chosen to be in steps of 2kpc along the major axis in figure [ fig : annulisgasi500 ] we plot the ratio of the gas mass surface density and spire500 surface brightness for all annuli in our galaxy sample that satisfy the criteria that the annuli was larger than the resolution of the spire500@xmath13 beam ( 36 ) .
the colors of the data indicate the position of the annuli in terms of the optical radius @xmath374 of each galaxy , as given in table [ tab : galaxies ] .
this figure demonstrates that within galaxies the radial profiles follow the same pattern as seen for the integrated luminosities ( i.e. as in figure [ fig : l500mgas_l500 ] ) , albeit with a larger scatter .
some of this scatter arises from faint background sources at large radii which have not been masked . as with the luminous galaxies
, we see that for high surface brightness regions @xmath417/i@xmath395 is linear , but when we move to lower i@xmath395 we find an increasing trend of @xmath417/i@xmath395 .
this trend is affected by both the intrinsic metallicities of the galaxies as well as the radial trends within galaxies , with a higher @xmath417/i@xmath395 ratio observed at larger radii .
this is emphasized by the horizontal lines which indicate the median @xmath417/i@xmath395 ratio for annuli between @xmath418 ( blue ) , @xmath419 ( purple ) , and @xmath420 ( red ) .
this radial @xmath417/i@xmath395 gradient is likely associated with the radial metallicity gradient within each galaxy , and when the observations are corrected for the metallicity gradient determined by @xcite for some of our sample , this gradient is reduced .
however the gradient is not totally removed , and has a large scatter introduced by the uncertainties in the metallicity gradient ( see * ? ? ? * for a discussion on these uncertainties ) . yet using the longer wavelength continuum emission is still a robust tracer of the total gas mass of galaxies , and the gas surface density for most of a galaxy .
the robustness of the sub - mm continuum can be seen in figure [ fig : ngc628radial ] , where we compare the radial variation of the spire500 mean surface brightness , @xmath421 , against the mean gas mass surface density , @xmath422 , in ngc628 .
using the radial determinations of the hi and h@xmath0 gas mass surface densities from @xcite , who stacked the heracles co observations to detect the low column density molecular gas at large radii , we can determine both the radial variation of the total gas mass and the molecular gas fraction in our sample .
using the same 15 elliptical annuli as @xcite , we have determined the radial variation of the spire500 surface brightness ( note that 15 under samples the 36 spire 500 beam by a factor of 3 ) . for illustration purposes , we have scaled @xmath421 by a factor of 25 to be on the same scale as the total gas mass surface density .
it is clear that the ir traces better the total gas mass surface density instead of either the h@xmath0 or hi gas alone in the inner radii .
however , at radii beyond 0.7 @xmath374 there is an increasing deviation of the ir brightness relative to the gas mass density .
a similar trend is seen for some of the other galaxies in our matched sample , such as ngc925 .
however the trend is not seen in all galaxies , with resolution and inclination affecting the observed trends in some objects ( see the attached figures in the appendix ) .
the point where the sub - mm surface brightness and gas mass density separate does not appear to directly correlate with the radius at which the gas becomes mostly atomic , emphasizing that the dust is not just tracing the molecular gas .
this conclusion is further supported by galaxies which are predominantly atomic at all radii .
neither does this separation appear to be related to the surface mass density of ngc628 becoming gas dominated . in figure
[ fig : ngc628radial ] we also indicate the stellar surface density determined from spitzer 3.6@xmath13 data by querejeta et al .
( in prep . ) using the data from the s4 g survey @xcite , arbitrarily scaled by 0.1 to be on the same scale as the gas mass density .
the disk of ngc628 becomes gas dominated beyond the radius at which the ir deviates from @xmath422 ( at @xmath423 ) .
the deviation of the sub - mm emission from the gas is more clearly seen in figure [ fig : ngc628gdr ] , where we plot the ratio of mean gas mass surface density to 500@xmath13 surface brightness as a function of radius for ngc628 . at the inner radii of the galaxy ,
the @xmath424 ratio is flat and then increases in slope beyond @xmath425 . to demonstrate that this change is significant
, we have shaded the data to indicate the fraction of pixels that are significant ( @xmath426 ) , with darker points indicating a larger fraction .
this shading is conservative as we are integrating in annuli , meaning that we are more sensitive to diffuse emission at larger radii .
even given this conservative estimate , it is clear that the measured values of @xmath424 ratio are significant out to @xmath423 .
one possibility for this deviation is that we are seeing a change in the dust to gas ratio at these radii , possibly related to a variation in the radial metallicity gradient at the larger radii .
however , @xcite measure gas - phase abundances out to @xmath427 for several of our galaxies , including ngc628 ( see their figure 7 ) , and no such changes in the abundance gradients are seen .
the expected variation of the gas to dust ratio is indicated by the inverse of the metallicity gradient determined by @xcite , using both the @xcite or @xcite calibrations , as shown by the dashed and solid lines , respectively , on figure [ fig : ngc628gdr ] .
these lines are both normalized to the value of @xmath424 at @xmath428 to emphasize the similarity ( or difference ) of the determined metallicity slopes compared to the observed variation of the @xmath424 ratio .
however , the lines also demonstrate the issues in interpreting the metallicity gradient .
the @xcite calibration follows reasonably well the observed @xmath424 ratio up to @xmath425 , beyond which a deviation from this slope is clearly visible . yet
the @xcite calibration shows a much steeper slope , and , while reasonable below @xmath429 , over - estimates the observed ratio between 0.4 and 0.8 @xmath374 , and no deviation of the @xmath424 ratio relative to this gradient is seen at larger radii . which gradient is more representative of the true gas - phase abundance gradient is uncertain , though @xcite find a metallicity in between the two calibrations shown here at @xmath430 using temperature insensitive metallicity calibration based on fir fine structure lines , suggesting that the gradient may lay between these lines .
the @xmath424 radial variations for the other galaxies in our sample are included in the online - only appendix .
for all galaxies this ratio lies between @xmath431 , with the ratio becoming larger at larger radii .
several galaxies also show a turn up in @xmath424 within @xmath432 , such as ngc6946 .
the increase in the @xmath424 in these galaxy centres is likely associated with the decreased central @xmath378 values observed by @xcite in several of our galaxies , including ngc6946 .
a central decrease in @xmath378 would mean that we overestimate @xmath422 at the center of our galaxies due to our assumption of a constant milky - way value , and hence over - estimate the @xmath424 ratio . for galaxies for which it is available ,
we have also plotted the inverse of the determined metallicity gradients as in figure [ fig : ngc628gdr ] . for several galaxies ( e.g. ngc3198 )
the metallicity gradient is sufficient to explain the observed radial trend in @xmath424 , with no significant deviations seen .
for some galaxies , i.e. ngc4254 and ngc4321 , the metallicity gradients are much steeper than the observed @xmath424 ratio . given the observed gradients in gas and stellar surface mass densities , it is possible that the metallicity gradients are much flatter than have been determined for these objects . finally for several galaxies the same excess at large radii relative to the metallicity gradient as in ngc628
is seen ( e.g. ngc925 and ngc6946 ) . for these galaxies which show a radial deviation in the @xmath424 ratio , like ngc628 and ngc925
, another possibile explanation is that the different conditions in the outskirts of the galaxies lead to different dust properties which result in different emissivities .
such a variation in the emissivity with radius for some of the kingfish galaxies has already been suggested by @xcite using sub - mm data and in m33 by @xcite .
however , at least some part of this deviation at large radii must arise from the decrease in the mean interstellar radiation field available to heat the dust at these radii , as demonstrated by the radial decrease in stellar density relative to gas density in figure [ fig : ngc628radial ] .
a weaker interstellar radiation field will lead to lower average dust temperatures and thus correspondingly weaker ir emission .
this can be seen in figure [ fig : ngc628radial_s250 ] , where we plot the same quantities as in figure [ fig : ngc628radial ] for ngc628 , but now with the shorter wavelength spire250@xmath13 band ( note the different scaling factor ) . while it is clear in this figure that a change in the ratio of gas mass to ir luminosity occurs at a similar radius , the ratio decreases much more rapidly .
this implies a change in the ir slope , and thus either a decrease in dust temperature or a flattening of the dust power - law emissivity .
a decreasing dust temperature would lead to a lower ir luminosity even without a variation of the dust to gas ratio . in any case
, it is clear from figure [ fig : ngc628radial ] and the other galaxies in the appendix that for the outskirts of many galaxies ( beyond @xmath433 ) , the ir emission no longer traces the total gas mass in the same manner . for the galaxies in our sample ,
the total gas mass is dominated by the atomic gas at these radii . for many of our galaxies a significant fraction of hi is found outside the optical radius ( a median of 40% for our sample ) , and for several objects emission can be found even outside 2@xmath374 , with an obvious ( and unrepresentative ) example being ngc3077 with its extended tidal streams @xcite . by defining our apertures purely on the extent of the optical and ir emission , it is obvious that such extended gas disks will not be accounted for in the above relations , with the relations missing a large fraction of the total gas mass in some galaxies , like the hi rich dwarfs like ddo154 .
however it is still an open question how much relevance this extended hi gas has on the galaxy as a whole , with all of the molecular gas and most of the star formation inside our chosen `` total '' galaxy apertures and it being uncertain whether this gas external to the galaxy will participate at all in future star formation events within the galaxy .
the evolution of the gas mass density over cosmic time plays a direct role in the evolution of galaxies through its link to the star formation rate . yet
measuring the gas mass directly of a large number of galaxies at high redshifts through line observations is technically difficult , and time consuming . as dust and gas
are intimately associated , the dust infrared continuum may provide a more feasible way to determine the total gas mass of galaxies at high redshift . to demonstrate the use of the ir continuum as a gas tracer we have compared the total gas mass and ir luminosity of a well - studied sample of 36 nearby galaxies ( average distance 10mpc ) observed with the kingfish , things , and heracles surveys .
these galaxies sample the peak of the mass function , with stellar masses from @xmath434 to @xmath435 , and , unlike the local merger - induced ulirgs , these galaxies have disk - like morphologies that are likely more representative of the main - sequence galaxies at high redshift observed in current and future deep sub - mm surveys . we find a strong correlation between the total gas mass and ir luminosity in all herschel bands , with the strongest and tightest correlation found for the longest wavelength ( spire 500@xmath13 ) .
the gas to luminosity ratio , @xmath436 , is found to increase with decreasing sub - mm luminosity , which can be ascribed to the declining metallicity associated with the lower galaxy masses of the low luminosity galaxies .
this declining metallicity will lead to a lower dust to gas ratio , and therefore higher gas to dust luminosity ratio .
we provide fits to the total gas mass versus ir luminosity relations for all herschel bands , which can be interpolated between to obtain an empirical relation to be used at any ir wavelength for determining the total gas mass ( whose coefficients are given in table [ tab : linearfitgas ] ) . to minimize the effects of metallicity
, we also fit only galaxies with stellar masses greater than @xmath437 in our sample , which also results in a significantly reduced dispersion . for the spire 500 band
we find that a linear relation is a reasonable approximation , and find that log@xmath438 $ ] , with a dispersion of only 0.118 dex .
the molecular gas mass surface density has been demonstrated to be more tightly associated with the sfr than the atomic gas , thus we also provide calibrations for the molecular gas .
we find similarly strong correlations , with the strongest correlation found for the pacs100 and pacs160 bands that lie at the peak of the ir emission , and are strongly correlated with the total ir luminosity .
this correlation between l@xmath439 and m@xmath403 appears to be driven mostly by the correlation of both the average dust temperature ( as measured by pacs160/spire500 ) and molecular gas fraction ( l@xmath405/m@xmath358 ) with galaxy mass and metallicity , with dwarf galaxies showing both warmer dust and lower molecular gas fractions . given this correlation
, we suggest that the l@xmath439m@xmath403 relation not be applied to higher redshifts . using the resolved nature of the ir and gas surveys we employed in this survey , we also explored the correlations using elliptical annuli , and find the same correlations as the integrated comparisons , albeit with larger scatter .
for most galaxies we find that the @xmath424 ratio increases with radius , as expected from the determined radial metallicity gradients .
however , for some galaxies the @xmath424 relation is observed to deviate from the metallicity gradient at larger radii , becoming much larger .
a drop in the mean dust temperature plays at least some role in this observed excess , given the decrease in stellar mass surface densities relative to gas surface densities , and supported by the observed steeper @xmath422 to @xmath440 relation relative to the @xmath441 radial relation .
this deviation in the @xmath424 ratio in the outskirts in galaxies , and the extended hi disks not visible in the ir observed in a fraction of our galaxies , suggest that the mass of extended gas in galaxies can not be straightforwardly determined from the ir emission .
however , even with these radial trends , the total gas mass of galaxies within their optical radius can be well determined within a factor of 2 by observations of the sub - mm dust emission , and typically to within 30% for typical ( log(m@xmath442 ) galaxies observed in high redshift samples . | we use a sample of 36 galaxies from the kingfish ( herschel ir ) , heracles ( iram co ) , and things ( vla hi ) surveys to study empirical relations between herschel infrared ( ir ) luminosities and the total mass of the interstellar gas ( h@xmath0 + hi ) .
such a comparison provides a simple empirical relationship without introducing the uncertainty of dust model fitting .
we find tight correlations , and provide fits to these relations , between herschel luminosities and the total gas mass integrated over entire galaxies , with the tightest , almost linear , correlation found for the longest wavelength data ( spire500 ) .
however , we find that accounting for the gas - phase metallicity ( affecting the dust - to - gas ratio ) is crucial when applying these relations to low - mass , and presumably high - redshift , galaxies .
the molecular ( h@xmath0 ) gas mass is found to be better correlated with the peak of the ir emission ( e.g. pacs160 ) , driven mostly by the correlation of stellar mass and mean dust temperature . when examining these relations as a function of galactocentric radius , we find the same correlations , albeit with a larger scatter , up to radius of @xmath1 ( containing most of a galaxy s baryonic mass ) .
however , beyond that radius , the same correlations no longer hold , with increasing gas ( predominantly hi ) mass relative to the infrared emission .
the tight relations found for the bulk of the galaxy s baryonic content suggest that total gas masses of disk - like ( non - merging / ulirg ) galaxies can be inferred from far - infrared continuum measurements in situations where only the latter are available , e.g. in alma continuum observations of high - redshift galaxies . |
consider a quantum state of some system consisting of many particles . this system could be a collection of cold atoms in an optical lattice , or of atoms in cavities , coupled by light , or entirely optical systems .
assume that one is capable of performing local projective measurements on that system , however there is no way to realize a controlled coherent evolution .
can one perform universal quantum computing in such a setting ?
perhaps surprisingly , this is indeed the case : the _ one - way model _ of refs .
@xcite demonstrates that local measurements on the _ cluster state _
a certain multi - particle entangled state on an array of qubits @xcite do possess this computational power .
the insight gives rise to an appealing view of quantum computation : one can in principle abandon the need for any unitary control , once the initial state has been prepared .
the local measurements a feature that any computing scheme would eventually embody then take the role of preparation of the input , the computation proper , and the read - out .
this is of course a very desirable feature : quantum computation then only amounts to ( i ) preparing a universal resource state and ( ii ) performing local projective measurements [ 26 ] .
but what about other entangled quantum states , different from cluster or graph states @xcite ?
can they form a resource for universal computation ?
is it possible to tailor resource states to specific physical systems ?
for some experimental implementations
e.g. , cold atoms in optical lattices @xcite , atoms in cavities @xcite , optical systems [ 11 - 13 ] , ions in traps @xcite , or many - body ground states it may well be that preparation of cluster states is unfeasible , costly , or that they are particularly fragile to finite temperature or decoherence effects .
also , from a fundamental point of view , it is clearly interesting to investigate the computational power of many - body states either for the purpose of building measurement - based quantum computers or else for deciding which states could possibly be classically simulated @xcite . interestingly , very little progress has been made over the last years when it comes to going beyond the cluster state as a resource for measurement - based quantum computation ( mbqc ) . to our knowledge , no single computational model distinct from the one - way computer has been developed which would be based on local measurements on an algorithm - independent qubit resource state .
the apparent lack of new schemes for mbqc is all the more surprising , given the great advances that have been made toward an understanding of the structure of cluster state - based computing itself .
for example , it has been shown that the computational model of the one - way computer and teleportation - based approaches to quantum computing @xcite are essentially equivalent @xcite .
a particularly elegant way of realizing this equivalence was discovered in ref .
@xcite : they pointed out that the maximally entangled states used for the teleportation need not be physical .
instead , the role can be taken on by virtual entangled pairs used in a `` valence bond '' @xcite description of the cluster state .
this point of view is closely related to our approach to be described below .
further progress includes a clarification of the temporal inter - dependence of measurements @xcite . in ref .
@xcite a first non - cluster ( though not universal , but algorithm - dependent ) resource has been introduced , which includes the natural ability of performing three - qubit gates .
recently , refs .
@xcite initiated a detailed study of resource states which can be used to prepare cluster states ( see section [ sec : propertiesdiscussion ] ) . in this work ,
we describe methods for the systematic construction of new mbqc schemes and resource states .
this continues a program initiated in ref .
@xcite in a more detailed fashion .
we analyze mbqc in terms of `` computational tensor networks '' , building on a familiar tool from many - body physics known by the names of matrix - product states , finitely correlated states @xcite or projected entangled pair states @xcite .
the problem of finding novel schemes for measurement - based computation can be approached from two different points of view .
firstly , one may concentrate on the _ quantum states _ which provide the computational power of measurement - based computing schemes and ask 1 . _
what are the properties that render a state a universal resource for a measurement - based computing scheme for a discussion , in particular in relation to ref . @xcite . ] ? _ secondly , putting the emphasize on _ methods _ , the central question becomes 1 .
_ how can we systematically construct new schemes for measurement - based quantum computation ?
is there a framework which is flexible enough to allow for the construction of a variety of different models ? _ both of these intertwined questions will be addressed in this work .
as our main result , we present a plethora of new universal resource states and computational schemes for mbqc .
the examples have been chosen to demonstrate the flexibility one has when constructing models for measurement - based computation .
indeed , it turns out that many properties one might naturally conjecture to be necessary for a state to be a universal resource can in fact be relaxed .
needless to say , the weaker the requirements are for a many - body state to form a resource for quantum computing , the more feasible physical implementations of mbqc become .
below , we enumerate some specific results concerning the properties of resource states .
the list pertains to question 1 given in the introduction .
* in the cluster state , every particle is maximally entangled with the rest of the lattice . also , the localizable entanglement @xcite is maximal ( i.e. one can deterministically prepare an maximally entangled state between any two sites , by performing local measurements on the remainder ) . while both properties are essential for the original one - way computer , they turn out not to be necessary for computationally universal resource states . to the contrary
, we construct _ universal states which are locally arbitrarily pure_. * for previously known schemes for mbqc , it was essential that far - apart regions of the state were uncorrelated .
this feature allowed one to logically break down a measurement - based calculation into small parts corresponding to individual quantum gates .
our framework does not depend on this restriction and resources with _ non - vanishing correlations _ between any two subsystems are shown to exist .
this property is common e.g. , in many - body ground - states .
* cluster states can be prepared step - wise by means of a bi - partite _ entangling gate _ ( controlled - phase gate ) .
this property is important to the original universality proof .
more generally , one might conjecture that resource states must always result from an entangling process making use of mutually commuting entangling gates , also known as a unitary _ quantum cellular automaton _
once more , this requirement turns out not to be necessary . *
the cluster states can be used as _ universal preparators _ : any quantum state can be distilled out of a sufficiently large cluster state by local measurements .
once more , this property is essential to the original one - way computer scheme .
however , computationally universal resource states not exhibiting this properties do exist ( the reader is referred to ref .
@xcite for an analysis of resource states which are required to be preparators ; see also the discussion in section [ sec : propertiesdiscussion ] ) .
more strongly , we construct universal resources out of which not even a single two - qubit maximally entangled state can be distilled . *
a genuine _ qu - trit _ resource is presented ( distinct , of course , from a qu - trit version of the cluster state @xcite ) .
measurement - based quantum computing as generalization of the one - way model as being considered in this work .
initially , an entangled resource state is available , different from the cluster state , followed by local projective measurements on all individual constituents in the regular not necessarily cubic lattice . in all figures ,
dark gray circles denote individual physical systems.,width=207 ] we will further see that there is quite some flexibility concerning the computational model itself ( addressing question 2 mentioned in the introduction ) : * the new schemes differ from the one - way model in the way the _ inherent randomness _ of quantum measurements is dealt with . *
we generalize the well - known concept of _ by - product operators _ to encompass any finite group .
e.g. we show the existence of computational models , where the by - product operators are elements of the entire single - qubit clifford group , or the dihedral group .
* we explore schemes where each logical qubit is encoded in _ several neighboring correlation systems _ ( see section [ sec : ctn ] for a definition of the term `` correlation system '' ) .
* one can find ways to construct schemes in which interactions between logical qubits are controlled by `` routing '' the qubits towards an `` interaction zone '' or keeping them away from it . * in many schemes , we adjust the layout of the measurement pattern dynamically , incorporating information about previous measurement outcomes as we go along . in particular , the expected length of a computation is random ( this constitutes no problem , as the probability of exceeding a finite expected length is exponentially small in the excess ) .
what are the properties from which a universal resource state derives its power ? after clarifying the terminology , we will argue that an answer to this question desirable as it may be faces formidable obstacles .
quantum computation can come in a variety of different incarnations , as diverse as e.g. , the well - known gate - model @xcite , adiabatic quantum computation @xcite or mbqc .
all these models turn out to be equivalent in that they can simulate each other efficiently . for measurement - based schemes , the `` hardware '' consists of a multi - particle quantum system in an algorithm - independent state and a classical computer .
the input is a gate - model description of a quantum computation . in every step of the computation
, a local measurement is performed on the quantum state and the result is fed into the classical computer .
based on the outcomes of previous steps , the computer calculates which basis to use for the next measurements and , finally , infers the result of the computation from the measurement outcomes @xcite . having this procedure in mind , we call a quantum state a _ universal resource _ for mbqc , if a classical computer assisted by local measurements on this states can efficiently predict the outcome of any quantum computation
. the reader should be aware that another approach has recently been described in the literature .
the cluster state has actually a stronger property than the one just used for the definition of universality : it is a universal preparator .
this means that one can prepare any given quantum state on a given sub - set of sites of a sufficiently large cluster by means of local measurements .
hence , cluster states could in principle be used for information processing tasks which require a quantum output .
@xcite referred to this scenario as _ cq - universality _
i.e. universality for problems which require a classical input but deliver a quantum output .
this observation is the basis of ref .
@xcite , where a state is called a universal resource if it possesses the strong property of being a universal preparator , or , equivalently , of being cq - universal . clearly , any efficient universal preparator is also a computationally universal resource for mbqc ( since one can , in particular , prepare the cluster state ) .
but the converse is not true , as our results show . indeed , while it proves possible to come up with necessary criteria for a state to be a universal preparator @xcite , we will argue below that the current limited understanding of quantum computers makes it extremely hard to specify necessary conditions for computational universality . in order to pinpoint the source of the quantum speedup
, we might try to find schemes where more and more work is done by the classical computer , while the employed quantum states become `` simpler '' ( e.g.,smaller or less entangled ) .
how far can we push this program without losing universality ?
the answer is likely to be intractable .
currently , we are not aware of a proof that quantum computation is indeed more powerful than classical methods .
hence , it can presently not be excluded that no assistance from a quantum state is necessary at all .
[ obs : assumption ] if one is unwilling to _ assume _ that there is a separation between classical and quantum computation ( i.e. , bpp @xmath0 bqp ) , then it is impossible to rule out any state as a universal resource .
it is , however , both common and sensible to assume superiority of quantum computers and we will from now on do so .
observation [ obs : assumption ] still serves a purpose : it teaches us that the only known way to rule out universality is to invoke this assumption ( this avenue was taken , e.g. , in refs .
@xcite ) .
[ obs : onlyway ] the only currently known method for excluding the possibility that a given quantum state forms a universal resource is to show that any measurement - based scheme utilizing the state can be efficiently simulated by a classical computer .
thus , the situation presents itself as follows : there is a tiny set of quantum states for which it is possible to prove that any local measurement - based scheme can be efficiently simulated .
on the other extreme , there is an even tinier set for which universality is provable . for the vast majority
no assessment can be made .
furthermore , given the fact that rigorously establishing the `` hardness '' of many important problems in computer science turned out to be extremely challenging , it seems unlikely that this situation will change dramatically in the foreseeable future .
we conclude that a search for necessary conditions for universality is likely to remain futile .
the converse question , however , can be pursued : it is possible to show that many properties that one might naively assume to be present in any universal resource are , in fact , unnecessary .
the current section is devoted to an in - depth treatment of a class of states known respectively as valence - bond states , finitely correlated states , matrix product states or projected entangled pairs states , adapted to our purposes of measurement - based quantum computing .
this family turns out to be especially well - suited for a description of a computing scheme . indeed , any systematic analysis of resources states requires a framework for describing quantum states on extended systems .
we briefly compile a list of desiderata , based on which candidate techniques can be assessed .
* the description should be _ scalable _ , so that a class of states on systems of arbitrary size can be treated efficiently .
* as quantum states which are naturally described in terms of one - dimensional topologies have been shown to be classically simulable @xcite , the framework ought to handle _ two- or higher dimensional topologies _ naturally . *
the basic operation in measurement - based computation are _
local measurements_. it would be desirable to describe the effect of local measurements in a local manner .
ideally , the class of efficiently describable states should be closed under local measurements .
* the class of describable states should include elements which show features that naturally occur in _ ground states _ of quantum many - body systems , such as _ non - maximal local entropy of entanglement _ or _ non - vanishing two - point correlations _ , etc . the description of states to be introduced below complies with all of these points .
we will introduce the construction in several steps , starting with one - dimensional matrix product states .
the new view on the processing of information is that the matrices appearing in the description of resource states are taken literally , as operators processing quantum information
. a _ matrix product state _ ( mps ) for a chain of @xmath1 systems of physical dimension @xmath2 ( so @xmath3 for qubits ) is specified by * an _ auxiliary @xmath4 dimensional vector space _ (
@xmath4 being some parameter , describing the amount of correlation between two consecutive blocks of the chain ) , * for each system @xmath5 a set of @xmath2 @xmath6-_matrices _ @xmath7 , j\in\{0\dots d-1\}$ ] . * two @xmath4-dimensional vectors @xmath8 representing _ boundary conditions_. the state vector @xmath9 of the matrix product state is then given explicitly by appears on the _ left _ of eq .
( [ eqn : linearmps ] ) . in linear algebra formulas ,
information usually flows from right to left : @xmath10 means `` @xmath11 is acted on by @xmath12 , then by @xmath13 '' . in the graphical notation to be introduce later , it is much more natural to let information flow from left to right : @xmath14{{\mbox{$| \psi \rangle$}}}}\ar[r]&{*+[f]{a}}\ar[r]&{*+[f]{b}}\ar[r ] & } } .\ ] ] the order in eq .
( [ eqn : linearmps ] ) anticipates the graphical notation . ]
@xmath15 \dots a_1[s_1 ] { \mbox{$| l \rangle$ } } \,\,{\mbox{$| s_1 , \dots , s_n \rangle$}}.\ ] ] from now on we will assume that the matrices are site - independent : @xmath7=a[j]$ ] , so the mps is translationally invariant up to the boundary conditions .
we take the freedom of disregarding normalization whenever this consistently possible .
let us spend a minute interpreting eq .
( [ eqn : linearmps ] ) .
assume we have measured the first site in the computational basis and obtained the outcome @xmath16 .
one immediately sees that the resulting state vector @xmath17 on the remaining sites is again a mps , where the left - hand side boundary vector now reads @xmath18{\mbox{$| l \rangle$}}.\ ] ] hence the state of the auxiliary system gets changed according to the measurement outcome .
so we find that the correlations between the state of the first site and the rest of the chain are mediated via the auxiliary space , which will thus be referred to as _ correlation space _ in the sequel . in the past
, the matrices appearing in the definition of @xmath9 have been treated mainly as a collection of variational parameters , used to parametrize ansatz states for ground states of spin chains @xcite .
however and that is the basic insight underlying our view on mbqc eq . ( [ eqn : firstcomp ] ) can also be read as an operator @xmath19 $ ] acting on some quantum state @xmath20 .
we will elaborate on this interpretation in section [ 1dtn ] . in order to translate eq .
( [ eqn : linearmps ] ) to the setting of 2-d lattices , we need to cast it into the form of a tensor network . setting @xmath21 and @xmath22_{i , j}:={\mbox{$\langle j |$ } } a { \mbox{$| i \rangle$}},\ ] ] we can write eq .
( [ eqn : linearmps ] ) as @xmath23_{i_0,i_1 } \dots a[s_n]_{i_{n-1 } , i_n } { r^\dagger}_{i_n}.\ ] ] while eq .
( [ eqn : tensormps ] ) is awkward enough , the 2-d equivalent is completely unintelligible . to cure this problem
, we introduce a graphical notation which enables an intuitive understanding beyond the 1-d case . in the following
, tensors will be represented by boxes , indices by edges : @xmath24{l}\ar[r]&},\\ a[s]_{l , r}&=&\xymatrix@c=5mm{\ar[r]&*+[f]{a[s]}\ar[r]&},\\ { r^\dagger}_l & = & \xymatrix@c=5mm{\ar[r]&*+[f]{r^\dagger}}\,.\end{aligned}\ ] ] needless to say , in the equation above , `` @xmath25 '' is the index leaving the box on the left - hand - side , `` @xmath26 '' the right - hand - side one .
connected lines designate contractions of the respective indices .
( [ eqn : linearmps ] ) now reads @xmath27{l}}\ar@{-}[r]&*+[f]{a[s_1]}\ar@{-}[r ] & \dots\ar@{-}[r]&*+[f]{a[s_n]}\ar@{-}[r]&*+[f]{r^\dagger } } .\ ] ] a single - index tensor can be interpreted as the expansion coefficients of either a `` ket '' or a `` bra '' .
sometimes , we will indicate what interpretation we have in mind by placing arrows on the edges : outgoing arrows designating `` kets '' , incoming arrows `` bras '' @xmath28{l}\ar[r]&}= { \mbox{$| l
\rangle$}},\quad \xymatrix@c=5mm{\ar[r]&*+[f]{r^\dagger}}= { \mbox{$\langle r |$}}.\ ] ] tensors with two indices @xmath29 can naturally be interpreted as operators . in the graphical notation
we often want to think of information flowing from the left to the right , in which case @xmath30 would be denoted as @xmath31&*+[f]{a}\ar[r]&}= a,\ ] ] i.e. with the l.h.s .
index being associated with a `` bra '' and the r.h.s one with a `` ket '' .
the following relations exemplify the definition : @xmath32{l}}\ar@{-}[r]&{*+[f]{r } } } } \,,\\ \nonumber \\
a { \mbox{$| l \rangle$ } } & = & { \xymatrix@c=5mm{{*+[f]{l } } \ar@{-}[r]&{*+[f]{a } } \ar[r ] & } } , \\ \nonumber \\ a b&= & { \xymatrix@c=5mm{\ar[r]&{*+[f]{b } } \ar@{-}[r]&{*+[f]{a}}\ar[r ] & } } , \\ \nonumber \\ { \operatorname{tr}}(a b)&= & \begin{xy } * ! c\xybox{\xymatrix@c=5mm@r=3 mm { & { * + [ f]{b } } \ar@{-}[r ] & { * + [ f]{a}}\ar@{- } ` /4pt[r]`[dl]`[dll]`[l][l ] & & \\ & & & & } } \end{xy}.\end{aligned}\ ] ] the formula for the expansion coefficients of a matrix product state finally becomes @xmath33{l } } \ar@{-}[r]&*+[f]{a[s_1 ] } \ar@{-}[r ] & \dots \ar@{-}[r]&*+[f]{a[s_n ] } \ar@{-}[r]&*+[f]{r^\dagger } } \,.\ ] ] this formula suggest a more `` dynamic '' interpretation of mps : the l.h.s . boundary conditions @xmath20 specify an initial state of the correlation system , which is acted on by the matrices of the mps representation .
the next paragraph is going to elaborate on this point .
we return to the discussion of the properties of matrix product states . above , it has been shown how to compute the overlap of @xmath9 with an element of the computational basis ( c.f .
eq.([eqn : tensormps ] ) ) .
the next step is to generalize this to any local projection operator . indeed ,
if @xmath34 is a general state vector in @xmath35 , we abbreviate @xmath36+{\mbox{$\langle \phi | 1 \rangle$}}\,a[1]=:a[\phi].\ ] ] one then easily derives the following , central formula @xmath37{l}{\ar@{-}[r]}[r]&*+[f]{a[\phi_1 ] } { \ar@{-}[r]}[r]&\dots{\ar@{-}[r]}&*+[f]{a[\phi_n]}{\ar@{-}[r]}&*+[f]{r } } \,\ , .\ ] ] now suppose we measure local observables on @xmath9 and obtain results corresponding to the eigenvector @xmath38 at the @xmath5-th site .
( [ eqn : transport ] ) allows us to re - interpret this process as follows .
initially , the @xmath4-dimensional correlation system is prepared in the state @xmath20 .
the result @xmath39 at the first site induces the evolution @xmath40{\mbox{$|
l \rangle$}}.\ ] ] from this point of view , a sequence of measurements on @xmath9 is tantamount to a processing of the correlation system s state by the operations @xmath41$].$ ] is not going to be unitary . choosing the bases in such a way
as to ensure unitarity is an essential part of the design of a computational scheme for a given resource . ]
an appealing perspective on mbc suggests itself : measurement - based computing takes place in correlation space .
the gates acting on the correlation systems are determined by local measurements .
intuitively , `` quantum correlations '' are the source of a resource s computational potency .
the strength of this framework lies in the fact that it assigns a concrete mathematical object to these correlations .
indeed , it will turn out that mbqc can be understood completely using this interpretation . to illustrate the abstract definitions made above
, we will discuss the linear cluster state vector @xmath42 in this section .
it is both one of the simplest and certainly the most important mps in the context of mbqc . what is the tensor network representation of @xmath42
recall that the cluster state can be generated by preparing @xmath1 sites in the state vector @xmath43 and subsequently applying the controlled-@xmath44 operation @xmath45 between any two nearest neighbors .
effectively , @xmath46 introduces a @xmath47-phase whenever two consecutive systems are in the @xmath48-state .
hence its expansion coefficients in the computational basis are given by @xmath49 where @xmath50 denotes the number of sites @xmath5 such that @xmath51 .
this observation makes it simple to derive the tensors of the mps representation .
we need a @xmath52-dimensional correlation system , which loosely speaking will convey the information about the state @xmath53 of the @xmath5-th site to site @xmath54 .
define the matrices @xmath55 $ ] by @xmath56&{*+[f]{a[0]}}\ar[r ] & } } & = & { \mbox{$| + \rangle$}}_r { \mbox{$\langle 0 |$}}_l , \\
\label{eqn : clustermatrix1 } { \xymatrix@c=5mm{\ar[r]&{*+[f]{a[1]}}\ar[r ] & } } & = & { \mbox{$| - \rangle$}}_r { \mbox{$\langle 1 |$}}_l .
\end{aligned}\ ] ] the intuition behind this choice is as follows . by the elementary relations @xmath57 the contraction in the middle of @xmath58&{*+[f]{a[s_1]}}\ar@{-}[r]&{*+[f]{a[s_2]}}\ar[r ] & } } \ ] ] will yield a sign of `` @xmath59 '' exactly if @xmath60 . indeed , setting the boundary vectors to @xmath61 one checks easily that @xmath62\dots a[s_1]{\mbox{$| l \rangle$ } } = 2^{-n/2 } ( -1)^p,\ ] ] which is exactly the value required by eq .
( [ eqn : clustercoefficients ] ) .
below , we will interpret the correlation system of a 1-d chain as a single logical quantum system .
for this interpretation to be viable , we must check that the following basic operations can be performed deterministically by local measurements : i ) prepare the correlation system in a known initial state , ii ) transport that state along the chain ( possibly subject to known unitary transformations ) and iii ) read out the final state . to set the state of the correlation system to a definitive value
, we measure some site say the @xmath5-th in the @xmath44-eigenbasis . throughout this work
, we will choose the notation @xmath63 , @xmath64 , and @xmath44 for the _
pauli operators_. denote the measurement outcome by @xmath65 . in case of @xmath66 , eq.([eqn : clustermatrix0 ] ) tells us that the state of the correlation system to the right of the @xmath5-th site will be @xmath67 ( up to an unimportant phase ) .
likewise , a @xmath68 outcome prepares the correlation system in @xmath69 , according to eq .
( [ eqn : clustermatrix1 ] ) .
it follows that we can use @xmath44-measurements for preparation . how to cope with the intrinsic randomness of quantum measurements
will concern us later .
secondly , consider the operators @xmath70&{*+[f]{a[+]}}\ar[r ] & } & = & 2^{-1/2 } ( \xymatrix@c=3mm{\ar[r]&{*+[f]{a[0]}}\ar[r ] & } + \xymatrix@c=3mm{\ar[r]&{*+[f]{a[1]}}\ar[r ] & } ) \nonumber \\ & \propto&{\mbox{$| + \rangle$}}{\mbox{$\langle 0
|$}}+{\mbox{$| - \rangle$}}{\mbox{$\langle 1 |$}}=h , \label{eqn : clustertransport+ } \\ \nonumber \\
\xymatrix@c=3mm{\ar[r]&{*+[f]{a[-]}}\ar[r]&}&\propto&h z , \label{eqn : clustertransport- } \end{aligned}\ ] ] where @xmath71 is the hadamard - gate .
we see immediately that measurements in the @xmath63-eigenbasis give rise to a unitary evolution on the correlation space . similarly
, one can show that one can generate arbitrary local unitaries by appropriate measurements in the @xmath64-@xmath44 plane .
below , we will frequently be confronted with a situation like the one presented in eqs.([eqn : clustertransport+],[eqn : clustertransport- ] ) , where the correlation system evolves in one of two possibilities , dependent on the outcome of a measurement .
it will be convenient to introduce a compact notation that encompasses both cases in a single equation .
so eqs . ( [ eqn : clustertransport+],[eqn : clustertransport- ] ) will be represented as @xmath72&{*+[f]{a[x]}}\ar[r ] & } } = h z^x.\ ] ] here @xmath73 corresponds to the outcome @xmath74 in an @xmath63-measurement , whereas @xmath75 corresponds to the outcome @xmath76 . in general , a physical observable given as an argument to a tensor corresponds to a measurement in the observable s eigenbasis .
the measurement outcome is assigned to a suitable variable as in the above example .
lastly , we must show how to physically read out the state of the purely logical correlation system .
it turns out that measuring the @xmath54-th physical system in the @xmath44-eigenbasis corresponds to a @xmath44-measurement of the state of the correlation system just after site @xmath5 .
indeed , suppose we have measured the first @xmath5 systems and obtained results corresponding to the local projection operator @xmath77 .
further assume that as a result of these measurements the correlation system is in the state @xmath78 : @xmath79{l}\ar@{-}[r]&*+[f]{a[\phi_1]}\ar@{-}[r]&\dots\ar@{-}[r]&*+[f]{a[\phi_i]}\ar[r ] & } = { \mbox{$| 0 \rangle$}}.\ ] ] using eq .
( [ eqn : clustermatrix1 ] ) we have that @xmath80{l}\ar@{-}[r]&*+[f]{a[\phi_1]}\ar@{-}[r]&\dots\ar@{-}&*+[f]{a[\phi_i]}{\ar@{-}[r ] } & * + [ f]{a[1]}\ar[r ] & } \\ & \propto & { \mbox{$| + \rangle$ } } { \mbox{$\langle 1 | 0 \rangle$}}=0.\nonumber\end{aligned}\ ] ] but then it follows from eq .
( [ eqn : transport ] ) that the probability of obtaining the result @xmath81 for a @xmath44-measurement on site @xmath54 is equal to zero . in other words : if the _ correlation system _ is in the state @xmath78 after the @xmath5-th site , then the @xmath54-th _ physical site _ must also be in the state @xmath78 . an analogous argument for
the @xmath48-case completes the description of the read - out scheme .
the graphical notation greatly facilitates the passage to 2-d lattices . here ,
the tensors @xmath82 $ ] have four indices @xmath82_{l , r , u , d}$ ] , which will be contracted with the indices of the left , right , upper and lower neighboring tensors respectively . after choosing a set of boundary conditions @xmath83 , the expansion coefficients of the state vector @xmath9 are computed as illustrated in the following example on a @xmath84-lattice : @xmath85{u } } & { * + [ f]{u } } \\ { * + [ f]{l}}{\ar@{-}[r]}&{*+[f]{a[s_{1,1}]}}{\ar@{-}[u]}{\ar@{-}[r]}&{*+[f]{a[s_{2,1}]}}{\ar@{-}[u]}{\ar@{-}[r]}&{*+[f]{r } } \\ { * + [ f]{l}}{\ar@{-}[r]}&{*+[f]{a[s_{1,2}]}}{\ar@{-}[u]}{\ar@{-}[r]}&{*+[f]{a[s_{2,2}]}}{\ar@{-}[u]}{\ar@{-}[r]}&{*+[f]{r } } \\ & { * + [ f]{d}}{\ar@{-}[u]}&{*+[f]{d}}{\ar@{-}[u ] } } } \end{xy}.\end{aligned}\ ] ] in the 1-d case , we thought of the quantum information as moving along a single correlation system from the left to the right .
for higher - dimensional lattices , a greater deal of flexibility proves to be expedient . for example , sometimes it will be natural to interpret the tensor @xmath86 as specifying the matrix elements of an operator @xmath12 mapping the left and the lower correlation systems to the right and the upper ones : @xmath87&{*+[f]{a}}\ar[u]\ar[r ] & \\ & \ar[u]&\\ } } \end{xy}.\ ] ] often , on the other hand , the interpretation @xmath88&\\ \ar[r]&{*+[f]{a}}\ar[r ] & \\ & \ar[u]&\\ } } \end{xy}\ ] ] or yet another one is to be preferred .
we have seen in section [ 1dtn ] that the correlation system of a one - dimensional matrix product state can naturally be interpreted as a single quantum system subject to a time evolution induced by local measurements .
it would be desirable to carry this intuition over to the 2-d case .
indeed , most of the examples to be discussed below are all similar in relying on the same basic scenario : some horizontal lines in the lattice are interpreted as effectively one - dimensional systems , in which the logical qubits travel from the left to the right .
the vertical dimension is used to either couple the logical systems or isolate them from each other ( see fig .
[ fig : flow ] ) .
the reader should recall that this setting is very similar to the original cluster state based - techniques .
clearly , it would be interesting to devise schemes not working in this way and the example presented in section [ sec : toric2 ] takes a first step in this direction .
once again the cluster state serves as an example .
one can work out the tensor network representation of the 2-d cluster state vector @xmath89 in the same way utilized for the 1-d case in section [ sec:1dcluster ] .
the resulting tensors are : @xmath90&{*+[f]{a[0]}}\ar[u]\ar[r ] & \\ & \ar[u ] & } } \end{xy } = { \mbox{$| + \rangle$}}_r{\mbox{$| + \rangle$}}_u\,{\mbox{$\langle 0 |$}}_l{\mbox{$\langle 0 |$}}_d , \\
\label{eqn:2dcluster1 } \begin{xy } * !
c\xybox{\xymatrix@c=3mm@r=3 mm { & & \\
\ar[r]&{*+[f]{a[1]}}\ar[u]\ar[r ] & \\ & \ar[u]&\ } } \end{xy } = { \mbox{$| - \rangle$}}_r{\mbox{$| - \rangle$}}_u\,{\mbox{$\langle 1 |$}}_l{\mbox{$\langle 1 |$}}_d , \\ { \mbox{$| l \rangle$}}={\mbox{$| d \rangle$ } } = { \mbox{$| + \rangle$ } } , \qquad { \mbox{$| r \rangle$}}={\mbox{$| u \rangle$}}={\mbox{$| 1 \rangle$}}.\end{aligned}\ ] ] an important property of eqs .
( [ eqn:2dcluster0 ] , [ eqn:2dcluster1 ] ) is that the tensors @xmath55 $ ] factor .
one could graphically represent this fact by writing @xmath91}&{*+[f]{a[0]}}{\ar@{-}[u]}{\ar@{-}[r ] } & \\ & { \ar@{-}[u ] } & } } \end{xy } = \begin{xy } * !
c\xybox{\xymatrix@c=3mm@r=3 mm { & & & & \\ & & { * + [ f]{+}}{\ar@{-}[u]}&\\ { \ar@{-}[r]}&{*+[f]{0}}&&{*+[f]{+}}{\ar@{-}[r]}&\\ & & { * + [ f]{0}}&\\ & & { \ar@{-}[u ] } } } \end{xy},\ ] ] where @xmath14{0}}\ar[r ] & } } = { \mbox{$| 0 \rangle$}},\,\ , { \xymatrix@c=5mm{{*+[f]{+}}\ar[r ] & } } = { \mbox{$| + \rangle$}}.\ ] ] in other words : the tensors @xmath55 $ ] effectively de - couple their respective indices .
based on this fact , we will see momentarily how @xmath44-measurements can be used to stop information from flowing through the lattice .
indeed , suppose three vertically adjacent sites are measured , from top to bottom , respectively in the @xmath44 , @xmath63 and @xmath44-eigenbasis : @xmath92&{*+[f]{a[z_u]}}\ar[u]\ar[r]&\\ \ar[r]&{*+[f]{a[x]}}\ar@{-}[u]\ar[r]&\\ \ar[r]&{*+[f]{a[z_d]}}\ar@{-}[u]\ar[r]&\\ & \ar[u ] & } } \end{xy}.\ ] ] denote the measurement results by @xmath93 .
as before , these numbers correspond to @xmath94 for @xmath95 and @xmath96 for @xmath97 , as well as @xmath73 for @xmath98 and @xmath75 for @xmath99 .
in fact , we are mainly interested in the indices of the middle tensor , as they will be the ones which carry the logical information . to this end eq .
( [ eqn : factor ] ) is of use , as it says that the upper and lower tensors factor and hence it makes sense to dis - regard all of their indices which do not influence the middle part .
it hence suffices to consider @xmath100{a[z_u]}}\ar@{-}[d]&\\ \ar[r]&{*+[f]{a[x]}}\ar[r]&\\ & { * + [ f]{a[z_d]}}\ar@{-}[u]&\\ } } \end{xy}.\ ] ] as a first step , we calculate @xmath101{0}}\ar@{-}[d]&\\ \ar[r]&{*+[f]{a[0]}}\ar[r]&\\ & { * + [ f]{+}}\ar@{-}[u]&\\ } } \end{xy } & = & \begin{xy } * ! c\xybox{\xymatrix@c=3mm@r=3 mm { & & { * + [ f]{0}}{\ar@{-}[d]}&\\ & & { * + [ f]{+}}\\ \ar[r]&{*+[f]{0}}&&{*+[f]{+}}\ar[r]&\\ & & { * + [ f]{0}}{\ar@{-}[d]}&\\ & & { * + [ f]{+ } } } } \end{xy } = 2^{-1 } { \mbox{$| + \rangle$}}{\mbox{$\langle 0 |$}},\end{aligned}\ ] ] having used eq .
( [ eqn : factor ] ) and the basic fact @xmath14{+}}{\ar@{-}[r]}[r]&{*+[f]{0 } } } } = { \mbox{$\langle 0 | + \rangle$}}=2^{-1/2}.\ ] ] a similar calculation where @xmath102 $ ] is substituted by @xmath103 $ ] yields @xmath104 .
hence , for @xmath105\propto a[0]+a[1]$ ] , we have @xmath101{0}}\ar@{-}[d]&\\ \ar[r]&{*+[f]{a[+]}}\ar[r]&\\ & { * + [ f]{+}}\ar@{-}[u]&\\ } } \end{xy } & \propto & { \mbox{$| + \rangle$}}{\mbox{$\langle 0
|$}}+{\mbox{$| - \rangle$}}{\mbox{$\langle 1 |$}}=h.\end{aligned}\ ] ] similarly , @xmath101{0}}{\ar@{-}[d]}&\\ \ar[r]&{*+[f]{a[-]}}\ar[r]&\\ & { * + [ f]{+}}{\ar@{-}[u]}&\\ } } \end{xy } & \propto & h z.\end{aligned}\ ] ] after these preparations it is simple to conclude that @xmath106{a[z_u]}}\ar@{-}[d]&\\ \ar[r]&{*+[f]{a[x]}}\ar[r]&\\ & { * + [ f]{a[z_d]}}\ar@{-}[u]&\\ } } \end{xy } \propto h z^{z_u+x+z_d}.\ ] ] this finding tells us how to transport quantum information along horizontal lines through the lattice .
namely by measuring the line in the @xmath63-eigenbasis to cause the information to flow from the left to the right and measuring vertically adjacent sites in the @xmath44-eigenbasis to shield the information from the rest of the lattice .
( [ eqn:2dtransport ] ) should be compared with eqs.([eqn : clustertransport+],[eqn : clustertransport- ] ) .
so up to possible corrections of the form @xmath107 , the procedure outlined above enables us to effectively prepare a 1-d cluster state within the 2-d lattice .
up to this point , we have reformulated the computational model of the one - way computer in the language of computational tensor networks .
this picture of one - way computation is educational in its own right .
however , to convincingly argue that the framework is rich enough to allow for quite different models , we have to explicitly construct novel schemes .
it is the purpose of this section to discuss a number of examples of new resources .
as before , important features will be highlighted as `` observations '' .
our first example is inspired by the _ aklt state _
@xcite , which is well - known in the context of condensed matter physics . the _ aklt model _
is a 1-d , spin-1 , nearest neighbor , frustration free , gapped hamiltonian .
its unique ground state is a matrix product state with @xmath52 and indeed , the aklt model motivated the first studies of such states @xcite .
the defining matrices of the mps description are : @xmath108&{*+[f]{a[0]}}\ar[r ] & } } & = & z,\\ { \xymatrix@c=5mm{\ar[r]&{*+[f]{a[1]}}\ar[r ] & } } & = & 2^{-1/2 } |0\rangle_r\langle 1|_l,\\ { \xymatrix@c=5mm{\ar[r]&{*+[f]{a[2]}}\ar[r ] & } } & = & 2^{-1/2 } |1\rangle_r\langle 0|_l\end{aligned}\ ] ] we will choose the boundary conditions to be @xmath109 . as a matter of fact , we will not work directly with the aklt state , but with a small variation , for which it turns out to be more straight - forward to construct a scheme for mbqc . in this modification ,
the matrix @xmath102 $ ] is given by the hadamard gate , instead of the pauli @xmath44 operator : @xmath110&{*+[f]{a[0]}}\ar[r ] & } } = h.\ ] ] this state shares all the defining properties of the original : it is the unique ground - state of a spin-1 nearest neighbor frustration free gapped hamiltonian ( see appendix [ sec : akltappendix ] ) .
against the background of our program , the obvious question to ask is whether these matrices can be used to implement any evolution on the correlation space . to show that this is indeed the case ,
let us first analyze a measurement in the @xmath111-basis , where @xmath112 . in a mild abuse of notation ,
we will hence write @xmath113 for state vectors in the subspace spanned by @xmath114 instead of @xmath115 . from eqs.([eqn : akltorg]-[eqn : aklth ] ) one finds that depending on the measurement outcome , the operation realized on the correlation space will be one of @xmath116 or @xmath117 . at this point , we have to turn to an important issue : how to compensate for the randomness of quantum measurement outcomes .
assume for now that we intended to just transport the information faithfully from left to right .
in this case , we consider the operator @xmath118 as an unwanted _ by - product _ of the scheme . the one - way computer based on cluster states has the remarkable property that the by - products can be dealt with by adjusting the measurement - bases depending on the previous outcomes , without changing the general `` layout '' ( in the sense of fig .
[ fig : flow ] ) of the computation @xcite . for more general models , as the ones considered in this work , such a simple solution seems not available .
fortunately , we can employ a `` trial - until - success '' strategy , which proves remarkably general .
the key points to notice are that i ) the three possible outcomes @xmath119 and @xmath44 generate a finite group @xmath120 and ii ) the probability for each outcome is equal to @xmath121 , independent of the state of the correlation system
. we will refer to @xmath120 as the model s _ by - product _ group .
now suppose we measure @xmath122 adjacent sites in the @xmath123-basis . the resulting overall by - product operator @xmath124 will be a product of @xmath122 generators @xmath125 .
so by repeatedly transporting the state of the correlation system to the right , the by - products are subject to a random walk on @xmath120 . because @xmath120 is finite
, every element will occur after a finite expected number of steps ( as one can easily prove ) .
the group structure opens up a way of dealing with the randomness .
indeed , assume that initially the state vector of the correlation system is given by @xmath126 , for some unwanted @xmath127 .
transferring the state along the chain will introduce the additional by - product operator @xmath128 after some finite expected number of steps , leaving us with @xmath129 as desired .
the technique outlined here proves to be extremely general and we will encounter it in further examples presented below . [
obs : randomness ] possible sets of by - product operators are not limited to the pauli group .
a way of compensating randomness for other finite by - product operator groups is to adopt a `` trial - until - success strategy '' , which gives rise to a random length of the computation .
this length is in each case shown to be bounded on average by a constant in the system size . by the preceding paragraphs
, we can implement any element of @xmath120 on the correlation space .
we next address the problem of realizing a phase gate @xmath130 for some @xmath131 . to this end , consider a measurement on the @xmath132-basis .
there are three cases * the outcome corresponds to @xmath133 . in this case
, we get @xmath134 on the correlation space and are hence done . *
the outcome corresponds to @xmath135 .
we get @xmath136 , which is the desired operation , up to an element of the by - product group , which we can rid ourselves of as described above . *
lastly , in case of @xmath78 , we implement @xmath71 on the correlation space . as @xmath137 ,
we can `` undo '' it and then re - try to implement the phase gate .
hence , we can implement any element of @xmath138 as well as @xmath134 on the correlation space .
this implies that @xmath139 is also realizable and therefore any single - qubit unitary , as @xmath140 is generated by operations of the form @xmath134 and @xmath141 .
the state of the correlation system can be prepared by measuring in the computational basis . in case one
obtains a result of `` @xmath81 '' or `` @xmath142 '' , the state of the correlation system will be @xmath78 or @xmath48 respectively , irrespective of its previous state .
a `` @xmath143''-outcome will not leave the correlation system in a definite state .
however , after a finite expected number of steps , a measurement will give a non-``0''-result .
lastly , a read - out scheme can be realized similarly ( c.f .
section [ sec:1dcluster ] ) .
ground states of one - dimensional gapped nearest - neighbor hamiltonians may serve as resources for transport and arbitrary rotations .
[ sec : aklt2d ] a universal resource deriving from the aklt - model.,width=245 ] several horizontal 1-d aklt - type states can be coupled to become a universal 2-d resource .
the coupling can be facilitated by performing a controlled - z operation , embedded into the three - dimensional spin-1 space , between vertically adjacent nearest neighbors .
more specifically , we will use the operation @xmath144 , which introduces a @xmath47-phase between two systems exactly if both are in the state @xmath145 .
the tensor network representation of this resource is given by @xmath146&{*+[f]{a[0]}}\ar[u]\ar[r ] & \\ & \ar[u ] & } } \end{xy } & = & h_{l\to r}\otimes { \mbox{$| + \rangle$}}_u{\mbox{$\langle 0 |$}}_d , \\ \begin{xy } * !
c\xybox{\xymatrix@c=3mm@r=3 mm { & & \\
\ar[r]&{*+[f]{a[1]}}\ar[u]\ar[r ] & \\ & \ar[u]&\ } } \end{xy } & = & 2^{-1/2 } { \mbox{$| 0 \rangle$}}_r{\mbox{$\langle 1 |$}}_l\otimes { \mbox{$| + \rangle$}}_u{\mbox{$\langle 0 |$}}_d , \\ \begin{xy } * !
c\xybox{\xymatrix@c=3mm@r=3 mm { & & \\ \ar[r]&{*+[f]{a[2]}}\ar[u]\ar[r ] & \\ & \ar[u]&\ } } \end{xy } & = & 2^{-1/2 } { \mbox{$| 1 \rangle$}}_r{\mbox{$\langle 0 |$}}_l\otimes { \mbox{$| - \rangle$}}_u{\mbox{$\langle 1 |$}}_d,\end{aligned}\ ] ] as one can check in analogy to sec . [ sec:2dcluster ] . here ,
@xmath147 to verify that the resulting 2-d state constitutes a universal resource , we need to check that a ) one can isolate the correlation system of a horizontal line from the rest of the lattice , so that it may be interpreted as a logical qubit and b ) one can couple these logical qubits to perform an entangling gate .
the first step works in complete analogy to section [ sec:2dcluster ] , see fig .
[ fig : aklt ] .
indeed , one simply confirms that @xmath100{a[z_u]}}\ar@{-}[d]&\\ \ar[r]&{*+[f]{a[s]}}\ar[r]&\\ & { * + [ f]{a[z_l]}}\ar@{-}[u]&\\ } } \end{xy } = \pm { \xymatrix@c=5mm{\ar[r]&{*+[f]{a[s]}}\ar[r ] & } } , \ ] ] where @xmath148 and @xmath149 denotes a measurement in the @xmath150-basis .
so measuring the vertically adjacent nodes in the computational basis gives us back the 1-d state , up to a possible sign .
a controlled-@xmath44 gate can be realized in five steps : @xmath151&{*+[f]{a[x]}}\ar@{-}[r ] & { * + [ f]{a[x]}}\ar@{-}[r]&{*+[f]{a[x]}}\ar@{-}[r ] & { * + [ f]{a[x]}}\ar@{-}[r]&{*+[f]{a[x]}}\ar[r ] & \\ \ar[r]&{*+[f]{a[z]}}\ar@{-}[r]\ar@{-}[u]\ar@{-}[d]&{*+[f]{a[z]}}\ar@{-}[u]\ar@{-}[d]\ar@{-}[r]&{*+[f]{a[y]}}\ar@{-}[u]\ar@{-}[d]\ar@{-}[r ] & { * + [ f]{a[z]}}\ar@{-}[u]\ar@{-}[d]\ar@{-}[r]&{*+[f]{a[z]}}\ar@{-}[u]\ar@{-}[d]\ar[r]&\\ \ar[r]&{*+[f]{a[x]}}\ar@{-}[r ] & { * + [ f]{a[x]}}\ar@{-}[r]&{*+[f]{a[x]}}\ar@{-}[r ] & { * + [ f]{a[x]}}\ar@{-}[r]&{*+[f]{a[x]}}\ar[r ] & } } \end{xy}.\nonumber\\\end{aligned}\ ] ] the pauli matrices @xmath152 are understood as being embedded into the @xmath153-subspace .
so , e.g. , @xmath63 denotes a measurement in the @xmath154-basis . when operating the gate , we first measure all sites of the upper and lower lines in the @xmath63-eigenbasis . in case
the result for the sites at position `` 0 '' ( refer to labeling above ) is different from @xmath67 , the gate failed . in that case
all sites on the middle line are measured in the computational basis and we restart the procedure five steps to the right .
otherwise , the systems labeled by a @xmath44 are measured .
we accept the outcome only if we obtained @xmath48 on sites @xmath155 and @xmath78 on sites @xmath156 should a different result occur , the gate is once again considered a failure and we proceed as above .
lastly , the @xmath64 measurement on the central site is performed . in case of a result corresponding to @xmath78
, it is easy to see that no interaction between the upper and the lower part takes place , so this is the last possibility for the gate to fail . let us assume now that the desired measurement outcomes were realized . at site @xmath157 on the middle line
, we obtained @xmath14{a[1]}}\ar[r ] & } } , \ ] ] which prepares the correlation system of the middle line in @xmath78 . at site @xmath59 , in turn , a hadamard gate has been realized , which causes the output of site @xmath59 to be @xmath158 .
the situation is similar on the r.h.s .
, so that the above network at site @xmath143 can be re - written as @xmath159 & { * + [ f]{a[+]}}\ar[r ] & \\ { * + [ f]{+}}{\ar@{-}[r]}&{*+[f]{a[y]}}{\ar@{-}[r]}\ar@{-}[u ] & { * + [ f]{+}}\\ \ar[r ] & { * + [ f]{a[+]}}\ar[r]\ar@{-}[u ] & \\ } } \end{xy}.\ ] ] we will now analyze the tensor network in eq.([eqn : akltentangling ] ) step by step . for proving its functionality , there is no loss of generality in restricting attention to the situation where the correlation system of the lower line is initially in state @xmath160 , for @xmath161 .
we compute for the lower part of the tensor network @xmath162{{\mbox{$| c \rangle$}}}}\ar@{-}[r]&{*+[f]{a[+]}}\ar[r]\ar[u ] & } } \end{xy } = x{\mbox{$| c \rangle$}}_r z^c { \mbox{$| + \rangle$}}_u.\ ] ] further , plugging the output @xmath163 of the lower stage into the middle part , we find @xmath162{+}}{\ar@{-}[r]}&{*+[f]{a[y]}}{\ar@{-}[r]}\ar[u ] & { * + [ f]{+}}\\ & { * + [ f]{z^c{\mbox{$| + \rangle$}}}}\ar@{-}[u ] } } \end{xy } \propto z^{c+y } ( { \mbox{$| 0 \rangle$}}+i{\mbox{$| 1 \rangle$}}),\ ] ] where @xmath164 reflects the outcome of the @xmath64-measurement on the central site : @xmath165 in case of @xmath166 and @xmath167 for @xmath168 .
lastly , @xmath169 & { * + [ f]{a[+]}}\ar[r ] & \\ & { * + [ f]{z^{c+y } ( { \mbox{$| 0 \rangle$}}+i{\mbox{$| 1 \rangle$}})}}\ar@{-}[u ] } } \end{xy } \propto s z^{c+y } x.\ ] ] in summary , the evolution afforded on the upper line is @xmath170 , equivalent to @xmath171 up to by - products .
this completes the proof of universality . for completeness ,
note that we never need the by - products to vanish for all logical qubits of the full computation simultaneously .
hence the expected number of steps for the realization of one- or two - qubit gates is a constant in the number of total logical qubits . in the following , we present two mbqc resource states which are motivated by kitaev s toric code states @xcite
this contrasts with a result in ref .
@xcite that mbqc on the planar toric code state itself can be simulated efficiently classically .
different from the other schemes presented , the natural gate in these schemes is a two - qubit interaction , whereas local operations have to be implemented indirectly . also , individual qubits are decoupled not by erasing sites but by switching off the coupling between them .
toric code states are states with non - trivial topological properties and have been introduced in the context of quantum error correction .
they have a particularly simple representation in terms of peps @xcite or ctns @xcite on two centered square lattices , @xmath172 } & { \ar@{-}[d ] } & & & \\ & { \ar@{-}[r]}{\ar@{-}[d ] } & { * + [ f]{k_v}}{\ar@{-}[r]}{\ar@{-}[d ] } & { * + [ f]{k_h}}{\ar@{-}[r]}{\ar@{-}[d ] } & { \ar@{-}[d ] } & & \\ { \ar@{-}[r ] } & { * + [ f]{k_v}}{\ar@{-}[r]}{\ar@{-}[d ] } & { * + [ f]{k_h}}{\ar@{-}[r]}{\ar@{-}[d ] } & { * + [ f]{k_v}}{\ar@{-}[r]}{\ar@{-}[d ] } & { * + [ f]{k_h}}{\ar@{-}[r]}{\ar@{-}[d ] } & { \ar@{-}[d ] } & \\ & { \ar@{-}[r ] } & { * + [ f]{k_v}}{\ar@{-}[r]}{\ar@{-}[d ] } & { * + [ f]{k_h}}{\ar@{-}[r]}{\ar@{-}[d ] } & { * + [ f]{k_v}}{\ar@{-}[r]}{\ar@{-}[d ] } & { * + [ f]{k_h}}{\ar@{-}[r]}{\ar@{-}[d ] } & \\ & & { \ar@{-}[r ] } & { * + [ f]{k_v}}{\ar@{-}[r]}{\ar@{-}[d ] } & { * + [ f]{k_h}}{\ar@{-}[r]}{\ar@{-}[d ] } & & \\ & & & & & & } } \end{xy}\ ] ] where @xmath173 } & & \\ & { * + [ f]{k_h[s]}}{\ar@{-}[ru]}{\ar@{-}[rd ] } & \\ { \ar@{-}[ru ] } & & } } \end{xy}&= & \begin{xy } * !
c\xybox{\xymatrix@c=3mm@r=3 mm { { \ar@{-}[rd ] } & & \\ & { * + [ f]{z^s } } { \ar@{-}[ru ] } & \\ & { * + [ f]{z^s } } { \ar@{-}[rd ] } & \\ { \ar@{-}[ru ] } & & } } \end{xy}\end{aligned}\ ] ] and @xmath173 } & & \\ & { * + [ f]{k_v[s]}}{\ar@{-}[ru]}{\ar@{-}[rd ] } & \\ { \ar@{-}[ru ] } & & } } \end{xy}&= & \begin{xy } * !
c\xybox{\xymatrix@c=3mm@r=3 mm { { \ar@{-}[rd ] } & & & \\ & { * + [ f]{z^s } } & { * + [ f]{z^s}}{\ar@{-}[rd]}{\ar@{-}[ru ] } & \\ { \ar@{-}[ru ] } & & & } } \end{xy}\ , \end{aligned}\ ] ] i.e. , @xmath174 and @xmath175 are identical up to a rotation by @xmath176 degrees .
let us first see how @xmath174 acts on two qubits in correlation space coming from the left .
the most basic operation is a measurement in the computational basis , which simply transports both qubits to the right ( up to a correlated @xmath44 by - product operator ) .
generalizing this to measurements in the @xmath64-@xmath44 plane , we find that @xmath177 & & \\ & { * + [ f]{k_h[\phi]}}\ar[ru]\ar[rd ] & \\ \ar[ru ] & & } } \end{xy}&= & \begin{xy } * ! c\xybox{\xymatrix@c=3mm@r=3 mm { \ar[rd ] & & \\ & { * + [ f]{zz(\phi)}}\ar[ru]\ar[rd ] & \\ \ar[ru ] & & } } \end{xy}\end{aligned}\ ] ] where @xmath178 is the angle with the @xmath44 axis , and @xmath179 ( note that this gate is locally equivalent to the cnot gate for @xmath180 . )
thus , the tensors in kitaev s toric code state have a _ two_-qubit operation as their natural gate in correlation space , rather than a _ single_-qubit gate . in mbqc schemes which base on these projectors ,
two - qubit gates are easy to realize , whereas in order to get one - qubit gates , tricks have to be used . in the first example , we obtain single - qubit operations by introducing ancillae : a @xmath181 controlled phase between a logical qubit and an ancilla in a computational basis state yields a local @xmath44 rotation on the logical qubit . in the second example , we use a different approach : we encode each logical qubit in _ two _ qubits in correlation space . using this nonlocal encoding ,
we obtain an easy implementation of both one- and two - qubit operations ; furthermore , the scheme allows for an arbitrary parallelization of the two - qubit interactions .
[ obs : kitaev ] there is no need to have a one - one correspondance between logical qubits and a single correlation system .
our first scheme consists of the modified tensor @xmath177 & & \\ & { * + [ f]{\tilde k_h[s]}}\ar[ru]\ar[rd ] & \\ \ar[ru ] & & } } \end{xy}&= & \begin{xy } * ! c\xybox{\xymatrix@c=3mm@r=3 mm { \ar[rd ] & & \\ & { * + [ f]{k_h[s]}}\ar[ru]\ar@{-}[rd ] & & \\ \ar[ru ] & & { * + [ f]{\sqrt{z}h}}\ar[rd]&\\ & & & & } } \end{xy } \label{eq : nys : mod1}\\ & = & \begin{xy } * ! c\xybox{\xymatrix@c=3mm@r=3 mm { \ar[rd ] & & \\ & { * + [ f]{\hspace*{1.2em}z^s\hspace*{1.2em } } } \ar[ru ] & \\ & { * + [ f]{\sqrt{z}hz^s } } \ar[rd ] & \\ \ar[ru ] & & } } \end{xy}\nonumber\end{aligned}\ ] ] [ with @xmath182 , arranged as in ( [ eqn : nys : toriccode ] ) where _ both _ @xmath174 and @xmath175 are replaced by @xmath183 .
the extra @xmath71 serves the same purpose as in other schemes : it allows to leave the subspace of diagonal operations and thus to implement @xmath63 rotations .
the need for the @xmath184 will become clear later ; it is connected to the fact that @xmath185 in the following , we show how this state can be used for mbqc .
the qubits run from left to right in correlation space in zig - zag lines in eq .
( [ eqn : nys : toriccode ] ) ; for the illustration in fig .
[ fig : toric1 ] , we have straightened these lines , and marked the measurement - induced @xmath181 interactions coming from the @xmath186 $ ] in ( [ eq : nys : mod1 ] ) by ellipses .
( the difference between filled and non - filled ellipses will be explained later . )
the @xmath187 operations of ( [ eq : nys : mod1 ] ) do not depend on the measurement and are thus hard - wired ; note that the order is reversed as we are considering @xmath71 and @xmath188 as two independent operations in the circuit .
let us first impose that all qubits are initialized to @xmath78 ; this corresponds to a left boundary condition @xmath78 in correlation space .
we will discuss later how to initialize the scheme .
every second qubit is an ancilla which will be used to implement one - qubit operations .
we first discuss the case of no pauli errors , and show later how those can be dealt with .
the implementation of single - qubit operations is illustrated in fig .
[ fig : toric1]a . there
, each ellipse denotes a possible @xmath181 interaction .
in particular , empty ellipses denote interactions which are switched off ( i.e. measured in the @xmath44 basis ) , while filled ellipses denote sites where one can measure in the @xmath64-@xmath44 plane to implement a @xmath181 gate .
if all interactions are switched off , all qubits are transported to the right , subject to the transformation @xmath187 . as @xmath189 ,
the ancillae are in the computational basis in every third step : these regions are hashed in fig .
[ fig : toric1]a . in these regions ,
a @xmath190 between ancilla and logical qubit ( corresponding to the filled ellipses in the figure ) results in a single - qubit @xmath44 rotation on the latter .
thus , in each block of length three as the one shown in fig .
[ fig : toric1]a , the transformation @xmath191 is implemented [ where @xmath192 , which allows for arbitrary one - qubit operations . in fig .
[ fig : toric1]b , the corresponding circuit is shown , which has been simplified using @xmath193 , and that diagonal matrices commute .
although the scheme has a natural two - qubit interaction , implementing an interaction between two adjacent _ logical _ qubits is complicated by the ancilla which is located inbetween . in order to obtain a coupling
, we first swap the logical qubit with the ancilla , then couple it to the now adjacent logical neighbor , and finally swap it back .
this is implemented by the measurement pattern shown in fig .
[ fig : toric1]c . again , empty ellipses correspond to switched off interactions , while the filled ellipses all implement @xmath194 gates , each of which together with two @xmath184 and two hadamards as grouped in the figure gives a cnot gate , cf .
( [ eq : nys : cnot ] ) .
this measurement pattern corresponds to the circuit shown in fig .
[ fig : toric1]d , where we have replaced each pair of cnots by a cnot and a swap . by merging each cnot with the two adjacent hadamards ,
we effectively obtain @xmath45 gates .
we thus remain with only diagonal gates on the two lower qubits ( except for the swap ) , i.e. the gates all commute and the circuit can thus be simplified to the one shown on in fig .
[ fig : toric1]e , proving that the sequence effectively implements a two - qubit interaction between the logical qubits .
note that the length of the complete sequence is compatible with the three - periodicity of the basis of the ancillae .
pauli errors in this scheme can be dealt with as usual : @xmath71 and @xmath184 are both in the clifford group , i.e. , paulis can be commuted through , and @xmath181 commutes with @xmath44 errors , while @xmath195 . finally , we show how to read out the logical qubits .
it holds that @xmath196 } & & \\ & { * + [ f]{h[+]}}{\ar[ru]}{\ar[rd ] } & \\ { \ar[ru ] } & & } } \end{xy}&= & \left|\begin{array}{c}0\\0\end{array}\right\rangle \left\langle\begin{array}{c}0\\0\end{array}\right|+ \left|\begin{array}{c}1\\1\end{array}\right\rangle \left\langle\begin{array}{c}1\\1\end{array}\right|\ , \\
\begin{xy } * !
c\xybox{\xymatrix@c=3mm@r=3 mm { { \ar[rd ] } & & \\ & { * + [ f]{h[-]}}{\ar[ru]}{\ar[rd ] } & \\ { \ar[ru ] } & & } } \end{xy}&= & \left|\begin{array}{c}0\\1\end{array}\right\rangle \left\langle\begin{array}{c}0\\1\end{array}\right|+ \left|\begin{array}{c}1\\0\end{array}\right\rangle \left\langle\begin{array}{c}1\\0\end{array}\right|\ , \end{aligned}\ ] ] i.e. , a measurement in the @xmath63 basis returns the parity of the ancilla and the logical qubit . if this is done when the ancilla is in a computational basis state , one effectively measures the logical qubit in the computational basis .
note that both the ancilla and the logical qubit are in a well - defined state afterwards and can thus be reused .
let us now turn towards the initialization procedure .
in contrast to the previous mbqc schemes , the read - out can not be used for initialization .
the reason is that the read - out only works if the ancilla qubit is initially in a computational basis state ; otherwise , it just projects onto the subspace spanned by @xmath197 or by @xmath198 . in the following ,
we demonstrate that it is still possible to initialize this scheme by taking a different perspective on how it encodes logical qubits .
therefore , we group each logical qubit with the ancilla above ( e.g. , the first two qubits in fig .
[ fig : toric1]a ) , and encode the new logical qubit in their parity note that this is what is really measured in the read - out . the following calculations are most conveniently carried out in a bell basis where each state is described as @xmath199 , where the @xmath200 qubit stores the sign of the bell state and the @xmath25 qubit the parity and thus encodes our logical qubit , i.e. @xmath201 the circuit transforming between the above encoding and the qubits in correlation space is @xmath202{toric3n}\ \ .\ ] ] using this decoding , it is straightforward to investigate what happens in the various steps of the mbqc scheme .
firstly , one can easily check that by measuring two consecutive couplings of the qubit pair in the @xmath63 basis , one prepares them in a maximally entangled state @xmath203 up to pauli errors , corresponding to @xmath204 in the encoded system . by pretending a pauli @xmath44 error on one of the qubits with @xmath205
, we effectively face the mixture @xmath206 , corresponding to @xmath207 . since the transformation ( [ eq : magictrafo ] )
is in the clifford group , pauli errors remain pauli errors in the encoded system . in the following , we will check how the circuit acts on initial states @xmath208 , where the sign can be different on each pair .
as we will show , all of them give the same output statistics , and thus the same holds for their mixture , i.e. the actual initial state .
these considerations are illustrated in fig .
[ fig : toric2 ] , where we take the circuits of fig .
[ fig : toric1 ] and compose them with the decoding and encoding circuits ( boxed ) in order to determine their action on the encoded system .
firstly , a @xmath190 gate on a pair gives a @xmath44 rotation of the encoded logical qubit , since the action of @xmath190 only depends on the parity ( fig .
[ fig : toric2]a ) . the action of the second @xmath181 rotation of fig .
[ fig : toric1]b which originally gave an @xmath63 rotation is shown in fig . [
fig : toric2]b .
the right hand side is obtained by using @xmath209 , @xmath210 , the fact that diagonal operators commute , and @xmath211 .
as we see from the simplified circuit , we obtain an @xmath63 rotation on the upper logical qubit , but with the rotation direction determined by the state of the @xmath212 qubit below : while @xmath213 results in a rotation @xmath214 , the state @xmath215 gives @xmath216 similarly , the circuit for the coupling of two logical qubits can be simplified as in fig .
[ fig : toric2]c : again , the coupling on the logical qubits is @xmath217 or @xmath218 depending on whether the second @xmath200 qubit is @xmath213 or @xmath215 .
therefore , the error introduced by the unknown state of each @xmath200 qubit results in a @xmath44 correction around each operation on the logical qubit above ( note that we can assume this also for @xmath44 rotations as they commute with the @xmath44 correction ) .
although the error itself is unknown and different for each logical qubit , it is consistent within each qubit , as it is always determined by the same ancilla .
thus , two subsequent @xmath44 errors cancel out , and one remains only with one @xmath44 correction on the logical qubit at the beginning and one at the end of the sequence .
the former has no effect since the initial state is @xmath219 , while the latter has no effect either since the encoded logical qubit is finally measured in the computational basis .
thus , the output statistics for the circuit is independent of the initial state @xmath220 of the phase qubits , and one can equally well start from their mixture @xmath221 which completes the argument . the second toric - code - like scheme is based on a very different idea . therefore , observe that the @xmath175 tensor can be written as @xmath222 & & \\ & { * + [ f]{k_v[s]}}\ar[ru]\ar[rd ] & \\
\ar[ru ] & & } } \end{xy}= \begin{xy } * !
c\xybox{\xymatrix@c=3mm@r=3 mm { \ar[rd ] & & & & & \\ & { * + [ f]{\textsc{copy}^\dagger}}\ar@{-}[r ] & { * + [ f]{h}}\ar@{-}[r]&{*+[f]{a[s]}}\ar@{-}[r ] & { * + [ f]{\textsc{copy}}}\ar[rd]\ar[ru ] & \\ \ar[ru ] & & & & & } } \end{xy}\ ] ] where @xmath223 is the copy gate latexmath:[$|0,0\rangle\langle0|+ no physical system associated to them ) , and @xmath12 the 1-d cluster projector , cf . eqs .
( [ eqn : clustermatrix0 ] ) and ( [ eqn : clustermatrix1 ] ) .
thus , @xmath175 takes two qubits in correlation space , projects them onto the @xmath225 subspace , implements the 1-d cluster map up to a hadamard , and duplicates the output to two qubits .
concatenating these tensors horizontally [ this takes place in ( [ eqn : nys : toriccode ] ) if all @xmath174 s are measured in @xmath44 , and one neglects pauli errors ] therefore implements a single logical qubit line , encoded in two qubits in correlation space . by removing the hadamard gate from @xmath175
, we obtain a 1-d cluster state encoded in two qubits which is thus capable of implementing any one - qubit operation on the logical qubit ; in particular , this includes intialization and read - out .
we thus define the tensor @xmath222 & & \\ & { * + [ f]{\tilde k_v[s]}}\ar[ru]\ar[rd ] & \\ \ar[ru ] & & } } \end{xy}= \begin{xy } * ! c\xybox{\xymatrix@c=3mm@r=3 mm { \ar[rd ] & & & & \\ &
{ * + [ f]{\textsc{copy}^\dagger}}\ar@{-}[r ] & { * + [ f]{a[s]}}\ar@{-}[r ] & { * + [ f]{\textsc{copy}}}\ar[rd]\ar[ru ] & \\ \ar[ru ] & & & & } } \end{xy}\ .\ ] ] then , the toric code state ( [ eqn : nys : toriccode ] ) with @xmath175 replaced by @xmath226 is universal for mbqc : initialization , one - qubit operations , and read - out are done exacly as in the 1-d cluster state .
the logical qubits are decoupled up to @xmath44 by - product operators in correlation space by measuring the @xmath174 tensors in the @xmath44 basis .
the @xmath44 by - products in correlation space correspond to @xmath44 errors on the encoded logical qubits and thus can again be dealt with as in the cluster . in order to couple two logical qubits ,
we measure a @xmath174 tensor in the @xmath64 basis and obtain a @xmath181 controlled phase gate in correlation space , which translates to the same gate on the logical qubits . note that this model has the additional feature that as as many controlled phases ( between nearest neighbors ) as desired can be implemented simultaneously . in the light of the discussion on the initialization of the first scheme
, one might see similarities between the two schemes , since in both cases the information is effectively encoded in pairs of qubits .
note however that in the first scheme , the information is stored in the parity of the two qubits , and the full @xmath227-dimensional space is being used ; the reason for this encoding came from the properties of the @xmath174 tensor used as a map in horizontal direction .
in contrast , the second scheme only populates the @xmath142-dimensional even parity subspace , and the qubit is rather stored in two copies of the same state ; finally , the encoding is motivated by the properties of the @xmath175 tensor as a map on correlation space in horizontal direction . in this section
, we will consider instances of _ weighted graph states _
@xcite forming universal resources . to motivate the construction , recall that the cluster state can be prepared by applying a controlled - phase gate @xmath228 with phase @xmath229 between any two nearest neighbors of a two - dimensional lattice of qubits initially in the state @xmath67 .
if one wants to physically implement this operation using _ linear optics _
@xcite , one encounters the situation that the controlled phase gate can be implemented only probabilistically , with the probability of success decreasing as @xmath178 increases .
it is hence natural to ask whether one can build a universal resource using gates @xmath230 , in order to minimize the probability of failure weighted graph state as a universal resource .
solid lines correspond to edges that have been entangled using phase gates with phase @xmath229 , dotted lines correspond to edges entangled with phase gates with @xmath231 .
this shows that one can replace some edges with weakly entangled bonds.,width=241 ] expanding the discussion presented in ref .
@xcite , we treat the weighted graph state shown in fig .
[ fig : weighted ] .
a tensor network representation of these states can be derived along the same lines as for the original cluster in section [ sec:1dcluster ] . set @xmath232 .
the relevant tensors are given by @xmath233&{*+[f]{a[0]}}\ar[lu]\ar[r]\ar[ru ] & \\
\ar[ru ] & & \ar[lu ] } } \end{xy } & = & { \mbox{$| + \rangle$}}_{ru}\,{\mbox{$| + \rangle$}}_{lu}\,{\mbox{$| + \rangle$}}_r{\mbox{$\langle 0 |$}}_{ld}{\mbox{$\langle 0 |$}}_{rd}{\mbox{$\langle 0 |$}}_l , \\
\nonumber\\ \hspace{-1 mm } \begin{xy } * !
c\xybox{\xymatrix@c=3mm@r=1 mm { & & \\
\ar[r]&{*+[f]{a[1]}}\ar[lu]\ar[r]\ar[ru ] & \\
\ar[ru ] & & \ar[lu ] } } \end{xy } & = & { \mbox{$| i \rangle$}}_{ru}\,{\mbox{$| i \rangle$}}_{lu}\,{\mbox{$| - \rangle$}}_r{\mbox{$\langle 1 |$}}_{ld}{\mbox{$\langle 1 |$}}_{rd}{\mbox{$\langle 1 |$}}_l .
\end{aligned}\ ] ] indices are labeled @xmath234 for `` right - up '' to @xmath235 for `` left - down '' . the boundary conditions are @xmath78 for the @xmath236-directions ; @xmath67 otherwise .
we will first describe how to realize isolated evolutions of single logical qubits in the sense of fig .
[ fig : flow ] .
again the strategy will be to measure the sites of one horizontal line of the lattice in the @xmath63-basis and all vertically adjacent systems in the @xmath44-basis .
the analysis of the situation proceeds in perfect analogy to the one given in section [ sec:2dcluster ] .
one obtains @xmath237{a[z_{i-1,u } ] } } & & { * + [ f]{a[z_{i+1,u } ] } } \\
\ar[r ] & { * + [ f]{a[x_i]}}\ar@{-}[lu]\ar@{-}[ru]\ar[r ] & \\ { * + [ f]{a[z_{i-1,d}]}}\ar@{-}[ru ] & & { * + [ f]{a[z_{i+1,d}]}}\ar@{-}[lu ] \\ } } \end{xy } = h s^{2x_i+z_i},\ ] ] where @xmath238 and @xmath239 denotes the _
@xmath240 gate_. the operators @xmath71 and @xmath241 generate the 24-element single qubit clifford group .
following the approach of section [ sec : aklt ] , we take this as the model s by - product group .
now choose some phase @xmath178 .
re - doing the calculation which led to eq .
( [ eqn:2dto1d ] ) , where we now measure in the @xmath242-basis instead of @xmath63 on the central node , shows that the evolution of the correlation space is given by @xmath134 , up to by - products . in complete analogy to section [ sec : aklt ] , we see that the model allows for the realization of arbitrary @xmath140 operations .
how to prepare the state of the correlation system for a single horizontal line and how to read read it out has already been discussed in section [ sec:1dcluster ] .
hence the only piece missing for universal quantum computation is a single entangling two - qubit gate .
the schematics for a controlled-@xmath44 gate between two horizontal lines in the lattice are given below .
we implicitly assume that all adjacent sites not shown are measured in the @xmath44-basis , @xmath243&{*+[f]{a[x]}}\ar@{-}[r ] & { * + [ f]{a[x]}}\ar@{-}[r ] & { * + [ f]{a[x]}}\ar[r ] & \\ & & { * + [ f]{a[y]}}\ar@{-}[lu]\ar@{-}[ru]\\ \ar[r]&{*+[f]{a[x]}}\ar@{-}[r]\ar@{-}[ru]&{*+[f]{a[x]}}\ar@{-}[r]&{*+[f]{a[x]}}\ar@{-}[lu]\ar[r ] & } } \end{xy}.\ ] ] the measurement scheme realizes a controlled-@xmath44 gate , where the correlation system of the lower line carries the control qubit and the upper line the target qubit . in detail one would proceed as follows :
first one performs the @xmath63-measurements on the sites shown and the @xmath44-measurements on the adjacent ones .
if any of these measurements yields the result `` @xmath81 '' , we apply a @xmath44-measurement to the central site and restart the procedure three sites to the right .
this approach has been chosen for convenience : it allows us to forget about possible phases introduced by other measurement outcomes .
still , the `` correct '' result will occur after a finite expected number of steps , so the overhead caused due to this simplification is only linear .
it is also not hard to see that most other outcomes can be compensated for
so for practical purposes the scheme could be vastly optimized .
now assume that all measurements yielded `` @xmath143 '' .
then a @xmath64-measurement is performed on the central site , obtaining the result @xmath244 . as we did in section [ sec : aklt2d ] , we assume that the ( lower ) control line is in the basis state @xmath160 , for @xmath161 .
the contraction of the lower - most three tensors gives @xmath245{{\mbox{$| c \rangle$}}}}\ar@{-}[r]&{*+[f]{a[x]}}\ar[u]\ar@{-}[r ] & { * + [ f]{a[x]}}\ar@{-}[r]&{*+[f]{a[x]}}\ar[r]\ar[u ] & \\ } } \end{xy } \\ \nonumber\\ & = & s^c { \mbox{$| + \rangle$}}_{lu}s^c { \mbox{$| + \rangle$}}_{ru } h{\mbox{$| c \rangle$}}_r , \nonumber\end{aligned}\ ] ] where as before @xmath246 .
we plug this result into the @xmath247 $ ] tensor : @xmath248{a[y]}}\ar[lu]\ar[ru]\\ { * + [ f]{s^c{\mbox{$| + \rangle$}}}}\ar@{-}[ru ] & & { * + [ f]{s^c{\mbox{$| + \rangle$}}}}\ar@{-}[lu ] } } \end{xy } \hspace{-5 mm } \\
\nonumber\\ & = & { \mbox{$| + \rangle$}}_{lu}\,{\mbox{$| + \rangle$}}_{ru } + ( -1)^{c+y}i(s\otimes s){\mbox{$| + \rangle$}}_{lu}\,{\mbox{$| + \rangle$}}_{ru}. \nonumber\end{aligned}\ ] ]
lastly , for @xmath249 , @xmath250&{*+[f]{a[x]}}\ar@{-}[r ] & { * + [ f]{a[x]}}\ar@{-}[r ] & { * + [ f]{a[x]}}\ar[r ] & \\ & { * + [ f]{s^x{\mbox{$| + \rangle$}}}}\ar@{-}[u ] & & { * + [ f]{s^x{\mbox{$| + \rangle$}}}}\ar@{-}[u ] } } \end{xy } = h z^x.\end{aligned}\ ] ] hence , the evolution on the upper line is @xmath251 equivalent to @xmath171 up to by - products .
we arrive hence at the following conclusion : [ obs : entanglement ] universal resouces may be prepared using commuting gates with non - maximal entangling power .
we will consider a second weighted graph state to exemplify yet another novel ingredient that one can make use of in measurement - based quantum computation : one can think of quantum information being transported in the correlation system of some systems on the lattice forming `` wires '' , in a way that gates are realized by bringing the `` wires '' together .
this is an element that is not present in the original one - way computer .
the subsequent example of a resource state has not been chosen for its plausibility in the preparation in a physical context , but in a way such that this idea of `` rerouting quantum information '' can very transparently be explained , see fig.[fig : rerouting ] .
weighted graph state where the gate is achieved by appropriately bringing two wires together in a `` rerouting process''.,width=226 ] the resource that we think about is defined by tensors that are fully translationally invariant in one dimension , and has period two in the orthogonal dimension , @xmath252 & \ar@{-}[d]&\ar@{-}[d ] & \ar@{-}[d]&\ar@{-}[d ] & \\
\ar@{-}[r]&{*+[f]{a}}\ar@{-}[r ] & { * + [ f]{a}}\ar@{-}[r]&{*+[f]{a}}\ar@{-}[r ] & { * + [ f]{a}}\ar@{-}[r]&{*+[f]{a}}\ar@{-}[r ] & \\ \ar@{-}[r]&{*+[f]{b}}\ar@{-}[r]\ar@{-}[u]\ar@{-}[d]&{*+[f]{b}}\ar@{-}[u]\ar@{-}[d]\ar@{-}[r]&{*+[f]{b}}\ar@{-}[u]\ar@{-}[d]\ar@{-}[r ] & { * + [ f]{b}}\ar@{-}[u]\ar@{-}[d]\ar@{-}[r]&{*+[f]{b}}\ar@{-}[u]\ar@{-}[d]\ar@{-}[r]&\\ \ar@{-}[r]&{*+[f]{a}}\ar@{-}[r]\ar@{-}[d ] & { * + [ f]{a}}\ar@{-}[d]\ar@{-}[r]&{*+[f]{a}}\ar@{-}[d]\ar@{-}[r ] & { * + [ f]{a}}\ar@{-}[d]\ar@{-}[r]&{*+[f]{a}}\ar@{-}[d]\ar@{-}[r ] & \\ & & & & & & } } \end{xy}.\ ] ] this is , we have two kinds of tensors : one set is given by @xmath253&{*+[f]{b[0]}}\ar[u]\ar[r ] & \\ & \ar[u ] & } } \end{xy } = { \mbox{$| + \rangle$}}_r{\mbox{$| + \rangle$}}_u\,{\mbox{$\langle 0 |$}}_l{\mbox{$\langle 0 |$}}_d , \\
\label{eqn : rerouting1 } \begin{xy } * !
c\xybox{\xymatrix@c=3mm@r=3 mm { & & \\
\ar[r]&{*+[f]{b[1]}}\ar[u]\ar[r ] & \\ & \ar[u]&\ } } \end{xy } = { \mbox{$| - \rangle$}}_r{\mbox{$| i \rangle$}}_u\,{\mbox{$\langle 1 |$}}_l{\mbox{$\langle 1 |$}}_d\end{aligned}\ ] ] whereas the other one is nothing but the familiar one for a 2-d cluster state as in eqs . ( [ eqn : rerouting0 ] , [ eqn : rerouting1 ] ) , with boundary conditions @xmath254 the resulting state is hence again a weighted graph state , where in one dimension every second edge is replaced by an edge prepared using a gate with non - maximal entangling power .
then , it is not difficult to see that , again with @xmath255 , @xmath256{a[z_u]}}\\ \ar[r]&{*+[f]{b[x]}}\ar@{-}[u]\ar[r]&\\ & { * + [ f]{a[z_d]}}\ar@{-}[u]\\ } } \end{xy } = hz^{x+z_d}s^{z_u},\ ] ] and @xmath256{b[z_u]}}\\ \ar[r]&{*+[f]{a[x]}}\ar@{-}[u]\ar[r]&\\ & { * + [ f]{b[z_d]}}\ar@{-}[u]\\ & \ar[u ] & } } \end{xy } = hz^{x+z_u}s^{z_d}.\ ] ] similarly , we can consider several corner elements in this resource .
we obtain @xmath257&{*+[f]{a[x]}}\ar[u]&{*+[f]{a[z_r]}}\ar@{-}[l]\\ & { * + [ f]{b[z_d]}}\ar@{-}[u]&\\ } } \end{xy } = hz^{x+z_d}s^{z_u},\ ] ] and similarly @xmath258 & \\ { * + [ f]{a[z_l]}}\ar[r]&{*+[f]{a[x]}}\ar[r]&\\ & { * + [ f]{b[z_d]}}\ar@{-}[u]&\\ } } \end{xy } & = & ( hsh)^{z_d } x^{z_l+x},\\[.2 cm ] \begin{xy } * !
c\xybox{\xymatrix@c=3mm@r=3 mm { & { * + [ f]{a[z_u ] } } \ar@{-}[d]&\\ \ar[r]&{*+[f]{b[x]}}\ar[d]&{*+[f]{b[z_r]}}\ar@{-}[l]\\ & & \\ } } \end{xy } & = & z^{x+z_r } s^{z_u},\\ \begin{xy } * !
c\xybox{\xymatrix@c=3mm@r=3 mm { & { * + [ f]{a[z_u ] } } \ar@{-}[d]&\\ { * + [ f]{b[z_l]}}\ar@{-}[r]&{*+[f]{b[x]}}\ar[r ] & \\ & \ar[u]&\\ } } \end{xy } & = & h z^{x+z_u+z_l } ( sz)^{z_u},\end{aligned}\ ] ] where we have again made use of the convention that @xmath73 corresponds to @xmath74 and @xmath75 to @xmath259 .
we need one more ingredient to the scheme , this is @xmath260{b[z_l]}}\ar@{-}[r]&{*+[f]{b[0]}}\ar[u]\ar[r]&\\ & \ar[u ] & } } \end{xy } & = & |+\rangle_r |+\rangle_u\langle 0|_d,\\ \begin{xy } * !
c\xybox{\xymatrix@c=3mm@r=3 mm { & & & \\ { * + [ f]{b[z_l]}}\ar@{-}[r]&{*+[f]{b[1]}}\ar[u]\ar[r]&\\ & \ar[u ] & } } \end{xy } & = & |-\rangle_r |i\rangle_u\langle 1|_d,\end{aligned}\ ] ] and @xmath261&{*+[f]{a[0]}}\ar@{-}[r]\ar[u]&{*+[f]{a[z_r]}}&\\ & \ar[u ] & & } } \end{xy } & = & |+\rangle_u\langle 0|_l \langle 0|_d,\\ \begin{xy } * ! c\xybox{\xymatrix@c=3mm@r=3 mm { & & & \\
\ar[r]&{*+[f]{a[1]}}\ar@{-}[r]\ar[u]&{*+[f]{a[z_r]}}&\\ & \ar[u ] & & } } \end{xy } & = & ( -1)^{z_r}|-\rangle_u\langle 1|_l \langle 1|_d.\end{aligned}\ ] ] putting these ingredients , and following an argument similar to the last subsection , we find that up to clifford group by - products , we can transport along the horizontal lines for both kinds of local tensors . we can also use the corner pieces to reroute as depicted in fig.[fig : rerouting ] , and bring routes together forming a `` gate '' imprinted in the lattice , actually , a controlled-@xmath241 gate .
it should be noted that it is not obviously possible to faithfully transport one qubit of information vertically through the resource .
loosely speaking , the entanglement between a site of type b and the site of type a directly above it is non - maximal ( this is indicated by dotted lines in fig . [
fig : rerouting ] ) .
interestingly , one can still perform a ( non - maximally entangling ) non - local gate over this connection .
gates in measurement - based quantum computation can be achieved by means of appropriate routing of quantum information in the lattice .
we will very briefly sketch a matrix product state on a 1-d chain of qubits , which i ) exhibits non - vanishing two - point correlation functions , ii ) allows for any unitary to be realized in its correlation system and iii ) can be coupled to a universal 2-d resource in a way very similar to the aklt - type example ( section [ sec : aklt ] ) .
the discussion will be somewhat superficial
however , given the extensive discussion of other models above , the reader should have no problems filling in the details .
choose an integer @xmath262 and define @xmath263 up to a constant , @xmath264 is a @xmath122-th root of @xmath63 .
the state is defined by the following relations : @xmath265&{*+[f]{a[s]}}\ar[r ] & } & = & { \mbox{$| s \rangle$}}_r { \mbox{$\langle s |$}}_l g,\end{aligned}\ ] ] and @xmath266 the two - point correlation functions for measurements on this state never vanish completely .
indeed , in appendix [ sec : correlations ] it will be shown that @xmath267 where @xmath268 . for @xmath63-measurements , we find @xmath269&{*+[f]{a[x]}}\ar[r ] & } } & = & z^x g\end{aligned}\ ] ] pursuing the strategy introduced in section [ sec : randomness ] , we set the by - product group to @xmath270 , so the group generated by @xmath44 and @xmath264 .
one can easily verify that @xmath120 is indeed a finite group , equivalent to the _ dihedral group _ of order @xmath271 .
it is now straight - forward to check that i ) measurements in the computational basis can be used for preparation and read - out ( as in section [ sec:1dcluster ] ) , ii ) general local unitaries can be realized by means of measurements in the equatorial plane of the bloch sphere ( as in section [ sec:1daklt ] ) and iii ) a 2-d resource is obtainable in a fashion similar to the one presented in section [ sec:2daklt ] . with similar methods , one can also find qubit resource states that have a local entropy smaller than unity . for completeness
, we mention yet another kind of resource : this is an imperfect cluster state where some edges are missing .
such a setting is clearly relevant in a number of physical situations : if the underlying quantum gates building up the cluster state are fundamentally probabilistic , such as in linear optical architectures , then one very naturally arrives at this situation when one aims at minimizing the need for feed - forward .
a similar situation is encountered in cold atoms in optical lattices , when in a mott state exhibiting hole defects some atoms are missing .
we do not present details of such arguments , which have been considered in ref.@xcite , based on ideas of _ edge percolation _ and renormalization @xcite .
we merely state the result for completeness .
note also that results that may be similar to these ones have been announced in ref .
@xcite .
we consider the setting where one starts from a 2-d or 3-d cubic lattice of size @xmath272 .
two neighboring vertices on the lattice are connected with an edge with probability @xmath50 .
the stochastic variables deciding whether or not an edge is present are assumed to be uncorrelated .
if @xmath273 holds , then it is not difficult to see that one can extract a 2-d renormalized lattice of smaller size : this means that one can find a function @xmath274 , such that one arrives at a cubic @xmath275 array almost certainly as @xmath276 , with the following property : within each of the elements of this array , there is a central site that is connected to the central site of the neighboring array .
since all the additional sites can be removed by means of @xmath44-measurements , we can treat this resource effectively as a 2-d cluster state of dimension @xmath275 , and refer to this as a _
perfect sublattice_. this state will not necessarily be exactly a cluster state , as it may contain vertices having a vertex degree of three , but which will nevertheless function as a graph state resource just as the cluster state does ( for details , see ref .
@xcite ) . also , @xmath277 is arbitrarily close to being linear in @xmath1 asymptotically .
however , an even stronger statement holds : whenever @xmath278 , for any @xmath279 , one can find a function @xmath274 with the following property : starting from a sublattice of a 3-d cubic lattice of size @xmath280 , one can almost certainly prepare a perfect sublattice of size @xmath275 .
the asymptotic behavior of @xmath122 can be chosen to satisfy @xmath281 that is , with an overhead that is arbitrarily close to the optimal scaling , one can obtain a perfect resource state out of an imperfect one , even if one is merely above the percolation threshold for a three - dimensional lattice , and not only for the two - dimensional lattice , see fig . [
fig : percolation ] .
the latter argument is technically more involved than the former , for details , see ref.@xcite .
this shows , however , with methods unrelated to the ones considered primarily in the present work , that also random aspects in the resource as such can be dealt with .
cubic lattice of a graph state corresponding to the situation where some edges are missing in a cluster state .
if the probability @xmath50 of having an edge is sufficiently high the processes independent , then a renormalized perfect sublattice can be found almost certainly , giving rise to a cluster state of smaller size . if @xmath273 , where @xmath282 is the percolation threshold for edge percolation in 2-d cubic lattices , then a renormalized lattice can be found almost certainly .
interestingly , even if @xmath283 , @xmath284 denoting the percolation threshold in 3-d , one can almost certainly construct a perfect sublattice , using an overhead that is arbitrarily close to being quadratic.,width=170 ]
in the final section of this work , we will show that one can find resource states for mbqc that differ substantially from the cluster in various entanglement properties .
this will be done by encoding each system of a resource into several physical particles .
we will not develop any new computational models and make no use of the computational tensor network formalism introduced before .
the study of encoded resource states was initiated in ref .
@xcite and later pursued more systematically in ref .
@xcite .
more concretely , the following statements will be proved : there exists a family of universal resource states such that * the local entropy of entanglement is arbitrarily small , * the localizable entanglement is arbitrarily small and , more strongly , * the probability of succeeding in distilling a maximally entangled pair out of the resource is arbitrarily small , even if one does not a priori fix the two sites between which the pair will be established .
in particular , the resource can not be used as a state preparator .
we start from a cluster state vector on @xmath272 systems , denoted by @xmath89 , referred to as logical qubits . as in ref .
@xcite , we want to `` dilute '' the cluster state , i.e. encode it into a larger system , by means of invoking the codewords @xmath285 for some parameter @xmath286 .
the argument relies only on the choice of @xmath287 as a code word in that we focus on its implications on the localizable entanglement , and for that argument , the state vector @xmath287 has the desired properties of small local entropy and permutation invariance .
for encoded one - way computation to be possible , any state vector orthogonal to @xmath288 may be taken , compare also ref . @xcite .
every qubit of the cluster is subjected to the encoding operation @xmath289 yielding the _ diluted cluster _
a set of physical qubits corresponding to one cluster bit will be called a _ block_. as before , by a _ local measurement scheme _ we mean a sequence of adaptive local projective measurements , local to the physical systems .
let us first show again in more detail that such an encoding constitutes no obstacle to universal quantum computation .
each of the code words is orthogonal , and for computation to be possible , we need to do local dichotomic measurements in the logical space . by ref .
@xcite , any two pure orthogonal multi - partite states on @xmath286 qubits can be deterministically distinguished using locc . by making use of the construction of ref .
@xcite , this can be done by an appropriate ordered sequence of adapted projective measurements @xmath291 on the sites of each codeword , giving rise to an arbitrary projective dichotomic measurement with kraus operators @xmath292 in the logical space , @xmath293 and @xmath294 .
hence , one can translate any single - site measurement on a cluster state into an locc protocol for the encoded cluster .
this shows that @xmath9 is universal for deterministic mbc .
this is the argument of ref .
@xcite ( see also ref .
@xcite for a more detailed and extensive discussion on one - way computing based on encoded systems ) . in the following we are going to show in more detail that despite this property , we are heavily restricted to use this resource to prepare states with a significant amount of entanglement between two constituents .
in fact , we can not even distill a perfect maximally entangled qubit pair beyond any given probability of success .
this means that these states are universal resources , but on the level of physical systems utterly useless for state preparation .
the given resource is , needless to say , not meant as a particularly feasible resource .
instead , we aim at highlighting to what extent as such the entanglement properties can be relaxed , giving a guideline to more general settings .
note first that the localizable entanglement @xmath295 in these resources can easily be shown to be arbitrarily small : the entropy for a measurement in the computational basis reads @xmath296 , where @xmath297\rightarrow[0,1]$ ] is the standard binary entropy function . using the concavity of the entropy function
, we find @xmath298 such that @xmath299 .
this means that for two fixed sites , the rate at which one can distill maximally entangled pairs by performing measurements on the remaining systems is arbitrarily small .
this can be seen as follows .
we will aim at preparing a maximally entangled state between any two constituents of two different blocks .
it is easy to see that within the same block , the probability of success can be made arbitrarily small .
we hence look at a locc distillation scheme , a _ measurement - based scheme _
, taking the input @xmath300 and producing outputs @xmath301 with probability @xmath302 , @xmath303 .
this corresponds to a locc procedure , where each of the measurements may depend on all outcomes of the previous local measurements .
let us assume that outcomes labeled @xmath304 for some @xmath305 are successful in distilling a maximally entangled state .
we start by exploiting the permutation symmetry of the code words .
choose a block @xmath5 of @xmath290 .
assume there exists a measurement - based scheme with the property that with probability @xmath50 , the scheme will leave _ at least one _ system of block @xmath5 in a state of maximal local entropy .
then there exists a scheme such that with probability @xmath50 , the scheme will leave _ the first _ system of block @xmath5 in a state of maximal local entropy . at some point of time
the scheme is going to perform the first measurement on the @xmath5-th block .
because of permutation invariance , we may assume that it does so on the @xmath286-th system of the block .
the remaining state is still invariant under permutations of the first @xmath306 systems .
hence there is no loss of generality in assuming that the next measurement on the @xmath5-th block will be performed on the @xmath306-st system .
if the local entropy of any of the unmeasured systems is now maximal , then the same will be true for the first one once again , by permutation invariance .
also , it is easy to see that the probability @xmath50 that a measurement - based scheme will leave any system of block @xmath5 in a locally maximally mixed state is bounded from above by @xmath307 let @xmath308 be the initial probability of obtaining the outcome @xmath48 for a @xmath44 measurement on this qubit , @xmath309 . clearly , @xmath310 we consider now a local scheme potentially acting on all qubits except this distinguished one , with branches labeled @xmath311 , aiming at preparing this qubit in a maximally mixed state .
let @xmath312 be the probability of the qubit ending up in a locally maximally mixed state . in case of success , so in case of the preparation of a locally maximally entangled state , we have that @xmath313 , in case of failure @xmath314 . combining these inequalities
, we get @xmath315 we can hence show that there exists a family of universal resource states such that the probability that a local measurement scheme can prepare a maximally entangled qubit pair ( up to l.u .
equivalence ) out of any element of that family is strictly smaller than @xmath279 .
let @xmath316 be the probability that a site of block @xmath5 will end up as a part of a maximally entangled pair .
this means that when we fix the procedure , and label as before all sequences of measurement outcomes with @xmath311 , one does not perform measurements on all constituents .
let @xmath317 denote the index set labeling the cases where somewhere on the lattice a maximally entangled pair appears , so the probability @xmath50 for this to happen is bounded from above by @xmath318 according to the above bound , @xmath319 , giving a strict upper bound of @xmath320 for the overall probability of success .
the family @xmath321 for @xmath322 is clearly universal , involves only a linear overhead as compared to the original cluster state and satisfies the assumptions advertised above .
in this work , we have shown how to construct a plethora of novel models for measurement - based quantum computation .
our methods were taken from many - body theory .
the new models for quantum computation follow the paradigm of locally measuring single sites and hence abandoning any need for unitary control during the computation .
other than that , however , they can be quite different from the one - way model .
we have found models where the randomness is compensated in a novel manner , the length of the computation can be random , gates are performed by routing flows of quantum information towards one another , and logical information may be encoded in many correlation systems at the same time .
what is more , the resource states can in fact be radically different from the cluster states , in that they may display correlations as typical in ground states , can be weakly entangled . a number of properties of resource states that we found reasonable to assume to be necessary for a state to form a universal resource could be eventually relaxed .
so after all , it seems that much less is needed for measurement - based quantum computation than one could reasonably have anticipated .
this new degree of flexibility may well pave the way towards tailoring computational model towards many - body states that are particularly feasible to prepare , rather than trying to experimentally realize a specific model .
this work has benefited from fruitful discussions with a number of people , including k. audenaert , i. bloch , h .- j .
briegel , j.i .
cirac , c. dawson , w. dr , d. leung , a. miyake , m. van den nest , f. verstraete , m.b .
plenio , t. rudolph , m.m . wolf , and a. zeilinger .
it has been supported by the eu ( qap , qovaqial ) , the elite - netzwerk bayern , the epsrc , the qip - irc , microsoft research , and the euryi award scheme .
what is the value of the two - point correlation function @xmath323 ? in this work , we have only introduced the behavior of the correlation system when subject to a local measurement of a rank - one observable . however , in order to evaluate the correlation function , we need `` measure the identity '' on the intermediate systems or , equivalently , trace them out .
without going into the general theory @xcite , we just state that tracing out a system will cause the completely positive map @xmath324\rho a[i]^\dagger\ ] ] to act on the correlation system . for the cluster state , using the fact that the bases @xmath325 and @xmath326 are unbiased , we can easily show that @xmath327 is the completely depolarizing channel , sending any @xmath300 to @xmath328 .
this causes any correlation function to vanish for @xmath329 .
how does the situation look like for the state vector defined by eq.([eqn : gstate ] ) ?
we compute : @xmath330 so for @xmath331 : @xmath332 where @xmath333 and @xmath334 . in other words : when acting on the computational basis , @xmath335 implements a simple two - state markov process , which remains in the same state with probability @xmath50 and switches its state with probability @xmath336 .
now , @xmath337 equals @xmath338 if an even number of state changes occurred and @xmath157 if that number is odd .
so for the expectation value we find @xmath339 in section [ sec : aklt ] we discussed an aklt - type matrix product state .
it was claimed that the state constitutes the unique ground - state of a spin-1 nearest neighbor frustration free gapped hamiltonian .
it must be noted that in this work , we have not introduced the technical tools needed to cope with boundary effects at the end of the chain .
there are at least three ways to make the above statement rigorous : a ) treat the statement as being valid asymptotically in the limit of large chains , b ) work directly with infinite - volume states @xcite , or c ) look at sufficiently large rings with periodic boundary conditions @xcite .
once one chooses one of the options outlined above , the proof of this fact proceeds along the same lines as the one of the original aklt state , as presented in example 7 of ref .
@xcite ( see also ref .
@xcite ) . indeed , using the notions of refs .
@xcite one verifies that @xmath340 a[i_2 ] ) { \mbox{$| i_1,i_2 \rangle$}}&\end{aligned}\ ] ] is injective .
further , if @xmath341 , it is checked by direct computation that @xmath342 .
all claims follow as detailed in refs .
@xcite . in particular ,
let @xmath343 be a positive operator supported on the vector space spanned by : @xmath344 set @xmath345 , where @xmath346 translates its argument @xmath5 sites along the chain .
then @xmath71 is a non - degenerate , gapped , frustration free , nearest neighbor hamiltonian ( called _ parent hamiltonian _ in ref .
@xcite ) , whose energy is minimized by the state at hand .
m. hein , j. eisert , and h .- j .
briegel , phys .
a * 69 * , 062311 ( 2004 ) ; r. raussendorf , d.e .
browne , and h .- j .
briegel , ibid . * 68 * , 022312 ( 2003 ) ; d. schlingemann and r.f .
werner , ibid .
* 65 * , 012308 ( 2002 ) .
m. fannes , b. nachtergaele , and r.f .
werner , commun .
phys . * 144 * , 443 ( 1992 ) ; y.s .
stlund and s. rommer , phys .
lett . * 75 * , 3537 ( 1995 ) ; u. schollwck , rev .
phys . * 77 * , 259 ( 2005 ) ; d. perez - garcia , f. verstraete , m.m . wolf , and j.i .
cirac , quant - ph/0608197 ; j. eisert , phys .
lett . * 97 * , 260501 ( 2006 ) .
f. verstraete and j.i .
cirac , cond - mat/0407066 ; s. richter ( phd thesis , osnabrck , 1994 ) , supervised by r.f .
werner ; f. verstraete , m.m .
wolf , d. perez - garcia , j.i .
cirac , phys .
* 96 * , 220601 ( 2006 ) .
nielsen and i.l .
chuang , _ quantum computation and quantum information _
( cambridge university press , cambridge , 2000 ) ; j. eisert and m.m .
wolf , _ quantum computing _ , in _ handbook of nature - inspired and innovative computing _ ( springer , new york , 2006 ) .
j. eisert , k. jacobs , p. papadopoulos , and m.b .
plenio , phys .
a * 62 * , 052317 ( 2000 ) ; d. collins , n. linden , and s. popescu , ibid .
* 64 * , 032302 ( 2001 ) ; d. gottesman , _ the heisenberg representation of quantum computers _ , in s.p .
corney et .
xxii int .
group theor .
( international press , cambridge , 1999 ) ; j.i .
cirac , w. dr , b. kraus , and m. lewenstein , phys . rev .
lett . * 86 * , 544 ( 2001 ) . | we introduce novel schemes for quantum computing based on local measurements on entangled resource states .
this work elaborates on the framework established in [ phys .
rev .
lett . * 98 * , 220503 ( 2007 ) , quant - ph/0609149 ] .
our method makes use of tools from many - body physics
matrix product states , finitely correlated states or projected entangled pairs states to show how measurements on entangled states can be viewed as processing quantum information .
this work hence constitutes an instance where a quantum information problem how to realize quantum computation was approached using tools from many - body theory and not vice versa .
we give a more detailed description of the setting , and present a large number of new examples .
we find novel computational schemes , which differ from the original one - way computer for example in the way the randomness of measurement outcomes is handled .
also , schemes are presented where the logical qubits are no longer strictly localized on the resource state .
notably , we find a great flexibility in the properties of the universal resource states : they may for example exhibit non - vanishing long - range correlation functions or be locally arbitrarily close to a pure state .
we discuss variants of kitaev s toric code states as universal resources , and contrast this with situations where they can be efficiently classically simulated .
this framework opens up a way of thinking of tailoring resource states to specific physical systems , such as cold atoms in optical lattices or linear optical systems .
[ theorem]definition [ theorem]lemma [ theorem]corollary [ theorem]property [ theorem]proposition [ theorem]remark [ theorem]example [ theorem]assumption [ theorem]observation h o |
the magnetic nano - disk ( thickness @xmath5 nm , diameter @xmath6 m ) was patterned by standard e - beam lithography and ion - milling techniques from an extended film of nimnsb grown by molecular - beam epitaxy on an inp(001 ) substrate@xcite .
a 50 nm thick si@xmath47n@xmath48 cap layer was deposited on top of the disk for protection and a broadband coplanar microwave antenna ( 300 nm thick au ) was subsequently evaporated on top of the patterned disk .
the width of the antenna constriction above the disk is 5 @xmath26 m ( see supplementary figure 2 ) . injecting a microwave current from a synthesizer inside
the antenna produces an in - plane linearly polarized microwave magnetic field @xmath8 , oriented perpendicular to the stripe direction . in order to apply two consecutive pulses separated by a variable delay @xmath42 ,
the carrier of the cw excitation is split into two branches , each of them being gated by independent mixers before being recombined .
a phase shift can be introduced between the two branches using a delay line . the typical rise and fall times of pulses
are 0.8 ns and the control on the delay @xmath42 is better than 0.1 ns .
the calibration of the amplitude of @xmath8 yields the value @xmath49 mt with an error bar of 25% for a 0 dbm input power in the antenna .
therefore , the microwave power range in fig.2a corresponds to microwave field amplitude ranging from 0.2 to 0.7 mt , the power in double pulses experiments of fig.3a to @xmath50 mt , and the power employed to read the polarity to @xmath51 mt . a @xmath18 pulse with settings ( @xmath27 ns , @xmath52 , @xmath53 dbm)@xcite is used to set the initial polarity state to @xmath12 .
in fact , such an initialization pulse does not affect the @xmath12 state and it transforms @xmath14 into @xmath12 ( see fig.2a ) .
similarly , a @xmath17 pulse ( @xmath27 ns , @xmath54 , @xmath53 dbm ) is used to set the initial polarity state to @xmath14 .
the mrfm setup@xcite is located inside a vacuum chamber ( @xmath55 mbar ) and operates at room temperature .
the cantilever is an olympus biolever ( spring constant @xmath56 mn / m ) with a 800 nm diameter sphere of soft amorphous fe ( with 3% si ) glued to its apex .
mrfm spectroscopy is achieved by placing the centre of the magnetic spherical probe above the centre of the nimnsb nano - disk .
the separation between the sample and the probe is @xmath57 m .
the probe senses the dipolar force @xmath58 proportional to the perpendicular component @xmath59 of the magnetization of the nano - disk ( see fig.1a ) .
ferromagnetic resonance spectra are obtained as a function of the microwave excitation frequency at a fixed bias field @xmath13 .
the microwave modulation is a cyclic absorption sequence , where the microwave power is switched on and off at the cantilever resonance frequency , @xmath60 khz .
the mrfm signal originates from the cyclic diminution of @xmath59 of the nano - disk synchronous with the absorption of the microwave field@xcite . the resulting force modulated at the mechanical resonance of the cantilever force produces a cantilever vibration amplitude enhanced by its quality factor @xmath61 , that is optically detected .
the calculations presented in figs.2 and 4 are based on linearised thiele s equation@xcite , an effective equation of motion for the vortex core position @xmath62 in the disk plane : @xmath63=0 .
\label{thiele}\ ] ] the first term is the gyroforce ( @xmath64 is the gyrovector ) , the second one is the damping ( @xmath65 ) , the third one is the restoring force ( @xmath66 is the stiffness coefficient ) , and the last one is the zeeman energy with the spatially uniform external field @xmath67 ( @xmath68 is the unit vector normal to the disk plane and details on @xmath26 and previously mentioned coefficients can be found in refs.@xcite ) . the resonant frequency of vortex core gyration is @xmath69 . from eq.[thiele ] one can find the instantaneous position @xmath70 and the velocity @xmath71 of the vortex core in the presence or absence of the harmonic excitation field @xmath72 , as derived in the supplementary documents of ref.@xcite . if guslienko s criterion on critical velocity@xcite is met at some time of the simulated pulse sequence ( @xmath73 ) , the vortex core is assumed to have switched .
experimental sample size , gyrotropic frequencies @xmath28 and @xmath29 associated to each polarity , and non - linear redshift of the frequency ( 3% ) are used in our calculation .
the minimal excitation amplitude @xmath74 to reach @xmath75 by the end of the pulse depends@xcite on the excitation frequency @xmath76 and on the pulse duration @xmath19 .
minimization of the energy @xmath77 with respect to @xmath76 yields the dependence on @xmath19 of the optimal frequency @xmath23 plotted in fig.2d and of the optimal pulse energy @xmath78 plotted in fig.2e : @xmath79 in this expression , @xmath80 is an experimental conversion factor between the input power in the antenna and @xmath81 determined from calibration .
therefore , two independent parameters are used to adjust the calculation with the data in fig.2e : the damping ratio @xmath82 to fit the overall shape ( position of minimum ) and the the critical velocity @xmath75 to fit the absolute value of the energy .
the optimal pulse duration @xmath38 corresponding to the minimum energy @xmath83 follows from eq.[e * ] : @xmath84 ( for the blue solid line in fig.2e , @xmath31 , @xmath85 ns , hence @xmath83 is reached for @xmath86 ns ) .
the fitted critical velocity in our nimnsb disk is @xmath87 m / s , in good agreement with the expected value@xcite of @xmath88 m / s ( the exchange constant of nimnsb is @xmath89 pj / m@xcite ) .
the same parameters are used in calculations of figs.2 and 4 . in fig.4 ,
the only new fitting parameter is the characteristic time @xmath90 of the vortex core free decay ( the relaxation is assumed to be exponential ) .
we also allow a fine adjustment of the phase shift between the two pulses in order to get the best agreement with the data ( 25 in the `` zero phase shift '' experiment and 190 in the `` @xmath43 phase shift '' experiment ; these small differences are ascribed to an imperfect delay line in the experimental pulse setup ) .
m , thickness @xmath5 nm ) .
a soft cantilever , with a spherical magnetic probe attached at its end and placed at a distance @xmath57 m from the sample , detects mechanically the vortex core dynamics .
the bias magnetic field @xmath13 is applied perpendicularly to the disk plane whereas microwave pulses of linearly polarized field @xmath8 are applied in the plane . *
b * , zero - field magnetic force image of the magnetic vortex , where the dark spot at the disk centre reveals the core prepared in the @xmath14 ( * c * ) polarity state .
the opposite configuration is the @xmath12 ( * d * ) polarity state.,width=453 ] and @xmath18 pulses ( see panel * b * ) to reverse the vortex core at @xmath91 mt measured as a function of power and frequency ( stepped by increments of 0.3 dbm and 1.2 mhz ) . for each pulse type , a coloured pixel ( @xmath17 : red shade , @xmath18 : blue shade ) marks a successful reversal .
the transparency gives the switching probability averaged over 16 attempts .
experiments corresponding to three values of the pulse duration @xmath19 are displayed . for each @xmath19
we define ( @xmath23,@xmath24 ) , the optimal working point of the @xmath18 pulse located at the bottom of the corresponding contour plot ( see @xmath92 ) .
* c * , numerical calculation ( see methods ) of the experiments presented in * a*. this calculation is not valid within the shaded area , where multiple vortex core reversals can occur . *
d , e * , experimental ( @xmath92 ) and calculated ( lines ) dependences on @xmath19 of the optimal frequency @xmath23 ( * d * ) and of the optimal pulse energy @xmath25 ( * e * ) .
the experimental points are obtained from the analysis of data sets similar to those presented in * a * , where @xmath19 is varied from 1 @xmath26s down to 3 ns ( the three @xmath92 close to the minimum energy in * e * are inferred from those displayed in * a * ) .
the best agreement is obtained for a damping ratio @xmath31 , i.e. , a characteristic decay time @xmath93 ns .
the absolute value of the energy is also fitted in the calculation , which enables to extract the critical velocity for vortex core reversal , @xmath87 m / s ( see methods ) .
error bars on @xmath23 and @xmath78 are absolute minima and maxima resulting from experimental uncertainties in * a * on the optimal working point ( @xmath23 , @xmath24 ) associated to each pulse duration.,width=453 ] separating the two pulses ( @xmath39 ns , @xmath40 dbm ) and of the carrier microwave frequency ( stepped by increments of 3 ns and 4 mhz , respectively ) .
the bias magnetic field is @xmath91 mt . the initial polarity state is @xmath12 in the left graphs ( * a , c * ) and @xmath14 in the right graphs ( * b , d * ) . as depicted in the right - side panels , the phase difference between the two pulses is zero in the upper graphs ( * a , b * ) and @xmath43 in the lower graphs ( * c , d*).,width=453 ] ns in the free regime . * b * , associated vortex core trajectory ( left ) and velocity ( right ) vs. time plotted for two @xmath18 pulses with settings @xmath94 ns and @xmath44 mhz ( see pixel @xmath92 in * a * ) .
for these settings , the vortex core is reversed when the phase difference between the pulses is equal to zero ( top , see star ) and not reversed when it is equal to @xmath43 ( bottom).,width=623 ] t ) and of the gyrotropic mode in the vortex state ( @xmath95 t ) .
dashed lines are analytical expressions ( eqs . 1 and 2 in ref.@xcite ) . *
b * , dependence on frequency of the damping ratio obtained from mrfm spectroscopic measurements in the small oscillation regime ( the typical power level is @xmath96 dbm , corresponding to @xmath97 mt ) .
dashed lines are guides to the eye .
lorentzian fits of the resonance peaks ( inset shows such a fit at a fixed field in the vortex state ) yield the centre frequency @xmath20 and the frequency linewidth @xmath98 together with the associated error bars , hence the damping ratio @xmath99 .
the damping ratio in the saturated state is equal to the gilbert constant @xmath100 . in the vortex state , the damping ratio is renormalized by topology ( eq.13b in ref.@xcite ) .
experimentally , it increases from about 0.005 to 0.008 as the frequency decreases from 450 mhz down to 190 mhz ( i.e. , the perpendicular field decreases from @xmath101 t down to @xmath102 t ) .
close to @xmath103 , the experimental value @xmath104 corresponds to a decay time @xmath105 ns and is in good agreement with the predicted value@xcite @xmath106\simeq0.007 $ ] ( @xmath107 nm and @xmath108 nm are respectively the disk and vortex core radii).,width=340 ] m wide constriction under which the nimnsb disk ( diameter @xmath6 m , thickness @xmath5 nm ) is located ( see insert ) .
the microwave antenna is wire bounded to a coplanar waveguide on the right side.,width=340 ] | in a vortex - state magnetic nano - disk@xcite , the static magnetization is curling in the plane , except in the core region where it is pointing out - of - plane@xcite , either up or down leading to two possible stable states of opposite core polarity @xmath0 .
dynamical reversal of @xmath0 by large amplitude motion of the vortex core has recently been demonstrated experimentally@xcite , raising fundamental interest@xcite for potential application in magnetic storage devices@xcite .
here we demonstrate coherent control of @xmath0 by single and double microwave pulse sequences , taking advantage of the resonant vortex dynamics in a perpendicular bias magnetic field@xcite .
optimization of the microwave pulse duration required to switch @xmath0 allows to experimentally infer the characteristic decay time of the vortex core in the large oscillation regime .
it is found to be more than twice shorter than in the small oscillation regime , raising the fundamental question of the non - linear behaviour of magnetic dissipation .
magnetic vortices are topological solitons with rich dynamical properties .
the lowest energy excitation of the vortex ground state is the so - called gyrotropic mode@xcite , corresponding to the gyration of the vortex core around its equilibrium position with a frequency in the sub - gigahertz range@xcite .
it is now established experimentally@xcite that the excitation of this gyrotropic motion leads to a dynamical distortion of the vortex core profile , as predicted by micromagnetic simulations and theoretical analysis@xcite .
this distortion increases with the linear velocity of the vortex core and opposes the core polarity , until the critical velocity @xmath1 ( @xmath2 is the gyromagnetic ratio of the magnetic material and @xmath3 its exchange constant ) is reached and the vortex core polarity is reversed@xcite . in zero magnetic field , dynamical control of the polarity
is difficult due to the degeneracy of the gyrotropic frequencies associated to opposite polarities @xmath4 , which can lead to multiple core switching@xcite .
still , selective core polarity reversal is possible using a circularly polarized microwave magnetic field because the sense of the core rotation is linked by a right - hand rule to its polarity@xcite .
control of polarity switching can also be achieved by precise timing of non resonant magnetic field pulses@xcite , in a similar fashion as domain wall propagation in magnetic nanowires@xcite .
resonant amplification@xcite of the vortex gyrotropic motion enables to reverse the core polarity with minimum excitation power@xcite , as it allows to concentrate the energy in a narrow frequency band . in this scheme
, the damping ratio is an important parameter because it controls the minimum amplitude of the resonant excitation required to switch the core@xcite . here
, it is shown that the damping ratio close to the reversal threshold is significantly larger than the one measured in the small oscillation regime .
we associate this to the non - linear nature of the reversal process@xcite .
investigation of vortex core reversal using time - resolved imaging techniques based on x - ray magnetic circular dichroism is very powerful , as it enables direct determination of the vortex core trajectory@xcite .
however , it requires to average many events and the interpretation can become more complicated due to multiple reversal processes . in order to investigate the coupling between resonant microwave pulses and vortex dynamics near the core reversal threshold
, we use a magnetic resonance force microscope ( mrfm ) in combination with a bias magnetic field applied perpendicularly to the disk plane , that induces two distinct resonant gyrotropic frequencies associated to opposite core polarities@xcite .
this frequency splitting is used for a simple , single shot reading@xcite of the polarity state using mrfm , before and after microwave pulses have been applied to the vortex - state nano - disk .
the mrfm detection setup@xcite , illustrated in fig.1a , is described in the methods section .
it allows to probe the magnetization dynamics in individual nanostructures@xcite , and in particular to spectroscopically determine the resonance frequency of the gyrotropic mode in a vortex - state nano - disk@xcite . the studied nano - disk with thickness @xmath5 nm and diameter @xmath6 m is made of nimnsb alloy ( see mfm image in fig.1b ) , an ultra - low damping epitaxial material@xcite ( typical gilbert constant @xmath7 is between 0.002 and 0.003 ) .
a gold antenna patterned on top allows to apply pulses of linearly polarized microwave magnetic field @xmath8 in the plane of the disk . in our experiments ,
we choose the bias perpendicular field @xmath9 mt in order to unambiguously discriminate the two different gyrotropic frequencies , or polarity states : resonant frequencies @xmath10 mhz and @xmath11 mhz respectively correspond to @xmath12 ( core antiparallel to @xmath13 , see fig.1c ) and @xmath14 ( core parallel to @xmath13 , see fig.1d ) .
the microwave power employed to read the polarity state with mrfm is weak enough ( @xmath15 dbm ) , so that @xmath0 is not reversed during the reading sequence@xcite .
we also emphasize that the bias perpendicular field @xmath16 mt is almost five times smaller than the static field required to switch the core polarity@xcite , so that no significant asymmetry between the two possible reversal processes ( @xmath12 to @xmath14 and vice versa ) is induced , as it will be clear from the data presented below .
first , we explore in fig.2a the efficiency of single microwave pulses to switch the vortex core , depending on their duration , frequency and power .
@xmath17 pulses are defined as single pulses that reverse the polarity from @xmath12 to @xmath14 and @xmath18 pulses from @xmath14 to @xmath12 ( see fig.2b ) .
the experimental data is acquired as follows : for @xmath17 ( @xmath18 ) pulses , the @xmath12 ( @xmath14 ) state is first reset using an initialization pulse whose result is known to be fully deterministic@xcite .
then , a single microwave pulse of given duration @xmath19 , frequency @xmath20 and power @xmath21 is applied , and the final polarity state is read using mrfm . an opaque coloured pixel marks pulse settings for
which reversals are recorded with a 100% success rate , while a blank pixel means that no reversal is recorded .
the contour plots presented in fig.2a with different shades of red ( blue ) show the superposition of the results for @xmath17 ( @xmath18 ) pulses with three different durations @xmath19 : 100 , 50 and 20 ns .
the frequency splitting introduced by the perpendicular bias field clearly appears in fig.2a . due to the resonant character of the investigated switching process , a pronounced minimum in the power level required to reverse the vortex core is observed at a frequency @xmath22 ( @xmath23 ) close to the gyrotropic frequency associated to the initial core polarity , for each duration of the @xmath17 ( @xmath18 ) pulses .
this frequency discrimination introduced by @xmath13 allows most of the recorded results to be fully deterministic .
the minimum power @xmath24 required to reverse the core polarity increases as the pulse duration @xmath19 decreases .
we have plotted the dependences on @xmath19 of the optimal frequency @xmath23 ( fig.2d ) and of the optimal pulse energy @xmath25 ( fig.2e ) . in these plots , @xmath19 ranges from 1 @xmath26s down to 3 ns , with experimental points obtained from the analysis of data sets similar to those presented fig.2a .
there is a clear minimum in the optimal pulse energy that occurs around @xmath27 ns .
the position of this minimum yields an estimation of the characteristic decay time .
if the pulse duration exceeds this characteristic time , the vortex core dynamics reaches steady - state before the end of the pulse , i.e. , the pulse duration is uselessly too long . on the contrary ,
if the pulse duration is shorter than this characteristic time , some energy is wasted outside the resonance line . due to the transient response of the vortex core to the excitation pulse ,
the optimal frequency @xmath23 measured for pulses shorter than 50 ns shifts to higher frequency@xcite ( see fig.2d ) , a general behaviour of any harmonic oscillator forced on time scales shorter than its decay time . to be more quantitative ,
we have performed a numerical calculation based on the analytical approach developed for zero applied magnetic field@xcite , with the new ingredient that due to the bias perpendicular field , @xmath28 and @xmath29 are distinct frequencies .
the trajectory of the vortex core submitted to the microwave field pulse is calculated using thiele s equation@xcite , and guslienko s criterion@xcite for core velocity is used to determine if the polarity is switched by the end of the pulse .
the behaviour of the optimal pulse energy as a function of @xmath19 can be reproduced with an analytical formula where the damping ratio @xmath30 is an adjustable parameter ( blue solid line in fig.2e ) . the best fit is obtained for @xmath31 , corresponding to a characteristic decay time @xmath32 ns . the predicted optimal pulse duration is @xmath33 ns ( see methods section for details ) .
fig.2c illustrates the good agreement of the model with the experimental data of fig.2a on the whole range of pulse parameters .
the significant result here is that the damping ratio @xmath31 inferred in the forced regime close to the core reversal threshold is more than twice larger than the value measured in the small oscillation regime of the gyrotropic mode , found to be @xmath34 ( corresponding to a decay time @xmath35 ns ) .
the damping ratio @xmath36 is itself about three times larger than the gilbert constant , measured from the linewidth in the perpendicularly saturated state of the nimnsb disk and found to be @xmath37 ( see supplementary figure 1 ) .
the difference between @xmath36 and @xmath7 corresponds to the expected increase due to topological renormalization@xcite .
this enhancement is produced by the strongly inhomogeneous spatial distribution of magnetization in the undistorted vortex state compared to the perpendicularly saturated state .
our experimental results of fig.2 give a quantitative estimation of the damping ratio @xmath30 for the vortex distorted by large oscillations@xcite .
in a further step , it would be interesting to evaluate how much of the observed increase of @xmath30 against @xmath36 is due to topological renormalisation between the distorted and undistorted vortex states . as can be seen in fig.2e ,
a practical consequence of this increase is to decrease the optimal pulse duration @xmath38 and to increase the corresponding minimum pulse energy .
next , we investigate in fig.3 the vortex core dynamics in the free decay regime by monitoring the core reversal induced by two consecutive microwave pulses . in these experiments
, the pulse duration is set to @xmath39 ns and the power to @xmath40 dbm , i.e. , slightly below the minimum power @xmath41 dbm required to reverse the core polarity @xmath0 for such a short pulse .
contrary to a single pulse , two pulses separated by a delay @xmath42 ( fig.3a : @xmath17@xmath42@xmath17 ; fig.3b
: @xmath18@xmath42@xmath18 ) can reverse @xmath0 . the striking oscillatory dependence on the pulses carrier frequency and on the delay @xmath42 observed in figs.3a and 3b enlightens the phase coherent coupling between the vortex gyrotropic motion and the microwave excitation . during the pulse duration
, the gyrotropic motion is forced at the pulse frequency , while in the free decay regime , the core oscillates towards its equilibrium position at its natural frequency , therefore acquiring a phase shift with respect to the excitation carrier . as a result ,
the efficiency of the second pulse to drive the vortex core to the reversal threshold , which is estimated out of ten attempts in fig.3 , depends on the microwave frequency and on the delay between pulses in an oscillatory manner ( the oscillation period scales as the inverse frequency detuning ) . to illustrate further this effect of coherence , similar experiments with a @xmath43 phase shift introduced between the two pulses
have been carried out .
it is clear that the regions where successful reversal are observed in figs.3c and 3d are complementary to those in figs.3a and 3b , respectively .
thus , phase control of the microwave excitation can trigger vortex core switching . using the same approach as before
, it is possible to calculate numerically such experimental phase diagrams .
an excellent agreement with experiments is achieved , as can be seen in fig.4a . in fig.4b , we have plotted the calculated vortex core trajectory and velocity as a function of time corresponding to two consecutive @xmath18 pulses at @xmath44 mhz separated by a 60 ns delay . in the top graphs ,
there is no phase shift between pulses and the time delay is such that the second pulse is efficient to amplify the gyrotropic motion from the beginning of the pulse , in contrary to the bottom graphs , where a @xmath43 phase shift is set between pulses . as a result
, the vortex core is not reversed by the end of the second pulse in the latter case , while it is in the former .
the obtained data sets also allow us to fit the free decay time , and the best agreement with the experiment is obtained for @xmath45 ns .
analysis of the difference of this value with @xmath46 should further shed some light on the precise non - linear nature of magnetic dissipation close to the reversal threshold and on its dependence on the amplitude of the vortex core motion .
this research was partially supported by the french grant voice anr-09-nano-006 - 01 , eu grants dynamax fp6-ist-033749 and master nmp - fp7 - 212257 .
the authors acknowledge fruitful discussions with v.s .
tiberkevich , a.n . slavin and k.y .
guslienko . |
for a knot @xmath0 in the @xmath1-sphere @xmath2 , the minimum number of crossings of a diagram among all diagrams of @xmath0 is called the _ crossing number_.
this is one of the most basic topological complexities of a knot .
another complexity of knots which we consider in this paper is the diameter of the set of boundary slopes .
let @xmath3 be the exterior of @xmath0 .
a compact connected surface properly embedded in @xmath3 is said to be _ essential _ if the surface is incompressible and boundary - incompressible .
the boundary of an essential surface appears on @xmath4 as a parallel family of non - trivial simple closed curves , and so , it determines an isotopy class of the curves , which is called the _ boundary slope _ of the surface . recall that the set of slopes is identified with the set of rational numbers with @xmath5 , where the meridian of @xmath0 corresponds to @xmath5 , by using the standard meridian - longitude system for @xmath0 .
then _ the diameter of the set of boundary slopes for @xmath0 _ is defined as the difference between the greatest one and the least one as rational numbers , except for the infinity ( i.e. , meridional ) slope .
this is well - defined since it is known that there are only finitely many boundary slopes @xcite and there exist at least two boundary slopes in general @xcite .
our main result is the following : [ thm : crossingnumberanddiameter : main ]
let @xmath0 be a montesinos knot without @xmath6-tangles .
then , we have @xmath7 where @xmath8 denotes the crossing number of @xmath0 and @xmath9 denotes the diameter of the set of boundary slopes for @xmath0 .
the equality holds if @xmath0 is alternating .
it is probable that this inequality holds for all knots in general .
for example , it is easily verified for torus knots . for a non - trivial torus knot
@xmath10 , assuming that @xmath11 are relatively prime with @xmath12 , it is known that the boundary slopes are @xmath13 and @xmath14 while the crossing number is @xmath15 .
thus , we have @xmath16 .
the equality holds when @xmath17 , equivalently , the knot is alternating .
moreover , as far as the authors observed , the equality holds for all alternating knots .
we here remark that , for alternating knots , the converse inequality @xmath18 can be shown . by the checkerboard construction for the reduced alternating diagram ,
we can obtain two surfaces , which are both essential by @xcite , and the difference between their boundary slopes is actually @xmath19 . to prove the theorem
, we mainly use the algorithm given by hatcher and oertel in @xcite , which enumerates all boundary slopes for a given montesinos knot . the algorithm in turn
is based on the algorithm by hatcher and thurston in @xcite , which enumerates all boundary slopes for a two - bridge knot .
another ingredient is the way to calculate the crossing number of a given montesinos knot obtained by lickorish and thistlethwaite in @xcite . in the next section ,
we introduce some notation and terminology .
then , we prove the main theorem in the last section .
the authors would like to thank masaharu ishikawa for suggesting the potential relationship between the crossing number and the diameter of the set of boundary slopes .
they also thank to thomas mattman for letting them know the work @xcite .
they are grateful to the referee for careful reading of the manuscript and useful comments
in this section , we prepare various notations and recall fundamentals .
we start with a _ rational tangle _ , which we regard as follows . see figure [ fig:1over2-tangle ] .
the left figure illustrates a sphere with four punctures , which is flattened like a `` pillowcase '' .
consecutive segments with slope @xmath20 are drawn from each of four punctures .
these segments form two arcs .
think about the interior of the sphere , and push these two arcs into the interior with four ends fixed .
see the right figure in figure [ fig:1over2-tangle ] .
then , we have a @xmath20-tangle in a @xmath1-ball .
similarly , we can obtain a @xmath21-tangle for arbitrary irreducible fraction @xmath21 including @xmath6 .
( 140,65 ) ( 0,8 ) ( 80,0 ) a _ montesinos knot _ is defined as a knot obtained by putting rational tangles together in a circle .
see figure [ fig : montesinosknot ] . a montesinos knot obtained from rational tangles @xmath22 , @xmath23 , @xmath24
, @xmath25 is denoted by @xmath26 , @xmath23 , @xmath24 , @xmath27 .
the number of tangles will be denoted by @xmath28 throughout this paper .
concerning such a tuple @xmath29 , @xmath23 , @xmath24 , @xmath27 , we will keep the following assumptions : * in general , by combining rational tangles , we have a montesinos link with one or more link components .
however , in this paper , we want to only consider montesinos knots ; the number of components is one .
thus we assume that the tuple must satisfy either of the two conditions ; ( a ) exactly one of the denominators of @xmath30 s is even or ( b ) all denominators of @xmath30 s are odd and
the number of odd numerators is odd .
* we always assume that none of @xmath30 s are @xmath6 .
furthermore , we always assume @xmath31 since if the number of tangles is two or less , the knot is found to be a two - bridge knot .
for the two - bridge knots , a result corresponding to the main theorem in this paper has been presented by mattman , maybrun and robinson in @xcite .
* we assume that each tangle is non - integral , since an integral tangle can be combined with an adjacent rational tangle .
this operation is a kind of normalization .
( 109,91 ) ( 0,0 ) in this subsection , we review the notions developed in the work of hatcher and oertel @xcite about the boundary slopes for montesinos knots . though we will try to give necessary explanations
so that this paper can be self - contained , it would be more preferable that the reader is rather familiar to their machinery .
also see our previous note @xcite for detail . throughout the following ,
let @xmath0 be a montesinos knot obtained from rational tangles @xmath22 , @xmath23 , @xmath24 , @xmath25 . in sequel , @xmath30 will denote a fraction or the corresponding rational tangle depending on the situation .
we first give an outline of the machinery developed by hatcher and oertel briefly .
take a simple unknotted loop in the 3-sphere @xmath2 , and set disjoint @xmath32 disks bounded by the loop . by the disks
, @xmath2 is divided into @xmath32 3-balls .
after appropriate isotopies , @xmath0 is divided into rational tangles @xmath33 , the intersections with the @xmath32 balls .
let @xmath34 be an essential ( meaning that , incompressible and boundary - incompressible ) surface properly embedded in the exterior of @xmath0 . by virtue of proposition 1.1 in @xcite , @xmath34 can be isotoped so that it is divided into subsurfaces @xmath35 in certain `` good position '' included in each of the @xmath32 balls
. then each @xmath36 is represented by an `` edgepath '' @xmath37 in certain `` diagram '' @xmath38 , and the whole @xmath34 is represented by an `` edgepath system '' @xmath39 in @xmath38 .
actually the boundary slope of @xmath34 is calculated from the corresponding @xmath39 in purely combinatorial way .
conversely an `` edgepath system '' @xmath39 in @xmath38 corresponds to a properly embedded surface in the exterior of @xmath0 . in @xcite ,
conditions for determining essentiality of the surface are fully described . in the following five subsubsections
, we will summarize the features of the diagram @xmath38 , edges , edgepaths and edgepath systems in @xmath38 , which we will use in the rest of paper . the diagram @xmath38 is described as the 1-skeleton of a triangulation of a region on a plane as illustrated in figure [ fig : diagram ] .
precisely a vertex of @xmath38 indicates either a point @xmath40 denoted by @xmath41 , a point @xmath42 denoted by @xmath43 where @xmath21 is an irreducible fraction , or a point @xmath44 denoted by @xmath45 .
( 70,210 ) ( -14,2.5 ) ( 58,175 ) ( 58,-10 ) ( -27,108)@xmath46 ( -13,104)(1,-1)10 ( -14,83)@xmath47 ( 82,82)@xmath48 ( 102,88)@xmath49 ( 45,205)@xmath50 ( 1,12)@xmath51 ( 23,14)(4,-1)15 ( 26,3)@xmath52 ( -11,50)@xmath53 ( 11,51)(1,0)25 ( 26,43)@xmath47 ( 10,94)@xmath54 ( 23,95)(4,-1)15 ( 30,82)@xmath55 ( -3,129)@xmath56 ( 11,131)(1,0)25 ( 33,134)@xmath48 ( 6,169)@xmath57 ( 20,171)(1,0)15 ( 33,173)@xmath58 ( 100,0 ) ( 100,85 ) ( -9,-1)(1,0)10 ( -23,-4)@xmath54 ( 121,-1)(-1,0)10 ( 124,-4)@xmath59 ( -9,100)(1,0)10 ( -23,97)@xmath56 ( 121,100)(-1,0)10 ( 124,97)@xmath60 ( 43,50)(1,0)10 ( 28,47)@xmath61 ( 121,50)(-1,0)10 ( 124,47)@xmath62 ( 61,33)(1,0)10 ( 46,30)@xmath63 ( 121,33)(-1,0)10 ( 124,30)@xmath64 ( 61,66)(1,0)10 ( 46,63)@xmath65 ( 121,66)(-1,0)10 ( 124,63)@xmath66 two vertices @xmath67 and @xmath68 in @xmath38 are connected by an _ edge _ if @xmath69 . such an edge is denoted by @xmath67@xmath68 .
an important class of the edges are the _ vertical edges _ , which connect the vertices @xmath70 and @xmath71 for arbitrary integer @xmath72 .
another important class of the edges are the _
@xmath5-edges _ , which connect the vertices @xmath73 and @xmath70 for integer @xmath72 .
there is another kind of edge called a _ horizontal edge _ , which connects @xmath67 and @xmath43 for each @xmath21 .
let @xmath74 be an edge @xmath67@xmath68 with @xmath75 , @xmath76 .
let @xmath77 and @xmath78 be integers satisfying @xmath79 and @xmath80 .
then , let @xmath81 denote a point on @xmath74 with @xmath82-coordinates @xmath83 .
this is called a _ rational point _ of the edge @xmath74 . for an @xmath5-edge @xmath46@xmath70 ,
let @xmath84 denote a point with @xmath85 on the edge , which is a rational point of the @xmath5-edge .
a _ partial edge _ is an edge @xmath86@xmath68 . on the other hand , a non - partial edge
is called a _
complete edge_. let @xmath74 be a horizontal edge @xmath67@xmath43 .
then , for integers @xmath87 and @xmath88 let @xmath89 denote a point with @xmath90 on the edge @xmath74 .
this is a rational point of the horizontal edge .
assume that a non-@xmath5-edge is oriented from right to left .
then it is said to be _ increasing _ or _ decreasing _ if the @xmath50-coordinate increases or decreases as a point moves along the edge in that direction .
that is , an edge in the region @xmath91 is increasing or decreasing if the @xmath50-coordinate increases or decreases as the @xmath49-coordinate decreases .
we assign @xmath92 or @xmath47 to an edge @xmath74 according to whether @xmath74 is increasing or decreasing respectively .
this is called the _
sign _ of the edge @xmath74 , and is denoted by @xmath93 . the _ length _ of a complete edge is set to be @xmath48 .
the length of a partial edge @xmath86@xmath68 is set to be @xmath94 .
let @xmath95 denote the length of an edge @xmath74 . roughly , an edgepath is a connected piecewise - linear path in the diagram @xmath38 .
there are two kinds of edgepaths ; non - constant edgepath and constant edgepath .
the first one is a path on @xmath38 starting from a vertex @xmath96 .
we will only consider such an edgepath with the following two conditions : 1 .
it must run from right to left in the weak sense , where vertical edges are allowed .
2 . it must be _ minimal _ , that is , it must not retrace back nor include two successive edges on a common triangle .
we call an edgepath of this kind a _ non - constant edgepath_. a non - constant edgepath is expressed by a sequence of vertices like @xmath97@xmath98@xmath24@xmath99@xmath100 , where @xmath101 . note here that the vertices are listed from right to left so that the direction of the sequence coincides with the direction of an edgepath in the diagram @xmath38 . also note that we allow that the last edge can be a partial edge
. an edgepath of the second kind is a single point on a horizontal edge @xmath96@xmath102 .
this is called a
_ constant edgepath_. the single point of a constant edgepath must be a rational point of the horizontal edge @xmath96@xmath102 .
we will always use the symbol @xmath103 for edgepaths . in the following ,
we collect some terminologies about edgepaths .
assume that the @xmath49-coordinate of the left endpoint of an edgepath is @xmath104 .
then the edgepath is said to be of _ type i _ , _ type ii _ or _ type iii _ according to which of @xmath105 , @xmath106 or @xmath107 the coordinate satisfies .
we here introduce a new notion ; a _ basic edgepath _ , which is not used in @xcite .
an edgepath is called a _ basic edgepath _ if the edgepath ends at the point with @xmath49-coordinate @xmath13 and includes no vertical edges .
we usually use symbol @xmath108 for basic edgepaths .
an edgepath is said to be _ monotonically increasing _
( respectively _ monotonically decreasing _ ) , if all non-@xmath5-edges in the edgepath are increasing ( resp . decreasing ) .
we ignore @xmath5-edges when we judge an edgepath is monotonic or not .
for the vertex @xmath67 @xmath109 , there is one increasing leftward edge and one decreasing leftward edge .
hence , there exists just one monotonically increasing basic edgepath and just one monotonically decreasing basic edgepath for a fixed fraction @xmath21 @xmath109 .
the _ length _ of an edgepath is the sum of lengths of edges in the edgepath .
naturally the length of a constant edgepath should be zero .
for an edgepath @xmath103 , @xmath110 denotes the length of the edgepath .
some examples of edgepaths for @xmath111-tangle are given in figure [ fig : edgepaths ] . in ( 1 ) ,
three basic edgepaths @xmath56@xmath112@xmath113 , @xmath54@xmath112@xmath113 and @xmath54@xmath114@xmath113 are drawn .
the first and third one are monotonically increasing and decreasing , while the second one is neither . in ( 2 ) ,
the edgepath @xmath56@xmath112@xmath114@xmath113 is not minimal and is not allowed , since @xmath112@xmath114@xmath113 consists of two edges of a common triangle . in ( 3 ) , a type i edgepath ( @xmath115+@xmath116)@xmath114@xmath113 and a type iii edgepath @xmath46@xmath56@xmath112@xmath113 are drawn .
a thick dot on the horizontal edge @xmath113@xmath117 in this figure is a constant edgepath , which is a rational point on the horizontal edge . in ( 4 ) , type ii edgepaths @xmath57@xmath56@xmath112@xmath113 , @xmath54@xmath56@xmath112@xmath113 , @xmath56@xmath54@xmath114@xmath113 and @xmath53@xmath54@xmath114@xmath113 are depicted .
note that the second one is also an example of a non - minimal edgepath , since the edgepath is not minimal at the vertex @xmath56 .
( 300,390 ) ( 0,200 ) ( 0,0 ) ( 65,115)(-1,-1)10 ( 67,112)@xmath56 ( 95,90)(-2,-1)15 ( 97,87)@xmath112 ( 116,75)(-1,0)20 ( 118,72)@xmath113 ( 104,62)(-2,1)15 ( 106,59)@xmath114 ( 65,43)(-1,1)10 ( 67,40)@xmath54 ( 50,192)(1 ) ( 150,200 ) ( 0,0 ) ( 65,115)(-1,-1)10 ( 67,112)@xmath56 ( 95,90)(-2,-1)15 ( 97,87)@xmath112 ( 116,75)(-1,0)20 ( 118,72)@xmath113 ( 104,62)(-2,1)15 ( 106,59)@xmath114 ( 200,192)(2 ) ( 0,0 ) ( 0,0 ) ( 3,70)(0,-1)15 ( -9,76)@xmath46 ( 65,115)(-1,-1)10 ( 67,112)@xmath56 ( 80,95)(0,-1)15 ( 70,99)@xmath112 ( 116,95)(-1,-1)20 ( 118,92)@xmath113 ( 116,75)(-1,0)10 ( 118,72)@xmath117 ( 104,62)(-2,1)15 ( 106,59)@xmath114 ( 65,43)(-1,1)10 ( 67,40)@xmath54 ( 50,-8)(3 ) ( 150,0 ) ( 0,0 ) ( 65,145)(-1,1)10 ( 67,142)@xmath57 ( 65,115)(-1,-1)10 ( 67,112)@xmath56 ( 95,90)(-2,-1)15 ( 97,87)@xmath112 ( 116,75)(-1,0)20 ( 118,72)@xmath113 ( 104,62)(-2,1)15 ( 106,59)@xmath114 ( 65,43)(-1,1)10 ( 67,40)@xmath54 ( 65,13)(-1,-1)10 ( 67,10)@xmath53 ( 200,-8)(4 ) by an _ edgepath system _ , we mean a collection of @xmath32 edgepaths . recall that @xmath32 denotes the number of tangles of a montesinos knot .
we will use symbol @xmath39 for edgepath systems .
we call the following conditions _ gluing consistency _ : the @xmath49-coordinates of the left endpoints of the edgepaths must coincide with each other , and the sum of @xmath50-coordinates of the left endpoints must be @xmath13 .
as we will see , this condition must be satisfied by an edgepath system which corresponds to a properly embedded surface in the exterior of a montesinos knot .
edgepaths in an edgepath system satisfying the gluing consistency have the common @xmath49-coordinate of the left endpoints .
therefore , similarly to edgepaths , edgepath systems satisfying the gluing consistency are also classified by the common @xmath49-coordinates .
that is , such an edgepath system is said to be of _ type i _ , _ type ii _ or _ type iii _ according to which of @xmath91 , @xmath118 or @xmath119 the common @xmath49-coordinate satisfies . in general
, we will only consider edgepath systems with gluing consistency , but we sometimes make an edgepath system which may not satisfy the gluing consistency .
we call the edgepath system a _ formal edgepath system_. for example , we will consider an edgepath system all of whose edgepaths are basic edgepaths , which we call a _ basic edgepath system_. such a basic edgepath system is formal .
we usually use symbol @xmath120 for basic edgepath systems .
we here observe a relationship between vertices and edges in @xmath38 and curve systems on a fourth - punctured sphere .
this enables us to describe a connection of edgepath systems and embedded surfaces in montesinos knot exteriors .
such a connection will be explained in the next subsubsection .
as mentioned before , segments of slope @xmath21 on a pillowcase form two arcs .
we call the arcs _
( a pair of ) @xmath21-arcs_. for instance , figure [ fig:1over2-arcs ] ( left ) depicts @xmath20-arcs .
for ease of drawing figures , we regard @xmath121 as a sphere and draw the four punctures and @xmath21-arcs on the plane .
see @xmath20-arcs in figure [ fig:1over2-arcs ] ( right ) .
some other examples of @xmath21-arcs are given in figure [ fig : somepairsofarcs ] .
( 140,75 ) ( 0,12 ) ( 80,0 ) ( 220,60 ) ( 0,0 ) ( 80,0 ) ( 160,0 ) we define that the set of @xmath78 pairs of @xmath21-arcs correspond to the vertex @xmath67 on @xmath38 .
please note that such multiple pairs can be drawn on the fourth - punctured sphere disjointly .
next , consider the integers @xmath122 , @xmath123 , @xmath124 and @xmath125 satisfying @xmath69 . then a combination of @xmath78 pairs of @xmath21-arcs and @xmath126 pairs of @xmath127-arcs
is set to correspond to a rational point @xmath81 .
also note that such a pair of @xmath21-arcs and a pair of @xmath127-arcs can be disjointly drawn on a fourth - punctured sphere .
a disjoint union of such pairs appears in figure [ fig : subsurfacemultisheets ] . besides
, a non - null - homotopic loop on a fourth - punctured sphere disjoint from @xmath21-arcs is topologically unique and is called a _
@xmath21-circle_. we can think a disjoint union of @xmath78 pairs of @xmath21-arcs and @xmath128 copies of @xmath21-circles with @xmath88 , @xmath87 , as seen in figure [ fig : subsurfacecap ] .
this corresponds to a rational point @xmath89 of a horizontal edge .
consider the same setting as in the explanation of the outline of hatcher - oertel s machinery at the beginning of this subsection .
then a subsurface @xmath36 lies in one of the @xmath28 @xmath1-balls .
now , we assume that the @xmath1-ball is actually a unit ball with the center @xmath55 .
we think about the intersection between @xmath36 and a sphere with the center @xmath55 and radius @xmath129 . as we vary @xmath130 within @xmath131 , the intersection changes . the subsurface @xmath36 is expressed by the sequence of such intersections between @xmath36 and concentric spheres . by virtue of proposition 1.1 in @xcite ,
the subsurface is deformed into a good position in morse theoretical sense .
this enables us to regard the intersection as a curve system , on any non - critical level @xmath130 .
the topological type of the intersection changes only at critical levels , and so , a subsurface can be represented by a sequence of curve systems and hence by an edgepath . in the following ,
we observe parts of the surfaces represented by each edge in the edgepath system .
a non - constant edgepath corresponds to a _
saddle subsurface_. an example of a saddle subsurface is expressed as a sequence of intersections with concentric spheres as illustrated in figure [ fig : subsurfacesaddle ] . in this figure ,
the intersection is empty for @xmath132 , @xmath20-arcs for @xmath133 , a disk bounded by @xmath20-arcs and @xmath13-arcs for @xmath134 , @xmath13-arcs for @xmath135 , a disk bounded by @xmath13-arcs and @xmath6-arcs for @xmath136 , and @xmath6-arcs for @xmath137 .
this is a subsurface corresponding to an edgepath @xmath46@xmath54@xmath112 .
( 380,90 ) ( 0,30 ) ( -7,0)@xmath138 ( 80,30 ) ( 83,0)@xmath136 ( 160,30 ) ( 150,0)@xmath139 ( 240,17 ) ( 243,0)@xmath134 ( 320,17 ) ( 310,0)@xmath133 a disk bounded by two pairs of arcs is called a _
saddle_. a pair of @xmath21-arcs and a pair of @xmath127-arcs can be disjointly drawn and give a saddle only if @xmath69 .
this is the reason why we think about an edgepath which consists of edges of the form @xmath67@xmath68 satisfying @xmath69 . note that , four punctures on each concentric sphere with radius @xmath140 and @xmath20-arcs on the sphere with radius @xmath141 form a @xmath20-tangle in the @xmath1-ball . thus , the boundary of the subsurface lies on the boundary of the exterior of @xmath20-tangle .
the example of a subsurface above is `` one - sheeted '' . in general
, a subsurface can have `` multiple sheets '' .
if a small meridian circle of a montesinos knot @xmath0 meets an essential surface in @xmath78 points , then the surface is said to be _
@xmath78-sheeted _ , or the _ number of sheets _ of the surface is @xmath78 at the point on the knot .
after isotoping the surface with keeping it properly embedded , the number of sheets is constant everywhere on the knot .
then , the number of sheets is defined for a surface . for an @xmath78-sheeted subsurface @xmath34 ,
the intersection between @xmath34 and a concentric sphere is in general a curve system including exactly @xmath78 pairs of arcs .
figure [ fig : subsurfacemultisheets ] illustrates an example of a saddle subsurface with multiple sheets . in this figure , there are @xmath1 sheets , and only one saddle appears .
the left figure is the union of one pair of @xmath6-arcs and two pairs of @xmath13-arcs , which is denoted by @xmath142 .
then , this subsurface corresponds to a partial edge @xmath143@xmath54 .
an edgepath including a partial edge must be multiple - sheeted .
( 250,75 ) ( 0,15 ) ( -7,0)@xmath144 ( 100,15 ) ( 101,0)@xmath145 ( 200,15 ) ( 190,0)@xmath146 note that , since @xmath67 corresponds also to a disjoint union of @xmath78 pairs of @xmath21-arcs for arbitrary positive integer @xmath78 , an @xmath78-sheeted parallel subsurface also corresponds to an edgepath without a partial edge .
a subsurface corresponding to a constant edgepath is illustrated in figure [ fig : subsurfacecap ] .
the intersection is empty for @xmath147 , a disk bounded by a @xmath13-circle for @xmath148 , a @xmath13-circle for @xmath149 , and the union of a @xmath13-circle and @xmath13-arcs for @xmath150 .
such a subsurface is called a _ cap subsurface_. note that the boundary of this subsurface lies on the boundary of the exterior of @xmath13-tangle in the @xmath1-ball . the intersection between the subsurface and a concentric sphere of radius @xmath150
is expressed by a rational point @xmath151 . and the subsurface is also denoted by @xmath151 , which means a constant edgepath .
( 280,80 ) ( 0,15 ) ( 5,0)@xmath150 ( 100,15 ) ( 101,0)@xmath149 ( 200,15 ) ( 213,0)@xmath148 for a saddle subsurface or a cap subsurface , the boundary of the subsurface lies on the boundary of the exterior of the corresponding rational tangle .
therefore , after the subsurfaces are glued each other , the boundary of the obtained surface lies on the boundary of the exterior of a montesinos knot . in short , the surface
is confirmed to be properly embedded . in order to calculate the boundary slope of an essential surface
, we calculate the _ total number of twists _ , or in short _ twist_. as well as the boundary slope , the twist also measures how many times the boundary of a surface runs around a knot in the counterclockwise direction of the meridian , while the boundary runs once along the knot .
the twist of a surface is obtained by first calculating the twist for each subsurface and then summing up the twists .
the twist of a subsurface is calculated as the sum of twists for each saddle in the subsurface . for a saddle corresponding to an @xmath5-edge like @xmath46@xmath70 , the twists appear on a saddle at four punctures and cancel out each other as in figure [ fig : twist ] ( left ) .
besides , for a saddle corresponding to a non-@xmath5-edge @xmath67@xmath68 , the twist is @xmath58 or @xmath52 as seen in figure [ fig : twist ] ( right ) .
the sign of the twist is determined by whether the edge is decreasing or increasing .
next , for saddles of a subsurfaces with multiple sheets as in figure [ fig : subsurfacemultisheets ] , a single saddle of an @xmath78-sheeted subsurface corresponding to a non-@xmath5-edge @xmath86@xmath152 has the twist @xmath153 , and the twist for @xmath86@xmath68 is @xmath154 . for a cap subsurface as in figure [ fig : subsurfacecap ] , since no saddles appear , the twist is @xmath13 . hence , the procedure to calculate the twist is summarized as follows . for each non-@xmath5-edge @xmath74 in an edgepath , we assign @xmath155 to the edge @xmath74 .
then the twist of an edgepath is the sum of such assigned values for edges in the edgepath , where a constant edgepath has twist @xmath13 .
the twist of an edgepath system is the sum of twists of all edgepaths in the edgepath system .
the twist is a substitution of the boundary slope and fits well with the algorithm .
we use the symbol @xmath156 for the twist , and @xmath157 , @xmath158 , @xmath159 and @xmath160 denote the twist of an edge @xmath74 , an edgepath @xmath103 , an edgepath system @xmath39 and a surface @xmath34 . since the boundary slope is defined so that the boundary slope of a seifert surface @xmath161 is @xmath13 , the boundary slope of an essential surface @xmath34 is calculated as the difference of the twist of the surface and that of a seifert surface , that is , @xmath162 .
( 200,86 ) ( 0,22 ) ( 100,0 ) we here collect some remarks about the previous subsubsections . *
since each rational tangle @xmath30 in a montesinos knot in this paper is non - integral , the starting point of an edgepath must be a vertex @xmath67 or a rational point on a horizontal edge @xmath67@xmath43 for a non - integral fraction @xmath21 .
on the other hand , figures [ fig : subsurfacemultisheets ] and [ fig : subsurfacecap ] correspond to edgepaths with the starting point on @xmath54@xmath59 .
hence , these figures are not appropriate in some sense .
these figures are used just for ease of drawing and explanation . * here we illustrates three - dimensional figures of subsurfaces corresponding to a non - constant edgepath and a constant edgepath ( see figure [ fig : partialsurfaces ] ) .
note that these subsurfaces lie in @xmath1-balls and top boundaries of these subsurfaces lie on the boundary spheres of the @xmath1-balls , which are not drawn in this figure . besides , the `` cap '' of the cap subsurface in figure [ fig : partialsurfaces](b ) is somewhat deformed from the subsurface illustrated in figure [ fig : subsurfacecap ] .
+ ( 330,120 ) ( 40,20 ) ( 0,0)(a ) a saddle subsurface ( 18,-10)corresponding to @xmath73@xmath54 ( 220,20 ) ( 180,0)(b ) a cap subsurface ( 200,-10)corresponding to @xmath151 * strictly speaking , there are some irregular essential surfaces : a conway sphere and an augmented surface .
although , we can ignore these surfaces since their boundary slopes are @xmath5 or the same as that of some regular essential surface .
see @xcite for detail . in this subsection
, we will explain the result given in @xcite on the calculation of the crossing number for a montesinos knot .
assume that @xmath163 is a non - integral fraction .
there is the unique continued fraction of @xmath163 as @xmath164 with @xmath165 where the integers @xmath166 , @xmath167 , @xmath24 , @xmath168 satisfy @xmath169 let @xmath170 denote this continued fraction .
for instance , a continued fraction expansion of @xmath171 is @xmath172 this continued fraction gives a diagram of a rational tangle @xmath163 , which is roughly a combination of vertical and horizontal half - twists .
see figure [ fig : alternatingdiagram47over36 ] for example .
if we connect two pairs of adjacent ends of the diagram , because of the condition on @xmath173 s , we obtain an alternating diagram of a two - bridge link .
thus , we call a diagram like this of a rational tangle an _ alternating tangle diagram _ of a rational tangle @xmath163 , which is denoted by @xmath174 . note that the number of crossings of the diagram is @xmath175 .
( 145,91 ) ( 0,0 ) in @xcite , lickorish and thistlethwaite determine the crossing number of the montesinos knots .
though , we must perform a technical operation which restricts the form of a diagram of a montesinos knot .
let @xmath0 be a montesinos knot @xmath176 with @xmath177 , where each @xmath178 is a non - integral fraction .
for @xmath179 , let @xmath180 denote the diagram of @xmath0 obtained by combining alternating tangle diagrams @xmath181 , @xmath182 , @xmath24 , @xmath183 in a natural way
. now , for any tuple of integers @xmath184 , @xmath185 , @xmath24 , @xmath186 with @xmath187 , the montesinos knot @xmath188 is isotopic to @xmath0 , since addition of such integers corresponds to rotations of the tangles .
thus , for instance , if @xmath189 includes both positive and negative fractions and @xmath190 for some @xmath191 , by subtracting @xmath48 from @xmath192 and adding @xmath48 to some negative @xmath193 , we can reduce the integral part @xmath194 $ ] of @xmath192 . through repeating such operations
, we finally have a tuple @xmath195 satisfying + ( a ) all @xmath196 s have the same sign + or + ( b ) @xmath197 includes both positive and negative fractions , and @xmath198 holds for all @xmath199 .
+ for @xmath179 satisfying ( a ) or ( b ) , we think about the diagram @xmath200 .
if @xmath201 have the common sign , then the diagram satisfies condition ( i ) in @xcite , that is , the diagram is alternating .
if @xmath202 holds for each @xmath199 , then each alternating tangle diagram @xmath203 in the diagram @xmath204 satisfies a technical condition in the condition ( ii ) .
therefore , @xmath204 satisfies the condition ( i ) or ( ii ) in @xcite to be a _ reduced montesinos diagram_. with this diagram ,
the crossing number of a montesinos knot @xmath0 is given by the following .
if a link @xmath205 admits an @xmath206-crossing , reduced montesinos diagram , then @xmath205 can not be projected with fewer than @xmath206 crossings . in short , in order to calculate the crossing number of a given montesinos knot @xmath0
, we have only to prepare a reduced montesinos diagram and count its crossings .
thus , by putting them all together , we have the following . [ prop : restrictedknotsanddiagrams ] for a montesinos knot @xmath176 , there is a tuple @xmath207 satisfying either @xmath208 or @xmath209 , such that the knot @xmath0 is isotopic to @xmath210 and has a diagram @xmath211 . furthermore , the diagram @xmath204 is a reduced montesinos diagram defined in @xcite and attains the crossing number of @xmath0 .
the key of the proof is the relation between the number of crossings of the alternating diagram of a rational tangle and lengths of monotonic edgepaths for the rational tangle .
roughly , summing up these relations gives the main theorem , since the crossing number of a montesinos knot is calculated as the sum of the numbers of crossings of the alternating diagrams of rational tangles by @xcite .
let @xmath212 denote the number of crossings of a tangle diagram @xmath204 .
the number @xmath213 of crossings of the alternating tangle diagram @xmath174 of a rational tangle @xmath163 is related to the lengths of monotonic edgepaths for @xmath163 by the following lemma .
let @xmath214 and @xmath215 be the monotonically increasing and decreasing basic edgepaths , which are determined uniquely for given @xmath163 . by adding @xmath5-edges , we obtain the monotonically increasing and decreasing type iii edgepaths @xmath216 and @xmath217 from the basic edgepaths .
note that , since @xmath5-edges do not contribute to the twist , we ignore @xmath5-edges when we judge a type iii edgepath is monotonic or not . according to the sign of @xmath163 , we can construct a monotonic type ii edgepath with an ending vertex @xmath54 , which is a monotonically increasing type ii edgepath @xmath218 if @xmath219 and a monotonically decreasing type ii edgepath @xmath220 if @xmath221 .
then , we obtain the following key lemma .
assume that @xmath163 is positive , and @xmath163 is represented by a continued fraction @xmath170 satisfying @xmath165 and the condition ( [ eq : conditionofuniquecontinuedfraction ] ) .
this continued fraction corresponds to the alternating diagram @xmath174 of the rational tangle @xmath163 .
we also prepare monotonic edgepaths @xmath216 and @xmath220 .
we divide these monotonic edgepaths by use of triangles defined as follows . for integers @xmath228 and @xmath88 ,
there exists a large triangle with vertices @xmath229 see figure [ fig : edgepathsintrianglel=3tok-1 ] . note here that an expression like @xmath230 denotes a rational number @xmath163 corresponding to the continued fraction , and hence , @xmath231 denotes a vertex @xmath96 of the diagram @xmath38 .
moreover , this triangle consists of some smaller triangles with vertices @xmath232 for @xmath233 .
the existence of such a triangle is explained roughly as follows .
an important fact here is that if @xmath234 and @xmath235 with @xmath69 , then @xmath236 holds for any @xmath237 .
this is easily shown for @xmath238 s , and then for @xmath239 s , for @xmath240 s , for @xmath241 s , and so on , inductively .
with the fact , the following features are easily confirmed .
there are edges @xmath242@xmath243 and @xmath243@xmath244 .
for the sequence of vertices @xmath245 , @xmath246 , @xmath24 , their @xmath50-coordinates are monotonic and their @xmath49-coordinates are monotonically increasing .
for all edges of the form @xmath243@xmath244 , if we regard them as line segments , then the slopes of the segments coincide with each other .
this is because we have @xmath247 their @xmath82-coordinates are @xmath248 and the slope of the segment between them is found to be @xmath249 which does not depend on @xmath199 .
hence , the vertices @xmath243 s for @xmath250 lie on the same line .
thus , we have confirmed the existence of the triangle mentioned as above . for @xmath258 ,
we focus on the triangle @xmath259 having the vertices @xmath260 see figure [ fig : edgepathsintrianglel=1or2 ] if @xmath261 , figure [ fig : edgepathsintrianglel=3tok-1 ] if @xmath262 .
which of the two large triangles in figure [ fig : edgepathsintrianglel=3tok-1 ] the triangle @xmath259 looks like depends on the parity of @xmath128 .
the large triangle @xmath259 includes two edgepaths @xmath263 one of the two edgepaths is monotonically increasing , and the other is monotonically decreasing .
note that @xmath264 can be a point when @xmath265 , and such an edgepath can be regarded as both increasing and decreasing .
let @xmath252 and @xmath255 denote the increasing one and the decreasing one .
the important fact here is that @xmath266 . for @xmath267 , we define @xmath268@xmath269@xmath270 as the triangle @xmath259 .
we use @xmath271 instead of @xmath272 .
though , the situation is similar to the previous case , and hence , the identity @xmath273 still holds .
see figure [ fig : edgepathsintrianglel = k ] . by paying attention to the fact that @xmath274
, we can find that @xmath216 is obtained by combining @xmath252 @xmath275 , and @xmath220 by combining @xmath255 @xmath276 .
furthermore , we have @xmath277 the argument is similar for negative @xmath163 .
for instance , in figure [ fig : edgepaths47over36 ] , @xmath216 and @xmath220 are depicted for @xmath296 . in the figure , for making it clear , the horizontal coordinates of the vertices
are somewhat altered from correct values , though , it does not matter in the argument .
note that @xmath297 coincides with @xmath298 .
the twists of monotonic basic edgepath systems give the upper and lower bounds for twists of basic edgepath systems , of type i edgepath systems and of type iii edgepath systems .
[ prop : twistofmonotonicedgepaths ] let @xmath305 and @xmath306 denote the unique monotonically increasing basic edgepath system and the unique monotonically decreasing basic edgepath system .
@xmath307 denotes a basic edgepath system , a type i edgepath system or a type iii edgepath system .
if @xmath308 holds for each @xmath309 , then we immediately have @xmath310 , or equivalently , @xmath311 .
hence , it suffices to show that @xmath312 holds for each edgepath @xmath313 . for a fixed starting point @xmath314 ,
all non - constant edgepaths lie in the region @xmath315 bounded by the monotonically increasing basic edgepath @xmath214 , the monotonically decreasing basic edgepath @xmath215 , and exactly one vertical edge as illustrated in figure [ fig : edgepaths47over36 ] . in this figure , @xmath315 is the union of @xmath316 triangles . as seen from the figure , roughly , the region @xmath315 consists of a sequence of triangles .
hence , no vertices of @xmath103 other than a rational endpoint lie in the interior of @xmath315 .
all increasing ( complete ) edges in the region end at a vertex on @xmath214 . for each intermediate vertex on @xmath214 , there are one or more increasing edges ending at the vertex .
a single edgepath contains at most one of them , and just one of them is always included in @xmath214 .
this implies that the number of increasing edges in an edgepath is less than or equal to that of the monotonically increasing basic edgepath . at the same time , the number of decreasing edges in an edgepath is more than or equal to that of the monotonically increasing basic edgepath , which is zero .
recall that the twist can be calculated as twice of the total length of decreasing non-@xmath5-edges minus the total length of increasing non-@xmath5-edges .
hence , we have @xmath317 .
the argument is similar for the monotonically decreasing basic edgepath .
if a tuple @xmath179 satisfies the condition ( a ) , then the montesinos knot @xmath318 is alternating .
conversely , by the following well - known fact , together with the another well - known fact that any montesinos knot without @xmath6-tangles is prime , an alternating montesinos knot is isotopic to a montesinos knot with a tuple satisfying not ( b ) but ( a ) .
if a link @xmath205 has a connected , irreducible , alternating diagram with @xmath206 crossings , then the number of crossings of any diagram of the link @xmath205 is @xmath206 or greater .
moreover , the number of crossings of any prime , non - alternating diagram is greater than @xmath206 .
assume here that all @xmath178 s are positive .
as mentioned above , by theorem 10 of @xcite , @xmath320 .
let @xmath321 and @xmath322 denote the monotonically increasing type iii edgepath system and the monotonically decreasing type ii edgepath system respectively whose edgepaths are constructed in the same manner as in subsection [ subsec : crossingsandmonotonicedgepaths ] . with respect to these , by lemma [ lem : lengthsandcrossing ] , we have @xmath323 .
one complete edge contributes @xmath324 to the twist .
the twists of two edgepath systems @xmath321 and @xmath322 have opposite signs . combining these facts
, we have : @xmath325 now , both @xmath321 and @xmath322 are minimal .
@xmath321 gives an essential surface by proposition 2.5 of @xcite since the sum of @xmath50-coordinates at @xmath118 of the edgepaths in the edgepath system @xmath321 is at least @xmath1 . in addition , @xmath322 corresponds to an essential surface by one of proposition 2.4 , 2.6 , 2.7 , 2.8 of @xcite since it is monotonic . for a type i or type iii edgepath system @xmath39
, we have @xmath326 by proposition [ prop : twistofmonotonicedgepaths ] .
if @xmath327 is type ii , there is another type ii edgepath system @xmath328 each of whose edgepath @xmath329 shares a basic edgepath with @xmath37 and ends at @xmath54 .
note that @xmath330 holds since the edgepath systems share a basic edgepath system and have the same total amount of signed length of vertical edges by the gluing consistency . by an argument similar to proposition [ prop : twistofmonotonicedgepaths ] with considering effect of vertical edges on twist
, we have @xmath331 .
hence , @xmath332 holds for type ii edgepath systems , and hence for edgepath systems of any type .
note that lemma [ lem : crossingnumberanddiameter : commonsign ] as it is can not be extended to the general montesinos knots .
for instance , @xmath334 has non - integral diameter @xmath335 . if a tuple @xmath179 satisfies the condition ( b )
, then we have the following lemma .
[ lem : crossingnumberanddiameter : distinctsigns ] for a tuple @xmath179 , assume that @xmath189 includes both positive and negative fractions , and @xmath202 holds for any @xmath199 .
then , for the montesinos knot @xmath176 , the following inequality holds .
first , we prepare monotonic type iii edgepath systems @xmath321 and @xmath337 . similarly to the argument in the proof of the previous lemma , we have @xmath338 note that , the condition @xmath202 is required to apply lemma [ lem : lengthsandcrossing ] .
we think about the range of twists .
the twists of the monotonic type iii edgepath systems are @xmath339 and @xmath340 .
though , the surfaces corresponding to these edgepath systems may be inessential .
we check that @xmath341 and @xmath342 are the lower and upper bounds of the twist @xmath156 .
the claim is true for type i and type iii edgepath systems by proposition [ prop : twistofmonotonicedgepaths ] . for a type ii edgepath system @xmath39 , by preparing @xmath343 as in the proof of the previous lemma , we easily have @xmath330 and @xmath344 by an argument similar to proposition [ prop : twistofmonotonicedgepaths ] .
hence , the twist is bounded by @xmath341 and @xmath342 also for any type ii edgepath system . thus , @xmath345 finally , combining two - bridge cases in @xcite , proposition [ prop : restrictedknotsanddiagrams ] , lemma [ lem : crossingnumberanddiameter : commonsign ] and lemma [ lem : crossingnumberanddiameter : distinctsigns ] , we have theorem [ thm : crossingnumberanddiameter : main ] .
k. ichihara and s. mizushima , bounds on numerical boundary slopes for montesinos knots , hiroshima math .
j. 37 ( 2007 ) , no .
2 , 211252 . w.b.r .
lickorish , an introduction to knot theory , graduate texts in mathematics .
new york , springer , 1997 . | it is shown that the diameter of the boundary slope set is bounded from above by the twice of the minimal crossing number for a montesinos knot . |
molecular clouds exhibit a hierarchical density structure with stars forming in the densest regions .
often , these star - forming complexes have an elongated , filamentary shape ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
in fact , recent observations by the _ herschel _ space observatory show that parsec - scale filaments are ubiquitous in the interstellar medium ( ism ) ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the filaments tend to extend out from dense star - forming hubs @xcite and , sometimes , these hub - filaments structures are part of a larger - scale pattern with parallel filaments @xcite .
the filaments exhibit a nearly universal width of @xmath1 pc ( measured by fwhm of the column density ) , independent of the central column density or length of the filament @xcite .
furthermore , the radial - dependence of the column density profile is flatter than expected for an isothermal cylindrical filament in hydrostatic equilibrium @xcite .
a number of explanations have been suggested for the observed density profiles : the filaments correspond to stagnant gas in locally colliding flows @xcite , they are isothermal equilibrium cylinders in pressure - balance with the external medium ( i.e. a radially - truncated cylinder in hydrostatic equilibrium ; * ? ? ? * ; * ? ? ?
* ) , or are supported by magnetic fields @xcite . thus , the origin of the filamentary cloud structure still remains unclear and heavily debated .
the clouds can be formed by compressive motions of gravitational and/or turbulent nature @xcite .
indeed , many numerical simulations produce filamentary structures by either isothermal driven turbulence ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , thermal instability in large - scale convergent flows ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) or behind a shock front ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , or by gravitational instabilities in self - gravitating sheets ( e.g. * ? ? ?
* ; * ? ? ?
some of the observational evidence points to a turbulent formation process for filaments and dense cores in the ism .
however , observations often show cores within clouds with linewidths that require little additional support beyond thermal pressure ( e.g. * ? ? ? * ; * ? ? ? * ) .
these quiescent cores are adequately modelled by bonnor - ebert spheres confining the cores by self - gravity and external pressure @xcite .
such objects show that gravity plays a more prominent role than assumed in the turbulent models .
magnetic fields are also important in the formation of the filaments .
studies comparing the orientation of the magnetic field with respect to the diffuse gas provide more insight ( e.g. * ? ? ?
optical polarization measurements of the taurus molecular cloud show that the magnetic field in the diffuse gas is orientated perpendicular to the axis of the b216 and b217 filaments @xcite .
this suggests that the filaments form as gas streams along the magnetic field lines ( e.g. * ? ? ?
. however , the long axis of the l1506 filament in taurus and of the @xmath2 ophiuchus clouds lies along the magnetic field @xcite . in this paper
we focus on the role of self - gravity and magnetic fields on the formation of filaments and cores within self - gravitating layers .
we neglect the effect of turbulent motions ( other than produced by gravitational and magnetic instabilities ) .
self - gravitating isothermal layers are unstable to perturbations if the exciting modes are of sufficiently long wavelength @xcite and they fragment into clumps or thin filaments @xcite .
when the layers are threaded by magnetic fields , fragmentation still occurs . using a linear perturbation analysis , @xcite
show that , for a layer with a thickness larger than the pressure scale height , the magnetic field suppresses the growth of perturbations perpendicular to the field .
perturbations along the field are unaffected so that parallel filaments form within the slab . on the other hand , for a layer with a thickness smaller than the scale height , filaments form with their axis orientated along the magnetic field .
the separation of the filaments is given by the wavelength of the fastest growing mode .
the filaments then condense into clumps and dense cores as long as the filaments are in a quasi - static equilibrium @xcite .
we extend the study of @xcite into the non - linear regime and discuss the resulting structures . in sect .
[ sect : initial_conditions ] we describe the initial conditions of the self - gravitating layer and the numerical model used .
then , we discuss the fragmentation of the layer in absence ( sect .
[ sect : no_magnetic_field ] ) and presence of a parallel or perpendicular magnetic field ( sect .
[ sect : magnetic_field ] and [ sect : perp_magnetic_field ] ) . in sect .
[ sect : comparison ] we interpret the structure of the molecular cloud g14.225 - 0.506 in terms of the discussed gravitational instability model . finally , we conclude and discuss our results in sect . [
sect : discussion_and_conclusions ] .
for our initial conditions we assume a self - gravitating isothermal layer that extends to infinity along the @xmath3-plane .
the layer is initially in hydrostatic equilibrium , so that the density distribution along the @xmath4-axis is given by @xcite @xmath5 with @xmath6 the pressure scale height , @xmath7 the isothermal sound speed , @xmath8 the gravitational constant and @xmath9 the unperturbed density at the midplane ( i.e. at @xmath10 ) .
the layer is truncated at @xmath11 and confined by a constant external thermal gas pressure @xmath12 . in this paper
we will not study the effect of the external pressure on fragmentation .
therefore , we only consider a layer bounded by a low external pressure , i.e. @xmath13 .
we adopt @xmath14 , so that @xmath15 . outside the layer ,
not only the pressure , but also the density , @xmath16 , is kept constant at a small fraction of the density at the boundary ( see later for exact value ) .
this means that the external temperature differs from the internal temperature .
we , therefore , modify the isothermal equation of state to be @xmath17 this reflects the difference in external and internal temperature with @xmath18 a scalar that is 1 inside the layer and 0 outside .
the layer and the external medium are threaded by a uniform magnetic field .
for parallel models the magnetic field is along the @xmath19-axis , while , for perpendicular models , it is along the @xmath4-axis .
the field strength is given by @xmath20 at the midplane with @xmath21 the magnetic pressure .
for our models we use @xmath22 . for simplicity , our simulations are performed dimensionless .
we adopt @xmath23 , @xmath24 and @xmath25 so that @xmath26 and the dynamical time @xmath27 . the free - fall time associated with gas at the midplane
is @xmath28 . for a layer with a central density of @xmath29 ( and thus a surface density of @xmath30 ) and a temperature of 10 k
, these values correspond to @xmath31 pc , @xmath32 ( for @xmath33 ) and @xmath34 myr . in the rest of the paper we will evaluate the dimensionless results in terms of this layer .
this set - up is similar to the one used in the linear analysis of @xcite allowing a comparison between analytically and numerically derived growth rates of the gravitational instability . to study the gravitational instability of a pressure - confined self - gravitating layer , we perturb the equilibrium state .
we do this in two different ways , i.e. we superimpose the density either with a sinusoidal density perturbation of a given wavelength or with randomly generated noise .
the former is used to determine the growth rates to compare with analytic values , while the latter shows how different wave modes interact with each other . from the linear analysis @xcite suggest that first the filaments form and only then the dense cores . to solve the ideal magnetohydrodynamics ( mhd ) equations we use the adaptive mesh refinement ( amr ) mhd code mg @xcite .
the basic algorithm is a second - order godunov scheme with a linear riemann solver @xcite . to ensure that the solenoidal constraint is met , a divergence cleaning algorithm is implemented in the numerical scheme @xcite .
a hierarchy of grids levels is used with a mesh spacing on grid level @xmath35 of @xmath36 with @xmath37 the cell size at the coarsest level .
while the two coarsest levels cover the entire domain , finer levels generally do not .
refinement is on a cell - by - cell basis and is controlled by error estimates based on the difference between solutions on different grid levels .
self - gravity is computed using a full approximation multigrid to solve the poisson equation .
the numerical domain is given by @xmath38 and @xmath39 where @xmath40 is either @xmath41 .
here , @xmath42 is the critical wavelength below which the gravitational instability is suppressed and @xmath43 the wavelength for which the growth rate is maximal ( see * ? ? ?
* ) . for our adopted values , @xmath44 and @xmath45 .
as the layer is assumed to be infinite , we set the boundary condition on both the @xmath19 and @xmath46-axis to periodic .
for the @xmath4-axis we use free - flow boundary conditions .
the resolution of the simulation is set by the vertical extent of the slab .
it is necessary to properly resolve the pressure scale height @xmath47 in order to find a proper balance between pressure gradients and self - gravity .
we resolve the scale height by at least 5 cells .
therefore , we use an amr mesh with the coarse level having 4 grid cells along the axis with the shortest length ( along the @xmath3-axis for the runs up to @xmath48 and along the @xmath4-axis for @xmath49 ) and 128 cells along the @xmath4-axis for the finest grid level .
this means that the number of additional grid levels varies with different models , i.e. from 2 extra grid levels for @xmath50 up to 5 grid levels for @xmath51 .
this resolution is more than sufficient to resolve the jeans length as suggested by @xcite to avoid artificial fragmentation .
in fact , artificial fragmentation is only expected above @xmath52 .
as we are interested in the gravitational instability in the equilibrium layer , we need to ensure that the external medium does not affect the gravitational instability in the layer in any way . by assuming a constant pressure in the external medium
, the external medium needs to be a vacuum .
( the gravitational force is zero if there is no pressure force . ) however , in a grid code , we can not set the density to zero .
for a non - zero external density , the self - gravity of the layer drags the external medium towards it . then the external medium affects the hydrostatic layer because gas is accreted onto the layer and because the ram pressure contributes to the total external pressure .
we find that the external density needs to be at most @xmath53 not to affect the equilibrium layer .
as the external pressure is @xmath54 , this means that the external sound speed is @xmath55 times larger than in the layer .
this significantly limits the stable numerical time step derived from the courant condition @xmath56 .
because of this constraint , we use a different approach , i.e. we artificially switch off the self - gravity in the external medium .
this is easily done as the different media have a different value of the scalar @xmath18 .
this intervention allows us to set the external medium density to @xmath57 without affecting the equilibrium layer and gives us a speed up of more than a factor of ten in computational time .
in order to determine the effect of the magnetic field on the density distribution and geometry of the filaments , we first study the onset and evolution of the gravitational instability in the absence of a magnetic field .
we start by confirming the theoretically derived growth rates and continue by examining the interaction of different wave modes .
( black ) , @xmath58 ( blue ) , @xmath59 ( red ) and @xmath60 ( green ) .
the dashed lines show the linear evolution using the theoretically derived growth rates . for a layer with a central density of @xmath29 ,
i.e. our default layer , a unit of time is @xmath61myr.,width=302 ] [ [ growth - rates ] ] growth rates + + + + + + + + + + + + figure [ fig : growth_rates_hydro ] shows the evolution of the maximum density perturbation amplitude for modes with a wavelength of @xmath62 , @xmath58 , @xmath59 and @xmath60 .
the initial amplitude of the perturbation is 1% of the local density . for a perturbation with a wavelength below @xmath58
, the amplitude of the perturbation gradually decreases .
the wave mode is damped and the self - gravitating layer is stable under such a perturbation .
this remains so until the wavelength of the perturbing mode reaches @xmath58 . this critical wavelength ( hence @xmath58 )
has a zero growth rate and the amplitude of the critical wave mode remains roughly constant for multiple dynamical time scales .
it also marks the transition to a layer that is unstable to perturbations with @xmath63 .
our simulations reproduce the theoretically derived linear growth rates of @xcite for these longer wavelength perturbation very well . during the linear growth phase
, the amplitude of the perturbation grows as @xmath64 .
for the fastest growing wave mode with wavelength @xmath59 , we find @xmath65 , which is close to the analytically derived value of @xmath66 @xcite . for @xmath67 , @xmath68 and is similar to the value inferred from fig . 1 of @xcite .
note that the linear growth phase ends when @xmath69 and the growth becomes non - linear .
this non - linear growth phase lasts until nearly all the gas of the layer is concentrated in a single dense filament which is starting to fragment into cores .
we do not have sufficient resolution to follow this fragmentation and , at this time , we end the simulation .
( blue ) , 1 ( green ) and 0.1 ( red ) model .
the dashed lines shows the linear growth of a wave mode with a wavelength @xmath59 . for our default layer , unit of time
is @xmath61myr.,width=302 ] [ [ filament - and - core - formation ] ] filament and core formation + + + + + + + + + + + + + + + + + + + + + + + + + + + we have confirmed the growth rates for different wave modes , but we need to understand how the different wave modes interact with each other . therefore , we superimpose the equilibrium density with random noise .
the maximal amplitude of the density noise is 10% of the local density .
we allow wave modes with a wavelength up to @xmath67 to develop within the layer , i.e. the numerical domain is @xmath70 .
figure [ fig : growth_rate_noise ] shows the maximum amplitude of the density perturbations .
initially , as we impose random noise at the grid level , the wave modes have a wavelength below @xmath58 .
these modes are stable and the amplitude thus decreases .
however , wave modes with longer wavelengths for which the layer is unstable are excited and gradually grow .
the fastest growing unstable wave mode , i.e. the mode with a wavelength @xmath59 , becomes dominant .
this can be clearly seen in fig .
[ fig : growth_rate_noise ] as the evolution of the maximum perturbation amplitude can be described by the linear growth phase of that wave mode .
-axis for a layer without magnetic fields at @xmath71 ( top ) and at @xmath72 ( bottom ) .
the boxes show filaments selected for further analysis . in terms of our default layer , the area shown is 8@xmath738 pc@xmath74 and the unit of surface density is 5@xmath7310@xmath75 .
, title="fig:",width=302 ] -axis for a layer without magnetic fields at @xmath71 ( top ) and at @xmath72 ( bottom ) . the boxes show filaments selected for further analysis . in terms of our default layer ,
the area shown is 8@xmath738 pc@xmath74 and the unit of surface density is 5@xmath7310@xmath75 .
, title="fig:",width=302 ] .
the local maxima are indicated by blue crosses.,width=302 ] figure [ fig : column_pl_hydro ] shows the resulting surface density integrated along the @xmath4-axis at two different times , i.e. at the beginning of the linear growth phase at @xmath71 and near the end of the non - linear growth phase at @xmath72 .
a network of interconnecting filaments with embedded cores has formed .
note that regions of high column density are already present early on in the linear phase .
in fact , there is a one - to - one correlation . to better understand the interplay between the different filaments and the positions of the cores
, we use a ridge finding algorithm to expose the skeleton of the network and to locate column density maxima ( e.g. * ? ? ?
the ridges and local maxima for the bottom panel of fig .
[ fig : column_pl_hydro ] are shown in fig .
[ fig : ridge_pl_hydro ] . note that filaments have a separation of about @xmath59 indicating that the filaments are indeed formed by the most unstable wavelength ( as already inferred from the growth rates ) .
although , in the absence of magnetic fields , the most unstable wave mode has the same growth rate in all directions , filaments predominantly form along the @xmath19 and @xmath46 axes .
this is most likely because the flow equations are evolved along the cartesian coordinate axes in the numerical code and thus excite the wave modes along these directions .
another interesting feature is that the local maxima , i.e. dense cores , lie at the intersections of the filaments .
this is reminiscent of the hub - filament structures discussed by @xcite and the network of filaments seen in the _ herschel _ observations of the rosette molecular cloud and pipe nebula @xcite .
such a network is also present in cloud formation simulations which include turbulence ( e.g. * ? ? ?
* ) , but is generated due to dynamical ( other than gravity ) and thermal instabilities .
other random noise initializations produce a similar network of filaments with a separation of about @xmath59 and dense cores at the intersections of filaments . [ [ radial - profile - of - filaments ] ] radial profile of filaments + + + + + + + + + + + + + + + + + + + + + + + + + + + several authors ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) characterise the filaments by mapping the observed column densities with plummer - like profiles .
they assume that the underlying density profile of the cylindrical filament is given by @xmath76^\frac{p}{2}},\ ] ] where @xmath77 is the cylindrical radius , @xmath78 the characteristic radius of the flat inner region and @xmath79 the power - law index of the profile at large radii . for @xmath80 ,
this profile describes an isothermal cylinder in hydrostatic equilibrium with @xmath81 the scale height @xcite .
the resulting column density profile ( assuming that the filament lies within the plane perpendicular to the integration ) is also a plummer - like profile given by @xmath82^\frac{p-1}{2}},\ ] ] where @xmath83 is the euler beta function with input values @xmath84 and @xmath85 and @xmath86 the projected radius . :
region h1 is on the top panel , while region h2 is on the bottom one .
we display the central column density at a radius of 0.0015 and not at a radius of 0.0 .
the solid circles show the column densities at a given radius from the filament axis averaged over the entire filament while the error bars are the dispersion on the mean .
the blue line is the best fit plummer - profile with a background column density , while the red and purple lines show the plummer - like profile for @xmath87 and @xmath88 , respectively . for our default layer ,
the unit of length is 1 pc and the unit of surface density is 5@xmath89.,title="fig:",width=302 ] : region h1 is on the top panel , while region h2 is on the bottom one .
we display the central column density at a radius of 0.0015 and not at a radius of 0.0 .
the solid circles show the column densities at a given radius from the filament axis averaged over the entire filament while the error bars are the dispersion on the mean .
the blue line is the best fit plummer - profile with a background column density , while the red and purple lines show the plummer - like profile for @xmath87 and @xmath88 , respectively . for our default layer ,
the unit of length is 1 pc and the unit of surface density is 5@xmath89.,title="fig:",width=302 ] we extract several filaments from the column density map at @xmath72 . for each filament ,
we derive the radial distributions perpendicular to the axis at each cell along the filament axis . by combining all the distribution along a filament axis
we obtain the average radial distribution of the filament .
note that , for curved filaments , radial profiles will cross each other , but this will not strongly influence the average radial distribution as long as the curvature is small .
we then fit the average distribution with the theoretical profiles of eq .
[ eq : column_density ] .
we add a constant background column density because the filaments are embedded in a layer .
figure [ fig : pl_hydro_profiles ] shows the profiles and fits for two filaments .
the best fit for region h1 is given by @xmath90 , @xmath91 and @xmath92 , while @xmath87 still produces a good fit .
however , an equilibrium model with @xmath80 can not reproduce the filament profile at all .
a similar result is found for region h2 where the best fit is @xmath93 , @xmath94 and @xmath95 .
these results agree well with observationally derived values of @xmath96 by @xcite and others .
the value of @xmath78 , however , varies significantly between the different filaments . for region
the inner flat region is about an order of magnitude larger than in region h2 . at @xmath97 with contour lines .
for the default layer , the units of density and length are 159.2 @xmath98 and 1 pc , respectively.,width=302 ] .
the open circles are along the @xmath46-axis and the solid ones along the @xmath4-axis . the blue lines ( solid along @xmath4-axis and dashed along @xmath46-axis ) are the best fit model for a schmidt - burgk profile , i.e. @xmath99 and @xmath100 , while the red line is the expected profile for an equilibrium cylinder with @xmath101 .
for our default layer , the unit of density is 159.2 @xmath98 . _
bottom : _ column density profile of the same figure .
the dots are the simulation values , while the blue , resp .
red , solid lines are the expected column density for the schmidt - burgk profile , resp . the ostriker profile .
the unit of surface density corresponds to 5@xmath89 , while the unit of length is 1 pc.,title="fig:",width=302 ] .
the open circles are along the @xmath46-axis and the solid ones along the @xmath4-axis .
the blue lines ( solid along @xmath4-axis and dashed along @xmath46-axis ) are the best fit model for a schmidt - burgk profile , i.e. @xmath99 and @xmath100 , while the red line is the expected profile for an equilibrium cylinder with @xmath101 . for our default layer ,
the unit of density is 159.2 @xmath98 .
_ bottom : _ column density profile of the same figure .
the dots are the simulation values , while the blue , resp .
red , solid lines are the expected column density for the schmidt - burgk profile , resp . the ostriker profile .
the unit of surface density corresponds to 5@xmath89 , while the unit of length is 1 pc.,title="fig:",width=302 ] the average radial column density profiles suggest that the filaments are far from equilibrium .
numerical simulations provide the opportunity to confirm this claim directly from the density distribution .
we examine different slices across the filament in region h1 .
figure [ fig : pl_hydro_crosssection ] shows one of these cross sections . although the center of the filament is nearly axisymmetric , the outer layers are flattened along the slab . an equilibrium profile
@xcite provides a good fit to the central region of the filament , but can not explain the asymmetry along the different axes ( see fig .
[ fig : pl_hydro_crosssection_slices ] ) .
this , however , does not imply that the structure is far from equilibrium . actually , the cross section is reminiscent of a modulated self - gravitating layer in equilibrium @xcite .
the density distribution of such a modulated layer is given by @xcite @xmath102 where @xmath103 is the midplane density of the unmodulated layer , @xmath104 the modulation factor between 0 and 1 and @xmath105 the scale height .
for @xmath106 we find the spitzer solution for a infinite layer , i.e. eq . [ eq : spitzer ] , while , for @xmath107 , it converges to the @xcite solution for an infinite cylinder . by applying a schmid - burgk profile to density slices through the center of the cross section shown in fig .
[ fig : pl_hydro_crosssection ] and along the @xmath46 and @xmath4 coordinate axes - axis is not at @xmath108 , but at @xmath109 due to the gridding of the numerical domain . ] , we find that the density profile of the filament is best reproduced with @xmath99 and @xmath100 ( see fig . [ fig : pl_hydro_crosssection_slices ] ) .
the fit is good for radii up to the scale height @xmath110 above which the density along the midplane is higher .
other slices of the filament can also be described by a schmid - burgk profile albeit with somewhat different values of @xmath104 and @xmath103 ( i.e. @xmath111 and @xmath112 ) .
this variation implies a density variation along the filament axis and reflects that cores are embedded within a filament . by integrating eq .
[ eq : schmid ] along the @xmath4-axis and using the derived parameters @xmath104 and @xmath103 , we find that the schmid - burgk profile reproduces not only the density profile , but also the column density profile ( see fig .
[ fig : pl_hydro_crosssection_slices ] ) .
again the fit is best for radii below the scale height , but it exhibits a flatter radial distribution than the @xcite profile .
this partly explains the flatter than the ostriker equilibrium radial profile observed in the mean column density distribution ( fig .
[ fig : pl_hydro_profiles ] ) .
some of the flattening is due to averaging profiles with marginally different parameters . , while the open ones are the values of @xmath103 normalized to the initial value . in terms of our default layer ,
a unit of time is 4.9myr .
, width=302 ] .
the vectors represent the mass - weighted velocities in the plane .
their size is an indication of their magnitude with the maximal velocity shown @xmath113 times the sound speed .
remember that the unit of column density is 5@xmath89 and that the sound speed is @xmath114 in our default layer.,width=302 ] [ [ evolution - of - the - filaments ] ] evolution of the filaments + + + + + + + + + + + + + + + + + + + + + + + + + + although the density profile shows a filament close to equilibrium , it is in fact dynamically evolving .
the filament profile changes with time , but can , at all instances , be described by a schmid - burgk profile ( see fig .
[ fig : sbevolution ] ) .
initially , the filament is only a small perturbation on top of the equilibrium layer ( i.e. @xmath115 ) , while it evolves towards an equilibrium cylinder at the end of the non - linear phase ( i.e. @xmath116 ) .
the evolution of a filament thus occurs through a sequence of quasi - equilibrium distributions .
note that @xmath103 does not remain constant during the evolution .
this suggests that gas does not accumulate in this filament segment .
figure [ fig : region1 ] indeed shows that , while gas flows towards the filaments , it is then diverted towards denser regions ( cores ) along the filament axis . the evolution of the filaments is set by the mass that the gravitational instability sweeps up .
the expected line mass is essentially determined by the wavelength of the fastest growing wavelength and given by @xmath117 with @xmath118 the column density of the equilibrium layer .
this value needs to be compared with the critical value for the line - mass of a stable cylinder , i.e. @xmath119 @xcite . for our model parameters , @xmath120 .
we can also derive line - masses for different cross - sections in filaments .
for example , we find @xmath121 for fig .
[ fig : pl_hydro_crosssection_slices ] ) .
while this is less than the predicted value , it still exceed the critical value of @xmath122 .
it is interesting to note that half of the line - mass lies within the central region bound by the scale height .
thus , the filament is collapsing radially while maintaining a quasi - equilibrium for radii up to the scale height . .
, width=302 ] [ [ evolution - of - the - dense - cores ] ] evolution of the dense cores + + + + + + + + + + + + + + + + + + + + + + + + + + + + as mentioned earlier , the dense cores arise at the junctions of the filaments .
the cores are elongated along the axis of the most dense filament . as they evolve , the cores become more centrally condensed ( see fig . [
fig : core_evol ] ) .
the cores collapse at a rate that is slower than expected for pressureless gravitational , i.e. homologous , collapse of a uniform sphere ( the free - fall time is between 0.1 and 0.2 for all the cores ) .
there are two main reasons .
firstly , the cores are embedded in and are formed out of cylindrical filaments .
mass accretion towards the cores is thus not isotropic but highly directional .
@xcite and @xcite show that the collapse time scales of non - spherical structures are longer than the corresponding spherical free - fall time scale . in the case of a cylindrical cloud
, the time - scale depends strongly on the aspect ratio of the cloud .
note that the above authors only consider the collapse of constant mass structures .
the mass of the filaments does change in our simulation . however , as the line mass of the filament is set by the fastest - growing wavelength , the mass of the filaments does not change significantly once the evolution is in the non - linear regime .
( black ) , @xmath58 ( blue ) , @xmath59 ( red ) and @xmath60 ( green ) .
the thin solid lines are for perturbations along the @xmath46-axis .
the dashed lines show the linear evolution using the theoretically derived growth rates . as before
, the unit of time corresponds to 4.9myr for our default layer.,width=302 ] secondly , the assumption of pressureless collapse is not valid .
the results for the filaments show that the inner region can be adequately described by an equilibrium profile in which self - gravity is nearly balanced by thermal pressure .
@xcite show that , in the case of equilibrium cylinders , this significantly reduces the collapse rate . as cores form out of filaments , it is likely that the thermal pressure plays a similar role in cores .
however , we can not easily test this hypothesis as there is no analytic solution for an equilibrium core within a cylindrical filament ( and certainly not for a core in a schmid - burgk like filament ) .
furthermore , the accretion of gas is not along one filament , but multiple ones complicating the expected structure even more .
-axis for the @xmath123 model at @xmath72 ( top ) and the @xmath124 , model at @xmath125 ( bottom ) . in terms of the default layer ,
the area shown is 8@xmath738 pc@xmath74 and the unit of surface density is 5@xmath89.,title="fig:",width=302 ] -axis for the @xmath123 model at @xmath72 ( top ) and the @xmath124 , model at @xmath125 ( bottom ) . in terms of the default layer ,
the area shown is 8@xmath738 pc@xmath74 and the unit of surface density is 5@xmath89.,title="fig:",width=302 ]
[ [ growth - rates-1 ] ] growth rates + + + + + + + + + + + + the analysis of @xcite shows that the gravitational instability in an equilibrium layer is modified by magnetic fields as they stabilize instabilities propagating perpendicular to the field ( when the external pressure is small ) .
instabilities along the magnetic field are not affected .
the effect of the magnetic field is expressed as a function of a dimensionless parameter @xmath126 which for our model parameters is given by @xmath127 .
thus , our models examine a range of @xmath126 between 0.215 and 21.5 . for values as low as @xmath128 ( i.e. @xmath129 )
the perpendicular perturbations are already strongly suppressed ( see e.g. fig . 4 of @xcite ) .
similar to the hydrodynamical simulations we first examine whether our numerical model reproduces the expected linear growth rates .
we only test this for @xmath130 ( i.e. @xmath131 ) . for perturbations along the magnetic field ( i.e. along the @xmath19-axis )
, we find , as expected , linear growth rates that are marginally smaller than in the hydrodynamical model ( see fig .
[ fig : growth_rates_blz ] ) .
the critical wavelength and the maximal wavelength are also the same as before . for perpendicular perturbations along the @xmath46-axis ,
the growth rates are significantly reduced . for @xmath59
we find @xmath132 or about 20 times lower than for an identical perturbation along the @xmath19-axis , while we have @xmath133 when @xmath67 .
note that the growth rate for @xmath67 is larger than for @xmath59 as @xmath59 is here the value for the fastest growing wave mode along the @xmath19-axis and not the @xmath46-axis .
the critical and maximal wavelength along the @xmath46-axis are shifted towards longer wavelengths , i.e. @xmath134 and @xmath135 .
the shown models for perturbations perpendicular to the magnetic field then represent models close to the critical and maximal wavelength .
the simulations confirm the behavior derived from the linear analysis and show that the magnetic field introduces a preferred direction for the growth of perturbations , even for high values of @xmath136 .
( top ) and for @xmath124 ( red ) and @xmath137 ( blue ) at @xmath125 ( bottom).,title="fig:",width=264 ] ( top ) and for @xmath124 ( red ) and @xmath137 ( blue ) at @xmath125 ( bottom).,title="fig:",width=264 ] [ [ filament - formation ] ] filament formation + + + + + + + + + + + + + + + + + + we superimpose the equilibrium layer with the same random noise perturbations as the hydrodynamical model .
figure [ fig : growth_rate_noise ] shows the normalized maximum perturbation amplitude for @xmath138 ( as well as for @xmath139 , i.e. the hydrodynamical model ) .
the linear phase of the instability shows , for all of the magnetic models , the typical growth rate of a mode with a wavelength of @xmath59 . that this specific wave mode is dominant also becomes apparent from the column density structure (
see fig .
[ fig : pl_bx_noise ] ) . as the perpendicular perturbations
are significantly damped for the @xmath124 model ( see above ) , only the modes parallel to the magnetic field become unstable .
four parallel filaments are then produced with an separation of about @xmath59 .
note that the @xmath124 model is shown at @xmath125 and not @xmath72 as it reaches similar densities at later times than in the @xmath123 model .
although the layer is threaded by a magnetic field , the column density structure of the @xmath123 model shows filaments both parallel and perpendicular to the magnetic field direction .
in fact , the column density structure is very similar to the structure seen in the hydrodynamical model .
actually , the positions of the filament ridges are identical in both models ( see fig .
[ fig : ridge_pl_bx ] ) . the ridge positions for the stronger magnetic field models ( i.e. @xmath137 and 0.1 ) are also interchangeable .
remember that the density initialization is identical for all models and the subsequent evolution is then determined by the growth of the unstable modes .
this means we can identify two different regimes , i.e. a quasi - hydrodynamical regime and a strong magnetic field one .
the separation between the two states lies somewhere between @xmath129 and 10 . and
the maximum magnetic field strength displayed is 6@xmath140g.,width=302 ] not only the density structure , but also the magnetic field structure is distinctively different in the two regimes . in the strong magnetic regime ,
the magnetic field is dynamically important and suppresses perturbations perpendicular to its orientation .
then gas flows are predominantly along the magnetic field lines resulting in filaments perpendicular to the magnetic field .
contrary , in the weak magnetic regime , the magnetic field is dynamically passive and is dragged with the gas .
consequently , the density structures lie not only perpendicular to the field lines , but also parallel and oblique . for the oblique filaments
an interesting phenomenon occurs : while the magnetic field is oblique to the filament axis in the outer regions of the filament , it is orientated along the filament axis within its central region ( see fig .
[ fig : mag_orientation ] ) .
the filaments are bound by fast - mode shocks in which the magnetic field increases perpendicular to the shock normal .
the formation of such filaments is described in @xcite who show that such filaments are in quasi - static equilibrium in their inner parts .
( red ) and @xmath141 ( purple ) plummer profile .
for the default layer , the units of column density and length are 5@xmath89 and 1 pc , respectively.,title="fig:",width=302 ] ( red ) and @xmath141 ( purple ) plummer profile . for the default layer , the units of column density and length are 5@xmath89 and 1 pc , respectively.,title="fig:",width=302 ] ( red ) and @xmath141 ( purple ) plummer profile
. for the default layer , the units of column density and length are 5@xmath89 and 1 pc , respectively .
, width=302 ] for region s1 .
the solid circles show the density along the @xmath4-axis and the open circles along the @xmath19-axis .
the solid line is the profile for an ostriker cylinder with @xmath142 . in terms of the default layer ,
the length unit is 1 pc and the density unit 159.2 @xmath143.,width=302 ] [ [ radial - filament - profile ] ] radial filament profile + + + + + + + + + + + + + + + + + + + + + + + the average radial column density profile for the filaments in the magnetized layers is similar to the profiles extracted for the hydrodynamical model ( see figs .
[ fig : pl_bhx_profile ] and [ fig : pl_blx_profile ] ) .
the distribution is best fitted by plummer profiles ( eq . [ eq : plummer ] ) with power - law indices @xmath144 and can not be fitted by an equilibrium profile with @xmath80 .
an exception is region w1 which can be fitted reasonably well by a large range of @xmath79-values , including @xmath80 .
the ratio of the central - to - external column density for region w1 is small .
the distribution then does not sample the power - law tail of the plummer profile properly as the distribution is dominated by the flat inner region and the background column density .
this results in a broad range of possible power - law indices for the plummer profile .
again , density slices through the filaments show that the density distribution of the central region can be approximated by an equilibrium schmid - burgk profile . for example
, the density distribution for a slice at @xmath97 in region w1 is best fitted with @xmath145 and @xmath146 . a slice through @xmath147 of region s1 is best fitted by an ostriker cylinder ( or a schmid - burgk profile with @xmath148 ) with @xmath142 ( see fig . [
fig : pl_blx_slice ] ) .
the selection of these filaments represents different stages during the formation process , i.e. the filament of region w1 is starting to form , while region s1 shows the end of the formation process .
although the interpretation from the density profiles seems inconsistent with the one inferred from the average column density distribution , it is not . for filaments exceeding the critical line - mass ,
the density distribution is described by an equilibrium profile only for radii below the scale height .
the excess mass is at radii above the scale height resulting in a radial profile flatter than @xmath0 .
as the plummer profile only measures the radial dependence at radii above the scale height , such a description misses the equilibrium distribution in the center of the filament .
a more important result is that the filaments can be approximated by a hydrodynamical equilibrium without adjusting the sound speed ( to take into account additional support provided by magnetic support ) .
this suggests that the magnetic field is not important in setting the density distribution inside of the filaments . in filaments perpendicular to the magnetic field , gas flows along the field lines
. then magnetic pressure gradients are not generated and are thus much smaller than pressure ( or density ) gradients . in the quasi - hydrodynamical regime
, filaments also form parallel to the magnetic fields . because of magnetic flux conservation
, a scaling of the magnetic field with density is established , i.e. @xmath149 . the magnetic pressure , then , has a similar slope as the thermal pressure .
however , as @xmath150 , the magnetic pressure gradient is still much smaller than the thermal pressure gradient and does not contribute to the total pressure force balancing self - gravity .
[ [ dense - core - formation ] ] dense core formation + + + + + + + + + + + + + + + + + + + + the formation of dense cores is different in the two regimes . while cores form at the junctions of filaments in the weak magnetic regime , the cores in the strong magnetic regime form due to fragmentation of the filament .
however , @xcite show that the filaments only fragments if the growth rate of the unstable axisymmetric is shorter than the radial collapse time .
this condition is only fulfilled if @xmath151 . for our initial conditions ,
the line mass is much larger than the critical value , i.e. @xmath152 , and , thus , we do not expect cores to form within the filaments due to fragmentation .
the condensations that we see in the filaments ( see fig . [
fig : pl_bx_noise ] ) are actually a result of the initial conditions that break the uniformity of the layer .
but for perpendicular magnetic fields .
the solid lines represent a hydrodynamical ( black ) , @xmath123 ( blue ) , 1 ( red ) and 0.1 ( green ) model . the unit of time for our default layer is 4.9myr.,width=302 ]
the previous section discusses the gravitational instability in a layer threaded by a parallel magnetic field . here
, we examine the fragmentation for a magnetic field perpendicular to the layer .
@xcite study the marginally stable modes of an isothermal layer with perpendicular magnetic field .
they find that the layer is unstable to gravitational perturbations when @xmath153 or simply @xmath150 , with the critical wavelength close to the hydrodynamical value .
this means that , contrary to a parallel magnetic field , a perpendicular magnetic field can stabilize the layer as long as it is strong enough .
figure [ fig : growth_rates_bz ] indeed shows that perturbations in models with @xmath154 are damped .
for these layers to become unstable , the magnetic support needs to disappear by e.g. ambipolar diffusion @xcite .
-axis at @xmath155 for a perpendicular magnetic field with @xmath123 . for the default layer , the unit of column density is 5@xmath89.,width=302 ] .
the error bars represent the dispersion on the average value at a given radius .
the solid lines are the best - fit ( blue ) , a @xmath87 and @xmath80 plummer profile .
the unit of column density of the default layer is 5@xmath89 and the unit of length is 1 pc .
_ bottom : _ density profile along the @xmath19 ( open circles ) and @xmath4-axis ( solid circles ) for a slice through the filament .
the blue lines show the schmid - burgk profile with @xmath156 and @xmath157 along the @xmath19 ( dashed ) and @xmath4 ( solid ) axis . for the default layer ,
the unit of length is 159.2 @xmath143.,title="fig:",width=302 ] .
the error bars represent the dispersion on the average value at a given radius .
the solid lines are the best - fit ( blue ) , a @xmath87 and @xmath80 plummer profile .
the unit of column density of the default layer is 5@xmath89 and the unit of length is 1 pc .
_ bottom : _ density profile along the @xmath19 ( open circles ) and @xmath4-axis ( solid circles ) for a slice through the filament .
the blue lines show the schmid - burgk profile with @xmath156 and @xmath157 along the @xmath19 ( dashed ) and @xmath4 ( solid ) axis . for the default layer ,
the unit of length is 159.2 @xmath143.,title="fig:",width=302 ] [ [ density - structure - and - filaments ] ] density structure and filaments + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the @xmath123 model has a similar , though slightly slower , growth rate as the hydrodynamical model .
this reduced growth rate is due to magnetic tension forces acting to straighten field lines ( which are bended because the flow in the layer is perpendicular to the field lines ) .
the mode associated with this growth still has a wavelength close to @xmath43 , as the resulting column density structure is very similar to the hydrodynamical model ( see fig . [
fig : pl_bhz_noise ] ) .
the filament ridges are actually in the same places . not only
does the column density structure look very similar , the column density and density profiles of the filaments show the same properties as in the previous models ( see fig .
[ fig : pl_bhz_prof ] ) .
the power index of the best - fit plummer profile has a value of @xmath158 , but , at the same time , the central density structure of the filaments is perfectly described by an equilibrium schmid - burgk profile .
again , this means that the magnetic field only contributes marginally to the dynamics .
the plasma @xmath159 near the center of the filament decreases but is still above @xmath137 . model .
the solid line shows the dependence for an isotropically contracting sphere . in terms of the default layer ,
the units of the magnetic field and density are given by 1.76@xmath140 g and 152.9 @xmath143.,width=264 ] [ [ dense - cores ] ] dense cores + + + + + + + + + + + while the magnetic field does not influence filaments to any significant degree , the evolution of the cores are affected by them .
dense cores still arise at the junctions of the filaments , but the density distribution of the cores is very much axisymmetric . in the hydrodynamical model ,
gas is predominantly fed to the cores along the filaments and not isotropic as suggested here . because of magnetic tension forces , gas in the vicinity of the cores flows directly towards the core instead of first to a filament and then towards the core .
the former produces a magnetic field that is less ` twisted ' .
the axisymmetry of the dense core is also observed in the scaling relation of the magnetic field and density relation , i.e. @xmath160 ( see fig . [
fig : bdens ] ) .
such a relation is obtained for an isotropically contracting sphere threaded by a frozen - in magnetic field .
although the magnetic energy increases more rapidly than the thermal pressure in the core ( i.e. @xmath161 ) , the magnetic field is never strong enough to prevent collapse .
the potential energy of a spherical core increases at the same rate as the magnetic energy during contraction .
initially , the gravitational energy is larger than the magnetic energy and thus remains so during the subsequent evolution .
recently , @xcite reported vla observations in nh@xmath162 of a massive star forming complex g14.225 - 0.506 .
the nh@xmath162 emission reveals a network of filaments that are aligned in parallel in projection on to the plane of the sky ( see fig .
[ fig : g14 ] ) .
the filaments appear to take two directions , one group is oriented at a position angle ( pa ) of 10@xmath163 , and the other is at a pa of 60@xmath163 .
the averaged projected separation between the adjacent filaments are between 0.5 pc and 1 pc .
polarimetric observations in the near infrared h - band toward the northern section of the complex reveal magnetic field orientations perpendicular to the main axis of the filaments .
the fragmentation of a layer threaded with a strong parallel magnetic field reproduces the g14.225 - 0.506 filament network ( see fig . [
fig : g14 ] ) .
this allows to infer the initial conditions of the initial , unperturbed layer .
the separation of the adjacent filaments is given by @xmath43 which results in a scale height of 0.04 - 0.08 pc for the unperturbed layer .
then the initial surface density is @xmath164 ( assuming a sound speed of 0.2 @xmath165 ) .
as the area covered by the filaments is 4.7@xmath73 8.7 pc@xmath74 , the unpertubed layer contains about 2900 - 5800 m@xmath166 .
this agrees well with the combined mass of 2600 m@xmath166 for the filaments in g14.225 - 0.506 @xcite .
also , an estimate of the magnetic field strength can be made .
for the strong parallel field model , @xmath167 and the required magnetic field is stronger than 12 - 25 @xmath140 g .
zeeman measurements of the magnetic field indicate that these values are towards the high end of , but within , the observed range in magnetic field strengths @xcite . the initial conditions needed to produce the density structure of the filamentary complex g14.225 - 0.506 by gravitational instability in the presence of a strong magnetic field are not unrealistic .
however , we can not exclude any other formation process such as e.g. colliding flows .
in this paper we have examined the gravitational instability in an isothermal equilibrium layer threaded by a magnetic field .
our main results are as follows . 1 .
an equilibrium isothermal layer threaded by parallel magnetic fields is always unstable for gravitational instabilities with wavelengths larger than @xmath42 . for perpendicular magnetic fields instabilities
only arise when @xmath150 .
the density structures of the unstable layer can be described by two different regimes , i.e. a hydrodynamical one and a strong magnetic one .
the network structure of filaments can then be used to estimate the magnitude of the magnetic field .
parallel filaments indicate that the magnetic field is strong with @xmath154 and that the field lines lie within the plane of the layer . on the other hand
, a filament - hub structure implies that the magnetic field is weak with @xmath168 . in this case
there is no connection between filament axes and the magnetic field , i.e. the magnetic field can lie parallel or perpendicular along the filament axes . 2 .
the filaments that form have a line - mass that exceeds the critical value for axisymmetric collapse and thus can not be in hydrostatic equilibrium .
however , at any time during their evolution , the central regions of the filaments are described by an equilibrium density structure . using a schmid - burgk ( 1967 ) density profile ( eq . [ eq : schmid ] ) ,
the evolution of a filament from a perturbation in the equilibrium layer ( @xmath115 ) towards an ostriker ( 1964 ) distribution ( @xmath148 ) is perfectly captured .
furthermore , because of the thermal support in the central region of the filament , the collapse time is much longer than the free - fall time scale ( similar to the results of @xcite for collapsing cylinders ) . +
although the central region of filaments is well described by equilibrium profiles , the distribution for radii above the scale height deviates from the equilibrium ( as that region contains an excess of mass ) .
the profile is then flatter than the expected power law of @xmath0 and similar to the results from _ herschel _ observations ( e.g. * ? ? ?
an additional flattening of the profile is obtained by averaging the column density profile along a filament with a central density variation .
an increase in the central density produces a larger central column density but also implies a smaller scale height ( i.e. @xmath169 ) . such a variation in the filament width
is actually observed along the l1506 filament in the taurus molecular complex @xcite . + by studying colliding flows ,
show that filaments also form due to dynamical and thermal instabilities in an initially sheet - like structure .
while their filaments exhibit the flat column density profile of the observations and our simulations , their formation and evolution does not follow quasi - equilibrium states .
argue that the filaments are just long - lived , persistent features of the flow .
however , the resolution of their simulation is not high enough to resolve the thermal scale height associated with the filaments ( e.g. h @xmath170 0.3 pc for their filament 1 , while the resolution is @xmath171 pc ) . without the proper resolution , thermal pressure forces can not balance gravity numerically .
we encounter the same problem when we analyze filaments with a too high central density ( @xmath172 in code units ) .
we also recognize the difficulty in resolving the thermal scale height in turbulent simulations where the density contrasts fluctuate rapidly .
3 . our simulations do not reproduce the other property of filaments , i.e. a near constant fwhm of @xmath1 pc .
as the central density of the filaments increases , the scale height continues to decrease .
as the scale height also depends on pressure support , i.e. @xmath173 , the pressure support within the filament needs to increase as it is collapsing .
our simulations show that the magnetic fields are not able to provide this extra support .
+ however , @xcite show that turbulence can be driven by accretion .
interstellar filaments indeed show that the filament velocity dispersion increases with column density and exceeds the thermal sound speed if they have a line - mass above the critical value @xcite .
different authors ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) show that the magnetic field does not prevent the formation and collapse of dense structures in molecular clouds .
however , these studies only consider a low level of magnetization . in case of a strong magnetic field , simulations show that the field dominates turbulence and inhibits the gravitational collapse .
the filaments in our simulations are purely hydrodynamical structures .
this means that the magnetic field does not play a significant role in setting the shape of the filaments , even in a low - beta plasma .
dense cores form at the same time as the filaments . in the hydrodynamical regime ,
the cores arise at the junctions of the filaments .
mass accretion of the cores depends on the orientation of the magnetic field , i.e. for perpendicular fields the accretion is roughly isotropic , while , for parallel fields , much of the gas is fed along the filaments and thus highly directional .
such non - isotropic mass accretion drastically changes the observational signatures of dense core collapse ( e.g. * ? ? ?
* ; * ? ? ?
enough mass is accreted for the cores to contract , but at rates slower than free - fall collapse .
this again suggests that thermal pressure ( and magnetic pressure to a lesser degree ) are able to roughly balance against self - gravity .
some caveats need to be taken into account when discussing these findings .
turbulent motions are ubiquitous in the ism and it is unlikely that a thick , quiescent equilibrium layers forms out of such a medium .
for example , simulations of colliding flows ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) show that the shock - bounded layer is prone to dynamical instabilities , such as kelvin - helmholtz and raleigh - taylor instabilities , but also to thermal instabilities .
however , @xcite show that , dense core profiles do not not depend strongly on the initial velocity perturbations , although their formation time scales are shorter for larger velocity fluctuations . for a thin isothermal layer ,
the nature of the gravitational instability changes and is very similar to an incompressible mode .
@xcite show that the density perturbations are then a result of a deformation of the boundary of the layer . when the layer is threaded by a strong magnetic field ( @xmath167 ) , filaments no longer form perpendicular to the magnetic field lines but parallel to them .
furthermore , the line - mass of the filaments is smaller than the critical value .
then the filaments do not longer collapse but are in equilibrium .
it is then expected that the filament fragments along its axis into cores with equal spacing between them .
we will study this possibility in a subsequent paper .
we thank tom hartquist , sam falle and phil myers for useful discussions , gemma busquet for providing the observational data of the g14.225 - 0.506 molecular cloud and the anonymous referee for his / her comments that improved the paper .
svl acknowledges support from the sma postdoctoral fellowship of the smithsonian astrophysical observatory .
the simulations for this work were run on the _ smithsonian institution high performance cluster _ ( si / hpc )
. alves , j. f. , lada , c. j. , & lada , e. a. 2001 , , 409 , 159 andr , p. , menshchikov , a. , bontemps , et al . 2010 , a&a , 518 , l102 + arzoumanian , d. , andr , p. , didelon , p. , et al . 2011 , a&a , 529 , l6 + arzoumanian , d. , andr , p. , peretto , n. , konyves , v. 2013 , , 553 , a119 audit , e. , & hennebelle , p. 2005
, , 433 , 1 ballesteros - paredes , j. , hartmann , l. , & vzquez - semadeni , e. 1999 , , 527 , 285 ballesteros - paredes , j. , & mac low , m .-
2002 , , 570 , 734 ballesteros - paredes , j. , klessen , r. s. , mac low , m .-
, & vazquez - semadeni , e. 2007 , protostars and planets v , 63 banerjee , r. , vzquez - semadeni , e. , hennebelle , p. , & klessen , r. s. 2009 , , 398 , 1082 basu , s. , & dapp , w. b. 2010 , , 716 , 427 bonnor , w. b. 1956 , , 116 , 351 busquet , g. , zhang , q. , palau , a. , et al . 2013 , , 764 , l26 collins , d. c. , padoan , p. , norman , m. l. , & xu , h. 2011 , , 731 , 59 crutcher , r. m. 2012 , , 50 , 29 curry , c. l. 2000 , , 541 , 831 de avillez , m. a. , & breitschwerdt , d. 2005 , , 436 , 585 dedner , a. , kemm , f. , krner , d. , et al .
2002 , journal of computational physics , 175 , 645 elmegreen , b. g. , & elmegreen , d. m. 1978 , , 220 , 1051 falle s. a. e. g. , 1991 , mnras , 250 , 581 falle , s. , hubber , d. , goodwin , s. , & boley , a. 2012 , numerical modeling of space plasma slows ( astronum 2011 ) , 459 , 298 fiege , j. d. , & pudritz , r. e. 2000 , , 311 , 105 fischera , j. , & martin , p. g. 2012 , , 542 , a77 galvn - madrid , r. , liu , h. b. , zhang , z .- y . , et al .
2013 , , 779 , 121 goldsmith , p. , heyer , m. , narayanan , g. , snell , r. , li , d. , & brunt , c. 2008 , apj , 680 , 428 gomez , g. c. , & vazquez - semadeni , e. 2013 , arxiv:1308.6298 goodman , a. a. , bastien , p. , myers , p. c. , & mnard , f. 1990 , , 359 , 363 goodman , a. a. , jones , j. t. , lada , e. a. , & myers , p. c. 1992 , , 399 , 108 heitsch , f. 2013 , , 769 , 115 heitsch , f. , hartmann , l. w. , & burkert , a. 2008 , , 683 , 786 inutsuka , s. , & miyama , s. m. 1992 , , 388 , 392 johnstone , d. , & bally , j. 1999 , apj , 510 , l49 keto , e. , rybicki , g. b. , bergin , e. a. , & plume , r. 2004 , , 613 , 355 kirk , h. , myers , p. c. , bourke , t. l. , et al .
2013 , , 766 , 115 klessen , r. s. , & hennebelle , p. 2010
, , 520 , a17 koyama , h. , & inutsuka , s
2000 , , 532 , 980 kudoh , t. , basu , s. , ogata , y. , & yabe , t. 2007 , , 380 , 499 kudoh , t. , & basu , s. 2008 , , 679 , l97 kudoh , t. , & basu , s. 2011 , , 728 , 123 ledoux , p. 1951
, annales dastrophysique , 14 , 438 li , p. s. , norman , m. l. , mac low , m .-
m . , & heitsch , f. 2004 , , 605 , 800 li , h .- b . , fang , m. , henning , t. , & kainulainen , j. 2013 , , 2585 liu , h. b. , jimnez - serra , i. , ho , p. t. p. , et al .
2012 , , 756 , 10 lindeberg , t. 1998 , international journal of computer vision , 30 , 117 menshchikov , a. , andr , p. , didelon , p. , et al .
2010 , a&a , 518 , l103 + molinari , s. , bally , j. , noriega - crespo , a. , et al .
2011 , , 735 , l33 myers , p. c. 1983 , , 270 , 105 myers , p. c. 2009 , , 700 , 1609 miyama , s. m. , narita , s. , & hayashi , c. 1987 , progress of theoretical physics , 78 , 1273 mizuno , a. , onishi , t. , yonekura , y. , nagahama , t. , ogawa , h. , & fukui , y. 1995 , apj , 445 , 161 nagai , t. , inutsuka , s .-
, & miyama , s. m. 1998 , , 506 , 306 nakajima , y. , & hanawa , t. 1996 , , 467 , 321 nakamura , f. , & li , z .- y . 2008 , , 687 , 354 nakano , t. , & nakamura , t. 1978 , , 30 , 671 ostriker , j. 1964 , , 140 , 1056 padoan , p. , & nordlund , .
2002 , , 576 , 870 palmeirim , p. , andr , p. , kirk , j. , et al .
2013 , , 550 , a38 peretto , n. , andr , p. , knyves , v. , et al .
2012 , , 541 , a63 pon , a. , toal , j. a. , johnstone , d. , et al .
2012 , , 756 , 145 schmid - burgk , j. 1967 , , 149 , 727 schneider , s. , & elmegreen , b. 1979 , apjs , 41 , 87 schneider , n. , csengeri , t. , hennemann , m. , et al .
2012 , , 540 , l11 smith , r. j. , shetty , r. , stutz , a. m. , & klessen , r. s. 2012 , , 750 , 64 smith , r. j. , shetty , r. , beuther , h. , klessen , r. s. , & bonnell , i. a. 2013 , , 771 , 24 spitzer , l. , jr .
1942 , , 95 , 329 tafalla , m. , myers , p. c. , caselli , p. , & walmsley , c. m. 2004 , , 416 , 191 toal , j. a. , vzquez - semadeni , e. , & gmez , g. c. 2012 , , 744 , 190 truelove , j. k. , klein , r. i. , mckee , c. f. , et al .
1997 , , 489 , l179 umekawa , m. , matsumoto , r. , miyaji , s. , & yoshida , t. 1999 , , 51 , 625 van loo s. , falle s. a. e. g. , hartquist t. w. , 2006 , , 370 , 975 van loo , s. , falle , s. a. e. g. , hartquist , t. w. , & moore , t. j. t. 2007 , , 471 , 213 van loo , s. , falle , s. a. e. g. , & hartquist , t. w. 2010 , , 406 , 1260 vzquez - semadeni , e. , ballesteros - paredes , j. , & klessen , r. s. 2003 , , 585 , l131 ysard , n. , abergel , a. , ristorcelli , i. , et al .
2013 , arxiv:1309.6489 | we examine the role of the gravitational instability in an isothermal , self - gravitating layer threaded by magnetic fields on the formation of filaments and dense cores . using numerical simulation
we follow the non - linear evolution of a perturbed equilibrium layer .
the linear evolution of such a layer is described in the analytic work of @xcite .
we find that filaments and dense cores form simultaneously .
depending on the initial magnetic field , the resulting filaments form either a spiderweb - like network ( for weak magnetic fields ) or a network of parallel filaments aligned perpendicular to the magnetic field lines ( for strong magnetic fields ) .
although the filaments are radially collapsing , the density profile of their central region ( up to the thermal scale height ) can be approximated by a hydrodynamical equilibrium density structure .
thus , the magnetic field does not play a significant role in setting the density distribution of the filaments .
the density distribution outside of the central region deviates from the equilibrium .
the radial column density distribution is then flatter than the expected power law of @xmath0 and similar to filament profiles observed with _
herschel_. our results does not explain the near constant filament width of @xmath1pc
. however , our model does not include turbulent motions .
it is expected that accretion - driven amplification of these turbulent motions provides additional support within the filaments against gravitational collapse .
finally , we interpret the filamentary network of the massive star forming complex g14.225 - 0.506 in terms of the gravitational instability model and find that the properties of the complex are consistent with being formed out of an unstable layer threaded by a strong , parallel magnetic field . |
the full understanding of the phase structure of qcd as the theory of strong interactions has become an important issue since the discovery of asymptotic freedom .
the possibility that high temperature qcd could show different properties from the theory at zero temperature has been addressed already in @xcite .
new features are also predicted for the theory at very high and intermediate baryon density @xcite , @xcite
. an essential role is played by the chiral symmetry which is supposed to be broken by the vacuum structure of the theory .
thus , before considering the high temperature phase transition with the chiral symmetry restoration , one has to deal with the problem of determining a nonzero order parameter which indicates the chiral symmetry breaking ( csb ) at zero temperature and density .
a strong simplification is obtained by considering the effective theory described by a linear @xmath0 model of scalar mesons coupled to an isospin multiplet of fermions with the same chiral symmetry group of the original action
. then csb corresponds to a mexican hat - shaped effective potential and the order parameter is identified with the vacuum expectation value ( vev ) of one of the scalar fields .
the infrared ( ir ) properties of a field theory such as the vacuum structure , can be analysed by means of the wilsonian renormalization group which generates a sequence of effective actions defined at some momentum scale @xmath1 , starting from the original action , by integrating out in the latter all the modes with frequency higher than @xmath1 .
this leads to the construction of a differential flow equation for the @xmath1-dependent action , generally known as exact renormalization group equation ( erge ) . in the past years many analytical and numerical approaches to the erge , applied to various theories , have been developed .
two recent extensive reviews of this topic are @xcite and a detailed erge analysis of the chiral phase transition can be found in @xcite . here
we shall consider just the simple case of the csb at zero temperature employing the sharp cut - off version of the erge , namely the wegner - hougton equation @xcite .
the cut - off dependence of the erge for the o(4 ) linear @xmath0-model has been addressed in@xcite where , however only the flow of the scalar potential is considered while the yukawa coupling is kept fixed .
in addition to the yukawa coupling flow , we are particularly interested in the behavior of the field renormalization .
in fact , if our model is really an effective theory which , at some scale , can naturally replace the original qcd action , then at this scale one expects the formation of the mesonic bound states characterized by a very small wave function renormalization function .
the reduction of qcd to the quark - meson action has been analysed before and a review of it is given in @xcite .
as expected the value of the scalar field renormalization is vanishingly small at the mesonic bound state formation scale which is found between 0.60 and 0.63 gev .
then , the flow of the field renormalization deserves particular attention .
the explicit form of the euclidean action at a scale @xmath1 is ( @xmath2 ) @xmath3 @xmath4 is a flavor doublet and the summation over the index @xmath5 is extended to @xmath6 colors . no csb fermion mass term is included and the action is fully invariant under chiral transformations . the presence of csb depends on the shape of the o(4 ) symmetric potential @xmath7 in the limit @xmath8 .
@xmath9 and @xmath10 are the @xmath0 and @xmath11 renormalization functions .
as long as @xmath7 has zero vacuum expectation value ( vev ) , we choose the following parametrization @xmath12 and consider the flow of the two parameters @xmath13 and @xmath14 neglecting all the irrelevant operators that could be generated during the flow . in the presence of a nonvanishing vev @xmath15 , according to @xcite , we reparametrize the potential in terms of @xmath15 and @xmath14 , recalling that @xmath13 can be expressed in term of these two parameters through the minimum condition .
the purpose of this double choice is practical , we take the couple of parameters which corresponds to the simplest set of flow equations . the wegner - houghton equation can be reduced to a set of coupled ordinary differential equations for the @xmath1 dependent parameters in ( [ eq : leq ] ) , i.e. @xmath16 , @xmath9 , @xmath10 , @xmath14 and @xmath13 or @xmath15 . in particular the equations for the parameters in the potential are deduced in @xcite , and the one for @xmath16 can be obtained by an analogous procedure .
the equations for @xmath9 , @xmath10 , ( which coincide as long as @xmath17 , according to the chiral simmetry ) are deduced following the procedure formulated in @xcite and already implemented in @xcite . in the following , the derivatives of @xmath9 and @xmath10 w.r.t .
the fields , which are supposed to be small , are neglected . the pion decay constant and the constituent fermion mass are two ir constraints for our equations .
in fact , in the limit @xmath8 we have @xmath18 gev and @xmath19 gev , as initial conditions for the equations .
if we start , according to @xcite , the flow at the uv scale @xmath20 gev with a convex potential ( @xmath21 ) , then at a certain scale @xmath22 , @xmath13 becomes negative generating a nonzero @xmath15 . in the following the three remaining initial conditions
imposed are @xmath17 at the csb scale @xmath22 and a particular value for @xmath9 at the uv scale ( the symmetry requires @xmath23 at the uv scale ) .
figure [ fig : uno ] shows , above @xmath22 ( fixed at the value @xmath27 gev ) , the flow of the scalar mass @xmath28 ( the subscript @xmath1 is omitted ) for two different initial values of @xmath29 and , below @xmath22 , the corresponding @xmath30 and @xmath31 .
the curves for @xmath30 and @xmath31 in the two cases are almost superposed indicating insensitivity to the uv condition on @xmath32 .
the value of @xmath22 is suggested by an ir stability criterion . in order to use a particular value of quark mass , @xmath24 , as the initial condition for the running quantity @xmath31 defined above
, @xmath31 must be practically flat around @xmath33 .
the same criterion should apply to @xmath34 but , since @xmath34 is smaller than @xmath24 , we find that @xmath30 at the scale @xmath35 is already a stable parameter . after fixing the ir condition @xmath36 with @xmath37 gev
, we have considered the ratio @xmath38 , plotted vs. @xmath22 in figure [ fig : due ] for two different values of @xmath24 .
the stability requirement indicates that @xmath22 should not be smaller than @xmath39 gev . in figures [ fig
: tre ] and [ fig : quattro ] the other running parameters are displayed for the same two initial values of @xmath32 that have been used in figure [ fig : uno ] .
it should be noted the small running of @xmath40 and @xmath41 , in the range of @xmath1 considered , and their large values which are clearly nonperturbative . as shown in figure [ fig : quattro ] ,
when going from the uv to the ir scale , the field renormalizations grow and , below @xmath22 , @xmath32 and @xmath42 increase differently .
since reasonably we expect , for the pion physics at scales around @xmath43 gev , field renormalization functions close to the unity , we must require very small values at @xmath20 gev .
namely we used @xmath44 and @xmath45 in the two cases here considered in figures [ fig : uno],[fig : tre ] , [ fig : quattro ] .
higher curves for @xmath46 correspond to the former case , lower curves to the latter .
the approximation here employed for the erge has been sufficient for exploring a nonperturbative region ( note that , as explained in @xcite , the renormalized couplings which include the field renormalization effect , are larger than those in figure [ fig : tre ] ) and it has yielded a very small ( vanishing ) scalar field renormalization around @xmath47 gev .
we hope to extend our analysis to the finite temperature case . 9 j.c .
collins and m.j .
perry , phys .
34 ( 1975 ) 1353 .
d. bailin and a. love , phys .
107 ( 1984 ) 325 .
m. alford , k. rajagopal and f. wilczek , phys .
422b ( 1998 ) 247 .
r. rapp , t. shafer , e.v .
shuryak and m. velkovsky , phys .
rev . lett .
81 ( 1998 ) 53 .
c. bagnuls and c. bervillier , exact renormalization group equations . an introductory review , preprint e - print arch.:hep - th/0002034 , feb 2000 .
j. berges , n. tetradis , c. wetterich , nonperturbative renormalization flow in quantum field theory and statistical physics , preprint mit - ctp-2980 , hd - thep-00 - 26 , e - print arch.:hep - ph/0005122 , may 2000 . submitted to phys .
j. berges , n. tetradis and c. wetterich , phys .
77 ( 1996 ) 873 .
jungnickel and c. wetterich , phys .
d53 ( 1996 ) 5142 .
j. berges , d .- u .
jungnickel and c. wetterich , phys .
d59 ( 1999 ) 034010 .
j. berges , qcd in extreme conditions and the wilsonian exact renormalization group , preprint mit - ctp-2829 , e - print arch .
: hep - ph/9902419 .
f.j . wegner and a. houghton , phys .
a8 ( 1972 ) 401 .
g. papp , b.j .
schaefer , h.j .
pirner and j. wambach , phys .
d61 ( 2000 ) 096002 . c. fraser , z. phys .
c28 ( 1985 ) 101 .
a. bonanno and d. zappal , phys .
d56 ( 1997 ) 3759 . | the wegner - houghton formulation of the exact renormalization group evolution equation is used in order to study the chiral symmetry breaking of the linear @xmath0-model coupled to an isospin doublet of quarks . a numerical investigation for a particular truncation of the equation which includes the scalar field renormalization function
is presented . |
d - instantons are responsible for several important nonperturbative effects in superstring theory .
firstly , they appear as euclidian dirichlet p - branes whose @xmath1 dimensional worldvolume wraps a supersymmetric @xmath1 cycle in the compactification manifold .
such configurations were discussed as euclidian membranes and fivebranes in calabi - yau compactifications of m - theory @xcite , instanton corrections to the hypermultiplet geometry near the conifold in iia on a calabi - yau manifold @xcite , d - string instanton corrections to @xmath2 and @xmath3 terms in type i string theory which are important for type i / heterotic duality @xcite , threshold corrections to @xmath2 terms and u - duality in toroidally compactified type ii theories @xcite . secondly , even in uncompactified ten dimensional iib superstring theory they make an appearance as @xmath4 branes @xcite , i.e d - branes with worldvolumes which are points in spacetime .
indeed in the context of bosonic string theory such d - instantons were the first dirichlet objects considered in the past .
the introduction of boundaries where the embedding coordinates satisfy dirichlet boundary conditions in all directions was used to implement point - like partonic behavior in dual model amplitudes which is a desirable feature for a string theory of strong interactions @xcite .
the fact that momentum is not conserved in the presence of dirichlet boundaries was used to define off - shell continuations of string scattering amplitudes @xcite .
it was also observed that the characteristic softness @xcite of string amplitudes in the high energy fixed angle scattering regime , does not hold in the presence of d - instantons @xcite .
polchinski @xcite showed in the context of bosonic strings that the new singularities which arise because of the pointlike nature of the open string boundary conditions cancel in a generalization of the fischler - susskind mechanism . in this scheme
the combinatorics of string perturbation theory in the presence of d - instantons changes and one has to include disconnected worldsheets of different topology whose boundaries are mapped into the same spacetime point . in the context of iib superstrings
d - instantons have several new properties .
a d - instanton is a bps solution which preserves half of the spacetime supersymmetries . in @xcite
the supergravity solution corresponding to a d - instanton was found .
the broken supersymmetries induce fermionic zero modes which have to be integrated over in order to restore spacetime supersymmetry .
this induces new effective ( thooft ) vertices in the low energy effective action . in @xcite
the instanton induced interactions were derived and the @xmath0 duality symmetry of iib superstring theory was used to conjecture a modular invariant completion of the @xmath2 term which contains tree level , one loop and instanton corrections . in @xcite these expressions
were related to similar terms in m - theory and in particular it was argued that they arise from a one loop calculation in eleven dimensions .
the material presented in this talk was obtained in collaboration with m.b .
green , h.h .
kwon and p. vanhove and the interested reader is referred to the original papers for a more detailed discussion .
the massless fields of iib superstring theory are given by the bosonic fields : a complex scalar @xmath5 , two second rank ast potentials @xmath6 @xmath7 , the zehnbein @xmath8 and a fourth rank ast potential @xmath9 with selfdual field strength and the fermionic fields : a complex spin @xmath10 dilatino @xmath11 and a complex spin @xmath12 gravitino @xmath13 of opposite chirality .
as usual in extended supergravity theories the scalar fields parametrize a coset , which for iib supergravity is @xmath14 . under @xmath15 transformations only the second rank ast transform as doublets .
the fermions @xmath16 and @xmath13 transform under the local @xmath17 with weight @xmath12 and @xmath10 respectively .
the metric and the fourth rank ast are inert under both transformations .
the scalars can be parameterized by a zweibein @xcite @xmath18 the group @xmath15 acts by matrix multiplication from the left and the local @xmath17 acts from the right and induces a shift @xmath19 .
we will use the @xmath17 gauge invariance to set @xmath20 .
a compensating gauge transformation accompanies an @xmath15 transformation in order to maintain the gauge condition . in this case a @xmath15 transformation of the complex scalar and the compensating @xmath17 transformation are given by @xmath21 it is useful to define a scalar field strength @xmath22 and a complex combination of the ast @xmath23 which have @xmath17 weight 2 and 1 respectively .
@xmath24}.\ ] ]
the classical d - instanton is a solution of the euclidian iib supergravity @xcite which preserves half the euclidian supersymmetry .
only the metric and the scalar fields are nontrivial in this background so the nontrivial part of the action is given by @xmath25 the bps constraint on the solution requires that @xmath26 the classical d - instanton solution is given , in the einstein frame , by setting @xmath27 . if the scalar is expressed as the sum of a constant term and an euclidian fluctuation , @xmath28 , the bps condition ( [ susyvar ] ) reduces to @xmath29 .
the dilaton equation of motion is solved by @xmath30 , @xmath31 .
the constant @xmath32 is quantized and given by @xmath33 where @xmath34 .
the action of a charge @xmath35 d - instanton is given by @xmath36 .
the charge @xmath35 d - instanton configuration preserves sixteen of the thirty - two supersymmetries .
the broken supersymmetries generate fermionic zero modes in the instanton background which have to be integrated over and they induce new vertices which soak up the zero modes @xmath37 where @xmath38 defines a tadpole of a field of iib supergravity which soaks up @xmath39 fermionic zero modes and @xmath40 .
in closed string perturbation theory processes in the presence of d - instantons are formulated by including worldsheets with boundaries , where the boundary is mapped to a point in spacetime corresponding to the position of the instanton . because the boundary conditions relate leftmoving and rightmoving worldsheet fields , a d - instanton preserves half of the closed string supersymmetry , i.e is a bps state .
the broken spacetime supersymmetry generators have the form of zero momentum open string fermion vertex operators and have a natural interpretation as generating fermionic zero modes in the instanton background . the simplest open - string world - sheet that arises in a d - brane process
is the disk diagram .
a tadpole diagram with one fermionic zero mode attached couples to the dilatino and the calculation in [ appa ] gives @xmath41 where @xmath42 is the holomorphic combination of the two dilatinos .
similarly one can evaluate the one point functions of the complex combination of the and two - forms , @xmath43 with two fermionic zero modes attached @xmath44 and the tadpole @xmath45 holomorphic gravitino with three fermionic zero modes attached .
of particular interest is the disk with four fermionic zero modes attached which couples to the graviton and the calculation in [ appb ] gives @xmath46 the results of the calculation in the appendix of tadpoles for the massless fields which absorb up to four fermionic zero modes are summarized with their weight under @xmath17 and the coupling in the following table [ cols="<,<,<,<",options="header " , ] there are a whole variety of instanton induced vertices which are in fact related to each other by supersymmetry .
the simplest case is given by a sixteen fermion term which is induced by sixteen disks with one holomorphic dilatino @xmath16 and one zero mode @xmath47 attached to each disk .
@xmath48 among many other terms we focus on a @xmath2 term which arises from four disks with a graviton and four fermionic zero modes attached to each disk .
@xmath49 the close connection between @xmath50 and @xmath51 terms is obviously a consequence of their common superspace origin which can be seen already in the rigid limit of the linearized theory .
consider a linear superfield @xmath52 ( where @xmath53 is a complex chiral @xmath54 grassmann spinor ) that satisfies the holomorphic constraint @xmath55 and the on - shell condition @xmath56 @xcite where @xmath57 ( recall that @xmath58 does _ not _ involve complex conjugation ) are the holomorphic and anti - holomorphic covariant derivatives that anticommute with the rigid supersymmetries @xmath59 the field @xmath60 has an expansion in powers of @xmath53 ( but not @xmath61 ) , describing the 256 fields in an on - shell supermultiplet .
@xmath62 where @xmath63 is the linearized fluctuation around a flat background with a constant scalar , @xmath64 .
note that because of the constraint @xmath56 the expansion terminates at @xmath65 .
the coefficients of the component fields are consistent with the conventions used in @xcite .
the terms indicated by @xmath66 fill in the remaining members of the ten - dimensional @xmath67 chiral supermultiplet , comprising ( in symbolic notation ) @xmath68 , @xmath69 , @xmath70 , @xmath71 and @xmath72 .
the linearized supersymmetric on - shell action has the form @xcite @xmath73,\ ] ] which is manifestly invariant under the rigid supersymmetry transformations , ( [ susys ] ) .
the various component interactions contained in ( [ actdef ] ) are obtained from the @xmath74 term in the expansion , @xmath75 = f(\rho_0 ) + \delta { \partial \over \partial \rho_0 } f(\rho_o ) + \half \delta^2 \left ( { \partial \over \partial \rho_0}\right)^2 f(\rho_o ) + \cdots.\ ] ] all terms are f - terms , i.e they are realized as integrals over half the superspace .
the fact that the lorentz index structure of the @xmath76 term in the instanton induced effective action is of the same form as the tree level @xcite and one loop @xcite perturbative terms together with invariance of the graviton under the @xmath0 duality of iib string theory make it very plausible to conjecture for the exact nonperturbative @xmath2 term @xcite .
@xmath77 where the nonholomorphic modular function @xmath78 is given by @xmath79 in the string frame the large @xmath80 expansion gives @xmath81 there are several interesting properties of this expansion
. there are terms corresponding to tree level , one loop and instanton contributions . the tree and one loop terms exactly agree with results in string perturbation theory . the fact that no higher perturbative corrections appear in ( [ expan ] ) together with the fact that all such terms have the form of integral over half the superspace , lead to the conjecture that the @xmath2 and related terms are protected from higher loop perturbative corrections by a nonrenormalization theorem .
recently substantial evidence for the validity of such a theorem has been obtained @xcite .
it is also interesting to note that it is possible to infer the value of certain integrals which are important for the calculation of the witten index for bound states of d0-branes @xcite from the modular function @xmath78 @xcite .
the nonholomorphic function @xmath78 is invariant under @xmath0 modular transformations ; this is because the graviton and hence @xmath76 is invariant under @xmath0 . for other terms
which are related to @xmath2 like the @xmath82 the modular function has to be generalized , since such terms transform with a nontrivial weight under the induced @xmath17 r - symmetry . for a term in the effective action which transforms with weight @xmath83
a natural generalization of ( [ modularguess ] ) is , @xmath84 this function transforms under @xmath0 with a @xmath17 weight @xmath85 @xmath86 the functions @xmath87 with different @xmath85 are related to @xmath88 by @xmath89 where @xmath90 is the nonholomorphic covariant derivatives which maps modular forms of weight @xmath91 into modular forms of weight @xmath92 . comparing the relations in ( [ actdef ] ) with ( [ moddef ] )
it becomes clear that the linearized superspace incorporated the action of the ordinary derivatives @xmath93 by expanding the function @xmath94 $ ] , but does not provide a covariantization .
one indication that the full nonlinear superspace should produce a structure like ( [ moddef ] ) can be obtained by considering the perturbative fluctuations around the instanton configurations which are supressed by powers of @xmath80 . for the @xmath82 term the lowest order contribution in the charge
one instanton sector comes from sixteen disks with one dilatino and one fermionic zero mode attached .
because the genus of the disk is @xmath95 and the insertion of a closed string vertex operator contributes one power of @xmath96 such a term contributes to order @xmath97 .
a term where two dilatini are on the same disk with two fermionic zero modes attached is of order @xmath96 and corresponds to a perturbative fluctuation .
there should be a contact term which contributes when the vertex operator of the two dilatini come close to each other .
such contact terms are suppressed by @xmath98 and should be responsible for the covariantization of the leading d - instanton terms . such terms with more than one closed string vertex operator on a disk together with disks with handles attached should produce the perturbative fluctuations around the instanton configuration [ expan ] .
in this section we briefly review the relation of @xmath50 term ( [ r4mod ] ) by evaluating the four - graviton one loop amplitude of eleven - dimensional supergravity compactified on a torus , @xmath99 , in the directions @xmath100 , @xmath101 .
the volume of the torus is @xmath102 and its complex structure , @xmath103 , may be expressed in terms of the components eleven - dimensional metric @xmath104 ( @xmath105 ) as @xmath106 and @xmath107 , where @xmath108 is the component of the iia one - form along the direction @xmath109 .
@xmath110-theory on @xmath99 is identified with @xmath111 on @xmath112 via setting @xmath113 .
the expression for the amplitude , obtained rather efficiently by considering a first - quantized super - particle in the light - cone gauge @xcite , has the structure , @xmath114 where @xmath115 is the graviton vertex operator for the @xmath116th graviton with momentum @xmath117 and is evaluated at a proper time @xmath118 around the loop ( and dimensional quantities are measured in units of the eleven - dimensional planck scale ) .
the trace is over the fermionic zero modes and the loop momentum in the directions @xmath119 and the integers @xmath120 and @xmath121 label the kaluza - klein momenta in the directions @xmath100 , @xmath101 of @xmath99 .
the trace over the fermionic modes gives an overall factor , @xmath122 , which contains eight powers of the external momenta and is the linearized form of @xmath50 .
the @xmath50 term in ( [ loopgen ] ) is obtained by setting the momenta equal to zero in the remainder of the loop integral .
it is convenient to perform a double poisson resummation in order to rewrite the sum over the kaluza
klein momenta as a sum over windings , @xmath123 and @xmath124 , of the loop around the cycles of @xmath99 , giving , @xmath125 where @xmath126 is contained in the zero winding term , @xmath127 , which has the divergent coefficient , @xmath32 .
it was argued in @xcite that in any regularization that is consistent with t - duality between the iia and iib string theories in nine dimensions this has to be replaced by a regularized finite value , @xmath128 @xcite
. presumably a microscopic description of m - theory ( such as matrix theory ) would reproduce this value . in the limit of zero volume ,
@xmath129 , this term disappears and only the term with coefficient @xmath130 survives .
similar results for the @xmath82 and related terms and the corresponding modular functions @xmath131 have been obtained in @xcite by evaluating different one loop amplitudes of @xmath110 compactified on @xmath99 .
i would like to thank m.b .
green , h.h .
kwon and p. vanhove for collaboration and discussions on matters presented in this talk .
i gratefully acknowledge the partiall support by a doe grant de - fg02 - 91er40671 , nsf grant phy-9157482 and a james s. mcdonnell grant 91 - 48 .
we give two examples of the calculation of one point functions on the half plane which give the tadpoles @xmath132 of closed string fields used in section [ secinter ] .
attaching one fermionic zero mode to the boundary leads to a non - vanishing one - point function for a combination of the two dilatinos from the @xmath133 and similarly for the dilatino @xmath134 the calculation of such correlation functions on the upper half plane is most easily done by doubling .
one uses the boundary conditions for the d - instanton to map the antiholomorphic fields of the vertex operators living in the upper half plane to holomorphic fields in the lower half plane and evaluates the correlation function using holomorphic correlators in the complex plane . putting ( [ dilatino1 ] ) and ( [ dilatino2 ] ) together the coupling of the dilatinos
is given by @xmath135 attaching two fermionic zero modes produces a coupling of the d - instanton to the antisymmetric rank 2 tensor fields from the and sector . for the antisymmetric tensor
the relevant amplitude is @xmath136 note that the @xmath137 can not be contracted into anything , the @xmath138 in @xmath139 acts as a current @xmath140 insertion which can be evaluated , @xmath141 where we defined @xmath142 .
only the first two terms proportional to @xmath143 contribute because the term proportional to just one @xmath144 matrix in the first line of ( [ bmunu1pt ] ) @xmath145 vanishes after the spinor wavefunctions @xmath146 and @xmath147 are antisymmetrized . in the last line we used the fact that @xmath148 which is true for any @xmath149 with non vanishing imaginary part by the residue - theorem .
for the antisymmetric tensor the relevant amplitude is given by @xmath150 again @xmath151 .
note that the first term vanishes because of the trace of @xmath144-matrices and bose - fermi symmetry of @xmath152 . to evaluate the second term we used the identity @xmath153 for ten dimensional @xmath144-matrices and the integral ( [ integral1 ] ) . attaching four fermionic zero modes
gives a coupling of the self - dual four form and the graviton to the d - instanton .
we focus here on the graviton where the relevant amplitude is given by @xmath155 using doubling and the well known correlation function for the superghosts and spin fields , in particular @xmath156 the operator @xmath157 acts on the spinor index associated with the i - th spin field as a lorentz transformation . as above only terms like @xmath158 are nonzero after antisymmetrization over the spinor wavefunctions @xmath159 .
for example the contributions of the form @xmath160 is given by @xmath161 acting on @xmath162 and @xmath163 acting on @xmath164 , which results in @xmath165 there are three other contributions in ( [ mm ] ) given by @xmath166 , @xmath167 and @xmath168 .
after antisymmetrization over all @xmath169 these terms add up to give like the result for the graviton four point function @xmath170 99 k. becker , m. becker and a. strominger , fivebranes , membranes and nonperturbative string theory , hep - th/9507158 , 456(1995)130 .
h. ooguri and c. vafa , summing up d instantons , hep - th/9608079 . c. bachas et al . ,
heterotic / type i duality and d - brane instantons , 9707126 ; e. kiritsis and n. a. obers , heterotic type i duality in d @xmath171 10-dimensions , threshold corrections and d instantons , 9709058 . c. bachas ,
_ heterotic versus type i _ , hep - th/9710102 ; + c. bachas and p. vanhove , _ the rules of d - brane calculus _ , cpth - s587.1297 .
e. kiritsis and b. pioline , on @xmath2 threshold corrections in iib string theory and ( p , q ) string instantons , 9707018 .
green , point - like states for type iib superstrings , hep - th/9403040 , 329(1994)435 .
green , dynamical pointlike structure and dual strings , 89(1977)89 .
a. cohen , g. moore , p. nelson , j. polchinski , semi off - shell string amplitudes , 281(1986)127 .
d.j . gross and p.f .
mende , string theory beyond the planck scale , 303(1988)407 .
green , the influence of worldsheet boundaries on critical closed string theory , 302(1993)29 ; + m. gutperle , multiboundary effects in dirichlet string theory , hep - th/9502106 , 444(1995)487 . j. polchinski , combinatorics of boundaries in string theory , hep - th/9407031 , 50(1994)6041 .
green , g.w .
gibbons and m.j .
perry , the instanton and seven - brane of type iib supergravity , hep - th/9511170 .
green and m. gutperle , effects of d - instantons , hep - th/9701093 , 498(1997)195 .
green and p. vanhove , d - instantons , strings and m - theory , hep - th/9704145 , 408(1997)122 . m.b .
green , m. gutperle and p. vanhove , one loop in eleven dimensions , hep - th/9706175 , 409(1997)177 .
green , m. gutperle and h. kwon , sixteen fermion and related terms in m theory on @xmath99 , 9710151 .
schwarz , covariant field equations of chiral n=2 d=10 supergravity , 226(1983)269 .
howe and p.c .
west , the complete n=2 d=10 supergravity , 238(1984)181 .
d.j . gross and e. witten , superstring modifications of einstein s equations 277(1986)1 .
green and j.h .
schwarz , supersymmetric dual string theory iii , 198(1982)441 .
i. antoniadis , b. pioline and t.r .
taylor , calculable @xmath172 effects , 9707222 .
n. berkovits , construction of @xmath2 terms in n=2 d=8 superspace , 9709116 . s. sethi and m. stern , d - brane bound states redux , 9705046 ; + p. yi , witten index and threshold bound states of d - branes , 505(1997)307 , 9704098 .
green and m. gutperle , d - particle bound states and the d - instanton measure , 9711107 .
a. kehagias and h. partouche , the exact quartic effective action for the type iib superstring , 9710023 . | an overview over effects of d - instantons in ten dimensional iib superstring theory is given , including the supergravity instanton solution , instanton induced effective interaction vertices , the conjectured @xmath0 invariant completion of such terms and the connection of such terms to a one loop calculation in eleven dimensional supergravity .
epsfig 16 ^ 16 # 1 hep - th/#1 # 1(#2)#3 _ nucl
. phys . _
* b#1 * ( # 2 ) # 3 # 1(#2)#3 phys . rept.*#1 * ( # 2 ) # 3 # 1(#2)#3_phys .
lett . _ * # 1b * ( # 2 ) # 3 # 1(#2)#3_phys .
rev . lett._*#1 * ( # 2 ) # 3 # 1(#2)#3_phys .
rev . _
* d#1 * ( # 2 ) # 3 # 1(#2)#3ann
. phys . *
# 1 * ( # 2 ) # 3 # 1(#2)#3_rev .
mod .
phys._*#1 * ( # 2 ) # 3 |
in this article , we consider the limiting behaviour for a class of stochastic recursions .
these recursions are natural approximations of continuous time stochastic equations .
they arise as models for fast , discretely evolving random phenomena @xcite and also as numerical discretizations of continuous stochastic equations @xcite .
the class is similar to the rough path schemes of @xcite ( see also @xcite , section 8.5 ) , but more general in the sense that the noise driving the recursion is not required to be a rough path , but may be an approximation ( or discretization ) of a rough path .
let @xmath0 and @xmath1 where @xmath2 is defined by @xmath3 for @xmath4 and @xmath5 for @xmath6 , @xmath7 .
for each @xmath8 , define @xmath9 by the recursion @xmath10 where @xmath11 , @xmath12 are noise sources and we use the notation @xmath13 to denote the matrix inner product .
let @xmath14 be a partition of a finite time interval @xmath15 $ ] , which gets finer as @xmath16 tends to infinity .
the vector @xmath17 should be thought of as an approximation of a random increment @xmath18 where @xmath19 is some given stochastic process ( a semi - martingale or fractional brownian motion , e.g. ) .
formally , the symbol @xmath20 means that the approximation gets better as @xmath16 tends to infinity .
likewise , the matrix @xmath21 should be thought of as some approximation of an iterated stochastic integral @xmath22 where @xmath23 denotes the outer product and where the notion of stochastic integration ( it , stratonovich or otherwise ) is given .
define the path @xmath24 \to\mathbb{r}^e$ ] by @xmath25 where @xmath26 is the largest mesh point in @xmath27 with @xmath28 [ note that we could equally define @xmath29 by linear interpolation , without altering the results of the article ] .
our objective is to show that the path @xmath29 converges to the solution of a stochastic differential equation ( sde ) driven by @xmath19 as @xmath16 tends to infinity . in the case where @xmath30 are the increments of a rough path ,
that is , @xmath31 and @xmath32 , the recursions we consider are precisely the rough path schemes defined in @xcite .
however , we only require that @xmath30 be _ approximations _ of rough paths .
this means the class of recursions we consider is much more general than the class of rough path recursions and includes many natural approximations that do not fall under @xcite .
this fact will be illustrated by the examples below . to see why such diffusion approximations should be possible ,
it is best to look at a few examples .
the most common variant of ( [ erecu ] ) is the `` first - order '' recursion , where @xmath33 , so that @xmath34 this resembles an euler scheme with approximated noise @xmath35 .
hence , it is reasonable to believe that there should be a diffusion approximation @xmath36 ( where @xmath37 denotes weak convergence of random variables ) , where @xmath38 satisfies the sde @xmath39 and @xmath40 denotes some method of stochastic integration ( e.g. , it , stratonovich or otherwise ) .
it turns out that the choice of approximating sequence @xmath41 of the increment @xmath42 has a huge influence as to what type of stochastic integral arises in the limit .
we now explore this idea with a few examples .
the first four examples are first - order recursions as in ( [ erec ] ) and the final two are higher order recursions , as in ( [ erecu ] ) .
suppose that @xmath43 is a @xmath44-dimensional brownian motion , let @xmath45 and define the partition @xmath27 with @xmath46 .
then clearly @xmath47 defines the usual euler
maruyama scheme on the time window @xmath48 $ ] .
it is well known that @xmath49 where @xmath38 satisfies the it sde @xmath50 this creates a feeling that any euler looking scheme , like ( [ erec ] ) , should produce it integrals . as we shall see in the next few examples , when some _ correlation _ is introduced to the random variables @xmath41 , this is certainly not the case .
less trivial recursions of the form ( [ erec ] ) have shown up in the areas of population genetics @xcite , econometric models @xcite , psychological learning models @xcite , nonlinear time series models @xcite and mcmc algorithms @xcite , to name but a few . here , we will list the example from @xcite ; our analysis follows that performed in @xcite . in @xcite ,
the authors consider the stochastic difference equation @xmath51 where @xmath52 and @xmath53 is a stationary sequence of random variables with @xmath54 , @xmath55 , @xmath56 and with mixing assumptions on the centered sequences @xmath57 and @xmath58 .
this recursion arises naturally in models for population growth and also gene selection , where the environment is evolving in a random way . since the equation ( [ eguess ] )
is linear , the solution can be written down explicitly . as a consequence ,
it is easy to directly identify the limiting behaviour of each term appearing in the solution , for instance with the help of prokhorov s theorem .
alternatively , we can incorporate the problem into the scope of this article by making ( [ eguess ] ) look more like the recursion ( [ erec ] ) .
we first write @xmath59 now if we replace @xmath60 and @xmath61 by their second - order taylor expansion , we obtain @xmath62 thus , if we set @xmath63 and @xmath64 then @xmath47 satisfies the recursion @xmath65 moreover , due to the assumptions on @xmath66 and @xmath67 , the functional central limit theorem for stationary mixing sequences implies that @xmath68 where @xmath69 are brownian motions with a computable covariance structure .
thus , we should expect a diffusion limit of the form @xmath70 by writing down the solution explicitly , it is shown in @xcite that this is indeed the case and the method of integration involves a correction term that is neither it nor stratonovich . in nonlinear scenarios ,
more sophisticated machinery is required @xcite , but this still entails quite heavy and often nonrealistic mixing assumptions on the stationary sequence .
the framework of martingale problems @xcite has proved quite suitable for this analysis @xcite . in @xcite , the authors beautifully address the case where the noise is a semi - martingale sequence , using the idea of a _ good _ sequence of semi - martingales .
the following example is taken directly from @xcite .
[ eggood ] let @xmath71 denote the periodic rate of return for a security with value @xmath72 .
it follows that @xmath73 the authors consider the case where @xmath71 is a semi - martingale difference sequence defined in such a way that , if @xmath74 denotes the partial sum process @xmath75 then @xmath76 where @xmath77 is a semi - martingale .
it is natural to expect a diffusion approximation @xmath78 with some undetermined method of integration @xmath79 . in @xcite ,
the authors provide a natural condition on @xmath74 that ensures this method of integration is it type , which is clearly the most natural .
sequences @xmath74 that satisfy this condition are called _ good _ semi - martingales .
thus , if @xmath74 is good , then @xmath80 where the integral is of it type .
the authors also permit for a class of semi - martingales which are a reasonable perturbation of a good semi - martingale .
for instance , suppose that @xmath81 , where @xmath82 is a good sequence of semi - martingales and @xmath83 is a sequence of semi - martingales with @xmath84 as @xmath85 ( hence @xmath86 ) .
now define @xmath87 where the integral is of it type and @xmath88_t$ ] where @xmath89 $ ] denotes quadratic covariation and suppose that @xmath90 as @xmath91
. then @xmath92 where @xmath93 satisfies the it sde @xmath94 so formally speaking , we have @xmath95 .
thus , two equally reasonable approximations of @xmath77 can yield two vastly different limiting diffusions .
this class of perturbed semi - martingales is comprehensive enough to cover virtually every diffusion approximation where the recursion is driven by a semi - martingale sequence .
the next example is a rather important one , which unfortunately does not fit into the classes of diffusion approximations already studied in the literature .
understanding the diffusion approximation for this example is one of the main motivations of this paper .
[ egfastslow ] let @xmath96 describe a chaotic dynamical system with invariant ergodic measure @xmath97 . define the fast slow system @xmath98 where @xmath99 and @xmath100 where @xmath101 denotes the torus @xmath102 and @xmath103 satisfies the centering condition @xmath104 .
if we assume that @xmath105 is a random variable with law @xmath97 , then the path @xmath106 becomes a random variable on cdlg space .
note that the assumption @xmath107 simply means that the chaotic dynamical system is started in stationarity .
fast slow systems of this type have been considered in @xcite and are fundamental to the understanding of naturally occurring physical systems with separated time scales @xcite .
previous attempts at diffusion approximations typically involve heavy mixing assumption on the dynamical system @xmath108 which are difficult to prove for most reasonable systems @xcite . in @xcite ,
the author develops an alternative it calculus , but only in the case where @xmath108 defines a partially hyperbolic dynamical system . in @xcite ,
the authors study the special case where the noise is additive .
this allows them to use path - space continuity properties to lift convergence of the partial sum process to convergence of @xmath47 .
let us see how fast slow systems fit into the recursion framework ( [ erec ] ) .
using a fourier expansion truncated at level @xmath44 , we can replace @xmath109 with the product @xmath110 where @xmath111 , @xmath112 , @xmath113 and similarly replace @xmath114 with @xmath115 .
hence , we obtain @xmath116 this clearly satisfies the recursion ( [ erec ] ) with @xmath117 and @xmath118 .
the limiting behaviour of the partial sums @xmath119 is well understood under extremely weak conditions on the dynamical system @xcite .
in particular , @xmath120 where @xmath121 is a multiple of brownian motion and @xmath122 .
thus , we would expect a diffusion approximation of the form @xmath123 in the situations that are already understood , namely partially hyperbolic dynamical systems @xcite or additive noise @xcite , the limiting stochastic integral shown to be neither it^ o nor stratonovich type .
thus , the interpretation of the integral in a more general setting is an important problem .
the more general family of recursions defined in ( [ erecu ] ) ( with @xmath124 ) arise when using a second - order approximation . naturally , it is easy to find examples from numerical analysis .
[ egtheta ] let @xmath19 be some stochastic process ( e.g. , fractional brownian motion ) and introduce the shorthand @xmath125 .
suppose we approximate a stochastic equation using a semi - implicit method of integration , for instance , @xmath126 it is easy to show that @xmath47 satisfies ( [ erecu ] ) with @xmath127 and @xmath128 .
for a simple stochastic process @xmath19 , like brownian motion , it is well known that the limit of this numerical scheme is @xmath129 where the integral is of stratonovich type .
but for more complicated objects like fractional brownian motion , it is not so simple @xcite .
thus , studying recursions of the type ( [ erecu ] ) can lead to a better understanding well - posedness for numerical schemes , that is , whether they are approximating the correct continuous time limit .
[ egsubdiff ] despite the article s title , its scope is not restricted to diffusions , in particular the results also concern sub - diffusions . in @xcite ,
the authors consider partial sum processes of the form @xmath130 where @xmath131 is a stationary dependent sequence of random variables and @xmath132 is some normalizing constant , such that @xmath133 in the skorokhod topology , where @xmath19 is fractional brownian motion with some hurst parameter @xmath134 depending on the correlation structure . with this in mind , it is natural to consider a recursion @xmath135 where @xmath21 is some approximation of an iterated integral defined using the sequence @xmath136 .
for instance , as indicated by example [ egtheta ] , if @xmath137 then the above recursion corresponds to a mid - point rule approximation of a stochastic integral .
in particular one would expect @xmath36 where @xmath138 and where the integral is of _ symmetric _ type @xcite , which is the natural limit of the mid - point scheme .
of course , this is only a guess and it is quite possibly wrong .
as we will see , the tools introduced in this article provide a natural basis for the investigation of such sub - diffusion approximations .
understanding such recursions is vastly important and could facilitate for the design of new methods for simulating stochastic differential equations driven by fractional brownian motion .
the technique employed in this article is similar in spirit to that found in @xcite , in that we will lift an approximation result for the noise signal into diffusion approximation for the recursion .
however , in our more general scenario , where we do not assume any particular probabilistic structure on the noise , we require not just an invariance principle for the noise but also for its _ iterated integral_. more precisely , define the noise signal @xmath139 which is the natural approximation of the limiting noise signal @xmath19 .
moreover , define the discrete iterated integral @xmath140 which is the natural approximation of the limiting iterated integral @xmath141 . in this paper
, we shall lift a limit theorem for the discrete pair @xmath142 into a diffusion approximation for the recursion @xmath47 .
in essence , the limiting behaviour of @xmath143 tells us what type of noise appears in the limiting stochastic integral and the limiting behaviour of @xmath144 tells us what type of stochastic integral we are talking about . looking back at example [ egfastslow ] , for instance
, this suggest that we can interpret the integral @xmath145 , provided we can identify the limit of the discrete iterated integral @xmath146 to derive this diffusion approximation technique , we use tools from _ rough path theory _ @xcite . for stochastic differential equations driven by piecewise smooth signals ,
the relationship between the noise and the solution is extremely well understood mostly thanks to rough path theory .
for the purpose of exposition , suppose that @xmath19 is some piecewise smooth stochastic process and that @xmath38 solves the equation @xmath147 where the integral is defined in the riemann
stieltjes sense .
it is well known that the map @xmath148 is not continuous in the sup - norm topology .
the theory of rough path proposes that we can build a continuous map from the noise to the solution , provided we know a bit more information about @xmath19 . in particular , suppose that we can define @xmath149 where the integral is again of riemann stieltjes type .
then one can show that the map @xmath150 is continuous in a topology called the @xmath151 topology ( known colloquially as the _ rough path topology _ ) .
this topology can be thought of as an extension of the @xmath152-hlder topology , defined on the space of objects similar to the pair @xmath153 .
the objects @xmath154 are called rough paths and the metric space of such objects is called the space of @xmath152-hlder rough paths .
this idea clearly has ramifications to the diffusion approximations . indeed , suppose that @xmath47 solves the stochastic equation @xmath155 for some smooth stochastic process @xmath143 and also define the iterated integral @xmath156 . since continuous maps preserve weak convergence , this suggests that a weak limit theorem for the pair @xmath142 in the @xmath151 topology can be lifted to a weak limit theorem for @xmath47 .
the general procedure can be summarized by two steps .
show that @xmath157 , where @xmath158 denotes convergence of finite - dimensional distributions .
show that the sequence is tight in the @xmath151 topology .
for instance , one could use a kolmogorov type argument , by checking estimates of the form latexmath:[\[\label{eestimates0 } \bigl ( \mathbf{e}\bigl|x^n(s , t)\bigr|^q \bigr)^{1/q } \lesssim|t - s|^\gamma\quad \mbox{and}\quad \bigl ( \mathbf{e}\bigl| \mathbb{x}^n(s , t)\bigr|^{q/2 } \bigr)^{2/q } \lesssim suitable @xmath152 and with @xmath160 large enough . since the map : _ rough path _
solution _ is continuous in the rough path topology , the conclusion from these two steps is that @xmath36 where @xmath38 is the solution to an sde whose form can be determined by the limit @xmath162 .
for instance , suppose that @xmath19 were a continuous semi - martingale and that @xmath163 where the above integral is stratonovich type .
then the limiting equation can be written @xmath164 this precise idea has proved useful in the areas of stochastic homogenization @xcite and equations driven by random walks @xcite .
unfortunately , for the recursion ( [ erecu ] ) the path @xmath47 does not satisfy a stochastic equation in the sense of rough path theory , so we can not simply apply the above procedure .
the objective of this article is to overcome this obstacle .
it turns out that the same two step procedure defined above , more or less still works .
all we have to do is replace iterated integrals with their discrete counterparts and replace step @xmath165 with the same statement _ up to some resolution_. that is , we need only check the estimates ( [ eestimates0 ] ) for all @xmath166 , which requires no continuity at all . in checking these discrete estimates , we obtain a tightness - like result for a discrete version of the hlder metric , defined ( on cdlg paths ) by @xmath167 this is of course _ always _ finite , since it is a maximum over a finite set , but the tightness result will tell us something about the asymptotics . at the heart of the proof is an approximation theorem ( theorem [ thmmod ] ) , which we believe to be useful in its own right .
the theorem allows us to approximate the recursion ( [ erecu ] ) with the solution to a stochastic differential equation driven by piecewise smooth paths .
this approximation theorem can be thought of as a generalization of _ the method of modified equations _ for sdes @xcite ( otherwise known as _
backward error analysis _
@xcite ) . in particular
, our approximation theorem has the advantage of being completely pathwise , without depending on the probabilistic properties of the stochastic process @xmath143 whatsoever . by approximating @xmath47 by the solution to a genuine stochastic equation
, we unlock the tools of rough path theory introduced above .
the outline of the paper is as follows . in section [ sresults ] ,
we sketch the main theorem of the paper . in section [ sapps ]
, we list a few applications . in section [ srough ] ,
we give a brief introduction to rough path theory and mention some results that are important to the present article . in section [ sdrp ] ,
we rigourously define rough paths recursions , these are the central objects to the article . in section [ sproperties ] , we derive the properties of rough path recursions that will be needed for the main theorem . in section [ sconvergence ]
, we prove the main theorem of the article , concerning weak convergence of rough path recursions .
in this section , we state the main theorem , avoiding the technical definitions that will be introduced in subsequent sections . in particular , the main theorem ( theorem [ thmmain ] ) can be stated and applied without requiring any knowledge of rough path theory and similarly for the approximation theorem ( theorem [ thmmod ] ) .
let @xmath168 be a partition of @xmath15 $ ] with mesh size @xmath169 . as stated above
, one should regard @xmath170 as an approximation of the increment @xmath171 likewise , one should regard @xmath172 as an approximation of the iterated integral @xmath173 the only consequence of this analogy is that it influences how we define the _ path _ corresponding to the incremental processes .
indeed , the increments can be anything at all , provided they satisfy the convergence properties stated in the theorem below .
to recap , the recursions we consider in this article are of the form @xmath174 where @xmath175 and @xmath176 for some @xmath177 and the implied constant is uniform in @xmath16 .
we now define the _ rough step - function _ @xmath142 corresponding to the increments @xmath178 .
if @xmath179 is the largest grid point in @xmath27 such that @xmath180 then @xmath181 we similarly define the incremental paths @xmath182 where @xmath183 is the largest grid point in @xmath27 such that @xmath184 .
it is easy to check that this is the natural choice , given the motivation ( [ einc1 ] ) and ( [ einc2 ] ) .
the main theorem is as follows .
[ thmmain ] let @xmath47 satisfy ( [ erecursion ] ) and let @xmath142 be cdl ` ag paths defined by ( [ epaths1 ] ) .
suppose that @xmath185 where @xmath19 is a continuous semi - martingale and @xmath186 is of the form @xmath187 where the integral is defined in the stratonovich sense and @xmath188 .
suppose that the pair @xmath142 satisfy the estimates @xmath189 \label{eestimates } \\[-8pt ] \nonumber \bigl(\mathbf{e}\bigl|\mathbb { x}^n \bigl(\tau ^n_j,\tau^n_k \bigr)\bigr|^{q/2 } \bigr)^{2/q } & \lesssim & \bigl|\tau^n_j - \tau ^n_k\bigr|^{2\gamma}\end{aligned}\ ] ] for all @xmath190 where @xmath191 , @xmath192 $ ] and the implied constant is uniform in @xmath16 .
then @xmath193 in the sup - norm topology , where @xmath38 satisfies the sde @xmath194 although we only require @xmath191 it is clear from @xmath195 $ ] that we always have @xmath196 .
[ rmkqsmall ] if the estimates ( [ eestimates ] ) hold for @xmath197 and all @xmath198 , then the condition @xmath199 can be relaxed to @xmath200 .
this follows using the standard techniques of @xmath201 rough paths ( see @xcite and @xcite , chapter 12 ) .
additional details will be given in remark [ rmkqlarge ] .
[ rmkdrift1 ] the result naturally extends to the case with an additional `` drift '' vector field @xmath202 @xmath203 in this setting , the limiting sde is given by @xmath204 this is a more natural way to treat the problem introduced in example [ egfastslow ] . as with remark [ rmkqsmall ] , this extension is a standard application of @xmath201 rough paths .
the next result is not so much a theorem as it is a guide for other theorems .
it applies to situations where the noise driving the limiting equation is not a semi - martingale , such as the sub - diffusions encountered in example [ egsubdiff ] .
[ thmmeta ] in the same context as above .
suppose that @xmath205 where @xmath19 is some continuous stochastic process and @xmath206 where @xmath207 denotes some constructible method of integration .
suppose moreover that @xmath142 satisfy the estimates ( [ eestimates ] ) .
then @xmath36 where @xmath38 satisfies the stochastic equation @xmath208 theorem [ thmmain ] and meta theorem [ thmmeta ] will be proved in section [ sconvergence ] .
the proof of the meta theorem indicates what we mean by a `` constructible method of integration '' . finally , the main tool used to derive the results above is the approximation theorem , which should be thought of as a pathwise version of backward error analysis ( or the method of modified equations ) @xcite .
the rate estimate depends on the _ discrete _ @xmath152-hlder norm @xmath209 which is the smallest number such that @xmath210 for all @xmath190 . since
this number can be achieved by taking the maximum over a finite set , it is clear that each @xmath209 is finite , regardless of the path @xmath211 .
we will always need some kind of asymptotic estimate on @xmath209 to make use of the approximation theorem .
[ thmmod ] suppose that @xmath29 is the path defined by the recursion ( [ erecursion ] ) and that the pair @xmath142 defined by ( [ epaths1 ] ) satisfy the estimates ( [ eestimates ] ) for some @xmath212 as in theorem [ thmmain ] .
then for each @xmath16 we can find a pair of piecewise smooth paths @xmath213 \to\mathbb{r } ^d\times\mathbb{r}^{d\times d}$ ] such that if @xmath214 solves @xmath215 where the integrals are of riemann stieltjes type , then @xmath216 where @xmath217 denotes the sup - norm and where the constant @xmath218 , where @xmath209 is the constant defined in ( [ ecn ] ) .
the proof of theorem [ thmmod ] is contained in section [ sproperties ] .
we will give one simple example to illustrate the idea behind this approximation theorem .
suppose that @xmath43 is a brownian motion and that @xmath219 it is easy to check that , for almost every brownian path , the constant @xmath209 defined in ( [ ecn ] ) is bounded uniformly in @xmath16 , for any @xmath220 .
it follows that we can find an equation driven by smooth paths , with solution @xmath214 such that @xmath221 for any @xmath220 .
we will now discuss some potential applications for the tools introduced above .
we start with a quite trivial and well known result , with the sole intention of illustrating how theorem [ thmmain ] should be used .
a continuous time version of this example can be found in @xcite .
it should be said that the following can easily be deduced from either @xcite or @xcite .
suppose that @xmath222 is an i.i.d .
sequence of random variables taking values in @xmath223 with @xmath224 and @xmath225 , with @xmath226 .
we will consider the random walk recursion @xmath227 with associated partition @xmath228 with @xmath229 .
if we define the path @xmath230 then it is well known that @xmath36 in cdlg space ( with sup - norm topology ) , where @xmath38 solves the sde @xmath231 where @xmath121 is standard brownian motion on @xmath223 .
the following lemma illustrates how to prove this using theorem [ thmmain ] .
first , we define the rough step function @xmath232 [ lem31 ] suppose that @xmath233 for some @xmath234 .
then the pair @xmath142 satisfy the assumptions of theorem [ thmmain ] with @xmath235 and @xmath236 .
in particular @xmath237 where @xmath231 where the integral is of it type . the moment condition on @xmath238 is much stronger than required by more traditional solutions to the problem .
this is due to the fact that the conclusion of the theorem is actually _ stronger _ than most traditional versions . in particular
, we could actually show that @xmath47 converges in ( a discrete version ) of the rough path topology , which is much stronger than the sup - norm topology .
we will not pursue such statements in this article .
proof of lemma [ lem31 ] from donsker s theorem , we already know that @xmath239 . to identify the limit of @xmath144 we simply write it as a stochastic integral . in particular , we see that @xmath240 where the integral is of left - riemann type ( hence it type ) .
that is , @xmath241 where @xmath242 is a partition of @xmath243 $ ] and the limit is taken as the maximum of @xmath244 tends to zero .
the theory of good semi - martingales @xcite provides a class of semi - martingale sequences for which the limit of a sequence of it integrals is an it integral .
since the partial sum process @xmath143 is clearly a martingale with respect to the filtration generated by the sequence @xmath245 , we can appeal to @xcite , theorem 2.2 .
in particular , since the quadratic variation @xmath246_t = n^{-1 } \sum_{i=0}^{\lfloor nt \rfloor-1 } \xi_i \otimes \xi_i,\ ] ] we have that @xmath247_t = d\lfloor nt \rfloor / n$ ] which is of course bounded uniformly in @xmath16 .
thus , @xmath143 is good and @xcite , theorem 2.2 , immediately tells us that @xmath248 in the sup - norm topology , where @xmath249 where the integrals are of it and stratonovich type , respectively , and we have converted between them in the usual way .
this is of course stronger than the finite - dimensional distribution result which we required , but the tools from @xcite make it quite easy to prove .
all that remains is to obtain the discrete tightness estimates ( [ eestimates ] ) with @xmath234 and @xmath197 .
since @xmath143 is a martingale , we can apply the burkholder davis
gundy ( bdg ) inequality @xmath250_{j / n , k / n}\bigr|^{q/2 }
\\ & = & n^{-q/2}\mathbf{e}\biggl{\vert}\sum_{i = j}^{k-1 } \xi_i\otimes\xi_i \biggr{\vert}^{q/2 } \leq n^{-q/2 } \mathbf{e } \biggl ( \sum_{i = j}^{k-1 } see that @xmath251 since @xmath144 is a stochastic integral ( or martingale transform ) it too is a martingale and hence we can again apply the bdg inequality . a similar argument yields @xmath252 and since @xmath234 , the interval @xmath253 $ ] is nonempty ,
so we do indeed satisfy the requirements of theorem [ thmmain ] .
it follows that @xmath36 where @xmath254 and we obtain the required expression by converting stratonovich to it . instead of showing how the tools can be used on each of the examples given in the
, we concentrate on the fast slow systems , since it is the least understood . the tools of this article are applied to fast slow systems in a companion paper @xcite ( see also @xcite ) , to yield new results for fast slow systems .
the dynamical system theory required is slightly too involved to be included in this paper , thus we will only sketch the ideas behind the result .
we will restrict our attention to the fast slow system @xmath255 the general case is treated in @xcite . setting @xmath256 , @xmath257 and @xmath258 we see that the rough step function is defined by @xmath259 we will also introduce the sigma algebra @xmath260 which is whatever sigma algebra we chose to go with the measure space @xmath261 . under `` sufficient '' mixing conditions on @xmath108 ,
the pair @xmath142 satisfy the assumptions of theorem [ thmmain ] with @xmath235 where @xmath262 and @xmath263 in particular , @xmath36 where @xmath264 where @xmath265 is defined precisely as @xmath266 , but in terms of @xmath103 .
sketch of proof to identify the limit of the pair @xmath142 , we proceed similarly to the random walk recursion case , namely identify the limit of @xmath143 and then lift it to @xmath144 . to identify the limit of @xmath143
, we will use a martingale central limit theorem on the _ time reversal _ of the partial sum process @xmath143 . by applying the _ natural extension _ of a dynamical system
, we can assume without loss of generality that the map @xmath108 is invertible .
now , fix a time window @xmath267 $ ] on which we will identify the limit of @xmath143 . by stationarity , we have that @xmath268 by setting @xmath269 , the above equals @xmath270 now , if we define the _ backward time _
partial sum @xmath271 then the above calculation show that @xmath272 under `` sufficient '' mixing conditions on the dynamical system , one can show that @xmath273 is a martingale with respect to the backward time filtration @xmath274 .
thus , using the central limit theorem for ergodic stationary @xmath275 martingale difference sequences @xcite it follows that @xmath276 where @xmath277 . and thus , @xmath278 using the time reversal property of brownian motion . now that we know the limiting behaviour of @xmath143 , we can use the tools from @xcite to identify the limiting behaviour of @xmath144 .
in particular , theorem 2.2 from @xcite allows us to identify the limit of integrals against the martingale @xmath273 , so all we have to do is rewrite @xmath144 in backward time , so that it becomes an integral driven by @xmath279 ( plus corrections ) .
using this idea , we show that @xmath280 where @xmath281 where the integral is of stratonovich type .
all that remains is to prove the discrete tightness estimates .
to do so , we again write the pair @xmath142 in terms of the martingale @xmath273 and stochastic integrals driven by @xmath279 .
since these are both martingales , we can apply the bdg inequality and the tightness estimates follow ( somewhat ) easily , using the ergodic properties and stationarity of @xmath108 .
it follows from theorem [ thmmain ] that @xmath36 where @xmath38 solves the sde @xmath282 it is important to note that although the diffusion approximation is essentially a consequence of the martingale central limit theorem , those martingale sequences only appear in _
backward time_. in particular , the fast slow system can not be written as an equation driven by a semi - martingale , so the theory of good semi - martingales can not be applied directly to the fast slow system .
the advantage of theorem [ thmmain ] is that even though it is not possible to apply martingale limit theory to the recursion , it is quite easy to do so to the _ noise _ processes @xmath142 . in the companion paper @xcite ,
the details are far more complicated than we present above .
for example , the backward time object @xmath273 is not in fact a good martingale , but rather a reasonable perturbation of a good martingale , as described in example [ eggood ] .
this makes matters more interesting
. we will now briefly comment on the connection between theorem [ thmmain ] and the tools introduced in @xcite . given a partition @xmath228 ,
consider the recursion @xmath283 where the increments @xmath284 are defined in such a way that the step - function @xmath285 were a semi - martingale with respect to some given sequence of filtrations , the random walk recursion provides a nice example .
it follows that the path @xmath29 solves the equation @xmath286 where the integral is of it type , as defined section [ srw ] .
suppose moreover that @xmath287 where @xmath74 is a good sequence of semi - martingales and @xmath84 . also define @xmath288_t , \qquad k^n(t ) = \int_0^t a^n(s ) \otimes da^n(s),\ ] ] where the integral is of it type .
suppose moreover that @xmath289 in the sup - norm topology , where the limits are continuous semi - martingales .
then @xcite , theorem 5.1 , states that @xmath193 where @xmath290 and the integrals are of it type .
let us see how this fits into theorem [ thmmain ] .
it is not hard to see that the assumption ( [ efours ] ) implies the assumption @xmath280 .
for instance ,
since @xmath291_t \\ & & { } + \int _ 0^t a^n(s-)\otimes da^n(s ) \\ & = & \int_0^t m^n(s- ) \otimes dm^n(s ) + a^n(t ) m^n(t ) - h^n(t ) + k^n(t).\end{aligned}\ ] ] combining the fact that @xmath74 is good with ( [ efours ] ) we see that @xmath280 where @xmath292_t -h(t ) + k(t).\end{aligned}\ ] ] taking the tightness estimates for granted , we see that in the case where @xmath293_t -h(t ) + k(t ) = \nu t$ ] , theorem [ thmmain ] reproduces the diffusion approximation ( [ ekplimit ] ) .
it is quite possible to extend the theorem [ thmmain ] so that it only requires , for instance , @xmath294 -h + k$ ] to be of bounded variation , which would yield a result closer to that of @xcite , but we do not pursue this here .
several recent articles have used rough path ideas to study numerical schemes for stochastic equations . to name a few ,
@xcite are all concerned with similar but typically higher order schemes than ( [ erecursion ] ) . in @xcite
, the authors also use the idea that a recursion can be approximated by an rde , but only for much higher order milstein - type schemes . in a recent preprint @xcite
, the authors consider euler - type schemes , again by approximating the recursion with a genuine rde .
the recursion considered in this article handles most numerical schemes for sdes , provided the driving noise is a random path with hlder exponent @xmath295 .
it is easy to see that the euler scheme @xmath296 fits into the framework of ( [ erecursion ] ) . using nothing more than a taylor expansion
, one can also show that another typical numerical scheme , the semi - implicit scheme , fits into ( [ erecursion ] ) .
this is defined by @xmath297 where @xmath298 $ ] . when @xmath299 this is of course the ( forward ) euler scheme , when @xmath300 this is the stratonovich mid - point scheme and when @xmath301 this is the backward euler scheme . in @xcite
, one can find a plethora of schemes that also fit into the class of recursions defined by ( [ erecursion ] ) , using similar arguments to that given below . in the context of numerical schemes , we see two key areas where the ability to identify weak limits is beneficial . well - posedness of numerical schemes . when the noise is not a semi - martingale , it may not be clear whether a limit exists and if it does how it should be interpreted .
theorem [ thmmeta ] provides a quick criterion for this situation .
in particular , since @xmath302 one need only identify the limit of @xmath144 . if the limit of @xmath144 corresponds to a reasonable type of integral ( it should correspond to the method of integration used by the numerical scheme ) then the limiting equation can be interpreted in the sense of that integral . numerical schemes that depend on an approximation of the noise , rather than the exact distribution
such situations arise if the noise is difficult to simulate and must instead be approximated , a common scenario when gaussianity is not present .
one also encounters this situation in the context of _ stochastic climate modeling _ , where ocean atmosphere equations are driven by an under - resolved source of noise with persistent correlations in time @xcite and also in _ data assimilation _ , where a perturbation of a stochastic observation is fed into the numerical simulation of a forecast model .
the article @xcite contains a brief overview of the latter idea . finally , the approximation theory above clearly has applications to determining the pathwise order of numerical schemes .
for example , suppose that @xmath47 is defined by the euler scheme @xmath303 since @xmath19 does not depend on @xmath16 the weak limit @xmath38 is determined by the weak limit @xmath162 of @xmath304 using theorem [ thmmod ] as well as the tools from rough path theory ( lemma [ thmitomap ] ) it is easy to show that @xmath305 where @xmath306 only depends on @xmath16 through the discrete hlder norm of @xmath144 . if @xmath19 were brownian motion , then one can trivially calculate moments of @xmath307 exactly , thus obtaining a rate of convergence is simple .
however , obtaining the _ optimal _ rate of convergence is slightly more subtle .
the topic of convergence rates will not be discussed further in this article but is the subject of a future article .
in this section , we will serve an appetizer in rough path theory .
for the full course , we recommend @xcite , which is closely aligned with the exposition below .
a rough path has two components to its definition , an algebraic one and an analytic one .
the algebraic component ensures that the objects @xmath308 do indeed behave like the increments they hope to imitate .
the analytic component describes the hlder condition that is required to construct solution maps . in the definition below , we _
always _ require that the exponent @xmath295 .
we use the notation @xmath309 for the step-2 tensor product algebra .
we say that @xmath310\times[0,t ] \to t^2(\mathbb{r}^d)$ ] is a _ rough path _
if for @xmath311 @xmath312 \label{erp2 } \\[-8pt ] \nonumber \mathbb{x}(s , t ) & = & \mathbb{x}(s , u ) + \mathbb{x}(u , t ) + x(s , u)\otimes x(u , t)\end{aligned}\ ] ] for all @xmath313 $ ] .
these are known as _
chen s relations_. if moreover we have that @xmath314 for all @xmath315 $ ] then @xmath316 is a _
@xmath152-hlder rough path_. the set of @xmath152-hlder rough paths will be denoted @xmath317 ; \mathbb{r}^d)$ ] .
every rough path defines a path @xmath310\to t^2(\mathbb{r}^d)$ ] by setting @xmath318 .
likewise , we could equally have defined rough paths as paths @xmath310 \to t^2(\mathbb{r}^d)$ ] and then simply taken chen s relations as a definition for @xmath319 .
this identification between paths and increments will be used frequently throughout the article .
we will make use of two metric spaces of rough paths .
first , @xmath317 ; \mathbb{r}^d)$ ] is a metric space when furnished with the metric @xmath320 } \frac { |x(s , t)-\tilde{x}(s , t)|}{|s - t|^\gamma } + \sup_{s\neq t \in[0,t ] } \frac { |\mathbb{x}(s , t)-\tilde{\mathbb{x}}(s , t)|}{|s - t|^{2\gamma}}.\ ] ] it is easy to check the interpolation inequality @xmath321 for any @xmath322 .
we also make use of the related @xmath152-hlder `` norm '' @xmath323 } \frac { |x(s , t)|}{|s - t|^\gamma } + \sup_{s\neq t \in[0,t ] } \frac{|\mathbb{x } ( s , t)|^{1/2}}{|s - t|^\gamma},\ ] ] which is by definition finite on @xmath317 ; \mathbb { r}^d)$ ] .
clearly , we have that @xmath324 for all @xmath315 $ ] and @xmath325 ; \mathbb{r}^d)$ ] .
the second metric space we make use of is the set of continuous rough paths @xmath310\to t^2(\mathbb{r}^d)$ ] endowed with the uniform metric @xmath217 . by this , we simply mean the sup - norm defined on functions with range @xmath326 ( with the ordinary euclidean norm on the range ) .
it is easy to see that this topology is equivalent to that generated by @xmath327 .
there is a good reason for using @xmath328 in addition to @xmath151 .
this is due to the relationship between the euclidean norm on and the carnot
caratheodory norm , defined on homogeneous groups .
this will be utilised in section [ sproperties ] . for @xmath329 ;
\mathbb{r}^d)$ ] and @xmath330 , there is a class of paths @xmath331\to\mathbb{r}^e$ ] known as _ controlled rough paths _ , for which one can define the integral @xmath332 .
we say that @xmath38 is an @xmath19-controlled rough path if @xmath333 has the form @xmath334 for all @xmath335 and @xmath336 , where @xmath337\to\mathbb{r}^{e\times d}$ ] is a @xmath152-hlder path . for a thorough treatment of controlled rough paths and their use in defining the above integrals ,
see @xcite , section 4 .
the integral is defined as a compensated riemann sum @xmath338 where @xmath339 \in\mathcal{p } } v^i \bigl(y(t_k ) \bigr)x(t_k , t_{k+1 } ) + \sum_{j=1}^e \bigl ( y'_j ( t_k ) \otimes\partial_j v^i \bigl(y(t_k ) \bigr ) \bigr ) \dvtx \mathbb { x}(t_k , t_{k+1})\ ] ] and @xmath340 denotes a partition of @xmath243 $ ] .
note that the integral is defined pathwise , for each @xmath329 ; \mathbb{r}^d)$ ] .
a controlled rough path @xmath38 is said to solve the rde @xmath341 with initial condition @xmath342 if it solves the integral equation @xmath343 for all @xmath344 $ ] . in this case
, we write @xmath345 .
when required , we write @xmath346 to denote the solution to @xmath347 with @xmath348 . for a thorough treatment of rdes , see @xcite , section 8 .
we will now state a few basis results concerning rdes , proofs can be found in @xcite . if @xmath349 and @xmath329 ; \mathbb{r}^d)$ ] then for each initial condition @xmath350 and any @xmath351 , there exists a unique global solution @xmath352 .
moreover , @xmath353 for all @xmath315 $ ] , where @xmath354 . to see that the remainder scales in this particular way , see the proof of @xcite , lemma 3.4 .
[ lemic ] if @xmath355 and @xmath356 , then @xmath357 for any @xmath358 , where the implied constant depends only on @xmath359 .
[ thmitomap ] suppose that @xmath360 and @xmath361 ; \mathbb{r}^d)$ ] satisfying @xmath362 .
then , on any time window @xmath15 $ ] , the map @xmath363 satisfies the following local lipschitz estimate @xmath364 where @xmath365 depends only on @xmath77 and @xmath108 .
the next lemma is a slight modification of a result in @xcite , hence we only sketch the proof .
let @xmath366 .
then the ball @xmath367 ; \mathbb { r}^d \bigr)\dvtx \vert \!\vert \!\vert\mathbf{x } \vert
\!\vert \!\vert _ { \alpha } \leq r \bigr\}\ ] ] is compact in the space @xmath317 ; \mathbb{r}^d)$ ] .
the proof is a standard modification of a similar statement found in @xcite .
fix a sequence @xmath368 .
use arzela ascoli to find a subsequence that converges in the sup - norm topology .
use the interpolation ( [ einterpolation ] ) between @xmath369 to show that this subsequence also converges in @xmath370 . since @xmath370 is a metric space ,
sequential compactness implies compactness .
the final result , which is a direct corollary of @xcite , theorem 17.3 , allows us to translate rde solutions to stratonovich sdes . [ lemrpsde ] suppose that @xmath360 , @xmath325 ; \mathbb{r}^d)$ ] and let @xmath371 .
suppose that @xmath19 is a continuous semi - martingale and that @xmath162 can be written @xmath372 for @xmath373 . where the integral on the right - hand side is defined in the stratonovich sense and where @xmath374
. then @xmath38 satisfies the sde @xmath375 in all of the results of this section , if in addition to @xmath295 , it is also known that @xmath152 can be taken arbitrarily close to @xmath376 ( as for brownian rough paths ) , then the condition @xmath377 can be relaxed to @xmath378 .
for instance , the versions of the results in any of @xcite will adhere to this .
in this section , we introduce _ rough path recursions _ , driven by _ rough step functions_. before proceeding with the definitions , we must introduce some assumptions and terminology . fix an interval @xmath15 $ ] and let @xmath379 be a partition of @xmath15 $ ] , that is @xmath380 .
we also introduce the _ mesh size _ @xmath381 .
for all the partitions considered in this article , we will assume that @xmath382 the first assumption is obviously natural , the second condition is effectively saying that the largest bin the partition does not shrink too much slower than the smallest bin in the partition .
given some @xmath383 $ ] , we will also use the notation @xmath384 to denote the largest mesh point @xmath385 with @xmath386 .
it follows from ( [ eassp ] ) that @xmath387 as @xmath85 .
we will now define rough step functions and rough path recursions rigorously . fix a partition @xmath27 of @xmath15 $ ] and suppose @xmath388 and @xmath389 for all @xmath390 . the _ rough step function _ above the increments @xmath391 is a path @xmath392\to t^2(\mathbb { r}^d)$ ] defined by @xmath393 where @xmath394 .
we similarly define the incremental paths @xmath395 where @xmath394 and @xmath396
. we will often employ the shorthand @xmath397 and @xmath398 .
we define the discrete @xmath399-hlder `` norm '' @xmath400 by @xmath401 in particular , we see that @xmath402 for all mesh points @xmath190 .
since it is a maximum over a finite set , the discrete hlder norm is finite for every fixed @xmath16 .
it will only play a role in an asymptotic sense .
[ drpr ] fix a sequence of partitions @xmath403 .
a _ rough path recursion _ @xmath404 with @xmath405\to\mathbb{r}^e$ ] is defined by @xmath25 , where @xmath406 and satisfying the recursion @xmath407 for all @xmath408 with @xmath409 and arbitrary @xmath410 and @xmath389 .
the remainder is assumed to satisfy the estimate @xmath411 where the implied constant is uniform in @xmath16 and @xmath412 is the rough step function over the increments @xmath391 .
we will use the notation @xmath413 .
the estimate ( [ erprremainder ] ) is picked as it seems to be the most naturally occurring upper bound in applications .
however , at the expense of a few extra constraints we could equally use @xmath414 for any increasing function @xmath415 and @xmath177 .
[ rmkinterpolation ] although we require that @xmath47 be constant in between mesh points , it is easy to see that all the properties of rough path recursions discussed in the sequel are still true if we assume that @xmath47 is defined by a reasonable interpolation between mesh points .
for example , even though the solution to an rde satisfies the recursion ( [ erps ] ) , it is not a rough path recursion .
however , it can be well approximated by a rough path recursion , and any convergence results for rough path recursion easily imply convergence results for the associated rde solution . as a more general course of action
, we could have defined a rough path recursion to any path satisfying ( [ erps ] ) as well as @xmath416 for some sequence of constants @xmath417 and @xmath418 .
it is clear that , assuming the right conditions on @xmath417 and @xmath97 , all the statements made in the sequel regarding rough path recursions are unaltered if we were to adopt this more general definition .
in this section , we discuss some useful properties of rough step functions and their associated rough path recursions .
the main result , lemma [ thmytilde ] , states that every rough path recursion can be approximated arbitrarily well by the solution to a rough differential equation
. at the heart of this result is the fact that for every rough step function @xmath412 , one can find a genuine rough path @xmath419 that agrees with @xmath412 on @xmath27 .
this is the content of lemma [ thmbxtilde ] . before stating the theorems
, we must introduce some terminology associated with _
geometric _ rough paths . for a more detailed exposition of this material , see @xcite and also @xcite , section 2.2 .
we first define @xmath420 , the step-2 nilpotent lie group , by @xmath421 where @xmath422 is the step-2 lie algebra over @xmath223 and where @xmath61 is the tensor exponential .
in particular , @xmath423 if and only if @xmath424 \bigr ) = a_\alpha e_\alpha+ \bigl(\tfrac{1}{2}a_\beta a_\kappa+ b_{\beta,\kappa } - b_{\kappa,\beta } \bigr)e_\beta\otimes e_\kappa,\ ] ] where @xmath425 denotes the canonical basis of @xmath223 and we employ the einstein summation convention .
it is easy to see that every element @xmath426 can be decomposed into @xmath427 where @xmath428 and @xmath429 .
the pair @xmath430 forms a group .
this group has a _ homogeneous metric _ known as the carnot
caratheodory metric , defined using geodesic paths . to make this precise ,
we first define the _ signature _ of a smooth path .
let @xmath431;\mathbb{r}^d)$ ] be the space of paths @xmath432\to\mathbb{r}^d$ ] with bounded variation . for @xmath433;\mathbb{r}^d)$ ] , the signature is defined by @xmath434 where the integral is constructed in the riemann stieltjes sense .
the carnot
caratheodory ( cc ) norm is defined by @xmath435 ; \mathbb{r}^d \bigr ) \mbox { and } g = \mathbf{s}(\gamma ) ( 0,1 ) \biggr\}.\ ] ] the following result ( which is a refinement of chow s theorem ) shows that the norm is well defined ( see @xcite , theorem 7.32 , for a simple proof ) .
[ lemgeodesic ] if @xmath428 then there exists a path @xmath436 \to \mathbb{r}^d$ ] with @xmath437 such that @xmath438 and @xmath439 .
the cc norm can also be `` compared '' with the usual euclidean norms in the following way .
suppose that @xmath428 can be decomposed into @xmath440 where @xmath441 and @xmath442 , then we have the comparison @xmath443 let @xmath444 be a rough step function on a partition @xmath445 and let @xmath446 $ ] .
for each @xmath16 , there exists @xmath447 ; \mathbb { r}^d)$ ] with @xmath448 moreover , we have that @xmath449 where the implied constant is uniform in @xmath16
. we will start by constructing @xmath450 for fixed @xmath451 $ ] with @xmath452 .
first , note that since @xmath453 , we have the decomposition @xmath454 where @xmath455 and @xmath456 is defined by @xmath457 we will now define @xmath458 first , define @xmath459 by a simple linear interpolation @xmath460 now , to define @xmath461 , we know from lemma [ lemgeodesic ] that there exists a path @xmath462 \to\mathbb { r}^d$ ] such that @xmath437 and @xmath463 and @xmath464 .
we set @xmath465
. we will now define @xmath450 for arbitrary @xmath466 $ ] with @xmath452 .
suppose without loss of generality that @xmath467 where @xmath468 and @xmath469
. then we define @xmath470 using chen s relations @xmath471 and @xmath472 we will now check that @xmath419 satisfies the requirements of the theorem .
first , we will show that chen s relations hold .
it is easy to see that chen s relations hold when restricted to the interval @xmath473 $ ] . indeed , since @xmath461 is a signature and since @xmath459 is an increment , both objects individually obey chen s relation .
thus , if we write @xmath440 , with @xmath441 and @xmath474 then @xmath475 therefore , @xmath476 and @xmath477 as required .
now , for arbitrary @xmath478 it follows immediately from the construction that @xmath479 satisfies chen s relations . using the shorthand @xmath480
, we will now prove that @xmath481 .
first , suppose that @xmath482 $ ] with @xmath452 . then using the comparison ( [ ecceq ] ) and the construction of @xmath483
, we have that @xmath484 & \lesssim & \bigl{\vert}g(s , t ) \bigr{\vert}_{\mathrm{cc } } = \frac{(t - s)}{(\tau ^n_{j+1}-\tau^n_j ) } \bigl{\vert}g\bigl(\tau^n_j,\tau^n_{j+1 } \bigr ) \bigr{\vert}_{\mathrm{cc}}\end{aligned}\ ] ] and again by ( [ ecceq ] ) we have that @xmath485 \label{efeb9a } & \lesssim & \bigl|x^n_{j , j+1}\bigr| + \bigl(\bigl|\mathbb{x}^n_{j , j+1}\bigr| + \bigl|x^n_{j , j+1}\bigr|^2 \bigr)^{1/2 } \\[-2pt ] & \lesssim & c_n \bigl(\tau^n_{j+1 } - \tau^n_j \bigr)^\gamma.\nonumber\end{aligned}\ ] ] it follows that @xmath486 where in the last inequality we use the fact that @xmath487 . by a similar argument , we can show that @xmath488 and hence @xmath489 now suppose @xmath315 $ ] with @xmath490 as above .
by chen s relations , we have that @xmath491 and @xmath492 this completes the proof .
we now have all the tools we need to prove the main result of this section , namely that rough path recursions can be well approximated by the solution to an rde .
[ thmytilde ] let @xmath413 .
let @xmath419 be any rough path satisfying the conditions of lemma [ thmbxtilde ] and let @xmath493 .
then @xmath494 for any @xmath446 $ ] , where the implied constant is uniform in @xmath16 .
we will again use the shorthand @xmath495 . for any @xmath344 $ ] , we have that @xmath496 where @xmath394 , and hence @xmath497 .
it follows from ( [ erdesoln ] ) that @xmath498 where @xmath499 . since @xmath500
, it follows that @xmath501 to estimate @xmath502 we need some new terminology . for each @xmath503 , define @xmath504 by @xmath505 that is , @xmath506 , where @xmath507 is the unique solution to the rde driven by @xmath419 with vector field @xmath508 , initialised at time @xmath183 with initial condition @xmath509 [ recalling that @xmath510 , as in definition [ drpr ] ] . in particular , we have that @xmath511 , @xmath512 and @xmath513 for any @xmath514 .
it follows that @xmath515 but since @xmath516 and @xmath517 , it follows from lemma [ lemic ] that @xmath518 by ( [ ezstep ] ) , we have @xmath519 where in the last line we have used the fact that @xmath520 agrees with @xmath412 on @xmath27 , as well as the recursive definition of the rough path scheme @xmath47 .
hence , we have that @xmath521 it follows from ( [ ez1 ] ) and ( [ ez2 ] ) that @xmath522 where in the last inequality we have used the assumption @xmath523 .
this completes the proof .
proof of theorem [ thmmod ] all that is required is to show that @xmath524 solves ( [ exz ] ) where @xmath525 ; \mathbb{r}^d)$ ] is derived in lemma [ thmbxtilde ] . by definition and by construction of @xmath419
we have that @xmath526 where @xmath527 is a piecewise smooth path ( obtained from the signature realizing @xmath483 ) and @xmath528 where the integral is of riemann stieltjes type and where @xmath529 is constructed by concatenating the increments @xmath459 , in particular @xmath529 is piecewise lipschitz .
by @xcite , theorem 12.14 , it follows that @xmath214 satisfies ( [ exz ] ) .
note that @xcite , theorem 12.14 , is basically lemma [ lemrpsde ] but under the assumption that the driving path is piecewise smooth rather than a semi - martingale . in section [ sconvergence ]
, we will employ the standard method of lifting weak convergence in the sup - norm topology to weak convergence in some @xmath152-hlder topology , using a tightness condition . in the _ continuous time _
setting ( which we can not use ) , the kolmogorov
lamperti criterion @xcite is the usual method for checking this tightness condition .
the following is a slight modification of a version of the criterion found in corollary a12 @xcite .
[ thmctskolm ] let @xmath530 define a sequence of rough paths .
suppose that @xmath531 .
then @xmath532 for any @xmath533 .
in particular , we have that @xmath534 and moreover @xmath535 is tight in the @xmath536 topology for every @xmath533 . in the case of geometric rough paths [ where @xmath412 takes valued in @xmath420 ] ,
the result is simply corollary a12 of @xcite . to extend the result to general rough paths ,
one simply applies the garcia
rumsey interpolation result to the components @xmath19 and @xmath162 individually .
this argument can be found in @xcite , corollary 4 .
obviously , this result can not be used directly on rough step functions , since step functions have no hope of satisfying the kolmogorov estimates .
fortunately , a discrete version of the above result turns out to be equally as useful .
we define the _ discrete tightness condition _ as @xmath538 this essentially says that the rough step functions are `` hlder continuous , '' provided we do not look at them too closely ( i.e. , near the jumps ) .
we will now show that the discrete tightness criterion can likewise be checked using a discrete version of the continuous kolmogorov criterion .
in particular , we need only check the estimate on the partition @xmath27 .
[ thmkolm ] suppose that @xmath539 for each @xmath540 uniformly in @xmath541 , for some @xmath542 $ ] .
then the discrete tightness condition ( [ edtight ] ) holds for any @xmath543 .
the idea behind the proof is to replace @xmath412 with the genuine rough path @xmath419 constructed in lemma [ thmbxtilde ] , which , as you recall , agrees with @xmath412 on @xmath27 . since @xmath544 to prove the discrete tightness condition ( [ edtight ] ) it is sufficient to check the hlder estimate ( [ ectskolm ] ) for the process @xmath419 and apply lemma [ thmctskolm ] .
hence , we need only verify that @xmath545 holds for each @xmath315 $ ] , uniformly in @xmath546 .
assume without loss of generality that @xmath315 $ ] and @xmath547 and @xmath548 ( note that the case @xmath549 $ ] is essentially a sub - argument of the arguments below ) . from chen s relations , we know that @xmath550 but from ( [ efeb9a ] ) , we see that @xmath551 by assumption , we have that @xmath552 the remaining term in ( [ efeb9 ] ) can be bounded similarly . by chen s relations ( and hlder s inequality ) , we also have that @xmath553 but from ( [ efeb10a ] ) we have that @xmath554 as above , it follows that @xmath555 the other terms in ( [ efeb10 ] ) can be bounded similarly .
this completes the proof .
the discrete criterion differs from the continuous case in the assumption @xmath556 , which was not required in the continuous case .
however , this assumption only becomes a restriction when the diffusion approximation is driven by a path with hlder exponent @xmath557 .
of course , one can always resolve the problem by treating the path as having the weaker hlder exponent . on the other hand , in these higher regularity situations
the iterated integrals become unnecessary and a much simpler theory of young integration ( with much weaker assumptions ) would suffice .
we can now prove the main result of the article .
[ thmweak ] suppose that @xmath558 and that @xmath412 satisfies the discrete tightness condition ( [ edtight ] ) for some @xmath446 $ ] .
then @xmath559 in the sup - norm topology .
first , let @xmath419 be the @xmath152-hlder rough path constructed in lemma [ thmbxtilde ] and let @xmath560 . to prove the theorem , it is sufficient to first show that @xmath561 in probability and second show that @xmath562 in the sup - norm topology , hence we will proceed as such . as usual
, we use the shorthand @xmath563 . from lemma
[ thmytilde ] , it follows that @xmath564 hence , @xmath565 but since @xmath566 as @xmath85 , we see that for any arbitrarily large @xmath567 @xmath568 finally , by taking @xmath569 , it follows from the discrete tightness condition that @xmath570 now we prove that @xmath562 in the sup - norm topology . due to the continuity of the map
@xmath571 , as stated in lemma [ thmitomap ] , it is sufficient to prove that @xmath572 in the @xmath151 topology .
it is therefore sufficient to first show that @xmath573 and second that @xmath574 is tight in the @xmath151 topology .
first , due to the regularity of @xmath419 between mesh points , it is easy to see that @xmath575 for some @xmath576 .
hence , by an argument similar to that found at the start of the proof , it follows from the discrete tightness condition that @xmath577 in probability . and
since by assumption @xmath578 it follows that @xmath579 .
we will now move onto the tightness argument . from lemma [ thmbxtilde ]
, we have the estimate @xmath580 .
it follows that @xmath581 and the tightness of @xmath419 in the @xmath151 topology follows from the discrete tightness condition .
this completes the proof .
we can now prove the theorems introduced in section [ sresults ] .
they are both immediate corollaries .
proof of theorem [ thmmain ] @xmath582by the discrete kolmogorov criterion ( lemma [ thmkolm ] ) , we obtain the discrete tightness criterion , and hence can apply theorem [ thmweak ] . to identify the limit @xmath345 , we simply apply lemma [ lemrpsde ] .
[ rmkqlarge][rmkdrift2 ] to prove the result with the relaxed assumption described in remark [ rmkqsmall ] , one simply replaces lemma [ thmitomap ] with the sharper version @xcite , theorem 12.10 , and the remaining argument is identical . to prove the result with additional drift , as described in remark [ rmkdrift1 ]
, we again replace lemma [ thmitomap ] with @xcite , theorem 12.10 , but now we must use the @xmath201-rough path @xmath583 with @xmath584 and arbitrarily small @xmath585 . the remaining argument
is identical .
proof of meta theorem [ thmmeta ] the proof is completely identical to the proof above , but we still need to `` interpret '' the limit @xmath345 .
this is a fairly nonrigorous statement and , therefore , has a fairly nonrigorous proof .
we are merely sketching an idea that would apply more rigorously in concrete situations . by definition ,
the limit @xmath38 solves the rde @xmath586 it is a general _ heuristic _ that if @xmath162 is constructed using some known construction then the integral in ( [ erdefinal ] ) is constructed similarly . for instance , suppose there is some `` method of integration , '' which is a bilinear operator @xmath587 for two continuous paths @xmath588;\mathbb{r})$ ] satisfying the obvious condition @xmath589 now suppose that @xmath162 is defined by @xmath590 for each @xmath591 .
then , using the theory of controlled rough paths @xcite , it can be shown that @xmath38 solves ( [ erdefinal ] ) if and only if @xmath38 is a fixed point of the equation @xmath208 the assumptions on @xmath592 are generic enough to include virtually any reasonable construction of an integration map ( for integrators with h " older exponent @xmath295 ) .
david kelly would like to thank i. melbourne for introducing the motivating problem and p. friz for several constructive comments . | in this article , we consider diffusion approximations for a general class of stochastic recursions .
such recursions arise as models for population growth , genetics , financial securities , multiplicative time series , numerical schemes and mcmc algorithms .
we make no particular probabilistic assumptions on the type of noise appearing in these recursions .
thus , our technique is well suited to recursions where the noise sequence is not a semi - martingale , even though the limiting noise may be .
our main theorem assumes a weak limit theorem on the noise process appearing in the random recursions and lifts it to diffusion approximation for the recursion itself . to achieve this ,
we approximate the recursion ( pathwise ) by the solution to a stochastic equation driven by piecewise smooth paths ; this can be thought of as a pathwise version of backward error analysis for sdes .
we then identify the limit of this stochastic equation , and hence the original recursion , using tools from rough path theory .
we provide several examples of diffusion approximations , both new and old , to illustrate this technique . ./style / arxiv - general.cfg |
we kindly acknowledge to a. cabello , m. grassl , p. horodecki , r. horodecki , s. weigert , and k. yczkowski for fruitful discussions .
this work was supported by grants conicyt pfb-0824 , msi rc130001 , fondecyt grants n@xmath138 1140636 , n@xmath1381120067 , and n@xmath138 3120066 .
g. c. and e. s. g. acknowledge financial support of conicyt .
z. hradil , j. rehacek , j. fiurasek , and m. jezek , _ maximum likelihood methods in quantum mechanics _ , in quantum state estimation , lecture notes in physics , m. g. a. paris and j. rehacek ( eds . ) , pp .
59 - 112 ( springer , 2004 ) .
we generated a pure state according to the haar measure and assumed the noise introduced by eq .
( [ noise - generation ] ) .
thereafter , the five probability distributions of this state were calculated . to these probabilities
we added the poisson noise characteristic of the photo - detection process and applied our method to generate an estimate @xmath139 of the pure state @xmath55 .
this estimate is obtained by normalising the vector provided by eq .
( [ solution ] ) . finally , the fidelity @xmath140 between these two states was calculated and averaged over all the generated states .
we considered poisson noise and @xmath141 .
this latter value is characteristic of our experimental setup . | a long standing problem in quantum mechanics is the minimum number of observables required for the characterisation of unknown pure quantum states .
the solution to this problem is specially important for the developing field of high - dimensional quantum information processing . in this work
we demonstrate that any pure @xmath0-dimensional state is unambiguously reconstructed by measuring 5 observables , that is , via projective measurements onto the states of 5 orthonormal bases .
thus , in our method the total number of different measurement outcomes ( @xmath1 ) scales linearly with @xmath0 .
the state reconstruction is robust against experimental errors and requires simple post - processing , regardless of @xmath0 .
we experimentally demonstrate the feasibility of our scheme through the reconstruction of 8-dimensional quantum states , encoded in the momentum of single photons . with the development of high - dimensional quantum information processing techniques @xcite , the total dimension @xmath0 of quantum systems employed in experiments increases at a fast pace .
since the total number @xmath2 of measurement outcomes required by conventional quantum tomography methods scales with @xmath3 @xcite , it is of paramount importance the search for tomographic protocols specially adapted to higher dimensions .
that is , schemes which require a lower @xmath2 and a reduced complexity of the post - processing methods .
thus , it is possible to consider _ a priori _ information about the set of states to be characterised .
for example , rank-@xmath4 quantum states are reconstructed , with high probability , with @xmath2 of the order of @xmath5 via compressed sensing techniques @xcite .
nearly matrix product states are determined with @xmath2 linear in @xmath0 and post - processing that is polynomial in the system size @xcite .
permutationally invariant quantum states of @xmath6 qubits ( @xmath7 ) are reconstructed with @xmath8 @xcite . in this work
we study the characterisation of unknown pure quantum states via projective measurements . in 1933 , w. pauli @xcite considered the unambiguous characterisation of pure states from probability distributions generated by the measurement of a fixed set of observables .
it has been shown that the number of observables must be larger than @xmath9 for @xmath10 @xcite . here
, we show that almost all pure quantum states can be characterised by a set of probability distributions generated by 4 observables , that is , projective measurements onto a fixed set of @xmath11 bases @xmath12 ( @xmath13 ) independently of the dimension @xmath0 . in our case @xmath14 , which is an improvement compared with quantum tomography based on mutually unbiased bases @xcite , sic - povm @xcite and compressed sensing @xcite , which require @xmath15 , @xmath16 and @xmath17 of the order of @xmath18 , respectively .
pure states can also be reconstructed via the expectation values @xcite of a fixed set of observables . in this case , @xmath19 observables are required at least , where @xmath20 is a quantity that scales with the logarithm of the dimension .
consequently , the value of @xmath2 is larger than in our proposal .
the exception to our first characterisation is a statistically unlikely null measure set @xmath21 of pure states .
such states can be characterised by adding a fifth measurement basis @xmath22 , which detects whether a pure state is in @xmath21 or not .
thus , it is possible to define a new set of @xmath11 bases to characterise the state .
these bases are similar to @xmath12 but defined on a lower dimensional subspace .
thereby , the set of pure states is decomposed into a finite number of disjoint sets of states such that each set is reconstructed with a particular set of five bases .
thus , any pure state can be reconstructed with projective measurements on no more than @xmath23 bases , that is , @xmath24 . an interesting feature of each set of @xmath23 bases is that they allow us to certify the initial assumption on the purity of the state to be reconstructed .
this class of certification can also be achieved by reconstructing pure states via expectation values @xcite but at the expense of a much higher @xmath2 than in our proposal .
we demonstrate the feasibility of our tomographic scheme with the experimental characterisation of eight - dimensional quantum states encoded on the transverse momentum of single - photons .
the preparation of the states is affected by white noise , which renders the states slightly impure , and the detection process by poison noise . in spite of these conditions the obtained fidelities are higher than @xmath25 with @xmath26 .
this result is in agreement with numerical simulations which consider realistic noise levels in the preparation of the unknown state as well as in the detection process .
our work shows that the effective characterisation of unknown high - dimensional pure quantum states by means of a reduced number of measurement outcomes is feasible .
a @xmath0-dimensional quantum state can be expanded as @xmath27 , where @xmath28 is the identity operator , @xmath29 form the generalised bloch vector and @xmath30 are a traceless , hermitian representation of the generators of the group @xmath31 such as the generalised gell - mann basis @xcite .
this consists of @xmath32 diagonal operators @xmath33 and @xmath34 non - diagonal operators @xmath35 and @xmath36 .
the latter are given by @xmath37 and @xmath38 , where @xmath39 .
any pure state @xmath40 satisfies the following set of @xmath41 equations @xmath42 interestingly , @xmath32 of the above equations univocally characterise a certain dense set of pure states .
we choose the set of @xmath32 equations with @xmath43 which allows us to solve eq .
( [ phases ] ) recursively , up to a null measure set of pure states . in order to calculate the traces of eq .
( [ phases ] ) we consider rank - one projective measurements associated to the eigenvectors of @xmath44 and @xmath45 with eigenvalues @xmath46 , that is , @xmath14 .
these @xmath47 vectors can be always sorted in four orthogonal bases . for @xmath48
these bases are given by @xmath49 where addition of labels is carried out modulo @xmath0 and @xmath50 $ ] .
for odd dimensions we considered the integer part of @xmath51 and every basis is completed with @xmath52 . defining @xmath53 ( @xmath54 ) as the probability of projecting the state @xmath55 onto the eigenvector of @xmath44 ( @xmath56 ) with eigenvalue @xmath46 , eq .
( [ phases ] ) becomes @xmath57 , where @xmath58 $ ] .
table [ table ] associates bases @xmath12 with the eigenvectors of @xmath44 and @xmath45 with eigenvalues @xmath46 and probabilities @xmath53 and @xmath54 , respectively .
.association between the four bases , @xmath59 generators and transition probabilities . [ cols="^,^,^ " , ] the equations @xmath57 can be recursively solved leading to @xmath60 where @xmath61 and @xmath62 is determined by normalisation .
if one of the coefficients @xmath63 vanishes then the system of equations can not be recursively solved . however , the remaining equations and the normalisation condition are enough to reconstruct the state .
this also holds in the case of two consecutive vanishing coefficients . for two
non - consecutive vanishing coefficients the system of equations has infinite solutions .
for instance , in @xmath64 the set of equations is @xmath65 , @xmath66 , @xmath67 and @xmath68 . if @xmath69 or @xmath70 then all left sides vanish .
thus , there are states that can not be singled out via @xmath71 .
these states form a manifold @xmath21 of dimension @xmath72 and , consequently , are statistically unlikely .
this means that , if pure states are randomly selected then with probability @xmath73 they would be out of @xmath21 .
thus , a total of @xmath47 measurement outcomes , corresponding to projective measurements onto states of bases @xmath71 , are informationally complete on the set of pure states up to the null measure set @xmath21 . the introduction of a fifth basis @xmath22 , the canonical basis which is first to be measured , allows us to determine whether a state belongs to the manifold @xmath21 or not depending on the number of vanishing coefficients detected .
if the state is in @xmath21 and has @xmath74 null coefficients , then we can reconstruct the state into the subspace associated to the @xmath75 non - vanishing coefficients .
this is done with the bases @xmath71 ( @xmath76 ) but defined for a subspace of dimension @xmath75 .
thereby , any pure state can be reconstructed via projective measurements onto , at most , @xmath23 orthonormal bases or equivalently with @xmath77 . in the @xmath78 qubits case , separable bases @xmath22 , @xmath79 and @xmath80 correspond to the eigenstates of the local operators @xmath81 , @xmath82 and @xmath83 , respectively .
entangled bases @xmath84 and @xmath85 can be mapped onto bases @xmath79 and @xmath80 respectively by applying the quantum fourier transform ( qft)@xcite twice .
this can be optimally implemented with the order of @xmath86 @xcite hadamard and conditional two - qubit phase gates .
thus , our tomographic scheme is reduced to simple local measurements and qft .
the latter has been experimentally realised in photonic qubits @xcite , nuclear magnetic resonance @xcite , neutral molecules @xcite , superconducting qubits @xcite , and trapped ions @xcite .
measurements onto the set of bases allow us to certify the assumed purity of the unknown state .
a state @xmath87 is pure if and only if the equation @xmath88 holds for every @xmath89 .
remarkably , these conditions for @xmath90 are enough to ensure that @xmath87 determines a pure state . indeed , given that @xmath87 is a quantum state then @xmath91 for a given operator @xmath92 .
so , every entry of @xmath87 satisfies @xmath93 , where @xmath94 is the @xmath95th column of @xmath92 . if @xmath90 the @xmath32 equations @xmath96 hold if and only if vectors @xmath94 and @xmath97 are parallel for every @xmath98 .
consequently , @xmath87 is pure .
the same holds for any set of five bases .
tomographic schemes reconstruct quantum states in matrix space . since quantum states form a null measure set in matrix space , noisy measurement results lead to matrices that do not represent quantum states . to overcome this problem experimental data
is post - processed with maximum likelihood estimation @xcite .
an important feature of our method is that it delivers a vector of the underlying hilbert space for any set of noisy probabilities , being the normalisation of this vector the only procedure required to obtain a pure state from noisy probabilities .
the scheme here proposed is based on a priori information about the purity of the state to be reconstructed .
this condition is difficult to realise in current experiments .
however , it is possible to generate nearly pure states such as @xmath99 here , the generation process is affected by white noise , which mixes the target pure state @xmath100 with the maximally mixed state @xmath101 .
the strength of the process is given by the real number @xmath102 .
this model is in agreement with our experimental setup .
the following lower bound for the fidelity holds @xmath103 where @xmath104 , @xmath105 is the probability distribution generated by measurements on the canonical basis @xmath22 and @xmath106 is the maximal amount of noise introduced by the photo - detection process . for a uniform state ( i.e. , @xmath107 for every @xmath108 ) of a composite system of @xmath78 qubits
we obtain @xmath109 .
thus , to keep a constant large lower bound while increasing the number of qubits , @xmath106 must decrease as @xmath110 .
fig .
[ fig1 ] shows the average fidelity @xmath111 onto the complete hilbert space and the square root of its variance @xcite as functions of @xmath78 considering white and poisson noises .
the setup employed to implement and test our proposal is depicted in fig .
[ fig : setup ] .
the state preparation stage has a continuous - wave laser operated at 690 nm and an acousto - optic modulator ( aom ) .
this is used to generate controlled optical pulses .
optical attenuators ( not shown for sake of clarity ) decrease the mean photon number per pulse to the single - photon level .
thereafter , single photons are sent through two transmissive spatial light modulators ( slm1 and slm2 ) , each one formed by two polarisers , quarter wave plates ( qwp ) and a liquid crystal display ( lcd ) .
slm1 and slm2 work in amplitude - only and phase - only modulation configuration , respectively .
slm2 is placed on the image plane of slm1 , and the 8-dimensional quantum system is generated with 8 parallel slits addressed on the slms .
after slm2 the non - normalised state of the transmitted photon is given by @xmath112 @xcite where @xmath113 represents the state of the single photon crossing the @xmath114th - slit .
here , @xmath115 is the transmission for each slit controlled by slm1 ; @xmath116 is the phase of each slit addressed by slm2 , and @xmath78 is a normalisation constant .
the different values of @xmath115 and @xmath116 are configured by the grey level of the pixels in the slms @xcite .
we addressed in the slms slits with the width of 2 pixels , and 1 pixel of separation between them , where each pixel is a square of 32 @xmath117 of side length.since a 3-qubit system is a 8-dimensional quantum system , state @xmath55 can be used to simulate a @xmath9-qubit system @xcite .
we generated the state @xmath118 $ ] with equal real probability amplitudes , the state @xmath119 $ ] analogous ( up to locc @xcite ) to the ghz state @xcite and the w state @xmath120 $ ] @xcite . to guarantee the purity of the spatial qudit states , it is necessary to observe a high visibility in the interference patterns in the far - field plane of the slms @xcite . the value of @xmath102 can be obtained from the relation @xmath121 , where @xmath122 is the observed visibility in the far - field plane . in our experiment
we have @xmath123 leading to @xmath124 .
thus the generated states have a purity of @xmath125 . in order to realise the projections onto the states of the @xmath23 bases @xmath12 we use two additional modulators , slm3 and slm4 , working in amplitude - only and phase - only , respectively , and a point - like avalanche photo - detector ( apd ) @xcite .
the slms in the projective measurement stage ( see fig
. 1 ) are addressed with slits whose amplitudes and phases are defined to implement the projections required by our method . at the detection plane ,
the single - photon detection rate is proportional to the probability of projecting the initial state ( @xmath126 , @xmath127 and @xmath128 ) into the required basis states of eq .
( [ base1 ] ) @xcite . from the experimental data we calculated the probability distributions associated to the 5 bases @xmath12 for the three initial states . with these probability distributions and using eq .
( [ phases ] ) , for the appropriate set of @xmath23 bases , we obtained a set of vectors , which were then normalised to obtain the final reconstructed states .
the fidelity @xmath129 of the initial states with respect to the expected ones are @xmath130 , @xmath131 and @xmath132 .
these were calculated by considering the effect of the poisson noise in the detection process and selecting the highest and smallest fidelity between the expected state and the estimated state for a particular noisy set of distributions . for comparison purposes we consider an experiment with a similar configuration @xcite where two pure states with non vanishing , real coefficients in @xmath133 were reconstructed by means of measurements on mutually unbiased bases achieving @xmath134 and @xmath135 .
note that we achieved higher fidelities with a total of @xmath136 projective measurements instead of @xmath137 , as in the compared case .
the states reconstructed with our method are shown in fig .
[ fig : a ] . fig .
[ fig : a](a ) , [ fig : a](c ) and [ fig : a](e ) exhibits real parts of reconstructed density matrices , compared with the expected ones ( insets ) .
fig .
[ fig : a](b ) , [ fig : a](d ) and [ fig : a](f ) show imaginary parts of the respective matrices , compared with the theoretical predictions ( insets ) .
we have shown that projections onto the eigenstates of a fixed set of @xmath11 rank-@xmath0 observables characterise all pure states of a @xmath0-dimensional quantum system up to a null measure set of dimension @xmath72 .
these @xmath47 measurements compares favourably with the typical @xmath3 scaling of measurements of known generic tomographic methods .
the addition of a fifth observable allows us to detect whether a state belongs to the null measure set or not .
if this is the case , then it is always possible to construct a new set of @xmath11 observables which determines unambiguously the state .
all sets of five observables allow us to certify whether the initial assumption on the purity of the state holds or not .
we experimentally demonstrated the feasibility of our scheme in the characterisation of states in @xmath133 .
we achieved high fidelities by means of a reduced number of measurement outcomes and a simple post - processing method . |
looking more closely at the klln function @xmath17 of three particles with masses @xmath18 , @xmath19 and @xmath20 , the analytic behaviour unfolds a rich spectrum of real and virtual processes .
the function is zero for any one of the thresholds @xmath21 allowing only positive values for the three masses , in the @xmath22 phase space the zeros of the klln function are located on the three planes spanned by each two of the three face diagonals @xmath23 , @xmath24 and @xmath25 . up to the general normalization one can visualise the phase space by cutting the first octant by a plane orthogonal to the space diagonal . on this plane
the first octant will appear as upright equilateral _
domain triangle _
[ vireoct ] ) , the corners representing the @xmath26 axes , and the opposite sides representing the coordinate planes @xmath27 .
the zeros of the klln function are found on the three lines connecting the midpoints of the sides of the triangle ( cf . fig .
[ viretri ] ) . the equilateral triangle enclosed by these three lines
is called the _
klln triangle_. while this triangle is the region where the klln function becomes negative , the other three triangles are regions with positive klln function , each of them containing a coordinate axis .
special values are @xmath28 \lambda(m_1 ^ 2,m_2 ^ 2,m_2 ^
2)&=&m_1 ^ 4 - 4m_1 ^ 2m_2 ^ 2\qquad\mbox{(medians ) } , \nonumber\\[7pt ] \lambda(m_1 ^ 2,m_2 ^ 2,0)&=&(m_1 ^ 2-m_2 ^ 2)^2\qquad\mbox{(sides)},\nonumber\\[7pt ] \lambda(m_1 ^ 2,0,0)&=&m_1 ^ 4 \qquad\mbox{(corners)}.\end{aligned}\ ] ] a negative value of the klln function means that the sum of two of the masses is larger than the third one .
this means physically that none of the three particles can decay into the other two .
the process is a virtual one , leaving at least one of the particles off - shell , i.e. with a momentum squared which is smaller than the squared mass of this particle . on the other hand
, a positive klln function means that exactly one decay channel is real .
taking for instance the region connected to the @xmath18 axis , all points in the lower left triangle in fig .
[ viretri ] represent possible decays through the channel @xmath29 because @xmath18 is larger than the sum @xmath30 .
therefore , one can conclude that the triangle in fig .
[ viretri ] gives structure and direction to processes involving three massive particles .
usually , three - particle interactions appear within a more complicated process . in this case
the masses are ( partially ) replaced by invariant squared momenta @xmath31 . on tree level
such a process is a cascade process as explained for instance in refs .
@xcite . for a cascade process
including @xmath32 final particles with masses @xmath26 ( @xmath33 ) the @xmath32-particle phase space @xmath34 can be factorized .
given for instance the momentum @xmath35 of the first particle emitted from this process , the remaining momentum @xmath36 is mediated by a virtual particle .
one can define an invariant mass @xmath37 with @xmath38 with the only condition that @xmath39 is time - like . separating the first phase space integration from the other ones and inserting the trivial identities @xmath40 one obtains @xcite @xmath41 as an example we consider the cascade decay @xmath42 .
the first part @xmath43 of the cascade phase space is given by the phase space @xmath44 of the previous ( simpler ) process @xmath45 where @xmath46 is replaced by @xmath47 .
for the second part @xmath48 one obtains in a similar manner @xmath49 assuming that one can perform the solid angle integrations for the @xmath3 boson ( @xmath50 ) and for the @xmath51 quark ( @xmath52 ) trivially , one is left with the three - particle phase space @xmath53 the example just introduced is an ideal playground for the implications caused by the klln functions in case of integrations over inner lines . even though the intermediate @xmath3 boson can be off - shell , the phase space has to be real .
therefore , the two conditions @xmath54 \lambda(p_w^2,m_c^2,m_s^2)&=&(p_w^2-(m_c - m_s)^2)(p_w^2-(m_c+m_s)^2)\ \ge\ 0\end{aligned}\ ] ] determine the phase space region for @xmath55 and at the same time an appropriate substitution to calculate the integral analytically . taking into account the known quark mass hierarchy
, the two conditions result in @xmath56 neglecting the masses of @xmath51 and @xmath57 quarks , the phase space simplifies to @xmath58 the square root in eq .
( [ intpw2 ] ) can be simplified by choosing one of the factors in eq .
( [ kallentbw ] ) as new variable , e.g.@xmath59 .
the second factor is then given by @xmath60 and the upper limit is given by @xmath61 .
however , @xmath62 is not the optimal choice . using instead
@xmath63 the square root simplifies to @xmath64 the integration can be performed by using @xmath65 , @xmath66 where @xmath67 finally , one can use @xmath68 to obtain @xmath69 to conclude this section , let us dwell on the parameter @xmath70 used in the substitution .
this parameter is related to the masses and the invariant momentum square by @xmath71 comparing with eqs .
( [ ebw ] ) and ( [ pbw ] ) one realizes that @xmath70 is the rapidity for the transition between the @xmath72 and @xmath73 quark rest frames , the exponential representation being @xmath74
dilogarithms appear after integrations like the one in eq .
( [ intpw2 ] ) are performed .
the `` classical '' dilogarithm is given by the integral representation @xmath75 the classical dilogarithm is a special case of a polylogarithm and is an analytical function in the whole complex plane except for the interval @xmath76 $ ] along the positive real axis where the branch cut is located .
starting from the branch point at @xmath77 the branch cut separates two different riemannian sheets of the multi - valued function .
there are two one - parameter and a couple of two- and three - parameter identities that allow to relate dilogarithms with different arguments @xcite .
the two one - parameter identities are given by @xcite @xmath78 an additional useful identity is @xmath79 [ cols=">,^,^,^ , < " , ] for the arguments of the dilogarithms the two identities ( [ dilog1 ] ) and ( [ dilog2 ] ) are involutions .
because of this , one can create a closed chain of transitions for the argument which constitutes a hexagon ( cf.fig .
[ hexagon ] ) . for real - valued arguments
this transition orbit helps to constrain the argument to a value lower or equal to @xmath80 , gaining a real value for the dilogarithm .
a pragmatic argument can be very helpful to choose an appropriate hexagon orbit transition : processes including only real particles have to lead to real phase space integrals .
therefore , a negative argument for the logarithms obtained together with the dilogarithm in calculations like @xmath81 indicate that the argument of the dilogarithm still is not in an appropriate form . a similar orbit , though for complex arguments , is given in refs .
@xcite for the bloch wigner dilogarithm @xmath82 which is mainly ( i.e. up to a double - logarithmic correction ) the imaginary part of the classical dilogarithm .
the identities , induced by the two involutions , are given by @xmath83 and @xmath84 . in ref .
@xcite these identities are applied to the ( complex ) root @xmath85 where the discriminant @xmath17 is again the klln function , assumed to be negative .
an orbit with constant value for @xmath86 given by @xmath87 is obtained by combining the two involutions with a third one , namely the calculation of the complex conjugate . as mentioned in ref .
@xcite , the points of the orbit can be obtained from the starting point @xmath88 by interchanging the squared masses @xmath89 , @xmath90 and @xmath91 .
this invariance describes a symmetry of the bloch wigner dilogarithm which is manifest already for the klln function . in the following we compare these two functions . + because the bloch wigner dilogarithm vanishes for positive values of the klln function , the support of the bloch wigner dilogarithm and the klln function within the domain triangle are both given by the klln triangle shown in fig .
[ viretri ] . a first glimpse on the topographical plot of both functions in fig . [ viretop ] might lead to the conjecture that these functions are the same up to normalization . in order to ( dis)prove this conjecture we have to compare the contour lines of the two functions
. the domain triangle is parametrized by @xmath92 and confined by the three straight lines connecting the points @xmath93 even though the domain triangle and , as a part of it , the klln triangle in fig .
[ viretri ] represent the kinematic situation in the most symmetrical way ( cf .
[ viretop ] ) , there is at least a second representation which is more appropriate for calculating .
one can solve the projective condition @xmath94 with fixed value @xmath95 for @xmath20 . in this case
the domain triangle is projected onto the @xmath96 plane . the resulting _ projective triangle _
is confined by the two axes and the line @xmath97 .
the contour line of the klln function with fixed ( negative ) value @xmath98 in the projective plane @xmath94 is given by @xmath99 for @xmath100 the contour degenerates to a triangle confined by the lines @xmath101 , @xmath102 and @xmath103 which is isomorphic to the klln triangle . for @xmath104 the contour lines run within this projective form of the klln triangle .
it can be easily seen that the minimal value of the klln function is found at @xmath105 with a value @xmath106 . in the parametrization ( [ mxmy ] ) of the domain triangle
the minimum is located at @xmath107 .
the value of the bloch wigner dilogarithm is maximal at the same point @xmath105 ( or @xmath108 ) , resulting in @xmath109 where @xmath110 is the clausen function . however ,
if taking an intermediate value for @xmath98 , a contour line for the klln function has not a constant height for the bloch wigner dilogarithm .
this is shown in fig .
[ virecon ] for three different values of @xmath98 .
in this brief note we have shown that the appearance of the kllen function is always related to the virtuality or reality of three - particle interactions .
we have derived kinematic relations , phase space limits and the relationship to lorentz boosts .
starting from the hexagon orbit for classical dilogarithms we have considered the invariance of the bloch wigner dilogarithm under a similar hexagon orbit , showing similar symmetry properties and the same support for the bloch wigner dilogarithm and the klln function .
however , we have disproved the conjecture that these two functions are related in a simple way .
this work was supported by the estonian institutional research support under grant no .
iut2 - 27 , and by the estonian science foundation under grant no .
s.g . acknowledges support by the mainz institute of theoretical physics ( mitp ) .
j. beringer _
[ particle data group collaboration ] , phys .
* d86 * ( 2012 ) 010001 e. byckling and k. kajantie , phys .
* 187 * ( 1969 ) 2008 e. byckling and k. kajantie , `` particle kinematics , '' john wiley & sons , 1973 l. lewin , `` dilogarithms and associated functions , '' macdonald , london , 1958 d. zagier , `` the dilogarithm function , '' in _ frontiers in number theory , physics , and geometry ii on conformal field theories , discrete groups and renormalization _ pierre cartier , bernard julia , pierre moussa and pierre vanhove ( eds . ) , springer , new york , 2007 , pp . | we enlighten relations between the klln function , allowing in a simple way to distinguish between virtual and real processes involving massive particles , and the dilogarithms occurring as results of loop calculations for such kind of processes .
mitp/14 - 032 + * virtual and real processes , the klln function , + and the relation to dilogarithms * + l. kaldame@xmath0 and s. groote@xmath1 + @xmath0 loodus- ja tehnoloogiateaduskond , fsika instituut , + tartu likool , the 4 , 51010 tartu , estonia + @xmath2prisma cluster of excellence , institut fr physik , johannes - gutenberg - universitt , + staudinger weg 7 , 55099 mainz , germany |
the vacuum stability requirement in the ( sm ) imposes a severe lower bound on the mass of the higgs boson , @xmath0 [ 1 - 4 ] , which depends on the mass of the top quark @xmath5 and on the cut - off @xmath6 beyond which the sm is no longer valid . roughly speaking ,
this is due to the fact that the top yukawa coupling , @xmath7 , drives the quartic coupling of the higgs potential , @xmath8 , to negative values at large scales , thus destabilizing the standard electroweak vacuum .
this bound is very relevant for us since it lies in a region that could be accesible for present ( lep , tevatron ) and future ( lep-200 , lhc ) accelerators .
it is therefore worthwile to re - examine this important issue taking into account recent improvements on the evaluation of the effective potential and the corresponding higgs mass @xcite .
this is the aim of this article . in previous works
, the stability bound was obtained from the tree level potential , improved by one - loop or two - loop renormalization group equations ( rge ) for the @xmath9- and @xmath10- functions of the running couplings , masses and the @xmath3-field ( see e.g. refs .
however , it has been shown that the one - loop corrections to the higgs potential are important in order to fix the boundary conditions for the electroweak breaking and calculate the higgs mass in a consistent and scale - independent way . as we will see , they are also significant to properly understand the whole structure of the potential . in section
2 we write the one - loop corrected higgs potential , giving the general conditions for extremals . in section
3 we apply the previous results to the determination of the realistic sm vacuum ( which implies a relationship between the boundary conditions of the sm parameters ) and the physical higgs mass up to next - to - leading - log order . for a given top mass , there is a one - to - one correspondence between @xmath0 and the boundary condition for @xmath11 .
we discuss also in this section the differences between our approach and previous calculations . in section 4
we study the form of the higgs potential for large @xmath3 and determine under which circumstances it develops a maximum and an additional minimum .
this allows us to define the conditions for vacuum stability in the sm , from which we extract the lower bound on @xmath0 as a function of @xmath5 and @xmath6 . in section 5
we present our complete numerical results .
typically we find that the lower bound on @xmath0 is @xmath12 lower than in the last previous estimate @xcite .
in addition , we perform a detailed comparison between the sm lower bounds on @xmath0 and the supersymmetric upper bounds on it .
it turns out that depending on the actual value of @xmath5 , the eventually measured higgs mass can discard the pure sm , the minimal supersymmetric standard model or both .
the renormalization group improved effective potential of the sm , @xmath2 , can be written in the t hooft - landau gauge and the @xmath13 scheme as @xcite [ veff ] v[(t),_i(t);(t)]v_0 + v_1 + , where @xmath14 runs over all dimensionless and dimensionful couplings and @xmath15 , @xmath16 are respectively the tree level potential and the one - loop correction , namely with m_i^2()=_i^2(t)-_i , and n_1=6 , & _
1=g^2(t ) , & _ 1=0 , & c_1= ; + n_2=3 , & _ 2=[g^2(t)+g^2(t ) ] , & _ 2=0 , & c_2= ; + n_3=-12 , & _
3=h_t^2(t ) , & _ 3=0 , & c_3= ; + n_4=1 , & _
4=(t ) , & _
4=m^2(t ) , & c_4= ; + n_5=3 , & _
5=(t ) , & _
5=m^2(t ) , & c_5= .
@xmath17 are the tree - level expressions for the masses of the particles that enter in the one - loop radiative corrections , namely @xmath18 , @xmath19 , @xmath20 , @xmath21 , @xmath22 .
@xmath23 $ ] is the one - loop contribution to the cosmological constant @xcite , which will turn to be irrelevant in our calculation . in the previous expressions
the parameters @xmath11 and @xmath24 are the sm quartic coupling and mass , whereas @xmath25 , @xmath26 , @xmath27 are the su(2 ) , u(1 ) and top yukawa couplings respectively .
all of them are running with the rge .
the running of the higgs field is [ field ] ( t)=(t ) _ c , @xmath28 being the classical field and @xmath29 , where @xmath30 is the higgs field anomalous dimension . finally the scale @xmath31 is related to the running parameter @xmath32 by [ mu ] ( t)=e^t , where @xmath33 is a fixed scale , that we will take equal to the physical @xmath34 mass , @xmath35 . it has been shown @xcite that the l - loop effective potential improved by ( l+1)-loop rge resums all lth - to - leading logarithm contributions .
consequently , we will consider all the @xmath9- and @xmath10-functions of the previous parameters to two - loop order , so that our calculation will be valid up to next - to - leading logarithm approximation . as has been pointed out in ref .
@xcite , working with @xmath36 ( and higher derivatives ) rather than with @xmath2 itself allows to ignore the cosmological constant to be a function of the @xmath3-field since the scale invariant properties of @xmath2 allow to perform the substitution either before or after taking the derivative @xmath37 @xcite .
we will turn to this point below . ]
term @xmath38 .
in fact , the structure of the potential can be well established once we have determined the values of @xmath3 , say @xmath39 , in which @xmath2 has extremals ( maxima or minima ) , thus we only need to evaluate @xmath36 and @xmath40 . from eq .
( [ veff ] ) , the value of @xmath39 is given by [ first ] .|_(t)=
_ ext(t)=0 ^2_ext(t)= .
the second derivative is given by [ second ] .|_(t)= _
ext(t ) = 2m^2+_i + ^2(t)_i , where we have used ( [ first ] ) . eq .
( [ second ] ) can be written in a more suggestive form as [ secondbis ] .|_(t)= _
ext(t ) & = & 2m^2+(_-4)^2(t ) + & + & ( _ m^2 - 2m^2 ) , where @xmath41 , @xmath42 and @xmath10 are one - loop @xmath9- and @xmath10-functions : [ betas ] 16 ^ 2_&=&12(^2-h^4_t+h_t^2 ) - ( 3g^2 + 9g^2 ) + , + 16 ^ 2_m^2&=&m^2 , + 16 ^ 2&=&3(h^2_t- g^2- g^2 ) .
the form of eq .
( [ secondbis ] ) reflects the scale invariance of @xmath2 .
indeed , it is straightforward to check that the requirement @xmath43 applied to the one - loop effective potential of eq .
( [ veff ] ) at any value of @xmath3 implies up to two - loop corrections . substituting ( [ p4inv ] ) and ( [ p2inv ] ) in ( [ second ] ) we obtain eq .
( [ secondbis ] ) . to conclude this section
, we would like to stress the fact that even though the whole effective potential is scale - invariant , the one - loop approximation is _
not_. therefore , one needs a criterion to choose the appropriate renormalization scale in the previous equations .
as was shown in @xcite , a sensible criterion is to choose the scale , say @xmath44 , where the effective potential is more scale - independent .
@xmath45 turns out to be a certain average of the @xmath46 masses .
actually , it was shown in @xcite , and we would like to emphasize it here , that any choice of @xmath45 around the optimal value produces the same results for physical quantities up to tiny differences .
we follow in this section the approach in ref .
@xcite , to which we refer the reader for further details .
the boundary conditions for the parameters of the higgs potential , @xmath47 , @xmath11 , are constrained by the fact that @xmath2 must develop an electroweak breaking minimum consistent with the experimental observations . in the framework of the standard model
realistic " minimum corresponds to our present vacuum .
accordingly , we must impose [ vev ] ( t_z)=v=(g_)^-1/2=246.22 gev , where @xmath48 is defined as @xmath49 and @xmath50 is the measured " vev for the higgs field - decay slightly modifying the relation ( [ vev ] ) , see e.g. ref .
@xcite . ] @xcite . on the other hand @xmath51 must be obtained from the minimum condition ( [ first ] ) . if we evaluate eq .
( [ first ] ) at the convenient scale @xmath44 , what we get is @xmath52 rather than @xmath53 .
then , @xmath53 is obtained through its rge [ vevstar1 ] ( t_z)=(t^ * ) .
the criterion to choose @xmath45 has been stated in the previous section .
we can trade @xmath52 by @xmath54 from the condition of minimum ( [ first ] ) , which translates , using ( [ vevstar1 ] ) and ( [ vev ] ) , into the boundary condition for @xmath47 : [ bcm ] & = ( t^*)^2(t^ * ) + ^2(t^*)\ { g^4(t^ * ) .
+ & + ^2 + & -.4h_t^4(t^ * ) } , where we have neglected the higgs and goldstone boson contributions .
the running higgs mass , @xmath55 , defined as the curvature of the scalar potential at the minimum , can be readily obtained from ( [ second ] ) evaluated at the scale @xmath56 , i.e. m_h^2(t^*)=.|_(t^ * ) = ( t^ * ) .
the scale invariance of the second derivative of the potential , @xmath57=0 $ ] , allows us to write @xmath55 at any arbitrary scale [ mhrun ] m_h^2(t)=m_h^2(t^ * ) . the physical ( pole )
higgs mass , @xmath58 , is then given by [ mhphys ] m_h^2= m_h^2(t ) + re[(p^2=m_h^2)-(p^2=0 ) ] , where @xmath59 is the renormalized self - energy of the higgs boson ( the @xmath32-dependence of ( [ mhphys ] ) drops out ) .
explicit expressions for @xmath60 and @xmath61 can be found in ref .
@xcite . as has been stated above
, the choice of @xmath45 , i.e. the scale at which we evaluate the minimum conditions , is not important for physical quantities , provided it is within a ( quite wide ) region around the optimal value .
this is illustrated for the higgs mass with a representative example in fig . 1 ( solid line ) .
the lack of flatness of @xmath0 reflects the effect of all non - considered ( higher order ) contributions in the calculation and , therefore , it is a measure of the total error in our estimate of @xmath0 . from fig .
1 we can see that the error is typically @xmath62 3 gev , which is the uncertainty we should assign to our results .
the dashed line is the corresponding result performing the previous calculations just with the ( rge improved ) tree - level part of @xmath2 in eq .
( [ veff ] ) , as was done in refs .
then , the higgs mass has a strong dependence on @xmath45 . choosing @xmath63 , as it was done in refs .
@xcite , results in an error in the estimate of @xmath0 , whose precise value depends on the top mass and is typically of @xmath64 , showing the need of a more careful treatment of the problem , as the one exposed above . finally , let us note that in the previous equations the top yukawa coupling , @xmath27 , enters in several places .
therefore , the higgs mass depends on the boundary condition chosen for @xmath27 , and thus on the top mass @xmath5 .
however the running top mass , defined as @xmath65 , does not coincide with the physical ( pole ) mass , @xmath5 . in the landau gauge the relationship between the running @xmath66 and the physical ( pole ) mass @xmath5 is given by @xcite [ mtphys ] m_t=\{1 + + ( ) ^2 } m_t(m_t ) . where @xmath67 , @xmath68 represent the masses of the five lighter quarks . summarizing the results of this section , for a given top mass and a boundary condition for @xmath11 ,
the boundary condition of @xmath47 is obtained from the electroweak breaking constraint , and is given by eq.([bcm ] ) .
then , the higgs mass , @xmath0 , can be calculated from eq.([mhphys ] ) .
the structure of maxima and minima of @xmath2 for large @xmath3 can be studied with the equations ( [ first ] ) and ( [ secondbis ] ) , evaluated at a scale @xmath31 within the region where @xmath2 is scale - invariant . as was discussed in @xcite and in the previous section , @xmath69 is always a correct choice ( other choices , such that @xmath70 , are equally valid and lead to the same results ) .
then the extremal condition ( [ first ] ) , neglecting the higgs and goldstone contributions , reads [ phiext ] ^2_ext= , with [ lambdahat ] & = & - \{6h_t^4- g^4 . + & & + & -&.(g^2+g^2)^2 } , . from ( [ phiext ] )
we see that , if @xmath2 develops an extremal for large values of @xmath3 , this must occur for a value of @xmath3 such that [ lext ] 0<[(t)=(t)]1 . on the other hand , for large values of @xmath3 eq .
( [ secondbis ] ) can be very accurately expressed as [ secblarge ] .|_(t)= _
ext(t ) & = & ( _ -4)^2(t ) . since near the extremum @xmath71 , and thus @xmath8 , is very small , we see from ( [ secblarge ] ) that depending on the sign of @xmath41 we will have a maximum or a minimum .
we have illustrated these features in fig . 2 with a typical example .
2a represents the evolution of @xmath8 ( dashed line ) and @xmath71 ( solid line ) with @xmath31 .
it is worth noticing that they do not cross the horizontal axis at the same value of @xmath31 , but they differ by more than one order of magnitude .
this is important since the point where the maximum of the potential is located , say @xmath72 , does correspond to @xmath73 rather than @xmath74 ( see eq .
( [ lext ] ) ) .
this is apparent in fig . 2b , where the scalar potential , @xmath1 , has been represented that has been plotted in fig .
2b has been chosen in order to give a continuous and faithful representation of @xmath2 in logarithmic units . ] . the open diamond and square in fig .
2a correspond to the position of the maximum and the minimum , respectively , of @xmath1 .
notice also that for values of @xmath3 very slightly higher than @xmath72 , the potential is negative and much deeper than the realistic minimum .
this is simply because for values of @xmath31 just beyond @xmath75 , the value of @xmath71 becomes negative and the potential is dominated by the contribution @xmath76 ( compare eqs .
( [ veff ] ) and ( [ lambdahat ] ) ) .
consequently , a sensible criterion for a model to be safe is to require one of the two following conditions : _ a ) _ : : the potential has no maximum . _
b ) _ : : the maximum occurs for @xmath77 , where we recall that @xmath6 is the cut - off beyond which the is no longer valid . in the following
we will assume @xmath78 gev . with this criterion [ in particular condition ( _ b _ ) ]
we see that the model represented in fig .
2 is acceptable for @xmath79 gev . beyond this scale
the stability of the vacuum requires the appearance of new physics .
note from this discussion that conditions ( _ a _ ) , ( _ b _ ) are not equivalent to require @xmath80 for @xmath81 , as is usually done . instead , the significant parameter is @xmath71 rather than @xmath82 .
we have represented in fig .
3 the evolution of @xmath71 for a set of models with the same value of @xmath5 as the previous one ( @xmath83 gev ) , but with different boundary conditions for @xmath11 , and thus different values of @xmath0 ( obtained as explained in sect . 3 ) .
the thickest line ( @xmath84 gev ) , which is tangent to the horizontal axis at @xmath85 gev , represents a limiting case : for @xmath86 gev the potential has no maximum and therefore is completely safe for any choice of @xmath6 [ see condition ( _ a _ ) above ] ; for @xmath87 gev the models are safe depending on the value of @xmath88 .
hence , for @xmath89 the lower bound on @xmath0 is insensitive to the value of @xmath6 .
the situation depicted in this example , i.e. for @xmath83 gev , typically occurs when the top mass is rather small , since then the top yukawa coupling is too small at large scales to maintain @xmath41 negative [ see eq . ( [ betas ] ) ] .
when the top mass is higher the situation is different and is illustrated with the case @xmath90 gev in fig .
we see that for each value of @xmath0 , @xmath91 crosses the horizontal axis at most once , and there is a value of @xmath0 for which the cross occurs at @xmath92 gev . in consequence , for @xmath93 gev there is a one - to - one correspondence between the choice of @xmath6 and the value of the lower bound on @xmath0 . finally , we would like to point out that generically the potential is _ not _ unbounded from below since whenever there is a maximum , there is an additional minimum for a larger value of @xmath3 , as illustrated in fig . 2b .
this is because @xmath71 gets back to the positive rage for large enough scales .
hence , the potential becomes eventually positive and monotonically increasing , although in some cases , e.g. for @xmath90
gev , this occurs for values of @xmath3 beyond @xmath94 gev .
as stated in the introduction and has become clear in previous sections , the lower bound on @xmath0 is a function of @xmath5 and @xmath6 .
however , apart from the previously estimated error @xmath62 3 gev in our calculation , there is an additional source of uncertainty coming from the value of @xmath95 , which enters in several places in the previous calculation .
the most recent estimate of @xmath95 gives @xcite [ alphas ] _
s=0.1240.006 . using the central value of ( [ alphas ] ) ,
we have represented in fig .
5 the lower bound on @xmath0 as a function of @xmath5 for different values of @xmath6 .
the form of the curves is easily understandable from the discussion of the previous section . in fig . 6 , which corresponds to the pure sm case , we have fixed @xmath6 at its maximum value , @xmath92 gev , and represented the lower bound on @xmath0 for the central value of @xmath95 in ( [ alphas ] ) ( diagonal solid line ) and the two extreme values ( diagonal dashed lines ) .
if we use the recent evidence for the top quark production at cdf with a mass @xmath96 gev @xcite , we obtain the following lower bound on @xmath0 : [ 174bound ] m_h > 12833 gev , i.e. @xmath97 95 gev ( 1@xmath98 ) . if the higgs is observed in the present or forthcoming accelerators with a mass below the bound of eq .
( [ 174bound ] ) , this would be a clear signal of new physics beyond the . comparing these bounds with the last evaluation performed in ref .
@xcite , we see that our values of @xmath0 are lower than in @xcite by an amount which increases with @xmath5 , going from @xmath99 gev for @xmath100 gev , to @xmath101 gev for @xmath102 gev . as has been discussed in sect . 3
, the main reason of this difference is the way in which the higgs mass was computed in ref .
accordingly , our results give more room to the higgs mass in the framework of the standard model .
it is very interesting to perform a comparison between the sm _ lower _ bounds on @xmath0 previously obtained and the supersymmetric _ upper _ bounds on @xmath0 @xcite .
we briefly recall that in the minimal supersymmetric standard model ( mssm ) the higgs quartic coupling is not a free parameter , but is given by a certain combination of the @xmath103 , @xmath104 gauge couplings .
thus , experimental data constrain the boundary condition for @xmath8 , which can not be as large as we want , leading to strong upper bounds on @xmath0 at large scales remains always positive and hence there are no corresponding lower bounds on @xmath0 from vacuum stability in the mssm . ] .
these bounds depend on three parameters ( besides @xmath5 ) : @xmath105 , i.e. the scale below which supersymmetry ( susy ) decouples ( from naturality reasons @xmath106 tev @xcite ) ; @xmath107 , i.e. the ratio @xmath108 of the two supersymmetric higgs doublets ; and @xmath109 , i.e. the mixing between stops , which is responsible for the threshold correction to the higgs quartic coupling . the larger threshold correction and @xmath107 , the less stringent the susy bounds .
therefore , the most conservative situation takes place considering maximum threshold correction ( which is achieved for @xmath110 ) and @xmath111 .
likewise , the larger @xmath105 , the less stringent the bounds ; but , as mentioned above , it is not sensible to consider @xmath105 much larger than @xmath112 tev . consequently , to be in the safe side , we have represented in fig .
6 the mssm upper bounds ( transverse solid and dashed lines ) , as recently obtained up to next - to - leading - log order in ref .
@xcite , in the most conservative situation with @xmath113 tev .
of course , since in the susy case @xmath105 indicates the appearance of new physics , the supersymmetric upper bounds must be consistent with the sm lower bounds for @xmath114 , as in fact they are .
however , setting @xmath92 gev , i.e. the pure sm case , as in fig . 6 , we find that this is not so for all the top masses , leading to an interesting situation .
we can distinguish three zones in fig .
6 : _ i ) _ : : for @xmath115 gev , i.e. the crossing area of the sm and mssm curves , the eventually measured higgs mass will be compatible either with the pure sm or with the mssm , but not with both at the same time ( unless @xmath116 with high accuracy ) .
accordingly , the experimental higgs mass either will discard the mssm or will be a clear signal of new physics beyond the sm compatible with the mssm . _
ii ) _ : : for @xmath117 gev , the situation is analogous , but there is a wider range of higgs masses ( area within the two curves ) compatible with both sm and mssm .
_ iii ) _ : : for @xmath118 gev , there is no region of higgs masses compatible with the sm and mssm simultaneously . on the contrary there
is a range of @xmath0 ( within the two curves ) which would discard both .
perhaps the most interesting situation is ( _ i _ ) , which occurs precisely for a top mass in the central value of the cdf experimental estimate . in order to facilitate the comparison between the pure sm and the mssm
, we give below fits of the corresponding bounds on @xmath0 , valid for 150 gev @xmath119 gev and within @xmath120 gev of error .
[ fits ] sm : m_h>127.9 + 1.92(m_t-174)-4.25 ( ) , = 10 ^ 19 gev : m_h<126.1 + 0.75(m_t-174)-0.85 ( ) , _ s=10 ^ 3 gev where @xmath0 and @xmath5 are expressed in gev .
_ note added _ : after this work was finished we received a preprint by g. altarelli and g. isidori @xcite , where the sm lower bound on @xmath0 is refined using a different approach .
our results are typically a few gev below theirs , though both are consistent within the estimated errors .
# 1 # 2 # 3 _ mod .
_ * # 1 * ( # 2 ) # 3 # 1 # 2 # 3 _ nucl .
_ * b#1 * ( # 2 ) # 3 # 1 # 2 # 3 _ phys .
_ * b#1 * ( # 2 ) # 3 # 1 # 2 # 3 _ phys
_ * # 1 * ( # 2 ) # 3 # 1 # 2 # 3 _ phys .
_ * d#1 * ( # 2 ) # 3 # 1 # 2 # 3 _ phys .
_ * # 1 * ( # 2 ) # 3 # 1 # 2 # 3 _ prog .
_ * # 1 * ( # 2 ) # 3 # 1 # 2 # 3 _ rev . mod .
_ * # 1 * ( # 2 ) # 3 # 1 # 2 # 3 _
z. phys .
_ * c#1 * ( # 2 ) # 3 99 m. lindner , 31 1986 295 .
m. sher , 179 1989 273 .
m. lindner , m. sher and h. w. zaglauer , 228 1989 139 .
m. sher , 317 1993 159 ; addendum : [ hep - ph/9404347 ] . c. ford , d. r. t. jones , p. w. stephenson and m. b. einhorn , 395 1993 17 .
j. a. casas , j. r. espinosa , m. quirs and a. riotto , preprint cern - th.7334/94 [ hep - ph/9407389 ] .
m. quirs , to appear in the proceedings of _ physics from planck scale to electroweak scale _ ,
warsaw , poland , 2124 september 1994 .
b. kastening , 283 1992 287 ; + m. bando , t. kugo , n. maekawa and h. nakano , 301 1993 83 ; 90 1993 405 .
s. peris , 16 1991 1505 .
h. arason et al . , 46 1992 3945 .
n. gray , d. j. broadhurst , w. grafe and k. schilcher , 48 1990 673 .
p. langacker , preprint upr-0624 t [ hep - ph/9408310 ] .
f. abe et al . , cdf collaboration , preprint fermilab - pub-94/097-e and 73 1994 225 .
y. okada , m. yamaguchi and t. yanagida , 85 1991 1 ; j. ellis , g. ridolgi and f. zwirner , 257 1991 83 ; h.e . haber and r. hempfling , 66 1991 1815 ; r. barbieri , m. frigeni and f. caravaglios , 258 1991 167 ; j.r .
espinosa and m. quirs , 266 1991 389 .
krasnikov and s. pokorski , 288 1991 184 ; m.a .
daz , t.a .
ter veldius and t.j .
weiler , vanderbilt preprint vand
th9414 , 1994 .
r. barbieri and g.f .
giudice , 306 1988 63 ; b. de carlos and j.a .
casas , 309 1993 320 .
g. altarelli and g. isidori , cern th.7351/94 .
fig . 1 : : plot of the physical higgs mass , @xmath0 , as a function of @xmath121 in a model with @xmath122 gev , @xmath92 gev and @xmath123
. the solid line corresponds to our approach using the full one - loop effective potential , @xmath2 , improved by two - loop rge .
the dashed line is the result using only the tree level part of @xmath2 , improved with two - loop rge , as was done in refs .
2a : : plot of @xmath8 ( dashed line ) and @xmath71 ( solid line ) as a function of the scale @xmath31 for @xmath6 and @xmath95 as in fig . 1
, @xmath83 gev and @xmath124 gev .
2b : : plot of the scalar potential , @xmath1 , corresponding to fig .
2a , represented in a convenient choice of units as described in the text .
fig . 3 : : plot of @xmath71 as a function of the scale @xmath31 for @xmath83 gev , @xmath92 gev and @xmath123 .
the curves are shown for intervales of 5 gev in @xmath0 in the range @xmath125 gev @xmath126 gev .
the thicker ( tangent ) line corresponds to the case @xmath84 gev .
fig . 4 : : the same as fig . 3 for @xmath90 gev and @xmath127 gev @xmath128 gev .
the thicker line , with @xmath129 , corresponds to the case @xmath130 gev .
fig . 5 : : sm lower bound on @xmath0 as a function of @xmath5 for @xmath131 and different values of @xmath6 in the range @xmath132 gev @xmath133 gev .
the values of @xmath6 for consecutive curves differ in two orders of magnitude .
fig . 6 : : diagonal ( thick ) lines : sm lower bound on @xmath0 as a function of @xmath5 for @xmath92 gev and @xmath123 ( solid line ) , @xmath134 ( upper dashed line ) ,
@xmath135 ( lower dashed line ) .
transverse ( thin ) lines : mssm upper bounds on @xmath0 for @xmath113 tev and @xmath95 as in the diagonal lines . | we re - examine the lower bound on the mass of the higgs boson , @xmath0 , from standard model vacuum stability including next - to - leading - log radiative corrections
. this amounts to work with the full one - loop effective potential , @xmath1 , improved by two - loop rge , and allows to keep control of the scale invariance of @xmath2 in a wide range of the @xmath3-field .
our results show that the bound is @xmath4 less stringent than in previous estimates .
in addition we perform a detailed comparison between the sm lower bounds on @xmath0 and the supersymmetric upper bounds on it .
it turns out that depending on the actual value of the top mass , @xmath5 , the eventually measured higgs mass can discard the pure sm , the minimal supersymmetric standard model or both .
10.5 mm = -1.5 cm 10.5 mm
@=11 stequation
@=12 -18.cm |
a core problem in modern physics , and one which has evolved to assume an unprecedented interdisciplinary character , is related to the study of open quantum systems ( oqss ) . in the broadest sense , these systems may be defined as consisting of some localized , microscopic , region that is coupled to an external environment by means of an appropriate interaction . in discussions of the crossover from quantum mechanics to classical physics ,
the role of such an environment is often invoked to describe the influence of a classical measuring apparatus . even if one is to remove such an apparatus ,
however , the description of most open systems may nonetheless be reduced to one in which the properties of some microscopic region are influenced by its coupling to its own , `` natural '' , environment .
in contrast to the aforementioned measurement problem , the influence of this environment can never be deleted but exists at all times , independent of any observer .
in essence , this environment functions as an `` intrinsic '' ( natural ) measuring apparatus , introducing a coupling between different states of the open system .
two very - different cases in which this situation applies are provided by the decay of unstable states in nuclei , and the transport of electrons through mesoscopic quantum dots . in the former case ,
the environment consists of a continuum of scattering wavefunctions outside of the nucleus , which may mediate the escape of either a neutron or a proton from that structure @xcite .
similarly , in discussions of quantum - dot transport , the discrete quantized states of an isolated cavity develop a broadening when the cavity is coupled to macroscopic reservoirs through the addition of appropriate leads . in this latter case ,
the environment is provided by the states of these reservoirs , and the broadening of the discrete cavity levels is strongly dependent upon the mutual overlap with these states , as mediated through the leads @xcite .
there are numerous examples of oqss , from a broad spectrum of disciplines within physics .
while obvious examples include nuclei , atoms , and molecules , a much broader range of systems is provided by lasers and optically - active media , biomolecules and molecular networks , nanophotonic structures , and mesoscopic electronic devices .
it is the latter systems ( namely mesoscopic devices ) that we focus on in this review , motivated by two of their important features .
the first of these is their capacity to exhibit a variety of rich phenomena arising from their environmental interaction , while the second is the ability to subject them to sophisticated external control . quite generally , mesoscopic devices are small metallic or semiconducting structures , in which carrier motion is constrained on a spatial scale that is comparable to , or even smaller than , the fundamental length scales associated with transport .
these length scales include the elastic mean free path , the phase - coherence length , and the inelastic scattering length , all of which may exceed the size of sub - micron scale devices at sufficiently low temperatures @xcite . in terms of their general structure ,
these devices consist of some central scattering region , in the form of a quasi - one dimensional wire or a quasi - zero - dimensional quantum dot ( qd ) , whose carriers may be injected into and extracted from in order to allow for transport .
such transport is achieved by connecting the central scattering region of the device to macroscopic charge reservoirs , by means of appropriate _ lead
_ structures . when compared with systems such as nuclei ,
the great advantage that mesoscopic devices offer is the capacity to externally _ control _ the key parameters of the system .
these include the system size , and thus the nature of the quantum states involved in transport , the strength of the coupling between the system and its environment , and the energy of the carriers involved in transport .
this control makes these structures ideally suited to the study of forefront issues associated with oqss .
we demonstrate this here for few - electron qds @xcite , which are the solid - state analog of scattering billiards , and for quasi - one - dimensional quantum point contacts ( qpcs ) , in which the presence of a strong environmental coupling may modify the very nature of the quantum states responsible for transport @xcite .
m in size . to illustrate the connection to the non - hermitian concept , we have also indicated in the figure how this mesoscopic device may essentially be viewed as consisting of a localized quantum system ( denoted here as the region ` @xmath0 ' ) that is embedded within an environment of scattering states ( denoted as the regions ` @xmath1 ' ) .
the coupling between regions @xmath0 and @xmath1 is regulated by means of the two qpc openings in the cavity , which correspond to the `` leads '' referred to in the main text . in this figure , we have also indicated the presence of a classical measuring apparatus that is _ outside _ of the @xmath3 space . while this measurement apparatus provides an `` environment '' that can be removed from the system , the influence of the environment @xmath1 , in contrast , can _ never _ be deleted . ] in this _ report on progress _ , we focus on providing a detailed discussion of the behavior of oqss in terms of the projection operator formalism of quantum mechanics . in this approach , as indicated in fig .
[ oqs ] , we consider the system under study to be comprised of a localized region ( @xmath0 ) that is embedded in a well - defined environment ( @xmath1 ) of scattering wavefunctions .
the complete function space satisfies the requirement @xmath2 , and is described in terms of a hermitian hamiltonian .
the hamiltonian describing the subspace of interest ( @xmath0 ) is non - hermitian , however , and has complex eigenvalues that define both the energy and broadening of its states @xcite .
these eigenstates are coherently coupled to each other , via an interaction that is mediated by the environment ( @xmath1 ) .
this projection operator formalism differs from that in which complex scaling is used to analyze non - hermitian systems , an approach that has been summarized in ref .
@xcite and that we will not discuss .
we also note that the role of the environment considered here is very different to that which is often invoked in discussions of decoherence , arising from the uncontrolled interaction of a quantum system with its environment ( such as an electron or phonon bath ) .
this latter problem has been treated in other works ( see , for example , the reviews of @xcite ) , where the feedback from the environment onto the quantum system is not explicitly considered .
this is in contrast with the situation here , where we are concerned , above all , with understanding the influence of the natural environment ( @xmath1 ) on the system ( @xmath0 ) .
this environmental influence remains even when sources of decoherence have been suppressed ( by lowering the temperature , for example ) . as such , the approach that we outline is expected to be useful in describing the behavior of _ small _ systems , in which the number of particles active in transport is small , and in which the number of quantum states they occupy is similarly restricted in number . under such conditions
we will see that , far from acting as a source of decoherence , the coupling to the environment can in fact mediate an extremely robust and coherent interaction .
the objective of this review is to focus on critical implications of non - hermiticity , namely the presence of _ exceptional points _ ( eps ) in the spectrum of oqss , their influence on the _ rigidity _ of the wavefunction phases , and their ability to initiate a _ dynamical phase transition _ ( dpt ) . as we discuss in further detail below ,
these features can be observed because of our ability to vary the strength of the environmental interaction in oqss , by means of some suitable control parameter .
eps are singular points in the continuum , at which two resonance states coalesce , that is where they exhibit a non - avoided _ crossing_.
this behavior should be contrasted with the well - known _ avoided _ crossings , exhibited by hermitian systems at so - called _
diabolic points_. due to the coalescence of the two resonance states at an ep , the time evolution of the system becomes undefined in its neighborhood . at the same time , the eps are branch points in the complex plane and are responsible for the generation of double poles of the @xmath4 matrix , with associated implications for transmission . an important quantity that indicates the range of influence of an ep is the phase rigidity of its two eigenfunctions . in isolated ( hermitian ) systems
the phases of the eigenfunctions are said to be rigid , a statement that expresses the well - known orthogonality of these functions . in open ( non - hermitian ) systems , however , the presence of the environmentally - mediated coupling between the eigenstates means that they are no longer orthogonal , but rather exhibit the property of _ biorthogonality_.
this allows the rigidity of the eigenfunction phases to be reduced , particularly near an ep where the rigidity actually vanishes .
this surprising behavior expresses the very strong influence of the environment on the localized system at an ep .
while the physical significance of eps for the dynamics of oqss has only recently begun to be studied ( see , for example , the review @xcite ) , they are nonetheless understood to be intimately connected to the observation of dpts in different open systems . the term dpt has been coined to refer to a phenomenon in which the quantum dynamics of the open system undergoes a phase transition between non - analytically connected states as a function of some suitable control parameter . here , time - dependent approaches to the description of the problem fail @xcite . instead ,
much like a conventional phase transition in , for example , magnetism or superconductivity , the physical behavior of the system on one side of the transition does not serve as a reliable indicator of that on the other .
the connection of dpts to the eps is provided by the phenomenon of _ width bifurcation _ , which results in a spectroscopic redistribution in the system into long - lived states with narrow linewidths and a much smaller number of short - lived , strongly - broadened , ones .
an important objective of this review is to connect the theoretical concepts of eps and dpts to the physical behavior manifested in experimental investigations of oqss .
one of the earliest works to explicitly invoke the notion of a dpt involved studies of spin swapping in atomic systems @xcite . in these experiments , the authors studied rabi oscillations due to spin flips in the @xmath5c@xmath6h system , and showed a transition to strongly damped motion by increasing the coupling to an environment formed by a spin bath . although not explicitly discussed as such , evidence of dpts is apparent , also , in earlier work on resonance trapping in microwave cavities @xcite , and on einselection @xcite in mesoscopic quantum dots . in both of these systems , the focus was on understanding the manner in which the states of a quantized cavity are affected by coupling them to an external environment .
the common phenomenon revealed in both cases was of width bifurcation , with certain eigenstates actually becoming narrower when the environmental coupling was increased over a specific range .
the accompanying short - lived eigenstates appear as a background with which the long - lived ones interfere , a behavior that has been observed experimentally in microwave cavities @xcite .
ultimately , the long - lived states may even become discrete , forming so - called _ bound states in the continuum _ @xcite . with the renewed interest in such problems , it is actually interesting to note that the appearance of long - lived `` merkwrdigen '' ( remarkable ) states is actually a problem that dates back to the earliest days of quantum mechanics @xcite !
elsewhere , evidence of a dpt has also been provided in work where a multi - level quantum dot was embedded into one of the arms of an aharonov - bohm interferometer , allowing the evolution of the transmission phase to be monitored across a sequence of resonance states @xcite .
these experiments revealed the presence of unexpected regularity in the measured scattering phases ( so - called phase lapses " ) , when the number of states occupied by electrons in the dot was sufficiently large .
while this behavior could not be fully explained within approaches based upon hermitian quantum theory @xcite , it has recently been established that the phase lapses can be attributed to the non - hermitian character of this mesoscopic system , and to a dpt that occurs as the number of electrons in the dot is varied @xcite .
the observed regularity arises from the overlap of the many long - lived states with the short - lived one , all of which are formed due to the dpt .
more recently , studies of transport in ballistic quantum wires may have also revealed a dpt , involving the formation of a protected channel for conduction under strongly nonequilibrium conditions . in these experiments
the environment is essentially provided by the phonon system , whose influence is controlled by means of the applied source bias and by the strong quantum confinement of the carriers induced within the wire @xcite . a well - known phenomenon from optics that may also involve a dpt is that of dicke superradiance @xcite .
this refers to the effect in which an ensemble of excited atoms within a cavity does not emit radiation randomly , but rather does so in a correlated manner when the atoms experience the same radiation field .
recently , superradiance has been demonstrated for solid - state systems , namely ensembles of self - assembled quantum dots @xcite and a dense semiconductor electron - hole plasma @xcite . in this review
, we will use the various experiments described above to connect the behavior exhibited by oqss to the key concepts of non - hermiticity , eps , and dpts .
we will also introduce the important concept of wavefunction phase rigidity , and the significance of time in open systems , in which we are unable to describe a dpt by means of time - dependent approaches .
we emphasize again that the focus of our review will be on discussions of the properties of open systems consisting of a small number of particles . as such
, we explore the behavior in a very different limit to that relevant , for example , in heavy nuclei .
these contain a much larger number of particles , as well as many closely - neighboring states , and are adequately described by the concepts of random - matrix theory @xcite .
in fact , with the exception of a few specific examples , we will not address phenomena arising in nuclear systems , preferring to note instead that this topic was recently excellently served by an associated review in this journal @xcite . due to the continued importance of mesoscopic systems for the investigation of fundamental quantum phenomena , we believe that the focus of our review will prove to be a particularly useful one .
the remainder of this paper is organized as follows . in the next section ,
we introduce the concept of small quantum systems that are coupled to an environment with which they interact .
we begin by reviewing some of the well known properties of isolated ( hermitian ) quantum systems , emphasizing concepts such as the real nature of their eigenvalues and the rigidity ( orthogonality ) of their eigenfunctions . following this
, we next introduce the projection - operator formalism for oqss , in which the total system is described in terms of two coupled subspaces ( @xmath0 & @xmath1 ) .
the complex character of the eigenvalues and eigenfunctions of the non - hermitian operator of the @xmath0 subspace is described , and its physical implications are discussed .
we introduce also the concept of gain and loss in oqss . in the last part of this section ,
some physical examples of the environmentally - mediated coupling of quantum states are presented , focusing on their manifestations in mesoscopic structures . in section iii
, we discuss the importance of eps to the behavior of oqss .
we formulate this discussion by first of all focusing on the interaction of two resonance states near an ep , following which we study the influence of an ep on the eigenvalues and eigenfunctions of the non - hermitian hamiltonian . in section
iiid we discuss the implications of eps for the @xmath4-matrix of the system , a problem with important implications for the analysis of transmission . in section iv , we consider the role of eps in giving rise to dpts in open systems subjected to some form of external control . in the presence of this control ,
dpts are found to occur when the range of influence of several different eps develops a sufficiently strong overlap .
the connection of dpts to the notion of width bifurcation is emphasized , and several experimental demonstrations of such phase transitions are presented .
these include demonstrations of resonance trapping in microwave cavities , studies of phase lapses in tunneling through quantum dots , and measurements of spin swapping in atomic systems .
emphasizing our belief that dpts are , in fact , an inherent feature of oqss in general , in section v we discuss further examples of physical problems in which dpts may occur . in section
vi , we conclude this review by summarizing its main points and by identifying important issues for further study .
two appendices to the paper are provided , in the first of which ( appendix a ) we make a few additional remarks on the non - hermitian hamitonian ( with and without using perturbation theory ) , and give expressions for the coupling matrix elements ( the partial width amplitudes ) between the system and the environment .
appendix b provides a list of principal symbols and acronyms used in the paper .
in this report on progress , we have reviewed the key physical attributes of oqss in the regime of closely - neighboring states and have described a number of different experimental systems in which they are manifested . at the most basic level , these attributes may be understood as arising from the fact that the hamiltonians needed to describe the dynamics of open systems are non - hermitian , in marked contrast to their hermitian counterparts that so accurately describe the behavior of isolated quantum systems . in the non - hermitian formalism , the open system is viewed as being comprised of a localized region ( @xmath0 ) , embedded in a well - defined environment ( @xmath1 ) of scattering wavefunctions , and the total system ( @xmath2 ) is hermitian .
the coupling between the two subspaces is found to mediate an effective interaction among the original states of the localized system .
this _ natural _
( _ intrinsic _ ) environment does not serve as a source of decoherence ( the role of which is not considered here ) and its influence can never be deleted from the system .
instead , it gives rise to complex eigenvalues , whose real component defines the energy of the state and whose imaginary part describes an effective level broadening
. the eigenfunctions also exhibit an important difference with those of hermitian systems , in that they are no longer orthogonal to one another but are instead biorthogonal .
physically , the biorthogonality expresses the fact that there is an environmentally - induced coupling between the different states of the system .
this should be contrasted with the case of hermitian quantum mechanics , which yields orthogonal stationary states that require a perturbation of the system in order to be coupled .
the description of open systems in terms of this approach is found to be well suited to the solution of a broad range of problems , providing proper care is taken to accurately identify the subspaces @xmath0 and @xmath1 . under such conditions , this approach is expected to provide a good description of small systems , in which the number of relevant quantum states , as well as the number of channels , is restricted . in larger systems that do not conform to these constraints , other approaches , such as random - matrix theory ,
should be more effective .
such approaches have been extensively treated in the literature and have therefore not been considered here .
we have seen in this review how the eigenstates of the non - hermitian hamiltonian exhibit fundamentally new behavior not encountered in hermitian physics .
the first example is provided by the appearance of singular points , referred to as exceptional points ( eps ) .
arising when an appropriate system parameter ( such as the coupling strength between the @xmath0 and @xmath1 subspaces ) is varied , the ep involves the coalescence of two eigenvalues of the non - hermitian operator and ( up to a phase ) their corresponding eigenfunctions .
although a point of measure zero , the ep exerts a strong influence on the spectral properties of the system , extending over a wider region than the singular point at which the two eigenvalues coalesce .
an important property that indicates the range of influence of the ep is the phase rigidity of the eigenfunctions . essentially , the phase rigidity is a measure of the biorthogonality of different eigenfunctions , and , therefore , of the strength of the environmentally - induced interaction between the different states . in the region of the ep
the phase rigidity is significantly reduced , expressing the fact that this interaction via the environment is very strong at the ep . as a result ,
the wavefunctions of these states are no longer orthogonal with one another but are equal , up to a phase . over the range where the phase rigidity is reduced ,
the two eigenstates undergo the phenomenon of width bifurcation , with one state becoming strongly broadened while the other becomes more stable as it decouples from the environment . under this condition ,
transport through the system occurs with high efficiency .
simultaneous with this , the eigenvalues may cross in energy .
this is in complete contrast to the behavior of hermitian systems , which may only ever exhibit an avoided level crossing when two of their eigenstates are close in energy .
the phenomenon of width bifurcation , in which the width of one state increases with increasing environmental coupling while that of the other decreases , is connected to a violation of the fermi golden rule .
this rule suggests that increased coupling should always lead to stronger broadening , in contrast to the behavior of the long - lived state produced by width bifurcation . the other important feature of oqss that we have discussed is their ability to undergo a dynamical phase transition ( dpt ) , when the ranges of influence of different eps overlap with one another .
the dpt involves a change in the dynamics of the system , which arises when width bifurcation occurs hierarchically for many states .
the simplest example of this phenomenon occurs when all states of the system are coupled to a single channel , resulting in the formation of just one strongly - broadened state .
the remaining states are then essentially stabilized , through a process in which they strongly decouple from the environment . in the two - channel case relevant for transport , essentially the same picture holds although now two of the original states develop the strong broadening , leaving all but these two states stabilized .
much like conventional phase transitions in thermodynamics , the details of the dynamics on either side of the dpt are not analytically connected to each other .
that is , the behavior on one side of the transition does not serve as a reliable indicator of that on the other .
both the width bifurcation caused by the eps , and the dpt that these give rise to , result from the fact that the coupling strength to the environment has an imaginary component .
experimental evidence in support of these various concepts has been provided throughout the course of this review , with a particular emphasis having been placed on their manifestations in mesoscopic systems . at the heart of
the non - hermitian scheme is the idea of environmentally - mediated coupling , which yields an effective interaction among the states of the quantum system .
we have seen how this coupling is manifested directly in experiments involving non - locally coupled quantum point contacts ( qpcs ) , in which an intervening continuum serves as the @xmath1 subsystem and mediates a coupling between remote pairs of qpcs .
we have described how this coupling gives rise to novel fano - resonance phenomenology , and have also observed that the environmentally - mediated interaction may provide a stronger coupling between quantum states than a simple tunneling overlap .
this may be attributed to the fact that the environmental interaction is supported by a large number of states that comprise a continuum , resulting in a much stronger effect than would be expected to arise from the direct interaction between the different states of @xmath0 .
several experimental demonstrations of dpts in physical systems were also presented .
these examples included : observations of resonance - trapping and its related phenomena , in open microwave cavities and qds ; a crossover from mesoscopic to universal behavior in the phase lapses found in single - electron tunneling through coulomb - blockaded qds ; observations of spin - swapping dynamics in ensembles of interacting spins , and ; energy - transfer processes in photosynthesis .
similar phenomena are observed also in optics , most notably in the form of the long - studied problem of dicke superradiance , as well as more recent investigations of systems with gain and loss . here ,
however , a better understanding of the equivalence between the optical - wave and schrdinger equations is needed before convincing conclusions can be drawn .
in addition , we have also highlighted several other examples of complex physical phenomena that may well be related to the occurrence of a dpt .
these include superradiance in qds , and the observation of protected - subband formation in the nonequilibrium transport of hot electrons through qpcs .
these observations by no means represent an exhaustive list , but rather suggest instead that dpts are a common property of oqss in general .
having summarized the general status of this field , we can now identify a number of important issues that should be resolved in future studies . while the different experiments described in this review provide several hints that dpts do indeed occur in oqss ,
the presence of the distinct time scales that are expected to accompany this phenomenon has yet to be directly demonstrated . according to the description of width
bifurcation given above , the environmentally - induced interaction in the vicinity of an ep is expected to result in the formation of eigenstates with very different lifetimes ; the first of these is very short and associated with the strongly - broadened state , while the other is much longer and related to the stabilized state .
experiments that can probe the existence of these different time scales are therefore desired .
a problem in attempting to design such experiments , however , comes from the fact that this problem must first be clearly formulated from a theoretical perspective . a simpler task may be to provide a demonstration of the width bifurcation that should serve as the signature of the different time scales , but , to the best of our knowledge , this has not yet been done experimentally , in either mesoscopic or any other physical systems .
time - resolved investigations of oqss may also be useful in exploring issues associated with irreversibility in the vicinity of the dpt .
the issues here can be discussed by referring to the features of fig .
9 , in which we show the system properties and their variation in the vicinity of two eps .
as either ep is approached from the left or the right of the figure , the width bifurcation begins at the ep and becomes maximal once @xmath7 . at this point
the phase rigidity returns to unity , indicating that the two eigenfunctions are orthogonal .
physically , this can be understood to result from the fact that the two states are now associated with very different time scales , and so no longer interact with one another . while the results of fig .
9 suggest that , by passing continuously through @xmath7 and the point of maximum bifurcation , it should be possible to return to two distinct states , in practice this is not possible .
in essence , one of the original states has now been `` lost '' to the environment , and the remaining stable state is therefore unable to interact directly with it .
this is completely consistent with the notion of a dpt ; the spectroscopic properties of the system on one side of the transition are not analytically connected to those on the other , and nonlinear terms in the schrdinger equation play an important role in this transition .
another important property that is of interest in characterizing the properties of oqss is the phase rigidity of their eigenfunctions .
the value of this parameter becomes vanishingly small near an ep , and it is useful also for establishing the range of influence of such points .
while the rigidity itself can probably not be measured directly in a physical system , we have seen here that it is related to a high efficiency of transport ( recall fig .
while this connection may not have been widely appreciated to date , it implies that measurements of the conductance of mesoscopic systems can provide a flexible probe of this important system property .
moreover , the enhancement of the conductance around a dpt is , by itself , of interest for possible applications . throughout this review ,
we have provided a number of different examples of dpts , occurring in a variety of physical systems . from these examples
we suggest that the dpt should be a very general characteristic of open systems , with broader consequences than have been appreciated to date . indeed , the stability properties of classical systems are also governed by non - hermitian degeneracies ( eps ) , as has been nicely shown in @xcite .
our hope therefore is that this review can stimulate interest in the study of this still relatively unexplored problem , and ultimately lead to a deeper understanding of its implications .
+ _ acknowledgment _ : we acknowledge the support of the max planck institute for the physics of complex systems , which enabled jpb to visit dresden twice during the preparation of this manuscript .
jpb is supported by the us department of energy , office of basic energy sciences , division of materials sciences and engineering ( award de - fg02 - 04er46180 ) .
+ _ note added in proof_. subsequent to the acceptance of our report we became aware of recently published work by ruderman et al . @xcite , demonstrating the possibility of a quantum dynamical phase transition in molecular chemical bond formation and dissociation .
the occurrence of this transition was connected to the non - hermitian nature of the hamiltonian in the molecular system , consistent with the dicussions of dynamical phase transitions given here .
d. k. ferry , a. m. burke , r. akis , r. brunner , t. e. day , r. meisels , f. kuchar , j. p. bird and b. r. bennett , j. phys . : condens .
matt . * 26 * , 043001 ( 2011 ) ; + d. k. ferry , a. m. burke , r. akis , r. brunner , t. e. day , r. meisels , f. kuchar , j. p. bird , and b. r. bennett , semicond .
sci . technol .
* 26 * , 043001 ( 2011 ) . c. dembowski , h. d. grf , h. l. harney , a. heine , w. d. heiss , h. rehfeld , and a. richter , phys .
86 * , 787 ( 2001 ) ; + c. dembowski , b. dietz , h. d. grf , h. l. harney , a. heine , w. d. heiss , and a. richter , phys .
90 * , 034101 ( 2003 ) .
a. guo , g.j .
salamo , d. duchesne , r. morandotti , m. volatier - ravat , v. aimez , g.a .
siviloglou and d.n .
christodoulides , phys .
lett . * 103 * , 093902 ( 2009 ) ; + c.e .
rter , k.g .
makris , r. el - ganainy , d.n .
christodoulides , m. segev and d. kip , nature physics * 6 * ( 2010 ) 192 ; + t. kottos , nature physics * 6 * ( 2010 ) 166 .
a. ruschhaupt , f. delgado , and j. g. muga , j. phys .
a * 38 * , l171 ( 2005 ) ; + r. el - ganainy , k. g. makris , d. n. christodoulides , and z. h. musslimani , optics lett . *
32 * , 2632 ( 2007 ) ; + k. g. makris , r. el - ganainy , d. n. christodoulides , and z. h. musslimani , phys .
lett . * 100 * , 103904 ( 2008 ) ; + z. h. musslimani , k.g .
makris , r. el - ganainy , and d. n. christodoulides , phys .
lett . * 100 * , 030402 ( 2008 ) .
j. schindler , z. lin , j.m .
lee , h. ramezani , f.m .
ellis , and t. kottos , j. phys .
a * 45 * , 444029 ( 2012 ) ; + m. liertzer , li ge , a. cerjan , a.d .
stone , h.e .
treci , and s. rotter , phys .
lett . * 108 * , 173901 ( 2012 ) ; + m. chitsazi , s. factor , j. schindler , h. ramezani , f.m .
ellis , and t. kottos , phys .
a * 89 * , 043842 ( 2014 ) .
r. lefebvre , o. atabek , m. sindelka , and n. moiseyev , phys .
lett . * 103 * , 123003 ( 2009 ) ; + i. gilary and n. moiseyev , j. phys .
b * 45 * , 051002 ( 2012 ) ; + r. uzdin and n. moiseyev , phys .
a * 85 * , 031804 ( 2012 ) .
h. ramezani , t. kottos , r. el - ganainy , and d.n .
christodoulides , phys .
a * 82 * , 043803 ( 2010 ) ; + h. ramezani , d.n .
christodoulides , v. kovanis , i. vitebskiy , and t. kottos , phys .
109 * , 033902 ( 2012 ) ; + h. ramezani , j. schindler , f.m .
ellis , u. gnther , and t. kottos , phys .
a * 85 * , 062122 ( 2012 ) ; + n. bender , s. factor , j.d .
bodyfelt , h. ramezani , d.n .
christodoulides , f.m .
ellis , and t. kottos , phys .
lett . * 110 * , 234101 ( 2013 ) ; + f. nazari , n. bender , h. ramezani , m.k .
morawej - farshi , d.n .
christodoulides , and t. kottos , optics express*22 * , 9574 ( 2014 ) .
m. scully , e.s .
fry , c.h.r .
ooi , and k. wodkiewicz , phys .
* 96 * , 010501 ( 2006 ) ; + a. svidzinsky , j.t .
chang , h. lipkin , and m. scully , j. mod .
optics * 55 * , 3369 ( 2008 ) ; + a. svidzinsky and j.t .
chang , phys .
a * 77 * , 043833 ( 2008 ) ; + m. scully , phys .
lett . * 102 * , 143601 ( 2009 ) ; + e. a. sete , a. a. svidzinsky , h. eleuch , z. yang , r. d. nevels , and m. o. scully , journ .
modern optics 57 , 1311 ( 2010 ) .
r. rhlsberger , k. schlage , b. sahoo , s. couet , and r. rffer , science * 328 * , 1248 ( 2010 ) ; + r. rhlsberger , special issue _ quantum physics with non - hermitian operators : theory and experiment _ , fortschr .
phys . * 61 * , 360 ( 2013 ) g.l .
celardo , f. borgonovi , m. merkli , v.i .
tsifrinovich , and g.p .
berman , j. phys .
c * 116 * , 22105 ( 2012 ) ; + a.i .
nesterov , g.p .
berman , and a.r .
bishop , special issue _ quantum physics with non - hermitian operators : theory and experiment _ , fortschr .
phys . * 61 * , 95 ( 2013 ) ; + d. ferrari , g.l .
celardo , g.p .
berman , r.t .
sayre , and f. borgonovi , j. phys .
c * 118 * , 20 ( 2014 ) .
kirillov , special issue _ quantum physics with non - hermitian operators : theory and experiment _ , fortschr . phys . * 61 * , 205 ( 2013 ) ; + o.n .
nonconservative stability problems of modern physics _ , de gruyter , 2013 . | this report on progress explores recent advances in our theoretical and experimental understanding of the physics of open quantum systems ( oqss ) .
the study of such systems represents a core problem in modern physics that has evolved to assume an unprecedented interdisciplinary character .
oqss consist of some localized , microscopic , region that is coupled to an external environment by means of an appropriate interaction .
examples of such systems may be found in numerous areas of physics , including atomic and nuclear physics , photonics , biophysics , and mesoscopic physics .
it is the latter area that provides the main focus of this review , an emphasis that is driven by the capacity that exists to subject mesoscopic devices to unprecedented control .
we thus provide a detailed discussion of the behavior of mesoscopic devices ( and other oqss ) in terms of the projection - operator formalism , according to which the system under study is considered to be comprised of a localized region ( @xmath0 ) , embedded into a well - defined environment ( @xmath1 ) of scattering wavefunctions ( with @xmath2 ) .
the @xmath0 subspace must be treated using the concepts of non - hermitian physics , and of particular interest here is : the capacity of the environment to mediate a coupling between the different states of @xmath0 ; the role played by the presence of exceptional points ( eps ) in the spectra of oqss ; the influence of eps on the rigidity of the wavefunction phases , and ; the ability of eps to initiate a dynamical phase transition ( dpt ) .
eps are singular points in the continuum , at which two resonance states coalesce , that is where they exhibit a non - avoided crossing .
dpts occur when the quantum dynamics of the open system causes transitions between non - analytically connected states , as a function of some external control parameter .
much like conventional phase transitions , the behavior of the system on one side of the dpt does not serve as a reliable indicator of that on the other .
in addition to discussing experiments on mesoscopic quantum point contacts that provide evidence of the environmentally - mediated coupling of quantum states , we also review manifestations of dpts in mesoscopic devices and other systems .
these experiments include observations of resonance - trapping behavior in microwave cavities and open quantum dots , phase lapses in tunneling through single - electron transistors , and spin swapping in atomic ensembles .
other possible manifestations of this phenomenon are presented , including various superradiant phenomena in low - dimensional semiconductors . from these discussions a generic picture of oqss emerges
in which the environmentally - mediated coupling between different quantum states plays a critical role in governing the system behavior .
the ability to control or manipulate this interaction may even lead to new applications in photonics and electronics . |
we study quadrature formulae of the form @xmath0=\sum_{i=1}^{n}a_{i , n}\,f(\tau_{i , n})\,,\quad 0\le \tau_{1,n}<\tau_{2,n}<\cdots<\tau_{n , n}\leq 1\ ] ] for approximate evaluation of the definite integral @xmath1:=\int\limits_{0}^{1}f(x)\,dx\,.\ ] ] our interest is in definite quadrature formulae .
let us recall some definitions .
quadrature formula is said to be _ definite of order @xmath2 _ , @xmath3 , if there exists a real non - zero constant @xmath4 such that its remainder functional admits the representation @xmath5:=i[f]-q_n[f]=c_r(q_n)\,f^{(r)}(\xi)\ ] ] for every @xmath6 $ ] , with some @xmath7 $ ] depending on @xmath8 .
furthermore , @xmath9 is called positive definite ( resp . ,
negative definite ) of order @xmath2 , if @xmath10 ( @xmath11 ) .
obviously , if @xmath9 is a definite quadrature formula of order @xmath2 , then @xmath9 has _ algebraic degree of precision _
@xmath12 ( in short , @xmath13 ) , i.e. , @xmath14=0 $ ] whenever @xmath8 is an algebraic polynomial of degree at most @xmath12 , and @xmath15\ne 0 $ ] . throughout this paper , by @xmath2-convex ( @xmath2-concave
) function @xmath8 we shall mean a function @xmath6 $ ] such that @xmath16 ( @xmath17 ) on the interval @xmath18 $ ] . the importance of definite quadrature formulae of order @xmath2 lies in the one - sided approximation they provide for @xmath19 $ ] when the integrand @xmath8 is @xmath2-convex ( concave ) .
if , e.g. , @xmath20 is a pair of a positive and a negative definite quadrature formula of order @xmath2 and @xmath21 is @xmath2-convex , then for the true value of @xmath19 $ ] we have the inclusion @xmath22\leq i[f]\leq q^{-}[f]$ ] .
this simple observation serves as a base for derivation of a posteriori error estimates and rules for termination of calculations ( stopping rules ) in automatic numerical integration algorithms ( see @xcite for a survey ) .
most of quadratures used in practice ( e.g. , quadrature formulae of gauss , radau , lobatto , newton - cotes ) are definite of certain order .
definite @xmath23-point quadrature formulae with smallest positive or largest negative error constant are called optimal definite quadrature formulae .
let us set @xmath24 it should be pointed out that it is fairly not obvious that the above infimums are attained or that the optimal definite quadrature formulae are unique .
the existence of optimal definite quadrature formulae was first proven by schmeisser @xcite for even @xmath2 , and for arbitrary @xmath2 and more general boundary conditions by jetter @xcite and lange @xcite .
the uniqueness has been proven by lange @xcite . for even @xmath2 , lange @xcite
has shown that @xmath25 for @xmath26 , where @xmath27 is the @xmath2-th bernoulli polynomial with leading coefficient @xmath28 .
schmeisser @xcite proved that the same result holds for optimal definite quadrature formulae with equidistant nodes .
the @xmath23-point optimal positive definite and the @xmath29-point optimal negative definite quadrature formulae of order @xmath30 are well - known : these are the @xmath23-th compound midpoint and trapezium quadrature formulae , respectively .
the case @xmath31 is exceptional , as for @xmath32 the optimal definite quadrature formulae are not known .
lange @xcite has computed numerically , for @xmath33 , the @xmath23-point optimal definite quadrature formulae of order @xmath34 and the @xmath23-point optimal positive definite quadrature formulae of order @xmath35 .
it is a general observation about the optimality concept in quadratures that , even though the existence and the uniqueness of the optimal quadrature formulae ( for instance , in the non - periodic sobolev classes of functions ) is established , the optimal quadrature formulae remain unknown . this fact severely reduces the practical importance of optimal quadratures .
the way out of this situation is to look for quadrature formulae which are nearly optimal , e.g. , for sequences of asymptotically optimal quadrature formulae .
let @xmath36 be a sequence of positive ( resp , negative ) definite quadrature formulae of order @xmath37 .
@xmath38 is said to be _ asymptotically optimal _ positive ( negative ) definite quadrature formula of order @xmath2 , if @xmath39 in @xcite schmeisser proposed an approach for construction of asymptotically optimal definite quadrature formulae of even order @xmath2 with equidistant nodes .
khler and nikolov @xcite have studied gauss - type quadrature formulae associated with spaces of splines with double and equidistant knots , and as a result obtained bounds for the best constants @xmath40 and @xmath41 . in particular , it has been shown in @xcite that for even @xmath2 the corresponding gauss - type quadrature formulae are asymptotically optimal definite quadrature formulae .
motivated by this result , in @xcite nikolov found explicit recurrence formulae for the evaluation of the nodes and the weights of the gaussian formulae for the spaces of cubic splines with double equidistant knots , and proposed a numerical procedure for the construction of the lobatto quadrature formulae for the same spaces of splines . according to @xcite , the gauss and the lobatto quadrature formulae for these spaces of splines are respectively asymptotically optimal positive definite and asymptotically optimal negative definite , of order @xmath35 .
although the evaluation of gauss - type quadrature formulae for spaces of splines ( also with single knots , because of their asymptotical optimality in certain sobolev classes , see @xcite ) is highly desirable , there is a serious problem occurring already with the splines of degree @xmath34 , and its difficulty increases with the splines degree : the mutual displacement of the nodes of the quadratures and the splines knots is unknown . for justifying the location of the quadrature abscissae with respect to the knots of the space of splines , additional assumptions are to be made .
for instance , in a recent paper @xcite ait - haddou , barto and calo extended the procedure from @xcite for explicit evaluation of the gaussian quadrature formulae for spaces of @xmath42 cubic splines with non - equidistant knots , assumming that the spline knots are symmetrically stretched . for another approach to the construction of gaussian quadrature formulae for @xmath43 cubic splines via homotopy continuation , see @xcite . in the present paper
we construct several sequences of asymptotically optimal definite quadrature formulae of order @xmath35 with explicitly given nodes and weights . for their construction
we make use of the euler - maclaurin summation formulae , associated with the midpoint and the trapezium quadrature formulae , replacing the values of the derivatives at the end - points by appropriate formulae for numerical differentiation ( this idea is not new , it can be traced in the book of brass @xcite , and , implicitly , has been applied already in @xcite ) .
thus , our quadrature formulae differ from the compound midpoint or compound trapezium quadrature formula by very little : they have only few different weights and/or involve few additional nodes .
we evaluate the error constants of our quadrature formulae , which , in view of their asymptotical optimality , are not essentially different .
our motivation for proposing not just two sequences of asymptotically optimal positive definite and negative definite quadrature formulae of order @xmath35 is that , when chosen appropriately , pairs of definite quadrature formulae of the same type furnish , similarly to the case of pairs of definite quadrature formulae of opposite type , error inclusions for @xmath19 $ ] whenever the integrand @xmath8 is @xmath35-convex or @xmath35-concave .
the rest of the paper is organized as follows .
section 2 provides the necessary facts about peano representation theorem for linear functionals , bernoulli polynomials and euler - maclaurin summation formulae . in section
3 we construct sequences of definite quadrature formulae of order @xmath35 . in section
4 we prove monotonicity of the remainders of some of our definite quadrature formulae under the assumption that the integrand is @xmath35-convex ( concave ) , and as a result obtain a posteriori error estimates . section 5 shows some numerical experiments , and section 6 contains some final remarks .
throughout the paper , @xmath44 will stand for the set of algebraic polynomials of degree not exceeding @xmath45 . by @xmath46 $ ] , @xmath3
, we denote the sobolev class of functions @xmath47:=\{f\in c^{r-1}[0,1]\,:\ , f^{(r-1 ) } \mbox { abs . continuous},\ ; \int_{0}^{1}\!|f^{(r)}(t)|\,dt<\infty\}\,.\ ] ] in particular , we have @xmath48\subset w^{r}_1[0,1]$ ] .
if @xmath49 is a linear functional defined in @xmath50 $ ] which vanishes on @xmath51 , then , by a classical result of peano @xcite , @xmath49 admits the integral representation @xmath52=\int_{0}^{1}k_r(t)f^{(r)}(t)\,dt,\qquad \quad k_r(t)={\cal l}\big[\frac{(\cdot - t)_{+}^{r-1}}{(r-1)!}\big],\ \ t\in [ 0,1]\,,\ ] ] where @xmath53 the function @xmath54 is called _ the @xmath2-th peano kernel of _ @xmath49 .
in the case when @xmath49 is the remainder @xmath55\,$ ] of the quadrature formula and @xmath56 , @xmath57 is also referred to as _
the @xmath2-th peano kernel of @xmath9_. an explicit representations of @xmath58 for @xmath59 $ ] is @xmath60.\ ] ] @xmath61 is called also _ a monospline _ of degree @xmath2 . from @xmath62=\int\limits_{0}^{1}k_r(q_n;t)\,f^{(r)}(t)\,dt\ ] ] it is clear that @xmath9 is a positive ( negative ) definite quadrature formula of order @xmath2 if and only if @xmath13 and @xmath63 ( resp .
, @xmath64 ) on @xmath18 $ ] . for more details on the peano kernel theory
we refer to @xcite . by appropriate integration by parts in the remainder of a quadrature formula @xmath9 with @xmath65
can be further expanded in the form @xmath66=\sum_{\nu = r}^{s}c_{\nu}(0)\,\big[f^{(\nu-1)}(1)-f^{(\nu-1)}(0)\big ] + \int\limits_{0}^{1}c_s(t)\,f^{(s)}(t)\,dt\,,\ ] ] with some functions @xmath67 depending on @xmath9 ( see @xcite for details ) . for
the sake of convenience , let us fix some notations . for @xmath68
, we set @xmath69 the @xmath23-th compound trapezium and midpoint quadrature formulae are denoted by @xmath70 and @xmath71 , respectively , i.e. , @xmath72:=\frac{1}{2n}\big[f(x_{0,n})+f(x_{n , n})\big]+\frac{1}{n}\ , \sum_{k=1}^{n-1}f(x_{k , n})\,,\qquad q_n^{mi}[f]:=\frac{1}{n}\,\sum_{k=1}^{n}f(y_{k , n})\,.\ ] ] the bernoulli polynomials @xmath73 are defined recursively by @xmath74 here , we shall need the explicit form of @xmath75 , @xmath76 , and shall exploit the fact that @xmath77\,.\ ] ] the @xmath78-periodic extension of @xmath79 on @xmath80 is denoted by @xmath81 and is called _ bernoulli monospline_.
the expansion with @xmath82 and @xmath83 yields the so - called euler - maclaurin summation formulae ( see , e.g. , ( * ? ? ?
* satz 98 , 99 ) ) . for easier further reference ,
they are given in a lemma : [ l1 ] assume that @xmath84 , where @xmath85
. then @xmath86\!=\!-\!\sum_{\nu=1}^{[\frac{s}{2}]}\
! \frac{b_{2\nu}(1/2)}{n^{2\nu}}\big[f^{(2\nu\!-\!1)}(1)\!-\!f^{(2\nu\!-\!1)}(0)\big ] \!+\!\frac{(-1)^{s}}{n^{s}}\!\int\limits_0^{1}\!\ ! \widetilde{b}_s\big(nx-\frac{1}{2}\big)f^{(s)}(x)dx\ ] ] and @xmath87=-\!\sum_{\nu=1}^{[\frac{s}{2 } ] } \frac{b_{2\nu}(0)}{n^{2\nu}}\big[f^{(2\nu\!-\!1)}(1)\!-\!f^{(2\nu\!-\!1)}(0)\big ] \!+\!\frac{(-1)^{s}}{n^{s}}\!\int\limits_0^{1}\!\widetilde{b}_s(nx)f^{(s)}(x)dx\,.\ ] ]
for verifying the asymptotical optimality of the definite quadrature formulae constructed in this section , we note that with @xmath88 reads as @xmath89 [ d3 ] for a given @xmath90 , @xmath91 , we denote by @xmath92\,$ ] and @xmath93\,$ ] the interpolatory formulae for numerical differentiation with nodes @xmath94 , which approximate @xmath95 and @xmath96 , respectively .
the formulae approximating @xmath97 and @xmath98 and obtained by reflection are denoted by @xmath99\,$ ] and @xmath100\,$ ] , i.e. , @xmath101=d_k(\mathbf{t})[g]\,,\qquad g(x)=-f(1-x)\,,\quad k=1,3\,.\ ] ] for the sake of brevity , we write @xmath102 for the collection of four formulae for numerical differentiation @xmath103,\,\widetilde{d}_1(\mathbf{t}),\ , d_3[\mathbf{t}],\,\widetilde{d}_3(\mathbf{t})\}$ ] . the second formula in lemma [ l1 ] with @xmath104 can be rewritten in the form @xmath105
-\frac{1}{12\,n^2}\big[f^{\prime}(1)-f^{\prime}(0)\big]+ \frac{1}{384\,n^4}\big[f^{\prime\prime\prime}(1)-f^{\prime\prime\prime}(0)\big]\\ & + \frac{1}{n^{4}}\int\limits_0^{1}\big[\widetilde{b}_4(nx)-b_4(1/2)\big]\,f^{(4)}(x)\,dx\\ = : & q_n^{\prime}[f]+ \frac{1}{n^{4}}\int\limits_0^{1}\big[\widetilde{b}_4(nx)-b_4(1/2)\big]\,f^{(4)}(x)\,dx\ , .
\end{split}\ ] ] clearly , @xmath106 is a negative definite quadrature formula of order @xmath35 , since , in view of , @xmath107\leq 0 $ ] .
however , @xmath106 is not of the desired form , as it involves derivatives of the integrand .
therefore , we choose a set @xmath102 of formulae for numerical differentiation to replace the values of @xmath108 and @xmath109 in @xmath106 , and thus to obtain a ( symmetric ) quadrature formula @xmath110 which involves at most @xmath111 nodes in addition to @xmath112 .
( in the sequel , we shall refer to @xmath113 to as a _ quadrature formula generated by @xmath102_. ) we have @xmath114=r[q_n^{\prime};f]+\frac{1}{12\,n^2}\,(l_1[f]-\widetilde{l}_1[f])- \frac{1}{384\,n^{4}}\,(l_3[f]-\widetilde{l}_3[f])\,,\ ] ] where @xmath115:=f^{(k)}(0)-d_k(\mathbf{t})[f]\,,\quad \widetilde{l}_k[f]:=f^{(k)}(1)-\widetilde{d}_k(\mathbf{t})[f]\,,\qquad k=1,\,3\,.\ ] ] the linear functionals @xmath116 and @xmath117 vanish on @xmath118 , hence @xmath119 $ ] vanishes on @xmath120 , too . from the definition of the peano kernels it is readily seen that @xmath121\,,\qquad k_4(\widetilde{l}_k;t)\equiv 0,\
\ t\in [ 0,1-t_4)\ , , \quad k=1,\,3\,.\ ] ] this implies the following important observation : [ p1 ] the fourth peano kernel of the symmetric quadrature formula @xmath113 generated by @xmath102 through satisfies @xmath122,\qquad x\in [ t_4,1-t_4]\,.\ ] ] as a consequence , @xmath123 is negative definite of order @xmath35 if and only if @xmath124 for @xmath125 . it should be pointed out that not every set @xmath102 of formulae for numerical differentiation generates a definite quadrature formula of order @xmath35 .
our first application of the above approach reveals a known result . the following @xmath29-point ( with @xmath126 )
asymptotically optimal negative definite quadrature formula of order @xmath35 was obtained in ( * ? ? ?
( 43 ) ) : @xmath127=&\frac{403}{1152n}\big[f(x_{0,n})\!+\!f(x_{n , n})\big]\!+\ !
\frac{159}{128n}\big[f(x_{1,n})\!+\!f(x_{n-1,n})\big]\\ & \!+\!\frac{113}{128n}\big[f(x_{2,n})\!+\!f(x_{n-2,n})\big]\!+\ ! \frac{1181}{1152n}\big[f(x_{3,n})\!+\!f(x_{n-3,n})\big ] \!+\!\frac{1}{n}\sum_{k=4}^{n-4}f(x_{k , n})\ , , \end{split}\ ] ] with an error constant @xmath128 schmeisser s quadrature formula is generated by @xmath129 .
as this is a known result , we do not enter into details .
for @xmath131 we have @xmath132=\frac{n}{2}\big[-11f(x_{0,n } ) + 18f(x_{1,3n})-9f(x_{2,3n})+2f(x_{1,n})\big]\,,\ ] ] @xmath133=27\,n^{3}\big[-f(x_{0,n } ) + 3f(x_{1,3n})-3f(x_{2,3n})+f(x_{1,n})\big]\,,\ ] ] and @xmath134 , @xmath135 are obtained from @xmath136 and @xmath137 by reflection . by @xmath102 generates a symmetric @xmath138-point quadrature formula @xmath139=\sum_{k=1}^{n+5}a_{k , n+5 } f(\tau_{k , n})\,,\ ] ] with nodes @xmath140 given by @xmath141 and weights @xmath142 ( left ) , and @xmath143 , @xmath144 ( right).,title="fig:",scaledwidth=47.0% ] ( left ) , and @xmath143 , @xmath144 ( right).,title="fig:",scaledwidth=47.0% ] in view of proposition , to verify that @xmath145 is a negative definite quadrature formula of order @xmath35 , we only have to show that @xmath146 for @xmath147 . by substituting @xmath148 , @xmath149
, this task reduces to @xmath150 the verification is straightforward , and we omit it .
the graphs of @xmath151 and @xmath143 , with @xmath152 , are depicted on figure 1 .
for the evaluation of the error constant @xmath153 $ ] , we make use of proposition [ p1 ] and the fact that @xmath154 is symmetric , hence @xmath155dt\\ & = \frac{1}{12\,n^{5}}\int_{0}^{1}\phi_1(u)\,du -\frac{b_4(1/2)}{n^{4}}\big(1-\frac{2}{n}\big)\ , . \end{split}\ ] ] a further calculation shows that @xmath156=-71/4320 $ ] and @xmath157 with this set of formulae for numerical differentiation we get through an @xmath159-point quadrature formula @xmath160=\sum_{k=1}^{n+3}a_{k , n } f(\tau_{k , n+3})\ ] ] with nodes @xmath161 and weights @xmath162 in view of proposition [ p1 ] , @xmath163 is a negative definite of order @xmath35 if and only if @xmath164 for @xmath165 . by change of variable @xmath148 ,
@xmath166 , this condition becomes @xmath167 and it is not difficult to verify that it is fulfilled . for the error constant @xmath168 $ ] we have @xmath169dt\\ & = \frac{1}{12\,n^{5}}\int_{0}^{2}\phi_2(u)\,du -\frac{b_4(1/2)}{n^{4}}\big(1-\frac{4}{n}\big)\ , , \end{split}\ ] ] and after evaluation of the integral of @xmath170 , we obtain @xmath171 we rewrite the first euler - maclaurin summation formula in lemma [ l1 ] with @xmath104 in the form @xmath172\ ! + \!\frac{1}{24\,n^2}\big[f^{\prime}(1)\!-\!f^{\prime}(0)\big]\!+\ ! \frac{1}{n^{4}}\!\int\limits_0^{1}\!\ ! \big[\widetilde{b}_4(nx-\frac{1}{2})-b_4(\frac{1}{2})\big]\,f^{(4)}(x)dx\\ & \!=:q_n^{\prime\prime}[f]+\frac{1}{n^{4}}\!\int\limits_0^{1}\!\ !
\big[\widetilde{b}_4(nx-\frac{1}{2})-b_4(\frac{1}{2})\big]\,f^{(4)}(x)dx\ , .
\end{split}\ ] ] here , @xmath173 is a negative definite quadrature formula of order @xmath35 , as , by , its fourth peano kernel @xmath174\,$ ] is non - positive . since @xmath173 is not of the desired form , we choose a set @xmath102 of formulae for numerical differentiation to replace the values of @xmath108 in @xmath175 $ ] , and thus to obtain a ( symmetric ) quadrature formula @xmath176 which involves at most @xmath111 nodes in addition to @xmath177 . by the same argument that led us to proposition [ p1 ] , here we have [ p2 ] the fourth peano kernel of the symmetric quadrature formula @xmath113 generated by @xmath102 through satisfies @xmath178\ , , \qquad x\in [ t_4,1-t_4]\,.\ ] ] consequently , @xmath123 is negative definite of order @xmath35 if and only if @xmath124 for @xmath125 .
for @xmath180 we have @xmath181=\frac{n}{3 } \bigg[-13f(x_{0,n})+36f(y_{1,n})-32f(x_{3,4n})+9f(x_{1,n})\bigg],\ ] ] and @xmath102 generates through a @xmath182-point quadrature formula @xmath183=\sum_{k=1}^{n+6}a_{k , n+6}\,f(\tau_{k , n+6})\ ] ] with nodes @xmath184 and weights @xmath185 by proposition [ p2 ] , to verify that @xmath186 is a negative definite quadrature formula of order @xmath35 , we only need to check whether @xmath187 @xmath188 , which , after the change of variable @xmath148 , becomes @xmath189 ( left ) , and @xmath190 , @xmath191 ( right).,title="fig:",scaledwidth=47.0% ] ( left ) , and @xmath190 , @xmath191 ( right).,title="fig:",scaledwidth=47.0% ] the latter condition is fulfilled , as is seen also on fugure [ fig:2 ] ( left ) . for the error constant @xmath192 , in view of proposition [ p2 ] ,
we have @xmath193dt\\ & = \frac{1}{12\,n^{5}}\int_{0}^{1}\phi_3(u)\,du -\frac{b_4(1/2)}{n^{4}}\big(1-\frac{2}{n}\big)\ , . \end{split}\ ] ] with further calculations we find @xmath194=-13/960 $ ] and @xmath195 below we give two further negative definite quadrature formulae generated through .
for @xmath197 we have @xmath181=\frac{n}{3 } \bigg[-21f(x_{0,n})+32f(y_{1,2n})-12f(y_{1,n})+f(x_{1,n})\bigg],\ ] ] and @xmath102 generates through another @xmath182-point quadrature formula @xmath183=\sum_{k=1}^{n+6}a_{k , n+6}\,f(\tau_{k , n+6})\ ] ] with nodes @xmath198 and weights @xmath199 the error constant of @xmath186 is @xmath200 with @xmath202 , @xmath203 generates a negative definite of order @xmath204 @xmath205-point quadrature formula @xmath206=\sum_{k=1}^{n+8}a_{k , n+8}\,f(\tau_{k , n+8})\ ] ] with nodes , weights and error constant given by @xmath207 @xmath208 @xmath209 clearly , the error constant @xmath210 is inferior to those of the preceding two quadrature formulae , which moreover involve two nodes less .
the reason for quoting this quadrature formula will become clear in section 4 . we rewrite the second euler - maclaurin summation formula in lemma [ l1 ] with @xmath104 in the form @xmath211 -\frac{1}{12\,n^2}\big[f^{\prime}(1)-f^{\prime}(0)\big ] + \frac{1}{n^{4}}\int\limits_0^{1 } \big[\widetilde{b}_4(nx)-b_4(0)\big]\,f^{(4)}(x)\,dx\\ = : & \widetilde{q}_n^{\prime}[f]+\frac{1}{n^{4}}\int\limits_0^{1 } \big[\widetilde{b}_4(nx)-b_4(0)\big]\,f^{(4)}(x)\,dx\ , .
\end{split}\ ] ] by , @xmath212 is a positive definite quadrature formula of order @xmath35 , and we choose a set of formulae for numerical differentiation @xmath102 to approximate @xmath95 and @xmath97 in @xmath212 , thus arriving at a new quadrature formula @xmath213 which involves at most @xmath111 nodes in addition to @xmath112 .
a statement analogous to propositions [ p1 ] , [ p2 ] holds true : [ p3 ] the fourth peano kernel of the symmetric quadrature formula @xmath113 generated by @xmath102 through satisfies @xmath214\ , , \qquad x\in [ t_4,1-t_4]\,.\ ] ] consequently , @xmath123 is positive definite of order @xmath35 if and only if @xmath215 for @xmath125 . below we construct three positive definite quadrature formulae generated through by different sets @xmath102 of formulae for numerical differentiation . with @xmath217
, @xmath102 generates through an @xmath218-point symmetric quadrature formula @xmath219=\sum_{k=1}^{n+7}a_{k , n+7}\,f(\tau_{k , n+7})\ ] ] with nodes and weights given by @xmath220 @xmath221 by proposition [ p3 ] , the verification that @xmath222 is positive definite of order @xmath35 reduces to @xmath223 for @xmath224 , which , after the change of variable @xmath148 , @xmath225 , becomes @xmath226 ( left ) , and @xmath227 , @xmath191 ( right).,title="fig:",scaledwidth=47.0% ] ( left ) , and @xmath227 , @xmath191 ( right).,title="fig:",scaledwidth=47.0% ] the graph of @xmath228 , depicted on fugure [ fig:3 ] ( left ) , shows that , indeed @xmath229 for @xmath225 . finally , in view of proposition [ p3 ] , for the error constant @xmath230 we have @xmath231dt\\ & = \frac{1}{12\,n^{5}}\int_{0}^{1/2}\psi(u)\,du -\frac{b_4(0)}{n^{4}}\big(1-\frac{1}{n}\big)\ , .
\end{split}\ ] ] the integral of @xmath228 is equal to @xmath232 , and a further simplification implies that @xmath233 the next two quadrature formulae are obtained through the same scheme .
we only give their nodes , weights and error constants , skipping the details on the verification of their definiteness and the calculations , as these go along the same lines as in the case we just considered .
with @xmath234 , @xmath102 generates through an @xmath138-point symmetric quadrature formula @xmath139=\sum_{k=1}^{n+5}a_{k , n+5}\,f(\tau_{k , n+5})\ ] ] with nodes and weights given by @xmath235 @xmath236 the error constant of @xmath145 is @xmath237 with @xmath239 , @xmath102 generates through an @xmath218-point symmetric quadrature formula @xmath219=\sum_{k=1}^{n+7}a_{k , n+7}\,f(\tau_{k , n+7})\ ] ] with nodes , weights and error constant given by @xmath240 @xmath241 @xmath242 we write the first formula in lemma [ l1 ] with @xmath104 in the form @xmath243+\frac{1}{24\,n^2}\big[f^{\prime}(1)-f^{\prime}(0)\big]- \frac{1}{384\,n^4}\big[f^{\prime\prime\prime}(1)-f^{\prime\prime\prime}(0)\big]\\ & + \frac{1}{n^{4}}\int\limits_{0}^{1 } \big[\widetilde{b}_4\big(nx-\frac{1}{2}\big)-b_4(0)\big]\,f^{(4)}(x)\,dx\\ = : & \widetilde{q}_n^{\prime\prime}[f]+ \frac{1}{n^{4}}\int\limits_{0}^{1 } \big[\widetilde{b}_4\big(nx-\frac{1}{2}\big)-b_4(0)\big]\,f^{(4)}(x)\,dx\ , .
\end{split}\ ] ] we choose a set @xmath102 for approximating the derivatives values in @xmath244 , thus obtaining a quadrature formula @xmath245 which involves at most @xmath111 nodes in addition to @xmath246 .
we have [ p4 ] the fourth peano kernel of the symmetric quadrature formula @xmath113 generated by @xmath102 through satisfies @xmath247\ , , \qquad x\in [ t_4,1-t_4]\,.\ ] ] consequently , @xmath123 is positive definite of order @xmath35 if and only if @xmath215 for @xmath125 . on using proposition [ p4 ] ,
we verify the definiteness and evaluate the error constant of @xmath113 .
we give below two positive definite quadrature formulae of order @xmath35 , constructed on the basis of . as the definiteness verification and the evaluation of the error constants are completely analogous to that in the preceding cases , they are skipped here .
with @xmath249 , @xmath250 , we obtain through an @xmath251-point symmetric quadrature formula @xmath252=\sum_{k=1}^{n+2}a_{k , n+2}\,f(\tau_{k , n+2})\,,\ ] ] which is positive definite of order @xmath35 .
the nodes and the weights of @xmath253 are @xmath254 @xmath255 the error constant of @xmath253 is @xmath256 with @xmath217 , @xmath257 , we obtain through an @xmath182-point positive definite of order @xmath35 quadrature formula @xmath183=\sum_{k=1}^{n+6}a_{k , n+6}\,f(\tau_{k , n+6})\,,\ ] ] with nodes , weights and error constant given by @xmath258 @xmath259 @xmath260
in this section we shall exploit the following general observation about definite quadrature formulae .
[ t1 ] let @xmath261 be a pair of positive ( negative ) definite quadrature formulae of order @xmath2 .
assume that , for some @xmath262 , the quadrature formula @xmath263 is negative ( positive ) definite of order @xmath2 .
then the following inequalities hold true whenever @xmath21 is an @xmath37-convex or @xmath2-concave function :
@xmath264|\leq\frac{c}{c+1}\,|r[q^{\prime\prime};f]|}$ ] ; 2 .
@xmath264|\leq c\,|q^{\prime}[f]-q^{\prime\prime}[f]|}$ ] ; 3 .
@xmath265|\leq ( c+1)\,|q^{\prime}[f]-q^{\prime\prime}[f]|}$ ] .
let us consider , e.g. , the case when @xmath266 and @xmath267 are negative definite and @xmath268 is positive definite , of order @xmath2 . without loss of generality
we may assume that @xmath8 is @xmath2-convex
. then @xmath269\leq 0 $ ] , @xmath270\leq 0 $ ] , and @xmath271\geq 0 $ ] , therefore @xmath272=(c+1)\,r[q^{\prime};f]-c\,r[q^{\prime\prime};f]\,,\ ] ] and hence @xmath273\leq-\frac{c}{c+1}\,(r[q^{\prime\prime};f])\,,\ ] ] which , in this case , is exactly claim ( i ) of theorem [ t1 ] . claim ( iii ) follows from @xmath274|-|r[q^{\prime};f]|\\ & \geq |r[q^{\prime\prime};f]|-\frac{c}{c+1}\,|r[q^{\prime\prime};f]| = \frac{1}{c+1}\,|r[q^{\prime\prime};f]|\ , , \end{split}\ ] ] and ( ii ) is a consequence of ( iii ) and ( i ) .
the proof of the case when @xmath266 and @xmath267 are positive definite and @xmath268 is negative definite of order @xmath2 is analogous , and we omit it .
[ r1 ] notice the non - symmetric roles of @xmath266 and @xmath267 in theorem [ t1 ] .
part ( i ) implies that for @xmath2-convex ( concave ) integrand @xmath8 , @xmath275\,$ ] furnishes a better approximation to @xmath19 $ ] than @xmath276 $ ] .
another observation is that , the smaller @xmath262 , the better a posteriori error estimates ( ii ) and ( iii ) we get .
hence , it makes sense to search for the best possible ( i.e. , the smallest ) @xmath277 for which @xmath278 is definite with the opposite type of definiteness to those of @xmath266 and @xmath267 . if @xmath279 , then , since @xmath280 , the assumptions of theorem [ t1 ] are fulfilled with @xmath31 and @xmath281 .
hence , for @xmath8 convex , we have the ( well - known ) inequalities : @xmath282|\leq \frac{1}{2}\,|r[q_{n}^{tr}[f]|}$ ] , @xmath283|\leq |q_{n}^{tr}[f]-q_{2n}^{tr}[f]|}$ ] , and @xmath284|\leq 2\,|q_{n}^{tr}[f]-q_{2n}^{tr}[f]|}$ ] .
theorem [ t1 ] is applicable to some pairs of the definite quadrature formulae of order @xmath35 , obtained in section 3 . in tables 1 and 2 below
, the notation @xmath285 stands for the quadrature , given in section 3.b.c , with a parameter @xmath286 .
[ t2 ] the assumptions of theorem [ t1 ] are fulfilled for the pairs @xmath287 of negative definite quadrature formulae and with the best possible constants @xmath288 , given in table 1 .
[ tab:1 ] .pairs @xmath289 of negative quadrature formulae of order @xmath35 and the corresponding best constants @xmath288 , satisfying the assumptions of theorem [ t1 ] . [ cols="^,^,^,^",options="header " , ] the values @xmath290 of @xmath23 in table 4 correspond to @xmath291 nodes used in total by @xmath266 and @xmath267 , and also to the values @xmath292 in ( * ? ? ?
* table 2 ) .
as is seen , the approximation error @xmath293 there and in table 4 behaves similarly .
* 1 . * in @xcite schmeisser proposed two sequences of asymptotically optimal positive definite quadrature formulae of order @xmath35 , which are of open type , i.e. , do not involve evaluations of the integrand at the end - points .
it is worth noticing that these quadrature formulae can be obtained via with a slight modification of @xmath294 ( and its reflected variant @xmath295 ) .
namely , formulae ( 45 ) and ( 47 ) in @xcite are obtained through with @xmath296=d_1(x_{0,n},x_{1,n},x_{2,n},x_{3,n},x_{4,n})[f]$ ] and @xmath296=d_1(x_{0,n},y_{1,n},x_{1,n},x_{2,n},x_{3,n})[f]$ ] , respectively . here ,
@xmath297 $ ] stands for the five - point formula approximating @xmath95 with nodes @xmath298 and with a _ fixed _ coefficient , equal to @xmath299 , in front of @xmath300 . with @xmath301=-6n\,f(x_{0,n})+\frac{46n}{3}\,f(y_{1,n})-17n\,f(x_{1,n})+ 10n\,f(y_{2,n})-\frac{7n}{3}\,f(x_{2,n})\ ] ] we obtain through an @xmath159-point ( @xmath302 ) positive definite quadrature formula of order @xmath35 , @xmath303=&\frac{23}{18n}\,[f(y_{1,n})+f(y_{n , n } ) ] -\frac{5}{12n}\,[f(x_{1,n})+f(x_{n-1,n})]\\ & \!+\!\frac{5}{6n}[f(y_{2,n})\!+\!f(y_{n-1,n } ) ] \!+\!\frac{29}{36n}[f(x_{2,n})\!+\!f(x_{n-2,n } ) ] \!+\!\frac{1}{n}\sum_{k=3}^{n-3}f(x_{k , n } ) \end{split}\ ] ] with error constant @xmath304 compared to the error constants of quadrature formulae ( 45 ) and ( 47 ) in @xcite when using the same number of nodes , say , @xmath45 , for @xmath305 the error constant of the above quadrature formula is better , i.e. , smaller .
yet , it is worse compared to the error constant of the @xmath45-point gaussian quadrature formula @xmath306 for the space of cubic splines with double equidistant knots , which has been constructed in @xcite and where we have ( see ( * ? ? ?
* corollary 2.3 ) , roughly , @xmath307 * 2 .
* one may wonder why table 1 does not contain pairs of negative definite quadrature formulae of order @xmath35 of the type @xmath308 or @xmath309 .
the reason is that , with the above combinations , quadrature formula @xmath310 can not be positive definite of order @xmath35 with @xmath262 . indeed , in the first case , according to proposition [ p1 ] , away from the neighborhoods of the end - points of @xmath18
$ ] , affected by the formulae for numerical differentiation applied to the construction of @xmath266 and @xmath267 , we have @xmath311 -\frac{c}{(n)^4}\,\big[\widetilde{b}_4(n\,t)-b_4(1/2)\big]\,,\ ] ] and @xmath312 as the first term is negative while the second term vanishes . in the second case , by propositions [ p1 ] and [ p2 ] we have away from the end - points @xmath311 -\frac{c}{(n)^4}\,\big[\widetilde{b}_4(n\,t-1/2)-b_4(1/2)\big]\,,\ ] ] and @xmath313 as the first term is negative while the second term vanishes .
* 3 . * for similar reasons , table
2 can not contain pairs of positive definite quadrature formulae of order @xmath35 of the type @xmath314 or @xmath315 .
indeed , away from the end - points of @xmath18 $ ] , in the first case one can see on the basis of proposition [ p4 ] that @xmath316 , while in the second case propositions [ p3 ] and [ p4 ] imply @xmath317 .
perhaps , the first results on monotonicity of the remainders of quadrature formulae are due to newman @xcite . for conditions for monotonicity of the remainders of quadratures , in particular of the remainders of compound and gauss - type quadratures , in terms of their peano kernels and the resulting exit criteria , we refer the reader to @xcite .
the quadrature formulae constructed here are not of compound type , and the method applied for proving monotonicity of their remainders by virtue of theorem [ t1 ] ( i ) is close to that applied in @xcite , i.e. , relies on the existence of common double zeros of the shifted bernoulli monosplines . * 5 . *
our choice to construct symmetric quadrature formulae here is for reasons of simplicity only ; otherwise , one can apply different formulae for numerical differentiation for approximating the derivatives evaluations at the end - points @xmath318 and @xmath78 , and thus obtaining non - symmetric definite quadrature formulae of order @xmath35 .
needless to say , the approach proposed here is applicable for the construction of definite quadrature formulae of higher order .
r. ait - haddou , m. barto , v.m .
calo , explicit quadrature formulae for @xmath42 cubic splines with symmetrically stretched knot sequences , j. comp .
( 2015 ) 543552 . | we construct several sequences of asymptotically optimal definite quadrature formulae of fourth order and evaluate their error constants . besides the asymptotical optimality , an advantage of our quadrature formulae is the explicit form of their weights and nodes . for the remainders of our quadrature formulae monotonicity properties
are established when the integrand is a 4-convex function , and a - posteriori error estimates are proven .
asymptotically optimal definite quadrature formulae , peano kernel representation , euler - maclaurin type summation formulae , a posteriori error estimates 41a55,65d30,65d32 |
the luminosity of colliding storage rings is one of the most important criteria of performance . in order to achieve high luminosity
, the colliding beams need to be aligned precisely at the collision point in all three dimensions . for a symmetric collider ,
the alignment of the electron and positron beams is ensured automatically by the symmetries of charge conjugation and time - reversal . since
an asymmetric collider consists of two different storage rings , precise alignment of the two beams at the collision can only be achieved through tuning of the two rings . in this paper ,
our goal is to show some effects on the luminosity when the two beams are not aligned well in transverse planes due to the coupling .
it was shown by edwards and teng@xcite that a two - dimensional coupled linear motion in a periodic and symplectic system can be parameterized with ten independent parameters as a block diagonalization of a one - turn matrix @xmath0 where @xmath1 is the rotation matrix and @xmath2 defines the symplectic transformation from the normalized coordinates to the physical coordinates .
these matrices can be further decomposed into @xmath3 and @xmath4 where @xmath5 is 2@xmath62 identity matrix and @xmath7 are 2@xmath62 symplectic matrices with a determinant of unity .
the angle @xmath8 is called coupling angle . here
we denote the bar as two - dimensional symplectic conjugate @xmath9 where @xmath10 is 2@xmath62 unit symplectic matrix @xmath11 recently , it is pointed out by sagan and rubin@xcite that there exists another solution of parameterization @xmath12 where @xmath13 . in the case of a strongly coupled lattice , for example , the interaction region of the low energy ring(ler ) of pep - ii , both solutions are needed for a complete parameterization of the region . here
we choose @xmath8 as a parameter of linear coupling because it has two very important properties : system is decoupled when @xmath14 and @xmath8 can be changed only by a skew quadrupole or solenoid as shown in fig .
[ fig : coupling ] from which the locations of the skew quadrupoles and solenoid are clearly seen .
in the normalized coordinate , the sigma matrix can be written as @xmath15 where @xmath16 and @xmath17 are the equilibrium emittances in the eigen planes ignoring the off - diagonal terms at the order of the damping increment per turn .
the sigma matrix in the physical coordinate is obtained with the transformation @xmath18 the corresponding equilibrium gaussian distribution is @xmath19 where @xmath20 is the vector of canonical coordinates . the beam profile in the configuration @xmath21 and @xmath22
is derived by integrating the canonical momentum @xmath23 and @xmath24 @xmath25 the result of the integration is @xmath26 where @xmath27 is the vector of configuration coordinates namely @xmath28 and @xmath29 is the inverse of the sigma matrix @xmath30 moreover , the symmetric matrix @xmath29 can be diagonalized by a rotation transformation as shown in fig .
[ fig : rot ] @xmath31 where @xmath32 is the tilted angle of the beam .
the parameters that describe the beam in two coordinate systems relate with each other as following @xmath33 and @xmath34 \nonumber \\ { 1\over\sigma_\eta^2 } & = & { 1\over2 } [ ( k_{xx } + k_{yy } ) - { k_{xx}-k_{yy } \over\cos2\psi}]\end{aligned}\ ] ] where @xmath35 are the elements of @xmath29 and @xmath36 and @xmath37 are the beam size along the long axis @xmath38 and short axis @xmath39 of the ellipse respectively .
for simplicity , we ignore the effect of a finite bunch length . given the two beam profiles , @xmath40 and @xmath41 , the luminosity can be written as @xmath42 where @xmath43 is the number of the colliding bunches , @xmath44 is the revolution frequency , and @xmath45 are the number of charges in each position and electron bunch respectively . at a low beam current ,
the beam distribution is nearly gaussian . using gaussian as an approximation
, we can evaluate the overlapping integral @xmath46 in order to analyze the result of the calculation , we rewrite the integral in terms of the geometrical parameters , @xmath36 , @xmath37 , and @xmath32 by substituting eq . [ eqn : geom ] into eq .
[ eqn : dets ] @xmath47 where @xmath48 and @xmath49 clearly , the luminosity is at its maximum when @xmath50 which is always the case in a symmetric collider due the symmetries .
when two flat beams have identical size , we have @xmath51 it is easy to see that the reduction of the luminosity due to the difference of the two tilted angles are much enhanced by the aspect ratio @xmath36/@xmath37 of the beams .
the luminosity depends upon all six geometrical parameters @xmath52 , and @xmath53 which describe the size and orientation of the beams .
it is impossible to extract these parameters directly from the luminosity alone . by transversely scanning the beam crossing the other one
, we could extract more constraints among them .
let s calculate the overlapping integral with an offset of centroid of a beam @xmath54 after a lengthy but straightforward algebra , we obtain @xmath55 the result agrees with eq . [
eqn : lumi ] when @xmath56 .
@xmath57 is a 2@xmath62 symmetric matrix that has the following elements @xmath58 where @xmath59 and @xmath60.\ ] ] please note that @xmath61 is again gaussian and is normalized to unity @xmath62 thus , similar to the treatment of the single beam profile , we can diagonalize @xmath57 with a rotation @xmath63 @xmath64 \nonumber \\ { 1\over\sigma_\eta^2 } & = & { 1\over2 } [ ( m_{xx } + m_{yy } ) - { m_{xx}-m_{yy } \over\cos2\psi}],\end{aligned}\ ] ] where we denote @xmath38 and @xmath39 as the principal axes . as an experiment
, we move one beam against the other one horizontally with a closed orbit bump at the collision point and measure the luminosity as a function of the offset .
the luminosity of the scan is proportional to @xmath65 where @xmath66 to simplify the calculation , let s assume that @xmath67 which is a very good approximation when the luminosity is well optimized . for flat beams and small @xmath32 s , we have @xmath68.\ ] ] we can see that @xmath32 makes @xmath69 smaller than the design value and the effect of @xmath32 is much enhanced by the aspect ratio .
similarly , we carry out the calculation for the vertical scan @xmath70 where @xmath32 makes @xmath71 slightly larger but is not enhanced by the aspect ratio as in the horizontal scan .
more general , we can scan the luminosity as a function of both horizontal and vertical offsets in a two - dimensional grid .
the result of the measurement is shown in fig .
[ fig : scan ] as contour plots of the specific luminosity .
the results of the fitting are @xmath72 @xmath73 m , @xmath74 @xmath73 m , and @xmath75 compared with the design values : @xmath76 @xmath73 m , @xmath77 @xmath73 m , and @xmath78 .
we have calculated the degradation of luminosity due to different tilted angles of colliding beams . for same tilted angles ,
the higher the aspect ratio is the more luminosity reduction will be .
furthermore , we have computed a general luminosity formula for off - centered beams .
the formula is used to understand the luminosity scan , especially for the two - dimensional scan .
we would like to thank w. kozanecki , m. minty , j. seeman , and u. wienands for many helpful discussions . | a formula of luminosity of asymmetric e+e- collider with coupled lattices is derived explicitly .
the calculation shows how the tilted angle and aspect ratio of the beams affect the luminosity . knowing the result of the calculation
, we measure the tilted angle of the luminous region of the collision for pep - ii using two dimensional transverse scan of the luminosity .
the method could be applied to correct the coupling at the collision point . |
the fate of heavy quark bound states as a probe of the deconfining properties of the thermal strongly interacting medium has a long history . in the seminal paper ref .
@xcite a suppression of the production rate of these states was predicted , being caused by the shortening of the screening length for color interactions in the quark - gluon plasma . in subsequent analyses
it was realized that the situation is in fact more involved : various effects can enhance or suppress this phenomenon , the final result being the outcome of a complex dynamical process ( see ref .
@xcite for a recent review of the theoretical and experimental aspects ) . here
we consider the consequences that the introduction of a magnetic background field may have on the screening lengths of the thermal medium .
the presence of strong magnetic backgrounds , with field strengths comparable to the qcd scale , is a situation common to many contexts , ranging from cosmology @xcite and non - central heavy ion collisions @xcite , with magnetic fields going up to @xmath3 tesla ( @xmath4 gev@xmath5 ) , to magnetars @xcite , where magnetic fields of the order of @xmath6 tesla are expected on the surface , and possibly higher in the inner core . how the properties of strongly interacting matter are modified by such magnetic fields has been the subject of many theoretical efforts , see refs .
@xcite for recent reviews .
as regards the effects more directly related to color interactions , various studies have considered the possible influence of an external magnetic field on the static quark - antiquark potential @xcite , which has been clarified by recent lattice results @xcite , and might have consequences relevant to the spectrum of heavy quark bound states @xcite . at zero temperature ,
the potential becomes anisotropic and the string tension @xmath7 is larger ( smaller ) in the direction orthogonal ( parallel ) to the magnetic field @xmath8 @xcite ; at finite @xmath9 , in particular in the region right below the pseudocritical temperature @xmath10 , the magnetic field induces a general suppression of @xmath7 @xcite , leading to an early onset of deconfinement , in agreement with the observed dependence of @xmath10 on @xmath2 @xcite in this paper we extend the study to the region of temperatures above @xmath10 , in order to investigate the effects of a magnetic background on the interactions between heavy quarks in the quark - gluon plasma . in this phase ,
the effective interaction is no longer confining and can instead be described by a screened coulomb form , with two different screening lengths / masses characterizing the ( color)electric and the ( color)magnetic sectors . studying the appropriate combinations of polyakov loop correlators by means of lattice qcd simulations at the physical point we estimated the screening lengths for @xmath11 and several values of the external magnetic field ( for @xmath12 ) . from these results
we conclude that the magnetic field induces a reduction of both screening lengths ( i.e. an increasing of the screening masses ) , which might induce a further suppression of the formation of heavy quark bound states in the thermal medium .
the paper is organized as follows . in section
ii we discuss our numerical setup and review the definition of the screening masses in terms of polyakov loop correlators . in section
iii we present our numerical results . finally ,
in section iv , we draw our conclusions .
the screening masses of strongly interacting matter have been historically introduced in perturbation theory by studying the pole structure of the finite temperature gluon propagator .
while this approach presents no difficulties for the computation of the leading order electric screening mass ( see e.g. @xcite ) , it was soon realized that perturbation theory gets into trouble at higher orders , or even at the leading order for the magnetic mass , because of the infrared sensitivity of the obtained expressions @xcite .
the natural way to overcome these difficulties and obtain nonperturbative results for the screening masses is to analyze the large distance behavior of gauge - invariant correlation functions . for this purpose ,
correlators of polyakov loops have been traditionally used @xcite and the two independent correlators that can be studied are @xmath13 where @xmath14 is the polyakov loop operator , which is defined in the continuum by @xmath15 in this expression @xmath16 is the number of colors and the symbol @xmath17 denotes the path - ordered exponential .
the correlator @xmath18 is often studied because of its connection with the free energy @xmath19 of a static quark - antiquark pair , that can be computed using the relation @xcite @xmath20 the study of @xmath21 is thus the finite temperature counterpart of the study of wilson loops at zero temperature , from which the potential energy of a static quark - antiquark pair can be extracted . since in any numerical study
the temporal extent of the lattice is always finite , it would also be possible to extract the static potential as the zero temperature limit of the free energy extracted from polyakov loops , however this second procedure is generically not numerically convenient because of the much larger statistical errors involved . and the real part of the correlator @xmath22 for @xmath23 ( in both cases we considered the connected correlators ) .
only the case @xmath24 is reported for clarity but also in the case of non - vanishing external magnetic field the behaviour of the correlators is qualitatively similar . ] in the following we will need also the somehow less used correlator @xmath25 ; a comparison of the behaviour of @xmath26 and @xmath27 for @xmath23 is shown in fig .
[ fig : corrsllandlldag ] , from which it is clear that the very large distance behavior of these two correlators is the same .
in fact , the two correlators turn out to be substantially different only in the confined phase , where the @xmath25 correlator is strongly suppressed also at short distances because of the confining properties of the medium vanishes exactly at all distances in the confined phase , because of the exact center symmetry . ] . under euclidean - time reversal @xmath28 the color - magnetic and color - electric gluon components , @xmath29 and @xmath30 , are respectively even and odd .
since the polyakov loop transforms under such a transformation as @xmath31 , the following combinations @xmath32 receive contributions only from the magnetic and electric sectors respectively .
these quantities can be further decomposed in eigenstates of the charge conjugation operator @xmath33 : under this symmetry @xmath34 and we thus obtain @xmath35 where the subscripts @xmath36 indicates the @xmath33 eigenvalues .
such a decomposition has been introduced in ref .
@xcite and it has been recently used in some lattice computations ( see e.g. @xcite ) . from the definitions in eq .
it immediately follows that the magnetic odd and the electric even sectors are trivial , i.e. @xmath37 and that the following relations hold true @xmath38 using the previous relations , the correlators needed to investigate the magnetic and electric sectors can be written in the form @xmath39 where the minus sign in the definition of @xmath40 is conventional and in the electric case no disconnected term is present because of the symmetry under charge conjugation .
it is convenient to rewrite the previous correlators in term of @xmath41 and @xmath21 and the final result is @xmath42- \left|\langle \mathrm{tr}l\rangle \right|^2\\ c_{e^-}=&-\frac{1}{2}\mathrm{re}\big[c_{ll}-c_{ll^\dag}\big ] \ ; . \end{aligned}\ ] ] notice that the mixed electric - magnetic correlator @xmath43 is zero by charge conjugation invariance , meaning that the real and imaginary parts of the polyakov loop fluctuate independently .
the electric and magnetic screening masses can now be extracted from the large distance behaviour of the correlators in eqs . .
in particular , in the very high temperature regime , one expects @xcite these quantities to scale as @xmath44 where @xmath45 and @xmath46 and @xmath47 are the electric and magnetic screening masses respectively . in the subsequent analysis
, we will use these expressions to extract the screening masses from the correlators computed on the lattice . in the presence of an external magnetic field
, we obviously have to take into account the explicit breaking of the rotational symmetry : correlators taken along directions parallel or orthogonal to the magnetic field can not be averaged , and two _ a priori _ different electric screening masses and magnetic screening masses have to be defined . in this work we have adopted a discretization of @xmath48 qcd based on the symanzik tree - level improved gauge action and the stout smeared rooted staggered action for the fermionic sector .
the partition function in the presence of a magnetic background @xmath2 is written as @xmath49})^{1/4}}\ , \ ] ] where @xmath50 is the functional integration over the @xmath51 gauge link variables , @xmath52 stands for the tree - level improved action @xcite : @xmath53 where @xmath54 and @xmath55 denote , respectively , the real part of the trace of @xmath56 and @xmath57 loops . finally , the staggered fermion matrix @xmath58 is written in terms of two times stout - smeared gauge links @xmath59
@xcite , with an isotropic smearing parameter @xmath60 , and the @xmath61 parallel transporters @xmath62 , which takes the presence of the external electromagnetic field into account ; in both cases the latin indices denote the position in the lattice and the greek indices denote the direction of the link . for a constant and uniform magnetic background directed along the @xmath63 direction ,
a possible choice of the @xmath61 phases is ( @xmath64 is the fermion charge ) @xmath65 while all the other @xmath61 link variables are set to 1 . for this choice to actually describe a uniform magnetic field on the lattice with periodic boundary conditions , it is necessary for the value of @xmath66 to satisfy the following quantization condition @xcite @xmath67 in our numerical simulations , we used for the bare parameters the values @xmath68 , @xmath69 and @xmath70 , which correspond @xcite to a lattice spacing @xmath71 and to physical values of the pion and strange quark masses ( isospin breaking is neglected ) .
we performed simulations on @xmath72 lattices , with @xmath73 , corresponding to a fixed spatial size of around 5 fm and to temperatures @xmath74 .
polyakov loop correlators have been measured on a set of around @xmath75 configurations for each temperature ( with measures separated by 5 molecular dynamics trajectories ) and , to reduce the statistical noise , a single step of hyp smearing @xcite has been applied to the temporal links , with the parameters of the hyp2-action , see ref .
correlators have been extracted for generic orientations ( i.e. not just along the lattice axes ) at @xmath76 , while in the presence of the background field we have considered separately correlators along the @xmath77 axis ( i.e. parallel to @xmath2 ) and in the whole @xmath78 plane ( i.e. orthogonal to @xmath2 ) , which in the following will be denoted respectively by @xmath79 and @xmath80 .
note that , since polyakov loops renormalize multiplicatively and no further distance - dependent renormalization enters the correlator , the screening masses defined by eq .
do not need any renormalization .
in fig . [ fig : corrs ] , we show an example of the electric and magnetic correlators for @xmath23 at @xmath81 and @xmath82 . at zero magnetic field
all possible orientations are displayed , while for non - zero field only orientations perpendicular or parallel to the magnetic field are considered .
for this reason , the three curves in each panel have different number of points .
the points at zero magnetic field approximately lay a single curve , indicating that the lattice violations of rotational invariance are small .
this holds true also for correlators defined in the plane perpendicular to the magnetic field , indicating that the residual @xmath83 subgroup of @xmath84 left unbroken by the presence of @xmath8 is well realized on the lattice for these correlator , and in short , that lattice artefacts are small .
the external magnetic field is expected to modify the correlators both by changing the screening masses and by inducing anisotropies in the correlators .
[ fig : corrs ] shows that both the electric and the magnetic correlators approach zero faster when an external magnetic field is present , a fact that implies that the screening masses are increasing functions of @xmath2 .
the anisotropy of the correlators is in general not easy to observe in the high temperature regime ( see also @xcite ) , since correlators decay very quickly and it is possible to estimate them with enough relative accuracy only for a short distance ; nevertheless , for the relatively low @xmath9 and high @xmath2 case shown in fig .
[ fig : corrs ] , the anisotropy is present and more pronounced in the magnetic correlator ( whose signal is larger than the electric one ) . in order to determine the screening masses from the correlators , numerical data have been fitted with the functional form in eq . , adopting a bootstrap approach to propagate the correlations between data .
several fit intervals have been investigated in order to assess the stability and reliability of the results and to estimate the systematic uncertainties associated with the fitting procedure . in table
[ tab : datapoints ] we report the numerical values obtained and in figs . [
fig : m_vs_eb]-[fig : m_vs_t ] we show a graphical representation of their behavior as a function of @xmath85 and @xmath9 .
at vanishing magnetic field we reproduced the known behavior of @xmath86 and @xmath87 : the electric screening mass is larger than the magnetic one and the ratios @xmath88 and @xmath89 are remarkably insensitive to the value of the temperature , something that _ a priori _ would have been expected to hold only at much higher temperatures .
our results are in good agreement with the corresponding temperatures and lattice spacing data presented in ref .
@xcite , where the same discretization was used and to which we refer for an in depth discussion of the @xmath24 case .
c|c|c|c|c|c ' '' '' @xmath90}$ ] & @xmath91}$ ] & @xmath92 & @xmath93 & @xmath94 & @xmath95 + 330 & 0.00 & 5.12(18 ) & 5.12(18 ) & 9.07(56 ) & 9.07(56 ) + & 0.26 & 5.16(17 ) & 4.92(16 ) & 8.71(61 ) & 9.50(60 ) + & 0.52 & 5.20(19 ) & 5.01(17 ) & 9.36(74 ) & 9.71(67 ) + & 0.78 & 5.17(19 ) & 5.16(16 ) & 9.15(56 ) & 9.51(66 ) + & 1.04 & 5.41(19 ) & 5.16(15 ) & 9.36(49 ) & 8.46(76 ) + & 1.30 & 5.77(19 ) & 5.32(17 ) & 10.39(60 ) & 9.69(47 ) + 250 & 0.00 & 4.70(17 ) & 4.70(17 ) & 9.54(58 ) & 9.54(58 ) + & 0.26 & 5.11(16 ) & 5.11(17 ) & 9.47(67 ) & 9.33(68 ) + & 0.52 & 5.12(18 ) & 5.22(21 ) & 9.64(57 ) & 9.66(60 ) + & 0.78 & 5.60(16 ) & 5.39(18 ) & 9.85(42 ) & 9.45(49 ) + & 1.04 & 5.98(16 ) & 5.60(18 ) & 10.20(64 ) & 9.57(60 ) + & 1.30 & 6.67(19 ) & 5.84(19 ) & 10.78(72 ) & 10.59(73 ) + 200 & 0.00 & 4.80(22 ) & 4.80(22 ) & 9.65(35 ) & 9.65(35 ) + & 0.26 & 5.61(21 ) & 5.59(20 ) & 9.19(54 ) & 9.84(68 ) + & 0.52 & 6.14(24 ) & 5.61(26 ) & 10.54(43 ) & 9.71(60 ) + & 0.78 & 6.59(18 ) & 6.30(17 ) & 11.55(52 ) & 11.88(61 ) + & 1.04 & 7.18(21 ) & 6.55(24 ) & 13.09(76 ) & 12.57(74 ) + & 1.30 & 7.70(25 ) & 6.93(23 ) & 12.84(55 ) & 12.52(43 ) + as previously anticipated from fig . [ fig : corrs ] , the effect of the magnetic field is to increase both the magnetic and the electric screening mass , as visible in fig .
[ fig : m_vs_eb ] . in both cases
the growth is roughly linear in the magnetic field and with similar slopes .
indeed the ratio @xmath96 turns out to be independent of both the magnetic field intensity and the temperature , as shown in fig .
[ fig : massratio ] .
in particular this means that that in the regime studied in this work the external magnetic field does not change the usual @xmath97 hierarchy . in the case of @xmath87 an anisotropy
is observed , as could have been guessed by fig .
[ fig : corrs ] , with the screening mass relative to the directions orthogonal to the external field being larger than the one in the direction parallel to the field . in the case of the electric screening mass
no such an anisotropy is observed , however this could be due to the fact that data for @xmath86 have larger relative error with respect to the ones for @xmath98 .
the different accuracy of these estimates is a consequence of the relation @xmath99 : magnetic and electric correlators have similar statistical ( absolute ) errors , but the extraction of @xmath86 is made difficult by the rapid decrease of the electric correlator . for different temperatures and external field intensity . ] we now try to determine a functional form that well describes the @xmath9 and @xmath2 dependence of the screening masses . from fig .
[ fig : m_vs_eb ] and fig .
[ fig : m_vs_t ] it follows that the main properties of the screening masses , at least in the explored range of temperatures and magnetic field intensities , are : _ i ) _ at @xmath24 the ratios @xmath89 and @xmath100 are independent of @xmath9 ; + _ ii ) _ for large magnetic field the screening masses grow linearly with @xmath2 . to these properties
, it is reasonable to add the requirement that the screening masses be analytic functions of the magnetic field @xmath8 . from
that it follows that , in the limit of small magnetic field intensity , the behaviors of @xmath101 and @xmath102 have to be quadratic in @xmath85 . in the high temperature phase it also seems reasonable to assume the only relevant dimensional parameters to be the temperature and the magnetic field intensity . before going on we can explicitly check , in a model independent way , that @xmath2 and @xmath9 are the only dimensional quantities that matter to describe the behavior of the screening masses , by showing that the ratios @xmath89 and @xmath100 depend only on the dimensionless combination @xmath103
this is indeed the case , as can be appreciated from fig .
[ fig : m_vs_eb_t2 ] , where the @xmath104 behaviour for small values of the magnetic field is also somehow more clearly visible than in figs [ fig : m_vs_eb]-[fig : m_vs_t ] . a simple ansatz that satisfies all the previous properties is @xmath105\ , \ ] ] where @xmath106 denotes the spatial direction ( i.e. @xmath107 or @xmath108 ) and an analogous expression can be used for the magnetic screening mass .
this functional form is analogous to the one used in ref .
@xcite for the case of the dependence of the chiral condensate on the magnetic field .
the three parameters that enter the ansatz in eq . have simple interpretations : @xmath109 is the @xmath24 value of the ratio @xmath89 ( that is known to be @xmath9-independent ) , @xmath110 is related to the slope of @xmath86 as a function of @xmath2 at fixed temperature for large magnetic field intensities , while @xmath111 is associated with the quadratic small-@xmath2 behavior of @xmath86 . the best fit values for the parameters entering eq . for both @xmath86 and @xmath87
are reported in table [ tab : modelres ] .
while for the magnetic screening masses reasonable values of the @xmath112 test are obtained , for the case of the electric masses the values of @xmath113 are somehow small , indicating that we are using more parameters than needed to fit the data at the current level of statistical accuracy .
indeed a simple linear dependence on @xmath85 is sufficient to describe the data for the electric masses , however this is just a consequence of the large error bars on the electric correlator .
c|c|c|c|c ' '' '' & @xmath114 & @xmath115 & @xmath116 & @xmath117 + ' '' '' @xmath118 & 4.964(82 ) & 0.137(19)@xmath119 & 0.141(55)@xmath120 & 1.06 + ' '' '' @xmath121 & 4.935(79 ) & 0.099(20)@xmath119 & 0.094(49)@xmath120 & 1.10 + ' '' '' @xmath122 & 9.24(21 ) & 0.120(47)@xmath119 & 0.069(38)@xmath120 & 0.63 + ' '' '' @xmath123 & 9.34(20 ) & 0.17(28 ) @xmath119 & 0.039(21)@xmath120 & 0.85 +
in this study we have investigated the effects of a magnetic background on color - screening phenomena taking place in the quark - gluon plasma . to that purpose , we have measured polyakov loop correlators for various temperatures , up to @xmath124 mev , and uniform magnetic fields going up to @xmath125 gev@xmath5 .
our results have been obtained at a single value of the lattice spacing , @xmath126 fm , and a refinement of the analysis , aimed at a continuum limit extrapolation , should be performed in the future .
we have shown that the magnetic field induces an increase of both the magnetic and the electric screening masses and , to some extent , also the appearance of an anisotropy in polyakov loop correlators .
the masses increase linearly with the magnetic field for moderate or large @xmath2 values ( i.e. for @xmath127 ) and a reasonable _ ansatz _ can be given to describe the connection of this regime with the expected quadratic behaviour for small values of @xmath2 , in which both screening masses are proportional to @xmath9 and to a function of @xmath103 .
indeed the influence of the magnetic field is stronger at lower temperatures and asymptotically vanishes in the large @xmath9 limit . on the other hand
, the ratio of the magnetic to electric screening masses turns to be independent of @xmath2 , within errors , with the magnetic screening mass always being smaller than the electric one .
the observed increase of the screening masses as a function of @xmath2 is in qualitative agreement with perturbative computations @xcite and with the behavior already observed below @xmath10 : the magnetic background tends to suppress the confining properties of the thermal medium below @xmath10 , and to enhance the screening of color interactions above @xmath10 . in both cases , one can interpret the effect in terms of the decrease of the pesudo - critical temperature @xmath10 as a function of @xmath2 @xcite , so that in the low temperature phase one approaches deconfinement as @xmath2 increases , while at high temperature one gets farther from the transition and color screening gets stronger by increasing @xmath2 .
following this line of reasoning , no particular critical behavior is expected in the high temperature phase when approaching the large magnetic field limit , since the system will just become more deconfined ( i.e. color interactions will be more and more screened ) ; this is in contrast to what happens below @xmath10 or to what might happen even at @xmath128 @xcite .
the increasing of the screening masses induced by the presence of an external magnetic field could in principle lead to a stronger suppression of heavy quark bound states in peripheral heavy - ion collisions and , more specifically , to a direct relation between suppression and centrality .
however , following the original argument presented in ref .
@xcite , an in deep discussion should consider also the modifications of the radius of the heavy quarks bound states as a function of @xmath2 , since it is the reduction of the screening length with respect to such radius which brings to the actual suppression ; a direct computation of quarkonia spectral functions in the presence of magnetic background is surely something that should be addressed by future lattice studies .
moreover , in order to assess the relevance of our results to heavy ion phenomenology , one should first of all know to which extent the magnetic field produced in non - central heavy ion collisions survives the thermalization process , not to mention all the other dynamical processes that make it difficult to safely predict the fate of heavy quark bound states even in the absence of external magnetic field .
v. skokov , a. y. illarionov and v. toneev , int .
j. mod .
a * 24 * , 5925 ( 2009 ) [ arxiv:0907.1396 [ nucl - th ] ] .
v. voronyuk , v. d. toneev , w. cassing , e. l. bratkovskaya , v. p. konchakovski and s. a. voloshin , phys .
c * 83 * , 054911 ( 2011 ) [ arxiv:1103.4239 [ nucl - th ] ] .
m. hasan , b. chatterjee and b. k. patra , arxiv:1703.10508 [ hep - ph ] .
g. s. bali , f. bruckmann , g. endrodi , z. fodor , s. d. katz , s. krieg , a. schafer and k. k. szabo , jhep * 1202 * , 044 ( 2012 ) [ arxiv:1111.4956 [ hep - lat ] ] .
p. weisz , nucl .
b * 212 * , 1 ( 1983 ) .
g. curci , p. menotti and g. paffuti , phys .
b * 130 * , 205 ( 1983 ) [ erratum - ibid .
b * 135 * , 516 ( 1984 ) ] . c. morningstar and m. j. peardon , phys .
d * 69 * , 054501 ( 2004 ) [ hep - lat/0311018 ] . | we present results for the ( color)magnetic and ( color)electric screening masses of the quark - gluon plasma in the presence of an external magnetic field .
the screening masses are extracted from the correlators of polyakov loops , determined by lattice qcd simulations at the physical point .
we explore temperatures in the range @xmath0 and magnetic field intensities up to @xmath1 .
we find that both screening masses are increasing functions of the magnetic field and that the dependence on @xmath2 becomes weaker for larger temperatures . in the case of the magnetic screening mass
a slight anisotropy is also observable . |
discovered in 1995 by both cdf and d , the top quark is by far the heaviest known fundamental particle @xcite .
the mass of the top quark ( @xmath1 ) is of particular interest , as radiative contributions involving both the top quark and the putative higgs boson contribute to the mass of the w boson .
thus , the masses of the top quark , the higgs boson and the w boson are not three independent parameters in the standard model ( sm ) . when and if the higgs boson is discovered , precision measurements of the masses of the w boson and the top quark will help make a key test of the sm , helping to answer whether the new find is indeed the sm higgs boson or some other , new scalar particle .
in addition , the heavy mass of the top quark , near the electroweak scale , indicates that the top quark may play a role in helping theorists disentangle possible new sources of physics @xcite .
this letter describes measurements of the top quark mass from the cdf and d collaborations using up to @xmath0 of data collected in run ii at the tevatron .
top quarks at the tevatron are produced predominantly in pairs , and decay almost always in the sm to a w boson and a b quark .
the topology of @xmath2events depends on the subsequent decay of the two w bosons . in the dilepton channel ,
each w boson decays leptonically , to an electron or muon and a neutrino .
the dilepton channel has the lowest background and only two jets in the leading order @xmath2decay , but suffers from underconstrained kinematics due to the two escaping neutrinos , as well as from having the lowest branching fraction among all decay channels . in the all - hadronic channel , the two w bosons decay hadronically to quarks .
the all - hadronic channel has the largest branching fraction and no neutrinos , but also contains no charged lepton to distinguish it from the large qcd background . in the lepton+channel channel
, one w boson decays hadronically and the other leptonically .
though there is an undetected neutrino , the kinematics of the system are still overconstrained .
two tricks are used often in @xmath2mass analyses to increase the signal - to - background and improve systematics .
each @xmath2event contains two btags ; if the secondary vertices from the decay of metastable b hadrons can be identified , jets arising from b quarks can be distinguished from jets arising from light flavor quarks .
this significantly cuts down on the number of background events , and also helps to match the jets observed in the detector to the quarks at the hard scatter level .
lepton+jets and all - hadronic events also contain at least one hadronically decaying w boson . the narrow decay width and well known w boson mass in these events can be used to constrain , _ in situ _ , the largest systematic in top quark mass measurements , the calibration and response of calorimeters to hadronic particles , also known as the jet energy scale ( jes ) .
due to the underconstrained kinematics , measurements of @xmath1 in the dilepton channel must integrate over some unknown quantities .
the d experiment has two dilepton measurements , each using @xmath3 of data . in the matrix weighting method ,
each charged lepton - jet pairing is given a weight for the expectation to find , within experimental resolutions , the leptons with the measured energy , given a top quark mass and the unknown top and and anti - top @xmath4 .
the @xmath4 values are integrated over using parton distribution functions , and a likelihood fit yields @xmath1 = @xmath5 @xcite , with systematics that are dominated by the jet energy scale . in the neutrino weighting algorithm ( nwa ) , the unknown pseudorapidities of the two neutrinos are integrated over .
the solutions for a given top quark mass are weighted by the agreement with the missing transverse energy in the detector .
the mean and rms of the top quark mass weight distribution are used as estimators for the true top quark mass . with @xmath3 of data ,
d measures @xmath6 @xcite . the above two measurements , while largely correlated
, are not completely correlated . a combination using the blue technique @xcite yields @xmath7 @xcite
. cdf uses the nwa measurement in the dilepton channel with @xmath8 of data .
the most probable top quark mass , and not the mean , is taken as the first estimator ; the distribution is often rather asymmetric , so these are not necssarily the same quantity .
the second observable is the @xmath9 , the scalar sum of , lepton @xmath4 values and jet @xmath4 values .
cdf measures @xmath10 @xcite .
the kinematics in the lepton+jets and dilepton channel are overconstrained , so there is no need to integrate over unknown quantities .
the overconstrained kinematics are also used to select the single best assignment of jets to the quarks at the hard scatter
the single assignment most consistent with the @xmath2hypothesis is used .
cdf has two such measurements with @xmath8 of data .
a measurement in the all - hadronic channel uses a neural network to increase the s : b and reduce the qcd background .
in addition , the w mass constraint is used to calibrate the jes , yielding @xmath11 @xcite , where , as in all such measurements that contain an _ in situ _ jes calibration , the statistical uncertainty also includes a component for the jes systematic that now scales with @xmath12 .
a measurement in the lepton+jets channel yields @xmath13 @xcite , and also includes an _ in situ _ calibration .
cdf also has the first - ever analysis combining measurements of the top quark mass across different decay toplogies into the same likelihood .
more - traditional combinations must assume correlations for systematics between measurements , as well as assume gaussian behavior of the separate likelihoods . by combining the measurements into the same likelihood
, these assumptions are not needed .
the combination of cdf s lepton+jets and dilepton anaylses described above yields @xmath14 @xcite .
a different class of top quark mass analyses , called matrix element ( me ) analyses , try to extract as much information as possible from every event .
all jet - parton assignments consistent with b - tagging are used in the likelihood , which makes use of leading order theoretical predictions for how @xmath2events are produced and decayed , as given by the matrix element .
typically , leptons are assumed to be perfectly measured , as are jet angles .
the energies of the partons at the hard scatter level are encoded in transfer functions , which give the probability to observe a jet with energy @xmath15 given a parton with energy @xmath16 .
the transfer functions are needed since analyses measure jets in the detector , but the matrix element knows only how to describe events at the parton level . in typical me analyses , the probability to observe @xmath17 in the detector , given some top quark mass and jes in the detector , is given by : @xmath18 where p gives the probability to observe x in the detector , given some top quark mass ( and jes in the detector , if the measurement includes an _ in situ _ calibration ) .
n is a normalization term that includes effects of efficiency and acceptance , as well as the changing @xmath2production cross section as a function of @xmath1 .
the integral over @xmath19 is an integral over the parton - level phase space .
the matrix element m is the leading order matrix element for @xmath2production with partons @xmath16 , given some top quark mass .
the transfer functions w give the probability to observe a jet with energy @xmath15 given a parton energy @xmath16 ( and possibly the jet calibration in the detector ) . finally , there are two terms in @xmath20 that come from the parton distribution functions and give the probability to observe the two incoming partons with the appropriate energy .
cdf has a me element in the dilepton channel using @xmath0 .
the analysis uses a novel evolutionary neural network at the selection stage to improve the _ a priori _ statistical uncertainty on the top quark mass by 20% .
normal neural networks are trained only to minimize misclassifcation .
as such , they can be used only to distinguish signal and background , not to improve directly the expected uncertainty on a measurement .
the analysis measures @xmath1 = @xmath21 @xcite .
cdf has a me element analysis in the lepton+jets channel using @xmath8 .
the analysis differs from typical me analyses via the modification of the propagators in the matrix element to account for the imperfect assumptions about perfectly measured angles and intermediate particle masses that make the multi - dimensional integral tractable .
the analysis also makes a cut on the peak likelihood to remove both background events as well as poorly modeled signal events where the object in the detector do not match the assumed partons at the matrix element level .
the analysis includes an _ in situ _ jes calibration , and measures @xmath1 = @xmath22 @xcite .
d also has a me analysis using @xmath23 .
unlike most lepton+jet analyses , this analysis includes events with 0 b - tags .
the events are separated by charged lepton type ( electron vs muon ) .
including an _ in situ _ jes calibration , d measures @xmath1 = @xmath24 @xcite .
as run ii progresses at the tevatron , top quark mass measurements are rapidly approaching systematic limits .
a new set of analyses are emerging from the tevatron that make very different assumptions to measure the top quark mass , and as such are sensitive to very different systematic uncertainties .
in one such measurement , d measures the top quark mass via a measurement of the @xmath2pair production cross section .
top quark pairs at the tevatron are produced nearly at threshold , so the cross section depends strongly on the top quark mass .
the analysis depends on theoretical inputs to model this relationship ; using @xmath23 of data @xcite , d measures @xmath1 = @xmath25 using a @xmath26-@xmath27 curve from kidonakis and vogt . using a curve from cacciari et al .
gives @xmath1 = @xmath28 .
the world average tevatron top quark mass from the tevatron as of march 2007 , @xmath1 = @xmath29 @xcite , already comes close to being a 1% measurement , and does not include most of the analyses describe in this letter .
figure [ tevplot ] compares the world average with measurements from both experiments .
cdf has a new combination of its own analyses , yielding @xmath1 = @xmath30 @xcite , and expects by the end of run ii to have a cdf - only combination of top quark mass measurements with a precision better than 1% , as indicated in figure [ cdfplot ] .
i would like to thank all those in the cdf and d top groups who came before me in run i and run ii ( including all conveners ) for setting the bar so high for top quark mass measurements .
in addition , many thanks to the organizers of the la thuile 2008 conference for all their hard work .
99 abe , f. _ et al_. the cdf collaboration .
lett . , * 74 * , 2626 ( 1995 ) .
abachi , s. _ et al_. the d collaboration .
lett . , * 74 * , 2632 ( 1995 ) .
hill , christopher t. and parke , stephen j. phys .
d. , * 49 * , 4454 ( 1994 ) . the d collaboration .
d conference note 5463 ( 2007 ) . the d collaboration .
d conference note 5347 ( 2007 ) .
lyons , l. , gibaut , d. and clifford , p. nucl .
inst . meth .
, * a270 * , 110 ( 1988 ) . the d collaboration .
d conference note 5460 ( 2007 ) . the d collaboration .
d conference note 5362 ( 2007 ) . the d collaboration .
d conference note 5459 ( 2007 ) .
the cdf collaboration .
cdf conference note 9206 ( 2008 ) . the cdf collaboration .
cdf conference note 9130 ( 2008 ) .
the cdf collaboration .
cdf conference note 9196 ( 2008 ) . the cdf collaboration .
cdf conference note 9165 ( 2007 ) .
the cdf collaboration .
cdf conference note 9214 ( 2008 ) .
the cdf and d collaborations .
cdf conference note 8735 , d conference note 5378 ( 2007 ) . | top quark mass measurements from the tevatron using up to @xmath0 of data are presented .
prospects for combined tevatron measurements by the end of run ii are discussed .
= 11.6pt |
for all life - forms on earth with the exception of certain viruses , genetic information is carried within the cellular nucleus via strands of strands of deoxyribonucleic acid ( dna ) .
the genetic information itself is encoded in the specific sequence of the nucleic acid bases : adenine ( a ) , thymine ( t ) , guanine ( g ) , and cytidine ( c ) .
dna is also a strong absorber of ultraviolet light leaving it highly susceptible to photomutagenic damage with the primary photoproducts being bipyrimidine dimers linking neighboring t bases . for all organisms ,
this susceptibility is compensated for in part through enzymatic repair actions that remove damaged segments along one strand using the complementary strand as a template for replacement .
such repair mechanisms are quite costly energetically . remarkably , however , dna is intrinsically photochemically stable as single bases are able to rapidly convert photoexcitation energy into thermal energy on a picosecond time scale through non - radiative electronic processes .
what remains poorly understood , is the role that base - pairing and base - stacking plays in the transport and migration of the initial excitation along the double helix . clearly , such factors are important since the uv absorption of dna largely represents the weighted sum of the absorption spectra of it constituent bases whereas the distribution of lesions formed as the result of photoexcitation are generally not uniformly distributed along the chain itself and depend strongly upon sequence , suggesting some degree of coupling between bases.@xcite given the importance of dna in biological system and its emerging role as a scaffold and conduit for electronic transport in molecular electronic devices , @xcite dna in its many forms is a well studied and well characterized system .
what remains poorly understood , however , is the role that base - pairing and base - stacking plays in the transport and migration of the initial excitation along the double helix.@xcite such factors are important since the uv absorption of dna largely represents the weighted sum of the absorption spectra of it constituent bases whereas the distribution of lesions formed as the result of photoexcitation are generally not uniformly distributed along the chain itself and depend strongly upon sequence , suggesting some degree of coupling between bases.@xcite recent work by various groups has underscored the different roles that base - stacking and base - pairing play in mediating the fate of an electronic excitation in dna .
@xcite over 40 years ago , lwdin discussed proton tunneling between bases as a excited state deactivation mechanism in dna@xcite and evidence of this was recently reported by schultz _
@xcite in contrast , however , ultrafast fluorescence of double helix poly(da).poly(dt ) oligomers by crespo - hernandez et al.@xcite and by markovitsi _
@xcite give compelling evidence that base - stacking rather than base - pairing largely determines the fate of an excited state in dna chains composed of a and t bases with long - lived intrastrand states forming when ever a is stacked with itself or with t. however , there is considerable debate regarding whether or not the dynamics can be explained via purely frenkel exciton models @xcite or whether charge - transfer states play an intermediate role .
@xcite here we report on a series of quantum dynamical calculations that explore the fate of a localized exciton placed on either the a side or t side of the b dna duplex poly(da)@xmath1.poly(dt)@xmath1 .
our theoretical model is based upon a @xmath0 lattice model we recently introduced @xcite that consists of localized hopping interactions for electrons and holes between adjacent base pairs along each strand ( @xmath2 ) as well as cross - strand terms linking paired bases ( @xmath3 ) and `` diagonal '' terms which account for the @xmath4 stacking interaction between base @xmath5 on one chain and base @xmath6 on the other chain ( @xmath7 ) in which @xmath8 denotes coupling in the 5-5 direction and @xmath9 coupling in the 3-3 direction .
[ fig1 ] shows the three - dimensional structure of poly(da)@xmath1.poly(dt)@xmath1 and the topology of the equivalent lattice model .
we also consider here the role of geometric or structural fluctuations in the electronic dynamics .
taking each link as fig . 1 as a specific electron , hole , or excitonic , hopping term , we arrive at the following single particle hamiltonian , @xmath10 where @xmath11 and @xmath12 are @xmath13 spinors that act on the ground - state to create and
remove an electron ( or hole ) on the @xmath5th adenosine or thymidine base along the chain .
the @xmath14 operators are the @xmath15 pauli spin matrices with @xmath16 and @xmath17 providing the mixing between the two chains . as discussed in ref .
@xcite , we used the highest occupied and lowest unoccupied ( @xmath4 and @xmath18 ) orbitals localized on each base as a orthonormal basis . for the single particle terms ( representing electron and hole transfer between bases ) , we use values reported by anatram and mehrez as determined by computing the coulomb integrals between homo and lumo levels on adjacent base pairs with in a double - strand b dna sequence using density functional theory ( b3lyp/6 - 31 g ) @xcite taking the geometries of each base from the b - dna structure . when @xmath19 , eq .
[ bandop ] is identical to the hamiltonian used by creutz and horvath @xcite to describe chiral symmetry in quantum chromodynamics in which the terms proportional to @xmath20 are introduced to make the `` doublers '' at @xmath21 heavier than the states at @xmath22 .
taking the chain to homogeneous and infinite in extent , one can easily determine the energy spectrum of the valence and conduction bands by diagonalizing @xmath23 where @xmath24 and @xmath25 are the valence band or conduction band site energies and intra - strand hopping integrals .
when @xmath26 , eq .
[ bandop ] is identical to the hamiltonian used by creutz and horvath @xcite to describe chiral symmetry in quantum chromodynamics in which the terms proportional to @xmath20 are introduced to make the `` doublers '' at @xmath21 heavier than the states at @xmath22 .
in particular , we note that when @xmath27 , the band closes at @xmath28 but has a gap at @xmath29 .
the single particle parameters are taken from anatram and mehrez as determined by computing the coulomb integrals between homo and lumo levels on adjacent base pairs with in a double - strand b dna sequence using density functional theory ( b3lyp/6 - 31 g ) @xcite .
parameters used in our model are presented in table [ params ] .
it is important to note thatg the asymmetry introduced with @xmath30 gives directionality between the 3- and 5- ends of the chain . introducing these parameters into eq .
[ bandop ] leads to 4 separate cosine shaped bands corresponding to conduction and valence bands localized along each chain as show in fig .
[ bands ] . the coupling between the conduction and valence bands is accomplished by introducing short - ranged coulomb and exchange interactions as well as `` dipole - dipole '' terms which couple geminate electron - hole pairs on different sites and considering only single excitations , @xmath31 where the @xmath32 are spin - symmetrized composite operators that create or remove singlet or triplet electron / hole pairs in configuration @xmath33 where @xmath34 where @xmath35 is the total spin .
@xcite in our model , we include three types of electron / hole interactions , an on - site direct coulomb @xmath36 , an on - site exchange term @xmath37 , and an inter - site singlet exciton transfer term : @xmath38 . for simplicity ,
we take the on - site coulomb interaction as @xmath39 and the on - site exchange interaction @xmath37 to be adjustable parameters with values @xmath40 ev and @xmath41 ev for both purines and pyrimidines .
we assume these interactions to be local since the distance at which the coulomb energy between an electron / hole pair equals the thermal energy in aqueous ionic media at 300k is on the order of the base - stacking distance .
furthermore , such values are certainly in the correct range for conjugated cyclic organic systems .
lastly , we estimated the coupling between geminate electron / hole pairs on different bases @xmath42 via a point - dipole approximation by mapping the @xmath43 transition moments onto the corresponding base in the b dna chain .
these are obtained from the isolated bases by performing single configuration interaction ( cis ) calculations using the gamess@xcite quantum chemistry package on the corresponding 9-methylated purines and 1-methylated pyrimidines after optimizing the geometry at the hf/6 - 31(d)g level of theory .
the use of the point - dipole approximation in this case is justified mostly for convenience and given the close proximity of the bases , multipole terms should be included in a more complete model .
@xcite as a result , the matrix elements used herein provide an upper limit ( in magnitude ) of the couplings between geminate electron - hole pairs .
most importantly , however , the point - dipole approximation provides a robust means of incorporating the geometric arrangement of the bases into our model.@xcite .charge - transfer and exciton transfer terms for at dna .
( b - form ) [ params ] [ cols="<,^,^,^",options="header " , ] ' '' '' ' '' ''
we now consider the fate of an initial singlet electron / hole pair placed either in the middle of the thymidine side of the chain or the adenosine side of the chain ( i.e. a localized exciton with the electron and hole starting on the same site ) .
we assume that such a configuration is the result of an photoexcitation at the appropriate photon energy ( 4.87 ev for the thymidine exciton and 5.21 ev for the adenosine exciton respecitively ) and based upon the observation that the uv absorption spectra largely represents the weighted sum of the uv spectra of the constituent bases . since these are not stationary states , they evolve according to the time - dependent schrdinger equation , which we integrate using the tchebychev expansion of the time - evolution operator .
@xcite ( top - left ) or @xmath44 ( top - right ) site . the bottom panels show the net charge on each site following excitation of @xmath45 ( bottom - left ) or @xmath44 ( bottom - right ) , title="fig : " ] ( top - left ) or @xmath44 ( top - right ) site . the bottom panels show the net charge on each site following excitation of @xmath45 ( bottom - left ) or @xmath44 ( bottom - right ) , title="fig : " ] ( top - left ) or @xmath44 ( top - right ) site . the bottom panels show the net charge on each site following excitation of @xmath45 ( bottom - left ) or @xmath44 ( bottom - right ) , title="fig : " ] ( top - left ) or @xmath44 ( top - right ) site .
the bottom panels show the net charge on each site following excitation of @xmath45 ( bottom - left ) or @xmath44 ( bottom - right ) , title="fig : " ] in the top two frames of fig .
[ fig2 ] we show the transient probability for finding an exciton placed on the adenosine ( left ) or thymidine ( right ) chain at time @xmath46 in some other excitonic configuration along either the adenosine or thymidine chain at some time @xmath47 later . in both cases , negligible exciton density is transferred between chains and the excitons rapidly become delocalized and scatter ballistically down the dna chain .
there are some striking differences , however , between the exciton dynamics in adenosine versus those in thymidine .
first , in comparing the @xmath48 matrix elements , one easily concludes that the exciton mobility along the thymidine chain is considerably greater than the mobility along the adenosine chain .
this can be see in fig 2 . comparing the time required for an excitonic wavepacket to reach the end of either chain . in adenosine
, the exciton travels nearly 5 base pairs in about 25 fs where as an exciton along the thymidine chain covers the same distance in about 10 fs .
this factor of two difference in the exciton velocity is commensurate with the @xmath491:2 ratio of the @xmath50 intrachain excitonic couplings .
secondly , we note that the adenosine exciton remains qualitatively more `` cohesive '' than the thymidine exciton showing a number of ballistic traverses up and down the adenosine chain over the 300 fs we performed the calculation .
one can also note that the exciton velocity in the 5-3 direction is slightly greater than in the 3-5 direction as evidenced by the exciton rebounds off site @xmath51 slightly sooner than it rebounds from site @xmath52 .
this is due to the asymmetry introduced in by the @xmath53 and @xmath54 terms .
all in all , one can clearly note a series of strong recurrences for finding the adenosine exciton on the original site @xmath44 every 100 fs .
the thymidine exciton dynamics are far more complex as the exciton rapidly breaks apart .
while few recursions can be noted , however , after the first ballistic traverse , the thymidine exciton no longer exists as a cohesive wavepacket and is more or less uniformly distributed along the thymidine side of the chain .
the excitonic dynamics only tell part of story . in the lower frames of fig .
[ fig2 ] we show the net charge taken as the difference between the hole density and electron density on a given base . in the case where the initial exciton is on the adenosine chain ,
very little charge - separation occurs over the time scale of our calculation . on the other hand , when the exciton is placed on the thymidine chain , the exciton almost _ immediately _ evolves into a linear combination of excitonic and charge - separated configurations .
what is also striking is that in neither case do either the electron or hole transfer over to the other chain even though energetically charge - separated states with the electron on the thymidine and the hole on the adenosine sides of the chain are the lowest energy states of our model .
@xcite it is possible , that by including dissipation or decoherence into our dynamics , such relaxation will occur , however , on a time scale dictated by cross - chain transfer terms . for the coupling terms at hand ,
electron or hole transfer across base pairs occurs on the time scale of 3 - 4 ps .
the difference between the excitonic dynamics following excitation of a vs. t can be quantitatively noted by comparing the curves shown in fig .
[ fig3 ] where we compare the projection of the time - evolved state onto the excitonic configurations of the chain on which the exciton was placed ( @xmath55 and @xmath56 ) compared to the projection onto the excitonic configurations of the other chain ( @xmath57 and @xmath58 ) . for the case
in which the adenosine chain was excited , approximately 75% of the total probability density remains as excitonic configurations along the adenosine chain .
in stark contrast , only about 40% of the initial thymidine exciton remains excitonic along the thymidine side of the chain .
the reason for the remarkable difference between the two chains stems from the difference in electron and hole mobility along the thymidine chain . indeed , comparing the electron and hole hopping terms given above , @xmath59 for the thymidine chain compared to @xmath60 for along the adenosine chain
this is manifest in the lower left panel of fig .
[ fig2 ] where we see almost immediately a negative charge remaining for a few fs on the site where the initial excitation was placed .
in contrast , no charge - separation is seen following excitation along the adenosine side of the chain .
a proper model of the electronic dynamics in dna must include some contribution from the solvent and environment . for dna in water at 300k , we assume that the electronic processes described in our model are fast compared to the structural and environmental fluctuations the dna lattice itself such that the parameters for the b - dna structure represent the average values for the system in aqueous media .
for example , the characteristic time scale for the relative lateral and longitudinal motions of bases in dna is 10 to 100 fs with amplitudes of 0.01 to 0.1 nm .
@xcite since electronic interactions between bases are sensitive to the fluctuations in the geometry of the dna structure .
factors such as salt concentration and other solvent media will have profound impact on the structure and hence on the model parameters . since short strands of dna are fairly rigid , the electronic coupling terms are likely most sensitive to the upon the dihedral angle , @xmath61 between adjacent bases . if we take the fluctuations in @xmath61 to be @xmath62 where @xmath63 is the reduced moment of inertia of the at base - pair and @xmath64 is the torsional frequency .
this gives an rms angular fluctuation of about 5% about the avg .
@xmath65 helical angle .
since this is a small angular deviation , we take the fluctuations in the electronic terms to be proportional to @xmath66 and sample these terms from normal distributions about b - dna average values . except averaging over ensemble of configurations.,title="fig : " ] except averaging over ensemble of configurations.,title="fig : " ] the effect of disorder on the excitonic dynamics are seen in fig .
[ fig4 ] . as in case of the homogeneous lattice
above , our initial state is a frenkel exciton on thymidine # 5 .
the right - hand plot of fig .
[ fig4 ] shows the probability of finding the time - evolved state in various excitonic configurations .
while the time - scale for the exciton break up is slower than in the homogenous lattice , one can deduce that even with static off - diagonal disorder , the thymidine exciton is subject to dissociate to charge - transfer ( i.e. polaron ) pairs on an ultrafast timescale .
the adenosine exciton appears to be more profoundly affected by the lattice disorder with some population transfered to the thymidine chain .
since the total exciton population is lowered upon introducing disorder , the remaining excited state population exists as charge - separated pairs .
analysis of the transient excited state wavepackets ( not - shown ) indicates a small but significant fraction of @xmath67 interstrand charge - transfer configurations .
lastly , we examine the relative mobilities of the electron and hole by calculating the r.m.s .
width of the electron and hole densities on either chain following excitation , the results of which are shown in fig .
[ fig6 ] . in the top two plots
are the r.m.s .
widths for the homogeneous lattice following excitation .
as discussed above , the hole is considerably more mobile than the electron along the thymidine side of the chain .
this is very much evidenced by the more rapid spread of the hole density compared to the the electron density . likewise , along the adenosine side , the conduction and valence bands have nearly identical band - widths and the electron and hole densities evolve very much in concert .
this scenario is preserved upon even averaging over configurations indicating that to within a 5% error in our parameter set or the to within the typical thermal fluctuations of a dna chain , our predicted dynamics are quite robust .
the results described herein paint a similar picture to that described by recent ultrafast spectroscopic investigations of ( da).(dt ) oligomers in that the initial excitonic dynamics is dominated by base - stacking type interactions rather than by inter - base couplings .
interchain transfer is multiple orders of magnitude slower than the intrachain transport of both geminate electron / hole pairs as excitons and independent charge - separated species . indeed , for an exciton placed on the adenosine chain
, our model predicts that exciton remains as a largely cohesive and geminate electron / hole pair wave function as it scatters along the adenosine side of the chain .
our model also highlights how the difference between the mobilities in the conduction and valence bands localized along each chain impact the excitonic dynamics by facilitating the break up of the thymidine exciton into separate mobile charge - carriers . in the actual physical system
, the mobility of the free electron and hole along the chain will certainly be dressed by the polarization of the medium and reorganization of the lattice such that the coherent transport depicted here will be replaced by incoherent hopping between bases .
the significance of the breakup of the exciton is twofold .
first , it is well recognized that photoexcitation of adjacent stacked pyrimidine bases leads to the formation of _ cis - syn _ cyclobutane pyrimidine - dimer lesions .
however , this dimerization occurs only in the triplet ( rather than singlet ) excited state .
@xcite consequently , spin - flip must occur _ ether _ via spin - orbit coupling or via recombination of polaron pairs.@xcite if we assume that the spins are decorrelated at some intermediate distance @xmath68 where the coulomb energy is equal to the thermal energy , photoexcitation of a thymine sequence could rapidly result in a population of triplet excitons formed by exciton dissociation followed by geminate recombination .
secondly , the process is reversible and triplet reactivation of the dimer can lead to repair of the lesion .
isolating the photoexcitation to the originally excited chain minimizes the potential mutagenenic damage to the dna sequence since it preserves the complementary chain as an undamaged back - up copy of the genetic information .
it is fascinating to speculate whether or not the isolation of a photoexcitation and its photoproducts to the original chain was an early evolutionary selection criteria for the eventual emergence of dna as the carrier of genetic information . in conclusion , we present herein a rather compelling model for the short - time dynamics of the excited states in dna chains that incorporates both charge - transfer and excitonic transfer .
it is certainly not a complete model and parametric refinements are warranted before quantitative predictions can be established . for certain ,
there are various potentially important contributions we have left out : disorder in the system , the fluctuations and vibrations of the lattice , polarization of the media , dissipation , quantum decoherence .
we hope that this work serves as a starting point for including these physical interactions into a more comprehensive description of this system .
this work was funded in part by grants from the robert welch foundation , and the national science foundation ( che-0345324 ) .
d. markovitsi , d. onidas , t. gustavsson , f. talbot , e. lazzarotto , collective behavior of franck - condon excited states and energy transfer in dna double helices . , journal of the american chemical society 127 ( 2005 ) 1713017131 .
s. o. kelley , j. k. barton , electron transfer between bases in double helical dna , science 283 ( 1999 ) 375 .
/ cgi / content / abstract/283/5400% /375[http://www.sciencemag.org / cgi / content / abstract/283/5400% /375 ] t. schultz , e. samoylova , w. radloff , v. h. ingolf , a. l. sobolewski , w. domcke , efficient deactivation of a model base pair via excited - state hydrogen transfer , science 306 ( 2004 ) 17658 . http://www.sciencemag.org / cgi / content / abstract/306/5702% /1765[http://www.sciencemag.org / cgi / content / abstract/306/5702% /1765 ] d. markovitsi , f. talbot , t. gustavsson , d. onidas , e. lazzarotto , s. marguet , molecular spectroscopy : complexity of excited - state dynamics in dna , nature 441 ( 2006 ) e7 .
http://www.nature.com / nature / journal / v441/n7094/pdf / nat% ure04903.pdf[http://www.nature.com / nature / journal / v441/n7094/pdf / nat% ure04903.pdf ] c. e. crespo - hernandez , b. cohen , b. kohler , molecular spectroscopy : complexity of excited - state dynamics in dna ( reply ) , nature 441 ( 2006 ) e8 .
http://www.nature.com / nature / journal / v441/n7094/pdf / nat% ure04904.pdf[http://www.nature.com / nature / journal / v441/n7094/pdf / nat% ure04904.pdf ] s. karabunarliev , e. r. bittner , polaron excitons and electron vibrational band shapes in conjugated polymers , the journal of chemical physics 118 ( 9 ) ( 2003 ) 42914296 .
s. karabunarliev , e. r. bittner , dissipative dynamics of spin - dependent electron
hole capture in conjugated polymers , the journal of chemical physics 119 ( 7 ) ( 2003 ) 39883995 .
s. karabunarliev , e. r. bittner , electroluminescence yield in donor - acceptor copolymers and diblock polymers : a comparative theoretical study , the journal of chemical physics 108 ( 29 ) ( 2004 ) 1021910225 .
s. karabunarliev , e. r. bittner , spin - dependent electron - hole capture kinetics in luminescent conjugated polymers , physical review letters 90 ( 5 ) ( 2003 ) 057402 .
http://link.aps.org/abstract/prl/v90/e057402 m. w. schmidt , k. k. baldridge , j. a. boatz , s. t. elbert , m.s.gordon , j. h. jensen , s. koseki , n. matsunaga , k. a. nguyen , s. j. su , t. l. windus , m. dupuis , j. a. montgomery , general atomic and molecular electronic structure system , j. comput .
chem 14 ( 1993 ) 1347 .
b. bouvier , t. gustavsson , d. markovitsi , p. millie , dipolar coupling between electronic transitions of the dna bases and its relevance to exciton states in double helices . ,
chemical physics 275 ( 2002 ) 7592 .
h. tal - ezer , r. kosloff , an accurate and efficient scheme for propagating the time dependent schr[o - umlaut]dinger equation , the journal of chemical physics 81 ( 9 ) ( 1984 ) 39673971 . | recent ultrafast experiments have implicated intrachain base - stacking rather than base - pairing as the crucial factor in determining the fate and transport of photoexcited species in dna chains .
an important issue that has emerged concerns whether or not a frenkel excitons is sufficient one needs charge - transfer states to fully account for the dynamics .
in we present an @xmath0 lattice model which incorporates both intrachain and interchain electronic interactions to study the quantum mechanical evolution of an initial excitonic state placed on either the adenosine or thymidine side of a model b dna poly(da).poly(dt ) duplex .
our calculations indicate that over several hundred femtoseconds , the adenosine exciton remains a cohesive excitonic wave packet on the adenosine side of the chain where as the thymidine exciton rapidly decomposes into mobile electron / hole pairs along the thymidine side of the chain . in both cases , the very little transfer to the other chain
is seen over the time - scale of our calculations .
we attribute the difference in these dynamics to the roughly 4:1 ratio of hole vs. electron mobility along the thymidine chain .
we also show that this difference is is robust even when structural fluctuations are introduced in the form of static off - diagonal disorder .
2g^(2 ) dna , molecular biophysics , excited states , lattice theory , quantum chromodynamics 87.14.gg , 87.15.mi , 87.15.he , 31.15.ar |
approximate bayesian computation ( abc ) represents an elaborate statistical approach to model - based inference in a bayesian setting in which model likelihoods are difficult to calculate ( due to the complexity of the models considered ) . since its introduction in population genetics @xcite
, the method has found an ever increasing range of applications covering diverse types of complex models in various scientific fields ( see , e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the principle of abc is to conduct bayesian inference on a dataset through comparisons with numerous simulated datasets .
however , it suffers from two major difficulties .
first , to ensure reliability of the method , the number of simulations is large ; hence , it proves difficult to apply abc for large datasets ( e.g. , in population genomics where tens to hundred thousand markers are commonly genotyped ) .
second , calibration has always been a critical step in abc implementation @xcite .
more specifically , the major feature in this calibration process involves selecting a vector of summary statistics that quantifies the difference between the observed data and the simulated data .
the construction of this vector is therefore paramount and examples abound about poor performances of abc model choice algorithms related with specific choices of those statistics @xcite , even though there also are instances of successful implementations .
we advocate a drastic modification in the way abc model selection is conducted : we propose both to step away from selecting the most probable model from estimated posterior probabilities , and to reconsider the very problem of constructing efficient summary statistics .
first , given an arbitrary pool of available statistics , we now completely bypass selecting among those .
this new perspective directly proceeds from machine learning methodology .
second , we postpone the approximation of model posterior probabilities to a second stage , as we deem the standard numerical abc approximations of such probabilities fundamentally untrustworthy . we instead advocate selecting the posterior most probable model by constructing a ( machine learning ) classifier from simulations from the prior predictive distribution ( or other distributions in more advanced versions of abc ) , known as the abc _
reference table_. the statistical technique of random forests ( rf ) @xcite represents a trustworthy machine learning tool well adapted to complex settings as is typical for abc treatments .
once the classifier is constructed and applied to the actual data , an approximation of the posterior probability of the resulting model can be produced through a secondary random forest that regresses the selection error over the available summary statistics .
we show here how rf improves upon existing classification methods in significantly reducing both the classification error and the computational expense . after presenting theoretical arguments
, we illustrate the power of the abc - rf methodology by analyzing controlled experiments as well as genuine population genetics datasets .
bayesian model choice @xcite compares the fit of @xmath1 models to an observed dataset @xmath2 .
it relies on a hierarchical modelling , setting first prior probabilities @xmath3 on model indices @xmath4 and then prior distributions @xmath5 on the parameter @xmath6 of each model , characterized by a likelihood function @xmath7 .
inferences and decisions are based on the posterior probabilities of each model @xmath8 .
while we can not cover in much details the principles of approximate bayesian computation ( abc ) , let us recall here that abc was introduced in @xcite and @xcite for solving intractable likelihood issues in population genetics .
the reader is referred to , e.g. , @xcite , @xcite , @xcite , @xcite and @xcite for thorough reviews on this approximation method .
the fundamental principle at work in abc is that the value of the intractable likelihood function @xmath9 at the observed data @xmath2 and for a current parameter @xmath6 can be evaluated by the proximity between @xmath2 and pseudo - data @xmath10 simulated from @xmath11 . in discrete settings ,
the indicator @xmath12 is an unbiased estimator of @xmath9 @xcite . for realistic settings ,
the equality constraint is replaced with a tolerance region @xmath13 , where @xmath14 is a measure of divergence between the two vectors and @xmath15 is a tolerance value .
the implementation of this principle is straightforward : the abc algorithm produces a large number of pairs @xmath16 from the prior predictive , a collection called the _ reference table _ , and extracts from the table the pairs @xmath16 for which @xmath17 . to approximate posterior probabilities of competing models , abc methods @xcite
compare observed data with a massive collection of pseudo - data , generated from the prior predictive distribution in the most standard versions of abc ; the comparison proceeds via a normalized euclidean distance on a vector of statistics @xmath18 computed for both observed and simulated data .
standard abc estimates posterior probabilities @xmath8 at stage ( b ) of algorithm [ algo : general ] below as the frequencies of those models within the @xmath0 nearest - to-@xmath2 simulations , proximity being defined by the distance between @xmath19 and the simulated @xmath18 s . selecting a model means choosing the model with the highest frequency in the sample of size @xmath0 produced by abc , such frequencies being approximations to posterior probabilities of models . we stress that this solution means resorting to a @xmath0-nearest neighbor ( @xmath0-nn ) estimate of those probabilities , for a set of simulations drawn at stage ( a ) , whose records constitute the so - called _ reference table _ , see @xcite or @xcite .
* generate a reference table including @xmath20 simulations @xmath21 from @xmath22 * learn from this set to infer about @xmath23 at @xmath24 selecting a set of summary statistics @xmath25 that are informative for model choice is an important issue .
the abc approximation to the posterior probabilities @xmath8 will eventually produce a right ordering of the fit of competing models to the observed data and thus select the right model for a specific class of statistics on large datasets @xcite .
this most recent theoretical abc model choice results indeed show that some statistics produce nonsensical decisions and that there exist sufficient conditions for statistics to produce consistent model prediction , albeit at the cost of an information loss due to summaries that may be substantial .
the toy example comparing ma(1 ) and ma(2 ) models in appendix and figure [ fig : trueppvssummaries ] clearly exhibits this potential loss in using only the first two autocorrelations as summary statistics .
@xcite developed an interesting methodology to select the summary statistics , but with the requirement to aggregate estimation and model pseudo - sufficient statistics for every model under comparison .
that induces a deeply inefficient dimension inflation and can be very time consuming .
-axis ) and on only the first two autocorrelations ( @xmath26-axis ) .
[ fig : trueppvssummaries ] , scaledwidth=50.0% ] it may seem tempting to collect the largest possible number of summary statistics to capture more information from the data .
this brings @xmath27 closer to @xmath8 but increases the dimension of @xmath18 .
abc algorithms , like @xmath0-nn and other local methods suffer from the curse of dimensionality ( see e.g. section 2.5 in @xcite ) so that the estimate of @xmath27 based on the simulations is poor when the dimension of @xmath18 is too large . selecting summary statistics correctly and sparsely is therefore paramount , as shown by the literature in the recent years .
( see @xcite surveying abc parameter estimation . ) for abc model choice , two main projection techniques have been considered so far .
first , @xcite show that the bayes factor itself is an acceptable summary ( of dimension one ) when comparing two models , but its practical evaluation via a pilot abc simulation induces a poor approximation of model evidences @xcite .
the recourse to a regression layer like linear discriminant analysis ( lda , @xcite ) is discussed below and in appendix section a. other projection techniques have been proposed in the context of parameter estimation : see , e.g. , @xcite . given the fundamental difficulty in producing reliable tools for model choice based on summary statistics @xcite , we now propose to switch to a different approach based on an adapted classification method .
we recall in the next section the most important features of the random forest ( rf ) algorithm .
the classification and regression trees ( cart ) algorithm at the core of the rf scheme produces a binary tree that sets allocation rules for entries as labels of the internal nodes and classification or predictions of @xmath28 as values of the tips ( terminal nodes ) . at a given internal node
, the binary rule compares a selected covariate @xmath29 with a bound @xmath30 , with a left - hand branch rising from that vertex defined by @xmath31 . predicting the value of @xmath28 given the covariate @xmath32 implies following a path from the tree root that is driven by applying these binary rules .
the outcome of the prediction is the value found at the final leaf reached at the end of the path : majority rule for classification and average for regression . to find the best split and the best variable at each node of the tree ,
we minimize a criterium : for classification , the gini index and , for regression , the @xmath33-loss . in the randomized version of the cart algorithm ( see algorithm a1 in the appendix ) , only a random subset of covariates of size @xmath34 is considered at each node of the tree .
the rf algorithm @xcite consists in bagging ( which stands for bootstrap aggregating ) randomized cart .
it produces @xmath35 randomized cart trained on samples or sub - samples of size @xmath36 produced by bootstrapping the original training database .
each tree provides a classification or a regression rule that returns a class or a prediction .
then , for classification we use the majority vote across all trees in the forest , and , for regression , the response values are averaged .
three tuning parameters need be calibrated : the number @xmath35 of trees in the forest , the number @xmath34 of covariates that are sampled at a given node of the randomized cart , and the size @xmath36 of the bootstrap sub - sample .
this point will be discussed in section [ sec : prac ] . for classification ,
a very useful indicator is the _ out - of - bag _ error ( * ? ? ?
* chapter 15 ) . without any recourse to a test set
, it gives you some idea on how good is your rf classifier .
for each element of the training set , we can define the out - of - bag classifier : the aggregation of votes over the trees not constructed using this element .
the out - of - bag error is the error rate of the out - of - bag classifier on the training set .
the out - of - bag error estimate is as accurate as using a test set of the same size as the training set . the above - mentioned difficulties in abc model choice drives us to a paradigm shift in the practice of model choice , namely to rely on a classification algorithm for model selection , rather than a poorly estimated vector of @xmath27 probabilities .
as shown in the example described in section 3.1 , the standard abc approximations to posterior probabilities can significantly differ from the true @xmath8 .
indeed , our version of stage ( b ) in algorithm [ algo : general ] relies on a rf classifier whose goal is to predict the suited model @xmath37 at each possible value @xmath38 of the summary statistics @xmath18 .
the random forest is trained on the simulations produced by stage ( a ) of algorithm [ algo : general ] , which constitute the reference table .
once the model is selected as @xmath39 , we opt to approximate @xmath40 by another random forest , obtained from regressing the probability of error on the ( same ) covariates , as explained below . a practical way to evaluate the performance of an abc model choice algorithm and to check whether both a given set of summary statistics and a given classifier is to check whether it provides a better answer than others .
the aim is to come near the so - called _ bayesian classifier _ , which , for the observed @xmath41 , selects the model having the largest posterior probability @xmath42 .
it is well known that the bayesian classifier ( which can not be derived ) minimizes the 01 integrated loss or error @xcite . in the abc framework , we call the integrated loss ( or risk ) the _ prior error rate _ , since it provides an indication of the global quality of a given classifier @xmath43 on the entire space weighted by the prior . this rate is the expected value of the misclassification error over the hierarchical prior @xmath44 it can be evaluated from simulations @xmath45 drawn as in stage ( a ) of algorithm [ algo : general ] , independently of the reference table @xcite , or with the out - of - bag error in rf that , as explained above , requires no further simulation .
both classifiers and sets of summary statistics can be compared via this error scale : the pair that minimizes the prior error rate achieves the best approximation of the ideal bayesian classifier . in that sense it stands closest to the decision we would take were we able to compute the true @xmath42 .
we seek a classifier in stage ( b ) of algorithm [ algo : general ] that can handle an arbitrary number of statistics and extract the maximal information from the reference table obtained at stage ( a ) .
as introduced above , random forest ( rf ) classifiers @xcite are perfectly suited for that purpose .
the way we build both a rf classifier given a collection of statistical models and an associated rf regression function for predicting the allocation error is to start from a simulated abc _
reference table _ made of a set of simulation records made of model indices and summary statistics for the associated simulated data .
this table then serves as training database for a rf that forecasts model index based on the summary statistics .
the resulting algorithm , presented in algorithm [ algo : abc - rf ] and called abc - rf , is implemented in the r package abcrf associated with this paper .
generate a reference table including @xmath20 simulation @xmath21 from @xmath22 construct @xmath35 randomized cart which predict @xmath23 using @xmath18 * draw * a bootstrap ( sub-)sample of size @xmath36 from the reference table * grow * a randomized cart @xmath46 ( algorithm a1 in the appendix ) determine the predicted indexes for @xmath19 and the trees @xmath47 affect @xmath19 according to a majority vote among the predicted indexes the justification for choosing rf to conduct an abc model selection is that , both formally @xcite and experimentally ( * ? ? ?
* chapter 5 ) , rf classification was shown to be mostly insensitive both to strong correlations between predictors ( here the summary statistics ) and to the presence of noisy variables , even in relatively large numbers , a characteristic that @xmath0-nn classifiers miss .
this type of robustness justifies adopting an rf strategy to learn from an abc reference table for bayesian model selection . within an arbitrary ( and arbitrarily large ) collection of summary statistics ,
some may exhibit strong correlations and others may be uninformative about the model index , with no terminal consequences on the rf performances . for model selection ,
rf thus competes with both local classifiers commonly implemented within abc : it provides a more non - parametric modelling than local logistic regression @xcite , which is implemented in the diyabc software @xcite which is extremely costly see , e.g. , @xcite which reduces the dimension using linear discriminant projection before resorting to local logistic regression .
this software also includes a standard @xmath0-nn selection procedure which suffers from the curse of dimensionality and thus forces selection among statistics .
the outcome of rf computation applied to a given target dataset is a classification vote for each model which represents the number of times a model is selected in a forest of n trees .
the model with the highest classification vote corresponds to the model best suited to the target dataset .
it is worth stressing here that there is no direct connection between the frequencies of the model allocations of the data among the tree classifiers ( i.e. the classification vote ) and the posterior probabilities of the competing models .
machine learning classifiers hence miss a distinct advantage of posterior probabilities , namely that the latter evaluate a confidence degree in the selected ( map ) model .
an alternative to those probabilities is the prior error rate . aside from its use to select the best classifier and set of summary statistics , this indicator remains , however , poorly relevant since the only point of importance in the data space is the observed dataset @xmath19 .
a first step addressing this issue is to obtain error rates conditional on the data as in @xcite .
however , the statistical methodology considered therein suffers from the curse of dimensionality and we here consider a different approach to precisely estimate this error .
we recall @xcite that the posterior probability of a model is the natural bayesian uncertainty quantification since it is the complement of the posterior error associated with the loss @xmath48 . while the proposal of @xcite for estimating the conditional error rate induced a classifier given @xmath49 @xmath50 involves non - parametric kernel regression , we suggest to rely instead on a rf regression to undertake this estimation .
the curse of dimensionality is then felt much less acutely , given that random forests can accommodate large dimensional summary statistics .
furthermore , the inclusion of many summary statistics does not induce a reduced efficiency in the rf predictors , while practically compensating for insufficiency . before describing in more details the implementation of this concept
, we stress that the perspective of @xcite leads to effectively estimate the posterior probability that the true model is the map , thus providing us with a non - parametric estimation of this quantity , an alternative to the classical abc solutions we found we could not trust .
indeed , the posterior expectation satisfies @xmath51 & = \sum_{i=1}^k \mathbb{e}[\mathbb{i}(\hat{m}(s({\mathbf{x}}^0))\ne m = i)|s({\mathbf{x}}^0)]\\ & = \sum_{i=1}^k \mathbb{p}[m = i)|s({\mathbf{x}}^0)]\times\mathbb{i}(\hat{m}(s({\mathbf{x}}^0))\ne i)\\ & = \mathbb{p}[m\ne \hat{m}(s({\mathbf{x}}^0))|s({\mathbf{x}}^0)]\\ & = 1-\mathbb{p}[m = \hat{m}(s({\mathbf{x}}^0))|s({\mathbf{x}}^0)]\,.\end{aligned}\ ] ] it therefore provides the complement of the posterior probability that the true model is the selected model . to produce our estimate of the posterior probability
@xmath52 $ ] , we proceed as follows : 1 .
we compute the value of @xmath53 for the trained random forest @xmath43 and for all terms in the abc reference table ; to avoid overfitting , we use the out - of - bag classifiers ; 2 .
we train a rf regression estimating the variate @xmath53 as a function of the same set of summary statistics , based on the same reference table .
this second rf can be represented as a function @xmath54 that constitutes a machine learning estimate of @xmath55 $ ] ; 3 .
we apply this rf function to the actual observations summarized as @xmath19 and return @xmath56 as our estimate of @xmath57 $ ] .
this corresponds to the representation of algorithm [ algo : posterior ] which is implemented in the r package abcrf associated with this paper .
* use the rf produce by algorithm [ algo : abc - rf ] to compute the out - of - bag classifiers of all terms in the reference table and deduce the associated binary model prediction error * use the reference table to build a rf regression function @xmath54 regressing the model prediction error on the summary statistics * return the value of @xmath56 as the rf regression estimate of @xmath57 $ ]
to illustrate the power of the abc - rf methodology , we now report several controlled experiments as well as two genuine population genetic examples .
the appendix details controlled experiments on a toy problem , comparing ma(1 ) and ma(2 ) time - series models , and two controlled synthetic examples from population genetics , based on single nucleotide polymorphism ( snp ) and microsatellite data .
the toy example is particularly revealing with regard to the discrepancy between the posterior probability of a model and the version conditioning on the summary statistics @xmath19 .
figure [ fig : trueppvssummaries ] shows how far from the diagonal are realizations of the pairs @xmath58 , even though the autocorrelation statistic is quite informative @xcite .
note in particular the vertical accumulation of points near @xmath59 .
table [ tab : mama ] in the appendix demonstrates the further gap in predictive power for the full bayes solution with a true error rate of 12% versus the best solution ( rf ) based on the summaries barely achieving a 16% error rate . for both controlled genetics experiments in the appendix ,
the computation of the true posterior probabilities of the three models is impossible .
the predictive performances of the competing classifiers can nonetheless be compared on a test sample .
results , summarized in tables [ tab : snp ] and [ tab : microsat ] in the appendix , legitimize the use of rf , as this method achieves the most efficient classification in all genetic experiments . note that that the prior error rate of any classifier is always bounded from below by the error rate associated with the ( ideal ) bayesian classifier .
therefore , a mere gain of a few percents may well constitute an important improvement when the prior error rate is low . as an aside
, we also stress that , since the prior error rate is an expectation over the entire sampling space , the reported gain may occult much better performances over some areas of this space .
figure [ fig : mama.post ] in the appendix displays differences between the true posterior probability of the model selected by algorithm [ algo : abc - rf ] and its approximation with algorithm [ algo : posterior ] . the original challenge was to conduct inference about the introduction pathway of the invasive harlequin ladybird _
( harmonia axyridis _ ) for the first recorded outbreak of this species in eastern north america .
the dataset , first analyzed in @xcite and @xcite via abc , includes samples from three natural and two biocontrol populations genotyped at 18 microsatellite markers .
the model selection requires the formalization and comparison of 10 complex competing scenarios corresponding to various possible routes of introduction ( see appendix for details and analysis 1 in @xcite ) .
we now compare our results from the abc - rf algorithm with other classification methods for three sizes of the reference table and with the original solutions by @xcite and @xcite .
we included all summary statistics computed by the diyabc software for microsatellite markers @xcite , namely 130 statistics , complemented by the nine lda axes as additional summary statistics ( see appendix section g ) . in this example
, discriminating among models based on the observation of summary statistics is difficult .
the overlapping groups of figure [ fig : cox_lda ] in the appendix reflect that difficulty , the source of which is the relatively low information carried by the 18 autosomal microsatellite loci considered here .
prior error rates of learning methods on the whole reference table are given in table [ tab : asian ] .
as expected in such a high dimension settings ( * ? ? ?
* section 2.5 ) , @xmath0-nn classifiers behind the standard abc methods are all defeated by rf for the three sizes of the reference table , even when @xmath0-nn is trained on the much smaller set of covariates composed of the nine lda axes .
the classifier and set of summary statistics showing the lowest prior error rate is rf trained on the 130 summaries and the nine lda axes .
figure [ fig : cox_viss ] in the appendix shows that rfs are able to automatically determine the ( most ) relevant statistics for model comparison , including in particular some crude estimates of admixture rate defined in @xcite , some of them not selected by the experts in @xcite .
we stress here that the level of information of the summary statistics displayed in figure [ fig : cox_viss ] in the appendix is relevant for model choice but not for parameter estimation issues . in other words ,
the set of best summaries found with abc - rf should not be considered as an optimal set for further parameter estimations under a given model with standard abc techniques @xcite .
.*harlequin ladybird data * : estimated prior error rates for various classification methods and sizes of the reference table.[tab : asian ] [ cols="^ , > , > , > " , ]
we here illustrate the potential of our abc - rf algorithm for the statistical processing of massive single nucleotide polymorphism ( snp ) datasets , whose production is on the increase within the field of population genetics . to this aim
, we analyzed a snp dataset obtained from individuals originating from four human populations ( 30 unrelated individuals per population ) using the freely accessible public 1000 genome databases ( i.e. , the vcf format files including variant calls available at ` http://www.1000genomes.org/data ` ) .
the goal of the 1000 genomes project is to find most genetic variants that have frequencies of at least 1% in the studied populations by sequencing many individuals lightly ( i.e. , at a @xmath60 coverage ) .
a major interest of using snp data from the 1000 genomes project @xcite is that such data does not suffer from any ascertainment bias ( i.e. , the deviations from expected theoretical results due to the snp discovery process in which a small number of individuals from selected populations are used as discovery panel ) , which is a prerequisite when using the diyabc simulator of snp data @xcite .
the four human populations included the yoruba population ( nigeria ) as representative of africa ( encoded yri in the 1000 genome database ) , the han chinese population ( china ) as representative of the east asia ( encoded chb ) , the british population ( england and scotland ) as representative of europe ( encoded gbr ) , and the population composed of americans of african ancestry in sw usa ( encoded asw ) .
the snp loci were selected from the 22 autosomal chromosomes using the following criteria : ( i ) all @xmath61 analyzed individuals have a genotype characterized by a quality score @xmath62 ( on a phred scale ) , ( ii ) polymorphism is present in at least one of the @xmath61 individuals in order to fit the snp simulation algorithm of diyabc @xcite , ( iii ) the minimum distance between two consecutive snps is 1 kb in order to minimize linkage disequilibrium between snps , and ( iv ) snp loci showing significant deviation from hardy - weinberg equilibrium at a 1% threshold @xcite in at least one of the four populations have been removed ( 35 snp loci involved ) . after applying the above criteria
, we obtained a dataset including 51,250 snp loci scattered over the 22 autosomes ( with a median distance between two consecutive snps equal to 7 kb ) among which 50,000 were randomly chosen for applying the proposed abc - rf methods . in this application , we compared six scenarios ( i.e. , models ) of evolution of the four human populations genotyped at the above mentioned 50,000 snps .
the six scenarios differ from each other by one ancient and one recent historical event : _
( i ) _ a single out - of - africa colonization event giving an ancestral out - of - africa population which secondarily splits into one european and one east asia population lineage , versus two independent out - of - africa colonization events , one giving the european lineage and the other one giving the east asia lineage .
the possibility of a second ancient ( i.e. , @xmath63 years ) out - of - africa colonization event through the arabian peninsula toward southern asia has been suggested by archaeological studies , e.g. @xcite ; _ ( ii ) _ the possibility ( or not ) of a recent genetic admixture of americans of african ancestry in sw usa between their african ancestors and individuals of european or east asia origins .
the six different scenarios as well as the prior distributions of the time event and effective population size parameters used to simulate snp datasets using diyabc are detailed in figure [ fig : outofaf ] .
we stress here that our intention is not to bring new insights into human population history , which has been and is still studied in greater details in a number of studies using genetic data , but to illustrate the potential of the proposed abc - rf methods for the statistical processing of large size snp datasets in the context of complex evolutionary histories .
rf computations to discriminate among the six scenarios of figure [ fig : outofaf ] and evaluate error rates were processed on 10,000 , 20,000 , and 50,000 simulated datasets .
we used all summary statistics offered by the diyabc software for snp markers @xcite ( see section g below ) , namely 130 summary statistics in this setting plus the five lda axes as additional summary statistics .
for all illustrations based on genetic data , we used the program diyabc v2.0 @xcite to generate the abc reference tables including a set of simulation records made of model indices , parameter values and summary statistics for the associated simulated data .
diyabc v2.0 is a multithreaded program which runs on three operating systems : gnu / linux , microsoft windows and apple mac os x. computational procedures are written in c++ and the graphical user interface is based on pyqt , a python binding of the qt framework .
the program is freely available to academic users with a detailed notice document , example projects , and code sources ( linux ) from : ` http://www1.montpellier.inra.fr/cbgp/diyabc ` .
the reference table generated this way then served as training database for the random forest constructions .
for a given reference table , computations were performed using the r package ` randomforest ` @xcite .
we have implemented all the proposed methodologies in the r package abcrf available on the cran .
single population statistics + ` hp0_i ` : proportion of monomorphic loci for population i + ` hm1_i ` : mean gene diversity across polymorphic loci @xcite + ` hv1_i ` : variance of gene diversity across polymorphic loci + ` hmo_i ` : mean gene diversity across all loci two population statistics + ` fp0_i&j ` : proportion of loci with null fst distance between the two samples for populations i and j @xcite + ` fm1_i&j ` : mean across loci of non null fst distances + ` fv1_i&j ` : variance across loci of non null fst distances + ` fmo_i&j ` : mean across loci of fst distances + ` np0_i&j ` : proportion of 1 loci with null nei s distance @xcite ` nm1_i&j ` : mean across loci of non null nei s distances + ` nv1_i&j ` : variance across loci of non null nei s distances + ` nmo_i&j ` : mean across loci of nei s distances three population statistics + ` ap0_i_j&k ` : proportion of loci with null admixture estimate when pop .
i comes from an admixture between j and k + ` am1_i_j&k ` : mean across loci of non null admixture estimate + ` av1_i_j&k ` : variance across loci of non null admixture estimated + ` amo_i_j&k ` : mean across all locus admixture estimates single population statistics + ` nal_i ` : mean number of alleles across loci for population i + ` het_i ` : mean gene diversity across loci @xcite + ` var_i ` : mean allele size variance across loci + ` mgw_i ` : mean m index across loci @xcite two population statistics + ` n2p_i&j ` : mean number of alleles across loci for populations i and j + ` h2p_i&j ` : mean gene diversity across loci + ` v2p_i&j ` : mean allele size variance across loci + ` fst_i&j ` : fst @xcite + ` lik_i&j ` : mean index of classification @xcite + ` das_i&j ` : shared allele distance @xcite + ` dm2_i&j ` : @xmath64 distance @xcite | approximate bayesian computation ( abc ) methods provide an elaborate approach to bayesian inference on complex models , including model choice .
both theoretical arguments and simulation experiments indicate , however , that model posterior probabilities may be poorly evaluated by standard abc techniques .
we propose a novel approach based on a machine learning tool named random forests to conduct selection among the highly complex models covered by abc algorithms .
we thus modify the way bayesian model selection is both understood and operated , in that we rephrase the inferential goal as a classification problem , first predicting the model that best fits the data with random forests and postponing the approximation of the posterior probability of the predicted map for a second stage also relying on random forests .
compared with earlier implementations of abc model choice , the abc random forest approach offers several potential improvements : _ ( i ) _ it often has a larger discriminative power among the competing models , _ ( ii ) _ it is more robust against the number and choice of statistics summarizing the data , _ ( iii ) _ the computing effort is drastically reduced ( with a gain in computation efficiency of at least fifty ) , and _
( iv ) _ it includes an approximation of the posterior probability of the selected model .
the call to random forests will undoubtedly extend the range of size of datasets and complexity of models that abc can handle .
we illustrate the power of this novel methodology by analyzing controlled experiments as well as genuine population genetics datasets .
the proposed methodologies are implemented in the r package abcrf available on the cran .
* keywords : * approximate bayesian computation , model selection , summary statistics , @xmath0-nearest neighbors , likelihood - free methods , random forests |
let @xmath1 denote the set of words of length @xmath2 generated using the alphabet @xmath3 .
let @xmath4 denote the * length of the longest strictly increasing subsequence * of @xmath5 , i.e. , the largest @xmath6 with a subsequence @xmath7 such that @xmath8 .
similarly , we consider the * length of the longest strictly decreasing subsequence * @xmath9 of @xmath10 .
our main goal is to introduce and study a discrete probability measure on young diagrams , in connection with the study of the distributions of lis and lds on uniform random words .
an additional goal is to provide a novel motivation for the @xmath0-theoretic schubert calculus combinatorics of @xcite .
there are analogies with the study of lis and lds in the * permutation case * , i.e. , when @xmath10 is chosen uniformly at random from the symmetric group @xmath11 .
the latter topic has attracted considerable attention ; we refer the reader to the surveys @xcite and the references therein . in the permutation case ,
random young diagrams are distributed according to the _ plancherel measure _ ( on irreducible representations ) of @xmath11 .
this discrete probability measure is the push - forward of the uniform distribution on @xmath11 , under the _ robinson - schensted correspondence_. schensted @xcite established that this correspondence encodes @xmath4 and @xmath9 symmetrically in the shape @xmath12 associated to @xmath10 . in @xcite ,
these ideas are applied to determine the asymptotics of the expectation of @xmath13 over @xmath11 ( solving the old `` longest increasing sequences problem '' ) , via a study of the `` limit typical shape '' under the plancherel measure . as a continuation of this theme
, we apply the hecke ( insertion ) algorithm of @xcite to define the young diagram @xmath14 for each @xmath15 ; using this we define _ plancherel - hecke measure_. our belief that this measure should actually be worthy of analysis was initially guided by our theorem that hecke symmetrically encodes @xmath4 and @xmath9 for @xmath15 , a generalization of schensted s theorem . during the course of our investigation , we found that many other aspects of the plancherel - hecke measure ( conjecturally ) also resemble those of the plancherel measure .
this paper records these results , both theoretical and computational , as a justification for further study .
briefly , this is how the two aforementioned measures compare : let @xmath16 . sending @xmath17
, we conjecture that for @xmath18 , our measure is concentrated around the limit typical shape under plancherel measure .
this _ plancherel curve _ plays an important role in @xcite . on the other hand , for @xmath19
we conjecture the measure is concentrated near the `` staircase shape '' .
in particular , a `` phase transition '' is suggested at @xmath20 .
as we tune @xmath21 , a symmetric deformation of the plancherel curve occurs . in view of the above mentioned result on the hecke algorithm , this transition phenomenon
is further evidenced by computations ( with contributions by o. zeitouni ) of the expectation of @xmath13 and @xmath22 as @xmath21 varies ; see section 5 and the appendix .
there have been earlier extensions of the permutation case to @xmath1 .
the limit distribution of the length of the longest _ weakly _ increasing / decreasing subsequence ( lwis / lwds ) on @xmath1 was found in work of @xcite , following the breakthrough @xcite on the limit distribution of @xmath13 on @xmath11 .
see also the more recent work @xcite .
however , analogous understanding of the distribution of lis and lds on @xmath1 appears to be less developed ; see , e.g. , @xcite for contributions .
as a point of comparison and contrast with our approach , previous work on @xmath13 , @xmath22 and @xmath1 utilizes the combinatorics of the _ robinson - schensted - knuth correspondence _ , which asymmetrically encodes @xmath23 and @xmath22 .
we offer an alternative viewpoint on the relationship between young diagrams and lis , lds .
new questions and conjectures are raised , stemming from the coxeter - theoretic viewpoint of @xcite ( which in turn generalizes ideas of @xcite ) .
this text expresses our desire to point out a natural link between the probabilistic combinatorics of lis , lds and the combinatorial algebraic geometry of @xmath0-theoretic schubert calculus .
in particular , we apply and further develop the _ jeu de taquin _ for _ increasing tableaux _ from @xcite , thereby giving another perspective on that work , distinct from the original one . in summary , we believe that the availability of these two disparate interpretations for @xcite provides something atypical to recommend @xmath0-theoretic tableau combinatorics , among the large array of interesting generalizations of the classical young tableau and symmetric function theories known today . we identify a partition @xmath24 with its young diagram ( in english notation ) ; set @xmath25 .
let @xmath26 denote the set of all young diagrams .
a filling of a shape @xmath12 with a subset of the labels @xmath3 is an * increasing tableau * if it is strictly increasing in both rows and columns .
let @xmath27 be the set of all increasing tableaux of shape @xmath12 .
we also need * set - valued tableaux * @xcite , which are fillings of @xmath12 assigning to each box a nonempty subset of @xmath28 such that the largest entry of a box is smaller than the smallest entry in the boxes directly to the right of it , and directly below it .
we call a set - valued tableau * standard * if each label is used precisely once
. let @xmath29 denote the set of all standard set - valued tableaux .
see figure [ fig : inc_set ] .
@xmath30 the * plancherel measure * on @xmath26 assigns to @xmath12 the probability @xmath31 let @xmath32 .
[ def : main ] the * plancherel - hecke probability measure * @xmath33 on @xmath26 is defined by letting @xmath34 be a random ( non - uniform ) young diagram with distribution @xmath35 [ prop : central ] the plancherel - hecke measure is well - defined as a probability distribution ; i.e. , the following identity holds : @xmath36 where @xmath37 there is an exact polynomial - time sampling algorithm @xmath38 terminating in @xmath39 operations , that induces @xmath33 from the uniform distribution on @xmath1 .
the core technical result of this paper is a generalization of the aforementioned theorem of schensted @xcite : [ thm : hard ] heckeshape simultaneously and symmetrically encodes @xmath4 and @xmath9 as the size of the first row and column of @xmath14 , respectively .
theorem [ thm : hard ] is obtained by establishing another new result , connecting @xmath40 to the `` @xmath0-infusion '' operation defined in @xcite .
this latter result is an analogue of the classical fact that connects the robinson - schensted correspondence to the ( ordinary ) _ jeu de taquin _ rectification procedure .
we prove of proposition [ prop : central ] in section 2 , after recalling the hecke algorithm of buch - kresch - shimozono - tamvakis - yong @xcite ( originally constructed to study degeneracy loci of vector bundles ) .
@xmath14 is the young diagram associated to @xmath10 under @xmath41 .
the proof of theorem [ thm : hard ] is given in section 4 .
we illustrate the identity ( [ eqn : ph2 ] ) for @xmath42 and @xmath43 .
there are nine partitions @xmath12 satisfying ( [ eqn : ph3 ] ) . these are @xmath44 then ( [ eqn : ph2 ] ) reads @xmath45 where the products on the righthand side of the equality are listed in order corresponding to the above partitions .
thus , the `` typical shape '' is @xmath46 , possessing nearly @xmath47 of the distribution .
the remainder of the above results will be illustrated in section 2 , after we define heckeshape .
theorem [ thm : hard ] has some immediate consequences , familiar from the permutation case .
[ cor : expectformula ] under the uniform measure on @xmath1 , we have @xmath48 in addition , @xmath49 two other consequences will be given in section 3 .
one gives a `` coxeter - theoretic '' generalization of the widely known erds - szekeres theorem @xcite .
another expands upon the discussion of patience sorting given in @xcite . in our experiments , heckeshape was reasonably efficient as a sampling algorithm .
for example , when @xmath50 sampling one young diagram takes on the order of seconds to minutes on current technology . for larger @xmath2
, we could sample one young diagram when @xmath51 in several hours on the same technology . a sample when @xmath52 took about one and a half days .
the memory demands were modest . in view of the apparent `` concentration '' suggested below
, one sample was enough to be of interest for our purposes , when @xmath2 is large .
there are classical antecedents of theorem [ thm : hard ] .
as stated earlier , schensted @xcite proved the analogous conclusion about the shape coming from the robinson - schensted correspondence for a permutation @xmath10 .
in contrast , knuth @xcite proved that the first row of the robinson - schensted - knuth algorithm ( rsk ) encodes the length of the longest weakly increasing subsequence ( @xmath23 ) of a word @xmath10 .
what is perhaps less well - known is that rsk also encodes @xmath9 as the length of the first column of the shape it associates to @xmath10 .
however , unlike heckeshape , it is asymmetric : @xmath4 is not encoded by the length of the first row ( as @xmath53 in general ) .
thus , the symmetry of hecke(shape ) makes it natural to analyze , as it seems desirable to simultaneously capture the statistics of @xmath13 and @xmath22 .
however , we do not have handy formulas for @xmath54 , @xmath55 or @xmath56 , such as the _ hook - length formula _ for @xmath57 .
( in the permutation case , the hook - length formula plays a crucial role , see @xcite . ) in small examples large prime factors appear , showing that such a formula is unlikely .
for instance , @xmath58 .
this issue is closely related to the open question of finding `` good '' determinantal expressions for _ grothendieck polynomials _
that being said , special cases exhibit connections to work of stanley @xcite on polygonal dissections , and of @xcite on generalized littlewood - richardson rules ( further discussion may appear elsewhere ) .
thus , the enumerative combinatorics of these numbers might of interest in their own right . it is not difficult to give recursions to calculate @xmath55 and @xmath56 that are useful in moderately large cases . are there efficient ( possibly randomized or approximate ) counting algorithms ? objectively ,
the lack of simple formulas to compute @xmath54 makes it trickier to apply standard approaches directly ; this is an admitted defect of our setting .
nevertheless , we believe the framework of problems described here is tractable .
in addition to the results below , in the appendix one gains useful and nontrivial information about the plancherel - hecke measure by exploiting the related work of @xcite . in this way
, the techniques of @xcite can be applied to the present context .
we organize our analysis by first setting @xmath59 and considering the limit behavior of @xmath33 when @xmath60 , as @xmath61 .
( the case @xmath62 is trivial . )
as is explained below , we conjecture that there is a * critical value * of @xmath21 , denoted @xmath63 : the behavior of @xmath33 is qualitatively different in the intervals @xmath64 $ ] and @xmath65 . at @xmath66 , further refinement of the analysis
is needed , as we transition from one state to the other . in the permutation case , to study the plancherel measure , it is useful to consider the most likely , or `` typical '' shape .
there , three facts are true
. first , in the large limit ( and after rescaling ) , a well - defined typical shape exists .
second , the expectation of the lis and lds of a large random permutation is encoded respectively in the length of the first row and column of the limit shape .
third , the plancherel measure is concentrated near the typical shape .
we conjecture that analogues of all three of the aforementioned features also hold for the plancherel - hecke measure . to be more precise , let the * typical shape * @xmath67 be the shape @xmath12 ( contained in @xmath68 ) maximizing @xmath54 .
this young diagram @xmath67 can be interpreted as a step - function .
it can furthermore be associated to a piecewise linear approximation @xmath69 .
finally , rescale by @xmath70 [ conj : thebigone ] * for any @xmath71 , there is a unique continuous function @xmath72 such that for any @xmath73 , @xmath74 as @xmath61 .
we call this @xmath75 the * limit typical shape*. * a `` phase transition '' occurs at @xmath76 : + for @xmath77 , @xmath75 is the line @xmath78 + for @xmath79 , @xmath75 is the * plancherel curve * , which is parametrically given by @xmath80 ( the curves defined by ( [ eqn : line ] ) and ( [ eqn : vershikcurve ] ) are declared to be identically @xmath81 for @xmath82 . ) * for @xmath66 : there is a constant @xmath83 such that if @xmath84 then if @xmath85 , @xmath75 is given by ( [ eqn : line ] ) .
otherwise , @xmath75 is given by a deformation of ( [ eqn : vershikcurve ] ) which is symmetric across the line @xmath86 .
in either case , the @xmath87 and @xmath88 intercepts are at @xmath89 where @xmath90 and explicitly , @xmath91 we have reasonable support for the cases ( i ) and ( ii ) of conjecture [ conj : thebigone ] .
_ heuristically _ , part ( ii ) of the conjecture says that when @xmath21 is large , and thus @xmath92 is `` close '' to @xmath2 , a random word is `` close to '' being a random permutation . for a permutation , schensted and
hecke behave the same .
hence the plancherel and plancherel - hecke measures ought to be maximized on the same shape .
when @xmath21 is small , the limit typical shape is a rescaling of the `` staircase shape '' which plays a distinguished role in the edelman - greene algorithm @xcite and the hecke algorithm ; see theorem [ thm : desc ] and its proof .
conjecture [ conj : thebigone](iii ) is more speculative , since we did not have as much computational evidence for the shape of @xmath75 .
there appears to be a continuous `` flattening '' of the plancherel curve to a line as we tune @xmath93 from @xmath94 to @xmath81 .
our data was insufficient to rule out the possibility that @xmath75 is simply a rescaling of the plancherel curve by a factor of @xmath95 .
explicitly describe the deformation of @xmath75 when @xmath66 , as @xmath93 varies .
our best estimate is that @xmath96 ( probably just @xmath97 ) .
however , the values of @xmath95 for @xmath93 relatively small can be experimentally estimated .
the table below was based on monte carlo estimates of @xmath98 for @xmath99 and @xmath100 and the estimates were stable throughout this range .
they closely agree with the conjecture for @xmath95 given above .
.estimates of @xmath95 for the @xmath66 case [ cols="<,<,<,<,<,<",options="header " , ] figure [ fig : pilesize ] shows that there is a definite shape describing the mean pile sizes as @xmath21 varies .
questions about the structure of this shape may be interpreted as enriched questions related to lis .
one can analyze such questions using the dichotomy of section 1.4 ; we do not pursue this here .
now we show that the size of the first row of @xmath101 computes @xmath4 .
let @xmath102 be the largest index such that the longest strictly increasing subsequence ending at @xmath103 has length @xmath104 .
for example , if @xmath105 then @xmath106 , @xmath107 and @xmath108 .
we now in fact prove the following claim , which is stronger than the lis assertion of theorem [ thm : hard ] ( cf .
7.23.10 ) ) : [ prop : refined_firstrow ] suppose @xmath15 , @xmath109 and @xmath110 is the insertion tableau of @xmath111 .
then the first row of @xmath110 is given by @xmath112 .
by induction on @xmath2 .
the base case @xmath113 is trivial .
suppose that the claim holds for @xmath114 . thus if @xmath115 is the insertion tableau of @xmath116 then by induction , the first row is given by @xmath117 where @xmath118 .
we consider the possibilities of what happens as we insert @xmath119 into the first row of @xmath115 .
we prove the desired conclusion holds for the case that @xmath120 for some maximally chosen @xmath121 ; other cases are similar .
so after inserting @xmath119 , the first row of @xmath110 is @xmath122 the assumption shows that the longest increasing subsequence @xmath123 in @xmath10 using @xmath119 is of length at least @xmath124 , since we can adjoin @xmath119 to the length @xmath104 subsequence ending at @xmath125 . on the other hand , @xmath123 is of length at most @xmath124 since if it were , say , of length @xmath126 , there must be a length @xmath124 increasing subsequence in @xmath127 ending at @xmath128 with @xmath129 and @xmath130 .
but this is a contradiction of the definition of @xmath131 since we would then have a length @xmath126 increasing subsequence ending at @xmath132 .
since we have just shown @xmath133 , it is now clear that @xmath134 for @xmath135 .
thus , the first row of @xmath110 satisfies the desired claim , and the induction follows . in order to prove that the first column of @xmath12 computes @xmath9
, we need to draw a connection to @xcite , where we developed a @xmath0-theoretic _ jeu de taquin _ theory . rather than repeat the setup in full here , for brevity , we refer the reader to that paper for the complete background on @xmath0-rectification and @xmath0-infusion used below . although what follows also constitutes a proof of our lis claims , we felt that including the direct proof via the stronger claim of proposition [ prop : refined_firstrow ] was worthwhile
. however , a similarly direct proof of our lds claim seems harder .
let @xmath136 be the staircase shape . also , let @xmath137 be the * permutation shape * consisting of @xmath2 single boxes arranged along an antidiagonal . given @xmath15 ,
define @xmath138 to be the tableau where @xmath139 is arranged from southwest to northeast .
also let @xmath140 be the * superstandard * young tableau , i.e. , the one whose first row is labeled @xmath141 , the second row is labeled by @xmath142 etc .
this latter tableau determines a _ particular _
@xmath0-rectification of @xmath143 , which by definition is @xmath0-@xmath144 , see ( * ? ? ?
* section 3 ) .
( an important subtlety in @xmath0-theoretic jeu de taquin is that @xmath0-rectification depends on the order in which it is performed , unlike the rectification of classical jeu de taquin .
however , the order defined by @xmath145 is particularly nice . )
the following result is an analogue of a classical result linking the robinson - schensted algorithm to the ( ordinary ) rectification of @xmath146 : [ thm : hecke_is_rect ] let @xmath15
. then @xmath0-@xmath144 is the insertion tableau of @xmath111 .
we induct on @xmath2 .
the base cases @xmath147 are easy .
we may assume that the steps of the @xmath148 that are defined by the `` inner '' labels @xmath149 results in a skew shape of the form @xmath150 , as depicted below : @xmath151 the induction hypothesis is that @xmath115 is the insertion tableau obtained by hecke inserting @xmath152 .
( in the depiction of @xmath115 from ( [ eqn : depiction ] ) , some of the boxes with labels @xmath153 may be empty . )
the non - underlined labels occupy the boxes of @xmath150 , whereas the underlined labels dictate the remaining steps to perform to complete the @xmath0-@xmath154 computation ( these steps are recalled below ) .
hence it remains to show that the tableau obtained by the hecke - insertion @xmath155 is the same as carrying out the @xmath0-@xmath154 indicated by ( [ eqn : depiction ] ) , i.e. , the operation @xmath156 to do this , we first develop a technical fact . in (
* ? ? ? * section 1.1 ) , we defined the procedure switch , which we restate now ( in a more convenient form ) .
let @xmath157 be the set of * mixed tableaux * , which , by definition , are tableaux of shape @xmath21 , each of whose boxes is filled with an entry from one of two alphabets , @xmath158 and @xmath159 , such that , within each row or column , the entries for each alphabet appear at most once .
( no increasingness condition is demanded . )
we also include the * null tableau * @xmath160 , as a special element of @xmath157 .
define an operator @xmath161 as follows .
given @xmath162 , consider the subshape @xmath145 of @xmath163 consisting of boxes whose entry is either @xmath164 or @xmath165 .
for each non - singleton connected component of @xmath145 , interchange the @xmath164 s and the @xmath165 s .
if this results in a ( non null ) mixed tableau , then the result is that tableau .
otherwise the result is @xmath160 . by definition @xmath166 .
[ eqn : a_switch ] let @xmath167 and @xmath168 .
then @xmath169 is given below , together with two different switch computations applied to it : @xmath170 on the other hand , @xmath171 the following lemma is easy to verify from the definitions : [ lemma : viable ] if @xmath172 and @xmath173 then the operators @xmath174 and @xmath175 commute , i.e. , @xmath176 is a relation in the algebra generated by switch operators on @xmath157 .
the procedure described in ( * ? ? ?
* section 3 ) for computing @xmath0-@xmath177 is to consider the entries of @xmath178 as being underlined , where the maximum entry of @xmath178 is @xmath179 , say , and the entries of @xmath180 as not underlined , where the maximum entry of @xmath180 is @xmath92 .
now perform the following sequence of @xmath181 operations , from left to right , as indexed by : @xmath182 we refer to this sequence of pairs ( interchangeably , the corresponding sequence of switch operators ) as the * standard switch sequence*. the technical fact we need is that @xmath0-@xmath183 can in fact be computed differently : a switch sequence is called * viable * if it is a `` shuffling '' of ( [ eqn : std ] ) , in the following sense : * every @xmath184 occurs exactly once , for @xmath185 and @xmath186 ; * for any @xmath185 , the pairs @xmath187 occur in that relative order ; and * for any @xmath186 the pairs @xmath188 occur in that relative order . this definition is explained by the proof of the following proposition : [ prop : viable ] any viable switch sequence can be used to calculate @xmath0-infusion . it is straightforward to show that one can obtain any viable switch sequence from the standard switch sequence ( [ eqn : std ] ) by repeated applications of the commutation relation of lemma [ lemma : viable ] . thus , in view of proposition [ prop : viable ] , to complete the induction it suffices to construct a viable switch sequence whose result is the same as @xmath189 .
( a caution : in @xcite it was shown that the standard switch sequence necessarily maintains increasingness along rows and columns of the members of each alphabet . we will not prove that a viable switch sequence also achieves this during the intermediate steps of a @xmath0-infusion .
however , this does not play a logical role in how we apply proposition [ prop : viable ] below . )
let @xmath190 , and for @xmath191 , let @xmath192 be the number which is inserted in row @xmath193 according to hecke insertion of @xmath119 into @xmath115 . during a bumping step of hecke insertion ,
let @xmath194 be the smallest number already in row @xmath193 which is greater that @xmath192 .
we say that a mixed tableau , obtained after some number of switch operations applied to @xmath195 , is in * row @xmath193 normal form * if : * the @xmath193-th row is of the form @xmath196 where @xmath192 has not yet moved from its initial position in column @xmath93 , or @xmath197 depending , respectively , on whether the hecke insertion @xmath189 terminates at row @xmath193 or after , or strictly earlier . here
the @xmath193-th row of @xmath115 has length @xmath104 ; and * all non - underlined symbols in rows @xmath198 and below have not moved from their initial positions .
having explained our general strategy , what remains is some tedious but straightforward case analysis to describe the viable switch sequence we use : our initial mixed tableau is in row @xmath199 normal form . now , suppose we have arrived , after some sequence of applications of ( a ) , ( b ) and ( c ) below , at a mixed tableau in row @xmath193 normal form .
there are three possibilities for a bumping step of hecke insertion of @xmath192 .
\(a ) _ @xmath192 is inserted , bumping @xmath194 : _ consider the switch sequence that moves @xmath192 to the left along row @xmath193 : specifically , using @xmath200 until it is directly above the @xmath194 in row @xmath198 , and then , starting from the right , swaps each box in row @xmath193 having an underlined label with the one directly below , which has a non - underlined label ( this can always be done since no label numerically equal to @xmath192 appears among the latter boxes , by assumption ) .
the result is therefore in row @xmath198 normal form since @xmath194 does nt move in this process , it is the unique box with a non - underlined label in row @xmath198 , and @xmath201 , as demanded .
note that the non - underlined labels of the @xmath193-th row of the mixed tableau we obtain after this process is the same as the @xmath193-th row of @xmath115 , with @xmath194 replaced by @xmath192 , as desired .
if @xmath202 and we started with @xmath203 so that @xmath204 and @xmath205 , then we begin by moving @xmath192 above @xmath206 : @xmath207 we conclude with @xmath208 , @xmath209 and then finally @xmath210 , resulting in @xmath211 which is in row @xmath198(@xmath212 ) normal form .
moreover , the non - underlined labels in row @xmath202 of this mixed tableau , namely @xmath213 agrees with the first row of @xmath214 , as desired .
\(b ) _ @xmath192 is not inserted because of a horizontal violation : _ ( that is , a label numerically equal to @xmath192 already appears in the row that we are hecke inserting into . ) proceed as in ( a ) by moving @xmath192 to the left , until it is directly above @xmath194 , i.e. , apply ( [ eqn : casea_startmoves ] ) . now , `` locally '' the situation in the column containing @xmath194 and the column to its left is : @xmath215 for some @xmath104 .
the former case shows rows @xmath193 and @xmath198 when @xmath216 does not appear in row @xmath217 , while the latter case also includes row @xmath217 if it does . to the right of the column containing @xmath194
, we swap boxes with underlined and non - underlined labels , as in ( a ) . then we perform the transformation @xmath218 respectively . after this transformation , we complete by swapping , right to left , the boxes to the left of the @xmath192 , as in ( a ) .
the result is in the demanded row @xmath198 normal form .
note that unlike ( a ) , row @xmath193 still has a box @xmath219 with an underlined label in it .
however , when we work on the row @xmath198 normal form , the reader can check that our descriptions will force that box @xmath219 to be filled by @xmath194 : e.g. , if we apply case ( a ) next , this will occur when we execute the switches ( [ eqn : casea_startmoves ] ) , if we apply case ( b ) next , this will occur when we execute either the switches ( [ eqn : casea_startmoves ] ) or ( [ eqn : case2_trans ] ) , whereas if we apply ( c ) next , the replacement will occur during the switches ( [ eqn : case3_trans ] ) below .
hence , in the end , we find that the non - underlined labels in row @xmath193 are the same as the @xmath193-th row of the hecke insertion of @xmath119 into @xmath115 , as desired
. if @xmath202 and we started with @xmath220 then moving the `` @xmath221 '' in row @xmath199 to the left , as in ( a ) , gives @xmath222 the remaining swaps give @xmath223 and the latter is in row @xmath224 normal form .
the mild complication of this case , as suggested above , is that row @xmath199 is not yet the same as the first row of the insertion tableau of hecke .
however , as we begin to work on this row @xmath221 normal form , we start by using @xmath225 ( beginning to move the @xmath226 in row @xmath221 to the left ) , giving : @xmath227 hence row @xmath199 now does agree with hecke insertion .
\(c ) _ @xmath192 will not be inserted because of a vertical violation ( and a horizontal violation does not occur ) : _ while hecke inserting @xmath192 into row @xmath193 of @xmath115 , @xmath228 is directly below a label ( numerically equal to ) @xmath192 that is in row @xmath217 .
this prevents one from replacing @xmath194 by the @xmath192 we are inserting .
( note that this case can only occur if @xmath191 . ) moreover , since there is no horizontal violation , the number immediately to the left of @xmath194 , say @xmath229 , satisfies @xmath230 .
notice , in order for a vertical violation to occur , the previous step of hecke inserting @xmath231 must have been an instance of case ( b ) or ( c ) ( since we are hecke inserting a number into row @xmath193 which also appears in row @xmath217 ) .
hence the box directly above and to the left of the @xmath192 contains an underlined label @xmath232 .
now we begin by switching all the boxes with underlined labels in row @xmath193 to the right of @xmath192 , with the non - underlined labels directly below them .
the remaining switches , which involve the rows @xmath217 to @xmath198 , in the column of @xmath192 and the column to its immediate left , are as given as follows : @xmath233 as in ( b ) , we finish by completing a sequence of swaps involving the columns to the left of the @xmath229 .
the result of this process is a tableau in row @xmath198 normal form . as at the conclusion of ( b )
, row @xmath193 of the resulting row @xmath198 normal form mixed tableau , still contains an underlined label . as in that case , this label will be switched with @xmath194 , as desired , during the forthcoming switches .
[ exa : ccase ] to give an example of ( c ) , the previous insertion must have been of type ( b ) or ( c ) , so consider the following example : @xmath234 after the first step , an insertion of type ( b ) , we reach row 2 normal form : @xmath235 as compelled by the conditions of a viable switch sequence , we switch @xmath236 with @xmath237 before we switch @xmath236 with 5 , and as a result , as described above , we get to row 3 normal form : @xmath238 the final result is : @xmath239 again , this tableau agrees with @xmath240 . note that after each of ( a ) , ( b ) and ( c ) , when we reach row @xmath198 normal form , @xmath241 is necessarily in a column weakly to the left of @xmath242 , because @xmath115 is an increasing tableau and @xmath242 together with the rows strictly below row @xmath198 are , at this point , still unaltered .
this observation guarantees that the switches as described above can actually be executed , i.e. , the above descriptions are well - defined . using similar analysis one can give switch sequences for the terminating steps of hecke insertion , such that one maintains row @xmath193 normal form for all @xmath243 , where @xmath6 is the number of rows of @xmath115 .
we leave the straightforward details to the reader .
these constructions then show , by induction on the number of rows @xmath6 , that we have a sequence of switch operations transforming @xmath150 into @xmath189 .
_ conclusion of the proof of theorem [ thm : hecke_is_rect ] : _ by the fact that @xmath115 is an increasing tableau , and by the definition of normal form , it is easy to see that the sequence of @xmath181 operations used forms a viable sequence , after suitable insertions of any trivial @xmath244 operators ( that is to say , switch operations which do not have any effect on the tableau )
. we can therefore apply proposition [ prop : viable ] as we claimed earlier .
continuing example [ exa : ccase ] , the switch sequence we obtain , by following the descriptions of the cases ( b ) and ( c ) that are needed is : @xmath245 this is not quite a viable sequence : although our constructions guarantee that it satisfies the second and third conditions to be a viable sequence , it fails the first , since , e.g. , @xmath246 does nt appear in the sequence , since this switch is never needed .
however , clearly we can simply insert this trivial switch , along with the others that are missing , giving the viable sequence : @xmath247 @xmath248 the action of this viable sequence on the original mixed tableau is therefore the same as the original switch sequence , which we highlight in boldface .
this viable sequence also happens to be the standard switch sequence , although it need nt be in general . hence @xmath249 in agreement with theorem [ thm : hecke_is_rect ] . in ( * ? ? ?
* theorem 6.1 ) we showed that the first row of @xmath0-@xmath250 has length @xmath4 , for any increasing tableau @xmath251 of shape @xmath252 .
so by theorem [ thm : hecke_is_rect ] , the first row of @xmath253 has length @xmath4 . by symmetry , (
* theorem 6.1 ) also implies that the first column of @xmath0-@xmath250 has length @xmath254 .
hence the lds claim follows .
this completes the proof of theorem [ thm : hard ] .
given @xmath255 , define @xmath256 .
the following is symmetry statement is immediate from theorem [ thm : main ] , since @xmath257 : [ thm : partialsymm ] let @xmath101 and @xmath258 .
then @xmath259 and @xmath260 where @xmath261 and @xmath262 are the conjugate shapes of @xmath12 and @xmath263 respectively .
a warning is needed : unlike robinson - schensted correspondence setting , with hecke , one _ can not _ conclude that the insertion tableaux associated to @xmath255 and @xmath256 differ only by a reflection across the main diagonal .
a counterexample is @xmath264 .
( in @xcite , the symmetry property of the robinson - schensted correspondence , was applied to prove the lds claim in the classical version of theorem [ thm : hard ] . )
give an explicit description of @xmath265 in terms of @xmath111 .
finally , greene @xcite has given an explanation of the other rows of the shape @xmath12 associated to a permutation @xmath266 under the robinson - schensted correspondence : @xmath267 equals the maximal size of a union of @xmath193 disjoint increasing subsequences of @xmath10 .
however , we could not find any extension of greene s theorem in the hecke context .
the naive tries do not work : since @xmath268 , the simplest case to analyze is when @xmath269 .
the example @xmath270 corresponds to @xmath271 ; this shows that it is not valid to merely replace `` increasing '' by `` strictly increasing '' in greene s theorem , since that would predict @xmath272 .
_ proof of theorem [ thm : desc ] : _ let @xmath273 be the word in @xmath274 of maximal coxeter length . hence @xmath275 .
this is the unique permutation in @xmath274 with this length .
we need the following lemma , which characterizes when @xmath14 is maximized .
[ lemma : w0 ] for @xmath15 , @xmath276 if and only if @xmath277 .
first suppose @xmath277 . under @xmath111 ,
the insertion tableau @xmath110 satisfies @xmath278 .
hence the shape of @xmath110 has at least @xmath279 boxes and so by proposition [ prop : central ] it must be @xmath280 .
conversely , if @xmath276 , then note that there is a unique increasing filling @xmath110 of that shape ( using @xmath281 in the first row , @xmath282 in the second row , etc ) . then it is well - known that @xmath278 . in view of the lemma [ lemma : w0 ] , the theorem will follow if we can show that @xmath283 ( we conjecture this to be true whenever @xmath284
. this would imply conjecture [ conj : thebigone ] for this entire range . )
. then either @xmath286 depending on whether the simple reflection @xmath287 ( say equal to @xmath288 ) is an * ascent * of @xmath289 or not . ( an * ascent * occurs at a position @xmath104 for a permutation @xmath290 if @xmath291 . )
provided that @xmath292 , @xmath290 has at least one ascent .
thus when @xmath293 , the probability that @xmath294 s introduction increases the coxeter length is at least @xmath295 .
let @xmath296 related to this , let @xmath297 be bernoulli distributed with parameter @xmath295 .
set @xmath298 clearly , @xmath299 we now show that when @xmath300 the righthand side of the inequality ( [ eqn : comparisonineq ] ) goes to zero as @xmath301 .
this is a simple application of ( a special case of ) bennet s large deviation inequality , see , e.g. , ( * ? ? ?
2.4.7 ) : suppose @xmath302 are independent , mean zero random variables with @xmath303 .
set @xmath304 .
then for @xmath305 we have @xmath306 to apply this to our setting , let @xmath307 .
hence @xmath308 .
then with @xmath309 @xmath310 the result then follows . in the above argument
, we interpreted @xmath15 as a random walk in @xmath274 that begins at the identity and works its way up in the weak bruhat order to @xmath311 . at each step
the probability of going up is _ at least _
@xmath295 ( as we have used ) , but is larger in general . however ,
since this probability varies , even for permutations with the same coxeter length , a more refined analysis is needed to push the argument we have used further , up towards @xmath76 .
_ proof of theorem [ thm : main ] : _ for @xmath312 we will apply an argument similar to that for theorem [ thm : desc ] .
given @xmath313 let @xmath314 let @xmath315 and set @xmath316 provided @xmath317 occurs , then @xmath318 with probability @xmath295 , and is equal to @xmath319 otherwise .
let @xmath320 and @xmath321 be discrete random variables , where @xmath322 is bernoulli distributed with parameter @xmath295 .
now , @xmath323 thus it will be enough to show that when @xmath324 then @xmath325 this is another application of the large deviation inequality ( [ eqn : bennet ] ) .
for @xmath326 we use a proof provided for us by o. zeitouni : @xmath327 , the expected length of the longest weakly increasing subsequence of @xmath15 ( with @xmath328 ) is known to satisfy @xmath329 see ( * ? ? ?
* theorem 1.7 ) .
the argument shows that the difference between the lis and lwis of @xmath10 is typically small .
let @xmath330 be the random variable for the value of @xmath331 of a random uniform word @xmath332 where @xmath333 is the integer part of @xmath334 , etc .
similarly define @xmath335 where @xmath13 replaces @xmath23 .
fix @xmath73 and let @xmath336 be large enough such that @xmath337 we need a `` graphical '' representation of a word in @xmath1 : consider a @xmath338 rectangle subdivided into unit squares
. in each of the @xmath2 columns , one places a single `` dot '' in one of the @xmath92 rows .
the set of such configurations is in obvious bijection with words in @xmath1 . given @xmath339 , draw @xmath340 smaller rectangles of dimension @xmath341 along an antidiagonal inside the @xmath338 rectangle , as depicted in figure [ fig : zeit ] below .
@xmath342 label the @xmath193-th southwest most box @xmath343 .
let @xmath344 be the random variable giving the number of dots in @xmath343 .
notice that the @xmath344 s are independent .
also , the dots inside @xmath343 define a word , and we can speak of @xmath345 and @xmath346 , the length of the longest strictly ( respectively , weakly ) increasing subsequence of that word . say that @xmath343 is * good * if the following conditions simultaneously hold : * @xmath347 ; * @xmath348 ; and * no two dots in @xmath343 have the same height ( hence @xmath349 ) .
now , we have @xmath350 and we claim for an @xmath351 sufficiently large , for @xmath352 , and for all @xmath2 large , we have : @xmath353 the proof is a standard argument : let @xmath354 be the indicator random variable which evaluates to @xmath199 if column @xmath93 has a dot that lies in the box @xmath343 that occupies that column ( hence with probability @xmath355 ) , and evaluates to @xmath81 otherwise . hence @xmath356 where the previous line is an application of chebyshev s inequality .
now take @xmath339 sufficiently large ( bigger than some @xmath351 ) so @xmath357 . assuming @xmath358 , if the event ( a ) occurs , then with probability at least @xmath359 , when @xmath2 is large , ( b ) holds , because of the definition of @xmath360 using ( [ eqn : inflim ] ) .
the probability of the event ( c ) not occurring is bounded above ( using a union bound ) by @xmath361 because @xmath362 .
so , @xmath363 for @xmath364 and @xmath2 large .
another standard argument with chebyshev s inequality shows with high probability , say at least @xmath359 , and for @xmath2 large , the number of good boxes is @xmath365 hence , with that probability , for @xmath15 @xmath366 since @xmath367 the @xmath328 case follows by taking @xmath368 , completing the proof of the theorem .
the goal of this appendix is to present a proof of the following result : [ thm : appendix ] let @xmath369 .
then @xmath90 where @xmath370 however , in order to prove this statement , we need to work with another variant of plancherel measure , utilized , e.g. , by @xcite and alluded to in section 1.5 of the main text .
our approach parallels the one developed in @xcite to prove @xmath371 in the permutation case , by utilizing work of @xcite .
a * semistandard young tableau * of shape @xmath372 with labels from @xmath373 is a filling of the young shape @xmath12 with these labels so that the entries weakly increase along rows , and strictly increase along columns .
for example , if @xmath374 and @xmath375 there are two such tableaux : @xmath376 and @xmath377 .
let @xmath378 denote the number of such tableaux .
define the * plancherel - rsk measure * @xmath379 on the set @xmath380 of young diagrams @xmath12 with @xmath2 boxes , by declaring that a random young shape @xmath34 occurs with probability @xmath381 we make no claims of originality in this definition .
indeed , this is the same measure studied in , e.g. , @xcite ; although there the measure is defined in terms of dimensions of irreducible @xmath11 and @xmath382 modules associated to @xmath12 ; the equivalence is well - known . the fact that @xmath379 is in fact a probability distribution follows from either schur - weyl duality , as in section 1.5 , or by the rsk algorithm , see , e.g. , ( * ? ? ?
* section 7.11 ) .
a crucial advantage of @xmath379 for the purposes of understanding @xmath98 , in comparison to plancherel - hecke measure , is that both @xmath57 and @xmath378 have simple multiplicative formulas .
this makes it more readily analyzed using ideas of @xcite , which we modify to the present setting .
given a box @xmath383 , define the * hook - length * associated to @xmath384 to be @xmath385 where @xmath386 is the number of boxes strictly to the right of @xmath384 , and in the same row , and @xmath387 is the number of cells strictly below @xmath384 and in the same column .
then we have ( * ? ? ?
* chapter 7 ) , the * hook - length formula * and * hook - content formulas * , respectively : @xmath388 where in the second formula @xmath389 is the * content * of @xmath384 , the column index of @xmath384 minus the row index of @xmath384 .
so for example , if @xmath390 , the contents are given by @xmath391 * young s lattice * is the poset structure on @xmath26 where @xmath392 if the shape of @xmath12 is contained in the shape of @xmath263 .
we write @xmath393 to denote a covering relation in this poset , i.e. , where @xmath263 is obtained from @xmath12 by adding a single box at a corner .
define a markov process on @xmath26 with the transition probabilities @xmath394 we need the following lemma , that in particular shows that plancherel - rsk measure is a markov measure with the above transition probabilities . *
@xmath395 * @xmath396 the claim ( i ) is equivalent to @xmath397 this follows from the following _ pieri rule _ for schur polynomials @xmath398 see ( * ? ? ?
* theorem 7.15.7 ) . here
@xmath399 is the * schur polynomial * , where the sum is over all semistandard young tableaux of shape @xmath12 with entries from @xmath3 , @xmath400 , and @xmath401 is the number @xmath165 s used in @xmath163 . in particular , ( i ) is immediate from @xmath402 . for ( ii ) , the claim is @xmath403 that is , @xmath404 , which is well - known ( and straightforward from the definitions ) .
work of biane ( * ? ? ?
* theorem 3 ) describes the typical shape under plancherel - rsk after the rescaling @xmath405 .
biane s theorem implies that @xmath406 but not @xmath407 briefly , we explicate how his work applies to our situation ( the reader is directed to the original source for details ) : biane works with the coordinate axes rotated @xmath408-degrees counterclockwise , as in , e.g. , @xcite .
his aforementioned theorem states that if @xmath409 is any sequence of ( rescaled and rotated ) young diagrams , then for any @xmath73 we have @xmath410 where the probability is computed with respect to @xmath379 .
also , @xmath411 is biane s limit shape , which has the property that it meets the line @xmath86 at a distance @xmath95 from the origin . in other words ,
the `` first column '' @xmath412 of @xmath411 satisfies @xmath413 for each @xmath2 , let @xmath414 be the length of the first column of @xmath415 .
from ( [ eqn : bianesresult ] ) it follows that for any @xmath73 , @xmath416 moreover , since it is known that for any @xmath15 the first column of the young diagram associated to @xmath417 equals @xmath9 , ( [ eqn : leq ] ) follows immediately from ( [ eqn : c ] ) and ( [ eqn : cn ] ) combined .
note that the above argument does not also prove ( [ eqn : geq ] ) since ( [ eqn : bianesresult ] ) does not rule out the possibility that @xmath409 consists of young diagrams with `` tails '' along the @xmath86 axis that both `` lengthen '' and `` thin out '' as @xmath418 .
therefore , it remains to verify ( [ eqn : geq ] ) . to do this ,
we modify an argument found in @xcite , which establishes the analogous assertion in the permutation case : consider the set @xmath419 of all sequences of young diagrams @xmath420 where @xmath421 for @xmath422 . for a young diagram @xmath12 ,
let @xmath423 denote the diagram obtained by adding a single box to @xmath12 , in the first column .
for each integer @xmath422 , define the indicator function @xmath424 by setting @xmath425 if @xmath426 , and setting @xmath427 otherwise .
studying the expectation of @xmath428 we have : @xmath429 where we have just used @xmath430 let @xmath431 note that by the hook - content formula we have @xmath432 where @xmath433 is the length of the first column of @xmath12 .
summarizing , we have @xmath434 where @xmath435 denotes the expectation of @xmath436 , i.e. , the expected length of the first column of a random shape with @xmath193 boxes , drawn under the plancherel - rsk measure .
notice also that since @xmath428 is an indicator random variable , we have @xmath437 therefore , combining ( [ eqn : summ1 ] ) and ( [ eqn : summ2 ] ) we obtain , by the cauchy - schwarz inequality , the following difference inequality : @xmath438 we claim that @xmath439 . to prove this , note the following facts about @xmath435 : * @xmath440 ; and * @xmath441 .
now define a linear interpolation : for @xmath442 $ ] , set @xmath443 note that for such @xmath104 , @xmath444 .
therefore , combining equations ( [ eqn : crucial1 ] ) and ( [ eqn : crucial2 ] ) we obtain @xmath445 since @xmath446 we have @xmath447 , hence the above differential inequality is equivalent to @xmath448
hence it follows that @xmath449 that is , @xmath450 now we care about @xmath451 .
we always have @xmath452 ( trivially ) , but for @xmath453 we have the better inequality @xmath454 .
therefore , @xmath455 for @xmath456 and @xmath457 for @xmath453 .
the result then follows .
ht was supported by an nserc discovery grant .
ay was supported by nsf grant dms 0601010 and a u. minnesota dtc grant during the spring 2007 ; he also utilized the resources of the fields institute , and of algorithmics incorporated , in toronto , while a visitor .
we would like to thank ofer zeitouni for allowing us to include his proof for the @xmath328 case of theorem [ thm : main ] and for his extensive help during this project .
we also thank alexander barvinok , nantel bergeron , alexei borodin , sergey fomin , christian houdr , nicolas lanchier , igor pak , eric rains , mark shimozono , richard stanley , dennis stanton , craig tracy and alexander woo for helpful correspondence .
bukrshtayo06 d. aldous and p. diaconis , _ longest increasing subsequences : from patience sorting to the baik - deift - johansson theorem _ , bull .
36*(1999 ) , 413432 .
j. baik , p. deift and k. johansson , _ on the distribution of the length of the longest increasing subsequence of random permutations _
, j. amer .
12*(1999 ) , 11191178 .
p. biane , _ approximate factorization and concentration for characters of symmetric groups _ ,
notices * 2001*(2001 ) , no . 4 , 179192 .
a. bjrner and f. brenti , _ combinatorics of coxeter groups _ , graduate texts in mathematics * 231 * , springer , new york , 2005 .
a. borodin , a. okounkov and g. olshanski , _ asymptotics of plancherel measures for symmetric groups _ , j. amer .
, * 13*(2000 ) , 481515 .
a. borodin and g. olshanski , _ asymptotics of plancherel - type random partitions _ , j. algebra * 313*(2007 ) , 4060 .
a. buch , _ a littlewood - richardson rule for the @xmath0-theory of grassmannians _ , acta math . , * 189 * ( 2002 ) , 3778 .
a. buch , a. kresch , m. shimozono , h. tamvakis and a. yong , _ stable grothendieck polynomials and @xmath0-theoretic factor sequences _ , math .
ann . , to appear , 2008 . .
a. dembo and o. zeitouni , _ large deviations and applications _ , handbook of stochastic analysis and applications , 351416 , statist .
textbooks monogr .
, * 163 * , dekker , new york , 2002 .
p. edelman and c. greene , _ balanced tableaux _ , adv .
63*(1987 ) , no . 1 , 4299 .
p. erds and g. szekeres , _ a combinatorial problem in geometry _ , composito math . * 2*(1935 ) , 463470
. s. fomin and c. greene , _ noncommutative schur functions and their applications _ , discrete math . *
193*(1998 ) , no . 1 - 3 , 179200 , selected papers in honor of adriano garsia ( taormina , 1994 ) . c. greene , _ an extension of schensted s theorem _ , adv
. math . * 14*(1974 ) , 254265 .
j. m. hammersley , _ a few seedlings of research_. in _ proc .
sixth berkeley symp . math .
statist . and probability , volume 1 _ , pp .
university of california press , 1972 .
c. houdr and t. litherland , _ on the longest increasing subsequence for finite and countable alphabets _ , preprint .
k. johansson , _ discrete orthogonal polynomial ensembles and the plancherel measure _ , ann .
( 2 ) * 153*(2001 ) , no . 1 , 259296 .
d. e. knuth , _ permutations , matrices and generalized young tableaux _ , pacific j. math . * 34*(1970 ) , 709727 .
a. lascoux and m. -p .
schtzenberger , _ structure de hopf de lanneau de cohomologie et de lanneau de grothendieck dune varit de drapeaux _ , c. r. acad .
* 295 * ( 1982 ) , no . 11 , 629633 . b. f. logan and l. a. shepp , _ a variational problem for random young tableaux _ , adv
. math . *
26*(1977 ) , 206222 . c. l. mallows , _ patience sorting _ , bull . inst .
appl . * 9*(1973 ) , 216224 . c. schensted , _
longest increasing and decreasing subsequences _ , canad . j. math . * 13*(1961 ) , 179191 r. p.
stanley , _ increasing and decreasing subsequences and their variants _ , proceedings of the international congress of mathematicians , madrid , spain , 2006 .
, _ enumerative combinatorics , volume 2 _ ( with an appendix by s. fomin ) , cambridge university press , 1999 . , _ polygon dissections and standard young tableaux _ , j. combin . theory ser .
a * 76*(1996 ) , 175177 .
h. thomas and a. yong , _ a jeu de taquin theory for increasing tableau , with applications to @xmath0-theoretic schubert calculus _ , preprint . c. tracy and h. widom , _ on the distributions of the lengths of longest monotone subsequences in random words _
, probab .
theory relat .
fields * 119*(2001 ) , 350380 . a. m. vershik and s. v. kerov , _ asymptotic behavior of the plancherel measure of the symmetric group and the limit form of young tableaux _ ,
nauk sssr * 223 * ( 1977 ) 10241027 ; english translation : soviet math
. dokl . * 233*(1977 ) , 527531 . , _ asymptotic behavior of the maximum and generic dimensions of irreducible representations of the symmetric group _ , funktsional .
i prilozhen .
* 19 * ( 1985 ) , no .
1 , 2536 . | we define and study the _ plancherel - hecke probability measure _ on young diagrams ; the _ hecke algorithm _ of [ buch - kresch - shimozono - tamvakis - yong 06 ] is interpreted as a polynomial - time exact sampling algorithm for this measure . using the results of [ thomas - yong 07 ] on _ jeu de taquin _ for _ increasing tableaux _ , a symmetry property of the hecke algorithm is proved , in terms of longest strictly increasing / decreasing subsequences of words .
this parallels classical theorems of [ schensted 61 ] and of [ knuth 70 ] , respectively , on the _ schensted _ and _ robinson - schensted - knuth _ algorithms .
we investigate , and conjecture about , the limit typical shape of the measure , in analogy with work of [ vershik - kerov 77 ] , [ logan - shepp 77 ] and others on the `` longest increasing subsequence problem '' for permutations .
we also include a related extension of [ aldous - diaconis 99 ] on _ patience sorting_. together , these results provide a new rationale for the study of increasing tableau combinatorics , distinct from the original algebraic - geometric ones concerning @xmath0-theoretic schubert calculus . |
the supernova remnant ( snr ) 1e0102.27219 ( hereafter e0102 ) is the brightest x - ray snr in the small magellanic cloud ( smc ) and was discovered with _ einstein _ @xcite . in x - rays , it has a ring - like appearance and its emission is dominated by h - like and he - like ions of o and ne .
it consists of a bright x - ray ring and a much fainter plateau with a sharp outer edge that surrounds the ring ( fig.[images ] ) .
this bright x - ray ring is caused by the reverse shock of the snr , while the outer boundary of the faint plateau marks the location of the blast wave .
the diameter of the bright ring is @xmath6.03em28 , and the diameter of the outer rim is @xmath6.03em44 .
the radio emission in the 6 cm wavelength band @xcite is located outside the bright x - ray emission but within the outer faint x - ray rim @xcite . in the optical band
, there is diffuse h@xmath1 emission surrounding the remnant , but a distinctive h@xmath1 ` hole ' at the location of the snr .
a filamentary shell of optical [ oiii ] emission with a diameter of @xmath6.03em30 is seen @xcite , which is presumably emission from the ejecta .
further outwards , the remnant is surrounded by a faint , diffuse , partial shell of [ oiii ] with a diameter of @xmath6.03em60 that is thought to be the result of photoionization by the precursor .
emission lines of o , ne , c , and mg have been detected in the uv and optical spectra @xcite , but h , he , and fe are notably absent .
@xcite estimated an age of @xmath6.03em1000 yrs for the remnant based on the optical diameter and the line of sight velocity range .
based on abundance estimates from optical and uv data , @xcite suggested that the progenitor was possibly a massive o star that , after becoming a wolf - rayet star , exploded as a type ib supernova .
more recently , @xcite argued that e0102 is the remnant of a type iil / b supernova with a progenitor that lost most of the h envelope before the explosion .
we present new observations with the _ far ultraviolet spectroscopic explorer _
( _ fuse _ ) to study the ovi emission in regions in the x - ray bright ring with differing distributions of [ oiii ] and x - rays .
in addition , we analyze x - ray observations with _ xmm - newton _ and _ chandra _ and compare the o emission in the far - uv ( fuv ) to the x - ray emission .
the _ fuse _ satellite was launched in june 1999 ; details of its operation and on - orbit performance are provided in the papers by @xcite and @xcite .
the _ fuse _ instruments cover the 905 1187 band with a nominal point source spectral resolution of @xmath7 .
the _ fuse _ telescope has four optical channels , each with its own mirror and grating .
two mirrors and gratings are coated with silicon carbide ( sic channels ) , and the other two mirrors and gratings are coated with al and a lithium fluoride overcoat ( lif channels ) .
one sic / lif channel pair is sensed by one microchannel plate intensified detector , and the other channel pair is sensed by a second detector .
the sic channels provide wavelength coverage from @xmath6.03em905 to 1100 , whereas the lif channels cover the wavelength range of @xmath6.03em1000 to 1187 .
each of the two detectors is divided into two segments ( 1a and 1b , 2a and 2b ) , so that 8 spectra are obtained for each pointing ( lif1a / b , sic1a / b , lif2a / b , sic2a / b ) . at the focal plane of each channel is an assembly holding three spectrograph apertures , dubbed lwrs , mdrs , and hirs . in time - tag data mode ,
spectra through all three apertures are recorded simultaneously , but only one aperture is designated as primary .
the primary science apertures for our program are the 4 @xmath8 20 mdrs apertures . after launch
, small thermally - induced distortions were discovered that cause misalignments of these apertures . while periodic alignment activities are performed to keep the channels approximately aligned ,
motions of 3 8 occur on an orbital timescale ( @xmath6100 minutes ) , preventing detailed co - alignment during a given integration .
the lif1 channel was used for guiding in the data presented here , so that the location of the source is fixed for this channel .
due to the drift , not all the counts from the source fall on the detector .
as this is not taken into account when the flux of the calibrated spectrum is computed , the flux of the sic and lif2 channels become lower than the real flux which is measured in the lif1 channel .
the spectra of the sic and lif2 channels can be corrected by scaling the continuum flux to the correct continuum flux as determined in the lif1 spectrum .
we observed three positions around the bright x - ray ring of e0102 .
the positions were selected to have differing distributions of [ oiii ] in the optical and ovii , oviii in the x - ray ( table [ obstab ] ) .
the observations were scheduled at times where the aperture position angles could be tailored to the positions being observed .
figure [ images ] shows the projected positions on a _ chandra _ image and on an _ hst _
wfpc2 [ oiii ] image .
the net _ fuse _ integration times are 12.4 ksec for the north pointing ( n ) , 15.6 ksec for the northeast pointing ( ne ) , and 8.2 + 5.2 ksec for the southeast pointing ( se , two visits ) .
we also performed a dedicated off - source position with mdrs for estimating the background .
however , the spectra from the background pointing show more significant contamination by background starlight than the on - source positions .
therefore , the background observation has not been subtracted from the source spectra .
the data have been processed by the calfuse 3.0 pipeline , and the resulting spectra are shown in figure [ fusespec ] .
the only emission lines detected from the snr are the broadened ovi doublet lines .
we have modeled these spectra using functions available in idl and routines from an idl library including the fitting function mpfit.03emcraigm / idl / idl.html _ ] that performs least - squares curve fitting .
line fluxes and derived surface brightnesses in ovi are provided in table [ oneovi ] ( see sect.[ovidoublet ] for more details ) .
there are numerous public archival x - ray data sets for e0102 obtained with _
chandra _
@xcite and _ xmm - newton_. we first considered using the advanced ccd imaging spectrometer ( acis ) data from the _ chandra _ archive .
_ chandra _ data have far superior spatial resolution in comparison with _ xmm - newton_. however , we have determined that the brightest portions of e0102 as observed with the acis - s array in full frame mode are affected by pile - up , while the acis - s sub - array data and acis - i array data with the remnant on - axis have poor photon statistics .
pile - up causes two or more photons to be detected as a single event . as the observed energy is the sum of the energies of two or more photons , the energy spectrum is thus distorted .
compared to _
chandra _
observations , _ xmm - newton _ observations are less affected by pile - up due to shorter frame - times and lower spatial resolution ( thus smearing emission from bright sources over a larger region of the detector ) .
_ xmm - newton _ is equipped with three european photon imaging cameras ( epic ) ; one is a pn - type ccd ( epic - pn ) and two are mos - type ccds ( epic - mos1/2 ) .
both the acis and the epics are sensitive in the energy band of @xmath6.03em0.3 10.0 kev
. for snrs , we are in general interested in soft x - rays ( @xmath9 kev ) .
because the sensitivity of _ xmm - newton _ in the energy range of 0.5 2.0 kev is higher and the spectral resolution of epic - mos detectors is better than that of acis , we decided to use _ xmm - newton _ data primarily , except for the examination of the faint outer rim .
the outer rim is much fainter than the bright ring and has much less pile - up .
_ chandra _ s spatial resolution helps to isolate the rim emission from the bright ring emission .
we selected archival data of three _ xmm - newton _ observations which were taken early in the mission ( april 2000 and april 2001 ) : obsids 0123110201 , 0123110301 , and 0135720601 .
because we want to study the emission lines , we use epic - mos1 and mos2 data which have better spectral resolution than epic - pn .
the net exposure times after removing times with soft proton flares are @xmath6.03em16 ks , 13 ks , and 20 ks for the observations 0123110201 , 0123110301 , 0135720601 , respectively .
the data were obtained in small window mode resulting in a short frame time ( 0.3 s ; for comparison , frame time of the _ chandra _ acis full frame mode is 3.2 s ) and negligible pile - up .
the data are processed with xmmsas 6.1.0 .
we extracted spectra of the reverse shock at the positions of the _ fuse _ apertures ( fig.[xmmspec ] ) .
the brightest lines in the x - ray spectra at these positions are the ovii triplet lines and the oviii ly@xmath1 line .
although the diffuse ovi emission lines are broad and faint , these line features are well detected in the _ fuse _ lif1a and lif2b spectra .
the sic1a and sic2b spectra also cover the spectral range around the ovi lines , however , the statistics of those data are poor due to the lower effective areas . also , the sic channels are more affected by the misalignment of the channels than the lif2b channel and the emission from the brighter parts of the remnant may have largely moved out of the aperture .
we see significant ovi emission in both lif1a and lif2b data .
the rescaling of the lif2b spectra by effective area and continuum flux shows that the lif2b seems to have mainly covered the same ovi emission as the guiding lif1a channel .
therefore , we combine the lif1a and lif2b spectra for each pointing and analyze the co - added spectra around the ovi lines ( 1020 to 1050 ) .
the strongest emission lines in the spectra of the e0102 pointings are terrestrial airglow lines . in the wavelength band around the ovi doublet @xmath01032 , 1038 ( 1031.926 , 1037.617 , respectively )
, there is a strong hi airglow line at 1025.7 , while the ovi emission is relatively faint , in part because of the broadening of the lines .
we use this hi airglow line to calibrate the wavelength and to confirm that the spectral resolution is consistent with the aperture size . the ovi doublet fluxes are derived by modeling the emission features between @xmath6.03em1030 and 1040 .
interstellar absorption is significant in the far - uv and needs be taken into account when determining the line fluxes .
we include absorption lines of and cii and correct the flux for reddening through extinction .
the absorption has been derived using h2ools developed by @xcite . for the column density of we use the milky way ( mw ) and smc values from @xcite ,
i.e. , median values from 13 stars in the northern part of the smc .
there is an evident absorption feature at @xmath6.03em1036 which is a cii @xmath01036.3367 , 1037.0182 doublet .
we model this feature as a line at @xmath6.03em1036 with free width .
we also model absorption by ovi along the line of sight , including both the mw and the smc components .
we combine night and day data ( instead of using only night data to reduce the airglow emission ) because of the poor photon statistics .
nevertheless , the sensitivity of _ fuse _ allows the broad oviemission lines from this faint diffuse source to be well detected ( see fig.[fusespec ] ) .
the spectra are binned with 0.3 bins ( corresponding to @xmath6.03em87 km s@xmath10 ) . using the fitting routine mpfit
, we fit the ovi @xmath01032 , 1038 lines with two broad gaussians , and the continuum is modeled as a constant .
the widths of ovi @xmath01032 , 1038 gaussian lines are tied together , and the relative positions of the lines are fixed to the wavelength separation : i.e. we fit the ovi @xmath111032 line ( free wavelength ) and tie the ovi @xmath111038 line to the ovi @xmath111032 wavelength + 5.691 .
we also considered fitting each line with many zero - width gaussians which fill the observed width of the broad lines , to see if we can find different velocity components in the lines .
the zero - width gaussians form pairs with fixed separation between the ovi @xmath111032 and the ovi @xmath111038 components . however
, the fits did not improve significantly .
the normalization of each line is a free parameter except for the n pointing data , which have much poorer statistics . in the case of the n data
, the normalization of the @xmath111038 line is fixed at half of the @xmath111032 line normalization ( corresponding to the optically thin limit ) . from the line energy and width
, we derive a velocity shift and fwhm .
the flux is corrected for extinction , using the galactic extinction curve of @xcite and @xcite as well as extinction curves of the magellanic clouds from @xcite . to obtain the de - reddened flux , we use @xmath12 = 0.08 , assuming that the extinction is caused half by the smc and half by the mw , as suggested by @xcite .
we compute the 90% confidence errors for the line energy , width , and flux .
the fwhm , shifts , and fluxes are listed in table [ oneovi ] .
figure [ fusespec ] shows the spectra around @xmath111036 with the fits overplotted and fit residuals .
the ovi emission is faintest in the n pointing and brightest in the se pointing . in both the ne and se pointings ,
the fits show that the gas has a velocity dispersion of about 900 km s@xmath10 , while the faint ovi line feature in the n pointing spectrum has a fwhm of @xmath13 km s@xmath10 .
this is most likely not thermal line broadening , but instead indicates a velocity structure in the remnant ( see sect.[specresult ] ) .
the rest smc velocity is @xmath6.03em165 km s@xmath10 @xcite , so the broad ovi emission lines are blue - shifted by @xmath6.03em100 km s@xmath10 in the ne pointing and by @xmath6.03em330 km s@xmath10 in the se pointing .
however , in the n pointing , the broad ovi emission line feature is red - shifted by @xmath6.03em220 km s@xmath10 .
this might indicate a non - symmetric expansion of the remnant . to combine the x - ray data with the fuv data , we perform an x - ray emission line analysis using the h - like and he - like lines of o , ne , and mg
. the x - ray spectra have been binned with a minimum of 20 counts per bin to allow the use of @xmath14 statistics .
we fit the epic - mos1/2 spectra using the x - ray spectral fitting package xspec .
the model consists of gaussian lines at the known line energies of o , ne , mg , and si ( i.e. ovii triplet at @xmath6.03em0.57 kev , oviii ly@xmath1 at 0.654 kev , neix triplet at @xmath6.03em0.91 kev , nex ly@xmath1 at 1.022 kev , mgxi triplet at @xmath6.03em1.34 kev , mgxii ly@xmath1 at 1.473 kev , sixiii triplet at @xmath6.03em1.84 kev , sixiv ly@xmath1 at 2.006 kev ) , a continuum - only model using the astrophysical plasma emission code ( apec , * ? ? ?
* ) , and line - of - sight absorption .
the foreground absorption model consists of a component for the foreground mw absorption with a fixed of @xmath15 @xmath16 and a component with smc abundances @xcite to be fit .
all the ly@xmath1 lines are well fit with zero - width gaussians .
the ccd resolution does not allow the components of the triplet lines ( ovii , neix , and mgxi ) to be resolved , but at least two zero - width gaussians are necessary to reproduce the line features .
therefore , we fit the triplet emission with two gaussians at the energies of the forbidden and the resonance lines which are the two stronger lines and span the energy range of the triplet . the forbidden and the resonance lines are fit separately ,
i.e. , their flux ratio is free in the fits , and the flux of the intercombination line is included in the fluxes of the fitted lines .
we use the sum of the fluxes as the flux of the triplet .
oxygen lines are the strongest lines in both the _ fuse _ and the _ xmm - newton _ spectra of e0102 .
we compare line fluxes obtained from _ fuse _ ( ovi doublet ) and epic - mos ( ovii triplet and oviii ly@xmath1 ) to line fluxes predicted by models .
the observed line ratios are listed in table [ ratiotab ] . for the comparison with the observed lines , we use line fluxes computed with neiline , which evaluates line emissivities for an ionizing or recombining plasma .
line fluxes are calculated by evolving the plasma at a given initial temperature for a given ionization timescale . as this calculation assumes a single ionization timescale for the plasma
, we also model line fluxes using a plane - parallel shock model with a single temperature and a distribution of ionization timescales appropriate for a plane - parallel shock ( vpshock model in xspec ; see * ? ? ?
we assume smc abundances for all elements except for o , ne , and mg for both models .
for the elements o , ne , and mg with enhanced abundances we use the numbers obtained from the global fits of the x - ray spectrum ( see sect.[vneifit ] ) . the diagrams in figure [ linediagn ] show the line ratios ovii triplet / oviii ly@xmath1 , ovi doublet / ovii triplet , and ovi doublet / oviii ly@xmath1 derived from _ fuse _ and _ xmm - newton _ spectra and the error estimates based on the 90% confidence intervals for the line fluxes as a function of the ionization timescale @xmath17 and temperature @xmath18 for the two models . in the vpshock model ,
the parameter @xmath2 is the upper limit of the ionization timescale that has a distribution characteristic for a plasma shocked by a steady plane - parallel shock .
for the three line ratios for o , we evaluate confidence regions for the ratios as a function of @xmath18 and @xmath2 .
the lines in black show the ratio between ovii and oviii observed in x - rays ( ovii triplet / oviii ly@xmath1 ) .
the cyan and orange lines compare ovi in far - uv with ovii and oviii ( ovi doublet / ovii triplet and ovi doublet / oviii ly@xmath1 ) .
the results show that in the case of the neiline model , the confidence regions for the ratios ovii triplet / oviii ly@xmath1 , ovi doublet / ovii triplet , and ovi doublet / oviii ly@xmath1 intersect , yielding a well - defined @xmath18 vs. @xmath17 region for each aperture : for n , @xmath19 kev and @xmath20 s @xmath4 , for ne , @xmath19 kev and @xmath21 s @xmath4 , and for se , @xmath22 kev and @xmath23 s @xmath4 . while the @xmath18-@xmath2 regions for ne and se overlap , the result for n indicates higher @xmath18 and lower @xmath2 .
however , the diagrams for the plane - parallel shock model show no overlap between the confidence regions of the line ratios .
we also computed and compared the line ratios of ovii triplet / oviii ly@xmath1 , neix triplet / nex ly@xmath1 , and mgxi triplet / mgxii ly@xmath1 to check whether different elements might indicate different plasma conditions .
the results are shown in figure [ tripladiagn ] .
as the diagrams show , for neiline , the confidence regions for o , ne , and mg have little overlap and there are no conditions where all three intersect .
there seems to be no single set of temperatures and ionization timescales that simultaneously describes the o , ne , and mg ratios .
this is consistent with previous results . in analyzing _
asca _ data for e0102 , @xcite have found it necessary to introduce separate plasma components for each element .
analysis of _ chandra _ high energy transmission grating spectrometer ( hetgs ) data has shown that the dispersed line emission images have different radii for lines of different elements , suggesting progressive ionization by the reverse shock propagating through the ejecta @xcite .
another explanation for the lack of overlap between the ions is that the x - ray emission arises from a distribution of shocks with different velocities as suggested for snr 1987a by @xcite . in the case of a plane - parallel shock model ( vpshock )
the three confidence regions overlap for @xmath24 kev and @xmath25 s @xmath4 .
however , when the ovi line results from _ fuse _ are included , there is no region of common temperatures and ionization timescales which satisfies the o line ratios .
this might be indicating that the bulk of the ovi emission arises from a different location with different plasma conditions than the location from which the bulk of the ovii and oviiiemission arises .
to analyze the global x - ray spectra of the three regions , we first fit the spectra in xspec using the vnei model with nei version 2.0 for a plasma in non - equilibrium ionization ( nei ) with variable abundances .
this model assumes a single ionization timescale for the emitting plasma and is a simplification compared to the plane - parallel shock model ( see * ? ? ?
* ) . the nei version 2.0 models use a newer atomic database ( astrophysical plasma emission database , aped , * ?
* ) than the versions 1.0 and 1.1 and reproduces more accurate line emissivities .
however , they do not contain inner - shell processes .
for the absorption we include two components : one component represents the foreground mw absorption with a fixed of @xmath15 @xmath16 and the other represents the smc absorption . for each region ,
the epic - mos1 and mos2 spectra for the three observations are fit simultaneously with a single model .
although we fit the spectra with various sets of initial parameters , models with a single vnei component fail to fit the spectra adequately . in all cases ,
the model predicts a lower flux than the observed spectrum next to the mgxi triplet line feature , exactly where the mgxii ly@xmath1 line would be expected .
because the fits for the x - ray bright ring are dominated by the strong ovii , oviii , neix , and nex lines , we conclude that the elements o , ne , and mg might not be ionized to the same extent , with mg being more highly excited than the other elements .
we therefore try a model with additional vnei components to represent different plasma components .
first , we include a component for the ism shocked by the blast wave : to estimate this component , we use _ chandra _ acis - s3 data ( obsids and ) and extract a spectrum in the x - ray plateau covering the outer blast wave , i.e. a ring centered on ra = 01@xmath26 04@xmath27 02.0@xmath28 , dec = 72 01 53.2 with inner radius of 19 and outer radius of 23 .
this emission is very faint and not affected by pile - up . in the _ xmm - newton_epic - mos1/2 data it is not possible to separate clearly the emission from the blast wave and the ejecta because of the poor spatial resolution of _ xmm - newton_. fitting the blast wave spectrum extracted from _
chandra _ data with one vnei component ( absorption component as above ) ,
we obtain : @xmath29 kev , @xmath30 s @xmath4 .
therefore , we fit the _ xmm - newton _ spectra of parts of the bright x - ray ring with a model including a vnei component for the ism with @xmath18 and @xmath2 fixed , and all the abundances set to the smc ism abundances @xcite . for the ejecta emission , we include two vnei components : a component for the emission dominated by o , fixing the parameters for @xmath18 and @xmath2 to the results from the line diagnostics ( see fig.[linediagn ] ) with all other element abundances set to zero , and a component accounting for the rest of the ejecta emission , consisting of ne and mg ( with abundances for the other elements set to zero ) , with free @xmath18 and @xmath2 . the same set of absorbing columns ( galactic and smc as described above ) is used for both components .
the fits are better than with a single vnei model , and the fit parameters are listed in table [ xmmpar ] .
the spectra of the obsid 0123110201 data and the best fit model , obtained by a simultaneous fit of all spectra of the three obsids , are shown in figure [ xmmspec ] .
if the line broadening in the _ fuse _ spectra were due to thermal motion , the fwhm of about 900 km s@xmath10 would correspond to an unreasonably high temperature of 3@xmath31 k. the analysis of the epic - mos data shows that the plasma temperature is about 10@xmath32 k. there are at least three scenarios that can explain this discrepancy .
first , it is possible that the collisionless heating of the electrons behind the reverse shock front is negligible .
@xcite argue that this is the case at the outer blast wave and that significant particle acceleration is required there . in such a case ,
the oxygen ions , and thus also ovi , have a much high temperature than the electrons which are responsible for the x - ray emission .
second , the electron temperature might have been underestimated by applying a single ionization timescale model ( vnei ) .
third , the broadening might be caused by turbulence or by bulk motions rather than by thermal motions .
@xcite derive expansion velocities of about @xmath331000 km s@xmath10 from _ chandra _ hetgs data for the bright x - ray ring .
the velocities measured from the hetgs data could be explained by geometric effects , if e0102 has , e.g. , a tilted , expanding ring geometry or a distorted shell with density variations .
@xcite suggest that the difference in velocity shifts in different parts of the remnant might indicate that the southeastern part of the remnant shell is approaching and the northwestern part is receding .
this is consistent with the velocity shifts obtained for the broad ovi emission lines which show blue - shifts in the ne and se pointings , whereas the lines in the n pointing indicate a red - shift ( sect.[ovidoublet ] ) .
the x - ray emission of the bright x - ray ring is mainly ejecta emission .
because the spectra in all three regions are not well fit by a single component nei model , we have fit the spectra with two nei model components with variable abundances of o ( first component ) and ne and mg ( second component ) for the emission from the ejecta shocked by the reverse shock , and an additional nei model to account for the emission from the ism shocked by the blast wave .
we calculate the unabsorbed flux ( 0.3 10.0 kev ) of the ejecta and shocked ism for each region and compare the flux fraction : the ratio between the surface brightness of the total ejecta emission and the ism emission , @xmath34 , is 6 , 3 , and 3 for regions
n , ne , and se , respectively .
the surface brightness of the ism component is @xmath35 erg @xmath16 s@xmath10 arcsec@xmath36 in all three regions . for comparison , the surface brightness of the emission of the shocked ism in the outer rim is @xmath37 erg @xmath16 s@xmath10 arcsec@xmath36 , in good agreement with the estimated ism component for the emission of the x - ray bright ring at the positions of the _ fuse _ pointings .
the differences between the total flux of the three regions in the x - ray bright ring seem to be caused by differences in the emission from the ejecta shocked by the reverse shock as well as by variations in absorption as obtained from the spectral fits .
the line ratio diagrams in figure [ linediagn]a show that the ovi seen in fuv and the ovii and ovii x - ray emission are compatible with the assumption of an ionizing plasma with a single ionization timescale @xmath2 , as calculated by neiline . in this case , the line diagnostics of the o emission obtained with _ fuse _ and _
xmm - newton _ show that the ionization timescale of the o gas differs in the three regions , increasing from n to se : @xmath38 s @xmath4 in the n region , @xmath39 s @xmath4 in the ne region , and @xmath40 s @xmath4 in the se region .
the temperatures in the three regions are comparable : @xmath18 = 0.3 0.5 kev .
this result may indicate that the density in the ejecta is higher in the southeastern part of the x - ray bright ring , consistent with the optical data which show that the [ oiii ] emission is strongest in the se .
if we instead assume a plane - parallel shock with a distribution of ionization timescales using the vpshock model , we do not obtain a consistent set of @xmath18 and the upper limit of @xmath2 for the ovi , ovii , and oviii lines .
however , the diagrams in figure [ tripladiagn]b indicate that the flux ratios of emission lines of o , ne , and mg in x - ray , and thus also ovii and oviii , are consistent with a common temperature and ionization timescale assuming the plane - parallel shock model .
as the diagrams in figure [ linediagn]b compare ovi in fuv with the oxygen emission in x - ray , the fact that ovi , ovii , and oviii show no overlap might suggest that the oxygen line emission in the fuv and the oxygen line emission in x - rays together are not consistent with the assumed model . as copious [ oiii ] emission is observed in the optical , there must also be denser shocked material and the ovi emission might be related to it as well .
therefore , multiple shocked plasmas or a range of velocities are possibly involved in the fuv and x - ray emission of e0102 .
these results show that we find a few possible scenarios for the emission from this snr .
however , as we are limited by the size of the aperture of _ fuse _ and the spatial resolution of _ xmm - newton _ we are not able to constrain whether the emission arises from a single shock or multiple shocks . moreover
, the shock can be ionizing or recombining and thus radiative .
although we have optical [ oiii ] images and x - ray images with ovii and oviii emission of high spatial resolution , we lack spatial information on ovi . if it were possible to obtain ovi imaging data we would be able to compare the distribution of o in the remant and thus to determine if the shock is ionizing or recombining .
furthermore , hydrodynamical models for reverse shocks are necessary to understand the distribution of elements as well as physical parameters like @xmath18 or @xmath2 in relatively young snrs like e0102 .
to derive the ejecta mass , we numerically estimate the volume of the snr included in the slit , assuming that the snr emission is from a thick shell . the outer radius of the shell is assumed to coincide with the outer edge of the bright x - ray ring :
@xmath41 = 160@xmath3310 = [email protected] pc ( for @xmath42 kpc , * ? ? ?
* ) ; the inner radius of the bright x - ray ring is taken to be @xmath43 = 120@xmath3310 = [email protected] pc .
the resulting volume , @xmath44 , of the observed emitting gas is the intersection between this thick shell and a projected rectangle with the slit size ( @xmath45 4@xmath820 = 6.77 pc@xmath46 @xmath47 ) .
the results are listed in table [ esttab ] .
we assume an error of @xmath6.03em20% for the estimated volumes , as there are systematic uncertainties including the actual shape of the shell .
the _ chandra _ hetgs spectra show that the plasma mainly consists of o , ne , and mg @xcite .
the epic - mos fits also show that the ejecta emission results from o , ne , and mg .
therefore , we assume a metal - rich plasma consisting of o , ne , and mg excited to h - like , he - like , or fully ionized states .
the xspec vnei emissivity normalization ( table [ xmmpar ] ) is @xmath48 , where @xmath49 is the distance to the source [ cm ] , @xmath50 is the hydrogen number density [ @xmath4 ] , and @xmath52 is the electron number density [ @xmath4 ] . in the case of a plasma with high overabundance of elements other than hydrogen , @xmath50 needs to be replaced by a reference density , @xmath53 , that is a sum over the number densities of the included elements , thereby taking the abundances into account .
the electron density is @xmath54 , where @xmath55 is the density of element i. the factor @xmath56 depends on the assumed mean charge state for each element ; we assume @xmath57 , for o , ne , and mg , respectively . assuming @xmath42
kpc@xmath58 cm and uniform density in the thick shell , we estimate @xmath59 , @xmath60 , and @xmath61 for the two components for each region ( see table [ esttab ] ) .
using these number densities , we obtain the o , ne , and mg mass for the entire remnant .
we include a filling factor @xmath62 to account for the unknown volume fraction of the x - ray emitting plasma .
the densities are proportional to @xmath63 .
assuming a thick shell with an outer radius of @xmath41 = [email protected] pc and an inner radius of @xmath43 = [email protected] pc , we obtain @xmath64 , @xmath65 , @xmath66 .
if we assume a thick ring with @xmath67 @xmath4 as estimated by @xcite , we derive : @xmath68 , @xmath69 , @xmath66 .
we estimate a few @xmath70 in o , indicating a massive progenitor . if we assume a volume fraction of unity , the oxygen mass @xmath71 is consistent with the result from _ chandra _ hetgs by @xcite who also assumed @xmath72 . by comparing their @xmath71
estimate with the models of @xcite , @xcite suggest that the mass of the progenitor of e0102 was @xmath6.03em32 @xmath70 .
@xcite compare abundance ratios based on optical and uv lines to values derived by @xcite and conclude that the e0102 progenitor is best described by a 25 @xmath70 model .
our mass estimates for o and mg as well as the mass ratios agree well with the 25 30 @xmath70 model @xcite , while the ne mass estimate is slightly higher than the model prediction . even for a filling factor as low as @xmath73 ,
the mass estimates are consistent with a massive progenitor @xcite . in these calculations , there are still some uncertainties that we have to be aware of . as the reverse shock
has not yet reached the center of the remnant , there is still unshocked material interior to the reverse shock . if the unshocked material has the same abundance distribution as the shocked material ( i.e. , well - mixed ejecta ) the ratios will not be much altered , but the estimated masses would be lower limits .
furthermore , the ionization potential increases with the atomic number ; the shock might be strong enough to ionize o all the way up to the he - like and h - like stages , but not as efficient in ionizing mg .
therefore , a part of mg could be in lower ionization stages and not emitting in x - rays . finally , assuming a too high volume filling factor can result in overestimating the total masses .
numerical calculations have shown that the ejecta shocked by the reverse shock form thin layers that can be deformed by rayleigh - taylor instabilities , decreasing the fraction of the observed ejecta in the apertures @xcite .
we have observed three regions in the bright x - ray ring of the o - rich snr 1e0102.27219 in the smc using the _ fuse _ mdrs ( 4@xmath820 ) aperture .
these three regions cover parts of the remnant ring with varying distributions of [ oiii ] in the optical , and ovii and oviii in the x - ray .
the ring is caused by the reverse shock propagating into the ejecta . in the northern region ( n ) , the _ fuse _ spectrum shows very faint but velocity broadened ovi emission . in the northeastern ( ne ) and the southeastern ( se ) parts of the snr ring
, there is significant ovi emission with broad ovi@xmath01032 , 1038 lines , indicating a velocity dispersion of about 900 km s@xmath10 .
the emission from the se pointing appears to be blue - shifted by @xmath6.03em200 km s@xmath10 relative to the emission from the ne pointing .
we have also analyzed the _ xmm - newton _
epic - mos1/2 spectra of e0102 in the same regions as in the _ fuse _
observations . for the combined analysis of the fuv and x - ray data , we consider two cases : 1 .
shocked plasma described by a single temperature and single ionization timescale ( using the models neiline and vnei ) and 2
. plane - parallel shock with a distribution of ionization timescales ( using the vpshock model ) . in the first case , combining ovi doublet emission from _ fuse _ data with ovii triplet and oviii ly@xmath1 in epic - mos1/2 spectra yields @xmath74 kev and @xmath38 s @xmath4 for the n region , @xmath74 kev and @xmath75 s @xmath4 for the ne region , and @xmath76 kev and @xmath5 s @xmath4 for the se region .
the ionization timescales of the o gas seem to increase towards the south ( @xmath77 ) , suggesting a higher density in the south .
the total epic - mos1/2 spectra are well fit with a three component non - equilibrium ionization model ; two vnei components are necessary to describe the ejecta emission ( one for the o plasma , one for ne + mg plasma ) and an additional vnei component is used to model the emission from the shocked ism .
the temperature of the ne + mg component is higher ( @xmath78 kev ) than the o component .
for the plane - parallel shock model , however , the observed ratios of ovi / ovii and ovi / oviii are not consistent with the ovii / oviii ratio , in that there is not a common region of allowed temperatures and ionization timescales .
when the triplet and ly@xmath1 ratios are considered for o , ne , and mg in the x - ray bandpass alone , the observed ratios do have significant overlap in temperature ( @xmath79 1.0 kev ) and ionization timescale ( @xmath80 s @xmath4 ) .
these results might indicate that the _ fuse _ and _ xmm - newton _ data are sampling multiple shocks with different conditions . to better understand these results in the future
, we still require information on the spatial distribution of ovi in fuv as well as hydrodynamic models describing the structure of the reverse shock . from the x - ray emission , we estimate the mass of o , ne , and mg in the entire remnant . making the assumption that the remnant is a thick shell , we obtain @xmath81 , @xmath65 , @xmath66 .
a comparison with previous measurements and progenitor models shows that our results are consistent with a massive progenitor .
we thank the anonymous referee for useful critiques and suggestions which materially improved this paper .
we are grateful to daniel dewey , john raymond , and ravi sankrit for helpful discussions .
this work is based on observations made with the nasa - cnes - csa _ far ultraviolet spectroscopic explorer_. _ fuse _ is operated for nasa by the johns hopkins university under nasa contract nas5 - 32985 . the presented work is also based on observations obtained with _ xmm - newton _ , an esa science mission with instruments and contributions directly funded by esa member states and nasa .
the work was supported by the nasa/_fuse _ grant nag5 - 12295 , cxc contract nas8 - 03060 , and nasa grant go1 - 2060x .
the _ chandra _
x - ray observatory center is operated by the smithsonian astrophysical observatory for and on behalf of the national aeronautics space administration under contract nas8 - 03060 .
close - ups of _ fuse _ spectra of n , ne , and combined se pointings .
lif1a and lif2b spectra are combined .
the thick dashed line shows the model ( see sect.[ovidoublet ] ) .
the thick lines at the bottom mark the rest wavelengths of the ovi lines .
the dotted line shows the absorption in an arbitrary scale . ,
scaledwidth=45.0% ] _ xmm - newton _ epic - mos1 ( solid ) and mos2 ( dashed ) spectra of the observation 0123110201 for the regions n , ne , and se .
fit model includes three vnei spectral components ( see sect.[vneifit ] ) .
the prominent o , ne , and mg lines are labeled in the n spectrum . , scaledwidth=45.0% ] lccccc + c0830101 ( n ) & 01:04:01.74 & 72:01:39.1 & mdrs & 2003 - 09 - 19 11:35:05 & 12 + c0830201 ( ne ) & 01:04:04.24 & 72:01:45.1 & mdrs & 2002 - 06 - 06 23:51:41 & 16 + c0830301 ( se1 ) & 01:04:04.04 & 72:02:01.9 & mdrs & 2003 - 08 - 29 07:54:23 & 8 + c0830302 ( se2 ) & 01:04:04.04 & 72:02:01.9 & mdrs & 2004 - 08 - 21 14:52:32 & 5 + + 0123110201 & 01:03:47.25 & 72:01:58.3 & em1/2 , sw & 2000 - 04 - 16 20:09:26 & 16 + 0123110301 & 01:03:47.19 & 72:01:58.3 & em1/2 , sw & 2000 - 04 - 17 04:43:36 & 13 + 0135720601 & 01:03:45.79 & 72:00:51.1 & em1/2 , sw & 2001 - 04 - 14 20:47:25 & 20 + + 5123 & 01:03:56.41 & 72:01:18.7 & acis - s & 2003 - 12 - 15 04:27:06 & 20 + 5130 & 01:04:12.31 & 72:01:51.2 & acis - s & 2004 - 04 - 09 13:08:57 & 19 + lcccccc n & 0.3 & 1540@xmath33350 & 380@xmath33180 & ( 1.2@xmath82 & ( 0.6@xmath83 & 1.59 + ne & 0.3 & 970@xmath33120 & 60@xmath3360 & ( 1.7@xmath84 & ( 1.3@xmath85 & 1.65 + se & 0.3 & 830@xmath33310 & 160@xmath3330 & ( 2.3@xmath86 & ( 2.0@xmath86 & 1.72 + lccc ( smc ) [ @xmath102 @xmath16 ] & 1.6 ( 1.3 2.0 ) & 0.1 ( 0.0 0.4 ) & 0.4 ( 0.2 0.7 ) + + @xmath104 [ kev ] & 0.40 & 0.38 & 0.30 + o ( solar ) & 93 ( 83 104 ) & 86 ( 80 94 ) & 99 ( 92 107 ) + @xmath105 [ @xmath106 s @xmath4 ] & 6.0 & 10 . & 20 .
+ @xmath107 [ @xmath108 & 2.4 ( 2.2 2.6 ) & 2.3 ( 1.5 3.2 ) & 3.7 ( 3.5 3.9 ) + + @xmath109 [ kev ] & 3.5 ( 2.7 3.7 ) & 1.5 ( 1.0 3.2 ) & 1.1 ( 0.8 1.6 ) + ne ( solar ) & 82 ( 75 89 ) & 86 ( 66 116 ) & 82 ( 77 94 ) + mg ( solar ) & 24 ( 20 28 ) & 33 ( 23 49 ) & 30 ( 22 40 ) + @xmath110 [ @xmath106 s @xmath4 ] & 1.8 ( 1.7 1.9 ) & 2.3 ( 1.5 3.7 ) & 3.0 ( 2.1 4.4 ) + @xmath111 [ @xmath108 & = @xmath107 & = @xmath107 & = @xmath107 + + @xmath112 [ @xmath113 & 3.0 ( 2.8 3.3 ) & 2.0 ( 1.8 2.3 ) & 2.3 ( 2.1 2.5 ) + + @xmath14 & 559.7 & 608.8 & 618.8 + d.o.f & 374 & 368 & 403 + reduced @xmath14 & 1.50 & 1.65 & 1.54 + lccc @xmath44 [ pc@xmath114 & [email protected] & [email protected] & [email protected] + @xmath44 [ @xmath115 cm@xmath114 & [email protected] & [email protected] & [email protected] + + @xmath117 [ @xmath4 ] & [email protected] & [email protected] & [email protected] + @xmath118 [ @xmath4 ] & [email protected] & [email protected] & [email protected] + @xmath119 [ @xmath4 ] & [email protected] & [email protected] & [email protected] + + @xmath120 [ @xmath70 ] & [email protected] & [email protected] & [email protected] + @xmath122 [ @xmath70 ] & [email protected] & [email protected] & [email protected] + @xmath123 [ @xmath70 ] & [email protected] & [email protected] & [email protected] + + @xmath120 [ @xmath70 ] & [email protected] & [email protected] & [email protected] + @xmath122 [ @xmath70 ] & [email protected] & [email protected] & [email protected] + @xmath123 [ @xmath70 ] & [email protected] & [email protected] & [email protected] + | we present far ultraviolet spectroscopic explorer ( _ fuse _ ) and x - ray multi - mirror mission ( _ xmm - newton _ ) data for the reverse shock of the o - rich supernova remnant ( snr ) 1e0102.27219 in the small magellanic cloud ( smc ) .
the _ fuse _
observations cover three regions with significantly different optical [ oiii ] intensities , all associated with the relatively bright part of the x - ray ring .
emission lines of ovi @xmath01032 , 1038 are clearly detected in the _ fuse _ spectra . by combining this ovi doublet emission with the ovii triplet and oviii ly@xmath1 fluxes from the _ xmm - newton _ spectra and assuming a non - equilibrium ionization ( nei ) model with a single ionization timescale for the spectra , we are able to find a narrow range of temperatures and ionization timescales that are consistent with the respective line ratios .
the increase of the ionization timescale @xmath2 from north ( @xmath3 s @xmath4 ) to south - east ( @xmath5 s @xmath4 ) is indicative of increasing density in the x - ray bright ring , in good agreement with the optical [ oiii ] emission which is strongest in the south - east .
however , if we assume a plane - parallel shock model with a distribution of ionization timescales , the ovi emission appears to be inconsistent with ovii and oviii in x - rays .
we also analyze the total _ xmm - newton _
epic - mos 1/2 spectra for the three regions .
the x - ray spectra are dominated by strong emission lines of o , ne , and mg , however , we detect an emission component that accounts for 14 25% of the flux and can be attributed to shocked ism .
we find that there is no consistent set of values for the temperature and ionization timescale which can explain the observed line ratios for o , ne , and mg .
this would be consistent with a structured distribution of the ejecta as the o , ne , mg would have interacted with the reverse shock at different times . |
in this paper we have investigated the effective action of the iib matrix model on the fuzzy @xmath0 background up to two loop level .
in particular we have studied the case in which the size of one of @xmath0 is different from that of the other . by the power counting and susy cancellation arguments
, we can identify the t hooft coupling and the large @xmath2 scaling behavior of the effective actions to all orders . in the large @xmath2 limit quantum corrections of the effective action
are found to be @xmath54 .
this analysis will be applied to all order corrections of the effective action .
we have found the minimum point of the effective action at the two loop level , and concluded that the same size @xmath0 with @xmath55 gauge group is favored . in the previous paper@xcite
we have concluded that the iib matrix model favors the 4 dimensional spacetime(fuzzy @xmath0 ) rather than 2 and 6 dimensional spacetimes @xmath3 and @xmath56 . combining the previous work and
the results in this paper suggests that the iib matrix model singles out not only _ four dimensionality of spacetime _ but also _ more symmetric spacetime_. we hope this suggestion is one of general features of the iib matrix model .
in addition we have developed a approach to calculate the amplitudes on the fuzzy @xmath0 .
this approach can not be applied to the @xmath3 background(@xmath25 ) case .
we wish that extensions of this approach enables us to calculate amplitudes on other homogeneous spaces , such as @xmath57 and @xmath58@xcite .
99 t. banks , w. fischler , s.h .
shenker and l. susskind , _ m - theory as a matrix model : a conjecture _ , phys . rev .
* d55 * 5112 ( 1997 ) , hep - th/9610043 . n. ishibashi , h. kawai , y. kitazawa and a. tsuchiya , _ a large - n reduced model as superstring _ ,
nucl . phys . *
b498 * ( 1997 ) 467 , hep - th/9612115 .
a. connes , m. douglas and a. schwarz , _ noncommutative geometry and matrix theory : compactification on tori _ , jhep 9802 : 003.1998 , hep - th/9711162 . n. seiberg and e. witten , _ string theory and non - commutative geometry _ ,
jhep * 9909 * ( 1999 ) 032 , hep - th/9908142 .
myers,_dielectric - branes _ ,
jhep 9912 ( 1999 ) 022,hep - th/9910053 . a. y. alekseev , a. recknagel and v. schomerus , _ brane dynamics in background fluxes and noncommutative geometry _ ,
jhep * 0005 * ( 2000 ) 010 , hep - th/0003187 .
h. aoki , n. ishibashi , s. iso , h. kawai , y. kitazawa and t. tada , _ non - commutative yang - mills in iib matrix model _ , nucl .
* 565 * ( 2000 ) 176 , hep - th/9908141 .
m. li , _ strings from iib matrices _ , nucl.phys .
* b499 * ( 1997 ) 149,hep - th/961222 . n. ishibashi , s. iso , h. kawai and y. kitazawa , _ wilson loops in non - commutative yang - mills _
b573 * ( 2000 ) 573 , hep - th/9910004 .
d. j. gross , a. hashimoto and n. itzhaki , _ observables of non - commutative gauge theories _
, adv.theor.math.phys .
* 4 * ( 2000 ) 893,hep - th/0008075 .
s. minwalla , m.v .
raamsdonk and n. seiberg , _ non - commutative perturbative dynamics _ , jhep * 0002 * ( 2000 ) 020,hep - th/9912072 .
a. dhar and y. kitazawa , _ non - commutative gauge theory , open wilson lines and closed strings _ ,
jhep * 0108 * ( 2001 ) 044 , hep - th/0106217 .
h. aoki , s. iso , h. kawai , y. kitazawa and t. tada , _ space - time structures from iib matrix model _
phys . * 99 * , 713(1998 ) ; hep - th/9802085 .
jun nishimura , graziano vernizzi , _ spontaneous breakdown of lorentz invariance in iib matrix model _ jhep 0004:015,2000 : hep - th/0003223 , k.n .
anagnostopoulos , j. nishimura , _ new approach to the complex action problem and its application to a nonperturbative study of superstring theory _
phys.rev.d66:106008 : hep - th/0108041 j. nishimura and f. sugino , _ dynamical generation of four - dimensional space - time in iib matrix model _
, jhep * 0205 * ( 2002 ) 001 , hep - th/0111102 .
h. kawai , s. kawamoto , t. kuroki , t. matsuo , s. shinohara , _ mean field approximation of iib matrix model and emergence of four dimensional space - time_,nucl.phys .
b647 ( 2002 ) 153 - 189 : hep - th/0204240 ; _ improved perturbation theory and four - dimensional space - time in iib matrix model _ , prog.theor.phys . 109 ( 2003 )
115 - 132:hep - th/0211272 .
y. kitazawa , _ matrix models in homogeneous spaces _
, nucl phys . *
b642 * ( 2002 ) 210 , hep - th/0207115 .
t. a. imai , y. kitazawa , y. takayama and d. tomino , `` quantum corrections on fuzzy sphere , '' nucl .
b * 665 * , 520 ( 2003 ) [ arxiv : hep - th/0303120 ] .
t. imai , y. kitazawa , y. takayama and d. tomino , `` effective actions of matrix models on homogeneous spaces , '' nucl .
b * 679 * , 143 ( 2004 ) arxiv : hep - th/0307007 . a new monte carlo event generator for high - energy physics . by s. kawabata ( kek , tsukuba ) .
kek - preprint-85 - 26 , jul 1985 .
44pp . published in comput.phys.commun.41:127,1986 also in tsukuba workshop : jlc 1990:239 - 249 ( qcd183:w82:1990 ) a new version of the multidimensional integration and event generation package bases / spring . by setsuya kawabata ( kek , tsukuba ) .
kek - preprint-94 - 197 , feb 1995 .
20pp . published in comput.phys.commun.88:309-326,1995 a. r. edmonds , _ angular momentum in quantum mechanics _ , princeton univ . press ( 1957 ) .
e. p. winger , group theory , academic press , new york ( 1959 ) g. ponzano , and t. regge , _ semiclassical limit of racah coefficients _ , spectroscopic group theoretical methods in physics ( 1968 ) | we continue our study of the iib matrix model on fuzzy @xmath0 . especially in this paper
we focus on the case where the size of one of @xmath1 is different from the other . by the power counting and susy cancellation arguments
, we can identify the t hooft coupling and large @xmath2 scaling behavior of the effective action to all orders .
we conclude that the most symmetric @xmath0 configuration where the both @xmath3s are of the same size is favored at the two loop level .
in addition we develop a new approach to evaluate the amplitudes on fuzzy @xmath0 .
@xmath4 [ cols= " < " , ] the table [ tbl : wig ] shows that @xmath5 comes closer to @xmath6 as we increase @xmath7 and @xmath8 .
it seems a bit surprising that this approximation works so well .
it will be clarified later why is the case .
we expect that @xmath6 and @xmath5 are equal in the large @xmath7 and @xmath8 limit and that this approximation is exact in this limit .
moreover in the limit @xmath7 and @xmath9 the summations in ( [ fwig ] ) can be approximated by integrations .
as an example let us focus on the case @xmath10 .
after rescaling variables @xmath11 , @xmath12 , and so forth and taking the limit @xmath13 , we obtain ] @xmath14 where the wigner asymptotic estimate changes the 6j symbols as @xmath15 with @xmath16 we now understand why this is a good approximation . from ( [ f ] ) , we can observe that the leading contribution to @xmath17 comes from the uv regions , i.e. , @xmath18 .
in such a region , ( [ wigest ] ) is certainly valid .
we can apply this approximation to @xmath19 cases as well .
the novelty here is the appearance of the ratio @xmath20 : @xmath21 @xmath22 is symmetric under the exchange of @xmath23 with @xmath24 , which is consistent with the argument in the sub - section [ sec : tree1loop ] .
numerically we can check whether this approximation is valid or not .
we perform the multiple integration ( [ fint ] ) twice analytically and eventually by monte - carlo integration@xcite@xcite .
the result is shown in fig.[fig : fr ] , which indicates this approximation is in good agreement with the method in the section [ subsec : intapprx ] .
we make a comment on the difference between the methods in the sections [ subsec : asymptotic ] and [ subsec : intapprx ] .
the former method requires more machine power with increasing @xmath23 , while the latter does not in the fig .
[ fig : effactionvsr ] . ] .
so it seems that the latter is more economical and gives better results .
it is noted that this approach can not applies to the @xmath3 background case , that is , @xmath25 case .
setting @xmath26 , @xmath6 becomes@xcite @xmath27 rescaling variables @xmath28 and taking the large @xmath7 limit , we have @xmath29 the factor @xmath30 vanishes , while the integration part diverges because of the ir contributions , i.e. @xmath31 .
this implies that ( [ f2d ] ) is ill - defined , so that the naive approximation is not valid for the case @xmath25 .
this limitation can be explained from another point of view . in the @xmath26 case
dominant contributions to @xmath6 comes from ir regions , which is to say , @xmath32 .
in such a region ( [ wigest ] ) is no longer valid .
this approximation can not be applied to @xmath3 background . in this case , however , 6j symbols are well - approximated by 3j symbols @xmath33 in this sub - section , we determine the structure of the effective action to all orders . by naive power counting
we estimate the @xmath34-th loop contribution to be @xmath35 since we obtain a factor of @xmath36 from 6j symbols in the interaction vertices(see @xcite ) . here
we also assume that the @xmath0 background breaks susy softly , and so that leading corrections of each @xmath34-th loop cancel . under this assumption
it appears that the @xmath34-th loop contribution becomes @xmath37 .
in fact this order estimation is correct in the two loop corrections ( [ 2loop ] ) .
this argument allows us to define the t hooft coupling of @xmath38 gauge theory as @xmath39 to all orders and so that the @xmath40th loop contributions are @xmath41 .
the above arguments give the effective action to all orders in terms of the t hooft coupling @xmath42 n n , \label{effaction}\\ & & \mbox{with } \quad { \tilde{h}}_4(r , \lambda^2 ) = { f(r ) \over 2 } { \lambda^2 \over ( 2 \pi)^2 } + o(\lambda^4).\end{aligned}\ ] ] our remaining task is to minimize the effective action in terms of @xmath43 , and @xmath44 or equivalently @xmath45 .
we can fix the t hooft coupling @xmath46 by minimizing the effective action ( [ effaction ] ) independently of @xmath47 ( but dependent on @xmath23 ) . by solving @xmath48 for @xmath45
we find , up to two loop level , @xmath49 then the effective action becomes @xmath50 this can be evaluated numerically in two ways , i.e. , the methods in the section [ subsec : asymptotic ] and [ subsec : intapprx ] ( fig.[fig : effactionvsr ] ) .
: the solid circles are obtained by the method described in the section [ subsec : asymptotic ] , while the solid line is obtained by the method of the section [ subsec : intapprx].,height=302 ] it is found from this figure that the @xmath51 configuration minimizes the effective action , which means that the configuration of the same size @xmath1 is favored over the different size configurations .
also it is apparent from ( [ effaction ] ) that the case @xmath52 is favored .
it is the most important result of this paper that the same size @xmath0 configuration with the @xmath53 gauge group is singled out by the iib matrix model .
it is worth noting that the effective action blows up , as we decrease @xmath23 below @xmath51 .
this tendency is consistent with the result in @xcite , that is , the configuration of the 4d limit(@xmath51 ) is favored rather than that of the 2d limit(@xmath25 ) .
we expect that higher order corrections do not change this result by the argument in @xcite . |
the riemann - hilbert problem ( rh - problem , for short ) is one of the most important classes of boundary value problems for analytic functions .
indeed , the dirichlet problem , the neumann problem , the mixed dirichlet - neumann problem , the problems of computing the conformal mapping and the external potential flow can be formulated as rh - problems .
the rh problem consists of determining of all analytic functions in a domain @xmath0 in the extended complex plane that satisfy a prescribed boundary condition on the boundary @xmath1 .
a boundary integral equation with continuous kernel for solving the rh problem has been derived in @xcite .
the kernel of the derived integral equation is a generalization of the well known _
so , the new kernel has been called the _ generalized neumann kernel_. the solvability of the boundary integral equation with the generalized neumann kernel has been studied for simply connected domains with smooth boundaries in @xcite , for simply connected domains with piecewise smooth boundaries in @xcite , and for multiply connected domains in @xcite .
it turns out that the solvability of the boundary integral equation with the generalized neumann kernel depends on the index @xmath2 of the coefficient function @xmath3 of the rh - problem on each boundary component of the boundary of @xmath0 .
however , the solvability of the rh - problem depends on the total index @xmath4 of the function @xmath3 on the whole boundary of @xmath0 .
this raises a difficulty in using the boundary integral equation with the generalized neumann kernel to solve the rh - problem in multiply connected domains .
such a difficulty does not appear for the simply connected case since the boundary of @xmath0 consists of only one component .
so far , the boundary integral equation with the generalized neumann kernel has been used to solve the rh - problem in multiply connected domains for special case of the function @xmath3 ( see ( * ? ? ?
* eq . ( 1.2 ) ) ) . for such special case ,
the boundary integral equation with the generalized neumann kernel has been used successfully to compute the conformal mapping onto more than @xmath5 canonical domains @xcite and to solve several boundary value problems such as the dirichlet problem , the neumann problem , and the mixed boundary value problem @xcite . in this paper
, we shall use the boundary integral equation with the generalized neumann kernel to solve a certain class of rh - problems in multiply connected domains considered by wegmann @xcite and known as the _ general conjugation problem_. we shall also use the integral equation to provide an alternative proof for the existence and uniqueness of the solution of the general conjugation problem .
let @xmath0 be the unbounded multiply connected circular domain obtained by removing @xmath6 disks @xmath7 from the extended complex plane @xmath8 such that @xmath9 ( see figure [ f : dom ] ) .
the disk @xmath10 is bounded by the circle @xmath11 with a center @xmath12 and radius @xmath13 .
we assume that each circle @xmath14 is clockwise oriented and parametrized by @xmath15 , \quad j=1,2,\ldots , m.\ ] ] let @xmath16 be the disjoint union of the @xmath6 intervals @xmath17 which is defined by @xmath18 the elements of @xmath16 are order pairs @xmath19 where @xmath20 is an auxiliary index indicating which of the intervals the point @xmath21 lies in .
thus , the parametrization of the whole boundary @xmath22 is defined as the complex function @xmath23 defined on @xmath16 by @xmath24 we assume that for a given @xmath21 that the auxiliary index @xmath20 is known , so we replace the pair @xmath19 in the left - hand side of ( [ e : eta-1 ] ) by @xmath21 , i.e. , for a given point @xmath25 , we always know the interval @xmath26 that contains @xmath21 . the function @xmath23 in ( [ e : eta-1 ] ) is thus simply written as @xmath27,\\ \hspace{0.3cm}\vdots\\ \eta_m(t),&t\in j_m=[0,2\pi ] .
\end{array } \right .\ ] ] [ 0.5 ] of connectivity @xmath6.,title="fig : " ] -3.4 cm @xmath28 + .30 cm @xmath29 + .20 cm @xmath30 -0.8 cm @xmath31 + .80 cm @xmath0 @xmath32 -0.7 cm @xmath33 + 0.8 cm
let @xmath34 denote the space of all real functions @xmath35 in j , whose restriction @xmath36 to @xmath37 $ ] is a real - valued , @xmath38-periodic and hlder continuous function for each @xmath39 , i.e. , @xmath40 in view of the smoothness of the parametrization @xmath23 , a real hlder continuous function @xmath41 on @xmath42 can be interpreted via @xmath43 , @xmath25 , as a function @xmath44 ; and vice versa .
so , in this paper , for any given complex or real valued function @xmath45 defined on @xmath42 , we shall not distinguish between @xmath46 and @xmath47 .
further , for any complex or real valued function @xmath45 defined on @xmath42 , we shall denote the restriction of the function @xmath45 to the boundary @xmath14 by @xmath48 , i.e. , @xmath49 for each @xmath39 .
however , if @xmath50 is an analytic function in the domain @xmath0 , we shall denote the restriction of its values to the boundary @xmath14 by @xmath51 , i.e. , @xmath52 for each @xmath39 .
let @xmath3 be a continuously differentiable complex function on @xmath42 with @xmath53 .
the generalized neumann kernel is defined by ( see @xcite for details ) @xmath54 when @xmath55 , the kernel @xmath56 is the well - known neumann kernel which appears frequently in the integral equations of potential theory and conformal mapping ( see e.g. @xcite ) .
we define also the following singular kernel @xmath57 which is closely related to the generalized neumann kernel @xmath58 @xcite , @xmath59 \(a ) the kernel @xmath58 is continuous with @xmath60 ( b ) when @xmath61 are in the same parameter interval @xmath26 , then @xmath62 with a continuous kernel @xmath63 which takes on the diagonal the values @xmath64 on @xmath34 we define the fredholm operator @xmath65 and the singular operator @xmath66 both operators @xmath67 and @xmath68 are bounded on the space @xmath34 and map @xmath34 into itself . for more details , see @xcite .
the identity operator on @xmath34 is denoted by @xmath69 .
the rh - problem for the unbounded multiply connected domain @xmath0 is defined as follows : + for a given function @xmath44 , search a function @xmath70 analytic in @xmath0 and continuous on the closure @xmath71 with @xmath72 such that the boundary values of @xmath70 satisfy on @xmath42 the boundary condition @xmath73=\gamma.\ ] ] the boundary condition in ( [ eq : rhp ] ) is non - homogeneous . when @xmath74 , we have the homogeneous boundary condition @xmath75=0.\ ] ] the solvability of the rh problem depends upon the _ index _ of the function @xmath3 on the boundary @xmath42 .
the index @xmath2 of the function @xmath3 on the circle @xmath14 is the change of the argument of @xmath76 along the circle @xmath14 divided by @xmath38 .
the index @xmath4 of the function @xmath76 on the whole boundary curve @xmath42 is the sum @xmath77 vekua ( * ? ? ?
* eq . ( 1.2 ) , p. 222 ) , gakhov ( * ? ? ?
* eq . ( 27.1 ) , p. 208 ) and mityushev ( * ? ? ?
* eq . ( 38.3 ) , p. 601 ) define the rh - problem with @xmath78=\gamma$ ] , i.e. , with the complex conjugate of the function @xmath3 .
this has the consequence that in some of the later results the index of the function @xmath3 occurs with the opposite sign as in @xcite .
we follow @xcite and define the space @xmath79 , the spaces of functions @xmath35 for which the rh problem have solution , by @xmath80\ , \mbox { on $ c$},\ , \psi \mbox { analytic in $ g$ , $ \psi(\infty)=0$}\}.\ ] ] we define also the space @xmath81 to be the space of the boundary values of solutions of the homogeneous rh problem , i.e. , @xmath82 to study the solvability of the rh - problem , we define the following boundary value problem as the homogeneous exterior rh problem on @xmath83 : + search a function @xmath84 analytic in @xmath85 and continuous on the closure @xmath86 such that the boundary values of @xmath84 satisfy on @xmath42 , @xmath87=0.\ ] ] it is clear that @xmath85 is not a domain .
in fact , it is the union of @xmath6 disjoint disks .
so , solving the homogeneous exterior rh problem ( [ eq : erhp ] ) is equivalent to solving @xmath6 rh problems in the disks @xmath88 .
the space of the boundary values of solutions of the homogeneous exterior rh problem ( [ eq : erhp ] ) is denoted by @xmath89 , i.e. , @xmath90 for @xmath91 , let @xmath92 be the subspace of @xmath93 of real functions @xmath94 such that @xmath95 where @xmath84 is analytic in the disk @xmath10 , i.e. , @xmath84 is a solution of the following homogenous rh - problem on the disk @xmath10 , @xmath96=0 \quad { \rm on}\quad c_j.\ ] ] the problem is a rh - problem in the bounded simply connected domain @xmath10 and the index of the function @xmath97 on @xmath11 is @xmath2 .
then we have from ( * ? ? ?
* eq . ( 29 ) ) @xmath98 ( note that the orientation of the circles @xmath14 is clockwise which changes the sign in ( * ? ? ?
* eq . ( 29 ) ) ) .
then , we have the following lemmas from @xcite .
[ l : s - sk ] the space @xmath89 is the direct sum of the subspaces @xmath99 , @xmath100 the space @xmath89 plays a very important rule in using the boundary integral equation with the generalize neumann kernel to solve the rh - problem especially when the problem is not solvable since it allows us to find the form of conditions that we should impose on @xmath35 to make the problem solvable . for simply connected domains ,
it was proved in ( * ? ? ?
* corollary 3 ) that the space @xmath34 has direct sum decomposition @xmath101 the decomposition means that if the rh - problem @xmath102=\gamma$ ] is not solvable , then there exists a unique function @xmath103 such that the rh - problem @xmath102=\gamma+h$ ] is solvable .
for simply connected domains , the decomposition is valid for general index @xmath4 of the function @xmath3 .
however , for multiply connected domains , the decomposition in general is not correct since we may have @xmath104 ( see @xcite . ) in this paper , we shall consider special case of the function @xmath3 for which we can prove that the decomposition is valid for multiply connected domains ( see theorem [ t : h = r+s ] below ) , namely , we assume the index of the function @xmath3 satisfies @xmath105 which implies that @xmath106 .
rh - problem with such special case of the index has wide applications . for example ,
our assumption on the index are satisfied for the rh - problem used in @xcite to develop a method for computing the conformal mapping onto more than @xmath5 canonical domains and for the rh - problems studied in @xcite .
furthermore , many other boundary value problems such as the dirichlet problem , the neumann problem , the mixed boundary value problem , and the schwarz problem can be reduced to rh - problems whose indexes satisfy our assumption ( see e.g. , @xcite ) . under the above assumption on the index , we have the following theorem from @xcite .
[ t : rhp - sol ] for @xmath106 , we have @xmath107 the following lemma follows from ( [ eq : sim - s - j ] ) , ( [ eq : s - sk ] ) and .
[ l : rhp - sol2 ] let @xmath108 for @xmath91 , then @xmath109 there is a close connection between rh problems and integral equations with the generalized neumann kernel .
the null - spaces of the operators @xmath110 are related to the spaces @xmath111 by ( see ( * ? ? ?
* theorem 11 , lemma 20 ) @xmath112 where @xmath113 is isomorphic via @xmath68 to @xmath114 .
for the function @xmath3 defined by , we have @xcite @xmath115 and @xmath116 in view of , eq .
implies that @xmath117 ( see also ) .
since @xmath113 isomorphic to @xmath114 , we have also @xmath118 [ t : h = r+s ] let @xmath108 for @xmath91 , then the space @xmath34 has the decomposition @xmath119 since @xmath120 the direct sum decomposition follows from ( [ eq : r - s- ] ) .
it follows from theorem [ t : rhp - sol ] that the non - homogeneous rh problem ( [ eq : rhp ] ) is in general insolvable for @xmath108 .
if it is solvable , then the solution is unique .
the following corollary which follows from theorem [ t : h = r+s ] provides us with a way for modifying the right - hand side of ( [ eq : rhp ] ) to ensure the solvability of the problem .
[ c : g+h ] let @xmath108 for @xmath91 , then for any @xmath44 , there exists a unique function @xmath103 such that the following rh problem @xmath121=\gamma+h\ ] ] is uniquely solvable .
solving the rh - problem requires determining both the analytic function @xmath70 as well as the real function @xmath122 .
this can be done easily using the boundary integral equations with the generalized neumann kernel as in the following theorem .
[ t : ie - rhp ] let @xmath108 for @xmath91 .
for any given @xmath44 , let @xmath123 be the unique solution of the integral equation @xmath124 then the boundary values of the unique solution of the rh problem ( [ eq : rhp2 ] ) is given by @xmath125 and the function @xmath122 is given by @xmath126/2.\ ] ] the theorem can be proved using the same argument as in the proof of ( * ? ? ?
* theorem 2 ) .
the uniqueness of the solution of the integral equation follows from the fredholm alternative theorem since @xmath127 .
the advantages of theorem [ t : ie - rhp ] are that it , based on the integral equation , provides us with formulas for computing the real function @xmath122 necessary for the solvability of the rh problem as well as the solution @xmath70 of the rh problem .
in this section , we shall consider a very important certain class of rh problems which has been considered by wegmann @xcite .
the indexes of this class of rh problems satisfied our assumption .
this class has been considered by many researchers and has many applications @xcite .
wegmann @xcite proves the following theorem .
[ t : weg ] for any integer @xmath128 and for any sufficiently smooth functions @xmath35 on the boundary of @xmath0 , the rh problem @xmath129=\gamma_{j}(t),\ ] ] has a unique solution consisting of an analytic function @xmath70 in @xmath0 with @xmath72 and @xmath130 real numbers @xmath131 for @xmath39 .
wegmann @xcite called the rh problem ( [ eq : rhp - w ] ) as the _ general conjugation problem _ since the case @xmath132 describes the problem of finding the conjugate harmonic function of a harmonic function with boundary values @xmath36 @xcite .
the general conjugation problem has been solved by wegmann @xcite using the method of _ successive conjugation _ which reduces the problem ( [ eq : rhp - w ] ) to a sequence of rh problems on the circles @xmath14 .
this method has been first applied by halsey @xcite for @xmath132 .
applications of this problem to compute conformal mapping have been given in @xcite for @xmath133 and in @xcite for @xmath132 .
in this paper , we shall present a method for solving the general conjugation problem for any @xmath128 .
the method is based on the boundary integral equation with the generalized neumann kernel ( [ eq : ie ] ) . however , we shall first rewrite the problem in a form suitable for using the integral equation .
the boundary condition can be written as @xmath134=\gamma_{j } -a_{j0}-\sum_{k=1}^{\ell}{\mathop{\mathrm{re}}}\left[(a_{jk}+{\mathrm{i}}b_{jk})e^{{\mathrm{i}}kt}\right ] , \quad j=1,2,\ldots , m.\ ] ] let @xmath3 be defined by @xmath135,\\ \hspace{0.3cm}\vdots\\ e^{{\mathrm{i}}\lambda_m}e^{{\mathrm{i}}\ell t},&t\in j_m=[0,2\pi ] .
\end{array } \right .\ ] ] let also @xmath136 be a function defined on the boundary @xmath42 where its values on @xmath14 is given by @xmath137 , \quad j=1,2,\ldots , m.\ ] ] hence , the general conjugation problem can be written as the rh - problem @xmath138=\gamma+\psi.\ ] ] by the existence and uniqueness of the solution general conjugation problem , the rh - problem has a unique solution .
remember that solving the rh - problem requires determining the analytic function @xmath70 and the unknown real function @xmath136 . determining the function
@xmath136 is equivalent to determining the @xmath139 real constants @xmath140 , @xmath141 , and @xmath142 in for @xmath39 , @xmath143 .
the index of the function @xmath3 given by is @xmath144 and hence the total index is @xmath145 .
thus our assumption is satisfied .
we shall use the integral equation with the generalized neumann kernel ( [ eq : ie ] ) to determine the analytic function @xmath70 as well as the real function @xmath136 in .
however , we need to show first that the function @xmath136 is indeed in @xmath89 . this can be proved by finding the explicit form of the functions of the space @xmath89 which will be given in the next section .
to find the explicit form of the space @xmath89 , we need the following theorem from @xcite . [
t : gak ] let @xmath146 and the boundary @xmath147 is the unit circle parametrized by @xmath148 , @xmath149 $ ] . for @xmath128 , the solution of the following homogenous rh problem on @xmath150 , @xmath151=0 , \quad \zeta(t)\in c,\ ] ] is given for @xmath152 by @xmath153\ ] ] where @xmath154 is an arbitrary real constant and @xmath155 are arbitrary complex constants .
[ l : sk ] let @xmath156 .
then @xmath157 if and only if it has the form @xmath158 , & \mbox{\rm on $ c_j$ } , \end{array } \right.\ ] ] where @xmath159 , @xmath91 , @xmath160 , are real constants . by the definition of the space @xmath161 , a function @xmath157 if and only if @xmath162 where @xmath84 is analytic in @xmath10 and @xmath163 is the restriction of the function @xmath3 given by to @xmath26 . using the definition of the function @xmath3 ,
the function @xmath164 is a solution of the homogeneous rh problem @xmath165=0\ ] ] on the disk @xmath10 .
we define an analytic function @xmath166 in @xmath10 by @xmath167 then @xmath166 is a solution of the homogeneous rh problem @xmath168=0\ ] ] on the disk @xmath10 .
the function @xmath169 is analytic on @xmath10 and maps the circle @xmath14 onto the unit circle and @xmath170 is the parametrization of the unit circle .
let the function @xmath171 be defined on the unit disk @xmath150 by @xmath172 then , it follows from that @xmath171 is a solution of the homogeneous rh problem @xmath173=0\ ] ] in the unit disk @xmath150 .
hence theorem [ t : gak ] implies that @xmath174\ ] ] where @xmath159 , @xmath91 , @xmath160 , are arbitrary real constants . since @xmath148 and @xmath175 , we obtain @xmath176.\ ] ] since @xmath177 , implies that @xmath178\ ] ] which can be written as @xmath179.\ ] ] hence follows from and .
[ t : s - gam ] let @xmath156 .
then @xmath103 if and only if it has the form @xmath180,\ ] ] where @xmath159 , @xmath91 , @xmath160 , are @xmath181 real constants .
the proof follows from lemmas [ l : s - sk ] and [ l : sk ] . in view of ( [ eq : s - gam ] )
, it is clear that the function @xmath136 in ( [ eq : psi ] ) is in the space @xmath89 . by the uniqueness of the function @xmath136 in ( [ eq : rhp-2 ] ) and the function @xmath122 in ( [ eq : rhp2 ] ) , we conclude that the functions @xmath136 and @xmath122 are identical , i.e. , @xmath182 where @xmath122 is given by ( [ eq : h ] ) .
since the function @xmath136 in ( [ eq : rhp-2 ] ) and the function @xmath122 given by ( [ eq : h ] ) are identical , the rh - problem ( [ eq : rhp - w2 ] ) or equivalently the general conjugation problem can be solved by the integral equation with the generalized neumann kernel ( [ eq : ie ] ) as in the following theorem .
[ t : rhpw - psi ] for any @xmath44 , the boundary values of the unique solution @xmath70 of the general conjugation problem are given by @xmath183 where @xmath184 is defined by , @xmath185 is the unique solution of the integral equation and the function @xmath186 is given by . to evaluate the @xmath130 unknown real constants @xmath140 , @xmath141 and @xmath142 in ( [ eq : rhp - w ] ) , @xmath39 , @xmath143 , we rewrite the function @xmath187 in ( [ eq : psi ] ) as @xmath188 by obtaining the real function @xmath122 from ( [ eq : h ] ) and since @xmath189 , we have the following theorem . [
t : rhpw - ajbj ] the values of the @xmath130 unknown real constants @xmath141 and @xmath142 in are given by @xmath190 for @xmath39 and @xmath143 where the function @xmath186 is given by . in the above two theorems
, we have used the integral equation ( [ eq : ie ] ) to develop a method for solving the rh problem ( [ eq : rhp - w ] ) .
we can also use theorem [ t : s - gam ] and corollary [ c : g+h ] to provide alternative proof for theorem [ t : weg ] for any @xmath44 . for any integer @xmath128 ,
let the function @xmath3 be given by .
then for any functions @xmath44 , it follows from corollary [ c : g+h ] that a unique function @xmath103 exists such that the rh problem @xmath191=\gamma+h\ ] ] is uniquely solvable . the function @xmath122 is given by ( [ eq : h ] ) where @xmath123 is the unique solution of the integral equation
. then theorem [ t : s - gam ] implies that unique values of the @xmath130 real constants @xmath159 , @xmath91 , @xmath160 , exist such that @xmath192.\ ] ] hence the rh problem can be written as @xmath193\right)=\gamma_{j},\ ] ] which in view of the definition of the function @xmath3 gives the proof of theorem [ t : weg ] where @xmath194 and @xmath195 , @xmath91 , @xmath160 .
we have presented a boundary integral method for solving the general conjugation problem which is a certain class of riemann - hilbert problems .
the method is based on a uniquely solvable boundary integral equation with the generalized neumann kernel .
we have also presented an alternative proof of theorem [ t : weg ] which has been proved previously in @xcite .
the numerical implementation of the above proposed method will be presented in future works .
al - hatemi , a.h.m .
murid and m.m.s .
nasser , _ a boundary integral equation with the generalized neumann kernel for a mixed boundary value problem in unbounded multiply connected regions .
value probl .
* 2013 * ( 2013 ) , article no .
54 .
murid , m.r.m .
razali and m.m.s .
nasser , _ solving riemann problem using fredholm integral equation of the second kind .
_ in : a.h.m .
murid ( ed . ) , proceeding of simposium kebangsaan sains matematik ke-10 , utm , johor , malaysia , 2002 , pp .
171178 .
nasser , a.h.m .
murid and s.a.a .
al - hatemi , _ a boundary integral equation with the generalized neumann kernel for a certain class of mixed boundary value problem .
_ j. appl . math .
* 2012 * ( 2012 ) , article i d 254123 , 17 pages .
nasser , a.h.m .
murid , m. ismail and e.m.a .
alejaily , _ boundary integral equations with the generalized neumann kernel for laplace s equation in multiply connected regions .
_ appl . math .
* 217 * ( 2011 ) , 47104727 . | we present a boundary integral method for solving a certain class of riemann - hilbert problems known as the general conjugation problem .
the method is based on a uniquely solvable boundary integral equation with the generalized neumann kernel .
we present also an alternative proof for the existence and uniqueness of the solution of the general conjugation problem . |
fractal dimensions such as hausdorff dimension , box - counting dimension and packing dimension are very useful in characterizing roughness or irregularity of stochastic processes and random fields which , in turn , serve as stochastic models in various scientific areas including image processing , hydrology , geostatistics and spatial statistics .
many authors have studied the hausdorff dimension and exact hausdorff measure of the image sets of markov processes and gaussian random fields .
we refer to taylor ( @xcite ) and xiao ( @xcite ) for extensive surveys on results and techniques for markov processes , and to adler ( @xcite ) and kahane ( @xcite ) for results on gaussian random fields .
let @xmath6 be a random field with values in @xmath7 , which will simply be called an @xmath8__-random field__. for any finite borel measure @xmath2 on @xmath9 , the image measure of @xmath2 under @xmath10 is defined by @xmath11 .
similarly , for every @xmath12 , the image set is denoted by @xmath13 .
this paper is concerned with the hausdorff and packing dimensions of the image measures and image sets of random fields which are , in a certain sense , comparable to a self - similar process . recall that @xmath14 is said to be @xmath15__-self - similar _ _
if , for every constant @xmath16 , we have @xmath17 and @xmath10 is said to have _ stationary increments _ if , for every @xmath18 , @xmath19 where @xmath20 denotes equality of all finite - dimensional distributions .
if @xmath10 satisfies both ( [ eq : ss0 ] ) and ( [ eq : si ] ) , then it is called an @xmath15__-sssi random field__. samorodnitsky and taqqu ( @xcite ) give a systematic account of self - similar stable processes .
the main results of this paper show that the hausdorff and packing dimensions of the images of an @xmath15-sssi random field @xmath10 are determined by the self - similarity index @xmath15 and essentially do not depend on the distributions of @xmath10 .
an important example of an @xmath15-sssi @xmath8-random field is fractional brownian motion @xmath21 of index @xmath15 @xmath22 , which is a centered gaussian random field with the covariance function @xmath23=\frac{1}{2}\delta_{i , j } ( \|s\|^{2h}+\|t\|^{2h } -\|t - s\|^{2h } ) $ ] for all @xmath24 , where @xmath25 if @xmath26 and @xmath27 otherwise .
it is well known ( see kahane ( @xcite ) , chapter 18 ) that for every borel set @xmath12 , @xmath28 where @xmath29 denotes the hausdorff dimension . on the other hand , talagrand and xiao ( @xcite ) proved that , when @xmath30 , the packing dimension analog of ( [ eq : bg60 ] ) fails in general .
xiao ( @xcite ) proved that @xmath31 where @xmath32 denotes packing dimension and @xmath33 is the packing dimension profile of @xmath34 defined by falconer and howroyd ( @xcite ) ( see section [ sec : pre ] for its definition ) .
results ( [ eq : bg60 ] ) and ( [ eq : xiao97 ] ) show that there are significant differences between hausdorff dimension and packing dimension , and both dimensions are needed for characterizing the fractal structures of @xmath5 .
there have been various efforts to extend ( [ eq : bg60 ] ) to other non - markovian processes or random fields , but with only partial success ; see kno ( @xcite ) , lin and xiao ( @xcite ) , benassi , cohen and istas ( @xcite ) and xiao ( @xcite ) . in order to establish a hausdorff dimension result similar to ( [ eq : bg60 ] ) for a random field @xmath10 ,
it is standard to determine upper and lower bounds for @xmath35 separately .
while the capacity argument ( based on frostman s theorem ) is useful for determining lower bounds , the methods based on the classical covering argument for establishing an upper bound for @xmath35 are quite restrictive and usually require strong conditions to be imposed on @xmath10 . as such , the aforementioned authors have only considered random fields which either satisfy a uniform hlder condition of appropriate order on compact sets or have at least the first moment . in particular , the existing methods are not enough , even for determining @xmath36^n),$ ] when @xmath21 is a general stable random field . given a random field @xmath37 and a borel set @xmath38 , it is usually more difficult to determine the packing dimension of the image set @xmath5 .
recently , khoshnevisan and xiao ( @xcite ) and khoshnevisan , schilling and xiao ( @xcite ) have solved the above problem when @xmath39 is a lvy process in @xmath7 .
however , their method depends crucially on the strong markov property of lvy processes and can not be applied directly to random fields .
this paper is motivated by the need to develop methods for determining the hausdorff and packing dimensions of the image measure @xmath4 and image set @xmath5 under minimal conditions on the random field @xmath10 . by applying measure - theoretic methods and the theory of packing dimension profiles , we are able to solve the problems for the hausdorff and packing dimensions of the image measure @xmath4 under mild conditions ( namely , ( c1 ) and ( c2 ) in section [ sec : measures ] ) .
the main results are theorem [ th : hdim1 ] and theorem [ th : main1 ] .
when @xmath10 satisfies certain uniform hlder conditions , theorems [ th : hdim1 ] and [ th : main1 ] can be applied directly to compute the hausdorff and packing dimensions of @xmath5 .
more generally , we also provide a method for determining the hausdorff dimension of @xmath5 under conditions ( c1 ) and ( c2 ) ( see theorem [ th : dimxe2 ] ) .
however , we have not been able to solve the problem of determining @xmath40 in general .
the rest of this paper is organized as follows . in section [ sec : pre ] , we recall the definitions and some basic properties of hausdorff dimension , packing dimension and packing dimension profiles of sets and borel measures . in section [ sec
: measures ] , we determine the hausdorff and packing dimensions of the image measure @xmath4 under general conditions ( c1 ) and ( c2 ) . in section
[ sec : set ] , we study the hausdorff and packing dimensions of the image set @xmath5 , where @xmath12 is an analytic set ( i.e. , @xmath34 is a continuous image of the baire space @xmath41 or , equivalently , @xmath34 is a continuous image of a borel set ) .
section [ sec : examples ] contains applications of the theorems in sections [ sec : measures ] and [ sec : set ] to sssi stable random fields , real harmonizable fractional lvy fields and the rosenblatt process . throughout this paper
, we will use @xmath42 to denote the inner product and @xmath43 to denote the euclidean norm in @xmath44 , no matter what the value of @xmath45 is . for any @xmath46 such that @xmath47 ( @xmath48 ) , @xmath49 = \prod^n_{j=1 } [ s_j , t_j]$ ] is called a _ closed interval_. we will use @xmath50 to denote an unspecified positive constant which may differ from line to line .
specific constants in section @xmath51 will be denoted by @xmath52
in this section , we recall briefly the definitions and some basic properties of hausdorff dimension , packing dimension and packing dimension profiles . more information on hausdorff and packing dimensions can be found in falconer ( @xcite ) and mattila ( @xcite ) . for any @xmath53 , the @xmath54-dimensional hausdorff measure of @xmath12
is defined by @xmath55 where @xmath56 .
the hausdorff dimension of @xmath34 is defined as @xmath57 for a finite borel measure @xmath2 on @xmath9 , its hausdorff dimension is defined by @xmath58 and its upper hausdorff dimension is defined by @xmath59 hu and taylor ( @xcite ) proved that @xmath60 the hausdorff dimensions of an analytic set @xmath12 and finite borel measures on @xmath34 are related by the following identity ( which can be verified by ( [ eq : hutaylor1 ] ) and frostman s lemma ) : @xmath61 where @xmath62 denotes the family of all finite borel measures with compact support in @xmath34 .
packing dimension was introduced by tricot ( @xcite ) as a dual concept to hausdorff dimension and has become a useful tool for analyzing fractal sets and sample paths of stochastic processes ; see taylor and tricot ( @xcite ) , taylor ( @xcite ) , talagrand and xiao ( @xcite ) , falconer and howroyd ( @xcite ) , howroyd ( @xcite ) , xiao ( @xcite , @xcite , @xcite ) , khoshnevisan and xiao ( @xcite , @xcite ) , khoshnevisan , schilling and xiao ( @xcite ) and the references therein for more information . for any @xmath53 , the @xmath63-dimensional packing measure of @xmath12
is defined as @xmath64 where @xmath65-@xmath66 is the set function on subsets of @xmath67 defined by @xmath68 the packing dimension of @xmath34 is defined by @xmath69 it is well known that @xmath70 for every set @xmath71 .
the packing dimension of a finite borel measure @xmath2 on @xmath9 is defined by @xmath72 and the upper packing dimension of @xmath2 is defined by @xmath73 in analogy to ( [ eq : setdim ] ) , falconer and howroyd ( @xcite ) proved , for every analytic set @xmath12 , that @xmath74 next , we recall some aspects of the packing dimension profiles of falconer and howroyd ( @xcite ) and howroyd ( @xcite ) . for a finite borel measure @xmath2 on @xmath9 and for any @xmath75 ,
let @xmath76 be the potential with respect to the kernel @xmath77 falconer and howroyd ( @xcite ) defined the packing dimension profile and the upper packing dimension profile of @xmath2 as @xmath78 and @xmath79 respectively .
further , they showed that @xmath80 and , if @xmath81 , then @xmath82 motivated by ( [ eq : setdim ] ) , falconer and howroyd ( @xcite ) defined the @xmath83-dimensional packing dimension profile of @xmath84 by @xmath85 it follows that @xmath86 by the above definition , it can be verified ( see falconer and howroyd ( @xcite ) , page 286 ) that for every borel set @xmath87 with @xmath88 , we have @xmath89 the following lemma is a consequence of proposition 18 in falconer and howroyd ( @xcite ) .
[ lem : con ] let @xmath2 be a finite borel measure on @xmath9 and @xmath12 be bounded and non - empty .
let @xmath90 $ ] be any one of the functions @xmath91 @xmath92 or @xmath93 in @xmath83 .
then @xmath94 is non - decreasing and continuous .
let @xmath6 be an @xmath8-random field defined on some probability space @xmath95 .
we assume throughout this paper that @xmath10 is separable ( i.e. , there exists a countable and dense set @xmath96 and a zero probability event @xmath97 such that for every open set @xmath98 and closed set @xmath99 , the two events @xmath100 and @xmath101 differ from each other only by a subset of @xmath97 ; in this case , @xmath102 is called a _
separant _ for @xmath10 ) and @xmath103 is @xmath104-measurable , where @xmath105 is the borel @xmath106-algebra of @xmath9 . for any borel measure @xmath2 on @xmath107 ,
the image measure @xmath4 of @xmath2 under @xmath108 is @xmath109 in this section , we derive upper and lower bounds for the hausdorff and packing dimensions of the image measures of @xmath10 , which rely , respectively , on the following conditions ( c1 ) and ( c2 ) .
analogous problems for the image set @xmath5 will be considered in section [ sec : set ] . 1 .
there exist positive and finite constants @xmath110 and @xmath111 such that @xmath112 for all @xmath113 , @xmath114 and @xmath115 , where @xmath116 , @xmath117 and @xmath118 are positive constants . 2 . there exists a positive constant @xmath119 such that for all @xmath120 and @xmath121 , @xmath122 where @xmath123 is a finite constant . since ( c1 ) and ( c2 )
play essential roles in this paper , we will now make some relevant remarks about them .
* condition ( c1 ) is a type of local maximal inequality and is easier to verify when the random field @xmath10 has a certain approximate self - similarity .
for example , if @xmath10 is @xmath110-self - similar , then condition ( c1 ) is satisfied whenever the tail probability of @xmath124 decays no slower than a polynomial rate ; see proposition [ prop : tail ] below and section [ sec : examples ] .
it can also be verified directly for gaussian or more general infinitely random fields by using large deviations techniques without appealing to self - similarity .
* there may be different pairs of @xmath125 for which ( c1 ) is satisfied .
we note that the formulae for hausdorff and packing dimensions of the images do not depend on the constant @xmath126 , it is @xmath127 that determines the best upper bounds for the hausdorff and packing dimensions of the image measures .
* for every point @xmath128 , the local hlder exponent of @xmath10 at @xmath129 is defined as @xmath130 condition ( c1 ) and the borel
cantelli lemma imply that @xmath131 almost surely ( see ( [ eq : lil ] ) below ) .
however , ( c1 ) does not even imply sample path continuity of @xmath10 . * in section [ sec : set ] , the following , slightly weaker , form of condition ( c2 ) will be sufficient : 1 .
there exist positive constants @xmath119 and @xmath132 such that ( [ eq : c2 ] ) holds for all @xmath120 satisfying @xmath133 and @xmath121 .
* condition ( c2 ) ( or ( c@xmath134 ) ) is satisfied if , for all @xmath135 ( or those satisfying @xmath136 ) , the random vector @xmath137 has a density function which is uniformly bounded in @xmath83 and @xmath129 . as shown by proposition [ prop : c2 ] below ,
( c2 ) is significantly weaker than the latter .
the following proposition gives a simple sufficient condition for an sssi process @xmath138 to satisfy condition ( c1 ) .
more precise information can be obtained if further distributional properties of @xmath10 are known ; see section [ sec : examples ] . [
prop : tail ] let @xmath139 be a separable , @xmath15-sssi process with values in @xmath7 . if there exist positive constants @xmath126 and @xmath140 such that @xmath141 and @xmath142 then there exists a positive constant @xmath143 such that for all @xmath144 , @xmath145 } \|x(t)\| \ge u \bigr\ } \le k_{{3,4 } } u^{-\beta}.\ ] ] in particular , condition is satisfied with @xmath146 and the same @xmath111 as in ( [ eq : x1tail ] ) . without loss of generality
, we can assume @xmath147 . since the self - similarity index @xmath148 , we have @xmath149 a.s .
let @xmath150 be a separant for @xmath151\}$ ] .
we assume that @xmath152 . for any @xmath153 , consider the random variables @xmath154 @xmath155 defined by @xmath156 for @xmath157 , let @xmath158 . by the stationarity of increments and self - similarity of @xmath10 and ( [ eq : x1tail ] )
, we derive that for any @xmath144 , @xmath159 thus , condition ( 3.4 ) of theorem 3.2 of moricz , serfling and stout ( @xcite ) is satisfied with @xmath160 , @xmath161 and @xmath162 .
it is easy to see that the non - negative function @xmath163 satisfies their condition ( 1.2 ) ( i.e. , @xmath164 and @xmath165 for @xmath166 ) with @xmath167 .
it therefore follows from theorem 3.2 of moricz , serfling and stout ( @xcite ) that there exists a constant @xmath143 ( independent of @xmath45 ) such that for all @xmath144 , @xmath168 letting @xmath169 yields ( [ eq : sstail ] ) , which , in turn , implies that ( c1 ) holds for @xmath146 .
next , we provide a necessary and sufficient condition for an @xmath170-random field @xmath14 to satisfy condition ( c2 ) ( or ( c@xmath134 ) ) . for any @xmath121 ,
let @xmath171 [ prop : c2 ] let @xmath172 be a random field with values in @xmath7 . condition ( or ( @xmath134 ) ) then holds if and only if there exists a positive constant @xmath173 such that for all @xmath121 and all @xmath120 @xmath174or for those satisfying @xmath175 , @xmath176 since @xmath177 as @xmath178 , condition ( [ eq : c2-con ] ) is significantly weaker than assuming that @xmath179 has a bounded density and can be applied conveniently to sssi processes .
we mention that ( [ eq : c2-con ] ) is also weaker than the integrability condition in assumption 1 on page 269 of benassi , cohen and istas ( @xcite ) . it can be shown that theorem 2.1 in benassi , cohen and istas ( @xcite ) still holds under ( [ eq : c2-con ] ) and their assumption 2 .
proof of proposition [ prop : c2 ] note that for every @xmath121 , the function @xmath180 is non - negative and is in @xmath181 .
the fourier transform of @xmath182 is @xmath183 in the above , @xmath184 for all @xmath185 .
since @xmath186 implies that @xmath187 , we have @xmath188 for all @xmath189 . here , and in the sequel , @xmath190 denotes the indicator function ( or random variable ) of the set ( or event ) @xmath191 . by fubini s theorem ,
we obtain @xmath192\\ & = & 2^d \int_{\r^d } \phi_r(x ) \e\bigl ( \mathrm{e}^{\mathrm{i } \l x , x(t ) - x(s)\rangle /\| t - s\|^{h_2 } } \bigr)\ , \mathrm{d}x .\end{aligned}\ ] ] hence , ( [ eq : c2-con ] ) implies condition ( c2 ) . on the other hand
, we have @xmath193 for all @xmath194 .
consequently , @xmath195 \le\p \bigl\ { \|x(s ) - x(t ) \| \le2\sqrt{d } \|s - t\|^{h_2 } r \bigr\ } .\ ] ] therefore , condition ( c2 ) implies ( [ eq : c2-con ] ) .
this completes the proof .
first , we consider the upper bounds for the hausdorff dimensions of the image measure @xmath4 . [ prop : hdimub1 ] let @xmath196 be a random field with values in @xmath7 . if condition is satisfied , then for every finite borel measure @xmath2 on @xmath197 , @xmath198\\[-8pt ] & & \dimh^ * \mu_{x } \le\min \biggl\{d , \frac1 { h_1 } \dimh^ * \mu \biggr\}\qquad \mathrm{a.s.}\nonumber\end{aligned}\ ] ] let @xmath199 be a constant . for any fixed @xmath200 and the sequence @xmath201 ( @xmath202 ) , it follows from condition ( c1 ) that for all integers @xmath203 , @xmath204 since @xmath205 , the borel
cantelli lemma implies that almost surely @xmath206 where @xmath207 depends on @xmath208 and @xmath83 . by fubini s theorem
, we derive that , for any finite borel measure @xmath2 on @xmath197 , almost surely ( [ eq : lil ] ) holds for @xmath2-a.a .
@xmath200 .
we now fix an @xmath209 such that ( [ eq : lil ] ) is valid for @xmath2-a.a .
@xmath200 and prove that both inequalities in ( [ eq : hupp1 ] ) hold . in the sequel , @xmath208 will be suppressed .
to prove the first inequality in ( [ eq : hupp1 ] ) , since @xmath210 holds trivially , we only need to prove that @xmath211 . without loss of generality , we assume @xmath212 and take any @xmath213 then , by ( [ eq : hutaylor1 ] ) , we have @xmath214 equivalently to ( [ eq : hupp11 ] ) ,
we have @xmath215 let us fix @xmath200 such that both ( [ eq : lil ] ) and ( [ eq:211 ] ) hold . for any @xmath216 , we choose @xmath217 such that @xmath218 for all @xmath219 . by ( [ eq : lil ] ) ,
we can write @xmath220\\[-8pt ] & \ge&\int_{\|t - s\| < 2^{-n_1 } } \i_{\{\|t - s\| \le r^{1/(h_1-\ep)}\ } } \,\mathrm{d}\mu(t ) .\nonumber\end{aligned}\ ] ] hence , we have @xmath221\\[-8pt ] & & { } + \int_{\|t - s\|\ge 2^{-n_1 } } \i_{\{\|t - s\| \le r^{1/(h_1 -\varepsilon ) } \ } } \,\mathrm{d}\mu(t ) .\nonumber\end{aligned}\ ] ] for the last integral , we have @xmath222 because the indicator function takes the value 0 when @xmath223 is sufficiently small .
it follows from ( [ eq:211 ] ) , ( [ eq:213 ] ) and ( [ eq:214 ] ) that with @xmath224 , @xmath225\\[-8pt ] & \le&\limsup_{r \to0 } r^{-\ga } \int_{\r^n}\i_{\ { \|x(t ) - x(s)\| \le r\ } } \,\mathrm{d}\mu(t)= 0.\qquad\mbox{}\nonumber\end{aligned}\ ] ] we have thus proven that ( [ eq:215 ] ) holds almost surely for @xmath2-a.a .
this implies that @xmath226 almost surely .
since @xmath227 and @xmath228 are arbitrary , ( [ eq : hupp1 ] ) follows . to prove the second inequality in ( [ eq : hupp1 ] ) , it is sufficient to show that @xmath229 a.s .
let @xmath230 be fixed as above .
we take an arbitrary @xmath231 . by ( [ eq : hutaylor2 ] )
, we have @xmath232 by using ( [ eq:212 ] ) , we derive that for @xmath233 , @xmath234\\[-8pt ] & & { } - \int_{\|t - s\|\ge 2^{-n_1 } } \i_{\ { \|t - s\| \le r^{1/(h_1- \varepsilon)}\ } } \,\mathrm{d}\mu(t ) .\nonumber\end{aligned}\ ] ] it follows from ( [ eq:227 ] ) , ( [ eq:214 ] ) and ( [ eq:226 ] ) that @xmath235 for all @xmath200 that satisfy ( [ eq:226 ] ) .
this implies that @xmath236 a.s . letting @xmath237 and @xmath238 yields the second inequality in ( [ eq : hupp1 ] ) .
this completes the proof of proposition [ prop : hdimub1 ] .
[ re : hdimub1 ] note that in ( [ eq : hupp1 ] ) , the exceptional null probability events depend on @xmath2 . for several purposes ,
it is more useful to have a single exceptional null probability event @xmath239 such that , for all @xmath240 , both inequalities in ( [ eq : hupp1 ] ) hold _ simultaneously _ for all finite borel measures @xmath2 on @xmath9 . by slightly modifying the proof of proposition [ prop : hdimub1 ] ( see ( [ eq:212 ] ) )
, one can show that this is indeed true if , for every @xmath241 and every compact interval @xmath242 , the sample function @xmath243 satisfies almost surely a uniform hlder condition of order @xmath244 on @xmath242 .
next , we show that condition ( c2 ) determines lower bounds for the hausdorff dimensions of the image measures of the random field @xmath10 .
[ prop : hdimlb1 ] let @xmath245 be an @xmath8-random field satisfying condition . then
, for every finite borel measure @xmath2 on @xmath246 , @xmath247 in order to prove the first inequality in ( [ eq : hlow1 ] ) , we fix any constants @xmath248 . since @xmath249 , it follows from ( [ eq : hutaylor1 ] ) that @xmath250 let @xmath200 be a fixed point such that ( [ eq : hlow3 ] ) holds . by ( c2 ) , we derive @xmath251\\[-8pt ] & \le & k_{{3,2 } } \mu(b(s , r^{1/h_2 } ) ) + k_{{3,2 } } \int_{\|t -s \| > r^{1/h_2 } } \biggl(\frac r { \|t - s\|^{h_2 } } \biggr)^d\mu(\mathrm{d}t ) .\nonumber\end{aligned}\ ] ] let @xmath252 be the image measure of @xmath2 under the mapping @xmath253 from @xmath9 to @xmath254 . then , by using an integration - by - parts formula and ( [ eq : hlow3 ] ) , we have @xmath255 for all
@xmath223 small enough , where the last inequality follows from ( [ eq : hlow3 ] ) and the fact that @xmath256 . combining ( [ eq : hlow4 ] ) and ( [ eq : hlow5 ] ) , we see that @xmath257 for @xmath223 small .
this , and the markov inequality , imply that for all @xmath45 large enough , @xmath258 it follows from the borel cantelli lemma that a.s .
@xmath259 for all @xmath45 large enough .
it should be clear the above implies that for all @xmath260 , @xmath261 almost surely .
thus , @xmath262 a.s . , and ( [ eq : hlow1 ] ) follows from the arbitrariness of @xmath263 . to prove the second inequality in ( [ eq : hlow1 ] ) , let @xmath264 . by ( [ eq : hutaylor2 ] ) , there exists a borel set @xmath265 such that @xmath266 and @xmath267 for all @xmath268 the proof above shows that a.s .
@xmath269 for all @xmath270 . since @xmath271 a.s .
, we derive @xmath272 a.s . and
the proof is completed .
combining propositions [ prop : hdimub1 ] and [ prop : hdimlb1 ] , we have the following theorem , whose proof is omitted . [
th : hdim1 ] let @xmath245 be an @xmath8-random field and let @xmath15 be a positive constant . if , for every @xmath241 , @xmath10 satisfies condition with @xmath273 , some @xmath274 and with @xmath275 , then for every finite borel measure @xmath2 on @xmath9 , @xmath276 we now study the problem of determining the packing dimensions @xmath277 and @xmath278 .
the following upper bounds for the image measures are proved by schilling and xiao ( @xcite ) .
[ lem : uppbd ] let @xmath196 be a random field with values in @xmath7 .
if condition is satisfied , then for every finite borel measure @xmath2 on @xmath197 , @xmath279 similarly to remark [ re : hdimub1 ] , we have the following . [ re : uppbd ] if , for every @xmath241 and every compact interval @xmath280 , @xmath243 satisfies almost surely a uniform hlder condition of order @xmath281 on @xmath242 , then almost surely both inequalities in ( [ eq : upp1 ] ) hold for all finite borel measures @xmath2 on @xmath197 . for the lower bounds of packing dimensions , we have the following proposition .
[ prop : lowerbd ] let @xmath245 be an @xmath8-random field satisfying condition . then
, for every finite borel measure @xmath2 on @xmath246 , @xmath282 we only prove the first inequality in ( [ eq : mdim1 ] ) ; the proof of the second one is similar .
we may , and will , assume that @xmath283 . for fixed @xmath284 , fubini s theorem implies that @xmath285 the integrand in ( [ eq:3lower ] ) can be written as @xmath286\\[-8pt ] & & \quad = { \p } \ { \|x(t ) - x(s)\| \le r \ } + { \e } \bigl\{r^d \|x(t ) - x(s)\|^{-d } \cdot\i_{\ { \|x(t ) - x(s)\| \ge r\ } } \bigr\ } .\nonumber % \end{aligned}\ ] ] by condition ( c2 ) , we obtain that for all @xmath24 and @xmath287 , @xmath288 denote the distribution of @xmath289 by @xmath290 .
let @xmath291 be the image measure of @xmath290 under the mapping @xmath292 from @xmath7 to @xmath254 .
the last term in ( [ eq:3lower2 ] ) can then be written as @xmath293\\[-8pt ] & \le & d \int_{r}^\infty\frac { r^d } { \rho^{d+1 } } \p\{\|x(t ) -
x(s)\| \le\rho\ } \,\mathrm{d}\rho,\nonumber\end{aligned}\ ] ] where the last inequality follows from an integration - by - parts formula . by ( [ eq:3lower2a ] ) and ( [ eq:3lower3 ] )
, we derive that the last term in ( [ eq:3lower2 ] ) can be bounded by a constant multiple of @xmath294\\[-8pt ] & & \quad\le \cases { k , & \quad if $ r \ge\|t - s\|^{h_2 } $ , \vspace*{2pt}\cr \displaystyle\frac{k r^d } { \|t - s\|^{h_2d } } \log \biggl(\frac { \|t - s\|^{h_2 } } { r } \biggr ) , & \quad if $ r < \|t - s\|^{h_2 } $ . } \nonumber % \end{aligned}\ ] ] it follows from ( [ eq:3lower2 ] ) , ( [ eq:3lower2a ] ) , ( [ eq:3lower3 ] ) and ( [ eq:3lower4 ] ) that for any @xmath295 and @xmath120 , @xmath296 for any @xmath297 by lemma [ lem : con ] , there exists @xmath241 such that @xmath298 it follows from ( [ def : dimprof ] ) that @xmath299 by ( [ eq:3lower ] ) , ( [ eq:3lower5 ] ) , ( [ eq:3lower6 ] ) and fatou s lemma , we have that for @xmath2-a.a .
@xmath300 , @xmath301\\[-8pt ] & & \quad\le k_{{3 , 6 } } \liminf_{r \to0 } r^{- { \ga}/{h_2 } } \int_{\r^n } \min\biggl\{1 , \frac{r^{d-\ep } } { \|t- s\|^{h_2(d -\ep ) } } \biggr\ } \,\mathrm{d}\mu(t)= 0 .\nonumber\end{aligned}\ ] ] by using fubini s theorem again , we see that almost surely @xmath302 hence , @xmath303 a.s .
since @xmath263 can be arbitrarily close to @xmath304 , we obtain ( [ eq : mdim1 ] ) .
the following is a direct consequence of propositions [ lem : uppbd ] and [ prop : lowerbd ] .
[ th : main1 ] let @xmath245 be an @xmath8-random field and let @xmath15 be a positive constant . if , for every @xmath241 , @xmath10 satisfies condition with @xmath273 , some @xmath274 and with @xmath275 , then for every finite borel measure @xmath2 on @xmath246 , @xmath305
we now consider the hausdorff and packing dimensions of the image set @xmath5 .
we will see that general lower bounds for @xmath35 and @xmath40 can be derived from the results in section [ sec : measures ] by using a measure theoretic method . for random fields which satisfy uniform hlder conditions on compact intervals , the upper bounds for @xmath35 and @xmath40 can also be easily obtained . however , it is difficult to obtain upper bounds for @xmath35 and @xmath40 under condition ( c1 ) alone .
we have only been able to provide a partial result on determining the upper bound for @xmath35 .
the analogous problem for @xmath40 remains open .
we will need the following lemmas .
lemma [ lem : edgar ] is from lubin ( @xcite ) , which is more general than theorem 1.20 in mattila ( @xcite ) .
[ lem : edgar ] let @xmath12 be an analytic set and let @xmath306 be a borel function . if @xmath291 is a finite borel measure on @xmath7 with support in @xmath307 then @xmath308 for some @xmath309 .
[ lem : setdim ] let @xmath12 be an analytic set .
then , for all borel measurable functions @xmath310 , we have @xmath311 denote the right - hand side of ( [ eq:41 ] ) by @xmath312 . by ( [ eq : setdim ] )
, we get @xmath313 .
next , for any @xmath314 , lemma [ lem : edgar ] implies that @xmath308 for some @xmath315 .
this and ( [ eq : setdim ] ) together imply @xmath316 . hence , ( [ eq:41 ] ) is proved .
the proof of ( [ eq:42 ] ) is similar and is therefore omitted .
we first consider the lower bounds for the hausdorff and packing dimensions of @xmath5 .
[ prop : lowerbd2 ] let @xmath245 be an @xmath8-random field that satisfies condition ( @xmath134 ) .
then , for every analytic set @xmath317 , @xmath318 since both @xmath29 and @xmath32 are @xmath106-stable ( see falconer ( @xcite ) ) , we may , and will , assume that the diameter of @xmath34 is at most 1 .
hence , condition ( c2@xmath319 ) will be enough to prove ( [ eq : setlb1 ] ) .
let us prove the first inequality in ( [ eq : setlb1 ] ) .
it follows from ( [ eq : setdim ] ) that for any @xmath320 , there exists a @xmath321 such that @xmath322 . by proposition [ prop : hdimlb1 ] ( which holds for any finite borel measure whose support has diameter @xmath323
, we have @xmath324 a.s .
this and ( [ eq:41 ] ) together imply that @xmath325 a.s .
since @xmath326 is arbitrary , the desired inequality follows .
next , we prove the second inequality in ( [ eq : setlb1 ] ) .
note that for any @xmath327 , by ( [ eq : setdimprofile ] ) , there exists a borel measure @xmath315 such that @xmath328 it follows from ( [ eq : mdim1 ] ) that @xmath329 a.s .
hence , by lemma [ lem : setdim ] , we have @xmath330 a.s . , which , in turn , implies that @xmath331 a.s . the proof is therefore completed .
the following proposition gives upper bounds for @xmath35 and @xmath40 .
[ prop : setupp1 ] let @xmath196 be an ( @xmath332)-random field . if for every @xmath241 , @xmath10 satisfies a uniform hlder condition of order @xmath244 on all compact intervals of @xmath9 almost surely , then , for all analytic sets @xmath12 , @xmath333 both inequalities in ( [ eq : setupp1 ] ) follow from remarks [ re : hdimub1 ] , [ re : uppbd ] and lemma [ lem : setdim ] . combining propositions [ prop : lowerbd2 ] and [ prop : setupp1 ]
yields the following theorem .
[ th : main2a ] let @xmath245 be an @xmath8-random field and let @xmath334 $ ] be a constant . if , for every @xmath241 , @xmath10 satisfies a uniform hlder condition of order @xmath335 on all compact intervals of @xmath9 and condition ( @xmath134 ) with @xmath275 , then , for all analytic sets @xmath71 , @xmath336 it is often desirable to compute @xmath40 in terms of @xmath337 .
the following is the packing dimension analog of ( [ eq : bg60 ] ) .
note that if @xmath30 , then the conclusion of corollary [ co:46 ] does not hold in general ; see talagrand and xiao ( @xcite ) . in this sense , it is the best possible result of this kind .
[ co:46 ] let @xmath245 and @xmath12 be as in theorem [ th : main2a ] .
if either @xmath338 or @xmath34 satisfies @xmath339 then @xmath340 a.s .
if @xmath341 , then ( [ eq : setprofile2 ] ) implies that for every analytic set @xmath84 , @xmath342 hence , theorem [ th : main2a ] yields @xmath343 a.s .
, as desired
. on the other hand , if an analytic set @xmath84 satisfies @xmath339 then ( [ eq : setprofile3 ] ) implies that @xmath344 .
hence , again , the conclusion follows from theorem [ th : main2a ] .
since many random fields do not have continuous sample functions and , even if they do , it is known that @xmath35 is not determined by the exponent of uniform modulus of continuity ( a typical example being linear fractional stable motion
see example [ ex : lfs ] below ) , there have been various efforts to remove the uniform hlder condition . however , except for markov processes or random fields with certain markov structure , no satisfactory method has been developed .
the main difficulty lies in deriving a sharp upper bound for @xmath35 . in the following ,
we derive an upper bound for @xmath35 under condition ( c1 ) .
this method is partially motivated by an argument in schilling ( @xcite ) for feller processes generated by pseudo - differential operators and , as far as we know , is more general than the existing methods in the literature .
[ lem : momentest ] let @xmath14 be a random field with values in @xmath7 . if condition holds for @xmath345 and @xmath126 , then , for all @xmath346 , @xmath347 and @xmath348 , @xmath349 where @xmath350 and @xmath351 is a constant independent of @xmath129 and @xmath352 .
we write @xmath353\\[-8pt ] & \le & k \int_0^\gamma u^{\gamma- 1 } \mathrm{e}^{- u } ( \gamma- u ) \min\{1 , ( h^{-h_1 } u)^{-\beta}\ } \,\mathrm{d}u , \nonumber\end{aligned}\ ] ] where the inequality follows from ( c1 ) .
it is elementary to verify that , up to a constant , the last integral is bounded by @xmath354 this proves ( [ eq : momentest ] ) .
[ prop : setdimuppbd ] let @xmath14 be a random field with values in @xmath7 .
suppose that the sample function of @xmath10 is a.s .
bounded on all compact subsets of @xmath9 .
if condition holds for @xmath345 and @xmath126 , then , for every analytic set @xmath12 that satisfies @xmath355 , @xmath356 without loss of generality , we assume that @xmath357^n$ ] . for any constant @xmath358 , there exists a sequence of balls @xmath359 such that @xmath360 for a constant @xmath361 , let @xmath362^n } \|x(t)\| \le m\}$ ] .
since the sample function of @xmath243 is almost surely bounded on @xmath363^n$ ] , we have @xmath364 note that @xmath365 and , by lemma [ lem : momentest ] , ( [ eq : covere ] ) and the fact that @xmath366 , we have @xmath367\\[-8pt ] & \le & \mathrm{e}^{2 m } k_{{4,1 } } \sum_{k=1}^\infty h_k^{\gamma } < \infty.\nonumber\end{aligned}\ ] ] it follows from ( [ eq : coverxe2 ] ) that @xmath368 almost surely on @xmath369 this implies that @xmath370 almost surely on @xmath371 .
letting @xmath372 first and then @xmath373 along the rational numbers proves ( [ eq : setdimupp1 ] ) . putting proposition [ prop : lowerbd2 ] and proposition [ prop : setdimuppbd ] together , we derive the following theorem . [
th : dimxe2 ] let @xmath14 be a random field with values in @xmath7 whose sample function is a.s
. bounded on all compact subsets of @xmath9 .
if there is a constant @xmath374 such that for every @xmath375 , @xmath10 satisfies conditions with @xmath273 and ( @xmath134 ) with @xmath376 , then for every analytic set @xmath38 that satisfies @xmath377 , @xmath378
the general results in sections [ sec : measures ] and [ sec : set ] can be applied to wide classes of gaussian or non - gaussian random fields .
since the applications to gaussian random fields can be carried out by extending xiao ( @xcite , @xcite ) , we will focus on non - gaussian random fields in this section .
if @xmath379 is a stable lvy process in @xmath7 , the hausdorff dimensions of its image sets have been well studied ; see taylor ( @xcite ) and xiao ( @xcite ) for historical accounts .
the packing dimension results similar to those in sections [ sec : measures ] and [ sec : set ] have also been obtained by khoshnevisan , schilling and xiao ( @xcite ) for lvy processes . in this subsection , we will only consider non - markov stable processes and stable random fields .
let @xmath380 be an @xmath54-stable random field in @xmath381 with the representation @xmath382 where @xmath383 is a symmetric @xmath54-stable ( s@xmath54s ) random measure on a measurable space @xmath384 with control measure @xmath385 and @xmath386 ( @xmath346 ) is a family of functions on @xmath387 satisfying @xmath388 for any integer @xmath202 and @xmath389 , the characteristic function of the joint distribution of @xmath390 is given by @xmath391 where @xmath392 ( @xmath393 ) and @xmath394 is the @xmath395-norm ( or quasi - norm if @xmath396 ) .
the class of @xmath54-stable random fields with representation ( [ eq : stablerep ] ) is broad . in particular
, if a random field @xmath380 is @xmath54-stable with @xmath397 or symmetric @xmath54-stable , and is _ separable in probability _
( i.e. , there is a countable subset @xmath398 such that for every @xmath346 , there exists a sequence @xmath399 such that @xmath400 in probability ) , then @xmath401 has a representation ( [ eq : stablerep ] ) ; see theorems 13.2.1 and 13.2.2 in samorodnitsky and taqqu ( @xcite ) . for a separable @xmath54-stable random field in @xmath381 given by ( [ eq : stablerep ] ) ,
rosinski and samorodnitsky ( @xcite ) investigated the asymptotic behavior of @xmath402^n } |x_0(t)| \ge u \}$ ] as @xmath403 ( see also samorodnitsky and taqqu ( @xcite ) ) .
the following lemma is a consequence of their result .
[ eq : stablesuptail ] let @xmath404 be a separable @xmath54-stable random field in @xmath381 given in the form ( [ eq : stablerep ] ) .
assume that @xmath401 has a.s .
bounded sample paths on @xmath363^n$ ] .
there then exists a positive and finite constant @xmath405 , depending on @xmath63 , @xmath406 and @xmath385 only , such that for all @xmath407 , @xmath408^n } |x_0(t)| \ge u \bigr\ } = k_{{5 , 1 } } .\ ] ] in the above lemma , it is crucial to assume that @xmath401 has bounded sample paths on @xmath363^n$ ] almost surely .
otherwise , ( [ eq : rosam93 ] ) may not hold , as shown by the linear fractional stable motion @xmath401 with @xmath409 ( see example [ ex : lfs ] below ) .
we define an @xmath63-stable random field @xmath14 with values in @xmath7 by @xmath410 where @xmath411 are independent copies of @xmath401 .
the following result gives the hausdorff and packing dimensions of the image measures of self - similar stable random fields .
[ th : stableimage ] let @xmath14 be a separable @xmath54-stable field with values in @xmath7 defined by ( [ def : xd ] ) , where @xmath401 is given in the form ( [ eq : stablerep ] ) .
suppose that @xmath401 is @xmath15-sssi and its sample path is a.s .
bounded on all compact subsets of @xmath9 .
then , for every finite borel measure @xmath2 on @xmath9 , @xmath412 moreover , for every analytic set @xmath12 that satisfies @xmath413 , we have @xmath414 it follows from the self - similarity and lemma [ eq : stablesuptail ] that @xmath10 satisfies condition ( c1 ) with @xmath415 and @xmath416 .
on the other hand , condition ( c2 ) with @xmath417 is satisfied because @xmath10 is @xmath15-self - similar and has stationary increments , and the @xmath63-stable variable @xmath418 has a bounded continuous density function .
therefore , both equalities in ( [ eq : stablehdim ] ) follow from theorems 3.8 and 3.12 .
finally , the last conclusion follows from theorem [ th : dimxe2 ] .
next , we consider two important types of sssi stable processes .
[ ex : lfs ] let @xmath419 and @xmath420 be given constants .
we define an @xmath63-stable process @xmath421 with values in @xmath381 by @xmath422 where @xmath423 is a symmetric @xmath54-stable random measure on @xmath381 with lebesgue measure as its control measure and where @xmath424 in the above , @xmath425 are constants with @xmath426 , @xmath427 and @xmath428 .
the @xmath63-stable process @xmath401 is then @xmath15-self - similar with stationary increments , which is called an @xmath429__-linear fractional stable motion__. if @xmath430 , then the integral in ( [ eq : repstable ] ) is understood as @xmath431)$ ] if @xmath432 and as @xmath433)$ ] if @xmath434 .
hence , @xmath401 is an @xmath63-stable lvy process .
maejima ( @xcite ) proved that if @xmath435 , then @xmath401 is a.s .
unbounded on any interval of positive length . on the other hand ,
if @xmath436 ( i.e. , @xmath437 and @xmath438 ) , then kolmogorov s continuity theorem implies that @xmath401 is a.s
. continuous . in the latter case ,
takashima ( @xcite ) further studied the local and uniform hlder continuity of @xmath401 .
his theorems 3.1 and 3.4 showed that the local hlder exponent of @xmath401 equals @xmath15 .
however , the exponent of the uniform hlder continuity can not be bigger than @xmath439 . now , let @xmath440 be the @xmath429-linear fractional stable motion with values in @xmath7 defined by ( [ def : xd ] ) .
it follows from theorem [ th : stableimage ] that if @xmath441 , then for every finite borel measure @xmath2 on @xmath254 , @xmath442 and for every analytic set @xmath443 , @xmath444 a.s .
note that the above dimension results do not depend on the uniform hlder exponent of @xmath10 .
there are several ways to define linear fractional @xmath63-stable random fields ; see kokoszka and taqqu ( @xcite ) .
for example , for @xmath445 and @xmath446 , define @xmath447 where @xmath423 is an s@xmath63s random measure on @xmath9 with the @xmath448-dimensional lebesgue measure as its control measure .
this is the stable analog of the @xmath448-parameter fractional brownian motion .
however , it follows from theorem 10.2.3 in samorodnitsky and taqqu ( @xcite ) that , whenever @xmath449 , the sample paths of @xmath450 are a.s .
unbounded on any interval in @xmath9 .
thus , the results of this paper do not apply to @xmath450 when @xmath449 .
in general , little is known about the sample path properties of @xmath450 . [
ex : lhsm ] given @xmath419 and @xmath420 , the harmonizable fractional stable field @xmath451 with values in @xmath381 is defined by @xmath452 where @xmath453 is a complex - valued , rotationally invariant @xmath63-stable random measure on @xmath9 with the @xmath448-dimensional lebesgue measure as its control measure .
it is easy to verify that the @xmath63-stable random field @xmath454 is @xmath15-self - similar with stationary increments .
it follows from theorem 10.4.2 in samorodnitsky and taqqu ( @xcite ) ( which covers the case @xmath409 ) and theorem 3 of nolan ( @xcite ) ( which covers @xmath455 ) that @xmath454 has continuous sample paths almost surely .
moreover , it can be proven that @xmath454 satisfies the following uniform hlder continuity : for any compact interval @xmath456 \subset\r^n$ ] and any @xmath241 , @xmath457 when @xmath458 , ( [ eq : harmholder ] ) is due to kno and maejima ( @xcite ) . in general , ( [ eq :
harmholder ] ) follows from the results in bierm and lacaux ( @xcite ) or xiao ( @xcite ) .
note that the hlder continuity of @xmath454 is different from that of the linear fractional stable motions .
applying theorem [ th : main2a ] to the harmonizable fractional stable motion in @xmath7 defined as in ( [ def : xd ] ) , still denoted by @xmath454 , we derive that for every analytic set @xmath459 , @xmath460 the results in this section are applicable to other self - similar stable random fields , including the telecom process ( lvy and taqqu ( @xcite ) , pipiras and taqqu ( @xcite ) ) , self - similar fields of lvy chentsov type ( samorodnitsky and taqqu ( @xcite ) , shieh ( @xcite ) ) and the stable sheet ( ehm ( @xcite ) ) .
we leave the details to interested readers .
we show that the results in sections [ sec : measures ] and [ sec : set ] can be applied to the real harmonizable fractional lvy motion ( rhflm ) introduced by benassi , cohen and istas ( @xcite ) . to recall their definition , let @xmath291 be a borel measure on @xmath461 which satisfies @xmath462 for all @xmath463 we assume that @xmath291 is rotationally invariant .
hence , if @xmath66 is the map @xmath464 , then the image measure of @xmath291 under @xmath66 can be written as @xmath465 where @xmath466 is the uniform measure on @xmath467 and @xmath468 is a borel measure on @xmath254 .
let @xmath469 be a poisson random measure on @xmath470 with mean measure @xmath471 and let @xmath472 be the compensated poisson measure . then , according to definition 2.3 in benassi , cohen and istas ( @xcite ) , a real harmonizable fractional lvy motion ( without the gaussian part ) @xmath473 with index @xmath420 is defined by @xmath474 as shown by benassi , cohen and istas ( @xcite ) , @xmath475 has stationary increments , as well as moments of all orders ; it behaves locally like fractional brownian motion , but at the large scale , it behaves like harmonizable fractional stable motion @xmath454 in ( [ eq : stableharm ] ) . because of these multiscale properties , rhflm s form a class of flexible stochastic models .
the following equation on characteristic functions of @xmath475 was given by benassi , cohen and istas ( @xcite ) : for all integers @xmath153 , all @xmath476 and all @xmath477 , @xmath478 \,\mathrm{d } \xi\,\nu(\mathrm{d}z ) \biggr),\ ] ] where @xmath479 in particular , for any @xmath120 and @xmath480 , ( [ eq : rhchf ] ) gives that @xmath481 where , for every @xmath482 , @xmath483 is defined by @xmath484 .
note that the function @xmath485 is non - negative and continuous .
moreover , up to a constant , it is the characteristic exponent of the infinitely divisible law in @xmath461 with lvy measure @xmath291 . for the proof of theorem [ th : rhflmimage ]
, we will make use of the following fact : there exists a positive constant @xmath486 such that @xmath487.\ ] ] this is verified by using the inequality @xmath488 for all @xmath489 $ ] .
[ th : rhflmimage ] let @xmath490 be a separable real harmonizable fractional lvy field in @xmath7 defined by ( [ def : xd ] ) , where @xmath475 is defined as in ( [ eq : rhflm ] ) . assume that @xmath485 satisfies the following condition : there exists a constant @xmath491 $ ] such that @xmath492 then , for every analytic set @xmath493 , @xmath494 it follows from proposition 3.3 in benassi , cohen and istas ( @xcite ) that for every @xmath241 , @xmath495 satisfies almost surely a uniform hlder condition of order @xmath496 on all compact sets of @xmath9 .
hence , the upper bounds in ( [ eq : rhflmhdim ] ) follow from proposition [ prop : setupp1 ] . in order to prove the desired lower bounds in ( [ eq : rhflmhdim ] ) , by proposition [ prop : lowerbd2 ] , it suffices to show that @xmath495 satisfies condition ( c@xmath134 ) with @xmath497 .
this is done by showing that there exists a positive function @xmath498 such that for all @xmath120 satisfying @xmath499 , we have @xmath500 this and the fourier inversion formula together imply that the density functions of @xmath501 are uniformly bounded for all @xmath502 satisfying @xmath499 . since the coordinate processes @xmath503 are independent copies of @xmath475 , it is sufficient to prove ( [ eq : g ] ) for @xmath504 .
note that , by ( [ eq : g ] ) , we can take @xmath505 for all @xmath506 .
for any @xmath507 such that @xmath508 , condition ( [ con : psi ] ) implies that @xmath509\\[-8pt ] & \ge & k |u|^\delta\int_{\|\xi\| \ge\gamma\|t - s\|^{-1 } } \psi\biggl(\frac{1 - \cos\langle t - s , \xi\rangle } { \|t - s\|^h \|\xi\|^{h+{n}/{2 } } } \biggr ) \,\mathrm{d}\xi,\nonumber\end{aligned}\ ] ] where @xmath510 is a constant whose value will be chosen later . by a change of variable @xmath511
, we see that the last integral becomes @xmath512\\[-8pt ] & & \quad \ge k \int_{\|\eta\| \ge\gamma } \frac{(1 - \cos\langle ( t - s)/\|t - s\| , \eta\rangle)^2 } { \|\eta\|^{2h+n } } \,\mathrm{d}\eta,\nonumber\end{aligned}\ ] ] where the inequality follows from ( [ eq : psi - trun ] ) , and we have used the fact that @xmath133 and taken @xmath513 large .
the last integral is a constant because the lebesgue measure is rotationally invariant .
thus , we have proven that for @xmath508 , @xmath514 therefore , when @xmath504 , ( [ eq : g ] ) holds for the function @xmath515 defined as @xmath505 if @xmath516 and @xmath517 if @xmath518 this completes the proof of theorem [ th : rhflmimage ] .
we mention that benassi , cohen and istas ( @xcite ) have introduced another interesting class of fractional lvy fields , namely , the moving average fractional lvy fields ( maflf ) . similarly to the contrast between linear fractional stable motion and harmonizable fractional stable motion ,
many properties of maflf s are different from those of rhflm s .
for example , the exponent of the uniform modulus of continuity of an maflf is strictly smaller than its local hlder exponent .
nevertheless , we believe that the arguments in this paper are applicable to maflf s .
this and some related problems will be dealt with elsewhere . given an integer @xmath519 and a constant @xmath520 , the hermite process @xmath521 of order @xmath385 is defined by @xmath522 where @xmath523 is a normalizing constant depending on @xmath385 and @xmath252 only and the integral @xmath524 is the @xmath385-tuple wiener
it integral with respect to the standard brownian motion excluding the diagonals @xmath525 , @xmath526 .
the integral ( [ eq : hermite ] ) is also well defined if @xmath527 ; the process is a fractional brownian motion for which the problem considered in this paper has been solved .
the hermite process @xmath528 is @xmath15-sssi and @xmath529 .
it is a non - gaussian process and often appears in non - central limit theorems for processes defined as integrals or partial sums of nonlinear functionals of stationary gaussian sequences with long - range dependence ; see taqqu ( @xcite , @xcite ) , dobrushin and major ( @xcite ) and major ( @xcite ) .
it follows from theorem 6.3 of taqqu ( @xcite ) that the hermite process @xmath528 has the following equivalent representation : @xmath530 where @xmath531 is a normalizing constant and @xmath532 is a centered complex gaussian random measure on @xmath381 with lebesgue measure as its control measure .
mori and oodaira ( @xcite ) studied the functional laws of the iterated logarithms for the hermite process @xmath528 .
lemma [ mwtail ] follows from lemma 6.3 in mori and oodaira ( @xcite ) .
[ mwtail ] there exist positive constants @xmath533 and @xmath534 , depending on @xmath385 only , such that @xmath535 } |y^{m , \kappa}(t)| \ge u \ } \le \exp(-k_{{5 , 6 } } u^{2/m } ) $ ] for all @xmath536 using lemma [ mwtail ] , one can derive easily a uniform modulus of continuity for @xmath528 .
[ lem : modcon ] there exists a finite constant @xmath537 such that for all constants @xmath538 , @xmath539 where @xmath540 . for every @xmath432 and @xmath347 , the self - similarity of @xmath528 and lemma [ mwtail ] together imply that @xmath541 hence , @xmath542 satisfies the conditions of lemmas 2.1 and 2.2 in cski and csrg ( @xcite ) with @xmath543 and @xmath544 .
consequently , ( [ eq : uniform ] ) follows directly from theorem 3.1 in cski and csrg ( @xcite ) . the case @xmath545 has recently received considerable attention .
the process @xmath546 is called the _ rosenblatt process _ by taqqu ( @xcite ) ( or _ fractional rosenblatt motion _ by pipiras ( @xcite ) ) .
its self - similarity index is given by @xmath547 .
this non - gaussian process in many ways resembles fractional brownian motion .
for example , since @xmath548 , fractional noise of @xmath546 exhibits long - range dependence .
besides its connections to non - central limit theorems , the rosenblatt process also appears in limit theorems for some quadratic forms of random variables with long - range dependence .
albin ( @xcite , @xcite ) has discussed distributional properties and the extreme value theory of @xmath546 . in particular , albin ( @xcite ) , section 16 , obtained sharp asymptotics on the tail probability of @xmath549}y^{2 , \kappa}(t)$ ] . pipiras ( @xcite ) established a wavelet - type expansion for the rosenblatt process .
tudor ( @xcite ) has recently developed a stochastic calculus for @xmath546 based on both pathwise type calculus and malliavin calculus .
we now consider the rosenblatt process @xmath550 with values in @xmath7 by letting its component processes be independent copies of @xmath546 .
the following result determines the hausdorff and packing dimensions of the image sets of @xmath550 .
let @xmath551 be a rosenblatt process in @xmath7 as defined above .
then , for every analytic set @xmath443 , we have @xmath552 by lemma [ lem : modcon ] , for any @xmath241 , @xmath550 satisfies a uniform hlder condition of order @xmath335 ( where @xmath553 ) on all compact intervals in @xmath254 . on the other hand , it is known that the random variable @xmath554 has a bounded and continuous density ( see davydov ( @xcite ) or albin ( @xcite ) ) .
thus , @xmath550 also satisfies condition ( c2 ) with @xmath555 .
therefore , the two equalities in ( [ eq : setfrm1 ] ) follow from theorem [ th : main2a ] .
the authors thank professor davar khoshnevisan for stimulating discussions on packing dimension profiles , professors patrik albin , david nualart and frederi viens for very helpful suggestions on the rosenblatt process and the anonymous referee for comments and suggestions which have led to significant improvement of the manuscript .
research of n .-
shieh was partially supported by taiwan nsc grant 962115m002005my3 .
research of y. xiao was partially supported by the nsf grant dms-0706728 and the national natural science foundation of china ( no . 70871050 ) .
kno , n. ( 1986 ) .
hausdorff dimension of sample paths for self - similar processes . in _ dependence in probability and statistics ( oberwolfach , 1985)_. _ progr . probab .
statist . _
* 11 * 109117 .
boston : birkhuser boston .
shieh , n .- r .
sample functions of lvy chentsov random fields . in _ probability theory and mathematical statistics .
proceedings of the 7th japan russia symposium _ ( s. watanabe _
et al . _ ,
eds . ) 450459 .
river edge , nj : world scientific .
xiao , y. ( 2004 ) .
random fractals and markov processes . in _ fractal geometry and applications : a jubilee of benoit mandelbrot _ ( m.l .
lapidus and m. van frankenhuijsen , eds . ) 261338 .
providence , ri : amer .
xiao , y. ( 2007 ) .
strong local nondeterminism and the sample path properties of gaussian random fields . in _ asymptotic theory in probability and statistics with applications _
lai , q. shao and l. qian , eds . ) 136176 .
beijing : higher education press . | let @xmath0 be a random field with values in @xmath1 . for any finite borel measure @xmath2 and analytic set @xmath3 , the hausdorff and packing dimensions of the image measure @xmath4 and image set @xmath5
are determined under certain mild conditions .
these results are applicable to gaussian random fields , self - similar stable random fields with stationary increments , real harmonizable fractional lvy fields and the rosenblatt process . |
two way relaying communications have recently attracted considerable attentions due to their various applications . in this communication scenario ,
two users attempt to communicate with each other with the help of a relay . to this end
, physical layer network coding ( plnc ) [ 1 ] along with the conventional df or af relaying strategy has been commonly considered [ 2 - 4 ] .
it has been shown that plnc can achieve within 1/2 bit of the capacity of a gaussian twrc ( two way relay channel ) and this is asymptotically optimal at high snrs [ 5 - 6 ] . in [ 2 - 3 ] , based on df startegy , two - step and three - step two - way relaying schemes are proposed . in the two - step scheme , in the first step , both users simultaneously transmit their messages , and the relay recovers both messages in turn , using a linear receiver structure like successive interference cancellation ( sic ) [ 7 ] . in the second step , the relay sends a combination of the recovered messages to the users .
the problem with the scheme proposed is that , when recovering one of the messages , the other message is considered as noise , which results in a performance loss . as a solution , an optimum ml decoder can be utilized at the relay at the expense of a very high complexity [ 3 ] .
the three - step df two - way relaying proposed in [ 2 ] requires three time slots that results in a throughput reduction . as an alternative
, the relay can exploit af strategy to simply amplify the received signal from the users , and then forward it to the users .
due to noise amplification in the relay , this scheme shows a poor performance [ 2 ] .
the novel relaying strategy known as compute - and - forward ( cmf ) , proposed by nazer and gastpar [ 8 ] , is proved to be efficient for multiuser communication scenarios .
the cmf strategy can exploit the interference to achieve a higher throughput .
cmf strategy is also known as a reliable physical layer network coding [ 9 ] . in cmf strategy ,
all sources transmit simultaneously .
each relay , based on its received signal ( a noisy and channel weighted combination of the users codewords ) and its knowledge of the channel coefficients , decodes an equation , which is an integer - linear combination of the users transmitted messages .
the integer coefficients of the equation are presented by a vector called an equation coefficient vector ( ecv ) .
the relay has to find the ecv with the highest possible rate .
the relay then transmits the decoded equation to the destination .
the destination recovers the desired messages by receiving sufficient number of decoded equations from the relays . for codewords , lattice codes
are commonly utilized , which can achieve the capacity of additive white gaussian noise ( awgn ) channels [ 10 ] . while cmf strategy has been considered in different scenarios in the literature , such as multi - antenna systems [ 11 ] , cooperative distributed antenna systems
[ 12 ] , multi - access relay channels [ 13 ] , generalized multi - way relay channels [ 14 ] , two transmitter multi - relay systems [ 15 ] , and finally multi - source multi - relay network [ 16 ] ; however , to our best knowledge the application of cmf in two - way relaying hasnt been considered so far , just from information theory aspect in [ 17 ] . in this paper
, we propose a new practical framework for two - way relaying based on cmf strategy , in which we use a linear receiver and a general lattice encoding previously proposed by nazer [ 8 ] .
we consider this framework for two cases .
first , we investigate the relays without the capability of sending any feedback to the users .
we call the corresponding proposed scheme as max compute - and - forward ( m - cmf )
. then we consider the relays that have feedback capability ; the related scheme is called as aligned compute - and - forward ( a - cmf ) . for the latter case , the power can be efficiently allocated to the users in a way to increase the computation rate through aligning the scaled channels to the integer coefficients , under a maximum power constraint for each user .
the proposed schemes , in contrast to df and af based schemes , can handle both the interference and noise , and thus enhance the network throughput considerably . to achieve a higher order of diversity , multiple relays along with a simple relay selection technique are employed .
we consider a block rayleigh fading channels between the users and the relays .
the channels have phase variations in addition to the amplitude variations . while the proposed schemes have been considered and work quite well for general complex gaussian channels with variation in both phase and amplitude , for the sake of simplicity and tractability of the analytical performance evaluation , in this paper , we focus on addressing the amplitude variation of the channels and do not consider the carrier phase offset for the analytical performance analysis . in the other words , we assume that the phase offset between two received users signals has been compensated at the best relay .
this makes the channels realized by the best relay be real - valued rayleigh channels .
this assumption was commonly used in the literature when considering the performance analysis of cmf based strategies , for instance please see [ 15 ] and [ 18 ] .
however for simulation evaluations , we consider general rayliegh fading channel ( complex gaussian coefficient ) with both phase and amplitude variations .
we analitically derive , an upper bound on the computation rate of each relay . then using this bound
, we derive a bound on the average sum rate and the outage probability of the proposed scheme .
based on the bound obtained for the outage probability , we derive the system diversity order .
numerical results show that a - cmf reaches the bound in all snrs , m - cmf is tight on the bound only in low snrs ; in the other words , a - cmf improves the compute rate at high snr . to have a fair comparison with m - cmf and the other schemes , a - cmf under a power constraint on the total powers of both users rather than on the power of each user
is also considered . as expected , the numerical results verifies that a - cmf under the new constraint performs better . for the latter case , a - cmf and m - cmf
are in fact compared under the same total system power .
we also evaluate the symbol rate of the system , and compare the results with those of the conventional schemes , which indicates the substantial superiority of the proposed schemes .
the remainder of this paper is organized as follows . in section
ii , the system model is described .
section iii presents the proposed method and the performance analysis is given in section iv .
numerical results are presented in section v. finally , section vi concludes the paper .
we consider a two way relay channel with two users and @xmath0 relays , as shown in fig .
1 . user @xmath1 , exploits a lattice encoder with power constraint @xmath2 to project its message @xmath3 to a length - n complex - valued codeword @xmath4 such that @xmath5 .
@xmath6 is considered as the maximum power constraint of the user @xmath7 ( @xmath8 ) .
we assume that each relay has a power constraint equal to @xmath9 that is more than or equal to @xmath10 .
the channel coefficient from user @xmath1 , to relay @xmath11 denoted by @xmath12 and assumed to be equal to the reverse link coefficient @xmath13 , follows a real - valued rayleigh distribution with variance @xmath14 , here we only focus on the the real channel . for the numerical part , we also consider more general complex gaussian channel . ] .
all channel coefficients for different @xmath15 and @xmath7 are assumed to be independent .
we assume a block fading such that the coefficients remain constant during total transmission time slots required for the message exchanges .
there is no direct link between two users .
the noise received at the @xmath15th relay and at the @xmath7th user , denoted by @xmath16 and @xmath17 respectively , are i.i.d . according to the zero - mean gaussian distribution with variance 1 .
in our proposed cmf based scheme , multiple access broadcast ( mabc ) protocol for two way transmission is used [ 19 ] . in the first time slot ( named multiple access phase ) , two users simultaneously transmit their codewords to the relays .
each relay receives a noisy linear combination of users codewords , and using the cmf strategy [ 8 ] , the relay decodes an equation , i.e. , an integer linear combination of users messages ( see subsection a ) . then in the second time slot ( broadcast phase ) ,
the best relay is selected ( see subsection c ) to transmit its decoded equation to the users .
finally , by receiving the equation , each user recovers the other user s message ( see subsection b ) .
the received signal in each relay @xmath15 in the multiple access phase can be written as @xmath18 + for each relay @xmath15 , vector @xmath19 and matrix @xmath20 are defined as @xmath21\end{aligned}\ ] ] and , @xmath22 based on cmf strategy [ 8 ] , each relay has to decode the equation coefficient vector ( ecv ) , i.e. @xmath23^t } \in { z^2}$ ] , such a way to maximize its computation rate , i.e. , the rate of recovering the equation , @xmath24 where @xmath25 .
the solution of ( 4 ) can be expressed [ 20 ] as scaling the received signal by the following factor @xmath26 that is the recovered equation is expressed as @xmath27 where @xmath28 shows the lattice quantizer function .
the computation rate @xmath29 of the equation @xmath30 is given by [ 8 ] @xmath31 to enable each user to recover the other user s message , the selected equation must not contain zero components . hence , @xmath32^t}$ ] and @xmath33^t}$ ] , as a solution of ( 4 ) , are not desirable
. please note that , from [ 15 , lemma 3 ] , the only vectors that have zero elements and can be the solution of ( 4 ) are @xmath34 and @xmath35 .
if the vector @xmath36 computed from ( 4 ) has nonzero components , it will be chosen as the coefficient vector of the selected equation .
otherwise , we do as follows . from ( 6 ) , the relay first recovers the @xmath37th message @xmath38 then , by removing the effect of the codeword @xmath39 from the received vector @xmath40 , the other user codeword , @xmath41 , is also recovered .
finally , the relay constructs the equation @xmath42 with nonzero coefficients for the transmission . + a necessary condition for the occurrence of @xmath43 , as the solution of ( 4 )
is @xmath44 from ( 7 ) , the computation rate of @xmath43 , can be computed as @xmath45 by removing @xmath39 from @xmath40 , the rate of recovering the other message in the relay @xmath15 can be easily found as @xmath46 hence , the rate of constructing the equation @xmath47 in each relay @xmath15 is given by @xmath48 for users power allocation , we consider two cases based on whether or not the relay is able to send some information feedback to the users , as follows : in the case that the relays havent feedback capability , the user @xmath37 transmits with the maximum possible power @xmath49 ( @xmath50 ) .
we call this case as max compute and forward ( m - cmf ) . in the another case
when there is feedback capability for the relays , we propose aligned compute and forward ( a - cmf ) scheme , as follows . in this scheme ,
a new power adaptation algorithm is exploited , in which the users adjust their transmitted powers in order to minimize the quantization noise @xmath51 between the scaled received signal at the best relay @xmath52 and its computed equation and hereby to increase the computation rate .
we define a power adaptation vector ( pav ) @xmath53 as @xmath54\end{aligned}\ ] ] based on the computation rate in ( 7 ) , in each relay @xmath15 , for the case @xmath55 or @xmath56 , the quantization noise @xmath57 can be written as @xmath58 where @xmath59 and @xmath60 . the ecv @xmath61 , pav @xmath53 , and scaling factor @xmath62 are the unknown parameters and should be optimally selected based on the maximization of the computation rate of the relay .
we consider an alternative manner to solve this optimization problem . in the first step , assuming the pav is known , the ecv and the scaling factor are computed from ( 4 ) and ( 5 ) . in the second step , similarly , by assuming ecv and scaling factor are known based on using the values computed at the previous step , the pav is calculated as follows : using kkt conditions by taking the derivation of ( 14 ) and putting the result equal to zero , we have @xmath63 where @xmath64 is the kkt coefficient related to the user @xmath37 . ( 15 ) leads to : @xmath65 hence , we have @xmath66 from kkt conditions , when the answer is in the feasible region or for @xmath67 , we have @xmath68 and when the answer is on the constraint , we have @xmath69 @xmath70 . in this case , we can easily derive that @xmath71 hence , we have @xmath72 the two described steps are iterated successively until the pav converges .
the above procedures are summarized in algorithm 1 .
the parameter @xmath73 used in the algorithm denotes the convergence telorance . in some applications , the total system power is more imprtant than the individual user power .
in addition , to have a fair comparison of the proposed scheme ( a - cmf ) with the other ones , they should be compared under the same total system power . as a result ,
in the following , we consider the a - cmf scheme under the power constraint on the total powers of the users .
in other words , we minimize ( 14 ) with the new constraint @xmath74 .
it is clear that in this case , the feasible region is larger and hence the performance should be better .
similar to the previous case ( separate power constraint ) , the kkt conditions results to @xmath75 where @xmath76 is the kkt coefficient .
this leads to @xmath77 from kkt conditions , when the constraint holds with equality , @xmath76 , obtained by solving the following equation , must be nonnegative .
@xmath78 this equation has not a straightforward answer and can be solved by bisection method [ 21 ] .
+ on the other hand , when the answer is inside the feasible region , we should have @xmath79 and the pav is given by @xmath80 when @xmath36 is either @xmath34 or @xmath35 , according to the computation rate in ( 12 ) , the optimum value of power for each user is the maximum possible value , i.e. we have @xmath81 and @xmath82 .
this algorithm is implemented in each relay .
then the relay with the highest computation rate ( given in ( 7 ) and ( 12 ) ) is selected as the best relay , as will be described in the following in section iii.c .
the pav of the best relay is sent to the users through a feedback channel in order the users to adjust their transmission powers .
( 1,0)470 * algorithm 1 : calculating pav for relay i * ( 1,0)470 initialize @xmath83 @xmath70 and @xmath84 1.ecv search : update @xmath61 and @xmath85 from ( 4 ) and ( 5 ) for fixed @xmath86 @xmath87 2.update @xmath88 @xmath70 for fixed @xmath61 and @xmath85 as follows for @xmath67 @xmath89 @xmath90 and for @xmath91 @xmath89 @xmath92 @xmath93 @xmath94 ( 1,0)470 the selected relay @xmath52 sends its recovered equation @xmath95 , with the rate @xmath96 , to both users .
each user by receiving this equation and having its own message can recover the other user s message .
the received signal by user @xmath15 is written as @xmath97 since the channel from the relay @xmath52 to the user @xmath98 , is a simple point to point channel , the achievable rate for the transmission of the equation @xmath95 is @xmath99 + the rate in ( 27 ) is achievable using cmf strategy [ 8 ] .
like as ( 2 ) and ( 3 ) , in one user computation case , the scaling factor @xmath100 and the recovered equation for user @xmath15 are given by @xmath101 @xmath102 the rate of recovering both messages in both users is easily given by @xmath103 from ( 30 ) , to maximize the rate of recovering both messages in both users , denoted by @xmath104 , the relay @xmath52 must be selected as @xmath105 the computation rate of the equation @xmath30 in each relay @xmath15 , i.e. @xmath29 , is upper bounded as @xmath106 ( note : for the sake of simplicity , the subscript index @xmath15 of the rates is removed . ) first we consider the case that the ecv @xmath107 or @xmath35 : + from ( 7 ) , we need to show that @xmath108 it is sufficient to show that @xmath109 which can be written as @xmath110 where @xmath111 $ ] and @xmath112 $ ] .
we define the variable @xmath113 as @xmath114 which with some straightforward simplifications and by defining @xmath115 and @xmath116 , results in @xmath117 without loss of generality , we assume that @xmath118 .
then @xmath119 for the right side of the inequality , we can minimize @xmath113 with respect to @xmath120 as follows @xmath121 which leads to @xmath122 by substituting ( 39 ) in ( 36 ) and some straightforward simplifications , we obtain @xmath123 thus , from ( 37 ) we get @xmath124 since we have @xmath125 the right - hand side of ( 35 ) can be written as @xmath126 this proves ( 35 ) , and then ( 33 ) .
now , we consider the case that @xmath127 or @xmath35 .
it is clear that @xmath128 hence , the theorem is proved .
the bound derived is tight , specially at low snrs .
please note that at low snrs , the ecv as a solution of ( 4 ) is usually either @xmath34 or @xmath35 [ 15 ] . in this case , from ( 36 ) @xmath113 can be written as @xmath129 which certainly is lower than @xmath130 , ( please note that we have assumed @xmath131 as the solution ) . for @xmath132 lower than the bound , i.e. @xmath133 , we have @xmath134 that leads to @xmath135 and @xmath136 .
hence , at low snrs , the rate is very close to the bound given in ( 32 ) with a high probability . according to this theorem
, we have @xmath137 using ( 47 ) and with the assumation of @xmath138 we can easily rewrite ( 31 ) as @xmath139 now , the best relay , after the multiple access phase , is selected based on ( 48 ) , using the approach similar to [ 22 ] . that is the relay @xmath140 ,
sets a timer with the value @xmath141 proportional to the inverse of its corresponding rate , i.e. @xmath29 . the first relay that its timer reaches zero ( which has the highest rate ) broadcasts a flag , to inform other relays , and is selected as the best relay for the broadcast phase .
from ( 30 ) , the outage probability of the proposed scheme can be computed as @xmath142 where @xmath143 denotes the target rate . according to the theorem 1 and with
the assumation of @xmath138 , we have @xmath144 moreover , from theorem 1 , a lower bound for the outage probability is derived as follows @xmath145 with the defination of @xmath146 and since @xmath100s are independent exponential random variables with the following cdf : @xmath147 we can find the cdf of @xmath148 as @xmath149 hence , the outage probability lower bound can be easily computed as @xmath150 from taylor series expansion , in high snrs , when @xmath151 , we can approximate ( 54 ) as @xmath152 hence , the acheivable diversity order from the outage bound with the definition @xmath153 [ 23 ] is equal to m , i.e. the number of relays . according to the theorem1 and with the assumation of @xmath138 , an upperbound on the sum rate conditioned on each channel realizations
can be derived as @xmath154 the unconditional sum rate can be computed by taking the expectation of ( 56 ) as @xmath155 where @xmath156 is the pdf of @xmath157 , which can be easily obtained from its cdf given in ( 52)-(53 ) . with some straightforward simplifications , leads to @xmath158 where @xmath159 and @xmath160 is the upper incomplete gamma function defined in [ 24 ] .
for numerical evaluation , target rate @xmath161 is considered .
the rayleigh channel parameters equal to @xmath162 , are assumed .
the parameter @xmath73 in algorithm 1 is setteled as @xmath163 . in fig .
2 , the outage probability of the proposed schemes along with the derived lower bound given in ( 54 ) , versus snr , is plotted for @xmath164 relays and for equal maximum transmit powers for the users and the relay . as observed , for both m - cmf and a - cmf schemes ,
the derived lower bound is quite tight especially at high snrs .
moreover , as expected , by the increase of the number of relays , the outage performance as well as the diversity order improve significantly . it is observed that the proposed schemes provides a diversity order of @xmath0 , i.e. , the number of relays employed . in fig .
3 , the average sum rates of the proposed schemes along with the derived upper bound in ( 58 ) are plotted for @xmath165 and for equal maximum powers for the users . as observed , the m - cmf reaches the bound only in low snrs , while the a - cmf approaches the bound in all snr values . in other words , a - cmf outperforms the m - cmf in high snr at the cost of using feedback transmission .
4 compares the symbol error rate ( ser ) of the proposed schemes with the ones introduced in [ 2 ] , including af and two - step df , for @xmath166 with bpsk modulation and for equal maximum transmit powers for the users .
our proposed scheme indiciates significantly better performance about 6db in ser equal to 0.02 .
please note that the ser has been evaluated by simulation , as it is not easy at all to analytically derive the ser when using the cmf based strategy , due to an integer optimization problem being solved numerically within this strategy .
5 compares the outage probability of the proposed schemes with the conventional strategies and also three - step df [ 2 ] , for @xmath167 and for equal maximum transmit powers for the users .
the same relay selection strategy is used for all schemes .
although all of the methods provide the same order of diversity , our proposed schemes demonstrates a better performance about 2db in high snr values .
6 compares the average sum rate of the proposed schemes with the conventional strategies , for @xmath168 and for equal transmit powers .
as it is observed , our proposed schemes perform significantly better than the conventional strategies in all snrs .
for example , in sum rate 4 , a - cmf has 4db and m - cmf has 2db improvement in comparison with the best conventional relaying scheme .
7 compares the average sum rate of the a - cmf scheme under two different power constraints , one in each user power and the other on the total power , for @xmath169 relays . for the first case ,
maximum transmission power of each user is considered to be equal to @xmath170 , i.e. @xmath171 , while for the second case , the maximum total transmission powers of both users is considered to be equal to @xmath172 . as it is observed from this figure , in the latter case
, the system has a better performance .
the reason is that in the second case , the feasible region of the optimization in ( 14 ) is larger than the one of the first case , that results in a higher rate .
when comparing with the other scheme , the latter constraint is more reasonable , as different schemes should be compared under the same total transmission powers .
since our results above indicate that the a - cmf under the maximum power constraint on each user transmission performs better than the other schemes , specially in term of the average sum rate , we did nt bring the comparison results with the other schemes , when the a - cmf is designed under the constraint on the total transmission power . in fig .
8 , we evalute the perfromance of proposed schemes along with the conventional strategies for @xmath167 when all link are modeled as complex gaussian channels with variance one .
the figure shows that the outage probability of the peroposed schemes , specially a - cmf , are better than those of the conventional strategies . in fig .
9 , the performance of proposed scheme has been evaluated when the channels variances are not identical .
this fig . shows the outage probability of proposed schemes and two - step df versus snr for different values of @xmath173 , where @xmath174 .
delta in fact indicates the difference between the two users channel variances . in this fig .
, we have @xmath167 . for a fair camparison , the sum of the two channels variances is set equal to two , i.e. @xmath175 . as can be observed , the lower delta makes better perfromance , however diversity order does not change with delta . from this fig . , the perfromance of the proposed schemes are better than two - step df , which shows the best performance among the conventional schemes ( please see fig .
as expected , the amount of the improvement decreases by the increase of the delta .
for example , in outage @xmath176 , while at delta equal to 0.5 , the proposed schemes have 1.8db better perfromance than the two - step df , the improvement is 1.4db at delta equal to 1 .
in this paper , based on cmf strategy , a novel two - way relaying scheme , for two cases of relays with and without capability of feedback transmission , is proposed that improves the network throughput significantly .
furthermore , a relay selection scheme is exploited to achieve a higher order of diversity through employing multiple relays . by theoretical analysis , an upper bound on the computation rate of each relay
is derived and based on that , a tight lower bound on the outage probability and an upper bound on the average sum rate of the system are presented .
our numerical results showed that the proposed scheme , in both cases of with and without using feedback , performs significantly better than the af and df strategies in terms of the outage probability , average sum rate , and the symbol error rate , and also provides a diversity order equal to the number of relays employed . 1 zhang , s. , liew , s.c . ,
lam . p.p . : hot topic : physical - layer network coding. mobicom 06 .
proc . of international conference on mobile computing and networking , 2006 ,
358 - 365 popovski , p. , yomo , h. : physical network coding in two - way wireless relay channels. icc 07 .
ieee international conference on communications , 2007 , pp .
707 - 712 zhou , q.f . ,
yonghui li , lau , f.c.m . ,
vucetic , b. : decode - and - forward two - way relaying with network coding and opportunistic relay selection , _ ieee trans .
commun . _ , 2010 , @xmath177 , ( 11 ) , pp .
3070 - 3076 tao cui , tracey ho , kliewer , j. : memoryless relay strategies for two - way relay channels , _ ieee trans . on commun . _ , 2009 , @xmath178 , ( 10 ) ,
3132 - 3143 wooseok nam , sae - young chung , lee , yong h. : capacity of the gaussian two - way relay channel to within 1/2 bit , _ ieee trans .
inf . theory _ , 2010 , @xmath179 , ( 11 ) ,
5488 - 5494 wilson , m.p . ,
narayanan , k. , pfister , h.d . ,
sprintson , a. : joint physical layer coding and network coding for bidirectional relaying , _ ieee trans .
inf . theory _
, 2010 , @xmath179 , ( 11 ) , pp .
5641 - 5654 varanasi , m.k . ,
guess , t. : optimum decision feedback multiuser equalization with successive decoding achieves the total capacity of the gaussian multiple - access channel. proceedings of the 31st asilomar conference on signals , systems and computers , 1997 , pp .
1405 - 1409 nazer , b. , gastpar , m. : compute - and - forward : harnessing interference through structured codes , _ ieee trans .
inf . theory _ , 2011 , @xmath178 , ( 10 ) ,
6463 - 6486 nazer , b. , gastpar , m. : reliable physical layer network coding , _ proceedings of the ieee _
, 2011 , @xmath180 , ( 3 ) , pp .
438 - 460 erez , u. , zamir , r. : achieving 1/2 log ( 1+snr ) on the awgn channel with lattice encoding and decoding , _ ieee trans .
inf . theory _ , 2004 , @xmath181 , ( 10 ) ,
2293 - 2314 jiening zhan , nazer , b. , erez , u. , gastpar , m. : integer - forcing linear receivers. ieee international symposium on information theory proceedings ( isit ) , 2010 , pp .
1022 - 1026 song - nam hong , caire , g. : compute - and - forward strategies for cooperative distributed antenna systems , _ ieee trans .
inf . theory _ , 2013 , @xmath182 , ( 9 ) , pp .
5227 - 5243 el soussi , m. , zaidi , a. , vandendorpe , l. : compute - and - forward on a multiaccess relay channel : coding and symmetric - rate optimization , _ ieee trans . on wireless commun .
_ , 2014 , @xmath183 , ( 4 ) , pp . 1932 - 1947
gengkun wang , wei xiang , jinhong yuan : outage performance for compute - and - forward in generalized multi - way relay channels , _ ieee commun .
letters _ , 2012 , @xmath184 , ( 12 ) , pp .
2099 - 2102 hejazi , m. , nasiri - kenari , m. : simplified compute - and - forward and its performance analysis , _ iet commun .
_ , 2013 , @xmath185 , ( 18 ) , pp .
2054 - 2063 chen , z. , fan , p. , letaief , k.b .
: compute - and - forward : optimization over multi - source - multi - relay networks , _ ieee trans . on veh
. tech _ , 2014 , @xmath186 , ( 99 ) , pp
. 1 hern , b. , narayanan , k. : an analysis of the joint compute - and - forward decoder for the binary - input two - way relay channel. 51st annual allerton conference on communication , control , and computing ( allerton ) , 2013 , pp .
1314 - 1320 tao yang , collings , i.b . :
asymptotically optimal error - rate performance of linear physical - layer network coding in rayleigh fading two - way relay channels , _ ieee commun .
letters _ , 2012 , @xmath184 , ( 7 ) , pp .
1068 - 1071 sang joon kim , mitran , p. , tarokh , v. : performance bounds for bi - directional coded cooperation protocols , _ ieee trans .
, 2008 , @xmath187 , ( 11 ) , pp .
5235 - 5241 lili wei , wen chen : efficient compute - and - forward network codes search for two - way relay channel , _ ieee commun . letter _ , 2012 , @xmath184 , ( 8) , pp .
1204 - 1207 boyd , s. , vandenberghe , l. : convex optimization ( cambridge university press , 2004 ) bletsas , a. , khisti , a. , reed , d.p . ,
lippman , a : a simple cooperative diversity method based on network path selection , _ ieee j. select .
areas commun .
_ , 2006 , @xmath188 , ( 3 ) , pp .
659 - 672 jafarkhani , h. : space - time coding theory and practice ( cambridge university press , 2005 ) alouini , m .- s . ,
goldsmith , a.j .
: capacity of rayleigh fading channels under different adaptive transmission and diversity - combining techniques , _ ieee trans . veh .
technol . _ , 1999 , @xmath189 , ( 4 ) , pp .
1165 - 1181 | * in this paper , a new two - way relaying scheme based on compute - and - forward ( cmf ) framework and relay selection strategies is proposed , which provides a higher throughput than the conventional two - way relaying schemes .
two cases of relays with or without feedback transmission capability are considered .
an upper bound on the computation rate of each relay is derived , and based on that , a lower bound on the outage probability of the system is presented assuming block rayleigh fading channels .
numerical results show that while the average sum rate of the system without feedback , named as max compute - and - forward ( m - cmf ) , reaches the derived upper bound only in low snrs , that of the system with feedback , named as aligned compute - and - forward ( a - cmf ) reaches the bound in all snrs
. however , both schemes approach the derived lower bound on the outage probability in all snrs . for the a - cmf , another power assignment based on applying the constraint on the total powers of both users rather than on the power of each separately ,
is introduced .
the result shows that the a - cmf performs better under the new constraint .
moreover , the numerical results show that the outage performance , average sum rate , and symbol error rate of the proposed schemes are significantly better than those of two - step and three - step decode - and - forward ( df ) and amplify - and - forward ( af ) strategies for the examples considered .
* * index terms- compute and forward , max compute - and - forward , aligned compute - and - forward , feedback , two - way relaying , relay selection , outage probability , average sum rate , symbol error rate . * |
identifying complete and reliable samples of agn has become a necessity for extragalactic surveys , whether the goal be the selection of agn candidates or the removal of agn `` contaminants '' .
only when armed with complete samples of agn will we be able to determine the role of obscured accretion in the build - up of the present day black hole mass function , or accurately characterize the star - formation history of the universe .
complete agn samples are also required to test proposed evolutionary theories in which black hole formation and star - formation are intimately linked by merger and feedback processes ( e.g. hopkins et al .
2006 ) , ultimately producing the correlation between black hole mass and bulge velocity dispersion ( ferrarese & merritt 2000 ; gebhardt et al .
unfortunately , the varied luminosities , accretion rates , orientations , and intrinsic obscurations of agn prevent any one selection technique from reliably identifying all of them .
for instance , while current uv , optical , and x - ray surveys are capable of detecting unobscured agn , they miss many of the obscured agn and nearly all of the compton - thick agn thought to dominate agn number counts at both low and high redshift ( e.g. gilli et al .
2007 ; daddi et al .
2007a , b ) .
likewise , only 10 - 15% of agn are radio - loud , making radio surveys relatively incomplete . the multiband imaging photometer ( mips ; rieke et al .
2004 ) and infrared array camera ( irac ; fazio et al .
2004 ) instruments aboard _
spitzer _ have provided sensitive surveys in multiple mid - ir bands . infrared selection with mips and irac data
is being used widely to select agn candidates independently of their optical and/or x - ray properties .
in addition to identifying agn in fields with little or no x - ray data , infrared selection criteria are capable of identifying heavily obscured agn missed in even the deepest x - ray fields ( e.g. donley et al .
2007 ) . as such
, ir selection has the potential to complement traditional agn selection methods and to yield a more complete census of agn activity . in this paper , we critically review the following infrared selection criteria : ( 1 ) irac color - color selection , ( 2 ) irac power - law selection , and ( 3 ) ir - excess selection .
the first selection method employs color cuts in two representations of irac 4-band mid - infrared ( mir ) color - color space ( lacy et al .
2004 ; stern et al . 2005 ) , the second identifies agn whose irac seds are well - fit by a power - law ( alonso - herrero et al .
2006 , donley et al .
2007 ) , and the third selects red galaxies with large infrared to uv / optical flux ratios ( daddi et al .
2007a , dey et al .
2008 , fiore et al .
2008 , polletta et al .
the first two criteria are based on the same principle : the hot dust near an agn s central engine reprocesses absorbed uv , optical , and x - ray emission into short - wavelength mir emission , filling in the gap between the stellar emission that peaks near 1.5 and the long - wavelength dust emission features that dominate the seds of star - forming galaxies .
the color - color and power - law selection criteria , however , differ in the range of mid - ir characteristics they include as possible agn indicators .
the third selection method identifies sources in which heavy obscuration with reemission in the infrared diminishes the optical emission and/or enhances the infrared emission .
this paper utilizes improved spectral templates with a sample of infrared color - selected , power - law galaxies ( plgs ) , and ir - excess galaxies in the ultra - deep goods - s field to test the reliability and completeness of these selection techniques over a wide range of sample properties . from this analysis
, we then quantify the contribution of these approaches plus _ spitzer _ identification of radio - intermediate and radio - loud agn to the x - ray selected agn population .
the paper is organized as follows . in 2 , we describe the selection of the color - selected , plg , and ir - excess samples .
the construction of high - reliability photometric redshifts is described in 3 , as are the overall redshift properties of the sample . in 4 , we briefly discuss the x - ray properties of our mips - selected sample .
the infrared color selection criteria of lacy et al .
( 2004 ) and stern et al .
( 2005 ) are discussed in 5 , where we compare and contrast the two selection criteria , investigate the behavior in color space of the star - forming templates , determine the redshift and flux dependencies of the color selection techniques , and investigate the properties of the most secure color - selected agn candidates . in 6 ,
we discuss the plg selection criteria , and in 7 we investigate the ir - excess sources .
finally , in 8 , we discuss the overall statistics of agn revealed by ir - based methods compared with x - ray selected samples .
a summary is given in 9 . throughout the paper ,
we assume the following cosmology : ( @xmath8,@xmath9)=(0.3 , 0.7 , 72 km s@xmath10 mpc@xmath10 ) .
we take as our initial goods - s sample all mips sources detected at 24 to a flux density of @xmath11 @xmath1jy . at this flux limit
, 99% of the mips sample are detected to @xmath12 .
there are several advantages to choosing a flux - limited mips sample .
first , agn ( and lirgs / ulirgs ) tend to be bright at 24 ( e.g. rigby et al . 2004 ) . selecting only those galaxies with @xmath11
@xmath1jy therefore retains all but the faintest agn while excluding 50 - 60% of irac - selected ir - normal star - forming galaxies at all flux densities .
second , a mips flux - limited sample is not subject to the complicated slope - dependent selection bias present in irac selected samples due to the significant variation in the sensitivity of the 4 irac bands ( see donley et al .
third , this selection gives a complete and well - defined sample of objects with extreme red @xmath13 $ ] colors .
the mips 24 catalog of the goods legacy team ( dickinson et al .
, in prep . )
is comprised of 948 sources in the mips cdf - s legacy field with @xmath14 @xmath1jy . while this mips depth can be obtained over the full cdf - s , we chose to limit this study to the goods region to take advantage of the super - deep irac imaging .
the relative depths of the limiting mips flux and the super - deep irac photometry ensure that essentially all mips sources have high s / n irac seds , allowing us to study in an unbiased way the irac properties of this flux - limited mips sample . to ensure that all agn candidates have sufficient x - ray coverage , we also required x - ray coverage of @xmath15 ks from the deep @xmath16 cdf - s x - ray dataset ( see giacconi et al .
( 2002 ) and alexander et al .
( 2003 ) ) . despite this relatively low cut ,
96% of the final mips sample have @xmath17 ms and 80% have @xmath18 ms .
the resulting sample was drawn from an area of 195.3 sq .
arcmin and contains 846 mips sources .
the cdf - s is one of the best - imaged fields in the uv , optical , nir , and x - ray .
we took advantage of this extensive multiwavelength dataset by creating an aperture - matched catalog using the uv - nir photometry of marzke et al .
( 1999 , @xmath19 ) , vandame et al . ( 2001 , @xmath20 ) , arnouts et al . ( 2002 , @xmath21 ) , combo17 ( wolf et al . 2004 ) , giavalisco et al . ( 2004 , @xmath22 ) , le fvre et al . ( 2004 , @xmath23 ) , and galex ( @xmath24 ) ( see prez - gonzlez et al . 2005 and the ucm extragalactic database for more details ) .
we then removed from our sample the 89 mips sources that had multiple optical counterparts that were ( 1 ) within a 2.5 search radius of the mips source and ( 2 ) separated by more than 0.5 .
( at @xmath25 , it is difficult to distinguish between multiple counterparts and an extended / irregular source with multiple components . ) while we do not restrict our sample to regions covered by goods acs imaging , 83% of the mips sources in our sample have deep acs coverage .
sextractor - selected irac sources were similarly matched to the mips sample after combining the super - deep irac data with data from the deep , wide - area , _ spitzer _ legacy program ( pi : van dokkum ) and the mips - gto irac program ( see prez - gonzlez et al .
2008 for further details ) . to ensure accurate mir seds , we removed from the ir plg and color - selected samples 32 sources with blended irac or mips photometry . of the remaining 725 mips sources that meet our criteria , 713 ( 98% ) have unique irac counterparts within a 2 search radius , and 699 ( 96% ) have @xmath26 irac detections in all 4 irac bands , allowing us to determine accurately the mir colors of essentially all members of the mips - selected sample . of the 12 sources without irac counterparts ,
6 were not detected due to blending with a nearby source , 4 had badly centered mips positions , and 2 had faint irac counterparts that fall below our catalog limit . after assigning irac counterparts , we separated the mips sample into three subsets : ir plgs , ir color - selected galaxies , and ir - normal galaxies .
we defined as plgs sources whose 4-band irac photometry is well - fit by a line of slope @xmath27 , where @xmath28 ( alonso - herrero et al .
2006 ; donley et al .
2007 ) . the effect of different cuts in @xmath29 will be discussed in 6 . to ensure a good fit
, we required the chi - squared probability p@xmath30 ( the probability that the fit would yield a chi - squared greater than or equal to the observed chi - squared ) to exceed 0.1 .
p@xmath30 tends either to lie close to 0.5 ( the probability that corresponds to a reduced chi - squared of 1 ) or is very small ( see bevington & robinson 2003 ) .
this selection , which identified 55 plgs , was done using the linfit task in idl .
this task takes the following 4-band input from irac : @xmath31 @xmath32 @xmath33*\delta f_{\nu}/f_{\nu}\ ] ] and returns the best fit slope , @xmath29 , and the chi - squared probability , p@xmath30 .
while sources selected via this method are dominated by the agn in the mid - ir , we do not require that the power - law extend into the optical .
consequently , many infrared - selected plgs are dominated by stellar emission at wavelengths short of @xmath34 , where the reprocessed emission from hot dust is suppressed because of dust sublimation .
color - selected galaxies were defined as sources that meet the agn irac color - cuts of lacy et al .
( 2004 ) or stern et al .
( 2005 ) , but that * do not meet the plg criterion*. as discussed in alonso - herrero et al .
( 2006 ) and donley et al .
( 2007 ) , both the plg and irac color cuts attempt to select luminous agn that outshine their host galaxies in the infrared , filling in the dip in a galaxy s sed between the short - wavelength stellar emission feature and the long - wavelength dust emission features . as such ,
the irac agn color selection regions contain , but also extend beyond , the power - law locus in color space .
while nearly all plgs meet the irac agn color selection criteria , not all color - selected galaxies meet the plg criteria . we therefore separate color - selected sources that can be identified via a power - law fit ( plgs ) from those that can not ( color - selected galaxies ) .
the color - selected sample consists of 210 sources , 188 of which meet the lacy et al .
criteria , 72 of which meet the stern et al .
criteria , and 50 of which meet both criteria . finally , we define ir - normal galaxies as sources that meet neither the irac plg nor the color selection criteria ; these sources comprise the remaining 448 galaxies in the mips sample .
as they are not the focus of this study , we have not checked their irac and mips photometry by eye .
instead , we estimate that the fraction of ir - normal galaxies with blended irac or mips photometry is similar to that found for the plg and color - selected samples , 11% , or @xmath35 galaxies .
we caution that ir - normal does _ not _ mean purely star - forming .
instead , ir - normal only indicates that any mid - ir emission from an agn is overwhelmed by emission from the host - galaxy .
in fact , many type 2 and seyfert - luminosity agn meet the ir - normal criteria ( stern et al . 2005 , donley et al .
2007 , cardamone et al . 2008 ) .
in addition to dividing the mips - selected sample into the 3 sub - samples discussed above , we identified ir - excess galaxies using the criteria of daddi et al .
( 2007a ) , dey et al .
( 2008 ) , fiore et al .
( 2008 ) , and polletta et al .
( we also searched for bright ulirgs that met the ir - excess criteria of yan et al .
( 2007 ) , but found none in our faint sample . )
daddi et al .
( 2007a ) selected galaxies in goods ( with a @xmath36 24 flux limit of 15 - 30 @xmath1jy ) whose total mir+uv star - formation rate ( sfr ) exceeds the dust - corrected uv sfr by a factor of @xmath37 , and estimate that at least @xmath3 of their sample are compton - thick agn .
dey et al .
( 2008 ) select sources with @xmath38 \ge 14 $ ] ( @xmath39 ) and @xmath40 @xmath1jy , criteria which yield a sample of both heavily obscured agn and star - forming galaxies .
fiore et al .
similarly require that @xmath41 , but also include an optical / nir criterion of @xmath42 and extend their selection to fainter 24 fluxes of @xmath43 @xmath1jy .
they estimate from simulations that 80% of the sources selected via these criteria are obscured agn .
polletta et al .
( 2008 ) focus only on the most luminous agn ( @xmath44 mjy ) with large infrared to optical flux ratios , whose irac and mips colors can be described by the following criteria : @xmath45 , and log @xmath46 + \rm{log } [ f_{\rm 24}/f_{\rm 3.6 } ] > 2 $ ] . because ir - excess sources tend to be optically - faint
, we use the approach of fiore et al .
( 2008 , private communication ) and estimate the r - band magnitudes by interpolating the acs v- and i - band data from the music catalog ( grazian et al .
2006 ) . despite the inherent uncertainties associated with this relatively simple method , the interpolated r - band magnitudes are in excellent agreement with ground - based r - band measurements for bright sources ,
and they greatly improve upon the uncertain r - band estimations at faint flux densities . because the music catalog is based on identifications at @xmath47 and @xmath48 ( the latter of which is universally bright for the ir - excess sources )
, the use of the music catalog for these sources also ensures that the correct optical counterpart is chosen , as verified by a visual inspection of the sources selected via the fiore et al .
the only disadvantage of this method is that , of the 195.3 sq .
arcmin of our survey , only 132.7 sq .
arcmin ( 68% ) are covered by the deep isaac k - band data , which was also used in the selection of the daddi et al .
our identification of the dey et al .
, daddi et al . , and fiore et al .
ir - excess samples is therefore limited to this region . of the 465/713 sources in our mips - selected sample that lie in the isaac field ,
10 meet the dey et al .
criteria , 52 meet the fiore et al . criteria , and 42 lie in the daddi et al . ir - excess sample ( the list of which was kindly provided by d. alexander , private communication 2008 ) . in addition , 71 mips sources have red ir / optical colors of @xmath41 ; we will refer to this sample as ir - bright / optically - faint. of the full sample of 713 sources , 5 meet the polletta et al .
criteria . the properties of these ir - excess galaxies , nearly all of which also meet the power - law or color - selection criteria outlined above , will be discussed in more detail in 7 .
while the redshift coverage in the cdf - s is amongst the highest in all cosmological fields , only 34% of the sources in our faint sample have spectroscopic redshifts .
as one of our main goals is to investigate the redshift - dependency of infrared color selection , we require both accurate and complete redshift information .
however , the mean magnitude of the sources without spectroscopic redshifts , @xmath49 , is very faint , making further spectroscopic follow - up challenging .
we therefore supplement the spectroscopic redshifts with photometric ones .
the sources that are the focus of this study generally have seds with weak stellar features and require extra care to fit accurate photometric redshifts .
to insure this accuracy , we have used two complementary photo - z approaches and have incorporated two stages of independent visual inspection by two reviewers , as described in appendix a. the final results , shown in figure 1 , support the high accuracy and completeness of the adopted redshifts for this faint sample . of the 713 mips - selected sources in our sample , 249 have secure spectroscopic redshifts from the vvds ( le fvre et al .
2004),vlt / fors2 ( vanzella et al . 2006 ) , k20 ( mignoli et al . 2005 ) , and szokoly et al .
( 2004 ) redshift surveys . for our photometric redshifts ,
the mean offset is @xmath50 , where @xmath51 , and the dispersion is @xmath52 , where @xmath53 .
eighty - nine percent of the sample have @[email protected] , and 95% have @[email protected] . as shown in table 1 , while the redshift completeness and accuracy are high for the ir - normal and color - selected samples , they drop significantly for the plgsample .
a comparison of their spectroscopic and photometric redshifts illustrates some of the issues .
three of the four plg outliers are best fit by a type 1 qso template , whose redshift is particularly difficult to constrain .
however , of the 18 plgs for which only photometric redshifts are available , only 2 ( 11% ) are fit by a type 1 qso template .
the remainder show optical features / breaks that make their redshift determination more secure . as such ,
we expect the overall accuracy of the plg redshifts to be higher than one would assume given the limited comparison with spectroscopic redshifts . with the addition of 400 photometric redshifts , computed as described above
, our redshift completeness is 91% .
the redshift distribution is shown in figure 2 .
also plotted in figure 2 is a redshift histogram that incorporates typical errors in the photometric redshifts . to produce this distribution , we simulated 10,000 redshift distributions in which the photometric redshifts were randomly varied according to the appropriate @xmath56 given in table 1 .
we plot in figure 2 the mean of the resulting distributions .
both redshift distributions show a strong peak at @xmath57 , and while the first shows a potential peak at @xmath2 , we can not confirm its presence due to the errors on the photometric redshifts .
we separate the sample in figure 2 into the plg , color - selected ( lacy et al . and stern et al . ) , and ir - normal subsamples . as expected , the number of ir - normal galaxies peaks at @xmath58 ( a well - known redshift peak in the cdf - s ) and at @xmath59 ( another known redshift peak )
we also detect a peak at @xmath60 similar to that found by desai et al .
( 2008 ) , whose strength increases if we apply their cut of @xmath61 @xmath1jy . in the cdf - s , however , this peak is dwarfed by the stronger peak at @xmath58 , even at large flux densities .
the redshift distribution of the ir - normal galaxies decreases rapidly at redshifts of @xmath62 , where only highly luminous star - forming galaxies are detectable .
in contrast , the plgs with redshift estimates have a relatively flat distribution in redshift space , as was found for the plgs in the cdf - n ( donley et al .
2007 ) .
the redshift distributions of the ( non - power - law ) lacy et al . and stern et al .
selected samples differ significantly , both from the ir - normal and plg samples , as well as from one another .
the stern - selected galaxies peak at @xmath63 whereas the lacy - selected galaxies show significant peaks at @xmath64 and @xmath65 , with very few galaxies falling in the @xmath66 regime originally probed by these selection methods ( see 5 ) .
in addition , at @xmath67 , nearly all ( 94% ) mips - selected sources meet the lacy agn selection criteria , regardless of their nature .
a large concentration of galaxies at @xmath68 was previously observed in this field by caputi et al .
( 2006 ) , and in a brighter sample of mips sources by desai et al .
( 2008 ) , and is most probably due to the 7.7 aromatic feature passing through the 24 band .
this behavior suggests that the mid - ir continua of the lacy color - selected galaxies contain substantial contributions from star - formation ( e.g. genzel et al . 1998 ) . finally , the mean redshifts of the ir - excess samples are as follows : @xmath69 for the daddi et al .
sources , @xmath70 for the dust obscured galaxies ( dogs , dey et al .
2008 ) , @xmath71 for the fiore - selected sources , @xmath72 for the one polletta source with a known redshift , and @xmath73 for the ir - bright / optically - faint sources . as discussed above , this concentration about a redshift of 2 ( when not a design of the selection as in daddi et al .
( 2007 ) ) is likely to be due at least in part to the passage of the 7.7 aromatic feature through the mips 24 band , suggesting a significant contribution from star - formation .
of the 713 mips sources , 109 ( 15% ) have x - ray counterparts in the alexander et al .
( 2003 ) or giacconi et al .
( 2002 ) catalogs . of these , 25 are plgs , 35 are color - selected galaxies ( 33 from the lacy criteria , and 12 from the stern criteria ) , and 49 are ir - normal . while the ir - normal galaxies therefore dominate the x - ray counts , the fraction of such sources is low : only 11% of the ir - normal galaxies have x - ray counterparts , as compared to 17% of the color - selected galaxies and 45% of the plgs .
we test for faint ( @xmath74 ) x - ray emission from the mips sources using the procedure outlined in donley et al .
the resulting detection fractions for the plg , color - selected , ir - normal , and ir - excess samples are given in table 2 .
with the inclusion of the weakly - detected x - ray sources , the detection fractions of the ir - normal , color - selected , and plg samples increase to 40% , 42% , and 64% , respectively .
as indicated in table 2 , however , while all strongly and weakly - detected plgs have agn x - ray luminosities of log @xmath75(ergs s@xmath10)@xmath76 , the same is true for only 74% of the color - selected galaxies , and 34% of the ir - normal galaxies . given the infrared luminosities implied by our 24 selection criterion , the portion of the sample with log @xmath77(ergs s@xmath10)@xmath78 will be heavily contaminated with star - forming galaxies ( ranalli et al .
2003 ) . while the x - ray detection fraction of plgs is relatively low ( though significantly higher than that of the color - selected galaxies ) , it is comparable to that of previously - selected samples ( alonso - herrero et al .
2006 , donley et al .
2007 ) . of the plgs in the 2 ms
cdf - n , 55% had high - significance x - ray counterparts .
however , only 15% remained undetected down to the 2.5@xmath56 detection level , suggesting that many plgs are likely to be heavily obscured x - ray sources whose fluxes fall below the current detection limits ( donley et al .
2007 ) . in the cdf - s , only 45% of the plgs have cataloged x - ray counterparts and 36% remain undetected down to @xmath79 . this slightly lower detection fraction is due at least in part to the lower x - ray exposure of the 1 ms cdf - s , and will be discussed further in 6 . the x - ray detection fraction of the luminous ir - excess sources selected via the polletta et al .
( 2008 ) sample is high ( 80% ) .
however , the same can not be said for the remaining ir - excess samples : only 30% of the dogs , 19% of the fiore et al .
sources , and 15% of the sources with high 24 to optical flux ratios have cataloged x - ray counterparts . by definition , none of the daddi et al .
sources have cataloged hard x - ray counterparts , although 3 have soft - band counterparts
. when the weakly - detected x - ray counterparts are included , these numbers rise to 100% for the polletta sources , 63% for the dogs , 44% for the daddi et al .
sources , 43% for the fiore et al .
sources , and 42% for the ir - bright / optically - faint sources .
there are several reasons why a re - examination of the color - color selection criteria is needed before applying these techniques to our sample .
first , thanks to the availability of high quality nir and _ spitzer _ mir spectra and photometry , we can now construct more accurate mir templates for both agn and star - forming galaxies than were available when the lacy et al . and stern et al .
agn selection criteria were initially defined .
second , the agn selection criteria of lacy and stern were initially designed for use with shallow surveys , and have not yet been properly tested over a range of both redshift and flux density ( but see cardamone et al .
third , photometric redshift techniques for _ spitzer_-detected galaxies have advanced sufficiently to allow nearly complete redshift estimation .
the positions of the mips - selected sample in irac color space are shown in figure 3 , where stars represent the plgs and circles represent the remaining mips sources .
as expected , all plgs meet the lacy criteria and all but 3 meet the stern criteria .
the relatively large scatter of the plgs on the stern plot probably arises in part from the use of adjacent color bands , whereas the power - law fitting tends to smooth over noise . to illuminate the strengths and weaknesses of the two color - color selection criteria in figure 3
, we separate the sources that meet both the lacy et al .
( 2004 ) and stern et al .
( 2005 ) agn selection criteria from those that meet only one of the two criteria .
the color - selected sources that meet both criteria primarily define an extension of the power - law locus to a blue slope of @xmath80 ( recall that our definition of plgs includes only those sources with red slopes of @xmath81 ) .
we will discuss in detail the effects of different power - law slope criteria in 6 .
the lacy - only sources occupy two regions in color - space .
the first , located in the lower portion of the lacy diagram , corresponds to the concentration in the upper - left corner of the stern et al .
selection region .
these sources were intentionally excluded from the stern selection region to minimize contamination from high - redshift ( @xmath82 ) star - forming galaxies .
the second concentration of lacy - only galaxies is scattered throughout the star - forming locus of the stern diagram .
are these normal , low - z star - forming galaxies , or obscured agn not selected by the stern criteria ?
the stern - only sources fall almost exclusively in the lower left portion of the stern wedge , and occupy the region dominated by low - redshift ir - normal galaxies in the lacy et al .
color diagram , suggesting that , like the lacy criteria , the stern et al .
criteria are likely to suffer from low - redshift star - forming galaxy contamination . because of the exclusion of @xmath68 star - forming galaxies , however , the stern et al .
criteria should perform better at high @xmath47 . to test this hypothesis
, we next consider the evolution in irac color - space of a number of high - quality star - forming and agn templates . as discussed above
, the availability of high - quality near- and mid - infrared data has allowed the construction of high - quality star - forming seds , particularly for luminous and ultra - luminous infrared galaxies ( lirgs and ulirgs ) . in figure 4
, we plot the redshift evolution of these templates in irac color - color space over @xmath83 , where purple and blue tracks represent the _ purely star - forming _ ulirg and lirg templates of rieke et al .
( 2008 , see table a1 ) , green tracks represent the spiral and starburst templates of polletta et al .
( 2007 ) and dale & helou ( 2002 ) , and red tracks represent the elliptical templates of silva et al .
large circles mark the tracks at @xmath84 , and small circles mark each integer redshift from @xmath85 to 4 .
while the star - forming templates generally avoid the power - law locus itself , they enter the lacy and stern selection regions at both low and high redshifts , tracing out the same regions in color space occupied by many of the color - selected agn , particularly those selected via only one of the two criteria . the templates therefore suggest potential star - forming galaxy contamination of the color - selected agn , as previously predicted by barmby et al .
( 2006 ) , donley et al .
( 2007 ) , and cardamone et al .
( 2008 ) , and indicate that the current agn selection regions may inadequately separate agn and star - forming galaxies .
our results can be compared with the simulations of sajina , lacy , & scott ( 2005 ) , who calculated mathematical models of galaxies from three spectral components : stars , aromatic features , and a continuum . for @xmath86 , we agree with their figure 8 ( upper left ) that starburst luminosity galaxies do not significantly invade the lacy agn color wedge ( see our figure 4 ) . however , we find that more luminous galaxies can invade this wedge much more seriously ( again see figure 4 )
. the difference may arise because the inputs to their models included few star - forming lirgs and only one ulirg ( arp 220 , an atypical case ) .
therefore , it is likely that the behavior of the most luminous star forming galaxies is not captured as accurately in their models as is that of lower luminosity ones . by @xmath85 , the typical _ spitzer _ 24 survey sensitivity limit reaches only to the bottom of the lirg range ,
so the lacy agn color wedge is more susceptible to contamination than was concluded by sajina et al .
( 2005 ) .
while the star - forming templates appear to trace quite well the positions of many of the color - selected galaxies in color - color space , figure 4 covers a wide range of redshifts ( @xmath83 ) . to understand better the overlap between the star - forming templates and the color - selected galaxies ( and plgs ) , we must take the redshift information into account .
we therefore break the sample down into smaller redshift bins , as shown in figures 5 and 6 .
we overplot on the color - color diagrams the redshift - appropriate colors of purely star - forming galaxies . to simplify the plots
, we do not plot each galaxy track separately , as was done in figure 3 , but instead draw 1 @xmath56 contours around the tracks , where @xmath56 is taken to be the median irac measurement error of the full mips sample .
we find that , unlike the plgs , the majority of the color - selected agn candidates fall within or very close to the contours for star - forming galaxies of similar redshifts .
thus , it is likely that their mid - ir seds are dominated by star formation .
this result suggests that simple mid - ir color - color cuts can not identify reliable agn samples without also including redshift - based or additional sed ( e.g. power - law ) criteria .
we discuss in appendix b the individual redshift intervals .
the summary of our findings can be found in table 3 , where we present the overall fraction of color selected galaxies and plgs that lie outside the 1 , 2 , and 3@xmath56 star - forming contours .
the top half of the table gives the fractions assuming no errors on the photometric redshifts ; the lower half incorporates 10% errors which for clarity are imposed on the templates rather than on the individual galaxy measurements . for the plgs , we give two percentages : the fraction of sources that lie outside the star - forming contours in lacy and stern color - space , respectively . as is clear from table 3 , the fraction of color - selected galaxies that lie outside the star - forming contours is lower than that of the plgs by a factor of 2 - 10 at all three levels of significance .
for instance , while 57% - 71% of plgs lie more than @xmath36 from the star - forming contours in lacy color space , the same can be said for only 5%-10% of the color - selected galaxies .
in addition , increasing the significance from 1@xmath56 to 3@xmath56 has a far greater effect on the reliability of the color - selected galaxies than on that of the plgs , especially in lacy color - space , indicating that the color - selected galaxies lie noticeably closer to the star - forming templates than do the plgs , as expected .
we note that this analysis depends on the templates being used , as the addition or removal of any one star - forming template will result in changes in the numerical results presented above and in the table .
the overall trends discussed above , however , will remain the same . while the majority of color - selected agn candidates lie inside the star - forming contours ( see table 3 ) , a number of non - power - law color - selected galaxies meet the agn selection criteria and have colors inconsistent with those of our star - forming templates .
are these agn , as predicted ?
how do their redshift distributions , x - ray detection fractions , and numbers compare to those of the plgs ? in the following discussion , we define as secure color - selected galaxies ( and plgs ) those agn candidates that lie @xmath87 away from the redshift - appropriate star - forming templates .
as discussed above , and as shown in figure 2 , the redshift distributions of the color - selected sources show a strong peak at @xmath88 , attributed to the 7.7 aromatic feature , a star - formation indicator .
in contrast , the plgs have a relatively flat redshift distribution , indicating little or no contribution from aromatic features .
if we do not account for errors in the photometric redshifts , the resulting sample of secure color - selected galaxies retains the large @xmath89 peak , suggesting a large star - formation contribution .
if we incorporate 10% errors on the redshifts of the star - forming templates when selecting our secure candidates , however , the redshift distributions of the remaining color - selected galaxies become relatively flat , with mean redshifts of @xmath90 and @xmath91 for the lacy and stern - selected sources , respectively , suggesting that a significant fraction of the high - redshift star - forming contaminants have been removed . for comparison ,
the mean redshifts and rms ranges of the secure plgsselected in lacy and stern color - space are @xmath92 and @xmath93 .
while the redshift distribution therefore suggests that many of the star - forming contaminants have been rejected from the secure sample , the x - ray detection fractions , shown in table 4 , suggest otherwise . at distances of 1 and 2 @xmath56 from the star - forming templates , the x - ray detection fraction of the plgs exceeds that of the color - selected galaxies by factors of @xmath94 .
this factor , however , should be taken with reservations ; x - ray sources are far more likely to have spectroscopic redshift estimates , and are therefore far more likely to be included in our secure sample , which requires redshift information .
while this has a minimal effect on the color - selected galaxies , for which the redshift completeness is high , it has a large effect on the plgs , boosting their probable x - ray detection fraction .
a better comparison is therefore with the ir - normal galaxy population , 11% of which are detected in the x - ray .
the x - ray detection fractions of the secure color - selected agn candidates ( 14 - 29% ) , exceed this value , but only by a factor of @xmath95 . in comparison
, the full sample of plgs has an x - ray detection fraction of 45% , which exceeds that of the ir - normal galaxies by a factor of 4 . only at the highest significance , @xmath36 from the star - forming contours , does the x - ray detection fraction of the secure color - selected galaxies approach that of the full plg sample .
this significant offset in x - ray detection fraction could be due either to a difference in intrinsic luminosity or to lingering contamination of the color - selected sample by star - forming galaxies . while the average x - ray luminosity of the secure plgs , log @xmath77(ergs s@xmath10)@xmath96 , exceeds that of the secure lacy and stern - selected sources detected in the x - ray , log @xmath77(ergs s@xmath10)@xmath97 and 43.9 , respectively , the offset is relatively small , suggesting that the large discrepancy in the x - ray detection fractions is not driven primarily by a systematic offset in x - ray luminosity of the relevant agn .
instead , it likely arises from the inclusion of star - forming galaxies , even in the @xmath98 secure color - selected population , suggesting that a larger cut in @xmath56 ( e.g. @xmath36 ) is required to define a reasonably secure sample . also shown in table 4 are the total number of x - ray selected galaxies in both the secure color - selected and plg samples . after applying completeness corrections for the fraction of lacy , stern , and plg x - ray sources that have redshifts and that can therefore be included in the secure sample ( 94% , 83% , and 76% , respectively ) , the number of secure x - ray selected plgs exceeds that of the secure x - ray detected color - selected galaxies , regardless of whether we define the secure sample as those sources that lie 1 , 2 , or 3@xmath56 from the star - forming contours
this indicates that _ the plg selection criterion identifies the overwhelming majority of secure agn candidates in irac color - space .
_ while the lacy and stern agn selection criteria were defined using relatively shallow surveys , these selection techniques are now being applied to samples with a range of flux densities ( e.g. cardamone et al .
how does the limiting flux affect the completeness and reliability of agn color selection ?
are the problems discussed above present in shallow as well as deep surveys ?
there are several reasons why we might expect to see a shift in reliability with flux .
first , the agn fraction of mips sources depends quite strongly on the 24 flux density ( e.g. treister et al .
2006 , brand et al .
for instance , while only @xmath99 of mips sources at our flux limit of 80 @xmath1jy are expected to be agn ( treister et al .
2006 ) , the fraction at 5 mjy is more than an order of magnitude greater ( @xmath100 ) . therefore , mips - selected shallow surveys should contain fewer star - forming galaxies , although those that remain are likely to be lirgs / ulirgs that enter the agn selection region in greater numbers than spirals and starbursts ( see figure 4 ) .
second , many of the problem sources that lie inside the agn color selection region but that also fall inside the star - forming contours have moderately high - redshifts ( @xmath101 ) , and are therefore likely to drop out of shallow surveys .
while shallow samples may therefore contain high - luminosity ulirgs , the exclusion of lower - luminosity high - redshift galaxies reduces the risk of contamination .
other flux cuts , such as the @xmath102 magnitude @xmath103 requirement of stern et al .
( 2005 ) , also prevent contamination at high-@xmath47 , as normal galaxies at this magnitude are detected only to @xmath104 . to investigate the effect of intermediate flux density cuts on the agn color selection , we show in table 5 the fraction of secure color - selected and plgs ( those that lie outside the 1@xmath56 star - forming contours ) as a function of flux density .
if we do not incorporate errors in the photometric redshifts into our definition of the secure sources , we find that the fraction of secure sources is relatively constant at @xmath3 regardless of flux density .
if we allow errors on the photometric redshifts when constructing our secure sample , the reliability of lacy - selected sources at @xmath105 @xmath1jy drops to 29% , and that of stern - selected sources drops to 21% . at @xmath106 @xmath1jy , the fraction of secure sources amongst both samples rises , but only to 50% .
regardless of our assumptions , therefore , the fraction of potential contaminants is still high ( @xmath3 ) , even at the highest fluxes probed by our survey .
because of the pencil - beam nature of the cdf - s survey , 95% of the sources in our sample have @xmath107 @xmath1jy , so our ability to comment on brighter samples is limited .
it is also worth noting that the brightest mips sources tend to lie above the power - law locus in lacy color - space and to the right of the power - law locus in stern space , regions where we expect minimal contamination from star - forming galaxies .
not surprisingly , these sources are also almost always detected in the x - ray , suggesting that these regions of irac color - space are the most secure . in a study of 77 agn candidates selected from the _ spitzer _ first - look ( lacy et al . 2005 ) and swire surveys ( lonsdale et al .
2003 ) via the lacy et al .
( 2004 ) criteria , lacy et al .
( 2007 ) found that 33% are unobscured type 1 quasars , 44% are type 2 agn , and 14% are dust - reddened type 1 quasars .
only 9% have star - forming or liner spectra .
is this relatively low contamination by star - forming galaxies consistent with our findings ?
there are three main factors that lead to the high reliability of the lacy et al .
( 2007 ) sample .
first , the sample members are very bright , with a typical 24 flux density of 5 mjy . as discussed above , treister et al .
( 2006 ) predict that 45% of sources at this flux density should be agn , regardless of their mir seds , compared to only 4% of sources at our flux limit of 80 @xmath1jy .
indeed , of the brightest 50 sources in the xfls ( whose median 24 flux density is 9.4 mjy ) , 58% are optically classified as agn ( lacy et al .
thus , the sample from which these color - selected agn was drawn contains far fewer star - forming galaxies than the deeper goods sample .
second , while the sources in the lacy et al .
( 2007 ) sample have redshifts ranging from @xmath108 to 4.27 , the median redshift of the sample is low : @xmath109 .
only 5 color - selected sources have redshifts in excess of @xmath110 , the redshift above which nearly all sources in our sample ( agn or star - forming ) meet the lacy et al . criteria .
we would therefore expect little or no contamination by _ high - redshift _ star - forming galaxies , the predominant source of contamination in our faint sample .
the third and most important reason for the high reliability of this sample , however , is the high plg fraction of the lacy - selected agn candidates . of the 77 sources in the lacy et al .
( 2007 ) sample , 59 are plgs as defined in 2 .
the 18 sources that are excluded by the plg criteria include the only two starburst galaxies detected in this subsample , one starburst / liner , an unclassified high - redshift galaxy with narrow uv emission lines , two composite galaxies , 8 type 2 agn ( the identification of 4 of which were based on a bpt analysis ) , 3 reddened type 1 agn , and 1 type 1 agn . if we relax the probability constraint of the plg criterion to p@xmath111 , as was done in alonso - herrero et al .
( 2006 ) , we recover 69 of the 77 lacy et al .
( 2007 ) sources selected for optical follow - up . excluded are the two starburst galaxies , one high - z unidentified galaxy , three type 2 agn , one reddened type 1 agn , and 1 type 1 agn whose slope of @xmath112 falls just short of our cut of @xmath27 .
the high reliability of this luminous color - selected sample is therefore consistent with our findings in 6 that while non - plgcolor - selected agn are subject to contamination by star - forming galaxies , sources selected via a power - law criterion are reliable .
thus far , we have focused primarily on the reliability of the agn color selection criteria of lacy et al .
( 2004 ) and stern et al .
in doing so , we have separated out the subset of sources that meet the power - law criteria of alonso - herrero et al .
( 2006 ) and donley et al .
( 2007 ) . here
, we discuss the plg selection itself , in particular the consequences of various choices of limiting power - law slope , @xmath29 .
our default cut of @xmath81 was chosen to match the spectral indices of typical agn ( e.g. alonso - herrero et al .
2006 , donley et al .
2007 ) . in the optical ,
agn have spectral slopes of @xmath113 to -2 ( sdss , ivezi et al .
2002 ) , with a mean value of @xmath114 ( neugebauer et al . 1979 , elvis et al .
1994 ) . in the irac bands , broad - line agn exhibit similar slopes , with a mean value of @xmath115 ( stern et al . 2005 ) .
in figures 7 and 8 , we plot plgs selected via cuts in @xmath29 ranging from + 0.5 to -1.5 . at high redshift ( @xmath116 )
the irac bands trace the blue side of the stellar bump .
it is therefore not surprising that at the bluest slopes ( @xmath80 ) , the plgsample is dominated by the population of high redshift x - ray non - detected star - forming galaxies discussed above . as the required slope reddens towards our cut of @xmath81 , the high - redshift star - forming galaxies gradually drop out of the sample , and the x - ray detection fractions rise from 21% at @xmath117 to 30% , 45% , 67% , 88% , and 80% at @xmath118 and -2.0 .
while choosing a redder cut in @xmath29 therefore increases the apparent reliability of the plg selection , it also decreases the number of galaxies selected , and may exclude interesting heavily obscured , x - ray non - detected agn like those seen in the cdf - n ( e.g. donley et al .
2007 ) .
the x - ray non - detected plgs in the current sample tend to be the faintest sources both in the irac bands and at 24 , where their mean flux density , 146 @xmath1jy , is over a factor of 2 lower than that of the x - ray detected sample , 334 @xmath1jy .
the low x - ray detection fraction of these faint sources may therefore be due simply to their systematically lower fluxes . at @xmath119 , however , the star - forming ulirg templates have irac seds that meet the plgcriteria , although dropping the power - law slope criterion to @xmath120 and @xmath121 raises this redshift range to @xmath122 ( depending on the template ) and to @xmath123 , respectively ( with the iras 22491 - 1808 template never reaching an @xmath29 of -1.5 ) .
it is therefore possible that the x - ray non - detected plgs ( which tend not to have redshift estimates , to be faint , and to lie at relatively high @xmath29 ) are high - redshift star - forming galaxies . to test this hypothesis , we plot in figure 9 the ratio of the mips 24 flux density to that of the 3.6 irac band . at the redshifts of interest ( @xmath124 )
, this flux ratio allows a direct comparison of the hot dust emission at 5 - 7 to the stellar emission at @xmath125 .
overplotted on the colors of our mips - selected sample are the redshifted colors of the agn templates of polletta et al .
( 2007 , see table 1 ) and the purely star - forming lirg and ulirg templates of rieke et al .
the mean colors of plgs without redshift estimates are given by large symbols placed at @xmath126 , the redshift above which contamination by star - forming galaxies is possible .
the circle , square , and triangle represent all plgs without redshifts , those that are detected in the x - ray , and those that are not , respectively .
as can be seen in figure 9 , the x - ray detected and x - ray non - detected plgs lacking redshifts have consistent 24 to 3.6 colors , suggesting that there is no significant difference between these two sub - samples .
in addition , while the colors of these sources are consistent with those of agn , they lie well above those of star - forming galaxies , not only at high redshift , but at all redshifts greater than @xmath127 .
in other words , the plgs that are undetected in the x - ray and that lack redshifts ( preventing us from testing their reliability as in 5.3 ) appear to have more hot dust emission than can be explained by purely star - forming templates , especially at high - z where star - forming contamination of the plg sample is most likely .
this result suggests that these sources are more likely to be agn than star - forming galaxies .
other lines of evidence , e.g. variability ( klesman & sarajedini 2007 ) , and x - ray properties ( steffen et al . 2007 ) , tend to support this conclusion . in addition , we note from figure 9 that a number of the other galaxies at @xmath128 have colors slightly redder than those predicted by star - forming galaxies at this redshift
. this behavior may be due to a minor issue with the templates , a small agn contribution to the mir flux density , or to reddening that exceeds that seen in our local lirg / ulirg templates . finally , the segregation of the x - ray detected and x - ray non - detected sources about the power - law locus , seen in figures 7 and 8 , warrants discussion , as this was not seen in the cdf - n ( donley et al .
this behavior appears to be due largely to the different selection methods and limiting fluxes of the two samples .
while the cdf - n plg sample was selected on the basis of irac fluxes , the current sample was selected from a flux - limited mips sample with far deeper irac data .
if we require the cdf - s sample to meet the irac detection limits of the cdf - n study , the plg sample size decreases by a factor of 2 and the x - ray detection fraction rises to 78% .
the segregation in the plg sample is also greatly reduced , due largely to the loss of many of the x - ray non - detected sources .
this high x - ray detection fraction of 78% , however , is surprising as it is significantly higher than the 55% found for the cdf - n plg sample . to test whether this change is a result of the much improved irac data
, we recalculated the spectral properties of the galaxies in the cdf - n using the most current irac data , which now includes the super - deep goods - n coverage not previously available .
drawing plgs from the same initial sample used in donley et al .
( 2007 ) , we find an updated x - ray detection fraction of 57% , a value nearly identical to that found previously . for consistency with the current sample , we further restricted the plg sample to those sources detected in the super - deep goods field , where the x - ray exposure is generally higher .
doing so raises the x - ray detection fraction to 65% .
the remaining offset in the x - ray detection fractions of the irac flux - limited plgs in the goods - n and goods - s fields , 13% , therefore appears to be due primarily to cosmic variance and small number statistics .
both high - redshift star - forming galaxies and agn can exhibit extremely red infrared to optical colors when heavily dust - obscured ( e.g. dey et al . 2008 ) . while the relative fractions of agn and star - forming galaxies among ir - excess samples is still a matter of debate , the fraction of dust - obscured galaxies ( dogs ) with agn - like power - law seds has been shown to decrease with decreasing flux from @xmath129 at @xmath130 mjy to @xmath131 at @xmath132 @xmath1jy ( dey et al . 2008 ) .
we might therefore expect a relatively small agn fraction amongst the faint ir - excess sources in the goods fields , whose mean 24 flux is 215 @xmath1jy .
fiore et al .
( 2008 ) , however , estimate that 80% of their red , ir - excess galaxies are heavily obscured , compton - thick agn . here ,
we probe the nature of the ir - excess sources and examine the overlap between the ir - excess galaxies and the x - ray , color - selected , and power - law samples discussed above .
the positions of the ir - excess galaxies in lacy and stern irac color - space are shown in figure 10 , and their overlap with the x - ray , plg , and color - selected samples is given in table 2 .
of the 101 ir - excess sources , 24 ( 24% ) are plgs and all but 7 meet either the lacy , stern , or power - law criteria .
at least 1/4 of the ir - excess galaxies therefore show evidence for agn - heated dust ( with the fraction rising to 100% for the luminous agn selected via the polletta et al .
( 2008 ) criteria , as expected ) .
not all plgs , however , have ir - excess colors . of the 55 plgs ,
only 44% meet one or more of the ir - excess criteria .
the same can be said for 33% of the irac color - selected galaxies , but for only 2% of the ir - normal galaxies . of the ir - excess sources ,
only 15% have cataloged x - ray counterparts , with the fraction rising to 45% when weak ( @xmath133 )
x - ray detections are considered ( see table 2 ) . of the x - ray cataloged sources , 47% are plgs , 47% are color - selected galaxies , and 1 ( 6% ) is ir - normal .
the statistics are quite different for those sources with no cataloged x - ray emission , where the power - law fraction drops to 19% and the color - selected fraction rises to 73% . while a strong argument can be made for the agn nature of the 32 ir - excess sources with cataloged x - ray counterparts and/or plgseds , what can be said about the remaining ir - excess sources ?
as discussed in 3 , the concentration of the redshifts of these sources about @xmath89 may be due to the passage of the 7.7 aromatic feature through the 24 band , suggesting a significant contribution from star - formation .
in addition , the vast majority of color - selected sources at @xmath89 ( of which these are a subset ) have irac colors consistent with those of star - forming galaxies or low - luminosity agn whose irac seds are dominated by the host galaxy . in combination with the lack of significant x - ray emission , these facts suggest that the remaining 68% of the ir - excess sample could be either star - forming galaxies , or extremely obscured agn . to distinguish between these two possibilities , we consider in more detail the sources selected via the daddi et al . and
fiore et al
. criteria . of the 88 ir - excess agn candidates selected in the isaac region of goods - s by daddi et al .
( 2007a ) , the list of which was kindly provided by d. alexander ( private communication , 2008 ) , 42 fall in our mips - selected sample .
twenty - three of the remaining 46 galaxies have mips fluxes that fall below our cut of @xmath134 @xmath1jy , and 6 galaxies lack mips and/or irac counterparts in our catalogs , indicating that they too are faint .
seven galaxies were excluded because of observable blending in the mips or irac bands , and an additional 10 were removed from our sample because of multiple optical counterparts .
the positions of the daddi et al .
sources in irac color - space are shown in figure 10 .
thirty - eight of the 42 daddi et al .
sources in our sample meet the lacy et al .
irac color - selection criteria .
only 10 meet the stern et al .
agn selection criterion , and only 5 are plgs ( 4 of which have relatively shallow slopes of only @xmath135 ) .
in addition , only 15 ( 36% ) meet any of the other ir - excess criteria .
this is not surprising , as the mir emission from the daddi et al .
ir - excess sources exceeds that of a typical star - forming galaxy by a factor of only @xmath136 ( daddi et al .
2007b ) . of the 42 ir - excess galaxies ,
18 have irac colors that lie @xmath137 from the redshift - appropriate star - forming contours in either lacy or stern color - space . only 3 and 1 , however , have colors that lie outside the @xmath138 and @xmath36 contours , respectively .
as discussed above , however , this indicates only that any agn activity can not be identified on the basis of the mir irac colors alone . to determine the nature of the daddi et al .
( 2007 ) sources in our sample , we therefore turn to the x - ray data .
while none of the ir - excess galaxies are individually detected in the hard x - ray band ( by definition ) , 3 have soft x - ray detections and 13 have faint ( @xmath139 ) x - ray counterparts .
twenty remain x - ray undetected , and 6 sources have a nearby x - ray counterpart that prevents an accurate test for low-@xmath56 x - ray flux . using the procedure outlined in steffen et al .
( 2007 ) , with the only change being our slightly different choice of source aperture radius , 2 , we coadded the three sources detected in the soft band , whose soft - band luminosities of @xmath140 ergs s@xmath10 indicate that they are agn .
we verify that our stacking method reproduces the results of daddi et al .
( 2007b ) for the same sample of 59 ir - excess galaxies used in that work .
while the coaddition of the 3 soft x - ray detected sources did not lead to a hard band detection , we place a @xmath141 limit on the hard - band flux that constrains their column density to @xmath142 @xmath143 , assuming an intrinsic photon index of @xmath144 .
a coaddition of the 13 weakly - detected sources leads to a @xmath145 hard - band detection , a @xmath146 soft - band detection , a hardness ratio of @xmath147 , and a photon index of @xmath148 .
if we assume that these sources are agn at their mean redshift of @xmath149 , the hard to soft flux ratio corresponds to an obscured column density of @xmath150 @xmath143 . at this modest
obscured yet compton - thin column density , the observed soft band flux is only attenuated by a factor of 2 , implying that the sources must have relatively low luminosities . indeed
, the rest - frame 2 - 10 kev absorption - corrected luminosity derived from the observed soft - band flux is only @xmath151 ergs s@xmath10 . however , daddi et al .
( 2007b ) argue that the coadded soft x - ray flux of their full ir - excess sample can be attributed to star - formation .
when they subtract this component from the detected fluxes , the hardness of the remaining x - ray emission implies a significantly larger ( e.g. compton - thick ) column density as well as a larger absorption - corrected luminosity . to test the origin of the x - ray emission , we plot in figure 11 the 2 - 10 kev luminosity and 25 power ( @xmath152 ) of a sample of starburst and agn - dominated galaxies and ulirgs drawn from franceschini et al .
( 2003 ) , ranalli et al .
( 2003 ) , and persic et al .
the 25 luminosities were extracted from the iras revised bright galaxy sample ( rbgs , sanders et al .
2003 ) when available , and from iso ( klaas et al . 2001 ) or the iras faint source catalog otherwise . as both the hard x - ray and mir flux densities of star - forming galaxies
trace the current star - formation rate ( e.g. ranalli et al .
2003 , franceschini et al .
2003 ) , these two luminosities are well correlated for starbursts and starburst - dominated ulirgs ( though there is a hint of a turnover to lower x - ray luminosities amongst the ulirg sample ) .
agn , however , show an increased x - ray output for their observed 25 flux density . to compare the daddi et al .
sample to these local starburst and ulirg samples , we convert the observed soft - band ( rest frame @xmath153 kev ) luminosity to a rest - frame 2 - 10 kev luminosity , and the observed 70 ( rest - frame @xmath154 ) flux density given by daddi et al .
( 2007b ) to a rest - frame 25 power .
the mean luminosities of the full daddi et al .
( 2007b ) ir - normal and ir - excess samples fall on the best - fit correlation for star - forming galaxies , confirming that their observed soft x - ray flux can be attributed to star - formation .
the x - ray luminosity of the x - ray weakly - detected ir - excess galaxies in our mips sample , however , is somewhat higher than predicted .
the agn origin of the soft x - ray emission is supported by the 2 - 10 kev luminosity implied by the soft x - ray flux , @xmath151 ergs s@xmath10 .
if the soft x - ray flux has a significant contribution from the agn , then it can not all be subtracted from the agn spectrum , resulting in column densities lower than estimated by daddi et al .
( 2007b ) .
however , the indicated agn contribution is modest , and is based on an average . therefore , it is likely that there will be a significant range of star - formation corrected absorbing columns , including some that are compton thick .
indeed , alexander et al . (
2008 ) show that a sample of 6 spectroscopically - confirmed compton - thick agn would be selected via the daddi et al .
( 2007 ) method , although 4 of the 6 have infrared excesses of a factor of @xmath155 , and thus represent the most extreme ir - excess sources .
a coaddition of the 20 x - ray non - detected daddi et al .
sources in our sample leads only to a marginal @xmath156 detection in the full band ( 0.5 - 8 kev ) , and @xmath157 and @xmath158 detections in the hard and soft bands , respectively .
a @xmath36 limit on the soft flux gives a 2 - 10 kev luminosity of @xmath159 ergs s@xmath10 , fully consistent with a star - forming origin ( see figure 11 ) .
( the 2 - 10 kev luminosity derived from the marginal full - band detection is even lower : @xmath160 ergs s@xmath10 ) .
the near - detection in the hard - band , however , suggests that these sources may have relatively hard x - ray spectra . by coadding the weakly and non - detected sources
, the photon index drops to @xmath161 , from a value of @xmath162 measured for the weakly - detected sources .
there is therefore marginal ( @xmath163 ) evidence that the x - ray non - detected sources have spectra harder than those of their weakly - detected counterparts .
does this hard spectrum confirm that these sources are agn , or might there be another explanation ? while it is well - known that heavily obscured agn exhibit hard x - ray spectra , star - forming galaxies can also produce such spectra . as discussed in persic & rephaeli ( 2002 ) and persic et al .
( 2004 ) , the hard ( @xmath164 kev ) x - ray emission of star - forming galaxies is dominated by low and high - mass x - ray binaries ( lmxbs , hmxbs ) , with the hmxb fraction increasing from @xmath165 at starburst luminosities to @xmath166 at ulirg luminosities . as shown by white ,
swank , & holt ( 1983 ) , hmxbs have x - ray spectra with @xmath167 , a cutoff energy of 20 kev , and an @xmath168-folding energy of @xmath169 kev ( see e.g. persic & rephaeli 2002 ) . at @xmath89 ,
the 0.5 - 2 kev and 2 - 8 kev x - ray bands sample the rest - frame 1.5 - 6 kev and 6 - 24 kev x - ray bands , and therefore should be minimally affected by the power - law cutoff . a luminous star - forming galaxy undergoing an isolated burst of star - formation
is therefore expected to display a hard x - ray spectrum of @xmath170 over the energies observed in our sample .
x - ray binary emission has been proposed as an explanation for the hard ( @xmath171 ) spectrum of arp 220 ( iwasawa et al .
2001 ) , the spectra of the apparently starburst - dominated ulirgs of ptak et al .
( 2003 ) and teng et al .
( 2004 ) , whose photon indices tend to lie at @xmath172 , and the starburst - dominated ulirgs of franceschini et al .
( 2003 ) .
while the hard x - ray properties of the ir - excess sources do not therefore require an agn origin , can the same be said of their mir properties ? to test the nature of these sources , we first plot in figure 12 the median seds of the full goods - s ir - normal and ir - excess samples , as given in figure 4 of daddi et al .
( 2007b ) .
although these high - redshift sources appear to be highly luminous ( see figure 11 ) , there are indications that local templates for lower - luminosity galaxies are appropriate for them , at least in the optical mir ( rigby et al .
it is therefore not surprising that the optical mir sed of the ir - normal galaxies is well - fit by the m82 sb template of polletta et al .
( the discrepancy between the observed sed and template at shorter wavelengths is likely due to a difference in the reddening and/or age of the stellar population . ) the ir - excess sed is also very well fit by the _ purely star - forming _ template of the lirg ngc 3256 ( rieke et al .
2008 ) , to which we have applied an additional reddening of @xmath173 . while this does not rule out an agn contribution to the daddi et al
. ir - excess sample , it does indicate that an agn need not be present to produce the observed seds of the ir - excess sources , and that their ir - excesses may simply be due to strong aromatic emission associated with their systematically higher ir luminosities , as suggested by daddi et al . ( 2007a ) .
if these ir - excess sources are dominated by star - formation , then it appears either that their ir sfrs have been overestimated or that their uv - derived sfrs have been underestimated .
the former scenario could be partially attributed to the larger than unity slope between 24 luminosity and pa@xmath29-derived sfr ( alonso - herrero et al .
2006 , calzetti et al . 2007 ) , which indicates that as the sfr increases , an increasing fraction of the resulting light is emitted in the mir .
a simple proportional relationship between sfr and ir luminosity ( e.g. that of kennicutt 1998 ) will therefore increasingly overpredict the sfr for more luminous infrared galaxies .
an underestimate in the uv - derived star - formation rates could be due to the inherent difficulties in determining accurate uv sfrs for luminous , heavily obscured galaxies ( e.g. goldader et al .
2002 , buat et al .
2005 ) . while uv extinction is known to correlate with luminosity ( see goldader et al .
2002 ; vijh , witt , & gordon 2003 ; buat et al .
2005 ) , daddi et al .
( 2007a ) find no difference between the average derived @xmath174 values of the ir - normal and ir - excess samples , despite the systematically higher 8 luminosities of the ir - excess sample ( see figure 2 of daddi et al .
2007b ) . under the assumption that the radio emission arises purely from star - formation ,
the radio luminosities of the 16 radio - detected ir - excess galaxies similarly indicate the uv sfrs have been underestimated by a mean factor of 6.4 , and a median factor of 3.5 ( see figure 12 of daddi et al .
a stack of all ir - excess galaxies , however , leads to radio sfrs that are consistent with those derived in the uv .
we have divided the 42 daddi et al .
ir - excess galaxies in our mips - selected sample into three subsamples : those that are x - ray detected ( 3 ) , those that are weakly - detected in the x - ray ( 13 ) , and those that remain undetected in the x - ray down to @xmath138 ( 20 ) . ( the remaining 6 sources lie too close to an x - ray source to test for faint emission . )
the 3 x - ray detected sources are agn with @xmath175 ergs s@xmath10 , but have relatively low obscuration ( @xmath176 @xmath143 ) .
a coaddition of the weakly - detected sources leads to a hard ( @xmath177 ) x - ray detection .
if these sources are obscured agn , the hardness ratio implies a column density of @xmath178 @xmath143 .
a hard detection , however , could be attributed either to obscured agn activity or to star - formation via hmxbs .
furthermore , the median sed of the full daddi et al .
( 2007b ) ir - excess sample is consistent with that of a local star - forming lirg , suggesting that the mir emission alone also can not be used to rule out a star - forming origin .
a comparison of the x - ray and mir luminosities of these weakly - detected sources does not conclusively distinguish between a star - forming or agn origin for the x - ray emission .
there are therefore 3 possible explanations for the members of this sample of ir - excess galaxies : ( 1 ) the sources are compton - thick agn whose soft x - ray emission can be attributed to star - formation and whose hard x - ray emission comes from the agn , ( 2 ) they are relatively low - luminosity , compton - thin agn , whose soft x - ray emission can not be entirely attributed to star - formation , or ( 3 ) they are star - forming galaxies whose soft _ and _ hard x - ray emission arise from star - formation . while the derived 2 - 10 kev x - ray luminosity of @xmath179 ergs s@xmath10 suggests explanation ( 2 ) , it is not sufficiently high to definitively rule out the remaining scenarios .
it is likely that no single possibility applies to all 13 galaxies , but that the possibilities define the range of their behavior .
the x - ray non - detected sources , when coadded alone and with the x - ray weakly - detected sample , also appear to have a relatively hard x - ray spectrum .
their x - ray luminosity , however , is significantly lower than that of the weakly - detected sources , @xmath159 ergs s@xmath10 .
if the hard spectrum arises from hmxbs , the properties of this sample would be consistent with a star - formation origin . in summary
, while we can not rule out an obscured agn origin for the daddi et al .
ir - excess sources in our mips - selected sample , their properties may also be consistent with a purely star - formation origin in the majority of cases .
further analysis , taking into account the additional 1 ms of x - ray exposure in the cdf - s , is therefore required to establish their nature .
fiore et al .
( 2008 ) select ir - excess galaxies with the following properties : @xmath41 and @xmath42 . using the music catalogs ( see 2.3 ) , we selected 64 sources that meet these criteria and that have 24 flux densities @xmath180 @xmath1jy .
however , 9 of these sources were removed from our mips sample because of visible blending in the mips and/or irac bands ( see 2 ) , which may have been responsible for their anomalously high ir to optical flux ratios ( unlike grazian et al .
( 2006 ) , we do not attempt to de - blend such sources ) .
a visual inspection of the remaining sources led to the removal of 3 additional sources with blended k - band fluxes , leaving 52 high - quality ir - excess sources .
the first fiore criterion was designed to select obscured agn with large x - ray to optical flux ratios ( see their figure 2 ) whose column densities tend to range from @xmath181 to @xmath182 @xmath143 ( see fiore et al . 2008 and references therein ) .
the @xmath183 criterion ensures that only extremely red objects ( eros ) fall in the sample .
the agn amongst ero samples tend also to be x - ray obscured with @xmath184 @xmath143 ( brusa et al .
obscured agn are therefore likely to be targeted by these criteria . however , these two selection criteria are known to identify both agn and star - forming galaxies ( e.g. alexander et al . 2002 , doherty et al .
2005 , dey et al .
we therefore examine whether these criteria are sufficiently stringent to exclude the possibility that the properties of these sources could arise from star formation .
the first of the three main arguments for the compton - thick agn nature of the fiore et al .
sources is the ability of the obscured agn template of iras 09104 + 41091 ( pozzi et al .
2007 ) to reproduce the extreme colors of these sources , and the comparable inability of the m82 and arp 220 star - forming templates to do the same ( see their figure 3 ) .
when fitting seds to the sources in their sample , fiore et al .
find that only 36% of their x - ray non - detected sources are best - fit by elliptical , spiral , m82 , n6090 , or arp 220 star - forming templates . of the sources in our mips - selected sample that meet the fiore et al .
criteria and have good redshift fits , we find an even lower fraction of sources best - fit by these templates : 20% . however , as shown in figure 13 , the purely star - forming iras 22491 template of polletta et al .
( 2007 ) satisfies the fiore et al .
criteria when modest additional reddening ( @xmath185 ) is allowed , indicating that the extreme colors of these sources can be reproduced not only by obscured agn , but by highly reddened star - formation as well . indeed , when we include this star - forming template , as well as the star - forming lirg / ulirg templates of rieke et al .
( 2008 ) , the fraction of fiore sources in our sample best - fit by a star - forming template rises from 20% to 66% . the second argument for the compton - thick nature of the fiore et al .
sources is the high fraction of heavily obscured agn ( @xmath186 @xmath143 ) required to reproduce the hardness ratios and counts of the stacked x - ray emission . using a simulation , fiore et al .
( 2007 ) estimate that 80% of their sources are obscured agn .
the prediction for the x - ray non - cataloged sources in our sample is somewhat lower : @xmath187 , although this fraction rises for the weakly - detected ir - excess sources to @xmath188 ( f. fiore , private communication , 2008 ) . to test this finding for the sources in our sample , we coadded the 36 x - ray non - cataloged fiore et al .
sources that lie far enough from known x - ray sources to allow an accurate test for faint x - ray emission .
the coaddition gave a soft detection ( 0.5 - 2 kev , 3.9@xmath56 ) , but only a weak hard detection ( 2 - 8 kev , 1.7@xmath56 ) .
the resulting hardness ratio , @xmath189 is only slightly harder than that of the 10 x - ray cataloged fiore et al .
sources , @xmath190 , whose column densities fall in the unobscured to highly obscured ( log @xmath191 ( @xmath143)@xmath192 ) range ( tozzi et al . 2006 ) , with a median value of log @xmath193 ( @xmath143)@xmath194 ) . if we stack only the 10 sources with weak x - ray counterparts ( that also lie sufficiently far from known x - ray sources ) , however , we find a @xmath195 hard - band detection , a @xmath196 soft - band detection , and a hardness ratio of @xmath197 , significantly harder than that of the x - ray cataloged sample . at a redshift of @xmath89
, this @xmath198 corresponds to a column density of @xmath199 @xmath143 .
the observed photon index , @xmath200 , indicates an obscured agn origin , as it is inconsistent even with the moderately hard spectra of hmxb - dominated star - forming galaxies ( @xmath201 ) .
stacking the remaining 26 x - ray non - detected fiore sources does not lead to a detection in any band .
it therefore appears plausible , at least in this bright subset of the fiore et al .
sample , that the hard x - ray flux can be attributed to a small number of obscured , yet mostly compton - thin , agn , as opposed to a large number of obscured , compton - thick agn . the third argument for the agn nature of the fiore et al .
sources is the significant ( factor of 30 ) offset between the ir and uv - derived sfrs .
however , the inherent difficulties in determining accurate uv sfrs for luminous , heavily obscured galaxies ( e.g. goldader et al .
2002 , buat et al .
2005 ) , as well as the systematic uncertainties of factors of 10 - 30 in the tir luminosity ( fiore et al .
2008 ) , make this the weakest of the 3 arguments . of the 52 fiore
sources in our mips sample , 10 ( 19% ) are therefore x - ray selected agn , and 10 ( 19% ) are weakly detected in the x - ray , with coadded properties consistent with their being obscured agn . as for the remaining 26 sources for which we can test for faint x - ray emission , we can not rule out the presence of compton - thick agn . however , the lack of coadded hard counts from this x - ray non - detected sample , the significant fraction of such sources that can be fit by star - forming templates , and the compton - thin nature of the x - ray detected fiore et al .
sources all suggest that many of these sources , which make up the remaining 56% of the sample , are instead star - forming galaxies .
our evaluation of the performance of the infrared selection methods allows a preliminary estimate of the overall role of _
spitzer_-discovered agn in the total population . of the 109 x - ray sources in the mips - selected sample , 95 have redshifts , and of these
, 73 ( 77% ) have agn - like x - ray luminosities of ( log @xmath77(ergs s@xmath10)@xmath202 ) .
we therefore assume a sample of 84 ( 77% of 109 ) x - ray selected agn in the mips - selected sample .
we now consider how many ir - selected agn can be added to this total .
of the 55 power - law agn in our sample , 30 lack cataloged x - ray counterparts .
if we assume that all of the x - ray weakly- and non - detected plgs are agn , the power - law selection criteria increases the number of known agn ( 84 ) by 36% . of the 25/30 x - ray
non - cataloged sources for which we could test for faint emission , 7 ( 28% ) show weak x - ray emission at the @xmath203 confidence level , and 18 show no sign of x - ray emission . correcting for the 5 sources with nearby x - ray counterparts
therefore gives an estimate of 8.4 weakly - detected sources . to place a conservative lower - limit on the contribution of x - ray non - detected plgs to the agn population , we select as agn those plgs that are either weakly - detected in the x - ray ( 8.4 ) or that have extremely red slopes of @xmath204 ( 8 , see 6 ) .
combining these two criteria results in a sample of 11.6 agn candidates missed in the x - ray catalogs , for a contribution of 14% .
we therefore conclude that plg selection increases the number of known mips - detected agn by @xmath205 . further adding the 3 color - selected galaxies that lie outside the 3@xmath56 star - forming contours and that lack cataloged x - ray counterparts ( after correcting for the fraction with redshifts and for which the distance from the star - forming contours could therefore be determined ) increases the contribution of irac - selected agn to @xmath206 . a search for weak x - ray emission from the full sample of color - selected galaxies results in the detection of 44 additional sources , 70% of which have x - ray luminosities typical of agn ( see table 2 ) . of these 30 agn candidates , however , 19 ( 63% ) lie at @xmath207 , the redshift above which nearly all ( 94% ) mips sources meet the lacy et al .
criterion , regardless of their nature ( agn / star - forming ) .
their selection as agn candidates is therefore not primarily a function of their irac colors , but of their x - ray properties . as such
, we do not add these additional @xmath208 agn candidates to our _ spitzer_-selected total .
radio / infrared selection , in which objects are selected for excess radio emission relative to that at 24 , provides an alternative way to identify agn independently of their optical and x - ray characteristics . in donley
( 2005 ) , radio - excess agn are defined as those sources with log @xmath209 .
unlike plgs , radio - excess agn tend to lie at @xmath210 , have seyfert - like x - ray luminosities of log @xmath75(ergs s@xmath10)@xmath211 , and have nir seds dominated by the stellar bump ( donley et al . 2005 ) . of the plgs detected in the cdf - n ,
only 3% meet the radio - excess criteria ( donley et al .
while there is therefore almost no overlap between these two agn populations , their x - ray detection statistics are very similar : only 40% of radio - excess agn are cataloged in the 2 ms cdf - n .
if we consider only those radio - excess agn with 24 flux densities in excess of 80 @xmath1jy , the x - ray detection fraction rises to @xmath212 . at this flux limit ,
the cdf - n sample is complete to radio - excess agn as defined by donley et al .
( 2005 ) .
in the cdf - n , the x - ray and mips detected radio - excess galaxies ( with x - ray exposures greater than 1 ms ) account for 3% of the total number of such sources , and their x - ray non - detected counterparts increase the number of known agn by 2% .
only 10 - 15% of agn , however , are radio - intermediate or radio - loud .
this small observed sample of radio - excess agn is therefore indicative of an underlying population at least 7 times larger , which would increase the known population of mips - detected agn by @xmath213 if a way could be found to identify them .
martinez - sansigre et al .
( 2006 ) also select high - redshift obscured agn candidates via a 24 and radio flux cut , although of the 21 agn candidates chosen by them , only 6 ( 29% ) are sufficiently radio loud to meet the infrared - to - radio selection criteria used above to define radio - excess agn .
their selection , however , also includes a 3.6 irac cut designed to select red galaxies . because it is designed for use in shallow surveys , only three galaxies in the goods region of the cdf
- n meet their mips and radio flux cuts ( using the radio data of richards 2000 ) , and none are red enough to meet all three criteria .
this selection method therefore does not contribute to the agn sample in the deep cdf - s . as discussed above , the ir - excess samples of daddi et al .
( 2007a ) , dey et al .
( 2008 ) , polletta et al .
( 2008 ) , and fiore et al .
( 2008 ) contain various fractions of agn and star - forming galaxies , with only the polletta et al .
selection criteria unquestionably identifying agn . of the polletta et al .
sources , however , 80% are detected in the x - ray catalogs , and the remaining source is a weakly - detected plg , whose contribution to the agn population has already been considered .
daddi et al .
( 2007a , b ) conclude that at least 50% of their ir - excess galaxies are compton - thick agn . if this hypothesis holds for the sources in our sample , the daddi et al .
selection criteria would contribute @xmath214 agn to the mips - selected sample in the isaac field , or @xmath215 agn to the full mips sample ( as the isaac region comprises only 68% of our full survey area ) . as discussed in 7.1 , however , it appears plausible that the properties of many of these sources can be attributed to star - formation .
we therefore add only the weakly - detected galaxies , whose mean x - ray luminosity suggests a likely agn origin . of the 13 weakly - detected daddi
ir - excess galaxies , 1 is a weakly - detected plg and 1 is a color - selected galaxy that lies @xmath216 from the star - forming contours , leaving 11 sources whose contribution is yet to be counted ( or 12.6 when we correct for the sources for which we could not test for weak x - ray emission ) . further scaling to the full sample region results in an additional contribution of 18.6 sources , or 22% .
f. fiore ( private communication , 2008 ) likewise concludes that @xmath217 of the x - ray non - cataloged sources in our sample are obscured agn .
if so , our mips sample in the isaac field should contain 25 x - ray weakly or non - detected fiore - selected agn .
of the 36/42 x - ray non - cataloged fiore sources in our sample ( which do not lie too close to a known x - ray source to test for faint emission ) , however , only 10 ( 28% ) are x - ray weakly - detected ( with properties indicative of heavy obscuration ) .
the remaining 26 sources show no evidence for x - ray emission , and have properties that may also be consistent with star - formation .
we therefore only consider the contribution from the 10 weakly - detected sources . of these 10 sources , 3
are weakly - detected power - law galaxies . correcting for the isaac field of view ( and the 6 sources for which we could not test for weak x - ray emission ) results in an additional contribution of 12.0 agn , or 14% . by combining the contribution of reliable power - law ,
color - selected , radio / infrared , and ir - excess agn candidates , we therefore estimate that _ spitzer_selected samples increase the known x - ray selected agn population by @xmath218 , down to a 24 flux density of 80 @xmath1jy .
in addition , the radio / infrared selection implies a @xmath219 contribution from radio - excess and radio - quiet agn yet to be identified .
the number of agn with 24 flux densities @xmath220 80 @xmath1jy is therefore @xmath5 larger than that of current samples detected both in the x - ray and at 24 in the deepest x - ray fields .
we plot in figure 14a the fraction of the mips sample comprised of mir - dominated agn as a function of 24 flux density .
we define as mir - dominated those agn that ( 1 ) meet the plg criteria , provided that they are also x - ray weakly - detected or have extremely red slopes of @xmath204 ( these criteria were used above to place a conservative lower limit on the contribution from plgs ) , ( 2 ) meet the irac color - color cuts of lacy et al .
( 2004 ) or stern et al .
( 2005 ) and lie outside the @xmath141 star - forming contours , ( 3 ) meet the polletta et al .
( 2008 ) criteria , ( 4 ) meet the fiore et al .
( 2008 ) criteria and are x - ray cataloged or weakly - detected , or ( 5 ) meet the daddi et al . ( 2007 ) criteria and are x - ray cataloged or weakly - detected . as above , we define as x - ray agn those sources with cataloged x - ray counterparts whose total x - ray luminosities ( when redshifts are available ) exceed @xmath221 ergs s@xmath10 . because the plg , daddi et al .
( 2007 ) , and fiore et al .
( 2008 ) selection criteria all identify agn candidates with mean redshifts of @xmath128 , the vast majority of the ir - selected agn lie at high redshift regardless of their flux density . as shown by brand et al .
( 2006 ) , the fraction of mir sources dominated by an agn drops with decreasing flux density .
we verify this trend , showing excellent agreement with the brand et al .
results ( see figure 13a ) , and confirm that it continues down to @xmath222 @xmath1jy . at lower flux densities , however , the fraction of mir - dominated x - ray agn begins to decrease at a much lower rate , and that of all mir - dominated agn plateaus at a value of @xmath223% .
this indicates that a non - negligible fraction of faint mir sources are not powered primarily by star - formation , but by their central engines .
the significant number of faint 24 sources that are agn - dominated in the mir is consistent with the finding of 6/19 such objects in the spectroscopy of faint 24 sources by rigby et al .
( 2008 ) . at these low flux densities
, it is also evident from figure 14a that the majority of the mir - dominated agn are not detected in the x - ray .
for instance , at 80 @xmath1jy , the total number of mir - dominated agn outnumbers that of mir - dominated x - ray agn by a factor of @xmath224 . using only the x - ray emission as a probe of agn activity will therefore result in a serious underestimation of the agn contribution to the mir flux density .
finally , we plot in figure 14b the fraction of all agn in the mips sample ( not only those whose mir flux is dominated by the agn ) , as a function of mips 24 flux density . to account for the contribution of the agn amongst the 14 x - ray sources lacking redshifts , we randomly add 11 , or 77% ( the agn fraction of the x - ray sample ) , of these 14 x - ray sources to the x - ray agn sample .
we further supplement the agn sample by including the assumed contribution from radio - excess and radio - quiet agn , 17% , or 14 sources . because these sources have not been individually identified in the cdf - s ,
we draw their flux densities randomly from the range of observed flux densities in the cdf - n ( @xmath225 @xmath1jy ) .
the x - ray non - detected agn still comprise the majority of all agn at the lowest 24 flux densities , and their contribution at larger flux densities is significant , raising the agn fraction to @xmath226 at all 24 flux densities .
as was seen for the mir - dominated agn , the total fraction of agn rises with increasing mir flux density , reaching a value of 37% at @xmath227 @xmath1jy , 15% higher than that of mir - dominated agn of the same flux density .
the x - ray agn fractions we find are somewhat higher than that of treister et al .
while the cause of this offset is not entirely clear , it is likely to be due at least in part to cosmic variance , as their 24 sample was drawn from goods - n , and ours from goods - s . once again , the vast majority of x - ray non - detected agn at all flux densities lie at @xmath128 , the mean redshift of the sources selected via the plg , daddi et al .
( 2007 ) , and fiore et al .
( 2008 ) criteria .
only the radio - selected agn samples lie at systematically lower redshifts of @xmath228 .
infrared selection of agn is a powerful technique . using new accurate star - forming and agn templates along with a flux - limited mips - selected sample drawn from the goods - s field , we critically review three mir selection criteria : ( 1 ) the irac color cuts of lacy et al .
( 2004 ) and stern et al .
( 2005 ) , ( 2 ) the power - law galaxy ( plg ) selection technique of alonso - herrero et al .
( 2006 ) and donley et al .
( 2007 ) , and ( 3 ) the ir - excess selection criteria of daddi et al .
( 2007a , b ) , dey et al .
( 2008 ) , fiore et al .
( 2008 ) , and polletta et al .
( 2008 ) . from this analysis
, we then quantify the contribution of _
spitzer_-selected agn to the x - ray selected agn population .
the main conclusions of this paper are as follows : * the majority of non - power - law irac color - selected agn candidates have ir colors consistent with those of redshift - appropriate star - forming templates . in comparison , the majority of plg agn candidates lie outside of the star - forming contours .
plg selection recovers the majority of high - quality agn candidates . * the reliability of agn irac color - color selection improves with increasing flux as high - redshift star - forming galaxies fall out of the sample .
nevertheless , the fraction of potential star - forming contaminants is still high ( @xmath3 ) at the highest fluxes probed by our survey ( @xmath229 @xmath1jy ) . *
a comparison of the 24 to 3.6 colors of the x - ray non - detected plgs to those of agn and star - forming templates suggests that the x - ray non - detected plgs , like their x - ray
detected counterparts , have more hot dust emission than can be explained by star - formation alone .
* an analysis of the daddi et al .
( 2007 ) ir - excess sources in our mips sample indicates that while these sources may be compton - thick agn , it is also possible that they are low - luminosity , compton - thin agn and/or luminous , highly - reddened star - forming galaxies . * an x - ray stacking analysis of the sources selected via the fiore et al .
( 2008 ) criteria indicate that @xmath230 of the sources are consistent with being obscured agn , and that the remaining 58% may be star - forming galaxies . * adding secure _ spitzer_-selected power - law , color - color , radio / ir , and ir - excess agn candidates to the deepest x - ray samples directly increases the number of known agn by @xmath218 , and implies a total increase of @xmath231% .
this fraction excludes the full contributions from the daddi et al . and
fiore et al .
agn candidates , whose nature is still uncertain . *
the fraction of mir sources dominated by an agn decreases with decreasing flux density , but only down to a 24 flux density of @xmath222 @xmath1jy . below this limit , the agn fraction levels out at @xmath7 .
this indicates that a non - negligible fraction of faint 24 sources are primarily powered not by star - formation , but by the central engine .
in addition , the majority of agn at low 24 flux densities are missed in the x - ray , indicating that x - ray emission alone can not be used to identify agn , especially amongst faint ir samples .
we thank m. polletta for providing templates , m. dickinson , and d. alexander for providing the list of daddi et al .
( 2007a , b ) sources .
we also thank d. alexander , f. fiore , m. lacy , d. stern , and the anonymous referee for discussions and comments that improved the paper .
finally , we thank caltech / jpl for support through contract 1255094 to the university of arizona .
g . acknowledges support from the spanish programa nacional de astronoma y astrofsica under grant aya 200602358 and aya 200615698c0202 , and from the ramn y cajal program financed by the spanish government and the european union .
alexander , d. m. , vignali , c. , bauer , f. e. , brandt , w. n. , hornschemeier , a. e. , garmire , g. p. , & schneider , d. p. 2002
, , 123 , 1149 alexander , d. m. , et al . 2003 , , 126 , 539 alexander , d. m. , et al . 2008 , arxiv e - prints , 803 , arxiv:0803.0636 alonso - herrero , a. , et al . 2006 , , 640 , 167 arnouts , s. , vandame , b. , benoist , c. , groenewegen , m. a. t. , da costa , l. , schirmer , m. , mignani , r. p. , & slijkhuis , r. 2002 , vizier online data catalog , 337 , 90740 barmby , p. , et al . 2006 , , 642 , 126 bevington , p.
r. , & robinson , d. k. 2003 , data reduction and error analysis for the physical sciences , 3rd ed .
, by philip r. bevington , and keith d. robinson .
boston , ma : mcgraw - hill , isbn 0 - 07 - 247227 - 8 , 2003 .
bolzonella , m. , miralles , j .-
m . , & pell , r. 2000 , , 363 , 476 bouchet , p. , lequeux , j. , maurice , e. , prevot , l. , & prevot - burnichon , m. l. 1985 , , 149 , 330 brand , k. , et al.2006 , , 644 , 143 brusa , m. , et al.2005 , , 432 , 69 buat , v. , et al.2005 , , 619 , l51 calzetti , d. , armus , l. , bohlin , r. c. , kinney , a. l. , koornneef , j. , & storchi - bergmann , t. 2000 , , 533 , 682 calzetti , d. , et al .
2007 , , 666 , 870 caputi , k. i. , et al .
2006 , , 637 , 727 cardamone , c. n. , et al .
2008 , , 680 , 130 daddi , e. , et al.2007a , , 670 , 156 daddi , e. , et al.2007b , , 670 , 173 dale , d. a. , & helou , g. 2002 , , 576 , 159 desai , v. , et al .
2008 , , 679 , 1204 dey , a. , et al.2008 , , 677 , 943 doherty , m. , bunker , a. j. , ellis , r. s. , & mccarthy , p. j. 2005 , , 361 , 525 donley , j. l. , rieke , g. h. , rigby , j. r. , & prez - gonzlez , p. g. 2005 , , 634 , 169 donley , j. l. , rieke , g. h. , prez - gonzlez , p. g. , rigby , j. r. , & alonso - herrero , a. 2007 , , 660 , 167 elvis , m. , et al.1994 , , 95 , 1 fazio , g. g. , et al .
2004 , , 154 , 10 ferrarese , l. , & merritt , d. 2000 , , 539 , l9 fiore , f. , et al.2008 , , 672 , 94 franceschini , a. , et al .
2003 , , 343 , 1181 gebhardt , k. , et al .
2000 , , 539 , l13 genzel , r. , et al .
1998 , , 498 , 579 giacconi , r. , et al .
2002 , , 139 , 369 giavalisco , m. , et al .
2004 , , 600 , l93 gilli , r. , comastri , a. , & hasinger , g. 2007 , , 463 , 79 goldader , j. d. , meurer , g. , heckman , t. m. , seibert , m. , sanders , d. b. , calzetti , d. , & steidel , c. c. 2002 , , 568 , 651 grazian , a. , et al .
2006 , , 449 , 951 hopkins , p. f. , hernquist , l. , cox , t. j. , di matteo , t. , robertson , b. , & springel , v. 2006 , , 163 , 1 ivezi , . , et al . 2002 , , 124 , 2364 iwasawa , k. , matt , g. , guainazzi , m. , & fabian , a. c. 2001 , , 326 , 894 kennicutt , r. c. , jr .
1998 , , 498 , 541 klaas , u. , et al .
2001 , , 379 , 823 klesman , a. , & sarajedini , v. 2007 , , 665 , 225 lacy , m. , et al.2004 , , 154 , 166 lacy , m. , et al.2005 , , 161 , 41 lacy , m. , petric , a. o. , sajina , a. , canalizo , g. , storrie - lombardi , l. j. , armus , l. , fadda , d. , & marleau , f. r. 2007 , , 133 , 186 le fvre , o. , et al .
2004 , , 428 , 1043 lonsdale , c. j. , et al .
2003 , , 115 , 897 martnez - sansigre , a. , rawlings , s. , lacy , m. , fadda , d. , jarvis , m. j. , marleau , f. r. , simpson , c. , & willott , c. j. 2006 , , 370 , 1479 marzke , r. , et al . 1999 ,
photometric redshifts and the detection of high redshift galaxies , 191 , 148 mcgregor , p. j. 1987 , , 312 , 195 mignoli , m. , et al .
2005 , , 437 , 883 neugebauer , g. , oke , j. b. , becklin , e. e. , & matthews , k. 1979 , , 230 , 79 prez - gonzlez , p. g. , et al .
2005 , , 630 , 82 prez - gonzlez et al .
2008 , apj , in press persic , m. , & rephaeli , y. 2002 , , 382 , 843 persic , m. , rephaeli , y. , braito , v. , cappi , m. , della ceca , r. , franceschini , a. , & gruber , d. e. 2004 , , 419 , 849 polletta , m. , et al .
2007 , , 663 , 81 polletta , m. , weedman , d. , hnig , s. , lonsdale , c. j. , smith , h. e. , & houck , j. 2008 , , 675 , 960 pozzi , f. , et al.2007 , , 468 , 603 prevot , m. l. , lequeux , j. , prevot , l. , maurice , e. , & rocca - volmerange , b. 1984 , , 132 , 389 ptak , a. , heckman , t. , levenson , n. a. , weaver , k. , & strickland , d. 2003 , , 592 , 782 ranalli , p. , comastri , a. , & setti , g. 2003 , , 399 , 39 richards , e. a. 2000 , , 533 , 611 richards , g. t. , et al .
2003 , , 126 , 1131 rieke , g. h. , et al .
2004 , , 154 , 25 rieke , g. h. , et al .
2008 , in prep .
rigby , j. r. , et al .
2004 , , 154 , 160 rigby , j. r. , et al . 2008 , , 675 , 262 sajina , a. , lacy , m. , & scott , d. 2005 , , 621 , 256 sanders , d. b. , mazzarella , j. m. , kim , d .- c . , surace , j. a. , & soifer , b. t. 2003 , , 126 , 1607 silva , l. , granato , g. l. , bressan , a. , & danese , l. 1998 , , 509 , 103 steffen , a. t. , brandt , w. n. , alexander , d. m. , gallagher , s. c. , & lehmer , b. d.2007 , , 667 , l25 stern , d. , et al.2005 , , 631 , 163 szokoly , g. p. , et al .
2004 , , 155 , 271 teng , s. h. , wilson , a. s. , veilleux , s. , young , a. j. , sanders , d. b. , & nagar , n. m. 2005 , , 633 , 664 tozzi , p. , et al.2006 , , 451 , 457 treister , e. , et al .
2006 , , 640 , 603 vandame , b. , et al .
2001 , arxiv astrophysics e - prints , arxiv : astro - ph/0102300 vanzella , e. , et al .
2006 , , 454 , 423 vijh , u. p. , witt , a. n. , & gordon , k. d. 2003 , , 587 , 533 white , n. e. , swank , j. h. , & holt , s. s. 1983 , , 270 , 711 wolf , c. , et al.2004 , , 421 , 913 yan , l. , et al .
2007 , , 658 , 778 lcccrrrr all & 713 & 649 & 249 & 0.012 & 0.15 & 11.3% & 4.6% + ir - normal & 448 & 424 & 187 & 0.012 & 0.12 & 7.2% & 2.8% + color - selected & 210 & 196 & 51 & 0.015 & 0.10 & 20.4% & 4.1% + power - law & 55 & 29 & 11 & -0.013 & 0.48 & 40.0% & 40.0% + lcccccccc total & 713 & 465 & 42 & 10 & 52 & 5 & 71 & 101 + x - rays & 109 & 76 & 3 & 3 & 10 & 4 & 11 & 15 + power & 25 ( 100% ) & 14 & 0 & 3 & 4 & 4 & 4 & 7 + color & 35 ( 79% ) & 28 & 3 & 0 & 5 & 0 & 6 & 7 + normal & 49 ( 65% ) & 34 & 0 & 0 & 1 & 0 & 1 & 1 + weak x - rays & 157 & 125 & 13 & 2 & 10 & 1 & 16 & 26 + power & 7 ( 100% ) & 5 & 1 & 1 & 3 & 1 & 4 & 4 + color & 44 ( 70% ) & 35 & 11 & 1 & 7 & 0 & 11 & 20 + normal & 106 ( 19% ) & 85 & 1 & 0 & 0 & 0 & 1 & 2 + no x - rays & 361 & 226 & 21 & 3 & 26 & 0 & 38 & 51 + power & 18 & 12 & 3 & 0 & 7 & 0 & 9 & 11 + color & 109 & 72 & 15 & 3 & 19 & 0 & 28 & 36 + normal & 234 & 142 & 3 & 0 & 0 & 0 & 1 & 4 + lccc lacy & 52% & 21% & 10% + stern & 40% & 18% & 8% + power - law & 93%,82% & 79%,61% & 71%,43% + + lacy & 29% & 12% & 5% + stern & 22% & 5% & 0% + power - law & 71%,68% & 61%,46% & 57%,39% + llrlrlr lacy & 21% & ( 19 ) & 26% & ( 10 ) & 42% & ( 8 ) + power - law ( lacy ) & 70% & ( 19 ) & 83% & ( 19 ) & 90% & ( 19 ) + stern & 19% & ( 5 ) & 25% & ( 3 ) & 40% & ( 2 ) + power - law ( stern ) & 67% & ( 16 ) & 72% & ( 13 ) & 85% & ( 11 ) + + lacy & 29% & ( 15 ) & 36% & ( 8 ) & 67% & ( 6 ) + power - law ( lacy ) & 81% & ( 17 ) & 89% & ( 16 ) & 88% & ( 15 ) + stern & 14% & ( 2 ) & 0% & ( 0 ) & 0% & ( 0 ) + power - law ( stern ) & 80% & ( 16 ) & 86% & ( 12 ) & 82% & ( 9 ) + ccccccc 80 & 52% & 40% & 93% , 82% + 100 & 54% & 44% & 96% , 84% + 150 & 53% & 43% & 94% , 78% + 200 & 60% & 63% & 100% , 76% + 300 & 53% & 25% & 100% , 72% + 400 & 52% & 50% & 100% , 66% + 500 & 56% & 50% & 100% , 71% + 80 & 29% & 21% & 72% , 68% + 100 & 29% & 24% & 76% , 72% + 150 & 25% & 21% & 73% , 63% + 200 & 32% & 27% & 92% , 76% + 300 & 31% & 25% & 90% , 72% + 400 & 39% & 50% & 88% , 66% + 500 & 50% & 50% & 85% , 71% + lll ell2 & 2 gyr old elliptical & ( 2 ) + ell5 & 5 gyr old elliptical & ( 2 ) + ell13 & 13 gyr old elliptical & ( 2 ) + s0 & spiral 0 & ( 1 ) + sa & spiral a & ( 1 ) + sb & spiral b & ( 1 ) + sc & spiral c & ( 1 ) + sd & spiral d & ( 1 ) + sdm & spiral dm & ( 1 ) + m82 & starburst & ( 1 ) + ngc 6090 & lirg / starburst & ( 1 ) + eso320-g030 & lirg / starburst & ( 3 ) + ngc 1614 & lirg / starburst & ( 3 ) + ngc 2639 & lirg / starburst & ( 3 ) + ngc 3256 & lirg / starburst & ( 3 ) + ngc 4194 & lirg / starburst & ( 3 ) + arp 220 & ulirg / starburst & ( 1 ) + arp 220 & ulirg / starburst & ( 3 ) + iras 12112 + 0305 & ulirg / starburst & ( 3 ) + iras 14348 - 1447 & ulirg / starburst & ( 3 ) + iras 17208 - 0014 & ulirg / starburst & ( 3 ) + iras 22491 - 1808 & ulirg / starburst & ( 3 ) + iras 22491 - 1808 & ulirg / starburst & ( 1 ) + mrk 231 & ulirg / seyfert 1 & ( 1 ) + sey1.8 & seyfert 1.8 & ( 1 ) + tqso1 & type 1 qso & ( 1 ) + sey2 & seyfert 2 & ( 1 ) + ngc 6240 & starburst / seyfert 2 & ( 1 ) + iras 19254 - 7245 & ulirg / seyfert 2 & ( 1 ) + iras 20551 - 4250 &
ulirg / buried agn & ( 1 ) + qso2 & type 2 qso & ( 1 ) +
we used two methods to determine photometric redshifts .
the first utilizes the extensive high - resolution template set of prez - gonzlez et al .
( 2008 ) , which was created by fitting stellar population synthesis and dust emission models to the @xmath2321500 galaxies in the cdf - n and cdf - s with secure spectroscopic redshifts .
when applied to all spectroscopically - detected irac - selected galaxies in the goods - n and goods - s , this template library returns photometric redshifts with @[email protected] for 88% of the sources , and @[email protected] for 96% , where @xmath51 ( prez - gonzlez et al .
because this method relies on star - forming templates , however , we do not expect it to provide equally reliable photometric redshifts for galaxies in which the stellar features are dominated by emission from an agn .
our second method is based on the chi - squared minimization routine hyperz ( bolzonella , miralles , & pell 2000 ) . with a suite of normal star - forming , lirg / ulirg , _ and _ agn templates , we can better account for the range of sources expected in our 24 selected sample , albeit with a smaller template library . to create this library , we started with a sample of 10 star - forming templates and 8 agn templates from silva et al .
( 1998 ) and polletta et al .
the polletta et al .
agn templates cover a range of intrinsic obscurations ( type 1 and type 2 ) and luminosities ( sey / qso ) .
we then supplemented this sample with 10 empirical star - forming lirg and ulirg templates .
the full template sample is listed in table 1 .
the lirg and ulirg templates , described in detail in rieke et al .
( 2008 , in prep . ) , significantly improve upon previous semi - empirical templates by constraining the seds between 1 and 6 with 2mass and irac photometry and by basing the seds from 5 to 35 on irs spectra .
using this template library , we ran hyperz on all robust optical mir data for the ir - normal and color - selected galaxies .
for the plgs , we removed all photometry longward of 3.6 ( irac channel 1 ) . including the full _ spitzer _
photometry for plgsprovided little or no additional constraint on potential redshifts , and limited severely the seds for which a good fit could be found .
we allowed @xmath47 to vary from 0 to 4 and @xmath233 to vary from 0 to 1.2 . while its shape is still relatively unconstrained
, a number of studies have suggested that the extinction curve of agn most closely resembles that of the smc ( e.g. richards et al .
we therefore assumed by default the smc extinction curve of bouchet et al .
( 1985 ) for the plgs and color - selected galaxies .
for the ir - normal galaxies , we assume the calzetti ( 2000 ) extinction curve , as this curve was modeled to represent the extinction properties of starburst galaxies . to prevent unrealistic redshift solutions
, we applied the redshift- and model - dependent absolute magnitude cuts of polletta et al .
finally , to increase the weight of the irac photometry , which solely defines the red slope of the stellar bump , we set the errors on the irac photometry to the measurement errors , as opposed to the total photometric errors .
this procedure is acceptable because many types of photometric error will have similar effects on the four irac bands , so we are making use of the overall internal consistency expected for the data . to improve the likelihood of obtaining a good fit for each source
, we then varied a number of these assumptions , and examined by eye the resulting fits to choose the most convincing redshift solution .
this visual inspection is an important characteristic of our work , and was made possible by the relatively small number of sources in our sample .
first , each plg and color - selected galaxy was fit by both the smc and calzetti extinction laws .
in most cases , the resulting redshift fits and solutions varied only slightly ( in which case we chose the smc - derived fit ) , but for some galaxies , one of the two extinction laws provided a clearly superior fit as determined by a visual inspection .
second , if the absolute magnitude of the resulting best - fit template did not meet the redshift - dependent absolute magnitude cut of polletta et al .
( 2007 ) , we allowed @xmath234 to vary between -23.7 to -17 for star - forming templates and -28.8 to -19 for agn templates ( e.g. polletta et al .
third , we allowed the irac errors to increase to their total estimated values by adding a 10% error to the flux ; in only 4 cases did this lead to a better fit . finally , for the plgs , we explored fits that did not include the 3.6 irac channel ; only 2 sources benefited from this change .
the final step in our redshift estimation was an independent review of the redshifts by two authors . by examining by eye the resulting redshift fits , we chose for each plg and color - selected galaxy the best hyperz redshift .
we then compared this to the redshift fit from the prez - gonzlez et al .
( 2008 ) technique , and replaced the former with the latter if it clearly provided a better fit .
for the ir - normal galaxies , we use the prez - gonzlez et al .
( 2008 ) redshifts by default , as these are optimized for normal galaxies .
we do , however , remove unconvincing redshift fits , and substitute a solid hyperz redshift if available .
for all sources , we only assign a photometric redshift if a convincing fit exists .
we rejected 76 sources with poor data or other problems that compromise our redshift determination .
the photometric redshifts classified as being of high quality are summarized in table 1 and shown in figure 1 .
the lowest redshift bin contains no plgsand 2 color - selected galaxies , both of which are detected in the x - ray , but with low observed x - ray luminosities of log @xmath77(ergs s@xmath10)@xmath236 and log @xmath75(ergs s@xmath10)@xmath237 , indicative of powerful starbursts or very low - luminosity agn .
the second redshift bin contains 3 plgs , 27 lacy - selected galaxies , and 4 stern - selected galaxies . of the lacy and stern - selected sources ,
only 41% and 25% are detected in the x - ray , respectively , compared to 100% of the plgs .
the average observed 0.5 - 8 kev x - ray luminosity of the x - ray detected color - selected galaxies , log @xmath77(ergs s@xmath10)@xmath239 , is consistent with agn activity , but is an order of magnitude less than that of the plgs in this redshift bin , log @xmath75(ergs s@xmath10)@xmath96 .
this is not surprising , as power - law selection preferentially identifies the most luminous agn where the emission from the central engine is able to overpower that of the host galaxy ( donley et al .
the agn that lie amongst the color - selected sample clearly represent a less luminous agn population than those selected via the plg criteria .
furthermore , of the 27 lacy - selected ( and 4 stern - selected ) galaxies , only 11 ( 2 ) lie outside the @xmath98 star - forming contours in their respective color - space , with the number dropping to 3 and 1 ( 2 and 0 ) at @xmath79 and @xmath141 .
this suggests that the majority of the color - selected sources in this redshift bin have infrared colors indicative of ir - normal galaxies . while all of the plgs lie outside even the @xmath141 contours in the lacy color - space , they lie inside the contours in stern - space , due primarily to the sdm template of polletta et al .
( 2007 ) , whose strong 3.3 aromatic feature passes through the 4.5 irac band at @xmath240 , causing the template to enter the stern agn selection region at @xmath241 . of the 2 plgs
whose colors overlap with this template , both have spectroscopic redshifts of @xmath242 , and would therefore be safe if we considered smaller redshift bins . removing this template
, however , would have no effect on the number of lacy or stern - selected sources that lie outside the @xmath98 contours .
we find a higher proportion of secure color - selected galaxies in the @xmath243 bin , with 10/12 of the lacy - selected sources and 6/13 of the stern - selected sources lying outside the @xmath98 star - forming contours .
this is not surprising , as it is at @xmath244 that the star - forming contours are best separated from the agn selection region .
if we extend our test to @xmath79 , however , the numbers drop significantly , to 5/12 and 1/13 .
the majority of the stern - selected galaxies in this redshift bin fill the lower - left corner of the agn selection region , the region populated by the stern - only sources whose lacy colors place them in the star - forming locus of color - space .
these star - forming galaxies are likely responsible for the fact that this is the only redshift bin in which the stern - selected sources outnumber the lacy - selected sources .
of the 5 plgs in this redshift bin , all lie outside the @xmath163 star - forming contours in both lacy and stern color - space , and while only 3 lie outside of the @xmath79 contours in stern color - space , all lie outside of the @xmath79 and @xmath141 colors in lacy color - space .
the number of color - selected galaxies rises significantly in the @xmath245 redshift bin .
the fraction of potential ir - selected agn , however , is by far the lowest at these redshifts , with only 6/38 lacy - selected sources and 3/31 stern - selected sources falling outside the @xmath98 star - forming contours .
in addition , only 3/38 lacy sources and 4/31 stern sources have x - ray counterparts , further suggesting that nearly all of the color - selected sources at this redshift are star - forming galaxies , and not agn .
in contrast , the 2 plgs found in this redshift bin fall outside of the @xmath98 contours in both the lacy and stern plots , and both are detected in the x - ray .
the largest number of lacy - selected sources , 77 , is found in the @xmath246 bin .
in addition , an extraordinarily high fraction , 50/77 , lie outside of the @xmath98 star - forming contours , a surprising fact given the x - ray detection fraction of only 9% .
have we discovered a significant population of @xmath128 obscured agn similar to those claimed by daddi et al .
( 2007 ) , or is there another explanation for this population ? the agn contours ( shown in green ) provide a better match to the ir colors of these color - selected galaxies than do the star - forming templates ( shown in blue ) . at this redshift , the primary difference between the irac regions of the star - forming and the low - luminosity agn templates is the strength of the co index , which is stronger in the star - forming galaxies than in the agn .
the mips selected sources therefore may have smaller co indices than those of our local templates .
one explanation for this offset is that an underlying agn continuum has diluted this feature . a lower co index , however ,
could also be attributed to evolution in the metallicity of the lirgs / ulirgs . at 1/3 solar metallicity
, the co index drops by 4 - 6 percentage points ( mcgregor 1987 ) , causing the contours to shift upwards in lacy color space by @xmath247 . while this lessens this offset between the colors of star - forming contours and the observed galaxies , it can not fully account for the observed discrepancy .
thus , if a change in the co index is invoked for the offset , it is likely that agn continua are also present .
however , other possible explanations exist .
if we incorporate into the contours a 10% error in ( 1+z ) ( recall that the measured @xmath56 for our photometric redshift fits was 0.15 and that 11% of the sources in our spectroscopic redshift sample have photometric errors @xmath248 ) the resulting contours are far - better matched to the ir - colors of the color - selected galaxies . of the lacy- and stern - selected galaxies ,
only 17% and 9% now lie outside of the @xmath98 contours , with the numbers dropping to 8% and 0% at @xmath79 and 4% and 0% at @xmath141 , respectively .
the templates with which the photometric redshifts were best fit can provide further insight into the sources in this redshift bin . of the 90 non - plgs in the @xmath2 bin , 75 ( 86% )
are best fit by a star - forming ulirg template , 9 ( 10% ) are best fit by a ulirg / hidden agn template , 1 ( 1% ) is best - fit by a type 2 agn template , and 1 ( 1% ) is best fit by a type 1 agn template .
in contrast , all of the plgs in this redshift range are best - fit by a type 2 agn template .
therefore , while we can not rule out the possibility that we have detected a sample of high - redshift , heavily obscured agn , the extremely low x - ray detection fraction of 9% , the much improved fit of the contours for which 10% errors in the photometric redshifts were included , and the overwhelming fraction of sources for which a purely star - forming ulirg provided the best fit to the sed suggest that it is more likely that the vast majority of color - selected galaxies in this redshift bin are star - forming galaxies , not agn .
the number of color - selected galaxies in the remaining two redshift bins is relatively low : 18 lacy - selected sources and 5 stern selected sources . of these , a large fraction ( 61% and 80% , respectively ) lie outside of the star - forming contours , which cover a comparatively small portion of the color - space . at @xmath79 and @xmath141 ,
the fractions drop to 39% and 11% ( lacy ) and 20% and 0% ( stern ) . at these redshifts , the x - ray detection fractions of both the color - selected sources and the plgs are low : 17% for the lacy sources , 40% for the stern sources , and 23% for the plgs . at @xmath250 , an unobscured agn with @xmath251
requires a rest - frame 0.5 - 8 kev luminosity of log @xmath77(ergs s@xmath10)@xmath252 to meet the flux limit within 1@xmath253 of the cdf - s aimpoint .
at @xmath254 , the required value rises to log @xmath75(ergs s@xmath10)@xmath239 , suggesting that if these sources are agn , they must have low luminosities , or high obscuring columns . not all sources in our sample have redshift estimates .
the redshift completeness for the lacy and stern color - selected samples is high ( 93% and 90% , respectively ) , but we have high - quality redshifts for only 49% of the plgs .
the difficulty in fitting redshifts to these sources stems largely from their faint fluxes : the power - law ( color - selected ) galaxies without redshifts are significantly fainter , @xmath256 ( @xmath257 ) , than those with redshift estimates , @xmath258 ( @xmath259 ) , suggesting that these sources may preferentially lie at high redshift .
the last panels of figures 5 and 6 show the irac colors of sources without redshift estimates .
overplotted are the star - forming contours for @xmath83 . of the sources without redshifts , all but 1 lie inside the lacy selection region and all but 9 lie inside the stern selection region .
this is not surprising , as sources with non - stellar continua are the hardest sources to fit .
of the 13 lacy - selected sources , 5 lie outside the @xmath98 star - forming contours , and 2 have x - ray counterparts . of the 26 plgs , 11 lie outside the @xmath98 star - forming contours , and 6
are x - ray detected .
there is a noticeable concentration of x - ray non - detected plgs towards the red end of the power - law locus .
these sources are discussed in more detail in 6 . | infrared selection is a potentially powerful way to identify heavily obscured agn missed in even the deepest x - ray surveys . using a 24 -selected sample in goods - s
, we test the reliability and completeness of three infrared agn selection methods : ( 1 ) irac color - color selection , ( 2 ) irac power - law selection , and ( 3 ) ir - excess selection ; we also evaluate a number of infrared excess approaches .
we find that the vast majority of non - power - law irac color - selected agn candidates in goods - s have colors consistent with those of star - forming galaxies .
contamination by star - forming galaxies is most prevalent at low 24 flux densities ( @xmath0 @xmath1jy ) and high redshifts ( @xmath2 ) , but the fraction of potential contaminants is still high ( @xmath3 ) at 500 @xmath1jy , the highest flux density probed reliably by our survey .
agn candidates selected via a simple , physically - motivated power - law criterion ( plgs ) , however , appear to be reliable .
we confirm that the infrared excess methods successfully identify a number of agn , but we also find that such samples may be significantly contaminated by star - forming galaxies . adding only the secure _ spitzer_-selected plg , color - selected , ir - excess , and radio / ir - selected agn candidates to the deepest x - ray selected agn samples directly increases the number of known x - ray agn ( 84 ) by @xmath4 , and implies an increase to the number of 24 -detected agn of @xmath5 .
finally , we show that the fraction of mir sources dominated by an agn decreases with decreasing mir flux density , but only down to @xmath6 @xmath1jy . below this limit , the agn fraction levels out , indicating that a non - negligible fraction ( @xmath7 ) of faint 24 sources ( the majority of which are missed in the x - ray ) are powered not by star formation , but by the central engine .
the fraction of all agn ( regardless of their mir properties ) exceeds 15% at all 24 flux densities . |
in contrast to simple metals such as pb , phase diagrams of unconventional superconductors often show a multitude of electronic phases of matter . in particular , magnetic order is ubiquitous in cuprate oxides and heavy fermion superconductor ( hfsc ) phase diagrams alike , as well as in some ruthenates and cobaltates .
the proximity , competition , or coexistence of two distinct types of electronic order at the fermi surface raised speculations that their driving mechanisms could be closely related.@xcite@xmath5 demler exotic superconductivity in heavy fermion materials @xcite usually appears near the quantum critical point @xcite where the magnetic ordering temperature is tuned to t = 0 by a variety of external parameters .
for example annealing , composition or magnetic field.trovarelli however , the most frequently used tuning parameter in hfsc is pressure.@xcite this is in variance with the cuprate family where the interplay of superconductivity and magnetism is balanced by adjusting the in plane charge density of the cuo@xmath6 layers .
the discovery of pressure - induced superconductivity in cein@xmath7@xcite stimulated exploratory synthesis of aucu@xmath7 superstructures by fisk , where the magnetic entropy might be further suppressed by crystallographic arguments.@xcite indeed , superconductivity in cerhin@xmath2,@xcite and ceirin@xmath2,ir was soon discovered , as well as coexistence of heavy fermion superconductivity and magnetism in cerh@xmath0ir@[email protected]@xmath5@xcite cecoin@xmath2,@xcite the _ primo _ compound in the 115 family of hfsc , has also been recently hole doped to an antiferromagnetic ground state by cd substitution on the in ( 1 ) site.pham despite a few examples of structurally tuned superconductivity in hfsc , the lack of a predictive theory suggests that there is still no clear understanding of how the delicate interplay of various degrees of freedom in these materials stabilizes superconducting , or a magnetic ground state .
the large quasiparticle mass enhancement in cecoin@xmath2 is reflected in two large nearly cylindrical pieces of the fermi surface and smaller 3d hole pockets.@xcite due to the pronounced @xmath8 - space inhomogeneity , hybridization of the @xmath9 electrons of co with the @xmath10 electrons of in results in a small density of states at the fermi energy , implying partially quasi - 2d electronic structure.@xcite by exploring a well defined non - hybridizing local moment such as nd , we were able to continuously tune the coupling in the lattice and consequently quasiparticle mass enhancement between 4f ions and conduction electrons .
as a result , we have obtained a rich phase diagram .
the magnetic ground state in nd@xmath0ce@xmath1coin@xmath2 smoothly evolves from local moment magnetism ( lmm ) on the nd - rich side to hfsc on the ce - rich side .
superconductivity coexists with other forms of electronic order , most likely magnetic in nature , in samples for nd concentrations between @xmath11 and @xmath12 .
we demonstrate that the delicate balance between coexisting ordered states can be smoothly tuned by magnetic fields near the magnetic - superconducting boundary .
single crystals of nd@xmath0cexcoin@xmath2 were grown from an excess in flux . magnetic susceptibility ,
specific heat and resistivity measurements were performed in a quantum design mpms xl 5 and ppms - 9 instruments respectively .
single crystals were thoroughly ground to a fine powder for structural measurements .
high resolution synchrotron powder x - ray patterns were taken at beamline x7a of national synchrotron light source at brookhaven national laboratory .
monochromatic synchrotron x - ray and gas - proportional position - sensitive detector were used to measure the powder diffraction data .
rietveld refinements were performed using gsas.@xcite the samples were manually aligned to measure the magnetization , heat capacity , or resistivity for fields applied along the appropriate axis .
m / h polycrystalline averages were calculated as @xmath13 $ ] and were used to obtain the high temperature effective moments .
temperature dependent magnetic susceptibility was used to estimate the relative ratio of ce / nd ions . at high temperatures ranging from 150 k to 350 k , h /
m curves were fit with curie - weiss ( cw ) law , @xmath14 where c is the curie constant and @xmath15 is the weiss temperature .
electrical contacts for in - plane resistivity measurements were made with epotek - h20e silver epoxy on thin plate - like crystals whose dimensions were measured by optical microscope with 10@xmath16 m resolution . the samples were previously etched in diluted hcl for several hours and thoroughly rinsed in ethanol in order to remove excess in .
since both ndcoin@xmath2 and cecoin@xmath2 are grown using identical temperature profiles and an identical ratio of starting materials nd ( ce ) , co and in,@xcite@xmath5@xcite one would expect a smooth change of the lattice parameters in the nd@xmath0ce@xmath1coin@xmath2 alloy series .
ce@xmath17coin@xmath2 . for all x ,
only the hocoga@xmath2 structure was detected , in addition of several small peaks of pure in from the flux.,height=211 ] indeed , this is confirmed by high resolution structural measurements which showed that the samples crystallized in tetragonal hocoga@xmath2 structure without any additional peaks introduced by ce alloying ( fig .
1 ) . selected regions of the powder x - ray spectra showed monotonic evolution and uniform sharpness independent of @xmath3 for both [ 111 ] and [ 003 ] peaks .
this implies that ce uniformly substitutes nd with the increase of x and that our samples are indeed alloys rather than a mixture of intergrown compounds ndcoin@xmath2 and cecoin@xmath2 .
the lattice parameters increase smoothly with ce substitution in accordance with vegard s law ( fig .
both [ 111 ] and [ 003 ] peaks shift uniformly with ce substitution .
( b ) tetragonal lattice parameters @xmath18 ( red symbols ) and @xmath19 ( blue symbols),height=384 ] magnetic susceptibility measurements did not detect magnetically ordered second phases , such as ndin@xmath7 or cein@xmath7 ( fig .
the cubic compounds cein@xmath7 and ndin@xmath7 order antiferromagnetically at 10 k and 6.3 k respectively.@xcite@xmath5@xcite magnetic ordering in ndcoin@xmath2 is depressed smoothly with the increase of ce , and we were able to follow the characteristic peak in m / h ( signature of the onset of the magnetic order ) down to lowest temperature of our magnetic measurement , t = 1.8 k for x = 0.6 .
curie - weiss analysis of polycrystalline magnetic susceptibility average at high temperatures showed at most a 4% deviation from the nominal ratio of ce@xmath20 and nd@xmath20 moments ( table 1 ) . as expected ,
the highest uncertainty in the nominal concentration of ce ( x ) is in the middle of the alloy series , for the highest chemical disorder .
combined together , these results demonstrate that ce uniformly substitutes nd in the entire doping range of nd@xmath0ce@xmath1coin@xmath2 , with a maximum @xmath21 = 0.04 . [
cols="<,<,<,<",options="header " , ]
m / h data of nd@xmath0ce@xmath1coin@xmath2 alloy series show substantial anisotropy ( fig .
3 ) at low temperatures and signature of magnetic order at t@xmath22 = 9 k. for field applied along the @xmath23-axis , @xmath24 the magnetic transition is rather broad . on the other hand for @xmath25
the transition is relatively sharp implying complex magnetic order .
ce@xmath1coin@xmath2 alloy series in 1koe for a)h @xmath26 @xmath23 and b)h @xmath26 @xmath27 axis .
magnetic susceptibility shows decrease with ce substitution , with easy axis along the crystalline @xmath27 axis .
the characteristic signature of antiferromagnetic order is observed down to @xmath28 above @xmath29 for @xmath30(t ) .
the smooth evolution of @xmath30(t ) with x could indicate that magnetic fluctuation spectrum along the @xmath27 axis is more relevant for tuning of the ground state .
, height=384 ] the thermodynamic properties of nd@xmath0ce@xmath1coin@xmath2 are shown in fig .
the magnetic ordering temperature in ndcoin@xmath2 is smoothly depressed from t@xmath22 = 9 k with increased ce concentration to 2.0 k by @xmath28 where ( c - c@xmath31)/t shows a broad peak ( fig .
interestingly , the magnetic ordering transition becomes sharper as @xmath3 is further tuned towards the superconducting boundary for @xmath32 . at @xmath33 superconductivity emerges .
with increasing @xmath3 , the t@xmath34 increases to the bulk t@xmath34 = 2.3 k of cecoin@xmath2 .
the magnetic entropy released upon emergence of lmm ( fig .
4a inset ) scales with nd concentration up to @xmath35 , implying that ce ions do not play a direct part in the formation of the lmm ground state . in the hfsc state for @xmath36 ,
the magnetic entropy at 5k is essentially invariant to changes of ce / nd ratio .
however , a hallmark of heavy fermion magnetism is observed for @xmath37 . this can be seen on the example of @xmath28
where only approximately @xmath38 rln2 is released below the magnetic ordering transition at 2 k. in the superconducting region at the ce -rich side ( fig .
4b ) additional thermodynamic anomalies @xmath39 , @xmath40 and @xmath41 emerge in ( c - c@xmath31)/t below the superconducting transition t@xmath34 for @xmath42 , @xmath43 , but above t@xmath34 for @xmath44 .
specific heat of lacoin@xmath2 was taken as the estimate of the lattice specific heat c@xmath31 .
magnetic entropy ( inset ) obtained from integral ( c - c@xmath31)/t in the same temperature range .
( b ) : thermodynamic and transport properties in the superconducting state , measured on the same sample for each ce concentration @xmath3 minus the lattice .
the lattice resistivity was estimated by taking values of electrical resistivity of lacoin@xmath45 with decrease of the ce / nd ratio , the superconducting temperature t@xmath34 is suppressed .
the additional thermodynamic anomaly increases in temperature ( a @xmath46 b @xmath46 c ) , as seen in the heat capacity data ( full symbols ) .
in - plane resistivity data ( open symbols ) are shown to identify superconducting transition.,height=384 ] by examining electronic transport we obtain further evidence for the presence of distinct types of electronic matter in the phase diagram . temperature dependent electrical resistivities normalized to their value at 300k ( fig .
5 a , b ) show rather small loss of the spin disorder scattering below neel temperature in ndcoin@xmath2 .
this is consistent with the lmm - type of order in rare earth intermetallic compounds where local moments do not become part of fermi surface upon cooling .
by @xmath47 the electronic scattering strongly increases due to a single ion kondo - type interaction and a logarithmic contribution of hybridizing ce@xmath20 ions submerged in the fermi sea .
eventually for @xmath37 a broad coherence peak develops in the lattice .
this marks the emergence of an additional kondo energy scale arising from the coherence in the heavy fermion lattice .
as expected , the coherence temperature t@xmath48 increases with the increase of ce@xmath20 ions .
looking from the ce - rich side , the onset of the superconducting transition in cecoin@xmath2 is depressed to 0.9 k and magnetic scattering increases by @xmath44 ( fig .
4b ) . ce@xmath1coin@xmath2 with current parallel to the [ 100 ] .
( a ) : resistivity curves normalized to their value at 300 k. ( b ) : magnetic contribution to the electrical resistivity ( @xmath49 - @xmath50 ) .
the lattice resistivity was estimated by taking values of electrical resistivity for lacoin@xmath2.,height=384 ]
heat capacity in the hfsc state for @xmath51 reveals two thermodynamic anomalies . only one ( c - c@xmath31)/t discontinuity , though , corresponds to the superconducting transition ( fig .
4b ) . one possible explanation for this would involve sample inhomogeneity and a distribution of the doping concentration .
this however is very unlikely due to the low residual resistivity , the clean high resolution synchrotron powder x - ray diffraction pattern with uniform sharpness of peaks for the whole range of @xmath52 and the compositional dependence of lattice parameters @xmath23 and @xmath53 in accordance with vegard s law ( fig.1 ) .
furthermore , one would expect the highest degree of metallurgical disorder around the middle of the doping range , for @xmath47 , and not near the ce end . finally , in superconducting materials with _
metallic _ type of bonding , metallurgical inhomogeneity would shift the bulk t@xmath34 of the _ whole _ sample , not only a fraction of the sample .
for example , in cecu@xmath54si@xmath6 this is indeed seen in high resolution studies of the local structure.@xcite moreover , as seen in the heat capacity data for @xmath55 , the two thermodynamic anomalies are comparable in size . thus secondary phases would have been easily detected in the analysis of powder x - ray spectra .
furthermore , as will be evident in fig .
7 , the smooth evolution of @xmath56 and @xmath57 argues against real space inhomogeneity .
our results imply that magnetic and superconducting phases coexist in the phase diagram , similarly to the situation in cd - doping of cecoin@xmath2.@xcite , for h @xmath26 @xmath27 - axis ( a ) and h @xmath26 @xmath23 - axis ( b ) . for each @xmath3
, c / t and @xmath49 data were taken on the same sample . the crossover from positive to negative magnetoresistance for h @xmath26 @xmath23 - axis suggests different nature of critical magnetic fluctuations below h@xmath34 = 50 koe and above h@xmath34 .
superconducting transition t@xmath34 corresponds to thermodynamic anomaly at higher temperature ( below h@xmath34 ) and at lower temperature ( above h@xmath34 ) for h @xmath26 @xmath23 - axis . a small feature in resistivity around 1 k for h = 70 koe
could indicate opening of the partial gap at the fermi surface in higher fields.,height=384 ] an alternative explanation involves two superconducting energy scales on different parts of the fermi surface and negligible interband scattering .
this could explain low temperature anomalies @xmath58 for @xmath42 and @xmath43 since zero resistivity is achieved at the high temperature transition and is maintained through the low temperature transition .
however , this scenario can not explain thermodynamic anomaly @xmath41 above the superconducting transition in nd@xmath59ce@xmath60coin@xmath2 ( fig .
we speculate that low temperature discontinuities @xmath39 and @xmath40 , high temperature discontinuity @xmath41 as well as sharp transition in [ c - c@xmath31]/t ( fig .
4(a ) ) for x = 0.7 may be connected with some form of fermi surface instability appearing concurrently on a different part of the fermi surface . in what follows
we show that this instability as well as the boundary between superconductivity and magnetism can be tuned by a magnetic field , in addition to composition @xmath3 .
application of magnetic field reverses remarkable behavior seen with progressive decrease in the ce / nd ratio , tuning the boundary between the two ordered states .
ce@xmath1coin@xmath2 at h = 0 : superconducting t@xmath34 ( blue symbols ) , antiferromagnetic neel temperature ( red symbols ) , coherence temperature of the kondo lattice ( purple symbols ) .
orange symbols represent second broad thermodynamic anomaly in c / t ( below t@xmath34 ) where resistivity shows small upturn ( above t@xmath34 and for x = 0.7 at h = 0 ) .
the local moment antiferromagnetism of ndcoin@xmath2 smoothly changes to heavy fermion antiferromagnetic state around @xmath47 .
magnetism and superconductivity meet around @xmath62 .
the neel ordering temperature monotonically decreases for @xmath63 . for @xmath64 heat capacity anomaly
most likely corresponds to spin density wave type of magnetic order , deep in the superconducting state.,height=268 ] the magnetic field depresses both @xmath65 and @xmath40 in nd@xmath66ce@xmath67coin@xmath2 ( fig 6 ) .
the suppression is rather anisotropic . for a field applied along the @xmath27- axis superconducting anomaly in heat capacity and increasingly broad resistivity transition
are suppressed to 0.5 k in 40 koe .
the anomaly @xmath40 is observed only below 10 koe , merging into a single thermodynamic transition for higher fields above 0.4 k , the lowest temperature of our measurement . on the other hand , for a field applied along @xmath23- axis
both thermodynamic anomalies show quite different field dependence .
superconducting t@xmath34 at 30 koe - defined by the simultaneous onset of zero resistivity and the start of the heat capacity anomaly - is suppressed to 1.4 k. the anomaly @xmath40 is observed below the t@xmath34 , just as we observe in h = 50 koe .
in contrast to h = 0 and h = 30 koe , zero resistivity in h = 50 koe corresponds to midpoint , rather than the onset of the heat capacity anomaly . in h
= 70 koe , nd@xmath67ce@xmath66coin@xmath2 electronic matter becomes equivalent to the @xmath44 sample at h = 0 koe : the main heat capacity transition is now above superconductivity .
electronic scattering for field applied in tetragonal plane ( fig .
3b ) first increases and then decreases in the normal state as the magnetic field is tuned through h@xmath34 = 50 koe . the resistivity transition in h
@xmath68 h@xmath34 is sharper implying a magnetic field induced phase transition for h @xmath69 @xmath23 -axis .
the size of the superconducting anomaly in ( c - c@xmath31 ) decreases in field relative to the size of additional thermodynamic anomaly b below t@xmath34 at h = 0 for field applied along both crystalline axes .
there is no loss of spin disorder scattering at the main heat capacity transition t@xmath70 for both x = 0.1 sample in h = 70 koe ( fig .
6b ) and for x = 0.83 sample in h = 0 ( fig .
4b ) . instead , a very small upturn is observed in fig .
4(b ) ( x = 0.83 , h = 0 ) and in fig .
6(b ) ( x = 0.1 , h = 70 koe ) in resistivity ( denoted by orange diamonds ) , a signature of the partial gapping of the fermi surface .
this is reminiscent of an itinerant spin - density - wave type - transition based on fermi - surface nesting in a heavy - electron band observed in ce(ru@xmath71rh@xmath72)@xmath6si@xmath6.@xcite it appears that nd@xmath0ce@xmath1coin@xmath2 can support simultaneously multiple electronic ordering states .
coexistence of magnetism in superconductivity in ce 115 superconductors has been induced by pressure in cerhin@xmath2 or by composition in cerh@xmath0ir@xmath1in@xmath2 , ceco(in@xmath0cd@xmath1)@xmath2 and cerh@xmath0co@xmath1in@xmath2.@xcite@xmath5pagliuso@xmath5@xcite@xmath5@xcite the low temperature specific heat and resistivity data taken in nd@xmath67ce@xmath66coin@xmath2 at h = 0 and in nd@xmath59ce@xmath60coin@xmath2 at h = 70 koe imply that t@xmath34 and t@xmath70 involve instabilities at different parts of the fermi surface .
the balance between these states is tuned only by degree of hybridization in the kondo lattice ( ce / nd ratio ) and by magnetic field .
though the presence of additional magnetic transition made it difficult to estimate accurate values for the electronic heat capacity coefficient @xmath73 , by taking @xmath73 = [ c - c@xmath31(t)]/t above t@xmath34 , we observe that the large jump at ambient pressure in the specific heat of cecoin@xmath2 @xmath74c / c(t@xmath34 ) = 4.35 is reduced as more nd enters into the matrix . for x = 0.98
we observe @xmath74c / c(t@xmath34 ) = 3.79 and by x = 0.1 this ratio decreases to 1.76 .
nevertheless , our results may indicate general trend that increased nd concentration decreases electron - boson coupling strength.@xcite in the standard paradigm , the suppression of heavy fermion antiferromagnetic order leads to superconductivity around the quantum critical point.millis@xmath5@xcite one aspect of the reduced ce / nd ratio in the lattice is a negative pressure . using the bulk modulus of cecoin@xmath2 b = 76 gpa,@xcite
we estimate that a rigid shift of the lattice parameters for @xmath44 and @xmath43 corresponds to 0.03 gpa of applied pressure , given that @xmath74t@xmath34 = 0.9 k ( 50% ) between these two concentrations , this result implies that chemical pressure effects are less relevant than electronic tuning through increased hybridization.sidorov this result applies not only at the magnetic - superconducting boundary where the contribution is only up to 2% , but also for the whole range of @xmath3 since changes of the ground state in nd@xmath0ce@xmath1coin@xmath75 are far more dramatic than what was observed in pressure - induced changes of the ground state in cecoin@xmath2.@xcite it would be instructive to compare our results with the reported coexistence of superconductivity and magnetism in cerhin@xmath2 under high pressure and to field - induced magnetic order in the superconducting state of la@xmath76sr@xmath67cuo@xmath77.@xcite@xmath5@xcite@xmath5@xcite in both materials , as well as in cecoin@xmath2 , there are nodes in the superconducting gap .
delicate balance between antiferromagnetic and superconducting coupling near a quantum phase transition is smoothly tuned by magnetic field which generates field - induced vortices that suppress superconductivity and enhance magnetic correlations.@xcite@xmath5zhang in addition to this , in nd@xmath0ce@xmath1coin@xmath2 suppression of d - wave superconductivity is certainly influenced by disorder , as seen in large increase in residual resistivity ( fig.5).@xcite the evolution of superconducting t@xmath34(x ) with the increase of nd is similar to ce@xmath0la@xmath1coin@xmath2.@xcite@xmath5@xcite this is consistent with dominant potential scattering and insensitivity of t@xmath34 to magnetic configuration of the rare earth ion , as observed in ce@xmath0r@xmath1coin@xmath2.@xcite
in conclusion , we have reported on nd substitution in cecoin@xmath2 .
nd substitution results in the rich phase diagram , controlled apparently only by ce / nd ratio , i.e. by the level of 4f - conduction electron coupling .
ground state of this alloy series evolves from lmm to hfsc state via magnetically ordered heavy fermion ground state .
small concentration of nd ( 5% ) induces magnetic order deep in the superconducting state .
we have demonstrated magnetic field tuning of the delicate balance between superconducting and magnetic ground state at the magnetic - superconducting boundary .
we invite further investigation into possible microscopic coexistence of magnetic and superconducting order parameters by nmr and neutron scattering measurements .
possible existence of short range order ( spin glass ) between local moment magnetism and heavy fermion superconductor near the middle of alloy series would offer model electronic system with rich interplay of superconductivity and magnetism comparable to high - t@xmath34 cuprate oxide superconductors .
we thank r. prozorov , s. l. budko , p. c. canfield , johnpierre paglione and myron strongin for useful communication .
this work was carried out at the brookhaven national laboratory , which is operated for the u.s .
department of energy by brookhaven science associates ( de - ac02 - 98ch10886 ) .
this work was supported by the office of basic energy sciences of the u.s .
department of energy .
a. llobet , a. d. christianson , wei bao , j. s. gardner , i. p. swainson , j. w. lynn , j .-
mignot , k. prokes , p. g. pagliuso , n. o. moreno , j. l. sarrao , j. d. thompson , and a. h. lacerda , phys .
lett . * 95 * , 217002 ( 2005 ) f. steglich , b. buschinger , p. gegenwart , m. lohmann , r. helfrich , c. langhammer , p. hellmann , l. donnevert , s. thomas , a. link , c. geibel , m. lang , g. sparn and w assmus , j. phys . condens .
matter * 9 , * 9909 ( 1996 ) b. lake , h. m. ronnow , n. b. christensen , g. aeppli , k. lefmann , d. f. mcmorrow , p. vorderwisch , p. smeibidl , n. mangkorntong , t. sasagawa , m. nohara , h. takagi and t. e. mason , nature * 415 * , 299 ( 2002 ) | the nd@xmath0ce@xmath1coin@xmath2 alloys evolve from local moment magnetism ( x = 0 ) to heavy fermion superconductivity ( x = 1 ) .
magnetic order is observed over a broad range of @xmath3 . for a substantial range of @xmath3 ( 0.83 @xmath4 x @xmath4 0.95 ) in the temperature - composition phase diagram
we find that superconductivity may coexist with spin - density wave magnetic order at the fermi surface .
we show that a delicate balance betwen superconducting and magnetic instabilities can be reversibly tuned by both the ce / nd ratio and magnetic field , offering a new and unique model electronic system . |
@xmath2he atoms adsorbed on a graphite surface is known to be an ideal two - dimensional correlated spin-1/2 fermion system .
a solidified commensurate phase of @xmath2he atoms is stabilized at a @xmath3 density of the underlying layer of @xmath4he atoms due to the substrate potential corrugation and thus a triangular lattice of @xmath2he atoms is formed , which is a realization of a gapless quantum spin liquid ( qsl).@xcite theoretically , this @xmath3 phase of spin-1/2 @xmath2he atoms has been studied by using the triangular - lattice heisenberg model with the multiple - spin exchange interactions.@xcite importance of the density fluctuations has recently been pointed out as well.@xcite a finite amount of vacancies of @xmath2he atoms can be introduced into this @xmath3 phase in a stable manner , where the vacancies can hop from site to site of the triangular lattice via quantum - mechanical tunneling motions even at absolute zero temperature .
the presence of such vacancies , called the zero - point vacancies ( zpvs ) , was predicted a few decays ago.@xcite quite recently , the experimental evidence for the zpvs has been reported in the monolayer of @xmath2he adsorbed on a surface of graphite preplated by a solid monolayer of @xmath4he:@xcite i.e. , heat capacity measurements of the system show an anomalous coexistence of a magnetic round - peak near 1 mk and a broad peak at several tens mk that are associated with the zpvs doped into the commensurate mott - localized solid .
the zpvs are maintained up to the doping of almost 20% of the lattice sites , which we call the doped mott region of monolayer @xmath2he .
theoretically , fuseya and ogata @xcite have proposed the triangular - lattice @xmath0-@xmath1 model with four - spin ring - exchange interactions as an effective model for the doped mott region of the system and obtained its ground - state phase diagram .
the low - energy excitations of the model were also discussed .
they have thereby argued that there is a new - type anomalous quantum - liquid phase , characteristic of the `` spin - charge separation '' , which may be relevant with the anomalous features observed in the doped mott region of the monolayer @xmath2he adsorbed on a graphite surface .
motivated by such developments in the field , we study in this paper the triangular - lattice @xmath0-@xmath1 model with the multiple - spin exchange interactions further .
in particular , we directly calculate the spin and density excitation spectra and single - particle spectra as well as the temperature dependence of the specific heat and uniform magnetic susceptibility by using an exact - diagonalization technique on small clusters .
we thereby consider the anomalous properties observed in the doped mott region of the two - dimensional liquid @xmath2he adsorbed on a graphite surface .
we will thus demonstrate that the double - peak structure actually appears in the temperature dependence of the specific heat , which is quantitatively consistent with experiment ; the low - temperature sharp peak comes from the spin excitations and high - temperature broad peak comes from the density excitations .
the spectral weight for the calculated spin excitation spectra is concentrated on a very low - energy region that scales with the exchange interactions , while that of the density excitations extends over an entire band width that scales with the hopping parameter of the vacancy . the clear separation between spin and density excitations in their energy scales
is thus found .
the accumulation of the low - energy spectral weight of the spin excitations comes from the frustrated nature of the spin degrees of freedom of the system ; i.e. , the ferromagnetic two - spin interactions compete with the antiferromagnetic four - spin interactions on the geometrically frustrated triangular lattice .
the single - particle excitation spectra suggest that the vacancies behave like fermionic quasiparticles dressed by the spin excitations , with the enhanced effective mass consistent with experiment .
preliminary results of our work have been presented in ref .
@xcite .
this paper is organized as follows . in sec .
ii , we present our model and method of calculation . in sec .
iii , we present our results of calculations for the specific heat , magnetic susceptibility , spin and density excitation spectra , and single - particle excitation spectra .
we compare our results with experiment in sec .
we summarize our work in sec . v. +
the triangular - lattice @xmath0-@xmath1 model with the multiple - spin exchange interactions is defined by the hamiltonian @xmath5 where @xmath6 is the projected annihilation operator of a fermion ( @xmath2he atom ) at site @xmath7 and spin @xmath8 @xmath9 allowing no doubly occupied sites , @xmath10 is the spin-1/2 operator , and @xmath11 @xmath12 is the number operator .
the summation in the @xmath0-@xmath1 part of the model is taken over all the nearest - neighbor pairs @xmath13 on the triangular lattice . @xmath14 and @xmath15 are the four - spin and six - spin exchange operators defined as @xmath16 and @xmath17 , respectively , where @xmath18 with the pauli spin matrix @xmath19 .
the summation is taken over all the possible combinations of four nearest - neighbor sites @xmath20 for @xmath14 and over all the equilateral hexagons @xmath21 for @xmath15 . in this paper , we study the dynamical properties of the model under the introduction of vacancies in the 4/7 commensurate solid phase of @xmath2he , i.e. , removal of particles ( @xmath2he atoms ) or addition of zpvs . we thus define the filling @xmath22 of particles as @xmath23 where @xmath24 is the total number of particles and @xmath25 is the total number of lattice sites in the system ; in particular , @xmath26 is referred to as `` half filling '' , which corresponds to the 4/7 solid phase . the noninteracting band structure and fermi surface at half filling of the tight - binding model with the nearest - neighbor hopping parameter @xmath0
are shown in fig . 1 ; we find no nesting features in the fermi surface and no singularities in the density of states for @xmath27 .
the nearest - neighbor hopping parameter @xmath0 and two - spin and four - spin exchange interaction parameters @xmath1 and @xmath28 have been estimated as follows:@xcite @xmath29 mk , @xmath30 mk , and @xmath31 .
we note that , in this parameter region , the two - spin exchange term favors the ferromagnetic spin polarization ( @xmath32 ) but the four - spin exchange term gives the antiferromagnetic spin correlations between neighboring spins , and thus we have the situation where the strong frustration in the spin degrees of freedom of the system appears .
geometrical frustration also appears on the triangular lattice when the interaction between spins is antiferromagnetic .
we have examined the effects of the six - spin exchange interaction term on the ground state and excitation spectra and found that the effects are very small , in particular when doped with vacancies .
this is because the @xmath15 exchange interaction is easily cut by the presence of vacancies .
we will therefore present the results at @xmath33 in this paper , the model of which we refer to as the @xmath0-@xmath1-@xmath28 model . throughout the paper ,
we use @xmath34 as the unit of energy unless otherwise stated and we set @xmath35 .
+ we use the lanczos exact - diagonalization technique on small clusters to calculate the ground state and excitation spectra of the model .
in particular , we calculate the dynamical spin and density correlation functions defined , respectively , as @xmath36 are the ground - state wave function and energy , respectively . @xmath37 and @xmath38 are the fourier transforms of the spin and particle - number operators defined , respectively , as @xmath39 where @xmath40 is the position of the lattice site @xmath7
. we also calculate the single - particle excitation spectrum defined as @xmath41 with the particle removal spectrum @xmath42 , @xmath43 , @xmath44 , are also shown .
they are at @xmath45 with @xmath46 and @xmath47 , where @xmath48 are @xmath49 for @xmath50 , @xmath51 for @xmath52 , @xmath53 for @xmath54 , @xmath55 for @xmath56 , @xmath57 for @xmath58 , @xmath59 for @xmath60 , @xmath61 for @xmath62 , @xmath63 for @xmath64 , @xmath65 for @xmath66 , @xmath67 for @xmath68 , and @xmath69 for @xmath44 . in the following
, we will in particular examine the cluster with two vacancies , i.e. , @xmath70 .
to calculate the temperature @xmath71 dependence of the specific heat @xmath72 , magnetization @xmath73 under uniform magnetic field @xmath74 , and uniform magnetic susceptibility @xmath75 , the hamiltonian for a smaller - size cluster of 12 sites ( see fig .
2 ) is fully diagonalized to calculate the partition function.@xcite we add the zeeman term @xmath76 to the hamiltonian eq .
( 1 ) when we calculate the magnetic response of the system . in the following , we will in particular examine this cluster with one vacancy ( @xmath77 ) and two vacancies ( @xmath78 ) .
the ground - state phase diagram in the parameter space of the present model at @xmath33 has been obtained by fuseya and ogata,@xcite which we have also reproduced successfully . here , we briefly review their results .
the phases obtained are as follows : the region of phase separation ( phase - i ) , the region of fermi liquid with strong spin fluctuations ( phase - ii ) , the region of new - type anomalous quantum liquid ( phase - iii ) , and the region of ferromagnetism ( phase - iv ) , where we follow their notations of the phases .
they have put special emphasis on the phase - iii , which has been argued to be the region of `` spin - charge separation '' and may be relevant with the anomalous features of the doped mott region of the @xmath2he monolayer . in the following ,
we in particular examine the region of this new - type anomalous quantum liquid ( phase - iii ) using the parameter values @xmath79 and @xmath80 , which we compare with the results of other regions when necessary , i.e. , the region of ferromagnetism ( phase - iv ) using @xmath79 and @xmath81 and the region of fermi liquid ( phase - ii ) using @xmath79 and @xmath82 .
+ + the calculated results for the temperature dependence of the specific heat @xmath72 at @xmath78 and @xmath83 are shown in fig .
we find that there appears a double - peak structure in @xmath72 at @xmath80 ( corresponding to the new - type anomalous quantum - liquid phase ) ; i.e. , a sharp peak at low temperatures and a very broad peak extending over high temperatures .
the double - peak structure is not clearly seen at @xmath81 and @xmath84 .
we should note here that the specific heat coefficient @xmath85 , where @xmath86 at low temperatures , can not be deduced from the present calculations since @xmath72 decays exponentially at low temperatures due to the discreteness of the energies of finite - size systems .
we will show in sec .
iii d that the low - temperature sharp peak comes from the excitation of the spin degrees of freedom of the system and the broad high - temperature peak comes from the excitations of the density degrees of freedom of the system . in other words ,
the width of the sharp low - energy peak scales with the exchange interactions between spins ( a combination of @xmath1 and @xmath28 ) and the width of the broad high - energy peak scales with the hopping parameter @xmath0 of the vacancy .
it is interesting to note that the double - peak structure in the specific heat @xmath72 due to the separation in their energy scales between spin and density degrees of freedom has previously been discussed in the context of the low - energy excitations in the hubbard ladder systems with charge ordering instability although the latter is for the insulating systems with a charge gap.@xcite in sec .
iv , we will compare the obtained double - peak structure with experiment@xcite .
we also calculate the temperature dependence of the entropy @xmath87 ( not shown here ) , which will be compared with experiment@xcite also in sec .
+ the calculated results for the temperature dependence of the uniform magnetic susceptibility @xmath88 are shown in fig .
we find that the temperature variation is strongly dependent on the value of @xmath28 : ( i ) when @xmath89 , @xmath88 is strongly enhanced in comparison with the curie susceptibility @xmath90 , resulting in the ferromagnetic spin polarization at low temperatures .
( ii ) when @xmath80 at which the frustration in the spin degrees of freedom is the largest , @xmath88 is slightly suppressed in comparison with the curie law . here , the ground state is highly degenerate due to the frustration of the spin degrees of freedom ; in our cluster , the degeneracy is 15 fold .
( iii ) when @xmath91 , @xmath88 is rapidly suppressed with decreasing temperatures . here
, the ground state of the system is spin singlet without degeneracy .
it should be noted that these results come basically from the finite - size effects of small clusters .
however , we may infer the intrinsic nature of the infinite - size system and its @xmath28 dependence from the low - energy behavior under the magnetic field .
the results for @xmath88 thus obtained are compared with experiment in sec .
+ the calculated results for the spin and density excitation spectra at @xmath80 are shown in fig .
the excitation spectra for the corresponding noninteracting infinite - size system , @xmath92 , are also shown for comparison .
we find the following .
the spectral weight for the spin excitations is concentrated on a very low - energy region of around @xmath93 .
this reflects the presence of a large number of nearly degenerate low - energy states coming from the frustrated nature of the spin degrees of freedom .
the low - energy spectral weight is extended over the entire brillouin zone , rather than special momenta , reflecting the spatially localized nature of the spin fluctuations .
the spectral weight for the density excitations , on the other hand , extends over a wide energy range of about @xmath94 ( entire band width ) , which is more or less resembles the spectrum of the noninteracting system . from these results , we may say that the spin and density excitations are clearly separated in their energy scales : i.e. , the spin excitations concentrate on the low energy regions , the width of which scales with the exchange interactions ( a combination of @xmath1 and @xmath28 ) , and the charge excitations extend over the entire band width , which scales with the hopping parameter @xmath0 of the vacancy . with increasing @xmath28
, we find the upward shift of the low - energy spectral weight of the spin excitations ; e.g. , at @xmath82 , we find the peaks at a higher - energy region of around @xmath95 , where the momentum dependence of the positions of the peaks becomes significant as well .
thus , the separation between the energy scales of the spin and density excitations becomes weaker .
we should note that the energy range where the spectral weight of the spin excitations accumulates , i.e. , @xmath96 , corresponds well to the temperature range where the low - temperature sharp peak in the calculated specific heat @xmath72 appears .
we should also note that the broad spectra extending over the entire band width correspond well to the very broad high - temperature peak in the calculated result for @xmath72 .
we may therefore conclude that the separation between the spin and density excitations in their energy scales is responsible for the double - peak structure of the temperature dependence of the specific heat .
+ the calculated results for the single - particle excitation spectra @xmath97 are shown in fig .
6 . from the results
, we can deduce the possible quasiparticle band structure and hence the fermi - surface topology .
we find the following .
there are broad and incoherent spectral features over a wide energy range corresponding to the total band width of the noninteracting dispersion , i.e. , @xmath98 , but there emerge the sharp quasiparticle - like peaks with a characteristic dispersion in the vicinity of the fermi energy .
this result is similar to the case of the square - lattice @xmath0-@xmath1 model near half filling.@xcite let us assume this to be the consequence of the presence of fermionic quasiparticles .
then , we find the quasiparticle band structure to be fitted well by the noninteracting band dispersion with a reduced hopping parameter @xmath99 or with an enhanced effective mass @xmath100 of the quasiparticle ( see fig . 7 ) . from the fitting , we find the value @xmath101 or @xmath102 at @xmath70 and @xmath80 .
the present results also suggests that the fermi - surface topology of the quasiparticles is equivalent to that of the noninteracting system since the quasiparticle band can be obtained only by assuming that the band width is reduced ( or the band mass is enhanced ) .
thus , the fermi surface is large ( i.e. , its area @xmath103 ) rather than small ( i.e. , its area @xmath104 ) .
the doping dependence of @xmath100 should be interesting , in particular whether @xmath100 diverges or not at @xmath105 .
however , we can not answer this question in our small - cluster study ; the behavior of @xmath100 even in the square - lattice @xmath0-@xmath1 model still remains to be a puzzle .
+ we may also assume that the quasiparticle band width shown in fig .
7 may scale well with the energy of the spin excitations , i.e. , a combination of the exchange parameters @xmath1 and @xmath28 , while the entire band width ( @xmath106 ) of the broad spectral features shown in fig .
6 scales with @xmath0 , as in the case of the square - lattice @xmath0-@xmath1 model near half filling.@xcite thus , we again find the separation between the spin and density degrees of freedom in their energy scales .
the quasiholes ( or quasiparticles as their conjugate ) are thus the vacancies dressed by the spin excitations .
we should note therefore that the spin and density degrees of freedom are not exactly separated in this sense , unlike in the tomonaga - luttinger liquid in the one - dimensional interacting fermion systems.@xcite only the energy scales are different .
further experimental and theoretical studies will be required to clarify the true low - energy physics of the system .
we may also point out that the enhanced effective mass of the quasiparticle band structure may partly be responsible for the enhancement of the effective mass @xmath107 determined from the specific heat coefficient @xmath85 although the latter can not be obtained from our finite - size calculations . in sec .
iv , we compare the effective mass obtained from the quasiparticle band dispersion with experiment .
the calculated results for the heat capacity @xmath72 at @xmath79 and @xmath80 ( see sec .
iii b ) at the fillings of @xmath77 and @xmath108 are compared with experiment in fig .
we here assume the value of @xmath0 determined so as to reproduce the higher - temperature peak observed in @xmath72 at @xmath77 , i.e. , @xmath109 mk , so that we have @xmath110 mk and @xmath111 mk with keeping the ratio @xmath0 : @xmath112 : @xmath28 = @xmath113 .
we also assume that the total area of the sample used in experiment ( 556 m@xmath114 ) is uniformly active and contributes to the heat capacity .
we should note that the results for the temperature region @xmath115 mk are not reliable because of the finite - size effects where the discreteness of the energies in the system gives the exponential decay of the heat capacity at low temperatures .
we then find the fair agreement with experiment ; in particular , the double - peak structure in @xmath72 , i.e. , the lower - temperature peak that comes from the spin excitations and higher - energy peak that comes from the density excitations of the system , are reasonably well reproduced .
more precisely , we find that our calculated results reproduce the experimental tendency that , near half filling , the lower - temperature peak is high but with increasing the vacancy concentration , the higher - temperature peak becomes larger and simultaneously the peaks shift to higher temperatures .
note that the specific heat coefficient @xmath85 ( or the effective mass ) can not be estimated from the present calculations of @xmath72 due to finite - size effects .
however , we find that the value of the enhanced effective mass @xmath102 estimated from the calculated quasiparticle band structure ( see sec .
iii e ) is consistent with the experimental value @xmath116 estimated from the observed temperature dependence of @xmath72 at @xmath117.@xcite here , we should note that the definition of @xmath100 in the experimental specific heat coefficient is two - fold : one is the value deduced from the lower - temperature peak and the other is the value deduced from the higher - temperature peak .
the two values are , however , not very different at least for @xmath118 , so that we can make comparison with our theoretical value .
then , it seems reasonable to assume that the renormalization of the band structure due to the spin excitations is mainly responsible for the observed enhancement of the effective mass.@xcite thus , stated differently , the specific heat coefficient @xmath85 should be determined predominantly by the spin excitations of the system .
the calculated results for the entropy @xmath87 are also compared with experimentally determined@xcite entropy in fig .
we again find the fair agreement in their general tendencies .
more precisely , we find that the calculated curves of @xmath72 cross the line of the entropy of @xmath24 free spins @xmath119 at @xmath12010 mk , at which the lower - temperature peak in @xmath72 terminates .
this result also supports that the lower - temperature peak in @xmath72 comes from the excitations of the spin degrees of freedom of the system .
the higher - energy peak should therefore come from the density degrees of freedom or motions of vacancies in the system .
note that the experimentally determined@xcite entropy is significantly smaller than @xmath119 even in the vicinity of half filling , @xmath121 , and even at temperatures of @xmath122 mk where the lower - temperature peak in @xmath72 terminates .
the missing entropy may reside in the region of much lower temperatures that the present experiment does not approach.@xcite + the calculated results for the magnetization under the uniform magnetic field are compared with experiment in fig .
we should note that the experimentally applied magnetic field , which is @xmath123 in our calculations , is very small in comparison with the energy scales of the @xmath2he system , so that the behavior of the magnetization under the uniform magnetic field is the same as that of the uniform magnetic susceptibility defined at @xmath124 .
we find that , although the finite - size effect is strong at low temperatures , the calculated magnetization at @xmath80 is consistent with experiment in the sense that the value is somewhat smaller than the value expected from the curie law @xmath125 . for more quantitative comparison , however , one would need the techniques appropriate for treating infinite - size systems , so that experimentally observed plateau - like behavior@xcite in the temperature dependence of the magnetization can be explained .
we have used an exact - diagonalization technique on small clusters to study the low - energy physics of the triangular - lattice @xmath0-@xmath1 model with the multiple - spin exchange interactions , whereby we have considered the anomalous properties observed in the doped mott region of the two - dimensional liquid @xmath2he adsorbed on a graphite surface .
we have calculated the temperature dependence of the specific heat , entropy , and uniform magnetic susceptibility , as well as the spin and density excitation spectra and single - particle spectra for the model , and have considered their implications .
\(1 ) the double - peak structure appears in the temperature dependence of the specific heat .
the result is quantitatively consistent with experiment .
the low - temperature sharp peak comes from the spin excitations and high - temperature broad peak comes from the density excitations .
\(2 ) the spectral weight for the spin excitations is concentrated on a very low - energy region , the width of which scales with the exchange interactions , while that of the density excitations extends over an entire band width , which scales with the hopping parameter of the vacancy . the clear separation between spin and density excitations in their energy scales is thus found .
\(3 ) the accumulation of the spectral weight of the spin excitations comes from the frustrated nature of the spin degrees of freedom of the system ; i.e. , the ferromagnetic two - spin interactions @xmath1 compete with the antiferromagnetic four - spin interactions @xmath28 on the geometrically frustrated triangular lattice .
\(4 ) the single - particle excitation spectra suggest that the vacancies behave like the fermionic quasiparticles dressed by the spin excitations , of which the effective band mass is estimated to be @xmath102 at @xmath70 , in consistent with the effective mass measured from the specific heat coefficient .
\(5 ) the temperature dependence of the spin susceptibility shows a suppressed curie - like behavior , reflecting the situation where the ground state is highly degenerate due to the frustrated nature of the spin degrees of freedom .
we hope that the present study will shed more light on the physics of the two - dimensional @xmath2he systems and stimulate further experimental and theoretical studies of the systems in greater details .
we have focused on the doped mott region of the monolayer @xmath2he in this paper .
however , it has recently been reported@xcite that the bilayer @xmath2he systems also contain rich physics concerning heavy fermions with quantum criticality , which we want to leave for future study .
enlightening discussions with professors hiroshi fukuyama and john saunders are gratefully acknowledged .
this work was supported in part by grants - in - aid for scientific research ( nos . 18028008 , 18043006 , 18540338 , and 19014004 ) from the ministry of education , culture , sports , science and technology of japan .
ts acknowledges financial support from jsps research fellowship for young scientists .
a part of computations was carried out at the research center for computational science , okazaki research facilities , and the institute for solid state physics , university of tokyo .
for a recent review , see h. fukuyama , j. phys .
. jpn . * 77 * , at press in no .
11 ( 2008 ) ; also see references therein .
m. roger , phys .
* 64 * , 297 ( 1990 ) .
m. roger , c. buerle , y. m. bunkov , a. s. chen , and h. godfrin , phys .
lett . * 80 * , 1308 ( 1998 ) .
g. misguich , b. bernu , c. lhuillier , and c. waldtmann , phys .
rev . lett .
* 81 * , 1098 ( 1998 ) .
t. momoi , h. sakamoto , and k. kubo , phys .
b * 59 * , 9491 ( 1999 ) . t. momoi , p. sindzingre , and n. shannon , phys . rev . lett . * 97 * , 257204 ( 2006 ) . c. buerle , y. m. bunkov , a. s. chen ,
d. j. cousins , h. godfrin , m. roger , and s. triqueneaux , physica b * 280 * , 95 ( 2000 ) . s. watanabe and m. imada , j. phys .
. jpn . * 76 * , 113603 ( 2007 ) .
a. f. andreev and i. m. lifshitz , sov .
jetp * 29 * , 1107 ( 1969 ) .
h. matsuda and t. tsuneto , prog .
. suppl . * 46 * , 411 ( 1970 ) .
y. matsumoto , d. tsuji , s. murakawa , h. akisato , h. kambara , and h. fukuyama , j. low temp .
phys . * 138 * , 271 ( 2005 ) .
y. matsumoto , d. tsuji , s. murakawa , c. buerle , h. kambara , and h. fukuyama , unpublished ( 2007 ) .
y. fuseya and m. ogata , preprint cond - mat/0804.4329 .
k. seki , t. shirakawa , and y. ohta , j. phys .
, in press ( 2008 )
. n. shannon , b. schmidt , k. penc , and p. thalmeier , eur .
j. b * 38 * , 599 ( 2004 ) .
n. elstner and a. p.
young , phys .
b * 50 * , 6871 ( 1994 ) .
y. ohta , t. nakaegawa , and s. ejima , phys .
b * 73 * , 045101 ( 2006 ) . r. eder and y. ohta , phys .
b * 50 * , 10043 ( 1994 ) .
r. eder , y. ohta , and s. maekawa , phys .
lett . * 74 * , 5124 ( 1995 ) .
see , e.g. , t. giamarchi , _ quantum physics in one dimension _ ( oxford university press , new york , 2004 )
. h. fukuyama , private communication .
s. murakawa _
_ , aip conf . proc . * 850 * , 311 ( 2006 ) .
m. neumann , j. nyki , b. cowan , and j. saunders , science * 317 * , 1356 ( 2007 ) . | using an exact diagonalization technique on small clusters , we study spin and density excitations of the triangular - lattice @xmath0-@xmath1 model with multiple - spin exchange interactions , whereby we consider anomalous properties observed in the doped mott region of the two - dimensional liquid @xmath2he adsorbed on a graphite surface . we find that the double - peak structure consistent with experiment appears in the calculated temperature dependence of the specific heat ; the low - temperature sharp peak comes from the spin excitations reflecting the frustrated nature of the spin degrees of freedom and high - temperature broad peak comes from the density excitations extending over the entire band width .
the clear separation in their energy scales is evident in the calculated spin and density excitation spectra .
the calculated single - particle excitation spectra suggest the presence of fermionic quasiparticles dressed by the spin excitations , with an enhanced effective mass consistent with experiment . |
chromium dioxide ( cro@xmath0 ) , which crystallizes in the tetragonal structure of rutile - type , is a well - known material because of its half - metallic nature with @xmath5 k.@xcite the origin of the ferromagnetic ( fm ) and half - metallic property of cro@xmath0 was explained in terms of the double - exchange model.@xcite due to the crystal field of distorted ( flattened ) cr - o@xmath6 octahedra , cr @xmath7 states are split into lower _ _ d__@xmath8 and higher _ _ d__@xmath9/__d__@xmath10 states . out of two @xmath11 electrons of cr@xmath12 ,
one occupies the lower _
_ d__@xmath8 that is localized , while the other occupies the higher
_ _ d__@xmath9/__d__@xmath10 that are delocalized near the fermi level ( @xmath13 ) due to the hybridization with o @xmath14 states .
then the double - exchange interaction arises from the hund coupling between localized _ _ d__@xmath8 and delocalized half - filled _
_ d__@xmath9/__d__@xmath10 states , so as to produce the fm and half - metallic properties .
in contrast to numerous reports on electronic and magnetic properties of cro@xmath0 , there have been relatively small number of studies on structural and lattice dynamical properties of cro@xmath0 .
especially , there is no experimental or theoretical report on the phonon dispersion curve for cro@xmath0 , except for a few raman studies.@xcite under the pressure , cro@xmath0 is known to undergo the structural transition from the ground state tetragonal cro@xmath0 ( t - cro@xmath0 ) to the orthorhombic cro@xmath0 ( o - cro@xmath0 ) of cacl@xmath0-type at p = 12 - 14 gpa.@xcite the question followed is whether there will an additional structural transitions from o - cro@xmath0 at higher pressure .
in fact , this question is not just for cro@xmath0 but also relevant to the structural stability issue of transition - metal ( tm ) dioxides ( tmo@xmath0 ) .
note that tmo@xmath0s show diverse structures depending on the tm elements.@xcite furthermore , magnetic properties of cro@xmath0 under the pressure are intriguing , such as ( i ) whether the half - metallic nature is preserved , and ( ii ) when cro@xmath0 becomes non - magnetic . in this work , to investigate the pressure - induced structural transitions of cro@xmath0
, we have studied phonon dispersions and total energies of relevant cro@xmath0 structures as a function of pressure .
based on the calculated phonon dispersions and the total energies , we have found three possible structural transitions with increasing pressure .
the first transition is consistent with the known transition from t - cro@xmath0 to o - cro@xmath0 . at this transition ,
fm and half - metallic properties are preserved , in agreement with previous reports of literature.@xcite the second transition is from o - cro@xmath0 to monoclinic cro@xmath0 ( m - cro@xmath0 ) of moo@xmath0 type , which is nonmagnetic ( nm ) .
the third transition is identified from m - cro@xmath0 to cubic cro@xmath0 ( c - cro@xmath0 ) of caf@xmath0-type .
interestingly , c - cro@xmath0 is a fm insulator even at the high pressure of p @xmath3 88.8 gpa .
note that the second and third structural transitions are our new findings for cro@xmath0 under the high pressure .
band structures and phonon dispersions of cro@xmath0 were obtained by employing the pseudo - potential band method and the linear response method , respectively , implemented in the quantum espresso code.@xcite the generalized gradient approximation ( gga ) is utilized for the exchange - correlation potential .
self - consistent electron and phonon band calculations were carried out after the full - relaxation of internal atomic positions and lattice parameters .
we have considered various structures of cro@xmath0 . at the ambient pressure ,
the stable phase is t - cro@xmath0 of rutile - type ( _ _ p__4@xmath0/_mnm _ ) , in which cr atoms are positioned at ( 0,0,0 ) and ( @xmath4 , @xmath4 , @xmath4 ) , while o atoms at @xmath15(u , u , 0 ) and @xmath15(@xmath4+u , @xmath16u , @xmath4 ) .
initial lattice constants and atomic positions adopted before the full - relaxation are a = b=4.421@xmath17 , c=2.916@xmath17 , and u=0.3043.@xcite for the high pressure phase of o - cro@xmath0 of cacl@xmath0-type ( _ pnnm _ ) , we have adopted a=4.3874@xmath17 , b=4.2818@xmath17 , c=2.8779@xmath17 , u@xmath18 = 0.299 , and u@xmath19=0.272.@xcite for candidate structural phases at the higher pressure , we considered m - cro@xmath0 of moo@xmath0 type ( _ _ p__2@xmath20/_c_)@xcite and c - cro@xmath0 of caf@xmath0-type ( _ _ fm__@xmath21_m_).@xcite in the latter , a cr atom is positioned at ( 0 , 0 , 0 ) , and o atoms at ( 0.25 , 0.25 , 0.25 ) , ( 0.25 , 0.25 , 0.75 ) .
( a ) fm t - cro@xmath0 of rutile type at the ambient pressure .
( b ) fm t - cro@xmath0 at p=34.0 gpa . notice the phonon softening at @xmath22 , which corresponds to the b@xmath2 mode .
the negative frequency here represents the imaginary part of the phonon frequency .
( c ) fm o - cro@xmath0 of cacl@xmath0 type at p=34.1 gpa . , width=8 ] soft phonon at @xmath22 for p=34.0 gpa .
the blue and light - blue circles represent cr ions , while the orange and red circles represent oxygen ions .
blue and orange ions are located at z=@xmath4 .
only the oxygen ions move in this mode .
, width=4 ] figure @xmath23 shows the phonon dispersions of t - cro@xmath0 and o - cro@xmath0 at the ambient and high pressures . as shown in fig .
[ phd_0_35](a ) , t - cro@xmath0 at the ambient pressure has regular phonon dispersions , reflecting the stable phase of t - cro@xmath0 at the ambient pressure .
in contrast , t - cro@xmath0 at p=34.0 gpa in fig . [ phd_0_35](b ) has a softening phonon mode at * q*=@xmath22 , indicating the structural instability of t - cro@xmath0 at this pressure .
the softened mode corresponds to b@xmath2 mode that is raman - active . as shown in fig .
[ b1 g ] , b@xmath2 mode generates the rotating motions of oxygen ions .
the resulting lattice displacements induce the structural transformation from t - cro@xmath0 of rutile type to o - cro@xmath0 of cacl@xmath0 type .
figure @xmath23(c ) provides the phonon dispersion of o - cro@xmath0 at p=34.1 gpa .
the phonon dispersion is regular , implying that o - cro@xmath0 is stable at this pressure .
therefore , fig .
[ phd_0_35 ] clearly demonstrates that there is a structural transition from t - cro@xmath0 to o - cro@xmath0 at the pressure of p@xmath24 gpa . in fig .
[ lattice](a ) , we plotted the calculated raman - active phonons of t - cro@xmath0 and o - cro@xmath0 as a function of pressure .
there are four raman modes ( b@xmath2 , e@xmath25 , a@xmath2 , b@xmath26 ) for t - cro@xmath0 , and six raman modes ( a@xmath25 , b@xmath2 , b@xmath26 , b@xmath27 , a@xmath25 , b@xmath2 ) for o - cro@xmath0.@xcite our data are consistent with experimental data up to p@xmath28 gpa.@xcite with increasing the pressure , one can clearly see the softening of b@xmath2 mode of t - cro@xmath0 , which indicates the structural instability of t - cro@xmath0 .
one can also notice two transition points .
the first one corresponds to the transition from t - cro@xmath0 to o - cro@xmath0 at p=9.8 gpa . at this transition
, cro@xmath0 keeps its fm and half - metallic properties.@xcite .
the second one corresponds to the transition at p=76.0 gpa .
the stable structure for p@xmath29 has not been identified yet .
the phonon anomalies at two transition points are also revealed in the variation of lattice constants of cro@xmath0 under the pressure . in fig .
[ lattice](b ) , the lattice constants are plotted as a function of pressure .
one can see anomalous behaviors of the lattice constants at the two transition points that are coincident with those in fig .
[ lattice](a ) . and o - cro@xmath0 versus pressure .
t- and o- stand for t - cro@xmath0 and o - cro@xmath0 respectively .
the lines connecting data are guide for eyes .
two phase transitions are noticed at p=9.8 gpa and p=76.0 gpa , which are marked by vertical lines .
( b ) calculated equilibrium lattice constants of t - cro@xmath0 and o - cro@xmath0 versus pressure .
, width=8 ] versus pressure . magnetic transition from the fm to
the nm phase occurs at p=76.0 gpa .
( b ) the phonon dispersion curve of nm o - cro@xmath0 at p=80.5 gpa .
, width=8 ] to investigate the second structural transition in more detail , we have examined the behavior of magnetic moment .
et al._@xcite once reported that there would be a magnetic transition in t - cro@xmath0 from half - metallic to nm at p@xmath30 gpa .
however , as discussed above , there occurs a structural transition from t - cro@xmath0 to o - cro@xmath0 at the low pressure of about p=10 gpa .
hence , in fig .
[ phd_80](a ) , we have examined the magnetic moment behavior for o - cro@xmath0 .
it is seen that there is a fm to nm transition at p=76.0 gpa , which is close to the second structural transition point .
the magnetic moment of o - cro@xmath0 suddenly drops at this transition point .
the half - metallic property persists up to this pressure .
this magnetic transition was also observed by kuznetsov _ _ et al.__@xcite , who obtained the transition pressure of p=53 gpa based on the pseudo - potential band method implemented in the vasp code .
it is thus tempting to conjecture that the second transition observed in fig . [
lattice](a ) corresponds to the magnetic transition in o - cro@xmath0 .
however , the phonon dispersion curve in fig . [ phd_80](b ) for nm o - cro@xmath0 at p=80.5 gpa shows the strong phonon softenings , indicating that even nm o - cro@xmath0 is unstable at the pressure of p@xmath31 gpa .
therefore , it is not possible that fm o - cro@xmath0 changes into nm o - cro@xmath0 with increasing the pressure .
there might be an additional structural transition in this pressure range .
two candidate structures after the transition are m - cro@xmath0 of moo@xmath0 type and c - cro@xmath0 of caf@xmath0 type .
the monoclinic structure of moo@xmath0 type is chosen from the expectation that , with increasing pressure , two @xmath11 electrons of cr become itinerant , and the local environment becomes similar to that of moo@xmath0 . in a similar system vo@xmath0 , a softening of phonon frequency was observed at * q*=r ( @xmath4 , 0 , @xmath4 ) , which corresponds to the atomic movements from the tetragonal structure of rutile - type to the monoclinic structure.@xcite indeed , the softening mode at * q * = r in fig .
[ phd_80](b ) is related to this structural transition , because the orthorhombic structure of cacl@xmath0 type is nothing but the distorted rutile - type structure .
figure [ 2soften ] depicts the normal mode of the softened phonon mode at * q * = r. the displacements generate cr - cr dimerization along the c - axis , which is consistent with the main distortions of the transition from the rutile - type to the monoclinic structure .
it is thus reasonable to expect the second transition to be from o - cro@xmath0 of cacl@xmath0 type to m - cro@xmath0 of moo@xmath0 type . concerning another candidate , c - cro@xmath0 of caf@xmath0 type ,
there were previous reports predicting the structural transitions from cacl@xmath0 type to caf@xmath0 type structure for cro@xmath0@xcite and ruo@xmath0.@xcite also , caf@xmath0 type is a typical structure of tmo@xmath32 .
for example , zro@xmath0 and hfo@xmath0 , which have relatively large cations , crystallize in caf@xmath0 type structure at high temperature .
@xcite structures versus volume .
data are fitted by the birch - murnaghan equation of state .
, width=328 ] to identify the additional structural transition at the higher pressure , we have compared the total energies of candidate structures in fig .
[ energy ] . from the total energy versus volume curves in fig .
[ energy ] , one can identify three structural phase transitions .
the first one is from fm t - cro@xmath0 to fm o - cro@xmath0 at the estimated pressure of p=12.2 gpa , which is consistent with phonon calculation in fig .
[ lattice ] . for p @xmath3 61.1 gpa ,
nm m - cro@xmath0 becomes the most stable , which corresponds to the second transition from o - cro@xmath0 to m - cro@xmath0 , as discussed in the phonon study of fig .
[ phd_80 ] . for p @xmath3 88.8 gpa
, c - cro@xmath0 , which is a fm insulator , becomes the most stable .
the more stable fm and insulating phase than the nm metallic phase of c - cro@xmath0 at this high pressure is extraordinary .
the magnetic moment of c - cro@xmath0 amounts to @xmath33@xmath34/cr , which is close to those in t - cro@xmath0 and o - cro@xmath0 .
note , however , that c - cro@xmath0 is an insulator not a half - metal .
we have confirmed this result by employing the all - electron full potential linearized augmented plane wave method ( flapw ) band method@xcite implemented in wien2k package too.@xcite the present result is different from that by srivastava _
et al._,@xcite who obtained the stable nm metallic phase of c - cro@xmath0 for p@xmath3590 gpa .
the different result is likely to come from their use of a simple tight - binding lmto band method . the transition from m - cro@xmath0 to c - cro@xmath0 is thought to originate from the increasing packing ratio .
there are six and eight oxygens around cr in m - cro@xmath0 and c - cro@xmath0 , respectively .
et al.__@xcite proposed several possible paths of structural transition from rutile to caf@xmath0 type structure in tmo@xmath36 .
interestingly , one of the paths is the same as the present structural transition path , rutile - type ( _ _ p__4@xmath0/_mnm _ ) @xmath37 cacl@xmath0-type ( _ pnnm _ ) @xmath37 moo@xmath0-type ( _ _ p__2@xmath20/_c _ ) @xmath37 caf@xmath0-type ( _ _ fm__@xmath21_m _ ) .
but they did not take into account the magnetic state .
we have studied the pressure effect on the structural properties of cro@xmath0 by performing the phonon dispersion and total energy band structure calculations . combining two analysis methods ,
we have found that there are three structural transitions with increasing pressure up to 100 gpa .
the first one is the transition from t - cro@xmath0 of rutile type to o - cro@xmath0 of cacl@xmath0 type at p@xmath1 10 gpa ( 9.8 gpa from phonon dispersion analysis , while 12.2 gpa from total energy study ) .
the fm and half - metallic properties of cro@xmath0 are preserved at this transition .
the second structural transition is from fm o - cro@xmath0 to nm m - cro@xmath0 , which corresponds to the lattice displacement of the phonon softening at r in o - cro@xmath0 .
the transition pressure is p=76.0 gpa from the phonon dispersion analysis , whereas p=61.1 gpa from the total energy study .
the third structural transition is from nm m - cro@xmath0 to fm c - cro@xmath0 at p= 88.8 gpa , which is accompanied by the metal to insulator transition .
99 b. i. min , h. j. f. jansen , and a. j. freeman , phys .
b * 33 * , 6383 ( 1986 ) .
b. blaha , k. schwarz , g. k. h. madsen , d. kvasnicka , and j. luitz , wien2k ( technische universitt wien , austria , 2001 ) . | to investigate the pressure - induced structural transitions of chromium dioxide ( cro@xmath0 ) , phonon dispersions and total energy band structures are calculated as a function of pressure .
the first structural transition has been confirmed at p@xmath1 10 gpa from the ground state tetragonal cro@xmath0 ( t - cro@xmath0 ) of rutile type to orthorhombic cro@xmath0 ( o - cro@xmath0 ) of cacl@xmath0 type .
the half - metallic property is found to be preserved in o - cro@xmath0 .
the softening of raman - active b@xmath2 phonon mode , which is responsible for this structural transition , is demonstrated .
the second structural transition is found to occur for p@xmath3 61.1 gpa from ferromagnetic ( fm ) o - cro@xmath0 to nonmagnetic ( nm ) monoclinic cro@xmath0 ( m - cro@xmath0 ) of moo@xmath0 type , which is related to the softening mode at * q * = r(@xmath4,0,@xmath4 ) .
the third structural transition has been newly identified at p= 88.8 gpa from m - cro@xmath0 to cubic cro@xmath0 of caf@xmath0 type that is a fm insulator . |
lattice calculations of the thermodynamical functions of quantum chromodynamics ( qcd ) indicate @xcite that , ( at zero net baryon density ) in the vicinity of a critical temperature @xmath4 mev , strongly interacting matter undergoes a rapid transition from a ( chirally broken , confined ) hadronic phase to a ( chirally symmetric , deconfined ) quark gluon plasma ( qgp ) .
the width of that transition region is presently only known to be in the range @xmath5 mev .
therefore , one can not yet conclude whether the transition is a first order phase transition ( @xmath6 ) , or merely a rapid increase of the entropy density associated with the change from @xmath7 hadronic to @xmath8 quark and gluon degrees of freedom .
one of the primary goals of relativistic heavy - ion physics is the creation and experimental observation of the predicted qgp phase of matter .
many signatures have been proposed such as electromagnetic radiation of thermal dileptons and photons @xcite , @xmath9suppression @xcite , jet quenching @xcite , strangelet formation @xcite , or disordered chiral condensates ( dcc s ) @xcite .
these signatures , however , do not depend directly on the actual form of the nuclear matter equation of state .
thermal electromagnetic radiation is , for instance , generic to any hot system , independent from its degrees of freedom ( as long as they have electromagnetic charge ) .
for example , it was shown @xcite that a hot hadron gas shines as brightly as a qgp . similarly , jet quenching and @xmath9suppression suppression as a function of transverse energy in pb+pb collisions at 158 agev @xcite . ]
are generic consequences of final state interactions in any form of dense matter @xcite .
finally , strangelet or dcc formation require very specific assumptions about the dynamical evolution of the system .
it is therefore of interest to study signals that are more directly related to the qcd equation of state .
signals of this type emerge from the influence of the equation of state on the _ collective _ dynamical evolution of the system .
such signals must be calculated within the framework of relativistic hydrodynamics @xcite , since that is the only dynamical model which provides a _ direct _ link between collective observables and the equation of state .
of course , the use of such an approach requires a strong dynamical assumption , namely that the equilibration rates are much larger than typical gradients of thermodynamic quantities in the system .
it now appears that radiative gluon energy loss in a qgp may be sufficiently large @xcite to support local equilibration on time scales less than 1 fm / c . in the following , we therefore neglect dissipative effects and assume the validity of ideal hydrodynamics to compute the collective evolution of the system .
it was shown in @xcite that the transition to the qgp softens the equation of state in the transition region , and thus reduces the tendency of matter to expand on account of its internal pressure .
this , in turn , delays the expansion and considerably prolongs the lifetime of the system .
in @xcite it was moreover shown that this effect leads to a reduction of the transverse directed flow in semi - peripheral collisions that can be readily tested experimentally at fixed target energies @xcite .
in @xcite the flow structure for the one dimensional expansion of a slab of matter ( `` landau expansion model '' @xcite ) was studied as a function of its initial energy density @xmath10 and of the width @xmath2 of the transition region in the equation of state . in particular ,
it was shown how , in addition to the softening of the equation of state , for @xmath11 the type of the hydrodynamical expansion solution changes qualitatively from a simple rarefaction wave to a rarefaction discontinuity .
such deflagration fronts @xcite have small velocities , @xmath12 , and thus the conversion of qgp matter into hadronic matter is considerably delayed .
in @xcite we also showed that for larger transition widths ( @xmath13 ) the lifetime of matter with temperature around @xmath14 is considerably reduced but cooler regions with @xmath15 remain long - lived . in the present work
we extend the investigations of @xcite to more realistic , 3dimensional geometries .
we consider systems with spherical symmetry ( `` fireball '' geometry ) , and with transverse cylindrical symmetry and boost - invariant initial conditions in longitudinal direction ( `` bjorken cylinder '' geometry @xcite ) .
for these symmetric geometries it is possible to use a modified one dimensional algorithm as e.g. presented in @xcite .
note that in contrast to the simple one dimensional landau geometry , single inclusive particle spectra and also two particle correlation functions can be calculated in a realistic way from the hydrodynamic solutions for the above geometries . in this paper
we show that the influence of the width @xmath2 of the transition on the dynamical evolution of the system in these 3dimensional geometries is very similar to that found in @xcite for purely one dimensional expansion .
in particular , the softening of the equation of state associated with a rapid cross - over region leads to a delay in the expansion and a prolonged lifetime of matter with temperature _ below _ @xmath14 . to explore observable consequences of the prolonged lifetime of the system
, we then calculate the inverse width of the two particle correlation functions in side and out direction , @xmath16 and @xmath17 .
we demonstrate that the initial energy density dependence of the system s lifetime is mapped closely by the excitation function of the ratio @xmath18 .
the phase transition leads to a time - delayed expansion that is reflected in an enhancement of the above ratio relative to its value obtained in the ideal gas case without a transition .
this signature of the qgp transition was first proposed by pratt @xcite who performed similar calculations for the fireball geometry and , independently , by bertsch et al .
@xcite who employed a kinetic model for the hadronization in the bjorken cylinder geometry .
these previous investigations assumed the existence of a sharp first order transition , @xmath6 .
we show here that this signature is a generic feature of a rapid cross - over transition and holds for @xmath19 as well .
the importance of the time delay as a signature of qgp formation is that it is among the few signatures that probe directly the equation of state of strongly interacting matter .
it tests whether there exists a `` soft region '' in the equation of state of ultra - dense matter @xcite which is most naturally characterized @xcite by a reduction of the local velocity of sound @xmath20 as a function of the local energy density . as discussed in detail in the next section ,
a very general feature of a transition ( that happens within a sufficiently narrow temperature interval @xmath2 ) is the existence of such a region . for matter passing through that region of energy densities during the expansion phase ,
the flow will temporarily slow down or even possibly stall under suitable conditions .
the main result of the present work is to identify initial conditions where the resulting stall is sufficiently long to be observable via pion interferometry @xcite . as we show , the maximum time delay may occur at ags as well as at rhic energies , depending on whether matter is initially stopped in a fireball or expands as in bjorken s longitudinally boost - invariant scenario .
the remainder of this paper is organized as follows . in section 2
we present the equation of state used in our investigations and explain our method to solve the hydrodynamical equations numerically for the above mentioned geometries . in section 3
we compare the hydrodynamical expansion solutions for different initial energy densities , for different values of @xmath2 , and for different values of degrees of freedom in the qgp and hadronic phase .
section 4 is devoted to the discussion of the lifetime of the system and how it can be measured via the widths of the side and out correlation function .
section 5 concludes with a summary and discussion of our results .
two appendices contain the derivation of the formulae employed to calculate the two particle correlation functions in the hydrodynamical framework .
natural units @xmath21 are used throughout this paper .
available lattice data for the entropy density in full qcd can be approximated by the simple parametrization @xcite @xmath22 ^ 3 \left ( 1 + \frac{d_q - d_h}{d_q+d_h } \ , \tanh \left [ \frac{t - t_c}{\delta t } \right ] \right)~,\ ] ] where @xmath23 is the entropy density at @xmath14 .
pressure @xmath24 and energy density @xmath25 follow then from thermodynamical relationships . for @xmath6 , the equation of state ( [ eos ] )
reduces to the mit bag equation of state @xcite with bag constant @xmath26 .
if one measures energies in units of @xmath14 and energy densities in units of the enthalpy density @xmath27 , the equation of state ( [ eos ] ) depends only on the ratio @xmath28 , and not on @xmath8 and @xmath7 separately . for @xmath29 ,
this ratio determines the latent heat ( density ) @xmath30 .
( here , @xmath31 is the energy density at the phase boundary between mixed phase and qgp , @xmath32 is that at the boundary between mixed and hadronic phase . ) for the case @xmath33 ( corresponding to an ultrarelativistic gas of pions ) and @xmath34 ( corresponding to two massless flavours of quarks and antiquarks , and eight massless gluons ) , the latent heat , @xmath35 gev @xmath36 , is large . on the other hand , including a resonance gas in the hadronic phase and/or reducing the effective number of degrees of freedom on the qgp side @xcite , @xmath37 may be taken as a ( perhaps more realistic ) lower limit , with a smaller latent heat @xmath38 . assuming that the high - temperature phase consists of gluons only ( such as expected for the `` hot - glue scenario '' @xcite ) this
would then correspond to about @xmath39 mev @xmath36 in physical units . to cover the range of uncertainty in the qcd equation of state we consider @xmath40 and @xmath37 , and @xmath41 and @xmath42 in our investigation of hydrodynamical expansion solutions . in fig . 1 we show ( a ) the entropy density and ( b ) the energy density as functions of temperature , and ( c ) the pressure and ( d ) the velocity of sound squared @xmath43 as functions of energy density for @xmath6 , @xmath42 , and an ideal gas with @xmath7 degrees of freedom for @xmath40 .
the corresponding plot for @xmath37 looks rather similar qualitatively and is not shown .
quantitative differences are : ( i ) the phase boundaries are different , @xmath44 shifts from @xmath45 to @xmath46 , @xmath47 from @xmath48 to @xmath49 , and @xmath50 from @xmath51 to @xmath52 . also , as mentioned above , ( ii ) the latent heat @xmath53 is reduced .
figs . 1 ( a , b ) present the thermodynamic functions in a form to facilitate comparison with lattice data .
present lattice data for full qcd can be approximated with a choice of @xmath2 in the range @xmath54 . in the hydrodynamical context , however , figs .
1 ( c , d ) are more relevant .
as can be seen in ( c ) , for @xmath6 the pressure stays constant , @xmath55 , in the mixed phase @xmath56 .
hydrodynamical expansion is , however , driven by pressure _ gradients_.
it is therefore the velocity of sound , fig . 1 ( d )
, that is the most relevant measure of the system s tendency to expand .
it represents the capability to perform mechanical work ( which is proportional to pressure gradients @xmath57 ) for a given gradient in energy density @xmath58 . for @xmath29 , the velocity of sound vanishes in the mixed phase ,
i.e. , mixed phase matter does not expand at all on its own account , even if there are strong gradients in the energy density .
this has the consequence that it does not perform mechanical work and therefore cools less rapidly . for finite @xmath2 ,
pressure gradients are finite , but still smaller than for an ideal gas equation of state , and therefore the system s tendency to expand is also reduced , cf .
fig . 1 ( d )
. hydrodynamics is defined by local energy
momentum conservation , @xmath59 under the assumption of local thermodynamical equilibrium ( the so - called `` ideal fluid '' approximation ) the energy
momentum tensor @xmath60 assumes the particularly simple form @xcite @xmath61 where @xmath62 is the 4velocity of the fluid ( @xmath63 is the 3velocity , @xmath64 , @xmath65 ) , and @xmath66 is the metric tensor .
the system of equations ( [ eom ] ) is closed by choosing an equation of state in the form @xmath67 , i.e. , as depicted in fig . 1 ( c ) .
in the ideal fluid approximation , the ( equilibrium ) equation of state is the _ only _ input to the hydrodynamical equations of motion ( [ eom ] ) that relates to properties of the matter under consideration and is thus able to influence the dynamical evolution of the system .
the final results are uniquely determined once a particular initial condition and a decoupling ( `` freeze - out '' ) hypersurface are specified .
the symmetry of the fireball and bjorken cylinder geometry affects that the system of four equations ( [ eom ] ) reduces to two independent equations . with the definition @xmath68 , where the index @xmath69 indicates the radial component of the energy
momentum tensor , the respective equations read @xmath70 & = & -\ ,
f(e , p , v , r , t ) \ , \ , , \\
\partial_t\ , m + \partial_r\ , ( mv+p ) & = & -\ , g(m , v , r , t ) \ , \ , .
\label{eom2b}\end{aligned}\ ] ] here , @xmath71 is the radial component of the velocity .
for the fireball geometry , @xmath72 and @xmath73 do not depend on the time @xmath74 explicitly , @xmath75 for the bjorken cylinder geometry , the above equations describe the system s transverse evolution at @xmath76 ( cf .
@xcite ) , and due to the assumption of longitudinal boost invariance @xcite , the hydrodynamical solution for arbitrary @xmath77 can be easily obtained by a lorentz boost with ( space time ) rapidity @xmath78 $ ] .
the functions @xmath72 and @xmath73 read in this case @xmath79 in order to solve ( [ eom2a],[eom2b ] ) , we employ sod s operator splitting method @xcite , i.e. , for each time step we first generate solutions of the hydrodynamical equations for @xmath80 . in this form ,
the equations are purely one dimensional and can therefore be solved with e.g. the relativistic hlle scheme presented in @xcite .
the performance of this algorithm for solving one dimensional hydrodynamical problems and with equations of state featuring phase transitions was shown to be excellent @xcite .
it has also been employed in @xcite to solve the landau expansion problem . in a second step , sod s method prescribes to correct @xmath81 and @xmath82 for the respective geometry by solving the ordinary differential equations @xmath83 more specifically , in a slight variation of sod s method we solve the finite difference equations @xmath84 where quantities with a tilde are the solutions of the hydrodynamical equations with @xmath80 generated previously with the relativistic hlle scheme .
this two step predictor corrector scheme is repeated for each time step to obtain the complete time evolution of the system .
2 shows temperature and laboratory energy density profiles calculated with sod s method for the expansion of ( a , b ) a sphere and ( c , d ) a bjorken cylinder ( for initial time @xmath85 ) with an ideal gas equation of state @xmath86 in comparison to profiles generated with the semi - analytic method of characteristics @xcite .
that method is a benchmark test for numerical algorithms as long as the hydrodynamical solution is continuous @xcite .
the grid spacing for the hlle scheme is taken as @xmath87 , the time step width for the hlle scheme and the corrector step ( [ correct ] ) is @xmath88 .
as one observes , agreement is excellent ( even on a logarithmic scale ) which gives confidence that sod s method as described above works for the more complicated equation of state ( [ eos ] ) as well ( which leads to discontinuous hydrodynamical expansion solutions for @xmath11 @xcite ) .
in this section we present the hydrodynamical expansion solutions for the fireball and bjorken cylinder geometry for different initial energy densities @xmath10 .
we assume @xmath10 to be homogeneous throughout the system .
we compare solutions for the equation of state ( [ eos ] ) with @xmath6 and @xmath42 with solutions for an ideal gas of equation of state with @xmath7 degrees of freedom .
we explicitly show results for @xmath89 and , where necessary , comment on the difference to the case @xmath90 . as in @xcite
we will use the notion `` lifetime '' for the intercept of a particular isotherm with the @xmath74axis .
note that in our comparison we fix the initial _ energy density _ rather than the initial temperature or entropy density , as sometimes assumed @xcite , because the latter are derived thermodynamic quantities while the former is determined by non - equilibrium energy loss mechanisms .
for instance , at rhic energies it is expected @xcite that semi - hard perturbative qcd processes determine the initial energy density in the range @xmath91 gev/@xmath92 .
this case is rather similar to the one dimensional landau expansion studied in @xcite .
differences are solely due to the spherical geometry ( i.e. , due to the extra terms ( [ fgfb ] ) in the equations of motion ) . in fig .
3 we show temperature profiles and isotherms in the @xmath93 plane for @xmath94 . of course , for this initial energy density , the case @xmath6 ( a , b ) is identical to the ideal gas case ( e , f ) .
note the delayed expansion in the case @xmath19 , figs . 3 ( c , d ) , due to the reduction in the velocity of sound , cf .
fig . 1 ( d ) .
this is quite similar to the one dimensional expansion ( cf .
figs . 3 ( c , i ) of ref .
@xcite ) , although the overall scale of the lifetimes is now considerably reduced since the system has two more spatial dimensions into which it can expand . fig .
4 shows the situation for an initial energy density @xmath95 which is close to @xmath96 . as in the one dimensional expansion ( cf .
fig . 5 of @xcite )
, the lifetime of the system is very long in the case @xmath6 , due to the small velocity of the deflagration front converting mixed phase matter into hadrons . as compared to the lifetime of the mixed phase in the one dimensional expansion , @xmath97 ( cf .
fig . 5 ( g ) of @xcite ) , that lifetime is now , however , only about @xmath98 , due to the more rapid expansion in three spatial dimensions . as in the one dimensional case ,
cooling is faster for @xmath19 ( cf .
fig . 5 ( i ) of @xcite ) .
while in that case , however , the lifetimes were only slightly shorter than for @xmath6 , now they are reduced by about a factor of 3 .
the fastest expansion is that for the ideal gas , in accord with the discussion of fig . 1 ( d ) . in fig .
5 the initial energy density is well above the mixed phase region , @xmath99 . as for the analogous one
dimensional expansion solution ( fig . 7 of @xcite ) the internal energy density and pressure are so large that , in the case of a transition to the qgp , figs .
5 ( a d ) , the system explodes rather than burns slowly as in the preceding case , fig .
nevertheless , the expansion is still somewhat delayed as compared to the expansion of an ideal gas , figs .
5 ( e , f ) . for the discussion of the expansion in the bjorken cylinder geometry it is instructive to first focus on purely longitudinal expansion .
longitudinal boost invariance with regard to boosts with @xmath78 $ ] @xcite implies that the longitudinal velocity is given by @xmath100 . in turn
, the longitudinal fluid rapidity is identical with the boost rapidity @xmath101 .
it is easy to show that in this case the hydrodynamical equations ( [ eom ] ) reduce to @xmath102 where @xmath103 is the proper time associated with fluid elements moving with @xmath100 .
the second equation implies that pressure gradients vanish along space
time hyperbolas @xmath104 and , for baryon - free matter , that the temperature is constant on these hyperbolas .
the first equation is an ordinary differential equation on the space
time hyperbolas and describes the cooling of the system on account of the longitudinal motion . for baryon - free matter
, the thermodynamical identities @xmath105 and @xmath106 imply the very simple cooling law @xmath107 which has the solution @xmath108^{-1}$ ] , _ independent _ from the underlying equation of state . for an ultrarelativistic ideal gas equation of state @xmath86 , @xmath109 ,
the solution of the longitudinal boost - invariant expansion problem is @xmath110^{-4/3 } \,\
, , \,\,\,\ , t(\tau ) = t_0 \ , [ \tau/\tau_0]^{-1/3 } \,\ , .\ ] ] for the equation of state ( [ eos ] ) with @xmath29 , one can easily solve ( [ bj ] ) analytically . if @xmath111 we have @xmath112^{-4/3 } + b \,\ , , & \tau_0 \leq \tau \leq \tau_q\,\ , , \\ ( \epsilon_q + p_c)\ , [ \tau/\tau_q]^{-1 } - p_c \,\ , , & \tau_q < \tau \leq \tau_h \,\ , , \\ \epsilon_h\ , [ \tau/\tau_h]^{-4/3 } \,\ , , & \tau > \tau_h \,\ , , \end{array } \right . \\
t(\tau ) & = & \left\ { \begin{array}{ll } t_0\ , [ \tau/\tau_0]^{-1/3 } \,\ , , & \tau_0 \leq \tau \leq \tau_q \,\ , , \\
t_c \,\ , , & \tau_q < \tau \leq \tau_h \,\ , , \\ t_c\ , [ \tau/\tau_h]^{-1/3 } \,\ , , & \tau > \tau_h \,\ , .
\end{array } \right . \end{aligned}\ ] ] here , @xmath113^{-3/4}$ ] is the ( proper ) time the system enters the mixed phase , and @xmath114 the time corresponding to entry of the hadronic phase .
note that the time spent in the mixed phase is linear proportional to the ratio @xmath28 .
also , the system does not cool at all in this phase , @xmath115 , due to the fact that no mechanical work is performed ( cf.the above discussion of fig . 1 ( d ) ) .
as another consequence , the energy density does not decrease with @xmath116 as in the qgp and hadronic phase , but only proportional to @xmath117 , precisely as in a ( one dimensional ) free - streaming expansion ( where no mechanical work is performed as well )
. for @xmath118 , the solution reads @xmath119^{-1 } - p_c \,\ , , & \tau_0 \leq \tau \leq \tau_h \,\ , , \\ \epsilon_h\ , [ \tau/\tau_h]^{-4/3 } \,\ , , & \tau > \tau_h \,\ , , \end{array } \right . \\
t(\tau ) & = & \left\ { \begin{array}{ll } t_c \,\ , , & \tau_0 \leq \tau \leq \tau_h \,\ , , \\
t_c\ , [ \tau/\tau_h]^{-1/3 } \,\ , , & \tau > \tau_h \,\ , .
\end{array } \right .
\label{t}\end{aligned}\ ] ] here , @xmath120 . for @xmath121 , the solution is identical to ( [ idgas ] ) . for finite @xmath2
there is , contrary to expectation , also a simple , semi - analytic solution which does _ not _ require to solve the equations of motion ( [ bj ] ) explicitly . due to the fact that the entropy density behaves as @xmath108^{-1}$ ] irrespective of the equation of state , and since the equation of state ( [ eos ] ) establishes ( for finite @xmath2 ) a one - to - one correspondence between temperature and entropy density ,
there is also a one - to - one correspondence between temperature and @xmath122 .
once @xmath123 and @xmath124 are known , one can easily calculate @xmath125 and @xmath126 from fundamental thermodynamical relationships . for illustrative purposes ,
we show in fig .
the time evolution of ( a ) energy density , ( b ) entropy density , ( c ) pressure , and ( d ) temperature assuming an initial energy density @xmath127 and @xmath89 .
solid lines are for @xmath6 , dotted lines for @xmath19 , and dashed lines for an ideal gas with @xmath7 degrees of freedom .
the change in the cooling law for the energy density when entering the mixed phase is clearly visible in fig .
6 ( a ) . also , fig . 6 ( b )
illustrates that the cooling law for the entropy density is independent from the equation of state , although the magnitude of the entropy density differs for the different equations of state because of our assumption of fixed initial energy density .
we mention that for @xmath19 , the parts of the system with temperature @xmath128 cool _ faster _ than for the ideal gas , cf .
fig . 6 ( d ) .
only parts cooler than @xmath129 survive longer than in the ideal gas case .
this is in agreement with the results of @xcite for the landau expansion model , where it was concluded that it is therefore less likely that electromagnetic radiation is a viable signature for the transition , in contrast to the conclusions of @xcite . on the other hand ,
it was speculated in @xcite ( referring to the conclusions of @xcite ) that , if the system freezes out at temperatures @xmath130 , the delayed expansion might be observable in side and outwards radii of two particle correlation functions .
this indeed happens to be the case , as will be shown in the next section . for given initial energy density @xmath10 and proper time @xmath131
, one can ask the question , how much longer it takes a system described by the equation of state ( [ eos ] ) to reach a certain freeze - out temperature @xmath132 in comparison to the expansion of an ideal gas ( [ idgas ] ) . since the ideal gas is supposed to freeze out at the same temperature @xmath133 and to have @xmath7 massless degrees of freedom , particle and entropy density at freeze - out are the same in both cases . for
@xmath6 one can easily answer this question analytically : for @xmath134 , the freeze - out time @xmath135 is prolonged by a factor @xmath136^{1/4}$ ] . for @xmath40 ,
the delay in cooling is about 87% ( as one can also see in fig .
6 ( d ) ) , for @xmath90 it is only about 32% .
note that this prolongation of the lifetime of the system is in agreement with the one for the landau expansion model at very high initial energy densities ( cf .
@xcite and fig .
10 below ) . only for initial energy densities around @xmath44 is the prolongation of the lifetime larger in the landau expansion ( see fig .
10 ) due to the fact that there exists a slow deflagration solution which delays the expansion even further .
such a solution does not occur in the purely longitudinal bjorken expansion .
this changes when we consider transverse motion as well . in the following we fix @xmath137
this choice is motivated by the fact that for au+au collisions at rhic , the transverse radius of the hot zone is of the order 5 fm , while the time scale of local equilibration is roughly given by the energy loss of a parton in strongly interacting matter , @xmath138 fm @xcite .
we have , however , also considered the case @xmath139 and @xmath140 ( which is motivated by an uncertainty principle argument ) and will comment on differences to the choice @xmath141 where necessary .
let us first consider the case @xmath142 , cf .
7 . due to the strong initial longitudinal motion
, the system cools rather quickly below temperatures of @xmath143 , even before the transverse rarefaction wave reaches the center of the cylinder .
the lifetimes are therefore solely determined by the longitudinal scaling expansion .
this causes the horizontal parts of the isotherms . for the @xmath144 isotherm and @xmath6 one expects from eq .
( [ t ] ) the lifetime @xmath145 , in good agreement with figs . 7 ( a , b , e , f ) .
( due to the numerical dissipation adherent to any finite difference scheme that solves the hydrodynamical equations , the system cools slightly faster in the numerical calculation , cf . figs . 7 ( b , f ) . )
the effect of longitudinal cooling is reduced for larger values of @xmath131 , cf .
( [ idgas ] ) .
for instance , in the case @xmath139 the center still has @xmath146 when the transverse wave reaches it .
the lifetimes for @xmath147 are therefore ( as expected ) a factor of 5 longer . in fig .
we show the situation for @xmath148 . as in the preceding case
, it takes some time before the transverse expansion wave reaches the center .
thus , the lifetimes of the highest temperatures in the system are solely determined by the longitudinal scaling solution presented above ( horizontal parts of the respective isotherms ) .
it is interesting to note that the lifetimes for @xmath6 and this particular @xmath10 are not that exceedingly long as they were in the case of one dimensional and spherical expansion .
these long lifetimes were due to the slow velocity of the deflagration front .
a deflagration solution , however , exists only for energy densities in the mixed phase , where matter is thermodynamically anomalous @xcite . here
, the strong longitudinal motion cools the system quickly below the respective energy densities , such that the transverse expansion proceeds as a ( comparatively fast ) simple rarefaction wave instead of the deflagration .
thus , the system cools even more quickly and the lifetimes are considerably reduced .
this effect is , however , compensated when higher initial energy densities are considered . in fig .
9 we show the solutions for @xmath149 .
this case is close to the initial conditions expected from mini - jet production at rhic energies .
now the initial energy density is sufficiently high that for @xmath6 the plasma enters the mixed phase at relatively late time ( @xmath150 fm for @xmath151 fm , see also fig .
6 ( d ) ) .
therefore , the time spent in the mixed phase is long enough in this case to allow a transverse deflagration front to form , cf . figs .
9 ( a , b ) . in comparison to the ideal gas case ( e , f ) , where no such solution exists , the expansion is therefore considerably prolonged .
such a time delay is observed also for @xmath19 , figs .
9 ( c , d ) , and is here due to the reduction of the velocity of sound in the transition region .
we note that for @xmath139 , the initial energy density where one observes this effect is smaller , @xmath152 .
on the other hand , for @xmath153 the initial time decreases with increasing initial energy density , enhancing the effect of longitudinal cooling , such that @xmath10 has to be very high ( @xmath154 ) for the onset of this effect .
then , however , a steady increase of the initial energy density affects a steady decrease of @xmath131 such that one observes a prolongation of the lifetimes over a wider range of energy densities ( see also fig . 13 below ) . it
is furthermore interesting to note that this phenomenon occurs at about the same initial energy densities ( in units of @xmath155 ) in the case @xmath37 ( cf .
finally , for very high energy densities , one recovers the case of the violent explosion , and the lifetimes are again reduced .
this happens at lower initial energy densities for the case @xmath156 , since the longitudinal motion is less effective in cooling the system .
for @xmath153 we have not found a decrease in the lifetime up to the highest initial energy density studied by us , @xmath157 .
the reason is , as mentioned above , that the increasingly more efficient cooling due to longitudinal motion compensates the increasing tendency of the system to explode transversally .
in this section we first discuss the lifetimes of differently hot parts of the system ( defined as the intercept of a particular isotherm with the @xmath74axis ) as a function of the initial energy density . assuming that the system freezes out at these temperatures
, we then show how the lifetimes can be inferred from two particle correlation functions . for convenience ,
we first show in fig .
10 the lifetimes as a function of @xmath10 for the one dimensional ( landau ) expansion studied in @xcite . figs . 10 ( a ) and ( b ) are for the case @xmath89 , ( c ) and ( d ) are for @xmath90 . the thick lines in ( a
, c ) show the lifetimes of matter with @xmath158 ( solid line ) , @xmath159 ( dotted ) , and @xmath160 ( dashed ) for @xmath6 , the corresponding lines in ( b , d ) are for finite @xmath19 . for comparison ,
the thin lines are the corresponding lifetimes for the expansion of an ideal gas with @xmath7 degrees of freedom ( they are identical in ( a ) and ( b ) as well as in ( c ) and ( d ) ) .
one clearly observes the maximum in the lifetime of mixed phase matter ( @xmath160 ) at @xmath161 in figs .
10 ( a , c ) .
as discussed in @xcite , this maximum vanishes for finite @xmath2 , figs . 10 ( b , d ) .
moreover , such hot matter does not survive as long as in the ideal gas case .
the lifetime of cooler matter with @xmath158 , however , stays longer than in the expansion of an ideal gas , independent of the value of @xmath2 .
note the change in the scale of @xmath162 by a factor of 3 between ( a , b ) and ( c , d ) .
this is due to the reduction of the latent heat for a smaller ratio @xmath28 , which in turn accelerates the expansion .
11 shows the corresponding diagram for the spherical expansion .
first of all , one notices that , at high @xmath10 , there is almost no difference between the lifetime of a system with a transition to the qgp and an ideal gas . inspecting fig . 5 , one observes , however , that the bulk of matter at finite @xmath69 does indeed live longer in the first case .
thus , for the spherical expansion , our definition of ` lifetime ' should rather be replaced by an average over the particular isotherm .
it is , however , not necessary to do so at this point , since the correlation functions considered subsequently will take this into account in a natural way .
second , one observes that the lifetimes do not grow as strongly for high @xmath10 as in the one dimensional expansion .
this is due to the fact that the system disperses its initial internal energy much more efficiently into kinetic energy in three dimensions than it does in one dimension .
this also leads to the reduction of the overall scale in the lifetimes as compared to fig .
third , it is noticeable that the _ increase _ of the lifetime for @xmath6 at @xmath163 ( where the lifetime is maximum ) as compared to the ideal gas case is about a factor of 2 bigger than in the one dimensional expansion .
the prolongation of the lifetime is thus most pronounced in spherical geometry .
unfortunately , the reduction of the lifetime in the case of a smooth transition is also rather strong , cf.figs .
11 ( b , d ) .
nevertheless , for freeze - out at @xmath158 the lifetime can still be longer by a factor of 2 in the case of a transition than in the expansion of an ideal gas .
moreover , in contrast to the one dimensional case , the spherical geometry leads to a ( broad ) maximum in the lifetime around @xmath44 . in fig .
12 we present the lifetimes for the expansion of a bjorken cylinder with initial time @xmath137 .
one still observes the distinguished maximum in the lifetime associated with the transition to the qgp as seen in the preceding fig .
11 . in this case , however , the overall scale is even smaller ( due to the fact that the system is from the very beginning quite effectively diluted by the longitudinal velocity field ) . moreover , one observes a shift in the maximum of the lifetime .
this was discussed in the preceding section and is due to the fact that the longitudinal motion compensates the tendency to explode transversally ( on account of a high initial energy density ) . for @xmath6
the maximum is now around @xmath164 for _ both _ values of @xmath28 . for @xmath19 ,
one still observes a maximum in the lifetimes , but as in fig .
11 it is also broader and less pronounced .
the corresponding plot for @xmath156 looks rather similar , and will therefore not be shown here .
the position of the maximum is , however , shifted to smaller @xmath165 , for reasons discussed above .
the situation for a dynamical @xmath153 is shown in fig .
13 . as one expects from the above discussion
, there is a steady increase of the lifetime up to the highest considered @xmath10 . to calculate two particle correlation functions we use the method developped by pratt @xcite
, sinyukov @xcite and others , and applied to hydrodynamics by the marburg group @xcite .
this method is essentially a straightforward generalization of the cooper
frye formula @xcite for single inclusive particle spectra to the case of correlation functions . to calculate single inclusive particle spectra or two particle correlation functions in the hydrodynamical framework
, one assumes that a fluid element decouples from the fluid evolution ( `` freezes out '' ) as soon as the particle density drops below a certain critical value where the collision rate becomes too small to maintain ( local ) thermodynamical equilibrium . in our case ,
the particle density depends on the temperature @xmath166 only .
therefore , freeze - out happens across isotherms in the space time diagram . to incorporate the freeze - out consistently into the solution of the hydrodynamical equations is a non - trivial problem .
hydrodynamics is obviously not applicable to determine the motion of particles that are already frozen out .
therefore , the hydrodynamical solution has to be restricted to a limited region of space time .
the main difficulty is that its boundary , i.e. , the freeze - out hypersurface , has to be determined _ dynamically _ , i.e. , simultaneously with the solution of the hydrodynamical equations .
a possible treatment of this problem was recently proposed by bugaev @xcite , but has so far not been applied in practical calculations .
commonly one circumvents this problem assuming the validity of the hydrodynamical description in the _ whole _ forward light cone , then solves the hydrodynamical equations , and determines the isotherms . finally , one employs the cooper
frye formula @xcite to calculate particle spectra along the isotherm corresponding to the particular freeze - out temperature ( for a detailed discussion of this approach see , for instance , @xcite ) .
we note that the cooper frye formalism holds rigorously only on space - like hypersurfaces . for time - like hypersurfaces it may yield negative numbers of particles emitted from the hypersurface ( corresponding to particles that do not freeze out but reenter the fluid ) . to cure this obviously unphysical result ,
modifications have been proposed in @xcite .
the approach of ref .
@xcite is , however , problematic since it does not describe correctly radiation from a static source .
when modifying the particle spectra as proposed in @xcite we found , however , for the cases under consideration virtually no deviation to the results obtained with the conventional approach of cooper and frye . therefore , we will use this well - established method throughout the following .
the two particle correlation function measures the coincidence probability @xmath167 of two ( identical ) particles with momenta @xmath168 relative to the probability of detecting uncorrelated particles from different events , @xmath169 in the following , the average 4momentum is denoted as @xmath170 and the relative 4momentum as @xmath171 . under the assumption that the particle source is chaotic and sufficiently large , and that the emitted particles are bosons ( with degeneracy factor @xmath172 ) the two particle correlation function can be written as @xcite @xmath173 \,\ , f\left(\frac{u \cdot k}{t}\right ) \right|^2 } { e_1 \ , \frac{{\rm d}n}{{\rm d}^3 { \bf p}_1 } \,\,\ , e_2\ , \frac { { \rm d}n}{{\rm d}^3 { \bf p}_2 } } \,\ , , \ ] ] where @xcite @xmath174 is the single inclusive momentum distribution .
@xmath175 is the bose
einstein distribution function , and @xmath176 the fluid 4velocity .
the integrals run over the freeze - out hypersurface . in general , that hypersurface
is represented by a 3parametric ( 4vector ) function @xmath177 , and the normal vector on the hypersurface is determined by @xmath178 where @xmath179 ( @xmath180 for @xmath181 an even permutation of @xmath182 ) is the completely antisymmetric 4tensor . while the ( freeze - out ) temperature @xmath166 is ( by definition ) assumed to be constant along the freeze - out hypersurface ,
the fluid velocity varies , @xmath183 . for symmetric systems ,
the number of independent variables which @xmath184 depends on can be reduced .
for the spherically symmetric fireball geometry , the orientation of the average momentum can be chosen arbitrarily , such that @xmath184 depends only on the modulus of @xmath185 .
our choice of coordinate system will be such that @xmath186 .
furthermore , @xmath187 can be decomposed into a so - called `` out '' component @xmath188 parallel to @xmath185 and a `` side '' component @xmath189 orthogonal to @xmath185 and @xmath190 .
rotational symmetry around the direction of @xmath185 allows us to choose @xmath191 , such that the correlation function depends only on three independent variables , @xmath192 .
for the bjorken cylinder geometry , we restrict our consideration to particles emitted at midrapidity , @xmath193
. rotational symmetry around the @xmath77axis allows us to choose the average transverse momentum as @xmath194 , and consequently , @xmath195 , @xmath196 .
again , @xmath197 is a function of three independent variables only . the explicit evaluation of ( [ c2 ] ) for the fireball and bjorken cylinder geometry is referred to appendices a and b , respectively .
as is well known @xcite , the width of the correlation function in out
direction is inversely proportional to the duration of particle emission , i.e. , to the lifetime of the source .
analogously , the inverse width of the correlation function in side
direction is a measure for the ( transverse ) size of the source . to be more precise , for fixed average ( transverse ) momentum @xmath198 we define side and out correlation functions as @xmath199 and @xmath200 , respectively .
we furthermore define a corresponding inverse width as @xmath201 , where @xmath202 is determined by @xmath203 , and analogously for @xmath17 .
we emphasize that we do not attempt a standard gaussian fit of the two particle correlation function , which would relate the inverse widths to the usual radii parameters .
first of all , we will be interested only in the generic shape of the correlation functions which is satisfactorily characterized by their inverse widths @xmath204 .
second , the functional form of the correlation functions is not a gaussian ( cf . the explicit formulae in appendix a and b ) .
the fit procedure would therefore only introduce unnecessary errors .
third , the radii parameters as well as the inverse widths are only _ proportional _ to the actual ( average ) size and lifetime of the system .
it is well known that hydrodynamical flow affects the radii parameters / widths , and , moreover , that this effect is sensitive to the choice of @xmath198 @xcite .
it is therefore tedious ( if not impossible ) to relate @xmath16 and @xmath17 to the real source size and lifetime .
this holds for our model calculations as well as for the experiment .
we can expect , however , that such effects are either irrelevant or largely cancel out if we consider the ratio @xmath205 . as can be inferred from the space
time diagrams in figs .
35 and 79 , while the size of the system is approximately constant , it is the lifetime that varies appreciably , depending on whether the system undergoes a phase transition or not .
thus , @xmath205 can be expected to be a good measure for the lifetime of the system . in the following ,
we choose the particles to be pions with mass @xmath206 mev , and as average momentum @xmath207 mev . on one hand
, this value is well within the typical experimental acceptance . on the other hand
, it is large enough to exhibit effects of the prolonged lifetime in the case of a transition to the qgp @xcite . to fix the @xmath208scale in mev we take @xmath209 fm as a typical initial radius of the system . in fig .
14 we show as an example side and out correlation functions corresponding to the hydrodynamical evolution of fig . 4 .
in light of the uncertainties in the actual freeze - out temperature , we calculate the correlation functions along isotherms for @xmath210 , as well as @xmath211 .
first of all , one immediately recognizes that in all cases the widths of the correlation functions correspond closely to the space
time structure of the corresponding isotherms .
for instance , the exceedingly long lifetime of the system in fig . 4 ( b )
is reflected in the comparatively small width of the out
correlation function fig . 14 ( b ) .
on the other hand , the transverse size is about the same in all cases , cf .
figs . 4 ( b , d , f ) , which reflects in nearly identical side correlation functions in figs . 14 ( a , c , e ) .
furthermore , with the exception of the case @xmath19 , the correlation functions are almost completely insensitive to the freeze - out temperature chosen .
this is intuitively clear since the corresponding space
time isotherms differ only marginally for @xmath6 and the ideal gas calculation , cf .
figs . 4 ( b , f ) , while there are larger differences in fig . 4 ( d ) .
note that the results shown in figs .
14 ( a , b , e , f ) correspond nicely to those in fig . 4 of @xcite , in spite of the fact that the latter calculation employs inhomogeneous initial conditions , a different way to solve the hydrodynamical equations , and a different treatment of the freeze - out . in fig .
15 we show the correlation functions corresponding to the hydrodynamical evolution of fig . 9 .
again , they adequately characterize the space time geometry of the source .
16 shows the experimentally measurable ratio @xmath205 as a function of @xmath10 for the spherical fireball geometry . comparing the results with fig . 11
, one observes that this ratio reflects closely the behaviour of the lifetime of the system , independent of details in the equation of state such as the width of the transition region @xmath2 or the latent heat of the transition ( which is proportional to @xmath28 ) . also , for the case of a first order transition , @xmath6 , figs .
16 ( a , c ) , the enhancement in @xmath205 over the ideal gas case is a factor of 3 to 7 ( for @xmath212 to @xmath213 ) at @xmath214 . in the case of a smooth transition , @xmath19 , figs.16 ( b , d ) ,
this is considerably reduced ( as expected from fig .
11 ) , but if the system freezes out at temperatures @xmath215 , there is still a factor of 2 enhancement over the ideal gas case . in figs .
17 , 18 we present the corresponding results for the bjorken cylinder expansion with @xmath141 and @xmath216 , respectively . in all cases
we find that the experimentally measurable ratio of correlation widths mirrors closely the dependence of the lifetime on initial conditions in figs . 12 , 13 . the most favourable case @xmath217 ( for a strong first order transition with a large latent heat , @xmath89 , cf .
17 ( a ) ) , may be reached with initial conditions expected at rhic energies . at that point
the enhancement of this ratio is about a factor of 2 above the ideal gas case and virtually independent of the freeze - out temperature . in the scenario with @xmath216 , fig .
18 ( a ) , the enhancement is somewhat smaller and is shifted toward higher initial energy densities because the initial pure bjorken expansion phase starts earlier than in the fixed @xmath137 case . for a smooth transition with @xmath19 , the maximum ratio is reduced to 2.5 and varies less rapidly with @xmath10 , but is still about @xmath218 larger than for the ideal gas expansion . in this case , however , a significant time delay can only be observed if the freeze - out occurs relatively late with @xmath219 . as seen in figs .
17 , 18 , for earlier freeze - out , smaller @xmath131 , or smaller @xmath28 the enhancement relative to the ideal gas case is significantly reduced and would be more difficult to observe . at energy densities estimated to be reached in cern sps pb+pb collisions ( @xmath220 in our units ) , we expect from our results that @xmath221 .
however , present data from cern sps @xcite indicate that the ( fitted ) out radii are rather similar to the side radii .
this does not contradict our results , because , as shown by schlei et al .
@xcite in the framework of a hydrodynamical calculation similar to ours , correlation functions constructed from thermal pions only give @xmath222 ( cf . especially @xcite ) , while the incorporation of long - lived _ resonance decays _ leads to a reduction of that ratio and good agreement with the measured radii .
we note that kaon interferometry @xcite is preferable , though experimentally more difficult , because only distortions of the interference pattern due to shorter lived @xmath223 resonances have to be taken into account .
in this paper we investigated the spherically symmetric expansion of a fireball as well as the cylindrically symmetric transverse expansion of a qgp with boost - invariant initial conditions along the beam axis .
the expansion was treated in the framework of ideal relativistic hydrodynamics and extends our systematic study @xcite of collective flow patterns with realistic equations of state .
the symmetries of the considered geometries allowed us to calculate the flow patterns using a simple modification of a well - tested one dimensional algorithm .
the emphasis of the present investigation was on how a rapid cross - over to the qgp in the equation of state influences the collective expansion dynamics .
in particular , since present lattice data only constrain the width of the transition region to the qgp to be in the range @xmath224 , it is important to test how such uncertainties may influence dynamical observables .
we also studied the effect of varying the ratio of degrees of freedom in the qgp and hadron phase ( i.e. essentially the latent heat of the transition ) on the system dynamics .
the results were compared to the expansion of an ultrarelativistic ideal gas ( without transition ) using fixed energy density initial conditions .
we focussed on the lifetime of the system as a function of initial energy density as an important _ collective _ observable that can discriminate between different equations of state .
as expected from previous one dimensional studies in the framework of the landau expansion model @xcite , we found that the lifetime of a spherical fireball is much longer in the case of a first order phase transition , @xmath6 , as compared to the expansion of an ideal gas without transition .
the prolongation of the lifetime in that case can be up to a factor of 4.5 to 9 ( for @xmath90 to @xmath213 , respectively ) , provided the initial energy density corresponds to that of mixed phase with a large fraction of qgp .
this time delay effect was originally pointed out by pratt @xcite . in the case of a smooth transition , @xmath225 ,
however , this time delay is drastically reduced . the lifetimes
( of matter with @xmath158 ) nevertheless still remain on the order of a factor of 2 longer as compared to the ideal gas expansion . for the bjorken cylinder expansion , it is also necessary to specify the initial ( proper ) time @xmath131 for the onset of hydrodynamic expansion . to explore uncertainties
associated with this additional degree of freedom , we investigated the cases @xmath226 , and a dynamical @xmath227 varying with the initial temperature .
the results were similar as for the spherical expansion , up to two important exceptions : ( a ) the maximum lifetimes emerged at higher initial energy densities ( the exact value of which depends on the choice of @xmath131 ) corresponding to qgp matter instead of mixed phase matter , and ( b ) the lifetimes were in general shorter .
both effects are explained by the very efficient cooling due to the initial longitudinal velocity profile associated with the boost invariance of the problem .
this effect causes an overall reduction of the lifetimes .
moreover , in order for slow ( van hove @xcite ) deflagration fronts to dominate the cooling mechanism in transverse direction , one has to start at higher initial energy densities to compensate for the longitudinal cooling .
otherwise , the longitudinal cooling reduces the energy densities too fast and the associated deflagration solution @xcite vanishes .
further cooling is then achieved ( in the usual way ) through a ( fast ) simple wave .
finally , we showed ( cf .
also @xcite ) that the prolongation of the lifetime can be observable via the ratio @xmath205 of inverse widths of two particle correlation functions in out and side direction .
this ratio follows the behaviour of the lifetimes rather closely
. the prolongation of the lifetime in the case of a transition to the qgp could therefore be in principle searched for using this observable .
the enhancement of that ratio is , of course , strongest in the case that the transition is first order with a large latent heat .
an interesting result is that , for the fireball geometry , the effect is maximum for energy densities achieved at the ags , while for the bjorken cylinder geometry , the maximum of @xmath205 occurs at energy densities presumably reached at the rhic collider .
there are several effects which may reduce the strength of the time
delay signal observable via the @xmath205ratios that will require further investigation .
first , for the fireball geometry it is important to extend our considerations to finite baryon number density . at finite chemical potential
the width of the cross - over region may be significantly larger than at zero chemical potential .
second , the decay of long - lived resonances can simulate time delay @xcite .
interferometry with kaons instead of pions is therefore preferable @xcite .
finally , while our investigations covered a wide range of uncertainties in the equation of state , our calculations have neglected effects of dissipation that tend in general to reduce the collective flow strengths predicted via ideal hydrodynamics .
for instance , bulk viscosity appears in the hydrodynamical equations of motion in a similar way as the pressure , and could in principle counteract any reduction of the velocity of sound in the transition region .
the main result of this paper is , nevertheless , that the generic time delay signature of qgp formation is remarkable robust to present uncertainties in the qcd equation of state
. + + * acknowledgments * + + we thank g. bertsch , a. dumitru , and e. shuryak for stimulating discussions on the lifetime of the qgp .
one of us ( d.h.r . ) benefitted considerably from discussions with b. schlei on the calculation of two particle correlation functions in the hydrodynamical framework .
we thank u. heinz and w. zajc for useful discussions on our results and d. keane for emphasizing that kaons might be the ideal probe to detect a prolonged lifetime with the star detector at rhic .
in this appendix we explicitly calculate the out and side correlation functions in the fireball geometry .
we first specify the parameters of the freeze - out hypersurface .
for the following discussion , we conveniently choose spherical coordinates @xmath228 in space .
let ( a ) @xmath229 parametrize the surface in the @xmath93 plane , @xmath230 , with the lower boundary corresponding to @xmath231 and the upper to @xmath232 .
let ( b ) @xmath101 parametrize the surface in the @xmath233 plane and ( c ) @xmath234 in the @xmath235 plane . in the @xmath93 plane ,
the surface is given by the corresponding isotherm shown in figs .
35 . due to spherical symmetry , however , the surface in the @xmath233 and @xmath235 planes is trivial .
we can simply identify @xmath236 , @xmath237 and vary them within the standard boundaries for the azimuthal and polar angle . as a consequence ,
the time @xmath238 and the radius @xmath239 at freeze - out depend only on @xmath229 , not on @xmath101 or @xmath234 .
the freeze - out surface is thus given by @xmath240 , where @xmath241 .
applying ( [ normal ] ) and choosing the sign to have @xmath242 pointing outwards along the isotherm , we obtain @xmath243 for the calculation of the correlation function we employ the boltzmann approximation . for an average momentum @xmath207 mev chosen in our calculations
the error introduced is negligible , in particular since we also consider the particles to be massive pions , @xmath244 mev , and choose @xmath245 mev . for the single inclusive pion spectrum ( [ single ] ) ,
one conveniently chooses @xmath246 , and obtains after inserting ( [ dsigma ] ) and performing the @xmath234 and @xmath101integrations @xmath247\ , \frac{{\rm d}t_f}{{\rm d}\zeta } \right\ } \ , \ , , \ ] ] where @xmath248 , @xmath249^{1/2}$ ] , and @xmath71 is the ( radial ) fluid 3velocity , @xmath250^{-1/2}$ ] . here
( and in the following ) , the @xmath229integration has to be done numerically along the respective isotherms obtained from the solution of the hydrodynamical equations . for further use , we define @xmath251 for the side correlation function , @xmath252 , and consequently @xmath253 where @xmath254^{1/2}$ ] .
since the single inclusive spectrum ( [ single2 ] ) does not depend on the direction of @xmath255 , the two single inclusive spectra in the denominator in ( [ c2 ] ) are equal .
furthermore , for the numerator one has to calculate the expressions @xmath256\,\ , , \\ { \cal i}_2 & \equiv & \int_{\sigma } { \rm d}\sigma \cdot k \ , e^{-u \cdot k /t } \ , \sin\ , [ \sigma \cdot q ] \,\ , .\end{aligned}\ ] ] for the calculation of @xmath257 we insert ( [ dsigma ] ) and perform the @xmath234integration with the help of eq .
( 3.715.18 ) of ref .
subsequently , the @xmath101integration can be done with an analytic continuation of either eq .
( 6.616.5 ) or ( 6.677.6 ) of @xcite and a suitable first derivative of these formulae .
the final result reads @xmath258 \ , \frac{{\rm d}t_f}{{\rm d}\zeta } \right\ } \,\
, , \nonumber\end{aligned}\ ] ] where @xmath259 , @xmath260 .
it is easy to show that for the side correlation function , @xmath261 by symmetry .
the final result thus reads @xmath262 with @xmath263 from ( [ i0 ] ) and @xmath257 from ( [ i1 ] ) . for the out
correlation function , @xmath264 , and consequently @xmath265 where @xmath266^{1/2},\,k^0 = ( e_1+e_2)/2 $ ] .
the @xmath234integration in the calculation of @xmath267 is now trivial .
for the @xmath101integration , we employ the angle addition theorem and eqs .
( 2.662.1,2 ) of @xcite , obtaining @xmath268 \,\frac{{\rm d}r_f}{{\rm d}\zeta } \right .
\nonumber \\ & & \hspace*{4 cm } \left . + k \left [ \cos \beta \ , \frac{\partial { \cal j}_0(a , b)}{\partial a } + \sin \beta \ , \frac{\partial { \cal j}_1(a , b)}{\partial a } \right ] \
, \frac{{\rm d}t_f}{{\rm d}\zeta } \right\ } \ , \ , , \end{aligned}\ ] ] where @xmath269 , and @xmath270 with @xmath271 and @xmath272 . in complete analogy one
obtains @xmath273 as @xmath274 \,\frac{{\rm d}r_f}{{\rm d}\zeta } \right .
\nonumber \\ & & \hspace*{4 cm } \left .
+ k \left [ \sin \beta \ , \frac{\partial { \cal j}_0(a , b)}{\partial a } - \cos \beta \ , \frac{\partial { \cal j}_1(a , b)}{\partial a } \right ] \ , \frac{{\rm d}t_f}{{\rm d}\zeta } \right\ } \ , \ , .\end{aligned}\ ] ] finally , the out
correlation function reads @xmath275
in this appendix we explicitly calculate the out and side correlation functions in the bjorken cylinder geometry .
convenient coordinates to work in are cylindrical coordinates @xmath276 in space . as in the preceding appendix ,
( a ) @xmath229 parametrizes the freeze - out surface in the @xmath93 plane ( at @xmath76 ) , @xmath277 , @xmath278 .
( b ) @xmath101 parametrizes the surface in the @xmath279 plane . due to boost invariance , that surface is simply the space
time hyperbola of constant @xmath280 .
note that at @xmath76 , @xmath281 and therefore @xmath282 depends on @xmath229 .
a natural choice for the parameter @xmath101 is the space time rapidity artanh@xmath283 $ ] ( as implied by our choice of symbols ) , such that @xmath284 , @xmath285 .
finally , ( c ) @xmath234 parametrizes the hypersurface in the @xmath235 plane and due to cylindrical symmetry can be identified with the polar angle , @xmath286 .
we note that due to boost invariance along @xmath77 , @xmath239 can not depend on @xmath101 .
the hypersurface 4vector thus reads @xmath287 with @xmath288 , i.e. , given by the isotherms in figs .
79 . the normal vector ( with proper orientation )
is readily calculated with ( [ normal ] ) , @xmath289 for the calculation of the single inclusive spectrum , we employ @xmath290^{1/2}$ ] , i.e. , @xmath291 , with the longitudinal ( particle ) rapidity @xmath292 $ ] .
furthermore , due to rotational symmetry around @xmath77 , we may choose @xmath293 .
thus , @xmath294 \ , \frac{{\rm d } r_f}{{\rm d } \zeta } + p_\perp \cos \phi \,\ , \frac{{\rm d } \tau_f}{{\rm d } \zeta } \right ) r_f(\zeta ) \ , \tau_f ( \zeta ) \ , { \rm d } \zeta \ , { \rm d } \eta\ , { \rm d } \phi\,\,.\ ] ] the fluid 4velocity in the bjorken cylinder expansion reads @xmath295 , where @xmath296 . with the ( longitudinal ) space time rapidity @xmath297 $ ] and the transverse fluid rapidity @xmath298 $ ] , one obtains @xmath299 note that longitudinal boost invariance implies that @xmath300 can not depend on @xmath101 , and is therefore given by the solution of the hydrodynamical equations at @xmath76 , @xmath301 $ ] , as provided in section 3 . for the single inclusive momentum distribution ( in boltzmann approximation )
we insert ( [ dsigmadotp ] ) and ( [ umu ] ) into ( [ single ] ) , perform the @xmath101integration with the help of eq .
( 3.547.4 ) of @xcite , and the @xmath234integration using the formula ( 3.937.2 ) @xcite , resulting in @xmath302 note that the final spectrum respects the symmetries of the problem , i.e. , it is azimuthally symmetric and boost invariant along @xmath77 .
for further use , let us define @xmath303 for the side
correlation function for particles at @xmath304 we may choose @xmath194 such that @xmath305 where @xmath254^{1/2}$ ] .
as in the previous case of the fireball geometry ( appendix a ) , one has to calculate @xmath257 and @xmath273 . while the latter vanishes again by symmetry , the former reads after using eq .
( 3.547.4 ) of @xcite in the @xmath101integration @xmath306 where @xmath307 , @xmath308 , and the functions @xmath309 ( which have to be evaluated numerically ) are defined by @xmath310\ , \cos\ , [ b\ , \sin \phi ] \,\
, , \\ \hat{i}_1 ( a , b ) & \equiv & \frac{1}{\pi } \int_0^\pi { \rm d } \phi\ , \cos \phi\ , \sinh\ , [ a\ , \cos \phi]\ , \cos\ , [ b\ , \sin \phi ] \equiv \frac{\partial \hat{i}_0(a , b)}{\partial a}\,\ , .\end{aligned}\ ] ] it is obvious ( see eqs .
( 8.411.1 , 8.431.4 ) in @xcite ) that @xmath311 is related to the bessel functions @xmath312 : @xmath313 the final result for the side correlation function reads ( as in the preceding appendix ) @xmath314 with @xmath257 from ( [ i1b ] ) and @xmath263 from ( [ i0b ] ) with the single inclusive spectrum ( [ single3 ] ) . for the out
correlation the choice of momenta is @xmath315 where @xmath266^{1/2},\,k^0 = ( e_1+e_2)/2 $ ] . as in the previous cases , the @xmath101 and @xmath234integrations separate , and the final result for @xmath257 and @xmath273 reads @xmath316 \ , \frac{{\rm d}r_f}{{\rm d}\zeta } \right .
\nonumber \\ & & \hspace*{3.9 cm } \left . + k \,\left[{\cal k}_0 ( \alpha,\beta)\ , { \cal j}_1 ( a , b ) + \hat{\cal k}_0(\alpha,\beta)\ , \hat{\cal j}_1(a , b ) \right]\ , \frac{{\rm d}\tau_f}{{\rm d}\zeta } \right\ } \ , \ , , \label{i1c } \\ { \cal i}_2 & = & 4 \pi \int_0 ^ 1 { \rm d}\zeta \ ,
r_f(\zeta ) \ , \tau_f(\zeta ) \left\ { - k^0\ , \left [ \hat{\cal k}_1(\alpha,\beta)\ , { \cal j}_0 ( a , b ) - { \cal k}_1(\alpha,\beta)\ , \hat{\cal j}_0(a , b ) \right ] \ , \frac{{\rm d}r_f}{{\rm d}\zeta } \right . \nonumber \\ & & \hspace*{3.9 cm } \left . + k \,\left[\hat{\cal k}_0 ( \alpha,\beta)\ , { \cal j}_1 ( a , b ) - { \cal k}_0(\alpha,\beta)\ , \hat{\cal j}_1(a , b ) \right]\ , \frac{{\rm d}\tau_f}{{\rm d}\zeta } \right\ } \ , \ , , \label{i2c}\end{aligned}\ ] ] where @xmath317 , @xmath318 , @xmath319 , @xmath320 , and ( see eqs .
( 8.431.4 , 8.431.5 , 8.432.1 , 8.476.4 ) in @xcite ) @xmath321\ , e^{-\alpha\ , \cosh \eta } \equiv { \rm re}\ , k_0(\alpha - i\beta ) \,\ , , \\ { \cal k}_1 ( \alpha,\beta ) & \equiv & \int_0^\infty { \rm d }
\eta\ , \cosh \eta\ , \cos\ , [ \beta \ , \cosh \eta]\ , e^{-\alpha\ , \cosh \eta } \equiv { \rm re}\ , k_1(\alpha - i\beta ) \equiv - \frac{\partial { \cal k}_0 ( \alpha,\beta)}{\partial \alpha}\,\ , , \\ \hat{\cal k}_0
( \alpha,\beta ) & \equiv & \int_0^\infty { \rm d } \eta\ , \sin\ , [ \beta \ , \cosh \eta]\ , e^{-\alpha\ , \cosh \eta } \equiv { \rm im}\ , k_0(\alpha - i\beta ) \,\ , , \\
( \alpha,\beta ) & \equiv & \int_0^\infty { \rm d } \eta\ , \cosh \eta\ , \sin\ , [ \beta \ , \cosh \eta]\ , e^{-\alpha\ , \cosh \eta } \equiv { \rm im}\ , k_1(\alpha - i\beta ) \equiv - \frac{\partial \hat{\cal k}_0 ( \alpha,\beta)}{\partial \alpha}\,\ , , \\ { \cal j}_0 ( a , b ) & \equiv & \frac{1}{\pi } \int_0^\pi { \rm d } \phi\ , \cos\ , [ b \cos \phi]\ , \cosh\ , [ a\ , \cos \phi ] \equiv { \rm re}\ , i_0(a+ib ) \,\
, , \\ { \cal j}_1 ( a , b ) & \equiv & \frac{1}{\pi } \int_0^\pi { \rm d } \phi\ , \cos \phi\ , \cos\ , [ b \cos \phi]\ , \sinh\ , [ a\ , \cos \phi ] \equiv { \rm re}\ , i_1(a+ib ) \equiv \frac{\partial { \cal j}_0(a , b)}{\partial a } \,\ , , \\
\hat{\cal j}_0 ( a , b ) & \equiv & \frac{1}{\pi } \int_0^\pi { \rm d } \phi\ , \sin\ , [ b \cos \phi]\ , \sinh\ , [ a\ , \cos \phi ] \equiv { \rm im}\ , i_0(a+ib ) \,\ , , \\ \hat{\cal j}_1 ( a , b ) & \equiv & \frac{1}{\pi } \int_0^\pi { \rm d } \phi\ , \cos \phi\ , \sin\ , [ b \cos \phi]\ , \cosh\ , [ a\ , \cos \phi ] \equiv { \rm im}\ , i_1(a+ib ) \equiv \frac{\partial \hat{\cal j}_0(a , b)}{\partial a } \,\ , .\end{aligned}\ ] ] the final result for the out correlation reads @xmath322 as before , cf . ( [ c2out ] ) , but with @xmath323 from ( [ i1c ] ) and ( [ i2c ] ) and @xmath263 from ( [ i0b ] ) with the single inclusive spectrum ( [ single3 ] ) .
99 e. laermann , proc . of `` quark matter 96 '' , may 20 24 , 1996 , heidelberg , germany ( to appear in nuclear physics a ) .
see , for instance : + s.a .
chin , phys . lett .
b 119 ( 1982 ) 51 , + l.d .
mclerran and t. toimela , phys .
d 31 ( 1985 ) 545 , + r.c .
hwa and k. kajantie , phys .
d 32 ( 1985 ) 1109 , + k. kajantie , j. kapusta , l. mclerran , a. mekjian , phys .
d 34 ( 1986 ) 2746 .
t. matsui and h. satz , phys .
b 178 ( 1986 ) 416 .
m. gyulassy and m. plmer , phys .
b 243 ( 1990 ) 432 .
c. greiner , d.h .
rischke , h. stcker , p. koch , phys .
d 38 ( 1988 ) 2797 .
k. rajagopal , f. wilczek , nucl .
b 399 ( 1993 ) 395 .
j. kapusta , p. lichard , d. seibert , phys .
d 44 ( 1991 ) 2774 .
m. gonin , proc . of ``
quark matter 96 '' , may 20 24 , 1996 , heidelberg , germany ( to appear in nucl .
a ) . c. gerschel and j. hfner , nucl .
phys . a 544 ( 1992 ) 513c . for a review ,
see : r.b . clare and d.d .
strottman , phys . rep . 141 ( 1986 ) 177 .
wang , m. gyulassy , m. plmer , phys .
d 51 ( 1995 ) 3436 , + r. baier , yu.l .
dokshitzer , s. peign , d. schiff , phys .
b 345 ( 1995 ) 277 , + r. baier , yu.l .
dokshitzer , a.h .
mueller , s. peign , d. schiff , preprint cern
th96/14 , 1996 ( unpublished ) .
. hung and e.v .
shuryak , phys .
lett . 75 ( 1995 ) 4003 .
rischke , s. bernard , j.a .
maruhn , nucl .
a 595 ( 1995 ) 346 .
rischke and m. gyulassy , nucl .
a 597 ( 1996 ) 701 .
bravina , n.s .
amelin , l.p .
csernai , p. levai , d. strottman , nucl .
phys . a 566 ( 1994 ) 461c .
rischke , y. prsn , j.a .
maruhn , h. stcker , w. greiner , preprint cu
tp695 , nucl
th/9505014 , to be published in heavy ion phys .
j. barrette et al .
( e877 collaboration ) , phys .
73 ( 1994 ) 2532 , nucl .
a 590 ( 1995 ) 259c , + y. zhang and j.p .
wessels ( e877 collaboration ) , nucl .
phys . a 590 ( 1995 ) 557c , + g. rai and the e895 collaboration , lbl pub5399 ( 1993 ) .
landau , izv . akd .
nauk sssr 17 ( 1953 ) 51 , in : `` collected papers of l.d .
landau '' ( ed .
haar , pergamon , oxford , 1965 ) , p. 569585
, + l.d .
landau and s.z .
belenkii , uspekhi fiz .
nauk 56 ( 1955 ) 309 , in : `` collected papers of l.d .
landau '' ( ed .
haar , pergamon , oxford , 1965 ) , p. 665700 .
l. van hove , z. phys .
c 21 ( 1983 ) 93 , + m. gyulassy , k. kajantie , h. kurki suonio , l. mclerran , nucl .
b 237 ( 1984 ) 477 .
bjorken , phys .
d 27 ( 1983 ) 140 .
s. pratt , phys .
c 49 ( 1994 ) 2722 , phys .
d 33 ( 1986 ) 1314 .
g. bertsch , m. gong , m. tohyama , phys .
c 37 ( 1988 ) 1896 , + g. bertsch , nucl .
phys . a 498 ( 1989 ) 173c .
blaizot and j.y .
ollitrault , phys .
d 36 ( 1987 ) 916 .
a. chodos , r.l .
jaffe , k. johnson , c.b .
thorn , v.f .
weisskopf , phys . rev .
d 9 ( 1974 ) 3471 .
neumann , d. seibert , g. fai , phys .
c 51 ( 1995 ) 1460 .
e. shuryak , phys .
lett . 68 ( 1992 ) 3270 , + k.j .
eskola and m. gyulassy , phys .
c 47 ( 1993 ) 2329 . l.d .
landau and e.m .
lifshitz , `` fluid mechanics '' ( pergamon , new york , 1959 ) .
g. baym , b.l .
friman , j.p .
blaizot , m. soyeur , w. czy , nucl .
phys . a 407 ( 1983 ) 541 .
sod , j. fluid mech .
83 ( 1977 ) 785 . v. schneider et al . , j. comput .
phys . 105 ( 1993 ) 92 .
rischke , y. prsn , j.a .
maruhn , nucl .
phys . a 595 ( 1995 ) 383 .
k. kajantie , m. kataja , l. mclerran , p.v .
ruuskanen , phys .
d 34 ( 1986 ) 811 , + m. kataja , p.v .
ruuskanen , l.d .
mclerran , h. v. gersdorff , phys .
d 34 ( 1986 ) 2755 . yu.m .
sinyukov , nucl .
phys . a 498 ( 1989 ) 151c .
schlei , u. ornik , m. plmer , r.m .
weiner , phys .
lett.b 293 ( 1992 ) 275 .
f. cooper and g. frye , phys .
d 10 ( 1974 ) 186 , + f. cooper , g. frye , e. schonberg , phys .
d 11 ( 1975 ) 192 .
bugaev , preprint university hannover , 1996 ( to be published in nucl .
s. bernard , j.a .
maruhn , w. greiner , d.h .
rischke , nucl .
a ( in press ) . yu.m .
sinyukov , z. phys .
c 43 ( 1989 ) 401 .
jacak ( na44 collaboration ) , nucl .
phys . a 590 ( 1995 ) 215c .
j. bolz , u. ornik , m. plmer , b.r .
schlei , r.m .
weiner , phys .
d 47 ( 1993 ) 3860 , + b. schlei , u. ornik , m. plmer , d. strottman , r.m .
weiner , los alamos preprint hep - ph/9509426 .
m. gyulassy and s.s .
padula , phys .
c 41 ( 1990 ) 21 .
schlei , los alamos preprint nucl - th/9605016 .
s. bernard , d.h .
rischke , m. gyulassy , in preparation .
padula and m. gyulassy , nucl .
phys . a 544 ( 1992 ) 537c .
gradsteyn and i.m .
ryzhik , `` table of integrals , series , and products '' , academic press , san diego , 1980 .
* figure captions : * + + * fig . 1 : * ( a ) the entropy density divided by @xmath324 ( in units of @xmath325 ) , ( b ) the energy density divided by @xmath326 ( in units of @xmath327 ) as functions of temperature ( in units of @xmath14 ) , ( c ) the pressure ( in units of @xmath155 ) , ( d ) the square of the velocity of sound as functions of energy density ( in units of @xmath155 ) .
the solid lines correspond to @xmath6 , the dotted curves to @xmath19 .
quantities for the ideal gas equation of state ( with @xmath7 degrees of freedom ) are represented by dashed lines .
the ratio of degrees of freedom in the qgp to those in the hadronic phase is @xmath89 .
the critical enthalpy density is @xmath328 gev @xmath36 for the case @xmath329 .
+ + * fig . 2 : * the expansion of ( a , b ) a fireball and ( c , d ) a bjorken cylinder ( with initial time @xmath137 ) for an ideal gas equation of state @xmath86 .
( a , c ) temperature ( in units of the initial temperature @xmath330 ) , ( b , d ) laboratory energy density ( in units of the initial pressure @xmath331 ) as functions of radial distance @xmath69 from the origin ( in units of the initial radius @xmath332 of the system ) .
the profiles are for times @xmath333 , @xmath334 , @xmath335 .
solid lines are obtained via the semi - analytical method of characteristics @xcite , dotted curves are produced with the relativistic hlle algorithm modified with sod s operator splitting method . + + * fig . 3 : * ( a , c , e ) temperature profiles for the fireball expansion for times @xmath336 and an initial energy density @xmath337 .
the profiles are alternatingly shown as full and dotted lines in order to better distinguish them .
( b , d , f ) show isotherms in the corresponding space time diagrams .
the isotherms are labelled with the corresponding temperatures in units of @xmath14 .
( a , b ) are calculated with @xmath6 , ( c , d ) with @xmath19 , and ( e , f ) with the ideal gas equation of state .
+ + * fig . 4 : * same as in fig . 3 , for @xmath148 .
profiles in ( a ) are for times @xmath338 , in ( c ) for @xmath339 , and in ( e ) for @xmath340 .
+ + * fig . 5 : * same as in fig . 3 , for @xmath341 . profiles in ( a , c , e ) are for times @xmath342 .
+ + * fig . 6 : * purely longitudinal expansion in the bjorken model for an initial energy density @xmath127 .
( a ) energy density ( in units of @xmath155 ) , ( b ) entropy density ( in units of @xmath343 ) , ( c ) pressure ( in units of @xmath155 ) , ( d ) temperature ( in units of @xmath14 ) as functions of ( proper ) time @xmath122 ( in units of @xmath131 ) .
solid lines are for @xmath6 , dotted for @xmath19 , dashed for the ideal gas .
different cooling laws @xmath344 are indicated . + + * fig . 7 : * same as in fig . 3 , but for the bjorken cylinder expansion with @xmath137
. profiles in ( a , c , e ) are for times @xmath345 .
+ + * fig . 8 : * same as in fig . 7 , for an initial energy density @xmath148 . profiles in ( a , c , e ) are for times @xmath346
. + + * fig . 9 : * same as in fig . 7 , for an initial energy density @xmath341 . profiles in ( a , c , e ) are for times @xmath347
. + + * fig .
10 : * lifetimes ( in units of the initial radius @xmath332 of the system ) in the one dimensional ( landau ) expansion ( cf . also @xcite ) as a function of the initial energy density ( in units of @xmath155 ) .
( a , b ) are for @xmath40 , ( c , d ) for @xmath90 .
the thick lines in ( a , c ) are for @xmath6 , in ( b , d ) for @xmath19 .
thin lines correspond to the ideal gas case .
solid lines are for @xmath158 , dotted for @xmath348 , dashed for @xmath160 .
+ + * fig .
11 : * the same as in fig . 10 , but for the fireball expansion .
+ + * fig .
12 : * the same as in fig . 10 , but for the bjorken cylinder expansion with @xmath349 .
+ + * fig . 13 : * the same as in fig . 12 , but for @xmath350 .
+ + * fig .
14 : * two pion correlation functions in ( a , c , e ) side and ( b , d , f ) out
direction for the fireball expansion with @xmath351 .
the average pion momentum is @xmath207 mev , the initial radius @xmath332 was fixed to be 5 fm .
the critical temperature was taken as @xmath352 mev , the pion mass is @xmath206 mev .
( a , b ) are for @xmath6 , ( c , d ) for @xmath19 , ( e , f ) for the ideal gas .
solid , dotted , and dashed lines are the correlation functions as calculated along the @xmath353 , and @xmath14 isotherm , respectively .
+ + * fig .
15 : * the same as in fig . 14 , but for the bjorken cylinder expansion with @xmath137 and @xmath354 .
the average transverse pion momentum is @xmath207 mev , the longitudinal momenta of the pions vanish .
+ + * fig .
16 : * the same as in fig . 11 , but for the ratio @xmath205
. + + * fig .
17 : * the same as in fig . 12 , but for the ratio @xmath205 .
+ + * fig .
18 : * the same as in fig . 13 , but for the ratio @xmath205 | the hydrodynamic expansion of quark gluon plasmas with spherical and longitudinally boost - invariant geometries is studied as a function of the initial energy density .
the sensitivity of the collective flow pattern to uncertainties in the nuclear matter equation of state is explored .
we concentrate on the effect of a possible finite width , @xmath0 , of the transition region between quark gluon plasma and hadronic phase . although slow deflagration solutions that act to stall the expansion do not exist for @xmath1 , we find , nevertheless , that the equation of state remains sufficiently soft in the transition region to delay the propagation of ordinary rarefaction waves for a considerable time .
we compute the dependence of the pion interferometry correlation function on @xmath2 , since this is the most promising observable for time - delayed expansion .
the signature of time delay , proposed by pratt and bertsch , is an enhancement of the ratio of the inverse width of the pion correlation function in out
direction to that in side direction .
one of our main results is that this generic signature of quark gluon plasma formation is rather robust to the uncertainties in the width of the transition region .
furthermore , for longitudinal boost - invariant geometries , the signal is likely to be maximized around rhic energies , @xmath3 agev .
16.5 cm -1 cm 22.4 cm columbia preprint cu
tp756 * the time
delay signature of quark gluon plasma formation in relativistic nuclear collisions * + dirk h. rischke and miklos gyulassy + + physics department , pupin physics laboratories , columbia university + 538 w 120th street , new york , ny 10027 , u.s.a . + + + june 1996 + |
cepheus b ( cep b ) is a molecular core located at the edge of the cepheus giant molecular cloud at a distance around 725 pc and lying @xmath11 above the galactic plane @xcite .
a handful of embedded young stars were found in cep b from radio continuum and infrared ( ir ) studies @xcite . the unobscured stellar ob association cep ob3 lies around cep b ; the younger subgroup , cep ob3b , lies closest to the cloud as shown in figure [ fig_map ] @xcite . for many years
, the cep ob3 association has been considered to be a good example of large - scale sequential star formation in accord with the model of @xcite where stellar winds and supernova remnants of an older stellar cluster compress and trigger a second generation of star formation in nearby molecular cloud cores @xcite .
the interface between the molecular cloud and the cep ob3b star association is clearly delineated by the optically bright hii region sharpless 155 ( s 155 ) , where cloud material is ionized and heated by the radiation field of the o7 star hd 217086 , b1 star hd 217061 and perhaps other cluster members @xcite . figure [ fig_map ] shows the spatial relationship of the cloud , hii region , and exciting stars . unlike in the orion nebula where the hii region lies between the cloud and ourselves , the photodissociation region at s 155 is favorably oriented to reveal the progression of star formation . following the triggered star formation model , we expect the surface of the cloud to be eroded by the early - type stars so that the cloud edge moves eastward across the observer s field of view with new stars emerging from the obscuring molecular cloud .
sources located within but near the edge of the molecular cloud would represent a new generation of star formation triggered by the hii region shock propagating into the cloud .
this scenario of triggered star formation has been recently strengthened by the discovery using the _ chandra x - ray observatory _ of @xmath12 lightly - obscured low - mass pre - main sequence ( pms ) members of the cep ob3b cluster located outside of the cloud , and a rich population of @xmath13 pms stars in the cluster embedded within cep b @xcite .
x - ray surveys are particularly effective in discriminating disk - free pms populations from galactic field stars which often badly contaminate infrared ( ir ) surveys of young stellar clusters ( see review by * ? ? ?
* ) . using 2mass counterparts of the x - ray sources , @xcite found that pms stars in the embedded cluster are more likely ( 26% @xmath14 4% ) to have @xmath15-band excesses from heated inner protoplanetary disks than stars in the unobscured cep ob3 region .
this supports both , youthfulness of the embedded cep b population and the prevalence of planet forming disks in the embedded cluster .
we seek here to elucidate the relationships of disks and environments in this region using mid - ir photometry from the _ spizer space telescope _ which is exceptionally well - suited for detecting disks around low - mass members of young clusters ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* and references therein ) .
measures of disk fractions as functions of stellar mass and age in different star - forming environments can provide better understanding of the evolution of circumstellar disks in different environments and thus have direct astrophysical importance in evaluating the conditions for planet formation .
x - ray pms selection is very complementary to ir surveys . as x - ray emission from pms stars
is based on enhanced solar - type magnetic reconnection events rather than disk or accretion processes , x - ray selection delivers rich and clean samples of diskless stars missed by ir selection @xcite .
the x - ray stars identified by @xcite in combination with discovered here @xmath8 disk - bearing stars constitute a valuable sample to measure protoplanetary disk properties in different radiative environments and to study the triggering process inside the cep b molecular cloud .
we use the spatial distribution of disks around young stars of the selected @xmath7 and @xmath8 sample to test concepts of cloud ablation and the radiative driven implosion ( rdi ) model of triggered star formation on the edges of hii regions ( e.g. , * ? ? ?
in [ data_reduction_section ] we report the @xmath8 observations using the infrared array camera ( irac ) detector .
classification on disk - bearing and diskless x - ray emitting pms stars is provided in [ yso_classification_section ] and the discovery of the new population of non-@xmath7 ir - excess members
is given in
[ non_xray_subsection ] .
the imf and mass completeness limits of the combined @xmath7 and @xmath16 stellar samples of different parts of the region are considered in [ imf_subsection ] .
the results on the disk evolution of the selected pms sample are presented in [ results_section ] .
we end in [ discussion_section ] with the implications of the new observational findings for the star formation process in the region .
we adopt a distance of 725 pc to the star forming complex and infer an age of @xmath17 myr for the cep ob3b stars and an age between @xmath18 and 2 myr for the embedded cep b stars , as discussed in appendix [ distance_age_subsection ] .
appendix [ lx_disks_section ] presents the relationship between x - ray activity , stellar mass and disks ; these results confirm those found in other young stellar clusters .
the x - ray observations of the cep b / ob3b and their data analysis are described in detail by @xcite .
the 30 ks exposure was obtained on 2003 march @xmath19 with the advanced ccd imaging spectrometer ( acis ) detector @xcite on - board nasa s _ chandra x - ray observatory _ @xcite as part of the acis instrument team s guaranteed time observions ( obsid # 3502 , g. garmire , pi ) . the imaging array ( acis - i ) consists of four abutted @xmath20 pixel front - side illuminated charge - coupled devices ( ccds ) covering about @xmath21 on the sky . following data reduction based on the _ acis extract _ idl script ,
@xcite obtained an x - ray catalog of 431 sources , most with @xmath22 positional accuracy ( figure [ fig_color_image ] ) . using 2mass counterparts , @xmath23 of the x - ray sources
are confidently associated with cluster pms members of the region .
the remaining x - ray sources are mostly extragalactic active galactic nuclei .
the mid - ir observation of cep b and cep ob3b was obtained on 2007 february 18 with the irac detector @xcite on nasa s _ spitzer space telescope _ in the 3.6 , 4.5 , 5.8 , and 8.0 @xmath24 m channels .
this is a general observer project ( program identification # 30361 ; j. wang , pi ) .
two adjacent fields subtending @xmath25 in channel pairs 3.6/5.8 @xmath24 m and 4.5/8.0 @xmath24 m were centered , respectively , on @xmath26 , @xmath27 ( j2000 ) and @xmath28 , @xmath29 ( j2000 ) .
the orientation of the fields long axis is @xmath30 .
these fields were imaged with a @xmath31 mosaic of adjacent positions separated by @xmath32 , and images were taken at 12 dithered positions for each of the 12 cells of the mosaic .
the images were taken in the high - dynamic - range mode with long effective exposures of 96.8 s in 3.6 , 4.5 , 5.8 @xmath24 m channels , and two 46.8 s exposures in the 8.0 @xmath24 m channel , as well as short effective exposures of 0.4 s in all 4 channels .
full coverage in all four channels is obtained for a @xmath33 area , which in turn covers @xmath34 of the @xmath35 _ chandra s _ acis - i field , as shown in figures [ fig_map ] and [ fig_color_image ] .
basic calibrated data ( bcd ) products from the spitzer science center s irac pipeline version s15.3.0 were automatically treated with the wcsmosaic idl package developed by r. gutermuth for the irac instrumental team . starting with bcd data products , the package mosaics individual exposures while treating bright source artifacts , cosmic ray rejection , distortion correction , subpixel offsetting , and background matching @xcite .
we selected a plate scale of 0.86@xmath36 for the reduced irac mosaics , which is the native scale divided by @xmath37 .
aperture photometry of irac sources was obtained using the iraf task phot . for the most of the x - ray and non - xray ( [ non_xray_subsection ] ) selected sources , an aperture radius of 4 pixels ( 3.44 ) was used with an adjoining sky annulus width of 1 and 6 pixels ( 0.86and 5.16 ) for the 3.6/4.5 and 5.8/8.0 @xmath24 m bands , respectively . for the crowded sources ,
aperture radii of 2 or 3 pixels ( 1.72and 2.58 ) and adjoining sky annulus width of 1 pixel ( 0.86 ) in all 4 bands were used .
we adopted zero point magnitudes ( @xmath38 ) of 19.670 , 18.921 , 16.855 , and 17.394 in the 3.6 , 4.5 , 5.8 , and 8.0 @xmath24 m bands , where @xmath39 @xcite .
the total aperture corrections applied to our measurements are : 0.185 , 0.175 , 0.165 , 0.250 in the case of 4 pixel aperture ; 0.275 , 0.305 , 0.515 , 0.735 in the case of 3 pixel aperture ; and 0.655 , 0.735 , 1.005 , 1.010 in the case of 2 pixel aperture in the 3.6 , 4.5 , 5.8 , and 8.0 @xmath24 m bands , respectively .
the reported photometric errors include poisson errors in the source and background emission plus a 2% uncertainty in the calibration of irac @xcite .
the initial list of point sources in these images was produced by running iraf task starfind with a low detection threshold in order to identify as many sources as possible , permitting false detections in this stage of the analysis .
an automated cross - correlation between the _ chandra _ source positions of @xcite and irac candidate source positions was made using a search radius of @xmath40 within @xmath41 of the acis field center , and a search radius of @xmath42 in the outer regions of the acis field where x - ray source positions are more uncertain due to the deterioration of the _ chandra _ telescope point spread function ( figure [ fig_color_image ] ) .
this was followed by a careful visual examination of each source in both bands to remove dubious sources and associations .
of the 431 x - ray sources within the @xmath35 _ chandra _ acis - i field @xcite , 396 ( 92% ) lie within the 3.6/5.8 @xmath24 m irac coverage and 412 ( 96% ) lie within the 4.5/8.0 @xmath24 m irac coverage . out of [ 396 , 412 , 396 , 412 ] x - ray sources in the [ 3.6 , 4.5 , 5.8 , 8.0 @xmath24 m ] channels , [ 384 , 399 , 369 , 385 ]
are identified with irac sources .
the irac detection rate of x - ray sources is thus @xmath43 for 3.6 and 4.5 @xmath24 m channels and @xmath44 for 5.8 and 8.0 @xmath24 m channels .
2mass associations are obtained as described in @xcite .
optical counterparts were identified from the @xmath45 photometric catalog of @xcite to the north - west and @xmath46 to the west . ] . unlike the search for the ir counterparts
, we lacked the ability to visually examine the @xmath7-optical matches .
we therefore compared the @xmath7-optical offset ( off@xmath47 in table [ tbl1 ] ) with the @xmath7-@xmath48 and @xmath7-irac positional offsets ( off@xmath49 and off@xmath50 ) .
accepted optical matches typically have @xmath51 . out of the total of 431 @xmath7 x - ray sources ,
368 are in the field of view of the optical survey and 321 of them have optical counterparts .
the detection rate of the x - ray sources in the optical survey is thus @xmath52 .
there are [ 12 , 13 , 27 , 27 ] x - ray sources in the [ 3.6 , 4.5 , 5.8 , 8.0 @xmath24 m ] channels lacking mid - ir counterparts .
roughly half of these lie in the interface region between the cep b molecular cloud and s 155 hii region where the irac point source sensitivity is reduced by the high background nebular emission from heated dust ( figure [ fig_map ] ) .
these are probably true cluster members , but are omitted from the science analysis below . the other half of these unidentified x - ray sources is uniformly distributed outside of the cep b cloud with weak x - ray fluxes and high x - ray median energies consistent with extragalactic background sources .
most of these are listed in table 4 of @xcite as likely extragalactic contaminants .
the results of this search for stellar counterparts of the @xmath7 sources are presented in table [ tbl1 ] .
listed by @xmath7 sequence number , the table gives offsets between the optical / ir stars and the x - ray sources , @xmath53 and irac photometry , and photometric flags .
the final column gives the aperture used in the photometry for each irac band ; a ` 9 ' value indicates that the x - ray source lies outside the irac coverage in that band or the irac source is too weak or strongly contaminated with no reliable photometry .
we base the evolutionary classification of x - ray emitting pms stars in the cepheus region on a comparison of their ir spectral energy distributions ( seds ) with the seds of pms stars in the well - studied ic 348 cluster in the perseus molecular cloud @xcite .
ic 348 stars have roughly the same @xmath17 myr age as cep ob3b stars .
the categories are : class i ( protostar , disk and envelope ) , class ii ( pms star and accretion disk ) , and class iii ( pms star with weak or absent disks ) . a simple characterization of ir seds involves the sed spectral slope @xmath54 of the sed @xmath55 over the range @xmath56 m .
class i sources have @xmath57 , class ii sources have @xmath58 , and class iii sources have @xmath59 . using optical and ir data for the ic 348 pms stars , @xcite showed the utility of the sed slope measured in the irac wavelength range from 3.6 to 8.0 @xmath24 m . comparing optical and irac seds to disk models calculated from a monte carlo radiative transfer code @xcite
, they find that class ii systems with optically thick disks have dereddened irac slopes @xmath60 , `` transition disk '' ( td ) systems with inner disk holes or optically thin inner disks have @xmath61 , and diskless class iii systems have @xmath62 .
@xcite also provide useful empirical templates of median seds of ic 348 pms stars for each evolutionary class .
we proceed with the disk classifications through the comparison of the observed cepheus seds in @xmath63irac ir bands to the ( de)reddened median sed templates of ic 348 pms stellar photospheres .
we further confirm these classifications through calculations of the sed slope @xmath54 , and locations in irac color - color diagrams .
results of our disk classification are shown in table [ tbl2 ] . first , a rough estimate of individual stellar masses of the cepheus pms stars is made by de - reddening star positions in the @xmath64 versus @xmath65 color - magnitude diagram to the @xmath66 myr and @xmath67 myr theoretical isochrons for the cep ob3b / s 155 and cep b subregions , respectively , based on the models of @xcite and ( * ? ? ?
* see appendix a ) .
these individual stellar photometric mass and extinction estimates are listed in table [ tbl2 ] .
although individual mass estimates may be subject to significant uncertainties and may not be always compatible with masses obtained by more accurate methods such as optical spectroscopy , this approach should be adequate to assign an ic 348 sed template ( table 4 of * ? ? ?
* ) to individual cepheus pms stars .
second , it is important to note that the ic 348 template seds are observed ( not de - reddened ) and exhibit an extinction of @xmath68 mag @xcite .
thus for each of the cepheus stars , an ic 348 template sed at the corresponding spectral sub - class is first reddened and de - reddened within the wide but finite range of extinctions using reddening relationships described in appendix a for @xmath69 and @xcite for irac bands .
the ( de)reddened ic 348 template which , after normalization to an observed @xmath64-band cepheus sed point , is the closest to an observed @xmath70-band cepheus point ( both @xmath64 and @xmath70-band points are considered to be indicators of a pure photospheric emission ) is recognized here as the best - fit to cepheus data .
third , the best - fit ic 348 ( de)reddened median template is visually matched to the observed cepheus sed , allowing classification of the cepheus stars . out of the total of 431 x - ray cepheus sources ,
our classification procedure yields 215 diskless class iii systems ( labelled `` nod '' in table [ tbl2 ] ) and 139 class ii or i disk - bearing systems ( `` dsk '' ) .
based on the membership analysis of @xcite , 24 x - ray sources without ir counterparts are possible extragalactic contaminants ( `` exg '' ) and 13 are foreground candidates ( `` frg '' ) .
forty chandra sources with ir detections which are unreliable or insufficient for classification have an uncertain classification ( `` unc '' ) .
we further subjectively classify 20 stars as class ii / iii transition disks ( labelled as `` td '' and `` dsk '' ) ; these follow the photospheric sed at wavelengths shorter than at least @xmath71 m .
four @xmath7 sources ( # # 314 , 322 , 328 , and 390 ) embedded in the cep b core have extreme sed excesses with @xmath72 and are classified as class i stars .
examples of diskfree , disk - bearing , and transition seds are shown in figures [ fig_sed_nod]-[fig_sed_tos ] .
the complete set of seds for 354 x - ray emitting cepheus pms stars classified here is available on - line as a figure set .
the on - line atlas contains 161 diskfree stars of the cep ob3b region ( labelled as `` nod '' and `` cepob3b '' in table [ tbl2 ] ) ; 75 disk - bearing stars of the cep ob3b ( `` dsk '' and `` cepob3b '' ) ; 36 diskfree stars of the s 155 region ( `` nod '' and `` s 155 '' ) ; 26 disk - bearing stars of the s 155 region ( `` dsk '' and `` s155 '' ) ; 18 diskfree stars in the cep b region ( `` nod '' and `` cepb '' ) ; and 38 disk - bearing stars in the cep b region ( `` dsk '' and `` cepb '' ) .
quantitative information derived from the fitting process is given in table [ tbl2 ] including the masses and absorptions estimated from the @xmath73 color - magnitude diagram ( columns 4 and 5 ) and the sed classification obtained by comparison with ic 348 templates ( columns 6 and 7 ) .
we performed least - squares linear fits to the @xmath74 values in the four irac wavelength band to obtain the observed ( not de - reddened ) sed spectral index , @xmath75 .
these are tabulated in column 2 of table [ tbl2 ] .
figure [ fig_slope_histogram ] shows the histogram of 354 @xmath75 values for the x - ray selected cepheus pms stars compared to 299 ic 348 pms stars from @xcite .
two important features are noted .
first , the cepheus histogram shows a bimodal distribution with peaks at @xmath76 and @xmath77 corresponding to disk - bearing and diskless star samples , respectively .
the ic 348 histogram has the same peaks , and similar bi - modality is seen in the pms populations of the cha i cloud @xcite and @xmath78 ori cluster @xcite .
second , in contrast to the ic 348 population , the cepheus region does not produce a rich td population in the interval @xmath79 .
the x - ray selected cepheus population has 20/139 ( @xmath80 ) tds among disk - bearing stars compared to 70/163 ( @xmath81 ) in ic 348 .
this apparent difference in td frequency is discussed in [ to_subsection ] .
comparison of @xmath8-irac source positions in the @xmath82 - [ 4.5]$ ] @xmath14 @xmath83 - [ 8.0]$ ] diagram with the results of models of stars with dusty disks and envelopes led @xcite and @xcite to propose this diagram as an excellent tool for pms classification .
@xcite show that similarly successful pms classification schemes can be obtained at shorter wavelengths using the @xmath82 - [ 4.5]$ ] @xmath14 @xmath84 - [ 5.8]$ ] and @xmath85 $ ] @xmath14 @xmath82 - [ 4.5]$ ] diagrams . in figure [ fig_ccds ] , we compare the sed - based classifications of cepheus x - ray stars derived in [ ic348_subsection ] with the expected loci of pms stars in irac color - color diagrams . here
we also begin comparison of the disk characteristics in three spatial subregions : the unobscured cep ob3b cluster , the optically - bright s 155 nebula , and the cep b molecular cloud core .
the regions are measured radially from the location of the hot molecular core at @xmath86 , @xmath87 ( j2000 ) .
figure [ fig_ccds ] shows the three irac color - color diagrams for the sources in the three subregions .
seventy - seven cep ob3b sources are missing from the plots due to inadequate photometry , often due to location outside of the full four - band coverage .
sources are also omitted from the s 155 and cep b areas when photometry at the longer 5.8 @xmath24 m and 8.0 @xmath24 m bands are absent or inaccurate due to bright nebular emission .
the positions of stars in the color - color diagrams generally agree with our previous classification of stars as disk - bearing and diskless pms stars based on comparison with ic 348 stars ( [ ic348_subsection ] ) , shown as blue and red symbols in figure [ fig_ccds ] .
the @xmath88 flag of table [ tbl2 ] indicates sources with their disk classification from [ ic348_subsection ] inconsistent with the simple color criterion that diskless stars have @xmath82 - [ 4.5 ] \la 0.2 $ ] , @xmath84 - [ 5.8 ] \la 0.2 $ ] or @xmath83 - [ 8.0 ] \la 0.2 $ ] .
several sources with such `` discrepant '' locations on color - color diagrams may represent additional cases of transitional disks .
due to the relatively short @xmath7 exposure and large distance to the cepheus region , the x - ray pms sample is only complete down to @xmath89 m@xmath5 for cep ob3b / s 155 class iii stars , less complete for cep ob3b / s 155 class ii stars , and even less complete for cep b embedded objects ( [ imf_subsection ] ) . as the stellar initial mass function peaks around @xmath90 m@xmath91 , hundreds of cepheus stars should be present in the @xmath7 field of view with their x - ray luminosities below the sensitivity limits of the @xmath7 observation of @xmath92 erg s@xmath93 @xcite . with an age around @xmath94 myr and
younger ( appendix a ) , a large fraction of the cepheus stars with masses below 0.5 m@xmath91 in our field will have optically thick disks detectable with irac .
selection by irac ir - excess should even detect many brown dwarfs in the region .
shallow irac exposures efficiently detected disks around brown dwarfs in the orion cloud at a distance of @xmath0 pc over several magnitudes in brightness ( * ? ? ?
* and references therein ) .
the sensitivity at the cepheus distance around @xmath95 pc will be reduced by @xmath96 mag , sufficient to detect a considerable fraction of the disk - bearing brown dwarfs .
thus some fraction of non - x - ray pms stars with lower - masses are expected to be detected in the irac survey within the @xmath7 field of view .
figure [ fig_ccd_cmd_nonxray]@xmath97 shows irac color - color diagrams for 774 irac sources in the @xmath7 acis - i field of view that are not associated with x - ray sources .
we restrict these diagrams to sources with photometric errors @xmath98 mag in all four irac bands . using the ir color - color classification criteria presented in
[ ccd_subsection ] , we identify 224 non - x - ray ir - excess stars . table [ tbl3 ] gives ir properties for the 224 non - x - ray ir - excess sources .
figure [ fig_ccd_cmd_nonxray]@xmath99 compares x - ray - selected and ir - excess - selected samples in the @xmath84 $ ] @xmath14 @xmath84-[8.0]$ ] color - magnitude diagram . as expected , the majority of these new pms candidates are fainter than the x - ray selected sample , although there is a considerable overlap in the distributions around @xmath100<12 $ ] mag ( @xmath101 m@xmath5 ) .
most of these new stars thus have masses in the @xmath102 m@xmath91 and some in sub - stellar range .
panels show that , despite the different selection criteria , the color distributions of these irac - selected stars are similar to those of the x - ray selected pms stars .
however , several of the weakest and reddest ( @xmath84 \ga 14 $ ] mag and @xmath84-[8.0 ] \ga 0.5 $ ] mag ) ir - selected sources may be unrelated extragalactic objects @xcite .
four probable extragalactic contaminants from the x - ray - selected sample lie near this region of the diagram ( grey diamonds in figure [ fig_ccd_cmd_nonxray]@xmath103 ) . in the unobscured cep
ob3b subregion , the combined x - ray - selected and ir - excess - selected sample should give a complete census of the disk - bearing population down to @xmath104 m@xmath5 , a nearly complete census of the disk - free population down to @xmath4 m@xmath91 , and a fraction of both the disk - bearing and disk - free populations into the brown dwarf regime ( [ imf_subsection ] ) .
but due to the limited point source sensitivity from the non - uniform ir nebular emission , the identified pms populations of the s 155 and cep b subregions are less complete .
this can be seen from the paucity of sources fainter than @xmath84>14 $ ] mag in s 155 and @xmath84>13 $ ] mag in cep b ( figure [ fig_ccd_cmd_nonxray]@xmath105 ) .
more accurate information on mass completeness limits of cepheus pms stars is given in the next section .
in the area covered by the @xmath7 field and all four bands of the irac mosaic , we examine the stellar imf separately for diskless and disk - bearing stars in each of the three sub - regions . for a given mass ,
the x - ray detection efficiency of class iii stars is somewhat high than that of class ii stars ( appendix [ lx_disks_section ] ) . to compensate for this effect , in this imf and the following disk evolution analyses
, we use the disk - bearing stellar sample that is the combination of the @xmath7 and non-@xmath7 ir - excess member young stars ( [ non_xray_subsection ] ) while retaining the diskless stellar sample as purely composed of x - ray stars . in this imf analysis section , we exclude x - ray stars lying outside of the 4 irac band mosaic , as well as the most massive young star in the region , o7vn star hd 217086 ( @xmath7 # 240 ) which is the only star with mass above @xmath106 m@xmath5 ( upper boundary of considered imf mass range ; figure [ fig_imf ] ) .
we also exclude 51 non-@xmath7 ir - excess possible members with inferred masses below 0.1 , as some of them may still be extragalactic background objects ( [ non_xray_subsection ] ) .
figure [ fig_imf]@xmath107 shows that the mass distributions of the cep ob3b sub - region stars , both diskless ( red ) and disk - bearing ( blue ) , nicely follow the shape of the galactic field imf , showing a salpeter powerlaw slope between @xmath108 m@xmath91 , and a peak around @xmath109 m@xmath5 .
the decline seen below @xmath109 m@xmath91 is attributable to our incompleteness limits and is probably not intrinsic to the cluster .
the pure @xmath7 class iii sample is complete down to @xmath4 m@xmath5 , while the combined @xmath7 and non-@xmath7 class ii sample is complete for @xmath110 m@xmath5 cep ob3b stars . assuming a galactic field imf shape
, we estimate the total pms population in the cep ob3b sub - region to be @xmath111 stars down to 0.1 m@xmath5 .
an important negative result here is that we do not confirm the tentative explanation by @xcite that the non - standard x - ray luminosity function of the cep ob3b is due to an anomalous ` bottom - heavy ' imf with an excess of @xmath112 m@xmath5 stars .
although a slight difference between the combined @xmath7 and @xmath8 sample and the galactic field imf is seen ( figure [ fig_imf]@xmath107 ) , it is not statistically significant .
figure [ fig_imf]@xmath113 shows that mass distributions of the @xmath114 sub - region stars , although with fewer stars than in cep ob3b , also follow the galactic field shape with mass completeness limits of @xmath4 m@xmath5 for both class ii and class iii stellar samples with an estimated total pms population of @xmath115 s 155 stars down to 0.1 m@xmath5 . the observed mass distribution of the embedded cep b cluster may be either top - heavy or , more likely , simply lacking unidentified low - mass disk - free pms stars due to @xmath7 sensitivity loss from obscuration ( figure [ fig_imf]@xmath116 ) .
a similar situation is noticed in the study of the embedded population of the bright - rimmed cloud ic 1396n @xcite . in the case of usual imf with still
unidentified low - mass stars , the observed cep b stellar population may be complete only somewhere above @xmath6 m@xmath5 , and the total intrinsic pms population of the cep b may reach a few hundred stars .
the triggered population is thus much larger than that identified by early radio and infrared techniques @xcite .
growing evidence has emerged from irac young cluster studies that more massive pms stars have shorter disk lifetimes .
@xcite find that disk fractions in the @xmath17 myr old ic 348 and @xmath78 ori clusters decrease from @xmath117 in brown dwarfs to @xmath118 in stars with @xmath119 m@xmath5 . in the older 5 myr old upper sco and ngc 2362 clusters , the disk fraction decreases from @xmath120 in @xmath121 m@xmath5 stars to @xmath122 in higher - mass stars @xcite .
the trend may not be universal : the disk fraction of @xmath119 m@xmath5 stars in the @xmath67 myr old cha i cluster is @xmath123 @xcite , much higher than that found in ic 348 and @xmath78 ori at similar ages .
the figure [ fig_mj_diskfraction ] compares the mass - dependent disk fractions of cep ob3b and the other three regions examined by @xcite .
our sample here is the combined 215 @xmath7 ( classified as `` dsk '' or `` nod '' in table [ tbl2 ] ) and 142 disk - bearing non-@xmath7 ( table [ tbl3 ] ) cep ob3b sources within the overlap area of 4 irac band and @xmath7 fields , omitting the most massive young star in the region , o7vn star ( @xmath7 # 240 ) with @xmath124 mag , and 37 non-@xmath7 ir - excess possible members of cep ob3b with inferred masses below @xmath125 m@xmath5 .
appendix [ lx_disks_section ] shows that the relatively short @xmath7 observation is only sensitive to stars down to @xmath126 m@xmath5 , so we have no cep ob3b coverage for very low mass class iii stars and brown dwarfs .
cep ob3b stars have average ages around @xmath17 myr old ( appendix [ distance_age_subsection ] ) similar to the ages in cha i , ic 348 , and @xmath78 ori ; possibly younger stars in the s 155 and cep b region are omitted here . following luhman et al .
, the absolute @xmath64-band magnitude @xmath127 serves as a proxy of mass .
our @xmath127 values are obtained using individual @xmath128 extinctions from tables [ tbl2]-[tbl3 ] , which in turn were derived through the evolutionary model - dependent procedure assuming the distance of @xmath129 pc and the age of @xmath66 myr .
individual magnitude accuracies should be sufficient to reliably assign individual sources to the coarse , 2 mag - wide @xmath127 bins .
errors on disk fractions have been estimated using binomial distribution statistics as described by @xcite .
for the cep ob3b stars , @xmath127 bins have been systematically shifted by @xmath130 mag from those of other regions in order to include in the second @xmath127 bin the @xmath89 m@xmath5 mass stars , stars for which the combined @xmath7 and non-@xmath7 cep ob3b sample is complete ( [ imf_subsection ] ) .
figure [ fig_mj_diskfraction ] shows that the disk fraction of the cep ob3b stars ( circles ) for the two highest mass bins is around 45% and agrees with the fractions seen in ic 348 , but appears higher than the comparison clusters for the lowest @xmath131 m@xmath91 mass bin .
the most plausible explanation for the reduced disk fraction in the @xmath102 m@xmath91 bin is incompleteness of the cep ob3b sample at low masses ( [ imf_subsection ] ) . here , our short @xmath7 observation is not sensitive to many low - mass class iii stars and the observational bias towards non-@xmath7 disk - bearing stars affects the sample selection .
we thus find no significant dependence of disk fraction on mass in the cep ob3b @xmath7 plus non-@xmath7 selected sample above @xmath4 m@xmath5 pc and thus younger age of @xmath132 myr for the cep ob3b cluster ( appendix [ distance_age_subsection ] ) . ] .
this agrees with the lack of mass dependence seen in ic 348 and @xmath78 ori in the @xmath133 m@xmath5 regime .
the increased fractions reported in these clusters were found in the @xmath134 m@xmath5 regime which is not covered by our cep ob3b selected sample .
however , we do not support studies in other clusters that report a rapid loss of irac - band disk emission in intermediate - mass stars .
these studies might consider possible sample incompleteness effects , such as an under - sampling of diskfree class iii low - mass members .
our inclusion of the x - ray selected sample is specifically designed to be nearly free of this bias .
we now consider spatial distribution of the disks around the mass - complete @xmath7 plus non-@xmath7 pms samples over the area covered by the 4 irac mosaics and the @xmath7 field .
figure [ fig_spat_distrib ] shows the spatial distribution , and table [ tbl_frac ] provides quantitative details for two mass strata , two azimuthal zones ( ne and sw ) , three radial subregions ( cep ob3b , s 155 , and cep b ) , and three radial layers discussed below ( inner , intermediate , and outer ) .
the azimuthal zones are separated by the line that roughly bisects the optically - bright ( figure [ fig_map ] ) and ir - bright nebular emission .
radial distances are measured from the hot core of the molecular cloud ( @xmath86 , @xmath87 ( j2000 ) ) .
below we consider the cepheus sample of 220 @xmath7 plus non-@xmath7 pms stars with stellar masses above @xmath4 m@xmath5 , the mass completeness limit for cep ob3b / s 155 stars ( [ imf_subsection ] ) , as well as its mass - stratified sub - samples .
the observed stellar population of the embedded cep b cluster may be complete somewhere above @xmath6 m@xmath5 ( [ imf_subsection ] ) and thus the analysis of the @xmath135 m@xmath5 stellar sub - sample may provide more realistic disk fraction estimates for this sub - region . examining first azimuthal dependence , we find that the ne zone of the region has @xmath136 times higher stellar surface density than the sw zone .
there is also an indication for possible slight increase in disk fraction from ne to sw zones of the cep ob3b / s 155 region , although the effect is not statistically significant ( table [ tbl_frac ] ) .
examining next the radial dependence , a dramatic trend is seen : the disk fraction increases from @xmath137% in the cep ob3b subregion to @xmath138% in the s 155 subregion to @xmath139% in cep b ( table [ tbl_frac ] ) .
this trend is elucidated in more detail in figure [ fig_disk_frac_spatial ] as a running average disk fraction . here the running disk fraction for the @xmath140 m@xmath5 mass stratum ( red ) is evaluated within a @xmath141 wide sliding window at @xmath142 intervals , while the @xmath143 m@xmath5 ( black ) and @xmath144 m@xmath5 ( blue ) mass strata are evaluated within a @xmath145 wide sliding window at @xmath141 intervals .
note these values are not independent because a given star is included in two adjacent values .
the effect could either be viewed as a smooth trend in disk fractions , or a step - function at 30% ( outer layer ) , 60% ( intermediate layer ) and @xmath139% ( inner layer ) disk fractions in @xmath146 pc radial bins ( grey bars ) .
it is important to note that the outer edge of the second bin ( intermediate layer ) is coincident with the projected position of the ionizing source of the region , o7v star hd 217086 .
if one adopt the simple relation between stellar age and disk fraction derived by ( * ? ?
* see their figure 14 ) , the cepheus disk fraction gradient corresponds to an age gradient from @xmath147 myr in the outer layer ( part of cep ob3b subregion ) , @xmath17 myr in the intermediate layer ( part of cep ob3b / s 155 region ) , and @xmath9 myr in the inner layer ( embedded cep b subregion ) .
this inferred age gradient agrees with that estimated from stellar ( @xmath14 disk ) properties in appendix [ distance_age_subsection ] . as noted earlier in
[ slope_subsection ] , the cepheus ob3b population appears deficient in transition disks compared to the @xmath17 myr old cluster ic 348 .
we investigate this further here , including comparison to the low - mass 1 myr old coronet cluster @xcite and the 4 myr old tr 37 cluster @xcite .
tr 37 , like our cep ob3b field , harbors an o7-type star .
it is important to recall that different studies of different star forming regions use different criteria for identifying tds . in the tr 37 and the coronet cluster studies ,
tds are identified with a photometric excess longward of @xmath148 m . in the study of ic 348 ,
small excesses around @xmath67 @xmath24 m define `` anemic disks '' systems @xcite .
our classification of cepheus stars here ( [ yso_classification_section ] ) relies on an intermediate criterion of excess longward of @xmath71 m .
thus , if the fraction of transition disks were intrinsically the same in all clusters , we expect the reported fractions to be largest in ic 348 , intermediate in our sample , smallest in tr 37 and the coronet
. furthermore , inter - cluster comparison of td fractions should also consider similar mass strata , as lower - mass pms stars may have flatter disks with more frequent inner holes than more massive pms stars ( e.g. * ? ? ?
* ) . from table 2 of @xcite , a td fraction among disk - bearing stars in ic 348 decreases from @xmath149 ( @xmath150 ) for m - type stars to @xmath151 ( @xmath152 )
for g - k - type stars , although this decrease is not statistically significant . the tr 37 study combined stars from late g to m2-type ( @xmath153 m@xmath5 ) and reported a @xmath154 ( @xmath155 out of @xmath156 ) td fraction among disk - bearing stars .
the coronet cluster study included only m0-m8 objects and reported a @xmath157 ( @xmath158 ) td fraction .
our classification of x - ray selected pms stars in the cep b / cep ob3b obtained 20 tds out of 139 disk - bearing stars giving a td fraction of @xmath159 .
most of these ( 13 out of 20 ) are in the unobscured cep ob3b subregion and none are present close to the cep b hot core .
to perform a mass - stratified analysis , we consider here only the most rich star sample , cep ob3b . from table
[ tbl2 ] , we find the cep ob3b td fraction may be mass - dependent with @xmath160 ( 8 out of 28 ) among @xmath161 m@xmath5 ( m - type ) disk - bearing stars and @xmath162 ( 5 out of 47 ) around more massive stars .
these results are roughly consistent with the previous studies , once the td definitions and mass ranges are taken into account .
our overall @xmath163 td fraction is intermediate between the @xmath164 reported for ic 348 and @xmath165 reported for tr 37 , as expected for an intrinsically constant td fraction .
our finding that m stars have a higher td fraction than more massive stars agrees with the indication for trend seen in ic 348 and the high fraction found in coronet m stars .
we thus conclude that the fraction of cep ob3b tds agrees with that of other clusters despite the apparent deficit noted earlier in figure [ fig_slope_histogram ] .
the observed picture of the cep b region is in many respects reminiscent , on a larger scale of several parsecs , of those of smaller bright - rimmed clouds ( brcs ) and cometary globules ( cgs ) found on the edges of giant h ii regions @xcite .
brcs are isolated clouds surrounded by ionized rims facing the exciting star(s ) with their dense cores close to the rims .
brcs are modeled as externally illuminated , photoevaporated and ablated into elongated head - tail morphologies by ultraviolet radiation of ob stars @xcite .
it is likely that pressure from the ionization shock front at the surface propagates through a globule and overcomes the magnetic , turbulent and thermal pressure that supports it against collapse , thereby triggering localized star formation .
this astrophysics of radiation - driven implosion ( rdi ) in molecular globules has been extensively studied ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
rdi models typically consider the triggering of a single star formation episode in a small ( @xmath18 pc ) cloud with tens of m@xmath5 of molecular gas .
the characteristic timescale for producing cometary morphologies and inducing gravitational collapse varies with initial conditions from @xmath125 to @xmath9 myr . in most cases , molecular , ir , h@xmath54 surveys of brcs trace only the most recently formed stars ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
in a few cases , @xmath7 observations have added the diskfree pms populations @xcite .
triggered brcs often show a few embedded mid - ir sources denoting protostars , while h@xmath54 , @xmath166 and x - ray surveys reveal small clusters of disk - bearing pms stars within and in front of the bright rim . in a few cases , spatial - age gradients in the stellar population are seen where the youngest stars are embedded and older stars are aligned toward the ionizing sources @xcite . this directly supports the rdi mechanism and implies that the existing clouds have been actively forming stars for several million years .
the cepheus region possesses all of the observational features of an rdi triggered star formation region : 1 .
the presence of exciting star(s ) and a molecular cloud surrounded by an ionized rim facing the exciting stars shown in figures [ fig_map]-[fig_color_image ] @xcite ; 2 .
the presence of a dense molecular core close to the rim @xcite ; 3 .
the spatio - temporal gradient of young stars oriented towards the exciting star(s ) shown in figures [ fig_spat_distrib]-[fig_disk_frac_spatial ] .
the major difference of the cepheus region from other brcs is its larger scale : its extent is @xmath167 pc linear size , its molecular mass is @xmath115 m@xmath5 , its observed pms x - ray stellar population is @xmath168 embedded and up to @xmath169 older unobscured members , as well as estimated intrinsic population of @xmath170 unobscured and possibly a few hundred embedded stars ( [ imf_subsection ] ) .
the large - scale morphology is also different from most other brcs . rather than lying on the edge of a circular hii region surrounding a concentrated ob association
, it is an extension of a giant molecular cloud protruding into a large evacuated region produced by a partially dispersed ob association .
another unusual case of a very large brc with star formation triggered over several million years is the high - latitute cometary globule cg 12 @xcite .
based on the rdi triggered star formation concept , we can propose some new insights into the star formation process of the cepheus region and make some simple model - independent estimates for a shock propagation velocity and a star formation efficiency ( sfe ) of the rdi process . in
[ spatial_domain_subsection ] and appendix a , we establish the spatio - temporal gradient of cepheus pms stars from the center of the cloud towards the hd 217086 o7vn star . assuming a distance around 725 pc , the innermost @xmath146 parsec adjoining the hot core of the cloud has the youngest ( @xmath171 myr ) stars with the highest seen disk fraction ( @xmath172 ) .
the intermediate @xmath173 parsec layer around s 155 has older ( @xmath17 myr ) stars with an intermediate disk fraction ( @xmath174 ) ( figure [ fig_disk_frac_spatial ] ) .
the outer layer of mostly diskless stars in the unobscured cep ob3b cluster has the lowest disk fraction ( @xmath175 ) pc , the ages of the cep b and s 155/cep ob3b subregions are older by @xmath176 myr but the @xmath9 myr difference between them remains ( appendix a ) . ] we consider two scenarios for rdi - induced star formation in this region . first , a relatively slow shock with speed @xmath177 km s@xmath93 may have passed through the molecular cloud , triggering star formation in the outer layers around @xmath17 myr ago and continuing to trigger star formation close to the cep b hot core today .
this slow shock speed is consistent with the @xmath178 km s@xmath93 propagation rate inferred from the age gradient of the pms stars in the globule ic 1396n , located @xmath179 pc projected distance of its ionizing o6e type star @xcite . in this scenario , stellar kinematic drift from their birthplaces plays an important role in relocating pms stars .
the o7vn star hd 217086 star is probably the principal ionization source but other ob stars ( figure [ fig_map ] ) , including the b1vn star hd 217061 currently located within the sw zone of the s 155 subregion , probably also play a role . if the original cloud were only slightly larger than the size observed today , the rdi shock would have slowly propagated through @xmath9 pc of the cep b molecular cloud over @xmath94 myr years , implying a shock velocity around @xmath177 km s@xmath93 .
the stars formed early in this process would have drifted @xmath180 ( @xmath181 pc ) in all directions , populating much of the @xmath7 field of view with triggered stars .
such star drifting would require a stellar velocity dispersion around 1 km s@xmath93 .
this is expected from the turbulent velocities within a cloud several parsecs in extent @xcite , which can be supplemented by dynamical interactions between protostars during star formation @xcite .
a second scenario involves the passage of a faster shock propagating at @xmath9 km s@xmath93 through a cloud that was originally much larger than the cep b cloud we see today . in this model ,
most of the original cloud material has ablated and most of the stars in the @xmath7 field formed in the cloud during the past @xmath94 myr .
present locations of the intermediate layer around s 155 and the inner layer around the molecular core ( separated by @xmath182 pc , figure [ fig_disk_frac_spatial ] ) may be directly associated with the passage of the rdi shock @xmath94 and @xmath183 myr years ago , respectively . star drift plays a less important role here .
a similar fast shock speed is observed in the triggered star formation in the molecular pillars of the eagle nebula ( m 16 ) and other bright rimmed clouds @xcite , and is consistent with the theoretical models of @xcite .
a critical discriminant between these scenarios is the sfe of the cloud . integrating the stellar masses in figure [ fig_imf ] , the total stellar mass of the cep ob3b disk - bearing population assuming a standard galactic field imf is @xmath184 m@xmath5 in our observed field . with the average disk fraction of @xmath185 ( [ mass_domain_subsection ] , table [ tbl_frac ] ) , we add @xmath186 m@xmath91 of diskfree stars and @xmath187 m@xmath91 from hd 217086 to give a total mass of the cep ob3b population around @xmath188 m@xmath5 .
a similar estimate for the s 155 population from information in figure [ fig_imf ] and table [ tbl_frac ] gives a total stellar mass around @xmath189 m@xmath5 . assuming that the observed cep b embedded population is complete only somewhere above @xmath6 m@xmath5 ( [ imf_subsection ] ) and follows a standard galactic field imf as well , we use star count and disk fraction information for the highest mass bin from figure [ fig_imf ] and table [ tbl_frac ] to obtain an estimated total mass of the cep b population of @xmath190 m@xmath5 .
the mass of the observed molecular gas in the cep b molecular portion of the cepheus molecular cloud is only @xmath115 m@xmath5 @xcite .
the inferred sfe for the embedded cep b population alone is thus around @xmath191 , at the top of typical sfe range of @xmath192 measured in other active star forming regions ( * ? ? ? * and references therein ) .
this is far above the sfe found in smaller molecular globules triggered by uv shocks ( e.g. * ? ? ?
* ) . for the total cepheus stellar population within the @xmath7 field ,
the apparent sfe is extremely high around @xmath10 .
this is unrealistically high and indicates that most of the cep ob3b stars were not formed from the presently seen cloud material but arose from an earlier generation of star formation from gas that is no longer present .
this supports the second scenario involving a fast shock passing through a much larger and more massive original cloud .
two other aspects of rdi triggered star formation can be discussed .
first , @xcite provided observational evidence for a non - standard imf where intermediate - mass stars are preferentially formed over lower mass stars in brcs . in the case of the embedded triggered pms population of ic 1396n , @xcite also found indications for a nonstandard imf biased towards higher mass stars , but cautioned about possible observation selection effects .
a similar situation arises here in the case of the embedded cep b population ( [ imf_subsection ] ) .
while the shape of the cep b stellar mass distribution is not definite yet , the stellar mass distribution of likely rdi triggered stellar population of the cep ob3b / s 155 region agrees well with a standard galactic imf , at least down to @xmath193 m@xmath5 ( [ imf_subsection ] ) .
second , most theoretical calculations of rdi triggering involve small globules and a single episode of triggering .
the @xmath17 myr range of stellar ages found in the cep b / cep ob3b region , and an even wider age spread found in cg 12 @xcite , implies repeated or continuing star formation over millions of years when the rdi mechanism occurs in a larger molecular cloud .
we present a @xmath8 irac observation of the cepheus b molecular cloud , the s 155 hii region on its periphery , and a portion of the nearby cep ob3b ob association .
the goals of this work are to study disk evolution of the nearly disk - unbiased combined @xmath7-acis and @xmath8-irac selected samples of pms stars and to provide new clues of the sequential triggered star formation process in the region . out of @xmath0 x - ray emitting pms stars in the region , 354 are classified as diskless or disk - bearing stars based on ir photometry : 161 , 36 , and 18 ( 75 , 26 , and 38 ) diskless ( disk - bearing ) stars in the cep ob3b , s 155 , and cep b subregions , respectively .
we immediately see that samples selecting only ir - excess stars miss the majority of stars outside the cep b molecular core . among all x - ray emitting disk - bearing systems , only 4 are class i protostars ; they lie in the younger embedded cep b cluster ( [ yso_classification_section ] ) .
in addition to the @xmath0 x - ray emitting pms stars in the region , we identify @xmath2 non-@xmath7 ir - excess low - mass members of the region ( [ non_xray_subsection ] ) . for the lightly - obscured cep ob3b and
s 155 clusters around the cloud , the combined @xmath7 and non-@xmath7 pms samples are complete down to @xmath4 m@xmath5 , while for the embedded cep b cluster the observed pms sample may be complete only above @xmath6 m@xmath5 ( [ imf_subsection ] ) .
the spatial distribution of the disks around stars of the combined @xmath194 pms samples reveals a dramatic spatio - temporal gradient : younger stars are clustered around the hot core of the molecular cloud , while older stars are dispersed in the direction of the primary ionizing o7-type star hd 217086 ( [ spatial_domain_subsection ] ) .
disk fractions fall from 80% to 30% across the s 155 hii region .
this finding constitutes strong endorsement for the basic model of triggered star formation in the region by passage of shocks driven by the ob stellar ionization and/or winds into the cep b molecular cloud .
our estimate of the shock velocity of @xmath9 km s@xmath93 agrees with empirical estimates obtained for a number of other shocked bright - rimmed clouds and with theoretical models .
the imf of the cep ob3b / s 155 triggered stellar population exhibits a standard galactic field shape ; we do not confirm our earlier suggestion of a bottom - heavy imf .
the unusually large apparent value of star formation efficiency of @xmath10 based on the present day molecular cloud indicates that most of the cep b molecular material has likely already been evaporated or transformed to stars .
contrary to the current rdi simulations of typical small globules which model a single triggering event over @xmath195 myr time period , our observational findings support a picture of multiple episodes of triggering over millions of years ( [ discussion_section ] ) .
using the nearly disk - unbiased combined @xmath7-acis and @xmath8-irac selected stellar sample , we also obtain several other results .
first , the disk fraction as a function of stellar mass in the cep ob3b cluster is similar to that of the ic 348 pms stars with ages similar to cep ob3b ( [ mass_domain_subsection ] ) .
second , the fraction of cep ob3b transition disks agrees with that of other clusters , such as tr 37 , coronet , and ic 348 ( [ to_subsection ] ) .
third , in agreement with the disk fraction analysis of [ spatial_domain_subsection ] , we obtain photometric ages of @xmath171 and @xmath17 myr for the embedded cep b and lightly - obscured cep ob3b clusters , respectively using the optical color - magnitude diagram ( appendix [ distance_age_subsection ] ) and assuming a distance of @xmath129 pc .
fourth , the effect of reduced x - ray emission in disk - bearing versus diskless stars which was originally established in studies of onc and taurus regions is confirmed in the cepheus systems ( appendix [ lx_disks_section ] ) .
the distance to the cep b / cep ob3 complex is somewhat uncertain . from the optical photometry of the early - type members of the cep ob3 association ,
@xcite and @xcite obtained a distance estimate of @xmath129 pc .
later @xcite combined near - ir photometry with the earlier optical photometry and spectroscopy data of bright members to derived a distance estimate of @xmath196 pc . a recent vlbi parallax measurement using the methanol maser in the cep a molecular core
gives a distance of @xmath197 pc @xcite .
a distance of 870 pc has been inferred for the nearby cep ob2 association @xcite . our x - ray study @xcite adopted the @xmath129 pc estimate while the optical studies of the cep ob3b by @xcite and @xcite adopted a distance of @xmath196 pc .
the chosen distance influences age estimates of the lower - mass stars as pms evolutionary isochrones on the hertzsprung - russell diagram ( hrd ) depend on absolute luminosities .
early photometric age estimates of the ob population indicated a relatively old cluster around @xmath198 myr @xcite .
for the x - ray pms stars in the @xmath7 field , we adopted an age of @xmath9 myr @xcite based on the hrd of five pms x - ray sources @xcite .
recently , @xcite obtained @xmath199- and @xmath200-band photometry data for an area including most of the @xmath7 acis - i field and measured a @xmath201 myr photometric age for the low - mass cep ob3b population .
it is unclear from any of these earlier datasets whether the age is uniform across the region .
we seek here to estimate stellar ages from the @xmath199- and @xmath200-band color - magnitude diagram for the @xmath7 x - ray pms stars using the photometry of @xcite listed in table [ tbl1 ] .
the @xmath48 near - ir photometry was first used with the combination of the synthesized @xcite ( for @xmath202 m@xmath5 ) and @xcite ( for @xmath203 m@xmath5 ) pms evolutionary isochrones to obtain individual extinction estimates for @xmath7 sources .
we obtained different sets of source extinctions ( @xmath204 ) using the @xmath64 versus @xmath65 color - magnitude diagram ( similar to the one shown in figure 5 of * ? ? ? * ) with two trial assumed distances ( 725 and 850 pc ) and three trial assumed ages ( 1 , 2 , and 3 myr ) .
the six sets of source extinctions were applied to correct the observed optical magnitudes of @xmath7 sources and to place them on the @xmath199 versus @xmath205 diagram . based on the values of @xmath206
mag and @xmath207 mag at @xmath208 mag adopted by @xcite and the reddening laws of @xcite , we used the reddening relationships @xmath209 , @xmath210 , and @xmath211 .
the cep b optical stars have an average extinction @xmath212 mag while s 155 stars have @xmath213 mag and cep ob3b stars have @xmath214 mag . within each of the areas , the extinctions of disk - bearing stars
are systematically higher than those of diskless stars .
as the inferred visual source extinctions differ by @xmath215 mag for the six trial combinations of distance and age , and the reddening vector on the optical color - magnitude diagram is almost parallel to the pms evolutionary tracks , the x - ray sources occupy very similar loci in the @xmath205 color - magnitude diagram for all six cases .
the x - ray sources are centered mostly around the @xmath94 myr isochrons assuming the distance of 725 pc , or the @xmath132 myr isochrone assuming the distance of 850 pc .
figure [ fig_cmd ] shows the color - magnitude diagram for 725 pc .
the plotted symbols stratify the 286 x - ray sources with optical photometry according to location in the star forming region and the presence of disks : 62 disk - bearing ( 143 diskless ) , 22 disk - bearing ( 31 diskless ) , and 17 disk - bearing ( 11 diskless ) for the cep ob3b , s 155 , and cep b subregions , respectively . in figure
[ fig_cmd ] , about thirty x - ray sources have @xmath216 locations inconsistent with any pms model tracks .
these outliers are not foreground or background stellar contaminants , as more than half of them have optically thick disks and very few field stars are predicted to be captured in the @xmath7 image .
most of these outliers have unusually high near - ir extinctions @xmath204 among the x - ray young stars seen in the optical bands . for the majority of the outliers , the @xmath204 values are comparable ( within @xmath217 ) to the absorption @xmath218 inferred from the x - ray median energy , @xmath219 .
@xmath219 is first converted to equivalent hydrogen column density assuming solar abundances using the @xmath220 relation found for the coup sample @xcite , and @xmath221 is converted to @xmath218 assuming a gas - to - dust ratio of @xmath222 @xmath223mag @xcite .
this may point to discrepant optical magnitudes and/or unusual conditions in these pms systems ; for example , some may posses disk at high inclination where the optical light is enhanced by scattering above the disk .
alternatively , these outliers may have ages overestimated by incorrect consideration of birth line and accretion effects ( e.g. * ? ? ? * ) . from a comparison of soft x - ray absorptions and fractions of @xmath224-band excess sources , @xcite presented evidence that the embedded cep b cluster is younger than the cep ob3b population within the @xmath7 acis - i field .
figure [ fig_cmd ] supports this age difference .
ignoring outliers mentioned above , about half of the cep b sources ( blue symbols ) are located closer to the 1 myr track compared to only about one - sixth of the cep ob3b ( red symbols ) .
however , the cep b sample is small because most of the x - ray selected members are embedded and do not have optical counterparts .
the result supports the youthfulness of the cep b population seen in the disk fraction discussed in [ spatial_domain_subsection ] .
we thus can not arrive at a clear conclusion on the distance to the star forming complex .
however we do establish a dependence of the inferred average age on the assumed distance . in this paper , we adopt ( not derive ) a distance to the cep b / cep ob3b complex of 725 pc , supported by the recent vlbi parallax measurement of cep a. we therefore adopt an age of @xmath17 myr for the cep ob3b stars and @xmath171 myr for the embedded cep b stars .
if the distance is closer to 850 pc , the inferred ages would be decreased by @xmath176 myr .
originally found by @xcite from rosat data , it is now empirically well - established that pms x - ray luminosities are strongly correlated with stellar mass , volume and surface area .
the @xmath225 relationship extends over 3 orders of magnitude range in x - ray luminosity .
the clearest relationships are seen in the _ chandra orion ultradeep project _ ( coup ) observation of the orion nebula cluster @xcite and in the _ xmm - newton extended survey of taurus _ ( xest , * ? ? ?
* ; * ? ? ?
. the astrophysical cause of this relationship is poorly understood , but probably is due to saturation of the magnetic dynamo in the fully convective stellar interior or on the surface of pms stars .
figure [ fig_lx_vs_mass ] shows the @xmath226 relationship for 222 unobscured cep ob3b pms stars compared to 457 lightly - absorbed ( @xmath227 mag ) orion coup stars from @xcite . here , @xmath228 represents the quantity @xmath229 , the x - ray luminosity computed in the @xmath230 kev band and corrected for obscuration . a similar , roughly linear ,
relationship is seen in the cep ob3b sample , but somewhat flatter than the relationship present in the coup sample .
this can be attributed to truncation effects : the cep ob3b observation is complete to @xmath231 erg s@xmath93 @xcite , while the coup observation is complete to @xmath232 erg s@xmath93 .
examination of figure [ fig_lx_vs_mass ] shows that the cep ob3b x - ray sample alone is nearly complete down to @xmath4 m@xmath5 and @xmath9 m@xmath5 for the class iii and class ii stars , respectively ( see figure [ fig_imf ] ) .
much of the scatter in both relations can be attributed to errors in coup and cep ob3 stellar masses ; less scatter is seen in the xest @xmath233 plot where masses are carefully evaluated from accurate spectroscopy .
x - ray members of the embedded cep b cluster show a similar @xmath234 relationship with the completeness limit above @xmath6 m@xmath5 .
a second relationship between x - ray emission and accretion was originally found from @xmath7 hrc ( high resolution camera ) data @xcite and has been confirmed in the coup and xest samples .
this is thought to arise from a suppression of time - integrated x - ray emission in accreting versus non - accreting pms systems .
this is a more subtle effect than the x - ray dependence on mass with an amplitude of only a factor of @xmath132 within the @xmath235 range in pms x - ray luminosities @xcite .
the astrophysical cause of the mild suppression of x - ray emission in accreting pms systems is also uncertain .
preibisch et al .
suggest that x - ray emission can not arise in magnetic field lines that are mass loaded with disk material .
@xcite argue that the outer magnetosphere of accretors is stripped by interaction with the disk .
@xcite propose that soft x - ray emission is attenuated by dense material in accretion columns .
@xcite find indications of shorter x - ray flare durations in class ii systems that may be due to distortion and destabilization of magnetic loop structures in accreting systems .
this effect is shown in figure [ fig_lx_vs_mass ] with running medians of accreting and non - accreting stars for the lightly - obscured cep ob3b and coup samples .
the running medians are calculated within a sliding window with width 0.3 dex in @xmath236 .
the errors on medians are median absolute deviations ( see appendix b of * ? ? ?
the sample selections are not identical due to different x - ray sensitivity limits and due to different limited spectroscopic / photometric measurements in orion and cep ob3b regions .
we show with the * @xmath237 * symbol 62 cep ob3b stars classified as disk - bearing ` dsk ' systems in [ yso_classification_section ] , and 142 coup stars classified as active accretors having ca ii 8542 @xmath238 line in emission with equivalent width of @xmath239 .
the @xmath240 symbols indicate 160 cep ob3b stars classified as diskless ` nod ' and 315 weakly accreting or non - accreting coup stars with absorption equivalent width of @xmath241 .
thirteen cep ob3b transition disk stars are marked by squares .
the factor of @xmath132 offset between the accreting and non - accreting stars in the coup samples is clearly seen to be superposed on the dominant @xmath233 relationship , but the effect is only marginally present in the cep ob3b sample .
the weakness of the offset here can be attributed to two effects .
first , the incompleteness at lower-@xmath228 values discussed above has a greater impact on the weaker-@xmath228 accreting stars than the stronger-@xmath228 non - accreting stars .
second , due to the absence of a spectroscopic survey of cep ob3b stars , we classify stars here using an infrared photometric disk indicator rather than a direct accretion indicator .
it has been established within the coup sample that the presence of a disk itself does not suppress x - ray emission , but active accretion must be present @xcite .
these two effects plausibly explain the less pronounced observed effect of the cep ob3b class ii x - ray emission suppression compared to that of the coup stars .
we thank the anonymous referee for very useful comments that improved this work .
we thank robert gutermuth ( cfa ) for development of spitzer data analysis tools , and patrick broos and leisa townsley ( penn state ) for their development of @xmath7 tools .
we benefited from scientific discussions with jeroen bouwman , thomas henning and colleagues at mpia .
this work is supported by the chandra acis team ( g. garmire , pi ) through the sao grant sv4 - 74018 and by guest observer jpl grant # 1287737 ( j. wang , pi ) .
k. l. was supported by grant ast-0544588 from the national science foundation .
this work is based on observations made with the _
spitzer space telescope _ , which is operated by the jet propulsion laboratory , california institute of technology under a contract with nasa .
we also use data products of the two micron all sky survey , which is a joint project of the university of massachusetts and the infrared processing and analysis center / california institute of technology , funded by nasa and nsf .
280 & 0.1 & @xmath242 & @xmath243 & 0.0 & @xmath244 & @xmath245 & @xmath246 & aaa000 & 0.1 & @xmath247 & @xmath248 & @xmath249 & @xmath250 & 4444 + 281 & 0.4 & @xmath251 & @xmath252 & 0.4 & @xmath253 & @xmath254 & @xmath255 & aaa000 & 0.5 & @xmath256 & @xmath257 & @xmath258 & @xmath259 & 4444 + 282 & 0.3 & @xmath260 & @xmath261 & 0.3 & @xmath262 & @xmath263 & @xmath264 & aaa000 & 0.1 & @xmath265 & @xmath266 & @xmath267 & @xmath268 & 4444 + 283 & 0.1 & @xmath269 & @xmath270 & 0.1 & @xmath271 & @xmath272 & @xmath273 & aaas00 & 0.1 & @xmath274 & @xmath275 & @xmath276 & @xmath277 & 4444 + 284 & 0.3 & @xmath278 & @xmath279 & 0.3 & @xmath280 & @xmath281 & @xmath282 & aaac00 & 0.2 & @xmath283 & @xmath284 & @xmath285 & @xmath286 & 2222 + 285 & & & & 0.3 & @xmath287 & @xmath288 & @xmath289 & aaaccc & 0.2 & @xmath290 & @xmath291 & @xmath292 & @xmath293 & 3333 + 286 & 0.2 & @xmath294 & @xmath295 & 0.2 & @xmath296 & @xmath297 & @xmath298 & aaa000 & 0.2 & @xmath299 & @xmath300 & @xmath301 & @xmath302 & 4444 + 287 & 0.4 & @xmath303 & @xmath304 & 0.5 & @xmath305 & @xmath306 & @xmath307 & uua00c & 0.6 & @xmath308 & @xmath309 & @xmath310 & & 3349 + 288 & 0.0 & @xmath311 & @xmath312 & 0.1 & @xmath313 & @xmath314 & @xmath315 & aaa000 & 0.2 & @xmath316 & @xmath317 & @xmath318 & & 4449 + 289 & 1.3 & @xmath319 & @xmath320 & 1.6 & @xmath321 & @xmath322 & @xmath323 & aaa000 & 1.6 & @xmath324 & @xmath325 & @xmath326 & @xmath327 & 4444 + 290 & 0.2 & @xmath328 & @xmath329 & 0.3 & @xmath330 & @xmath331 & @xmath332 & aaa000 & 0.3 & @xmath333 & @xmath334 & & & 2299 + 291 & 0.5 & @xmath335 & @xmath336 & 0.5 & @xmath337 & @xmath338 & @xmath339 & aaa000 & 0.3 & @xmath340 & @xmath313 & @xmath341 & @xmath342 & 4444 + 292 & 0.4 & @xmath343 & @xmath344 & 0.4 & @xmath345 & @xmath346 & @xmath347 & aaa000 & 0.3 & @xmath290 & @xmath348 & @xmath349 & & 4449 + 293 & 0.4 & @xmath350 & @xmath351 & 0.4 & @xmath352 & @xmath353 & @xmath354 & aaa0s0 & 0.3 & @xmath355 & @xmath275 & @xmath356 & @xmath357 & 4444 + 294 & 0.3 & @xmath358 & @xmath359 & 0.5 & @xmath360 & @xmath361 & @xmath362 & aaa000 & 0.3 & @xmath363 & @xmath364 & @xmath326 & @xmath365 & 4444 + 295 & 0.1 & @xmath366 & @xmath367 & 0.2 & @xmath368 & @xmath369 & @xmath370 & aaa000 & 0.5 & @xmath371 & @xmath372 & @xmath373 & & 2229 + 296 & & & & & & & & & & & & & & 9999 + 280 & @xmath374 & 4 & 0.6 & 0.2 & dsk & & 000 & cepob3b & 0000 + 281 & @xmath375 & 4 & 0.5 & 0.5 & nod & & 000 & cepob3b & 0000 + 282 & @xmath376 & 4 & 0.4 & 0.8 & dsk & & 000 & cepb & 0000 + 283 & @xmath377 & 4 & 0.5 & 0.7 & nod & & 000 & cepob3b & 0000 + 284 & @xmath378 & 4 & 0.5 & 0.4 & nod & & 000 & cepob3b & 0000 + 285 & @xmath379 & 4 & 0.6 & 0.3 & dsk & & 000 & cepb & 0000 + 286 & @xmath380 & 4 & 0.5 & 0.5 & dsk & & 000 & s155 & 0000 + 287 & @xmath381 & 3 & 0.1 & 0.5 & nod & & 300 & cepb & 0000 + 288 & @xmath382 & 3 & 1.0 & 1.0 & nod & & 300 & s155 & 0000 + 289 & @xmath383 & 4 & & & frg & & 333 & cepob3b & 0000 + 290 & @xmath384 & 2 & & & frg & & 333 & cepb & 0000 + 291 & @xmath385 & 4 & 0.5 & 0.2 & dsk & td & 011 & cepob3b & 0000 + 292 & @xmath386 & 3 & 0.8 & 0.7 & dsk & td & 301 & s155 & 0000 + 293 & @xmath387 & 4 & 0.5 & 0.8 & nod & & 000 & cepob3b & 0000 + 294 & @xmath388 & 4 & 0.7 & 2.1 & nod & & 000 & cepob3b & 0000 + 295 & @xmath389 & 3 & 0.7 & 0.4 & dsk & & 300 & cepb & 0000 + 296 & & 0 & & & unc & & 333 & cepb & 0000 + 1 & 343.90431 & 62.619685 & @xmath390 & @xmath391 & @xmath392 & baacc0 & @xmath393 & @xmath394 & @xmath395 & @xmath326 & cepob3b & 1.3 & 0.20 + 2 & 343.90796 & 62.613150 & @xmath396 & @xmath397 & @xmath398 & aaa000 & @xmath399 & @xmath400 & @xmath401 & @xmath402 & cepob3b & 1.3 & 0.30 + 3 & 343.90916 & 62.601677 & @xmath403 & @xmath404 & @xmath405 & aaa000 & @xmath406 & @xmath407 & @xmath408 & @xmath409 & cepob3b & 1.2 & 0.43 + 4 & 343.91215 & 62.631087 & @xmath410 & @xmath411 & @xmath412 & aaa000 & @xmath413 & @xmath414 & @xmath415 & @xmath416 & cepob3b & 0.7 & 0.27 + 5 & 343.91421 & 62.627495 & @xmath417 & @xmath255 & @xmath418 & aaa000 & @xmath419 & @xmath420 & @xmath421 & @xmath422 & cepob3b & 0.7 & 0.61 + 6 & 343.91472 & 62.634968 & @xmath423 & @xmath424 & @xmath425 & aaa000 & @xmath298 & @xmath418 & @xmath426 & @xmath427 & cepob3b & 0.7 & 0.27 + 7 & 343.92033 & 62.641813 & @xmath428 & @xmath429 & @xmath430 & aaa000 & @xmath431 & @xmath432 & @xmath433 & @xmath434 & cepob3b & 1.0 & 0.24 + 8 & 343.92861 & 62.671552 & @xmath435 & @xmath436 & @xmath437 & aaa000 & @xmath438 & @xmath439 & @xmath440 & @xmath441 & cepob3b & 0.5 & 0.22 + 9 & 343.93061 & 62.645682 & @xmath442 & @xmath443 & @xmath444 & aaacc0 & @xmath445 & @xmath274 & @xmath446 & @xmath447 & cepob3b & 1.3 & 0.24 + 10 & 343.93414 & 62.631564 & @xmath448 & @xmath449 & @xmath450 & aaa000 & @xmath451 & @xmath452 & @xmath453 & @xmath454 & cepob3b & 0.5 & 0.25 + + dsk & 28 & 14 & 10 & 9 & 5 & 4 + nod & 34 & 14 & 14 & 4 & 4 & 3 + disk frac .
& @xmath455 & @xmath456 & @xmath457 & @xmath458 & @xmath459 & @xmath460 + + dsk & 10 & 11 & 4 & 3 & 9 & 4 + nod & 16 & 13 & 1 & 3 & 2 & 1 + disk frac . & @xmath461 & @xmath462 & @xmath463 & @xmath464 & @xmath465 & @xmath463 + + dsk & 38 & 25 & 14 & 12 & 14 & 8 + nod & 50 & 27 & 15 & 7 & 6 & 4 + disk frac . &
@xmath466 & @xmath467 & @xmath468 & @xmath469 & @xmath470 & @xmath471 + + class & inner & & intermediate & & outer & + & ( 0@xmath472 3 ) & & ( 4@xmath472 7 ) & & ( @xmath473 ) & + + dsk & 9 & & 30 & & 10 & + nod & 7 & & 24 & & 20 & + disk frac . & @xmath474 & & @xmath475 & & @xmath476 & + + dsk & 13 & & 17 & & 2 & + nod & 3 & & 13 & & 13 & + disk frac . & @xmath477 & & @xmath478 & & @xmath479 & + + dsk & 22 & & 47 & & 12 & + nod & 10 & & 37 & & 33 & + disk frac . & @xmath480 & & @xmath481 & & @xmath482 & + co emission from @xcite .
o stars ( cyan circles ) , b0-b3 ( green ) , b4-b9 ( red ) are from @xcite .
the black dashed line demarcates the older sub - group cep ob3a from cep ob3b .
north is up , east is to the left .
( b ) closeup of the cep b region showing the @xmath483 irac image in the 4.5 @xmath24 m band .
the ionizing sources of the region , the o7 star hd 217086 and b1 star hd 217061 , are marked .
the magenta contours show the @xmath484co emission measured by @xcite with the central hot core marked .
the @xmath35 @xmath7 acis - i field is outlined in cyan .
the h ii s 155 interface between the cep b molecular cloud and unobscured cep ob3b association is schematically indicated in cyan .
[ fig_map],width=672 ] acis - i image in the @xmath485 kev band ( blue ) is superposed on the @xmath8 irac composite image in the 3.6 @xmath24 m ( green ) and 5.8 @xmath24 m ( red ) bands .
the @xmath35 @xmath7 acis - i field is outlined in blue and the @xmath25 field of the @xmath8 irac mosaic in 3.6/5.8 @xmath24 m bands is outlined in green .
[ fig_color_image],width=672 ] ( diamond ) and irac - band ( square ) flux points with usually small errors .
the dashed and solid lines give the original and ( de)reddened ic 348 median sed from @xcite fitted to the data .
the top two lines of the panel labels give information from table [ tbl2 ] .
the third line gives the spectral class of the ic 348 median sed template from @xcite and reddening applied to the original template sed to fit the observed cepheus source sed .
see the electronic edition of the journal for the full atlas of ir seds of the cepheus x - ray young stars .
[ fig_sed_nod],width=672 ] for ( a ) x - ray stars in the cepheus field and ( b ) comparisons in ic 348 .
colors indicate pms evolutionary classes : diskfree class iii stars ( red ) , disk - bearing class i and ii stars ( blue ) , and transition disk class ii / iii stars ( green ) .
[ fig_slope_histogram],width=624 ] and short irac channels ( bottom ) .
colors indicate pms evolutionary classes : class iii ( red ) , class i and ii ( blue ) , and class ii / iii transition disk stars ( green ) .
the reddening vectors show @xmath486 mag using the extinction law from @xcite .
legends indicate number of stars with available photometry .
[ fig_ccds],width=672 ] -[4.5]$ ] versus @xmath83-[8.0]$ ] color - color diagrams employed to select pms candidates for ( a ) cep ob3b , ( b ) s 155 , and ( c ) cep b subregions . dashed lines show classification criteria .
bottom : @xmath84 $ ] versus @xmath84-[8.0]$ ] color - magnitude diagrams for ( d ) cep ob3b , ( e ) s 155 , and ( f ) cep b subregions showing x - ray emitting pms stars ( @xmath240 ) and non - x - ray pms candidates ( circles ) .
panel ( d ) also shows x - ray emitting extragalactic contaminants ( diamonds ) .
reddening vectors represent @xmath486 mag .
[ fig_ccd_cmd_nonxray],width=672 ] ( red ) and combined class ii @xmath7 and non-@xmath7 ( blue ) pms stars in the ( a ) cep ob3b , ( b ) s 155 , and ( c ) cepb sub - regions .
the numbers of stars are estimated within mass bins of 0.2 dex width and @xmath487 equivalent poisson errors are shown .
annotation labels indicate the corresponding to each bin number of detected @xmath7 class iii stars ( red ) , as well as @xmath7 @xmath3 non-@xmath7 class ii stars ( blue ) .
the dashed lines represent the scaled versions of the galactic field imf @xcite .
[ fig_imf],width=528 ] -band magnitude for four @xmath17 myr stellar clusters : cep ob3b ( circles with 1@xmath78 error bars ) ; cha i ( solid line ) ; ic 348 ( dashed line ) ; and @xmath78 ori ( dashed - dotted line ) .
the cep ob3b stellar sample is the combined class ii / iii @xmath7 and class ii non-@xmath7 pms stars from this work .
data for the three other clusters are from @xcite .
the top axis shows an approximate scaling to stellar mass .
[ fig_mj_diskfraction],width=624 ] plus non-@xmath7 mass - complete selected pms samples superposed on a grey - scale 5.8 @xmath24 m band @xmath8-irac m@xmath5 stars , and panel ( b ) shows @xmath488 m@xmath5 stars .
the o7vn star hd 2170786 is marked by the cyan square - circle , and the molecular hot core by the cyan cross .
the cyan dashed locus delineates the s 155 subregion .
the @xmath7 field of view is outlined by the large cyan square , and the ne and sw azimuthal zones are demarcated by the magenta line .
[ fig_spat_distrib],width=672 ] plus non-@xmath7 sample of pms stars .
the three subregions are separated by dashed lines and the green vertical arrow indicates the projected location of the primary ionizing star hd 217086 .
mass - stratified sub - samples are color coded : @xmath489 m@xmath5 ( red ) , @xmath490 m@xmath5 ( black ) , and @xmath491 m@xmath5 ( blue ) . an interpretation based on three constant disk fraction layers
is indicated by horizontal grey bars with their disk fraction levels consistent with those from table [ tbl_frac ] .
[ fig_disk_frac_spatial],width=672 ] color - magnitude diagram showing the cep b / cep ob3b x - ray selected pms stars individually corrected for extinction .
disk - bearing stars are plotted as open squares , and diskfree stars as filled circles .
colors indicate subregion location : cep ob3b ( red ) , s 155 ( green ) , and cep b ( blue ) .
solid lines show pms isochrones for ages 1 , 2 , 3 , 4 , 5 myr and the zero - age main sequence from evolutionary tracks of @xcite assuming @xmath492 pc .
dashed lines show mass tracks 0.3 , 1.0 , and 3.0 m@xmath5 .
the reddening vector shows @xmath493 mag .
[ fig_cmd],width=624 ] symbols are diskless or non - accreting stars , * @xmath237 * are optically thick disk or accreting stars , and * @xmath494 * are transition disk objects .
running medians for @xmath495 are shown as dashed ( solid ) lines for diskless or non - accreting ( optically thick disk or accreting ) stars .
[ fig_lx_vs_mass],width=672 ] | the cepheus b ( cep b ) molecular cloud and a portion of the nearby cep ob3b ob association , one of the most active regions of star formation within 1 kpc , have been observed with the irac detector on board the _ spitzer space telescope_. the goals are to study protoplanetary disk evolution and processes of sequential triggered star formation in the region . out of @xmath0 pre - main sequence ( pms )
stars selected with an earlier _ chandra x - ray observatory _ observation , @xmath1 are identified with mid - infrared sources and most of these are classified as diskless or disk - bearing stars .
the discovery of the additional @xmath2 ir - excess low - mass members gives a combined _ chandra_@xmath3_spitzer _ pms sample that is almost complete down to @xmath4 m@xmath5 outside of the cloud , and somewhat above @xmath6 m@xmath5 in the cloud .
analyses of the nearly disk - unbiased combined @xmath7 and @xmath8 selected stellar sample give several results .
our major finding is a spatio - temporal gradient of young stars from the hot molecular core towards the primary ionizing o star hd 217086 .
this strongly supports the radiation driven implosion ( rdi ) model of triggered star formation in the region .
the empirical estimate for the shock velocity of @xmath9 km / s is very similar to theoretical models of rdi in shocked molecular clouds .
the initial mass function ( imf ) of the lightly obscured triggered population exhibits a standard galactic field imf shape .
the unusually high apparent value of @xmath10 star formation efficiency inferred from the ratio of star mass to current molecular gas mass indicates that most of the cep b molecular cloud has been already ablated or transformed to stars .
contrary to the current rdi simulations , our findings indicate that star formation triggering by hii region shocks is not restricted to a single episode but can continue for millions of years .
other results include : 1 .
agreement of the disk fractions , their mass dependency , and fractions of transition disks with other clusters ; 2 .
confirmation of the youthfulness of the embedded cep b cluster ; 3 .
confirmation of the effect of suppression of time - integrated x - ray emission in disk - bearing versus diskless systems . |
the evolution of galaxies is influenced by how quickly gas is consumed by the formation of stars .
many empirical star formation recipes assume that the star formation rate depends on the gas surface density ( e.g. , @xcite ) .
however , observations of spiral galaxies show that the molecular gas is concentrated along spiral arms , implying that star formation is triggered by the spiral structure itself .
the strongest evidence for the important role of spiral structure is based on a gravitational stability threshold , below which star formation ceases @xcite .
nevertheless , above this threshold , azimuthally averaged data have shown that these empirical star formation laws are obeyed ( @xcite , @xcite ) . in this paper
we make use of recent high resolution mapping to estimate the rate and efficiency of star formation along individual spiral arms .
we aim to investigate the influence of the spiral structure itself on the process of star formation .
m51 is a magnificent example of a nearby grand design spiral galaxy .
because of its large angular size and strong two - armed spiral structure , it has been the focus of studies of spiral density wave driven star formation ( e.g. , @xcite , @xcite ) .
while the stellar density contrast in the spiral arms in m51 is high compared to many spiral galaxies @xcite , it is still far below the density contrasts seen in molecular or ionized gas .
the distribution of molecular gas and associated star formation are particularly sensitive to the underlying spiral density waves propagating through the galaxy .
@xcite in their fourier analysis of the k - band light distribution measured significant deviations from bi - symmetric ( @xmath1 ) symmetry in the galaxy .
because shocks occur in molecular gas , even weak perturbations in the stellar density distribution can cause large density variations in the molecular gas distribution . by measuring the shape and location of the peak densities seen in molecular gas , and comparing the shapes of the spiral arms with the star formation rates measured from the pa@xmath0 emission , we probe for differences in star formation rate and efficiency that might be caused by the spiral density waves themselves . in this paper
we take advantage of recent high angular resolution observations of m51 to measure the star formation efficiency along individual spiral arms .
the bima survey of nearby galaxies ( bima - song , @xcite ) is the first systematic interferometric survey of the disks of nearby galaxies in the co(j=1 - 0 ) emission line .
prior to bima - song , the overwhelming majority of galaxies observed in co had been observed with single dish telescopes , at much lower angular resolution . with a typical beam full width half max of @xmath2 ,
bima - song represents a major improvement in our ability to resolve the molecular gas distribution in a significant sample of nearby galaxies . to measure the gas depletion time , which is inversely related to the star formation efficiency
, we combine a measurement of the gas density with an estimate of the star formation rate . to measure the star formation rate , we use archival nicmos data in the pa@xmath0 recombination line , at 1.87@xmath3 m .
the pa@xmath0 recombination line is better suited toward probing the possibly obscured central regions , since the near infrared is less affected by dust extinction .
the high angular resolution of the nicmos images on board hst facilitates continuum subtraction in crowded regions . in section 2
, we describe the archival observations used for this study . to measure quantities along spiral arms
, we must first determine their locations .
in section 3.1 we describe our procedure for doing this and discuss our measurements that describe the shape of the arms . in section 3.2
we discuss the relative differences in gas density , star formation rate and gas depletion time measured in different spiral arm regions . in section 3.3
we discuss the effect of a possible additional spiral density wave on the gas density and on the symmetry of the spiral arms previously identified .
a summary and discussion follows . for m51
we adopt a distance of 7.7 mpc @xcite , so @xmath4 on the sky corresponds to a distance of 37 pc .
nicmos narrow band ( f187n and f190n ) images were obtained as part of the nicmos gto program @xcite with a mosaic of nine fields covering an area of about @xmath5 .
see @xcite for more information about these observations .
the f187n and f190n filters , with effective wavelengths of 1.87 and 1.90@xmath3 m contain pa@xmath0 line and continuum images , respectively .
we scaled and subtracted the f190n continuum images from the f187n images and combined them to form one mosaic .
the pixel scale of the nicmos images is @xmath6 , so one pixel is 7.6 pc long in m51 .
the nicmos images were calibrated using zero points listed in the hst data handbook .
specifically , we used @xmath7 erg @xmath8 s@xmath9/(dn s@xmath9 ) for the continuum subtracted f187n narrow band images .
the bima - song database contains a collection of spatial velocity data cubes . for this study
, we used an image made of the intensity integrated over all velocity channels .
for m51 , large scale fluxes were recovered in the interferometric data from single dish observations @xcite . to compare the co image to the pa@xmath0 image
we must ensure that the same areas of the sky can be measured on the same scale .
therefore , we resampled the co image to be at the same pixel scale as the pa@xmath0 mosaic .
both images are inclination corrected and rotated so that the minor axis is at a p.a .
of 170 degrees .
the resulting images are shown in figures 1 and 2 .
to determine the location of the spiral arms , we model the spiral density waves in the disk with the following function , that describes a logarithmic spiral @xmath10 where @xmath0 describes how tightly wound the arm is , @xmath11 is an angular offset , and @xmath12 is an integer that corresponds to the number of spiral arms . in the case of m51 ,
the spiral structure is predominantly two armed ( @xmath1 ) .
one maximum is then described by @xmath13 for each region we measured parameters @xmath14 and @xmath11 by choosing several positions of peak intensity along the spiral arms in the co intensity map .
all positions were measured from the center of the galaxy , which was determined from the centroid of an h band ( f160w ) image of m51 that was observed simultaneously with the pa@xmath0 data @xcite . since the nuclear region of m51 contains little molecular gas ( see figure 1 ) , the h band centroid accurately measures the location of the stellar density peak .
we defined the angle @xmath15 to be toward the right in figures 1 and 2 . from these measurements ,
sets of coordinates , ( @xmath16 ) , were measured in each arm .
linear regression was used to determine @xmath14 and @xmath11 in each region .
different regions in the spiral arms were fit separately , as a single spiral did not provide an adequate fit to the entire galaxy .
the fits to the 8 selected spiral sub - regions are illustrated in figure 1 , and the corresponding parameters that describe them , are listed in table 1 . in figure 2
we compare our fit to the spiral arms traced in co to the pa@xmath0 emission . while the pa@xmath0 is coincident with the molecular gas in some of the spiral arms , in other regions an offset is observed . @xcite and @xcite noted an offset between co and h@xmath0 in the sense that h@xmath0 emission was at somewhat larger radii than the co emission .
we find that the pa@xmath0 emission is offset from the co emission at radii 1.2 to 2.8 kpc in the spiral arms 2b and 2c .
however , we can not tell if there is an offset between the pa@xmath0 emission and co emission in spiral arm 1c because the edge of the pa@xmath0 image is near the position of the co emission .
we do note hints of pa@xmath0 emission , to the south of spiral arm 1c , lying on the very edge of the nicmos mosaic .
if there is pa@xmath0 emission to the south of the mosaic , then a significant offset between the co and pa@xmath0 is present in this arm .
in addition to observing offsets between co and pa@xmath0 , we can look for deviations from point symmetry by plotting the reflection of the spiral arms determined in table 1 . in figure 2
we show the fit to the spiral arms , rotated through an angle of 180 degrees . in the case of perfect @xmath1 symmetry
, the over - plotted curve would lie on top of the peak intensities .
we observe that the most inner and outer spiral arm regions appear to be symmetric , but for intermediate radii , the observed spiral arms are offset from the curves . to verify this , in figure [ offsets ] we plot @xmath17 as a function of @xmath18 for the a , b , and c spiral arms listed in table 1 , with the northern spiral arm rotated by an angle of 180 degrees .
we note that the deviation from point symmetry is present over much of the inner part of the galaxy , but is most apparent for the b spiral arms , where the offset , @xmath19 degrees . in this section
we present our measurements of the star formation rate ( sfr ) and co mass , per unit length , along each segment of the spiral arms .
figures 4 , 5 , and 6 show the molecular gas density in each spiral arm segment and the star formation rate per unit length along the spiral arm segments .
in addition , we calculated the gas depletion time from the ratio of the co mass to the star formation rate . to measure these quantities , we used our own idl procedure to measure the fluxes in particular regions .
each region was centered in radius on the position of a logarithmic spiral curve determined from our study described in section 3.1 ( see figures 1,2 ) .
the regions were defined by intervals in @xmath18 and @xmath17 , with @xmath20 radians and @xmath21 pixels or approximately 400 pc .
the same regions were used to measure fluxes in both the co image and the pa@xmath0 image , so that a one to one comparison could be made at each location .
we ensured that @xmath22 was large enough to include the pa@xmath0 emission , where offsets between pa@xmath0 and co were observed .
the measured fluxes were divided by arc length , so that integrated fluxes were computed per pc along each arm .
background sky values were determined in the pa@xmath0 image by taking the median of the regions adjacent to each box , on both sides of the spiral arms .
only spiral arm regions 1a , 1b , 2a , 2b , and 2c are compared , as regions 1c , 1d , and 2d do not lie on the nicmos pa@xmath0 mosaic .
we next convert to physical units assuming an intrinsic flux ratio of @xmath23 for case b recombination at a temperature of @xmath24k and a density @xmath25 @xmath26 ( line ratios are given in @xcite ) .
then we converted the h@xmath0 fluxes to an star formation rate , using a factor from @xcite , @xmath27 the co data were converted to m@xmath28 pc@xmath29 using a co to h@xmath30 conversion factor of @xmath31 h@xmath30 @xmath8 ( k km / s)@xmath9 @xcite . in figures
[ gdt1 ] , [ gdt2 ] , and [ gdt3 ] , we show the molecular gas mass , star formation rate , and gas depletion time , per unit length , for arms covered by both images .
the plots are arranged such that the top plots show the molecular gas mass per pc , along the spiral arm segment .
the middle plots show the star formation rate per pc , as measured from the pa@xmath0 mosaic .
the bottom plots show the ratios of the points in the two plots above them . this quantity is known as the gas depletion time , the time in which all of the molecular gas would form into stars , if the star formation rate was to remain constant . in figure [ gdt1 ] , spiral arm 1a was placed next to its opposite spiral arm , 2a , and the same is true for arms 1b and 2b , in figure [ gdt2 ] . in figure
[ gdt3 ] , spiral 2c is shown alone , because we were unable to measure pa@xmath0 emission for spiral arm 1c .
slightly negative star formation rates , the result of imperfect sky subtraction in the pa@xmath0 image , are assumed to be sites where essentially no stars are being formed .
figures [ gdt1 ] and [ gdt2 ] can be used to look for differences in the gas density and star formation efficiency between a spiral arm and its partner on the opposite side of the galaxy . in figure [ gdt1 ]
we note that the gas density is twice as high on nw side ( 1a ) than the se side ( 2a ) . however , the star formation efficiency is about the same with a gas depletion time @xmath32 years . in figure [ gdt2 ]
we note a similarity , in the sense that the gas density is again twice as high in the western spiral arm ( 1b ) as in the eastern spiral arm ( 2b ) . both spiral
arms 1b and 2b have a jump in gas depletion time ( a decrease in star formation efficiency ) , at a radius of 1.5 kpc . at radii smaller than 1.5 kpc ,
the gas depletion time in spiral arm regions 1b and 2b is similar to that in 1a and 2a . at larger
radii the gas depletion time is several times larger than within this radius .
star formation in the outer parts of spiral arms 1b and 2b is inefficient compared to that in the inner parts of the same arms . past a radius of 2 kpc
the gas depletion time drops back to @xmath33 years , and remains similar in spiral arm 2c .
we suspect that the star formation efficiency in arm 1c would be similar to that in 2c , although we are unable to observe it .
except for the outer parts of spiral arms 1b and 2b at a radius of about 1.5 to 2kpc , the star formation efficiency or gas depletion time is remarkably similar in all arms , and approximately @xmath34 yr , despite large variations in gas density and the star formation rate . in the outer parts of 1b and 2b the gas depletion time is longer , @xmath35 years , despite the high molecular gas densities in these arms .
in this section we consider a dynamical model that can explain the degree of deviation from @xmath1 symmetry that we have measured in the spiral arms of m51 . by expanding galaxy images in fourier components
, @xcite noted that many galaxies exhibit hidden three armed components and consequently suggested that multiple spiral density waves often propagate simultaneously in galaxy disks .
@xcite in their fourier expansion of the near - infrared images of m51 saw both @xmath36 and @xmath37 components as well as the dominant 2-armed structure .
@xcite found that the galaxy was lopsided at all radii but also measured a weak peak in the @xmath37 component of the k - band images at a radius of about @xmath38 .
this radius is approximately at the location where we have found the largest asymmetry in the spiral structure , although from figure 3 we note that the asymmetry is present throughout most of the galaxy .
we now consider the possibility that more than one spiral density wave pattern is present in the disk of m51 .
below we estimate the likely position of molecular gas along spiral arms resulting from more than one perturbation .
we assume that the perturbations in the gravitational potential resulting from spiral modes can be described in terms of fourier components @xmath39 where @xmath40 and @xmath41 are amplitude and phase shift , and @xmath42 describes an @xmath12 armed logarithmic spiral ( see section 3.1 ) .
because spiral density waves travel at pattern speeds above the sound speed in the interstellar medium , the passage of spiral modes in the disk causes shocks in the gas .
the concentration of molecular clouds along spiral arms is interpreted to be a result of these galactic scale shocks .
shocks in the gas are likely to occur near maxima in the stellar density field , approximately corresponding to minima in the gravitational potential . these are found at angles where @xmath43 or @xmath44 this equation is also valid at potential maxima where shocks will not occur .
we assume that one spiral component dominates ; in the case of m51 this would be the @xmath1 or two - armed mode . in the absence of other modes , the potential minima ( density maxima )
occur near angles @xmath45 where @xmath46 is an integer .
we expand equation [ extreme ] about the angle @xmath47 , for @xmath48 , where the offset , @xmath49 , is the difference between the location of density maximum caused by only one @xmath1 wave and that resulting from two spiral density waves .
to first order in @xmath50 , @xmath51 = 0\ ] ] where we have assumed only one spiral density wave with @xmath52 , in addition to the dominant two - armed one .
we solve for @xmath53 @xmath54 substituting for @xmath47 ( equation [ phimaxdef ] ) and taking the limit @xmath55 @xmath56 we use our model to estimate the angular offset that would be measured between a spiral arm and its opposite one after a rotation of 180 degrees . for a dominant two - armed mode , each arm has a different integer @xmath46 in equation ( [ dphi ] ) ; @xmath57 or @xmath58 . only when @xmath12 is odd will the angular offsets be asymmetric , causing reflection symmetry about the origin to be broken .
the angular offset between one spiral arm and the other that has been rotated about @xmath59 is then given by subtracting the @xmath57 offset from that of the opposite one ( @xmath60 ) , @xmath61 for odd @xmath12 .
we expect that the angular difference depends upon the amplitude of the additional spiral density wave and on its wave number .
the angular offset should increase as the amplitude of the additional spiral density wave increases . in a region where the amplitudes of the two dominant spiral density waves are constant
, we expect that the angular offset would be more open ( less tightly wound ) than the spiral density waves themselves .
we now relate the size of the angular offset to the one expected based on observed density perturbations or observed fourier components . in the wkb approximation @xmath62 where @xmath63 is magnitude of the @xmath12th fourier component of the mass surface density @xcite , so that @xmath64 substituting this into equation([dphi ] ) and using the ratios @xmath65 , @xmath66 , and @xmath67 at @xmath68 measured by @xcite we estimate @xmath69 where the angles are given in radians . assuming @xmath70 , and for a maximum value of the angular offsets ( @xmath71 with respect to @xmath41 ) the angular shift could be as large as 60 degrees for @xmath36 , and 20 degrees for @xmath37 .
we compare the estimated model angular offsets to those we measured in m51 .
recalling from figure [ offsets ] , the offset measured between the 1b and 2b spiral arms is @xmath72 degrees .
so the @xmath37 fourier component observed by @xcite is sufficiently large to account for the angular offset we measured between these two arms .
a three - armed mode , in addition to the dominant two - armed one , consistent with perturbations seen in the k - band fourier components , could explain the asymmetry in the locations of the two spiral arms seen in molecular gas .
figure [ offsets ] shows that while the asymmetry is present over much of the galaxy , the maximum offset occurs where @xcite observe a peak in the @xmath37 mode .
the @xmath36 mode , on the other hand , is strong over much of the entire galaxy and does not single out any particular radial region , though it could be responsible for asymmetry at larger and smaller radii , where the @xmath37 amplitude is weak .
the decrease of the angular offset at small and large radii ( @xmath73kpc and @xmath74kpc ) can be explained by the decrease in amplitude of the @xmath37 wave .
the nearly constant value of the angular offset between @xmath58 and @xmath75 kpc suggests that the wave numbers of the @xmath1 and @xmath37 modes are approximately the same size ( @xmath76 , see equation [ offset_eqn ] ) . if there is both @xmath1 and @xmath37 waves , then at least one density peak of the @xmath37 wave will lie in between the dominant @xmath1 arms .
we might expect inter - arm molecular gas and associated inter - arm star formation in the same region where we see the deviations from bi - symmetry .
a more careful inspection of figure 1 shows that there is a patch of interarm star formation within spiral arm 2b to the north east of the nucleus .
as we have done for the molecular gas distribution ( see section 3.1 ) , we crudely fit a spiral shape to the the pa@xmath0 emission in this region .
the resulting fit is shown in figure [ tri ] , along with the same arm rotated by 120 and 240 degrees . from this figure
, we can determine where the @xmath37 and @xmath1 waves are likely to constructively add and where they interfere .
the @xmath37 wave is coincident with the dominant spiral arm in the west ( 1a-1b ) , but is offset from the spiral arm in the southeast ( 2a-2b ) .
we find that the @xmath37 wave can account for the molecular gas density enhancement in the western spiral arms , relative to the eastern spiral arms ( see figures [ gdt1 ] , [ gdt2 ] , and [ gdt3 ] ) .
the addition of an extra wave should also introduce an asymmetry in the velocity field .
if this asymmetry is sufficiently large then it could be detected in the co velocity field .
to search for this we examined the mean velocity mom1 bima map of m51 .
we first subtracted the systemic velocity of the galaxy from the velocity field and then compared this field to the same field that has been rotated by 180 degrees and multiplied by @xmath77 .
we find that the line of sight velocity on the southern side of the galaxy 2kpc from the nucleus is about 20 - 30 km / s above that on the opposite side .
the velocity asymmetry seems to be largest in the same region where we see the other asymmetries .
the magnitude of the velocity asymmetry should be something that our simple model could predict .
however , it depends on the pattern speed of the two waves and the details of the gas dynamics so a simulation is probably required to make a prediction sufficiently accurate to compare to the complex streaming motions seen in the co channel maps .
in this paper we have fit the spiral arms of m51 as traced in co within 3kpc of the nucleus with logarithmic spirals .
we have detected an asymmetry between the locations of the arms on one side of the galaxy compared to those on opposite side .
the asymmetry corresponds to a maximum angular offset of about @xmath78 degrees between one arm and the opposite one rotated by 180 degrees , and is most prominent between 1 and 2.2 kpc from the nucleus .
the molecular gas density in this region is also asymmetric with one spiral arm about twice as dense as that on the opposite side . in the same radial region
, there is a patch of star formation ( seen in the pa@xmath0 images located between the dominant spiral arms to the north east of the galaxy nucleus . to explain the angular offset seen between one arm and the opposite one rotated by @xmath59
, we consider a model that contains two spiral density waves ; a weak @xmath79-armed wave in addition to the dominant two - armed one .
when the additional wave is odd - armed , we predict that the position of the two dominant arms deviate from point symmetry in the manner we have observed in m51 .
we find that an additional @xmath37 mode , consistent with the radial location and strength of the @xmath37 fourier components previously measured at k - band by @xcite , is strong enough to account for the angular offset we have measured between the two dominant arms .
this additional spiral density wave also is consistent with the lopsided distribution of molecular gas , and the inter - arm star formation to the north east of the galaxy nucleus . despite molecular gas density variations of a factor of a few and corresponding star formation rate variations ( measured from the pa@xmath0 emission )
, we see little variation in the computed gas depletion times .
this suggests that the star formation efficiency is not largely affected by the large gas density variations caused by the presence of an additional spiral density wave .
between 1 and 2.2 kpc in m51 we see large scale deviation from pure @xmath1 or bi - symmetric structure : an angular offset between the spiral arms , differences in density on one side compared to the other side , and inter - arm star formation . all of these three asymmetries are most pronounced in the same radial region suggesting that they are related .
we have suggested here that this phenomena can be explained with one simple unifying model that includes an additional weak 3-armed spiral density wave .
the additional wave does not strongly influence the star formation efficiency , however because of the large gas density variations caused by it , the wave does influence the star formation rate .
in particular , large molecular gas densities and associated high star formation rates are likely to occur where and when the 3-armed and 2-armed mode both reach their density maxima simultaneously .
we expect that the 3-armed wave has a different pattern speed ( angular rotation rate ) than the 2-armed wave .
consequently the star formation rate which is highest when the 3 and 2 armed patterns constructively add , must be highly variable .
even low amplitude waves can cause large changes in the gas density . despite the low estimated amplitude of the 3-armed wave ( @xmath80 ) it is likely that it is causing large variations in the molecular gas density , and consequently in the star formation rate .
the simultaneous propagation of 2 and 3 armed structure must cause associated large variations in the local star formation rate . since our model contains more than one spiral density wave we would more strongly support scenarios that include multiple and transient spiral density waves ( e.g. , @xcite ) rather than those that focus on a single dominant quasi - steady mode ( e.g. , @xcite ) .
the model containing two and three armed spiral density waves , that we have proposed to explain the asymmetries in the central 3kpc of m51 , does not account for all the phenomena we have described in this paper . for example , this model does not account for the drop in star formation efficiency observed in the outer parts of spiral arms 1b and 2b between 1.5 and 2 kpc from the nucleus .
it also does not account for the offsets between the co and pa@xmath0 emission which are largest at radii greater than 2kpc .
similar offsets have previously been discussed by @xcite who have offered a variety of explanations including the triggering of star formation by spiral density waves , uneven heating and disassociation by the resulting star formation and differential clumping of the molecular gas .
however , it will be tempting to search for an additional explanation for these offsets and the decrease in the star formation efficiency we see in the outer arms 1b and 2b that also involve more than one spiral density wave .
spiral structure exists in m51 over a large range of radius . in section 3.1
we attempted to fit as large as possible pieces of the spiral structure with as few as possible individual components .
however , it is not clear from figure 1,2 or 3 that the ranges of radius over which we fit individual arm pieces correspond to separate spiral density waves , traveling at individual pattern speeds .
it is likely that the inner spiral arms of m51 rotate at a different pattern speed than the outer ones , and the interplay between these waves might provide an alternative explanation for the offsets observed between the co and pa@xmath0 emission or the low star formation efficiency in the outer parts of spiral arms 1b and 2b .
future work can further test the possibility that multiple spiral density waves propagate in galaxies and explore the role that these multiple modes exert on the distribution of gas and resulting pattern of star formation .
future studies can extend this type study carried out here to more galaxies , extend the multi - wavelength coverage and exploit the velocity fields that are available in the co and hi data .
we thank joel green , judith pipher , dan watson , bruce elmegreen , alar toomre , mousumi das and peter teuben for stimulating discussions .
we are grateful to tamara helfer for providing us with access to the full bima song galaxy sample data in advance of publication .
cccc 1a & -7.60 & -0.058 & @xmath81 + 1b & -4.08 & -6.10 & @xmath82 + 1c & -5.88 & -6.55 & @xmath83 + 1d & -4.88 & -6.09 & @xmath84 + 2a & -5.54 & -3.58 & @xmath85 + 2b & -4.86 & -3.44 & @xmath86 + 2c & -9.00 & -5.00 & @xmath87 + 2d & -2.98 & -1.78 & @xmath88 + | in the inner 3kpc of m51 we find that logarithmic spirals provide good fits to the peak intensities in molecular gas observed by bima in the co ( j=1 - 0 ) emission line along the spiral arms .
however , we measure significant asymmetries between the location and density of the arms observed on one side of the galaxy compared to those on the opposite side . between a radius of 1 and 2.2 kpc the gas distribution traced in co
is lopsided with densities twice as large in one arm compared to the opposite one . in the same region ,
the spiral arms are offset by 20 degrees from the position of the arm on the opposite side of the galaxy after a rotation of 180 degrees .
we use the ratio of co emission to that in the pa@xmath0 hydrogen recombination line to estimate the efficiency of star formation along each arm . except for a narrow region at about 2kpc where star formation is surprisingly inefficient , the gas depletion time is approximately 200 million years despite large variations in the molecular gas density in the arms .
we account for the deviations from two armed bi - symmetric structure with a simple model that includes an additional three - armed spiral density wave .
this model accounts for the angular offset between the arm on one side compared to its opposite , the lopsided distribution in molecular gas , and inter - arm star formation to the north east of the galaxy nucleus .
since the star formation efficiency is unaffected by the variations in gas density , and the variations in gas density can be accounted for by the presence of an additional three - armed density wave , we suggest that the star formation rate is variable , and is highest where and when the maxima of the two spiral density waves coincide or constructively add .
the inner region of m51 provides good evidence for the presence of more than one spiral density wave and a resulting variable rate of star formation . |
the evolutionary stage of a powerful extragalactic radio source is currently thought to be related to its linear size .
statistical studies of the population of radio sources in the first stages of their individual evolution are fundamental for a comprehensive understanding of the radio emission phenomenon and its duty - cycle .
+ compact steep - spectrum ( css ) and gigahertz - peaked spectrum ( gps ) radio sources are intrinsically small - sized ( linear size , ls @xmath115 - 20 kpc ) , powerful ( @xmath2 w hz@xmath3 ) extragalactic objects generally associated with distant ( [email protected] ) galaxies and quasars .
their main characteristic is the steep radio spectrum ( @xmath5 0.5 , @xmath6 ) in the optically thin regime , that flattens and turns over at low frequencies , between 1 ghz and 30 mhz @xcite .
the genuine youth of these objects was strongly supported by the determination of both the kinematic ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and radiative ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) ages of a dozen of the most compact objects ( ls @xmath1 100 pc ) which turned out to be about 10@xmath710@xmath8 years . given their young age and the intrinsically small size , css / gps sources provide us with a unique opportunity to study how the radio emission evolves and which role the ambient medium plays on their growth during the first stages of their evolution .
a dense , inhomogeneous environment may slow down the expansion of the radio source in the case one jet interacts with a dense cloud .
although the confinement may not last for the entire source lifetime , it may cause an underestimate of the source age .
+ the radio morphology of css / gps radio sources closely resembles that of fanaroff - riley type - ii radio galaxies @xcite , but on much smaller scale .
their radio structure is usually termed _
symmetric _ in the sense that the radio emission is found on the two opposite sides of the core ( when detected ) , giving rise to the classification of either compact symmetric objects ( csos , * ? ? ?
* ) , if they are smaller than 1 kpc , or medium - sized symmetric objects ( mso , * ? ? ?
* ) if they extend on scales up to 15 - 20 kpc .
however , a large fraction of css and gps sources have a very asymmetric two - sided morphology ( e.g. * ? ? ?
* ) where one of the lobes is much brighter and closer to the core than the other .
this kind of asymmetry can not be explained in terms of beaming effects and path delay , suggesting that the two jets are piercing their way through an inhomogeneous medium ( e.g. * ? ? ?
* ; * ? ? ?
a strong indication of asymmetries produced by a jet - cloud interaction comes from the detection of atomic hydrogen in absorption only against the brighter ( and closer to core ) lobe in the css objects 3c49 and 3c268.3 @xcite , and in the restarted source 3c236 @xcite .
moreover , evidence that the ionized gas may be inhomogeneously distributed around the radio source was provided by the detection of asymmetric free - free absorption against the two lobes in three of the most compact radio galaxies ( ls @xmath9 16 pc ) : j0428 + 3259 , j1511 + 0518 @xcite , and oq208 @xcite .
all these indications suggest that in a substantial fraction of small objects the radio source is not uniformly enshrouded by a homogeneous environment and the two jets may experience different conditions during their propagation through the interstellar medium .
although the gas is not dense enough to frustrate the jet expansion for the entire source lifetime ( e.g. * ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , it may slow down the source growth if a jet - cloud interaction takes place .
the high fraction of asymmetric intrinsically - compact radio galaxies strongly supports this scenario .
+ in this paper we present results of global - vlbi and vla observations at 5 ghz aimed at determining the radio morphology and the polarimetric properties of 10 out of the 16 sources of the css sample selected by @xcite from the peacock and wall catalogue @xcite . furthermore , the combination of these data with the 1.7 ghz images already published by @xcite allows the analysis of the spectral index distribution across the source , which is crucial in order to unambiguously constrain the nature of each source component .
+ the paper is organized as follows : section 2 describes the radio data and the data reduction ; in section 3 we report the results and a description of each radio source , while discussion and summary are presented in sections 4 and 5 , respectively .
+ throughout this paper , we assume @xmath10 km s@xmath3 mpc@xmath3 , @xmath11 , @xmath12 , in a flat universe . the spectral index is defined as @xmath13(@xmath14 ) @xmath15 .
global vlbi observations at 5 ghz ( 6 cm ) were carried out on september 16 and 17 , 1991 for a total of 30 hours , using the mark ii recording system @xcite with 1.8 mhz bandwidth in left circular polarization ( lcp ) , at the stations reported in table [ station ] .
the data were correlated with the jpl
caltech vlbi correlator at the california institute of technology in pasadena .
+ each source was observed in snap shot mode for about 1.01.5 hour of global time plus 1.52 hour of each subnetwork ( stations in europe and us were observing two different sources ) . for about 24 hours the vla ( in a configuration )
was used in phased
array mode as part of the network , greatly enhancing the sensitivity of the observations .
this allowed us to have also simultaneous conventional interferometric data ( see section 2.2 ) .
in addition , the merlin array was jointly observing , but various failures allowed us to use small parts of the vlbi data taken at cambridge and knockin only .
the correlated vlbi data were read into the astronomical image processing system ( ` aips ` ) package , developed at the national radio astronomy observatory ( nrao ) where fringe fitting , amplitude calibration , editing , self - calibration and imaging were performed .
the flux density scale was calibrated following @xcite , using b0016 + 731 , b0235 + 164 and b0851 + 202 ( oj287 ) as calibration sources , with adopted flux densities of 1.70 , 1.74 and 3.37 jy respectively , as derived from the vla simultaneous observation ( see sect .
+ our earlier 1.7-ghz images @xcite were used as reference in order to solve for phase ambiguities and to constrain initial self - calibration .
the standard techniques of editing , mapping and self calibration were applied .
gain self
calibration was applied only once at the end of the process , adopting a solution interval longer than the scan length ( 30 min ) , in order to remove residual systematic errors and to fine tune the flux density scale .
the gain corrections obtained from self calibration were in general within 2% .
the 1@xmath16 noise level on the images , measured far from the source , is in the range of 0.2 0.9 mjy / beam , often not far from the thermal noise ( @xmath00.17 mjy / beam for a typical observation ) .
the dynamic range , defined as the peak brightness over r.m.s noise level ( @xmath17/1@xmath16 ) , varies from @xmath0200 to @xmath05000 .
the resolution achieved is about 2 milliarcseconds
. + lccl antenna & diam.(m ) & t@xmath18 ( k ) & sens .
( k / jy ) + wsrt ( nl ) & 43 & & + jodrell bank ( uk ) & 25 & 38 & 0.12 + knockin ( uk ) & 25 & 38 & 0.1 + cambridge ( uk ) & 32 & 38 & 0.23 + onsala ( s ) & 26 & 50 & 0.06 + effelsberg ( d ) & 100 & 75 & 1.45 + noto ( i ) & 32 & 30 & 0.16 + haystack ( us ) & 37 & 75 & 0.165 + green bank ( us ) & 43 & 30 & 1.93 + ovro ( us ) & & & 0.21 + vla ( us ) & 25 ( 130 ) & 35 & 0.12 ( 2.3 ) + vlba@xmath19nl ( us ) & 25 & 40 & 0.12 + vlba@xmath19fd ( us ) & 25 & 40 & 0.12 + vlba@xmath19pt ( us ) & 25 & 40 & 0.12 + vlba@xmath19kp ( us ) & 25 & 40 & 0.12 + vlba@xmath19la ( us ) & 25 & 40 & 0.12 + [ station ] lcllccrccccrcc source & opt .
m & z & scale & @xmath20 & ls & @xmath21 & @xmath22 & @xmath23 & @xmath24 & @xmath25 & ref .
+ & & & & pc / mas & arcsec & kpc & jy & jy & % & & mhz & + ( 1)&(2)&(3)&(4)&(5)&(6)&(7)&(8)&(9)&(10)&(11)&(12)&(13 ) + 0223 + 341 & q & 21.3r & 2.910 & 7.898&0.56&4.42&1.75&1.63&@xmath90.6 & 0.38 & @xmath9100 & 1 + 0316 + 161 & g & 23.4v & 0.907 & 7.830&0.22&1.72&2.89&2.51&@xmath90.2 & 0.81 & 800 & 2 + 0319 + 121 & q & 18.0v & 2.662 & 8.079&0.04&0.32&1.37&1.33 & 5.5 & 0.17 & @xmath9100 & 3 + 0404 + 768 & g & 22.0r & 0.598 & 6.662&0.13&0.87&2.95&2.64&@xmath90.6 & 0.48 & 350 & 1 + 0428 + 205 & g & 18.0r & 0.219 & 3.507&0.10&0.36&2.38&2.00&@xmath90.2 & 0.38 & 1150 & 4 + 1225 + 368 & q & 21.4r & 1.973 & 8.486&0.06&0.51&0.76&0.73&@xmath90.5 & 0.88 & 1750 & 4 + 1358 + 624 & g & 19.8r & 0.431 & 5.590&0.05&0.28&1.77&1.67&@xmath90.5 & 0.70 & 550 & 4 + 1413 + 349 & ef & - & - & 8.041&0.03&0.24&0.92&0.87&@xmath90.4 & 0.53 & 850 & 4 + 1600 + 335 & g & 23.8r & - & 8.041&0.07 & 0.42&2.58&2.55&@xmath90.5 & 0.15 & 2400 & 5 + 2342 + 821 & q & 20.2r & 0.735 & 7.285&0.18&1.31&1.31&1.29&@xmath90.7 & 0.83 & 500 & 5 + [ vla_parameter ] during the vlbi observation , the vla in a - configuration was observing in phased array mode for 24 hours at 5 ghz , allowing us to obtain also images with a resolution of about 0.4 arcsecond for all the sources . in this way we got accurate flux density measurements of each source and calibrators and we could search for extended radio emission with angular size up to 10 arcsec . furthermore , we obtained vla polarization information for all the sources .
the calibration of the instrumental polarization could not be very accurate since no suitable d - terms calibration source was observed on a wide range of parallactic angles and we had to use three different sources covering together about 150@xmath26 .
the absolute flux density scale was set on 3c286 , observed for a few minutes , for which we assumed a flux density of 7.38 jy at 5.0 ghz .
in this section we present the total intensity vlbi images at 5 ghz for the 10 sources discussed in this paper and listed in table [ vla_parameter ] , together with spectral index images obtained by comparing these 5-ghz data with those at 1.7 ghz presented in @xcite . for each source
we provide a description of the main characteristics . + in general , for each source we obtained images with a set of different restoring beams and samplings in order either to match the resolution of our 1.7-ghz images , or to get all the details of the structures .
+ the @xmath27-coverages at 1.7 and 5 ghz for the source 1413 + 349 are shown in fig .
[ figuvcov ] as an example : the 5-ghz visibilities are restricted to the region covered by the 1.7-ghz data .
sources with higher declinations have a better match between the 1.7-ghz and 5-ghz @xmath27-coverages .
+ high resolution images at 5 ghz of each source are presented in figs .
[ 0223 ] [ 2342 ] . for the source 0316 + 161 we present also the high resolution image at 1.7 ghz with the identification of the source core , which was not provided in the paper by @xcite . in the case of 0223 + 341 , where one lobe is located at about 0.5 arcsec from the main component , and 1600 + 335 , where the central component is surrounded by an extended low surface brightness feature ,
a lower resolution image is presented next to the high - resolution one .
+ source parameters ( total flux density and deconvolved size ) have been generally measured by means of the ` aips ` task jmfit for marginally resolved components . when a component could not be properly fitted with a gaussian profile ,
tvstat was used to measure the flux density , while the angular size was measured from the lowest contours and it corresponds to 1.8 times the size of the full width at half maximum ( fwhm ) of a conventional gaussian covering a similar area @xcite .
+ sub - components are referred to as north ( n ) , south ( s ) , east ( e ) , west ( w ) , central ( ce ) , jet ( j ) , and flat spectrum core ( c ) when detected , following the labelling used in @xcite .
the parameters of the sub - components are given in table [ component ] .
+ to produce the spectral index distribution across each source , we created low - resolution images with the same restoring beam , _
uv_-range and image sampling used for the 1.7-ghz image .
images at 1.7 and 5 ghz were obtained using all the baselines shorter than 50 m@xmath28 .
these images were then convolved with a larger beam ( usually between 6 and 10 mas ) , in order to highlight features corresponding to baseline - length shorter than 20 m@xmath28 , where the mismatch of the _ uv_-coverages is less significant .
we must note that the addition of the vla array in the 5 ghz observations reduces the gap at the short spacings , providing a good set of short baselines that allows a good sensitivity to extended , low - surface brightness structures .
+ the spectral index images were produced with the ` aips ` task comb .
blanking was done considering only the pixels of the input images with values above five times the rms measured on the off - source image plane .
image registration was performed using the task lgeom in ` aips ` , by comparing the location of optically - thin , compact bright features .
this led to satisfactory results in case of compact regions , while artificial gradients may be present in the extended structures , precluding an accurate analysis of these regions .
furthermore , we do not provide spectral index images of the more extended regions , like the relatively large and low surface brightness lobes in 0223 + 341 , 0316 + 162 , and 0428 + 205 , given that most of the information in the 1.7-ghz images relies on the merlin baselines @xcite , which are not sampled by the 5-ghz vlbi observations .
[ spix ] , for all the sources we show the greyscale spectral index images between 1.7 and 5 ghz superimposed on the 5-ghz contours obtained from the low resolution images .
+ no vla a - configuration images are presented since the resolution of about 0.4 arcsec is not enough to resolve the targets , and only the largest object in our sample , 0319 + 121 , is marginally resolved .
+ here we provide a description of the radio morphology of the observed sources .
in addition to the _ local _ spectral index derived on the spectral index images ( fig . [ spix ] ) , we computed the _ total _
spectral index integrated on the whole component by means of the total flux density measured on the high - resolution images at 5 and 1.7 ghz .
the formal error on both the _ total _ and _ local _ spectral index is about 0.1 , and it has been calculated assuming the error propagation theory . however , we note that the use of the total flux density in computing the spectral index may cause an artificial steepening due to the absence of the shortest spacing at the higher frequency , implying that the formal errors associated to the extended components may be a lower limit
. the _ total _
spectral index values are reported in table [ component ] .
+ .size and flux density of the components in the vlbi images .
columns 1 and 2 : source name and component label ; cols . 3 , 4 , and 5 : deconvolved major and minor axes , and the position angle of the major axis derived from the fit to the image .
the values reported are the fwhm of the fitted gaussian component ; col . 6 : vlbi flux density at 5 ghz ; col . 7 : spectral index computed from the full - resolution images at 1.7 and 5 ghz ; col . 8 : the equipartition magnetic fields ( see section 4.3 ) .
for consistency with @xcite , when the redshift is not available we assume @xmath29 in the determination of the magnetic field and the value is reported in italics .
[ cols= " < , < , > , > , > , > , > , > " , ] [ fit ] the slope obtained for css / gps sources is a little steeper than that for fri / frii , although within the uncertainties , suggesting a lower contribution from the extended structures .
another possibility is that the css / gps sources have a slightly larger core emission .
this may be due either to an intrinsically brighter core , or to the additional contribution from the jet base that can not be disentangled from the core emission .
the slope obtained for fri / frii sources is in agreement with what was found by @xcite for a different sample of fri / frii objects . in the case of the blazar population from the lbas ,
the slope is close to 1 , indicating that the contribution from extended structures is negligible and the core dominates the source emission due to doppler beaming effects , as it was already pointed out by @xcite .
+ on the basis of the observed parameters , and assuming minimum energy conditions ( particle energy is equal to the magnetic energy ) , we derive the physical properties for each component of the radio sources .
we computed the equipartition magnetic field @xmath30 , the minimum energy density @xmath31 , and the minimum internal pressure @xmath32 using standard formulae @xcite .
we assumed that the source components have an ellipsoidal volume with a filling factor of unity , and an equal distribution between proton and electron energies .
an average optically - thin spectral index of 0.7 has been adopted .
we found that the minimum energy density @xmath33 is between 10@xmath34 and 10@xmath35 erg @xmath36 , the minimum internal pressure @xmath37 ranges between 10@xmath34 and 10@xmath38 dyne @xmath39 , while the equipartition magnetic field accounts for a few mg up to about 120 mg ( table [ component ] ) .
the latter value has been found in the compact core of the core - jet radio source 0319 + 121 , which is also at very high redshift .
the higher values are found in compact components and are similar to those found in the hotspots and core region of other css sources @xcite , whereas in extended components , like lobes and jets , we found lower values .
+ several studies of the polarization properties of css sources ( e.g. * ? ? ?
* ; * ? ? ?
* ) have shown that sources smaller than a few kpc have very low values of polarized emission at frequencies below 8.4 ghz .
this result has been interpreted by assuming that compact sources are highly depolarized by the dense interstellar medium that enshrouds the radio emission and acts as a faraday screen .
support to this idea is the detection of some level of polarized emission as we consider higher frequencies , while the sources are completely unpolarized at the lower frequencies .
+ from our vla observations , we found that 9 out of the 10 sources studied here are unpolarized at 5 ghz , in good agreement with the expectation .
the radio source 1358 + 624 , unpolarized in our 5 ghz observations , turned out to be polarized at 15 ghz @xcite , suggesting that strong depolarization is taking place in this source .
+ only the quasar 0319 + 121 turned out to be significantly polarized , with a fractional polarization of about 5.5% .
since polarimetric observations are available at one frequency only , we can not determine the rotation measure ( rm ) .
however , if we assume a low rm , we obtain that the magnetic field is parallel to the vlbi jet axis , as it is observed in quasars .
the presence of significant percentage of polarized emission , together with the core - jet structure , suggests that this source may be oriented at a small angle with respect to the line of sight and its linear size may be foreshortened by projection effects .
we presented the results from global - vlbi observations at 5 ghz of a sample of 10 out of 16 css sources from the sample presented by @xcite .
the majority of the targets have a two - sided structure and the radio emission is dominated by lobes , jets , and hotspots .
one sources , 0319 + 121 , has a core - jet morphology and the radio emission comes mainly from the core component .
this object is the only one with significant polarized emission .
high polarization level is uncommon in css sources , where the interstellar medium enshrouding the radio source acts as a faraday screen causing severe depolarization of the radio emission .
+ the determination of the nature of the source components has been provided by the availability of two - frequency observations which allowed the study of the spectral index distribution across the whole source .
the core has been unambiguously identified in 7 sources . among them , the six sources with a two - sided structure have large flux density and/or arm - length asymmetries . in three cases
the brightest lobe is the farthest from the core , in agreement with what is expected in case of projection effects .
the evidence that the ambient medium may play a decisive role in producing asymmetric objects is supported by the galaxies 0316 + 161 , 0404 + 768 and 0428 + 205 , where the brightest lobe is also the closest to the core .
the interaction with a dense medium may slow down the jet growth , lowering adiabatic losses and enhancing radiative losses . as a consequence
the synchrotron emission from the interacting jet is enhanced .
if the jet - environment interaction is not a sporadic phenomenon in the lifetime of css radio sources , it may provide a possible explanation of the high fraction of css objects in flux density limited sample .
+ the influence of the ambient medium on the source expansion has been proved on a few individual objects only and no statistical study has been performed so far .
multi - frequency vlba observations of a sample of the most compact css / gps objects ( with linear size of a few parsecs ) have been already performed in order to investigate the role played by environment during the very first stage of the source growth .
the results will be used to draw a complete and reliable picture of the radio source evolution .
we thank the anonymous referee for reading the manuscript carefully and making valuable suggestions .
we thank the european vlbi network and the united states vlbi network for carrying out the observations .
the wsrt is operated by the netherlands foundation for research in astronomy with the financial support of the netherlands organization for scientific research ( nwo ) .
the national radio astronomy observatory is operated by associated universities , inc .
, under contract with the national science foundation .
this research has made use of the nasa / ipac extragalactic database ( ned ) , which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
dd acknowledges the commission of the european union for the award of a fellowship .
funding for sdss - iii has been provided by the alfred p. sloan foundation , the participating institutions , the national science foundation , and the u.s .
department of energy office of science .
the sdss - iii web site is http://www.sdss3.org/. sdss - iii is managed by the astrophysical research consortium for the participating institutions of the sdss - iii collaboration including the university of arizona , the brazilian participation group , brookhaven national laboratory , university of cambridge , carnegie mellon university , university of florida , the french participation group , the german participation group , harvard university , the instituto de astrofisica de canarias , the michigan state / notre dame / jina participation group , johns hopkins university , lawrence berkeley national laboratory , max planck institute for astrophysics , max planck institute for extraterrestrial physics , new mexico state university , new york university , ohio state university , pennsylvania state university , university of portsmouth , princeton university , the spanish participation group , university of tokyo , university of utah , vanderbilt university , university of virginia , university of washington , and yale university . | global vlbi observations at 5 ghz have been performed to study the source morphology in 10 compact steep - spectrum ( css ) sources selected from the peacock & wall catalogue with the aim of finding asymmetric structures produced by the interaction with the ambient medium .
the combination of these data and earlier 1.7-ghz observations allows the study of the spectral index distribution across the source structure and the unambiguous determination of the nature of each component . in seven sources we detected the core component with a flat or inverted spectrum . in six sources
the radio emission has a two - sided morphology and comes mainly from steep - spectrum extended structures , like lobes , jets , and hotspots .
only one source , 0319 + 121 , has a one - sided core - jet structure . in three out of the six sources with a two - sided structure the flux density arising from the lobes
is asymmetric , and the brightest lobe is the one closest to the core , suggesting that the jets are expanding in an inhomogeneous ambient medium which may influence the source growth . the interaction between the jet and the environment may slow down the source expansion and enhance the luminosity due to severe radiative losses , likely producing an excess of css radio sources in flux density limited samples .
the lobes of the other three asymmetric sources have a brighter - when - farther behaviour , in agreement with what is expected by projection and relativistic effects .
simultaneous vla observations carried out to investigate the polarization properties of the targets detected significant polarized emission ( @xmath05.5% ) only from the quasar 0319 + 121 . [ firstpage ] galaxies : active galaxies : jets galaxies : nuclei quasars : general radio continuum : galaxies radio continuum : general |
let @xmath0 be an elliptic curve defined over the field of rationals @xmath1 with discriminant @xmath15 . for a prime @xmath4 where @xmath0 has good reduction , i.e. @xmath16 , we denote by @xmath17 the reduction of @xmath0 modulo @xmath4 .
let @xmath18 be the finite field with @xmath4 elements and @xmath19 be the group of @xmath18 points over @xmath17 . for @xmath16 , we know @xmath20 where @xmath21 is the trace of the frobenius morphism at @xmath4 .
by hasse s theorem we know @xmath22 . for a fixed positive integer @xmath2 ,
we define the following prime counting function @xmath23 now , for a pair of integers @xmath24 , let @xmath25 be the elliptic curve defined by the weierstrass equation @xmath26 also , for @xmath27 , we define the family of curves @xmath28 by @xmath29 now , let us recall barban - davenport - halberstam conjecture for primes in arithmetic progression in short interval .
( barban - davenport - halberstam)[conj_1 ] let @xmath30 .
let @xmath31 and @xmath32 be real numbers .
suppose that @xmath33 , @xmath34 , and @xmath35 are positive real numbers satisfying @xmath36 and @xmath37 .
then @xmath38 under the above conjecture , david and smith proved that [ thmm_1 ] let conjecture [ conj_1 ] be true for some @xmath39 . if @xmath40 and that @xmath41 , then for any odd integer @xmath2 , we have @xmath42 where @xmath43 denotes the usual @xmath4-adic valuation where @xmath44 and @xmath45 is the kronecker symbol . also , by taking another average over @xmath46 , a similar result was unconditionally proven by chandee , david , koukoulopoulos and smith @xcite .
improving a result of martin , pollack and smith @xcite , in a work with balasubramanian @xcite , we showed that the function @xmath47 is @xmath48 on an average and the average approaches @xmath48 reasonably fast .
using an approach first used by banks and shparlinski @xcite , balog , cojocaru and david @xcite , akbary and felix @xcite , in @xcite parks proved that the average result in _ theorem [ thmm_1 ] _ is true even if we significantly relax the lower bound conditions on @xmath6 and @xmath7 . to be precise , he proved [ par ] let @xmath49 and assume for intervals of length @xmath50 that the barban- davenport - halberstam conjecture holds for @xmath51 where @xmath52 is given in ( [ eq_1.5 ] ) . in an earlier work with balasubramanian @xcite , we proved results related to distribution of the function @xmath3 .
more precisely , we proved that [ th_1 ] let @xmath28 be as defined as in ( [ eq_1.6 ] ) and @xmath2 be a positive integer greater than @xmath53 . if @xmath54 be a positive integer such that @xmath55 and @xmath56 for some @xmath11 , then for @xmath57 @xmath58 where the @xmath59constant in the last error term is independent of @xmath60 .
using an approach similar to parks @xcite , in this paper , we improve _ theorem [ th_1 ] _ as follows : [ th_01 ] let @xmath61 be a small positive number and @xmath10 be a positive integer .
suppose @xmath62 and @xmath63 and @xmath64 , then @xmath65 where the @xmath59constant in the last error term is independent of @xmath60 . alternatively , under conjecture [ conj_1 ] , we can state the above theorem in the following form : [ th_03 ] suppose conjecture [ conj_1 ] be true for some @xmath66 .
let @xmath67 be a non negative integer and @xmath68 .
also let @xmath69 and @xmath70 for a odd positive integer @xmath2 with @xmath71 .
then , for @xmath72 @xmath73 where @xmath28 is as before and @xmath74 * remark : * although , in @xcite , theorem _ [ par ] _ is claimed to hold for @xmath75 , the correct upper bound for @xmath6 and @xmath7 should be of the order @xmath76 . the crucial difference between proof of theorem [ th_01 ] and theorem [ th_1 ] is proposition [ lm_0.1 ] , which is stated in section 3 .
in this proposition , we have better estimate of the number of curves of the form @xmath77 with @xmath78 , which simultaneously reduces modulo a given set of distinct primes @xmath79 to fixed set of curves of the form @xmath80 , @xmath81 for @xmath82 in our previous paper with balasubramanian @xcite , we estimated number of curves satisfying above conditions using a technique essentially due to fouvry and murty @xcite , which involves partitioning a rectangle of size @xmath83 into boxes of size @xmath84 and using chinese reminder theorem to merge congruence condition over different primes together . while in proposition [ lm_0.1 ] we use estimates of sums of suitable multiplicative characters .
* acknowledgements : * i would like to thank chantal david , dimitris koukoulopoulos and amir akbary for their useful advices .
i also thank james parks for spending time in clarifying some of my doubts that were essential for completing the work . *
funding : * this work was done while working as a postdoctoral fellow at centre de recherches mathematiques , universit de montreal .
let @xmath85 be a negative discriminant . using the class number formula [ p. 515 ,
@xcite ] , the _ kronecker class number _ for a discriminant @xmath85 can be written as @xmath86 where @xmath87 is the kronecker symbol @xmath88 and @xmath89 . using deuring s theorem @xcite we get @xmath90 where the sum is over the @xmath91-isomorphism classes of elliptic curves .
define , @xmath92 with these notations defined , we recall the following lemma from [ lemma 2.1 , @xcite ] [ lm_1.1 ] let @xmath2 be a positive integers and @xmath93 and @xmath94 are defined as above . also let @xmath95 be defined by ( [ eq_2.1 ] ) and ( [ eq_2.2 ] ) . then a. @xmath96 b. for @xmath97 , @xmath98 we also need the following two theorems : [ largesieve ] let @xmath99 be positive integers and let @xmath100 is a sequence of complex numbers . for a fixed @xmath101 , we let @xmath102 be a dirichlet character modulo @xmath103 . then @xmath104 for the proof of the above theorem , see [ chapter 27 , @xcite ] . + the second theorem is due to friedlander and iwaniec @xcite , which bounds the fourth power moment of dirichlet characters .
( * friedlander - iwaniec * ) [ fourthpower ] let @xmath103 and @xmath2 be positive integers .
let @xmath102 denote a dirichlet character modulo @xmath103 , with @xmath105 denoting the principal character
. then @xmath106
let @xmath107 be a positive integer .
we have , @xmath108 for any non - negative integer @xmath67 , breaking the sum into two parts , the right hand side can be written as @xmath109 for @xmath110 , consider the expression @xmath111 now , for a curve @xmath0 with @xmath112 , the curve @xmath0 is counted @xmath113 times in ( [ b_1 ] ) . while , the same @xmath0 is counted @xmath114 times if we consider the expression @xmath115 using stirling s approximation , is easy to see that @xmath116 for @xmath110 .
thus @xmath117 for @xmath118 , using a similar argument , one can also show that @xmath119 also , for @xmath120 , @xmath121 we now consider the first term of ( [ eq_24 ] ) .
note that , the primes in the range of summations in ( [ eq_24 ] ) are not distinct . recalling the definition of @xmath122 , the stirling number of the second kind , which equals to the number of ways of partitioning a set of @xmath123 elements into @xmath124 non empty sets , we get @xmath125 to simplify the first factor on the right hand side , we use the equality @xmath126 . see [ ( 4.1.3 ) , p. 60 , @xcite ] . with this
, @xmath127 now we denote the left hand side of ( [ eq_27 ] ) by @xmath128 and the first term of the right hand side of ( [ eq_19 ] ) by @xmath129 .
also we call the left hand side of ( [ eq_2.7 ] ) by @xmath130 .
then , in view of ( [ eq_27 ] ) and ( [ eq_2.7 ] ) , we get the following set of relations @xmath131 now , we state the following proposition , whose proof will be completed in section 4 .
[ lm_0.1 ] let @xmath28 be as above .
let @xmath61 be a small positive number .
suppose @xmath2 be a positive integer such that @xmath62 with @xmath132 and @xmath133 , then @xmath134 now , by proposition [ lm_0.1 ] @xmath135 whenever @xmath132 and @xmath136 .
now , we replace @xmath137 by @xmath138 where @xmath139 are some constants to be determined later using ( [ eq_4 ] ) .
also note that @xmath140 $ ] .
then , in view of ( [ b_2 ] ) and proposition [ lm_0.1 ] , the expression in ( [ eq_24 ] ) equals to @xmath141 for some real numbers @xmath142 + only thing that remains to be shown is that @xmath143 are equals to @xmath144 , as defined in ( [ eq_10 ] ) . for that
, we have the following lemma . [ lm_2 ] consider @xmath145 as variables satisfying the identities in ( [ eq_4 ] ) .
then , the solution of the equation @xmath146 in variables @xmath147 are given by @xmath148 see [ lemma 3.2 , @xcite ] for the proof of the above lemma . finally , combining ( [ b_1 ] ) , ( [ eq_009 ] ) and lemma [ lm_2 ] , we have @xmath149 for @xmath150 and @xmath151 .
putting @xmath152 , @xmath153 and @xmath154 , @xmath155 , from ( [ eq_11 ] ) we get , @xmath156 for @xmath157 and @xmath158 + also , for @xmath154 , @xmath155 , from ( [ eq_11 ] ) we have @xmath159 for @xmath132 and @xmath13 .
+ we use ( [ eq_7 ] ) and ( [ eq_12 ] ) to replace @xmath160 in the right hand side of ( [ eq_12 ] ) by @xmath161 .
now , using lemma [ lm_1.1]a , we get @xmath162 for @xmath132 and @xmath13 .
further , we recall that @xmath163 .
this proves theorem [ th_01 ] . assuming conjecture [ conj_1 ] and the proof of [ theorem 3 , @xcite ] , we have that @xmath164 for odd integer @xmath2 , where @xmath52 is given by ( [ eq_1.5 ] ) . combining ( [ eq_13 ] ) with ( [ eq_11 ] ) , we complete the proof of theorem [ th_03 ] .
before proceeding with the proof of the proposition , we define some standard notations
. + let @xmath165 be a vector of @xmath10 distinct primes such that @xmath166 for @xmath167 .
so , the primes in question are effectively of the order @xmath2 . throughout the rest of the proof
, @xmath175 denotes a curve over a finite field @xmath18 .
also @xmath25 denotes curve over @xmath176 as defined in ( [ eq_1.6 ] ) . +
further , we know , two elliptic curves @xmath177 and @xmath178 are isomorphic over @xmath179 if and only if there exists a @xmath180 such that @xmath181 and @xmath182 .
hence , the number of elliptic curves over @xmath179 isomorphic to @xmath177 is @xmath183 further , @xmath184 we plan to count the number of curves @xmath190 whose reductions modulo @xmath191 are @xmath192 for all @xmath191 .
then , the inner summation on left hand side of ( [ eq_010 ] ) can be written as @xmath193 where @xmath194 denotes the number of integers @xmath195 such that @xmath196 @xmath197 such that @xmath198 now , @xmath199 most of the times and in particular when @xmath200 .
so , we write ( [ aliquotaut ] ) as @xmath201 define @xmath202 we plan to complete the estimation of @xmath203 first .
later , we show that the same estimation technique can be modified suitably to give required upper bound to @xmath204 .
+ for this part of the proof related to the estimation of @xmath203 , we are essentially going to follow the approach of parks @xcite , except possibly the different range of summation over primes .
we give a proof of the above lemma later in this section .
+ now , using ( [ eq_001 ] ) , we write the inner summation in the first sum in @xmath210 as @xmath211 using the bound @xmath212 , one can show that @xmath213 for @xmath167 .
this together with ( [ alithree ] ) gives before we proceed with estimating @xmath204 as defined in ( [ eq_003 ] ) , we give a proof of lemma [ errorimprove ] .
later we are going to use the same proof and the discussions above to give a bound on @xmath204 . @xmath231\prod_{i=1}^\ell\left(\frac{\chi_i(s_i)\chi_i'(t_i)}{p_i-1}\right){\mathcal{a}}(\overline{\chi_1\cdots \chi_\ell } ) { \mathcal{b}}(\overline{\chi_1'\cdots \chi_\ell'})\nonumber\\ = & z_1(p , s , t)+z_2(p , s , t)+z_3(p , s , t)+z_4(p , s , t)\label{thefourcs}\end{aligned}\ ] ] now , for a fixed prime @xmath242 @xmath79 , the second product in ( [ holdboundone ] ) , let @xmath243 be the set of positive integers such that @xmath244 for @xmath245 .
thus , @xmath246 let @xmath247 denote the number of representation of @xmath248 as a product of @xmath209 positive @xmath7 smooth integers .
then , @xmath249 thus , @xmath250 now , @xmath251 is a primitive character modulo @xmath252 .
now we extend the sum in @xmath253 to a sum over all primitive characters modulo @xmath254 for all modulus @xmath255 . using theorem [ largesieve ]
, we get @xmath256 next , we plan to modify the previous proof of lemma [ errorimprove ] to give a upper bound on @xmath204 .
+ recall , @xmath278 _ case 1 : @xmath279 for all @xmath280 . _
+ then the corresponding rational curves look like @xmath25 where @xmath281 or @xmath282 . + in that case ,
the contribution corresponding to @xmath283 is bounded by @xmath284 if , either @xmath285 or @xmath286 , then the curve has complex multiplication .
hence , by kowalski @xcite , there are only @xmath287 many primes such that @xmath288 .
so , the contribution corresponding to @xmath289 is bounded by @xmath290 _ case 2 : @xmath291 for some @xmath292 and @xmath293 for some @xmath294 . _ + the number of possible subsets @xmath295 of @xmath296 such that @xmath279 for all @xmath297 is bounded by @xmath298 .
take one such subset @xmath295 and without loss of generality , assume that @xmath299 with @xmath300 in that case , the contribution corresponding to the set @xmath295 in ( [ e1 ] ) is bounded by @xmath301 where @xmath302 again , we write @xmath311 as @xmath312 where @xmath313\prod_{i = e+f+1}^\ell\left(\frac{\chi_i(s_i)\chi_i'(t_i)}{p_i-1}\right)\hat{{\mathcal{a}}}(\overline{\chi_{e+f+1}\cdots \chi_\ell } ) { \hat{\mathcal{b}}}(\overline{\chi_{e+f+1}'\cdots \chi_\ell'})\\ \hat{z}_2(\hat{p},\hat{s},\hat{t})&=\frac{1}{2^\ell}\sum_{\substack{\chi_i=(\chi_i')^6=\chi_0 \imod{p_i}\\ { \rm for}\ : 1\leq i\leq \ell\:{\rm and } \\ \exists
1\leq j \leq \ell \:{\rm s.t.}\ : \chi_j'\neq \chi_0 \imod{p_j}}}\prod_{i = e+f+1}^\ell\left(\frac{\chi_i(s_i)\chi_i'(t_i)}{p_i-1}\right)\hat{{\mathcal{a}}}(\overline{\chi_{e+f+1}\cdots \chi_\ell } ) { \hat{\mathcal{b}}}(\overline{\chi_{e+f+1}'\cdots \chi_\ell'})\\ \hat{z}_3(\hat{p},\hat{s},\hat{t})&=\frac{1}{2^\ell}\sum_{\substack{\chi_i'=\chi_i^4=\chi_0 \imod{p_i}\\ { \rm for}\ : 1\leq i\leq \ell\:{\rm and } \\
\exists 1\leq j \leq \ell \:{\rm s.t.}\ : \chi_j\neq \chi_0 \imod{p_j}}}\prod_{i = e+f+1}^\ell\left(\frac{\chi_i(s_i)\chi_i'(t_i)}{p_i-1}\right)\hat{{\mathcal{a}}}(\overline{\chi_{e+f+1}\cdots \chi_\ell } ) { \hat{\mathcal{b}}}(\overline{\chi_{e+f+1}'\cdots \chi_\ell'})\\ \hat{z}_4(\hat{p},\hat{s},\hat{t})&=\frac{1}{2^\ell}\sum_{\substack{\chi_i^4(\chi_i')^6=\chi_0 \imod{p_i}\\ { \rm for}\ : 1\leq i\leq \ell\:{\rm and } \\
\exists 1\leq r , s \leq
\ell \:{\rm s.t.}\ : \chi_r\neq \chi_0 \imod{p_r},\\\chi_s'\neq \chi_0 \imod{p_s } } } \prod_{i = e+f+1}^\ell\left(\frac{\chi_i(s_i)\chi_i'(t_i)}{p_i-1}\right)\hat{{\mathcal{a}}}(\overline{\chi_{e+f+1}\cdots \chi_\ell } ) { \hat{\mathcal{b}}}(\overline{\chi_{e+f+1}'\cdots \chi_\ell'})\end{aligned}\ ] ] since , the primes @xmath191 s are distinct , we also have @xmath317 replacing @xmath6 , @xmath7 and @xmath10 by @xmath318 , @xmath319 and @xmath320 respectively in the proof of ( [ jtwosevenbound ] ) , we get the following inequality @xmath321 for some @xmath262 . since @xmath323 , observe that we get a savings of a factor of @xmath324 in @xmath325 , ( [ eq_015 ] ) and ( [ eq_016 ] ) compared to the upper bounds for corresponding expressions in the proof of lemma [ errorimprove ] .
also , in view of ( [ eq_017 ] ) , we need to assume @xmath326 to make the corresponding error term sufficiently small in proposition [ lm_0.1 ] .
a. akbary and a. t. felix , on invariants of elliptic curves on average .
acta arith .
* 168 * ( 2015 ) , no . 1 , 31 - 70 .
r. balasubramanian and s. giri , the mean - value of a product of shifted multiplicative functions and the average number of points of elliptic curves .
j. number theory * 157 * ( 2015 ) , 37 - 53 .
r. balasubramanian and s. giri , poisson distribution of a prime counting function corresponding to elliptic curves , to appear in internat .
notices , arxiv:1503.01018 [ math.nt ] .
a. balog , a .- c . cojocaru and c. david , average twin prime conjecture for elliptic curves , amer . j. math . * 133 * ( 2011 ) , no . 5 , 1179 - 1229 .
w. d. banks and i. e. shparlinski , sato - tate , cyclicity , and divisibility statistics on average for elliptic curves of small height .
israel j. math . 173
( 2009 ) , 253 - 277 .
v. chandee , c. david , d. koukoulopoulos and e. smith , the frequency of elliptic curve groups over prime finite fields .
* 68 * ( 2016 ) , no . 4 , 721 - 761 . h. davenport , multiplicative number theory .
third edition . revised and with a preface by hugh l. montgomery .
graduate texts in mathematics , 74 .
springer - verlag , new york , 2000 .
c. david and e. smith , elliptic curves with a given number of points over finite fields , _ compositio math .
_ * 149 * ( 2013 ) , 175203 . c. david and e. smith , corrigendum
to `` elliptic curves with a given number of points over finite fields '' , _ compositio math . _
* 150 * ( 2014 ) , no . 8 ,
m. deuring , die typen der multiplikatorenringe elliptischer funktionenkorpr .
humbury , * 14 * ( 1941 ) , no . 1 , 197 - 272 .
e. fouvry and m. r. murty , on the distribution of supersingular primes , canad . j. math .
* 48 * ( 1996 ) , 81104 .
j. friedlander and h. iwaniec , the divisor problem for arithmetic progressions._acta arith .
_ 45 ( 1985 ) , 273 - 277
. h. iwaniec and e. kowalski , analytic number theory , colloquium publications , vol .
53 , american mathematical society .
e. kowalski , analytic problems for elliptic curves , j. ramanujan math .
* 21 * ( 2006 ) , 19114 .
g. martin , p. pollack and e. smith , averages of the number of points on elliptic curves .
algebra number theory * 8 * ( 2014 ) , no .
4 , 813836 .
j. parks , amicable pairs and aliquot cycles on average .
j. number theory * 11 * ( 2015 ) , no .
6 , 1751 - 1790
. j. parks , a remark on elliptic curves with a given number of points over finite fields . scholar a scientific celebration highlighting open lines of arithmetic research , 165 - 179 , contemp .
, * 655 * , amer . math .
soc . , providence , ri , 2015 .
s. roman , the umbral calculus .
new york : academic press , pp .
59 - 63 , 1984 . | let @xmath0 be an elliptic curve defined over @xmath1 and let @xmath2 be a positive integer .
now , @xmath3 counts the number of primes @xmath4 such that the group @xmath5 is of order @xmath2 . in an earlier joint work with balasubramanian
, we showed that @xmath3 follows poisson distribution when an average is taken over a family of elliptic curve with parameters @xmath6 and @xmath7 where @xmath8 and @xmath9 for a fixed integer @xmath10 and any @xmath11 . in this paper we show that for sufficiently large @xmath2 , the same result holds even if we take @xmath6 and @xmath7 in the range @xmath12 and @xmath13 for any @xmath14 .
# 17mu(@font mod#1 ) |
the fractional brownian motion with hurst index @xmath0 is a centered gaussian process @xmath1 with covariance function @xmath2=\tfrac{1}{2}(|s|^{2h}+|t|^{2h}-|t - s|^{2h}).\ ] ] it follows from ( [ def : fbm ] ) that @xmath3 is self - similar with index @xmath4 and has stationary increments .
unless @xmath5 ( i.e. , @xmath6 is brownian motion ) , @xmath3 is not markovian .
moreover , it is known that @xmath6 has long - range dependence if @xmath7 and short - range dependence if @xmath8 ( see samorodnitsky and taqqu @xcite ) .
these properties have made @xmath3 not only important theoretically , but also very popular as stochastic models in many areas including telecommunications , biology , hydrology and finance .
weak convergence to fractional brownian motion has been studied extensively since the works of davydov @xcite and taqqu @xcite . in recent years
many new results on approximations of fractional brownian motion have been established .
for example , enriquez @xcite showed that fractional brownian motion can be approximated in law by appropriately normalized correlated random walks .
meyer , sellan and taqqu @xcite proved that the law of @xmath6 can be approximated by those of a random wavelet series . by extending stroock @xcite , bardina _
et al . _
@xcite and delgado and jolis @xcite have established approximations in law to fractional brownian motions by processes constructed using poisson processes .
let @xmath9 be a standard poisson process , and for all @xmath10 , define the processes @xmath11 by @xmath12 stroock @xcite proved that as @xmath13 tends to zero , the laws of @xmath14 converge weakly in the banach space @xmath15 $ ] ( i.e. , the space of continuous functions on @xmath16 $ ] ) to the law of brownian motion .
delgado and jolis @xcite proved that every gaussian process of the form @xmath17 where @xmath18 is a one - dimensional brownian motion and @xmath19 a sufficiently regular deterministic kernel , can be weakly approximated by the family of processes @xmath20 their result can be applied to fractional brownian motion .
in addition , bardina and jolis @xcite proved that as @xmath13 tends to @xmath21 , the family of two - parameter random fields @xmath22 defined by @xmath23 converges in law in the space of continuous functions on @xmath24^{2}$ ] to the standard brownian sheet .
bardina , jolis and tudor showed in @xcite that as @xmath13 tends to @xmath21 , the family of two - parameter random fields @xmath25 converges in law to the two - parameter gaussian process @xmath26 where @xmath18 is a standard brownian sheet and the deterministic kernel @xmath27 can be separated by the integration variables and satisfies certain conditions . as examples , the authors include the fractional brownian sheet , among others . for more information ,
see bardina , jolis and rovira @xcite , where an approximation to the @xmath28-parameter wiener process by a @xmath28-parameter poisson process was provided , and bardina and bascompte @xcite , where two independent gaussian processes were constructed using a unique poisson process .
we note that in the serial works @xcite , the dimension of the parameter set is always the same for the approximating and the approximated processes .
naturally , we will be interested in the problem of whether we can approximate the @xmath28-parameter fractional brownian motions by @xmath29-parameter poisson processes if @xmath30 . the purpose of this paper is to study this problem in the case of @xmath31 and @xmath32 .
we find that for a special deterministic kernel function which can not be separated with respect to the integration variables , the answer is affirmative .
below , we introduce the deterministic kernel function . in order to study a non - gaussian and non - stable process arising as the limit of sums of rescaled renewal processes under the condition of intermediate growth , gaigalas @xcite , page 451 , introduced the function @xmath33 , defined as follows . for @xmath34 and @xmath35 @xmath36}(u - y)\,\mathrm{d}u= \cases
{ t , & \quad$-y < x , y<-t$,\cr x+t+y,&\quad$-t - y < x\leq - y , y<-t$,\cr -y , & \quad$-y < x , -t\leq y<0,$\cr x , & \quad$0<x\leq - y , -t\leq y<0,$\cr 0 , & \quad otherwise . } \ ] ] note that kaj and taqqu @xcite , page 388 , interpreted the function @xmath37 in the context of the infinite source poisson model as a function of the starting time @xmath38 and the duration @xmath39 of a session that measures the length of the time interval contained in @xmath40 $ ] during which the session is active . therefore , @xmath41 measures the length of the time interval contained in @xmath40 $ ] during which the session with duration @xmath39 and finishing time @xmath42 is active . obviously , by this definition , @xmath43 if and only if @xmath44 and @xmath45 , that is , @xmath46 . in this paper
, we define @xmath47}(u - y)\,\mathrm{d}u-\int_{-s}^0\mathbf{1}_{[0 , x]}(u - y)\,\mathrm{d}u \\ & = & \int_{-t - s}^{-s}\mathbf{1}_{[0 , x]}(u - y)\,\mathrm{d}u=\int_{-t}^0\mathbf{1}_{[0 , x]}(u - y - s)\,\mathrm{d}u = h(t , x , y+s)\nonumber\end{aligned}\ ] ] for all @xmath48 , @xmath49 , @xmath50 and any given @xmath51 .
then , according to the definition of an integral of random measure ( see @xcite , chapter 3 ) , we can directly verify that for @xmath52 , @xmath53 where @xmath54 is a gaussian random measure on @xmath55 with control measure @xmath56 ( see section [ 2 ] for its definition ) , @xmath57 and the notation @xmath58 denotes identification of finite - dimensional distributions .
in fact , the last equality is taken from gaigalas @xcite , page 454 , although the constant @xmath59 is omitted in gaigalas representation . therefore , @xmath60=\frac{t^{2h}}{2}.\ ] ] from the representations ( [ intr-1 ] ) and ( [ intr-3 ] ) , inspired by ( [ intr-11 ] ) , it seems reasonable that the law of the process @xmath3 can be approximated by that of some process similar to @xmath61 with kernel @xmath62 . in this paper
we define a sequence of processes @xmath63\}_{n\geq1}$ ] as follows : @xmath64 for @xmath65 . here , @xmath66 is a poisson process with intensity @xmath67 ( see definition [ d21 ] ) .
the main purpose of this paper is to show that the law of @xmath68 converges to the law of @xmath3 for @xmath65 .
we note that the kernel @xmath62 in ( [ intr-5 ] ) can not be separated by the arguments @xmath69 , unlike the kernel function in ( [ intr-11 ] ) .
this difference is not trivial . as we will see in remark [ rem3.1 ] , it causes many real technical difficulties .
the rest of the paper is organized as follows .
section [ 2 ] is devoted to introducing the necessary definitions , notation and the main result . in section
[ 3 ] we prove the family of processes @xmath70 given by ( [ intr-5 ] ) is tight in @xmath15 $ ] . in section
[ 4 ] we prove that as @xmath67 tends to infinity , the finite - dimensional distributions of @xmath71 converge weakly to those of the fractional brownian motion @xmath3 with @xmath72 and hence @xmath71 converges weakly in @xmath15 $ ] to the fractional brownian motion @xmath3 .
we now give the definitions of the brownian sheet and poisson processes on @xmath73 .
let @xmath74 be the borel algebra on @xmath73 .
@xmath75 and @xmath76 denote a @xmath77-finite measure and the lebesgue measure on @xmath73 , respectively .
[ d21 ] given a positive constant @xmath78 , a random set function @xmath79 on the measure space @xmath80 is called the _ poisson random measure with density measure _ @xmath81 if it satisfies the following conditions : a. for every @xmath82 with @xmath83 , @xmath84 is a poisson random variable with parameter @xmath85 defined on the same probability space ; b. if @xmath86 are disjoint and all have finite measure , then the random variables @xmath87 are independent ; c. if @xmath88 are disjoint and @xmath89 , then @xmath90 a.s .
if @xmath79 is a poisson random measure with density measure @xmath91 , then we define @xmath92\times[0,t ] ) , & \quad$s\ge0,$ $ t\ge0,$\cr n([0,s]\times[t,0]),&\quad$s\ge0,$ $ t\le0,$\cr n([s,0]\times[0,t]),&\quad$s\le0,$ $ t\ge0,$\cr n([s,0]\times[t,0 ] ) , & \quad$s\le0,$ $ t\le0 $ } \ ] ] and call @xmath93 the two - parameter poisson process with intensity @xmath94 in @xmath73 .
it is not hard to see that @xmath95 is independent with @xmath96 , which is the ordinary two - parameter process in @xmath97 , and for any @xmath98 , @xmath99 has the same finite - dimensional distributions as those of @xmath100 consider a random set function @xmath101 on the measure space @xmath102 such that : a. for every @xmath82 with @xmath103 , @xmath104 is a centered gaussian random variable defined on the same probability space with variance @xmath105 ; b. if @xmath106 are disjoint and have finite measure , then the random variables @xmath107 are independent ; c. if @xmath108 are disjoint and @xmath109 , then @xmath110 a.s .
we then call @xmath101 a _ gaussian random measure on @xmath73 with control measure _ @xmath75
. in particular , if @xmath101 is a gaussian random measure on @xmath73 with control measure @xmath111 , then we define @xmath112\times[0,t ] ) , & \quad$s\geq0,$ $ t\geq0,$\cr w([0,s]\times[t,0]),&\quad$s\geq0,$ $ t\leq0,$\cr w([s,0]\times[0,t]),&\quad$s\leq0,$ $ t\geq0,$\cr w([s,0]\times[t,0]),&\quad$s\leq0,$
$ t\leq0 $ } \ ] ] and call @xmath113 the _ two - parameter brownian sheet in _ @xmath73 . similarly , we have that @xmath114 is independent of @xmath115 , which is the ordinary brownian sheet in @xmath97 , and for any @xmath116 , we have @xmath117 hence , from ( [ intr-3 ] ) it is easy to check that @xmath118 let @xmath119 if @xmath49 and @xmath120 if @xmath121 .
we have the following conclusion , which essentially parallels bardina and jolis @xcite , theorem 1.1 .
the proof is omitted .
[ pre - lem-1 ] suppose that @xmath122 is a two - parameter poisson process with intensity @xmath123 in @xmath73 .
for any @xmath124 and @xmath125 , let @xmath126 for any @xmath127 , @xmath128 .
the finite - dimensional distributions of @xmath129 then converge weakly to those of a two - parameter brownian sheet @xmath130 .
naturally , ( [ pre-1 ] ) and ( [ pre-2 ] ) suggest that we consider the following approximation of @xmath6 for @xmath65 : @xmath131 for @xmath132 and @xmath133 $ ] , where @xmath134 is the two - parameter poisson process with intensity @xmath67 .
the main result of this paper is as follows .
[ s4-thhm1 ] for any fixed @xmath51 , the law of the process @xmath135\}$ ] given by ( [ pre-3 ] ) converges weakly to the law of @xmath136\}$ ] for @xmath137 in @xmath15 $ ] .
[ r21 ] if we define @xmath138 , then @xmath139 has a more natural physical interpretation .
it measures the length of the time interval contained in @xmath40 $ ] during which the session with duration @xmath39 and finishing time @xmath38 is active .
define @xmath140 where @xmath141 .
it is then easy to see that theorem [ s4-thhm1 ] holds for @xmath142 .
note that @xmath143 is a given positive number . in the sequel
we will treat it as a constant .
in addition , since the parameter @xmath59 can not affect our discussion , we will take it to be @xmath123 in order to simplify matters .
we now introduce some auxiliary notation . in the sequel ,
@xmath144 and @xmath145 will denote positive infinity and negative infinity , respectively . for all @xmath146 , @xmath147 , by cairoli and walsh @xcite
, we define @xmath148\times(y , y']\bigr).\ ] ] for any @xmath149 $ ] , @xmath150 and @xmath151 , we let @xmath152 define a function @xmath153 as follows : @xmath154}\,\mathrm{d}x_2\,\mathrm{d}y_2,\ ] ] where @xmath155 , @xmath156 and @xmath157 is a measurable function such that the integral is meaningful . obviously , if @xmath158 , then @xmath159
let @xmath161\}$ ] be the process defined by ( [ pre-3 ] ) . the purpose of this section is to prove the tightness of the processes @xmath162 .
[ s3-prop1 ] the family of laws of the processes @xmath162 is tight in @xmath160 $ ] . to prove the proposition , we need the following lemmas . [ s3-lem1 ]
let @xmath163 and @xmath164 .
for a non - negative function @xmath165 , if @xmath166 , then for any @xmath167 , @xmath168 \nonumber \\[-8pt]\\[-8pt ] & & \quad \leq2\bigl(i_1(n , f)+i_2(n , f)\bigr ) , \nonumber\end{aligned}\ ] ] where @xmath169}\,\mathrm{d}x_1\,\mathrm{d}y_1\,\mathrm{d}x_2\,\mathrm{d}y_2.\quad\end{aligned}\ ] ] let @xmath170 $ ] . by fubini s theorem ,
the left - hand side of ( [ s3-lem1 - 0 ] ) is equal to @xmath171 define @xmath172 and @xmath173 .
then , by ( [ intr-6 ] ) , @xmath174 equals @xmath175\nonumber \\[-8pt]\\[-8pt ] & & \quad={\mathbb{e}}\bigl[(-1)^{\delta_{(0,0)}n_{n}(x_{1},|y_{1}| ) + \delta_{(0,0)}n_{n}(x_{2},|y_{2}|)}\bigr].\nonumber\end{aligned}\ ] ] note that @xmath176 is equal to the sum of the increments of the poisson process over some disjoint rectangles , and the rectangles which contribute to the value of @xmath177 are those which appear only once .
since two - parameter poisson processes have independent increments , after some simple calculation , we obtain that on @xmath178 , @xmath179\nonumber \\[-8pt]\\[-8pt ] & & \quad= { \mathbb{e}}\bigl[(-1)^{\delta_{(0,|y_{1}|)}n_{n}(x_{1},|y_{2}|)}\bigr]{\mathbb{e}}\bigl[(-1)^{\delta_{(x_{1},|y_{1}| ) } n_{n}(x_{2},|y_{2}|)}\bigr]{\mathbb{e}}\bigl[(-1)^{\delta_{(x_{1},0)}n_{n}(x_{2},|y_{1}| ) } \bigr]\nonumber \\ & & \quad= \exp\{-2n(x_1y_1-x_2y_2)\}.\nonumber\end{aligned}\ ] ] using the same method as above , we obtain that @xmath180\}\qquad\mbox{on } \omega_2 , \\ \label{s3-l1 - 4 } i(n , x_{1,2},y_{1,2})&=&\exp\{-2n [ x_2(y_1-y_2)-(x_1-x_2)y_1]\}\qquad\mbox{on } \omega_3 , \\ \label{s3-l1 - 5 } i(n , x_{1,2 } , y_{1,2})&=&\exp\{-2n(x_2y_2-x_1y_1)\}\qquad\mbox{on } \omega_4.\end{aligned}\ ] ] substituting ( [ s3-l1 - 2])([s3-l1 - 5 ] ) into ( [ s3-l1 - 1 ] ) and using a change of the integration variables if necessary , we can easily obtain ( [ s3-lem1 - 0 ] ) .
[ s3-lem2 ] if @xmath181 , then @xmath182 defined by ( [ s3 - 1 ] ) is such that @xmath183 define @xmath184 , @xmath185 and @xmath186 .
let @xmath187 , @xmath188 and @xmath189 denote the integral ( [ s3 - 1 ] ) , where @xmath178 is replaced by @xmath190 , @xmath191 and @xmath192 , respectively .
then , @xmath193 using the elementary inequality @xmath194 from ( [ s3 - 1 ] ) , we have that @xmath195 where @xmath196 is @xmath197and @xmath198 is @xmath199 since , for any @xmath200 , @xmath201 @xmath196 is bounded by @xmath202}\,\mathrm{d}x_1\,\mathrm{d}y_1\,\mathrm{d}x_2\,\mathrm{d}y_2.\ ] ] integrating with respect to @xmath203 and then with respect to @xmath204 in the above integral , we obtain the following bound : @xmath205 furthermore , in the region @xmath190 , it follows from ( [ s3-lem1-a0 ] ) that @xmath206 by the same argument as above , we then have that @xmath207 therefore , from ( [ s3-lem1 - 21 ] ) , ( [ s3-lem1-a2 ] ) and ( [ s3-lem1-a3 ] ) , it follows that @xmath208 we now consider @xmath209 .
note that @xmath210for any @xmath211 .
define @xmath212 and @xmath213 as @xmath196 and @xmath198 , respectively , with @xmath191 instead of @xmath190 .
then , @xmath214 integrating in @xmath212 with respect to @xmath215 and @xmath216 , and in @xmath213 with respect to @xmath217 and @xmath218 , respectively , we obtain that @xmath219 since @xmath220 , we have @xmath221 similarly , for any @xmath222 , we have that @xmath223 therefore , using a similar approach to the one above , we have @xmath224 combining ( [ s3-l2 - 1 ] ) with ( [ s3-l2 - 2])([s3-l2 - 4 ] ) , we get ( [ s3-l2 - 0 ] ) . from ( [ s2 - 3 ] ) and lemma [ s3-lem2 ]
, we can immediately get the following corollary . [ s3-c1 ] for each @xmath155 , @xmath225 .
observe that for every non - negative measurable function @xmath165 , by ( [ intr-7 ] ) and ( [ s3 - 2 ] ) , @xmath226 the following lemmas concern @xmath227 .
we focus on the case where @xmath228 .
[ s3-lema ] for any @xmath229 , let @xmath230 and @xmath231 .
there then exist a non - negative function @xmath232 on @xmath233 such that for all @xmath155 , @xmath234 and a positive constant @xmath235 which depends only on @xmath4 , such that for all @xmath155 , @xmath236 by ( [ def : h ] ) and ( [ def : g ] ) , @xmath237 let @xmath238 and @xmath239 .
then , @xmath240 where @xmath241}\,\mathrm{d}x_2\,\mathrm{d}y_2 , \\ i_{22 } & = & n^2\int_{y_1}^{-s - x_1}\!\!\int_{-s - y_2}^{\infty}\phi _ t(x_2,y_2)\sqrt{x_2|y_1y_2| } { \mathrm{e}}^{-2n[x_1(y_2-y_1)-(x_2-x_1)y_2]}\,\mathrm{d}x_2\,\mathrm{d}y_2 , \\ i_{23 } & = & n^2\int_{-s - x_1}^{-s}\!\int_{x_1}^{\infty}\phi _ t(x_2,y_2)\sqrt{x_2|y_1y_2| } { \mathrm{e}}^{-2n[x_1(y_2-y_1)-(x_2-x_1)y_2]}\,\mathrm{d}x_2\,\mathrm{d}y_2.\end{aligned}\ ] ] for @xmath242 , from ( [ s3-la-1 ] ) , we have that @xmath243}\,\mathrm{d}x_2\,\mathrm{d}y_2 \nonumber \\ & \leq & \sqrt{\frac{t+s}{s}}n^2\int_{y_1}^{-s - x_1}\!\int _ { x_1}^{-s - y_2}\frac{1}{x_2^{1/2-h}}|y_2| { \mathrm{e}}^{-2n[x_1(y_2-y_1)-(x_2-x_1)y_2]}\,\mathrm{d}x_2\,\mathrm{d}y_2\nonumber \\[-8pt]\\[-8pt ] & \leq & t^{h-1/2}\sqrt{\frac{t+s}{s}}\frac{n}{2}\int _ { y_1}^{-s - x_1}{\mathrm{e}}^{-2nx_1(y_2-y_1)}\,\mathrm{d}\nonumber y_2 \\ & \leq & \frac{t^{h-1/2}}{4x_1}c_s=:\psi_1(x_1,y_1),\nonumber\end{aligned}\ ] ] where @xmath244 is a constant .
similarly , for @xmath245 , @xmath246}\,\mathrm{d}x_2\,\mathrm{d}y_2 \nonumber \\ & \leq & c_sn^2\int_{y_1}^{-s - x_1}\!\int_{-s - y_2}^{\infty } ( -s - y_2)^{h-1/2}|y_2| { \mathrm{e}}^{-2n[x_1(y_2-y_1)-(x_2-x_1)y_2]}\,\mathrm{d}x_2\,\mathrm{d}y_2\nonumber \\[-8pt]\\[-8pt ] & \leq & ( -s - y_1)^{h-1/2}c_s\frac{n}{2}\int_{y_1}^{-s - x_1 } { \mathrm{e}}^{-2nx_1(y_2-y_1)}\,\mathrm{d } y_2 \nonumber \\ & \leq & \frac{(-s - y_1)^{h-1/2}}{4x_1}c_s=:\psi_2(x_1,y_1)\nonumber\end{aligned}\ ] ] and @xmath247}\,\mathrm{d}x_2\,\mathrm{d}y_2\nonumber \\ & \leq & c_sn^2\int_{-s - x_1}^{-s}\int_{x_1}^{\infty}\frac { x_1}{x_1^{3/2-h}}|y_2| { \mathrm{e}}^{-2n[x_1(y_2-y_1)-(x_2-x_1)y_2]}\,\mathrm{d}x_2\,\mathrm{d}y_2 \\ & \leq & c_s\frac{1}{4x_1^{3/2-h}}=:\psi_3(x_1,y_1).\nonumber \ ] ] in addition , for @xmath248 , @xmath249}\,\mathrm{d}x_2\,\mathrm{d}y_2 \nonumber \\[-8pt]\\[-8pt ] & \leq & c_s\frac{-s - y_1}{4x_1^{2-h}}=:\psi _ 4(x_1,y_1).\nonumber\end{aligned}\ ] ] define @xmath250 as @xmath251 obviously , @xmath252 is positive and ( [ s3-la-01 ] ) follows from ( [ s3-la-2])([s3-la-7 ] ) .
note that @xmath253 with some basic calculations , we obtain the following results : @xmath254 from the above integrals , ( [ s3-la-02 ] ) follows with @xmath255 .
[ s3-lem4 ] for @xmath229 and each constant @xmath256 , let @xmath257 and @xmath258 . there
then exist a non - negative function @xmath259 on @xmath260 such that for all @xmath155 , @xmath261 and a positive constant @xmath262 which only depends on @xmath4 , such that @xmath263 from ( [ s3-la-1 ] ) , we have that @xmath264 where @xmath265}\,\mathrm{d}x_2\,\mathrm{d}y_2 , \\ \hat{i}_{22 } & = & n^2\int_{y_1}^{-t - s}\!\!\!\int_{-x_1}^{-s - y_2}\frac{x_2+t+s+y_2}{x_2^{3/2-h}}\sqrt{x_1|y_1y_2|}{\mathrm{e}}^{-2n[x_1(y_2-y_1)-(x_2-x_1)y_2]}\,\mathrm{d}x_2\,\mathrm{d}y_2.\end{aligned}\ ] ] for @xmath266 , since @xmath267 , @xmath268}\,\mathrm{d}x_2\,\mathrm{d}y_2 \nonumber \\[-8pt]\\[-8pt ] & \leq & \sqrt{\frac{|m|}{s}}\frac{t}{4x_1^{2-h}}.\nonumber\end{aligned}\ ] ] therefore , from ( [ s3-l4 - 2 ] ) and ( [ s3-l4 - 3 ] ) , it follows that @xmath269 let @xmath270 . then , ( [ s3-l4 - 01 ] ) holds .
furthermore , by some basic calculations , @xmath271\,\mathrm{d } y_1\qquad \\ & & \quad= \sqrt{\frac{|m|}{s}}\biggl[\frac{t^{2h}}{4(1-h)(3 - 2h)}+\int_{0}^{t}\mathrm{d } x_1\int_{-t - s - x_1}^{-t - s}\frac{t(x_1+t+s+y_1)}{2x_1^{4 - 2h}}\,\mathrm{d}y_1\qquad \nonumber \\ & & \quad\hphantom{=\sqrt{\frac{|m|}{s}}\biggl[}{}+ \int_{t}^{\infty}\!\!\mathrm{d}x_1\int_{-t - s - x_1}^{-s - x_1}\frac{t(x_1+t+s+y_1)}{2x_1^{4 - 2h}}\mathrm{d}y_1\biggr]\qquad \nonumber \\ & & \quad= \frac{1}{4(1-h)(3 - 2h)(2h-1)}\sqrt{\frac{|m|}{s}}t^{2h}.\qquad\nonumber\end{aligned}\ ] ] hence , ( [ s3-l4 - 02 ] ) holds for @xmath272 .
[ s3-lem5 ] for @xmath229 , @xmath273 , let @xmath274 and @xmath275 .
there then exist a non - negative function @xmath276 on @xmath277 such that for all @xmath155 , @xmath278 and a constant @xmath279 which is independent of @xmath280 and @xmath4 , such that @xmath281 where @xmath262 is the constant in lemma [ s3-lem4 ] . from ( [ s3-la-1 ] )
, it follows that @xmath282}\,\mathrm{d}x_2\,\mathrm{d}y_2 \nonumber \\ & = & \frac{tn^2}{x_1^{3/2-h}}\int_{y_1}^{m}\!\!\sqrt{x_1}|y_1|\int _
{ x_1}^{\infty } { \mathrm{e}}^{-2n[x_1(y_2-y_1)-(x_2-x_1)y_2]}\,\mathrm{d}x_2\,\mathrm{d}y_2 \nonumber \\[-10pt]\\[-10pt ] & = & \frac{t}{2x_1^{2-h}}\int_{|m|}^{|y_1|}\frac{nx_1|y_1|}{|y_2| } { \mathrm{e}}^{2nx_1|y_2|}\,\mathrm{d}|y_2|{\mathrm{e}}^{-2nx_1|y_1| } \nonumber \\[-2pt ] & = & \frac{t}{2x_1^{2-h}}{\mathrm{e}}^{-2nx_1|y_1|}nx_1|y_1| \int_{nx_1|m|}^{nx_1|y_1|}\frac{1}{w}{\mathrm{e}}^{2w}\,\mathrm{d}w.\nonumber\end{aligned}\ ] ] let @xmath283 for @xmath284 .
then , @xmath285 is continuous on @xmath286 and @xmath287 , @xmath288 .
hence , there is a constant @xmath279 , which is independent of @xmath289 and @xmath4 , such that @xmath285 is bounded by @xmath290 .
note that on @xmath291 , because @xmath292 and @xmath293 we have @xmath294 for all @xmath155 .
therefore , ( [ s3-l5 - 2 ] ) yields @xmath295 let @xmath296 . by calculations similar to those in ( [ s3-l4 - 4 ] ) , ( [ s3-l5 - 02 ] ) holds for @xmath297 .
[ s3-prop2 ] for any @xmath229 , there exists a constant @xmath19 , independent of @xmath298 but dependent on @xmath143 , such that @xmath299 \leq k t^{2h}.\ ] ] using the same notation as in lemmas [ s3-lem1][s3-lem5 ] , taking @xmath300 and observing that from ( [ s3-la-1 ] ) , @xmath301 if @xmath302 is not in the set @xmath303 we can rewrite the left - hand side of ( [ s3-p1 - 0 ] ) by @xmath304\hspace*{35pt } \nonumber \\[-10pt]\\[-10pt ] & = & { \mathbb{e}}\biggl[\biggl(n\int_0^n\!\!\!\int_{-n}^0\bigl(\tilde{\phi } _
t(x , y)+\hat{\phi}_t(x , y)+\bar{\phi}_t(x , y)\bigr ) \sqrt{x|y|}(-1)^{n_{n}(x , y)}\,\mathrm{d}x\,\mathrm{d}y \biggr)^{2}\biggr ] , \nonumber\end{aligned}\ ] ] which is bounded by @xmath305 note that @xmath306 .
lemma [ s3-lem1 ] , corollary [ s3-c1 ] and lemma [ s3-lema ] imply that @xmath307\\[-10pt ] & \leq&\biggl(\frac { 37}{8}+2k_1\sqrt{\frac{1+s}{s}}\biggr)t^{2h},\nonumber\end{aligned}\ ] ] and lemma [ s3-lem1 ] , corollary [ s3-c1 ] and lemma [ s3-lem4 ] imply that @xmath308 furthermore , from lemma [ s3-lem1 ] , corollary [ s3-c1 ] and lemma [ s3-lem5 ] , we have @xmath309 therefore , ( [ s3-p1 - 1])([s3-p1 - 5 ] ) yield that @xmath310t^{2h}.\ ] ] taking @xmath311 $ ] then completes the proof of proposition [ s3-prop2 ] . finally , we prove proposition [ s3-prop1 ] .
proof of proposition [ s3-prop1 ] to prove the tightness of @xmath162 in @xmath160 $ ] , it suffices to show that for some @xmath312 there exist two constants @xmath313 and @xmath314 such that for any @xmath315 $ ] , @xmath316 \\ & & \quad\leq\tilde { m}(t - t')^{\eta},\end{aligned}\ ] ] which follows from the criterion given by billingsley ( see @xcite , theorem 12.3 ) and the fact that our processes are null at the origin . without loss of generality ,
let @xmath317 .
note that from ( [ def : g ] ) , we have @xmath318 by proposition [ s3-prop2 ] , it is easy to check that the above inequality holds for @xmath32 , @xmath319 and @xmath320 $ ] , which completes the proof of proposition [ s3-prop1 ] .
[ rem3.1 ] compared with the proofs of tightness in bardina et al .
@xcite , we have found that under the condition that the kernel @xmath157 can be separated by its arguments @xmath69 , the calculation of @xmath321 is transformed to the calculation of @xmath322 which is relatively simple ; see the proofs of lemmas 3.1 and 3.2 in @xcite and the proof of lemma 3.1 in @xcite .
however , in our case , the kernel @xmath157 can not be separated by the arguments @xmath69 , so we need to discuss @xmath227 . from our proof
, we can see that the calculation of @xmath227 is more complicated and delicate than that of @xmath182 .
in addition , the fact that the kernel @xmath157 can not be separated by the arguments @xmath69 also creates some difficulties in the identification of the limit law ; see the proof of ( [ s4-p1 - 8 ] ) in the next section . by proposition 10.3 in @xcite ,
page 149 , proposition [ s3-prop1 ] also shows that @xmath323 , n\in\mathbb{n})=1.$ ]
in this section , we proceed with the identification of the limit law
. we will prove the following proposition .
[ s4-prop1 ] the finite - dimensional distributions of @xmath324\}$ ] defined by ( [ pre-3 ] ) converge weakly , as n tends to @xmath144 , to those of a fractional brownian motion @xmath325\}$ ] with hurst index @xmath65 . for each @xmath326 , @xmath327 and @xmath328 , we define @xmath329 it suffices to prove that for any @xmath330 , as @xmath331 , @xmath332-{\mathbb{e}}[\exp ( \mathrm{i}\xi u)]|\to0.\end{aligned}\ ] ] for @xmath333 , let @xmath334 and @xmath335 where @xmath336 is given by lemma [ pre - lem-1 ] .
let @xmath337 below , we estimate @xmath338 , @xmath339 and @xmath340 , respectively .
\(1 ) we estimate @xmath341 .
noting that @xmath342 is a non - negative measurable function on @xmath277 , we can find a sequence of elementary functions @xmath343 such that @xmath344 then , by the dominated convergence theorem , it follows from the fact that @xmath345 ^ 2\,\mathrm{d}x\,\mathrm{d}y < \infty$ ] that as @xmath346 , @xmath347 for any @xmath348 , define @xmath349 by lemma [ pre - lem-1 ] , we can readily verify that for fixed @xmath350 , as @xmath331 , @xmath351-{\mathbb{e}}\biggl[\exp\biggl(\mathrm{i}\xi \sum_{j=1}^{k}a_{j}b^{m , j}\biggr)\biggr]\biggr|\to0\end{aligned}\ ] ] because @xmath352 is essentially a linear combination of increments of @xmath129 defined by ( [ pre-2 ] ) , and @xmath353 is the same linear combination of the corresponding limits of increments of @xmath129 .
we further define @xmath354-{\mathbb{e}}\biggl[\exp\biggl(\mathrm{i}\xi\sum_{j=1}^{k}a_{j } y_{n}^{m , j}\biggr)\biggr]\biggr| , \\
j_{13}(t , m ) & : = & \biggl|{\mathbb{e}}[\exp(\mathrm{i}\xi u(t))]-{\mathbb{e}}\biggl[\exp\biggl(\mathrm{i}\xi\sum_{j=1}^{k}a_{j}b^{m , j}\biggr)\biggr]\biggr|.\end{aligned}\ ] ] then , for any @xmath355 , @xmath356 below , we will show that for all fixed @xmath357 , there exists some @xmath358 such that for any @xmath359 @xmath360 as @xmath346 . to this end , let @xmath361 .
define @xmath362\times[-t_j - s , 0)}(x , y ) , \\
\hat{f}_{m , j}(x , y)&:=&f_{m , j}(x , y)\mathbf{1}_{[0 , t]\times[-t , -t_j - s)}(x , y).\end{aligned}\ ] ] by ( [ s4-p1 - 3 ] ) , as @xmath346 , @xmath363\times[-t , 0].\ ] ] define @xmath364.\ ] ] then , by lemma [ s3-lem1 ] , @xmath365 ^ 2 & \leq & n^2e\biggl[\biggl(\int_{0}^{t}\!\!\!\int_{-t}^{0}\bigl(\hat{f}_{m , j}(x , y)+\tilde{f}_{m , j}(x , y)\bigr)\sqrt{x|y|}(-1)^{n_{n}(x , y)}\,\mathrm{d}x\,\mathrm{d}y\biggr)^2\biggr]\quad \nonumber \\ & \leq & 2\bigl(i(n , \hat{f}_{m , j})+i(n , \tilde{f}_{m , j})\bigr ) \\ & \leq & 4\bigl(i_1(n , \hat{f}_{m , j})+i_2(n , \hat{f}_{m , j})+i_1(n,\tilde{f}_{m , j})+i_2(n , \tilde{f}_{m , j})\bigr ) .
\nonumber\end{aligned}\ ] ] lemma [ s3-lem2 ] and ( [ s4-p1 - 4 ] ) show that for any @xmath155 , as @xmath366 @xmath367 note that @xmath368 a.e . as @xmath346 . from lemma [ s3-lema ] we know that @xmath369 and that @xmath370 by the dominated convergence theorem , as @xmath346 , @xmath371 in a similar way , we know there are @xmath372 , @xmath373 such that for all @xmath155 , as @xmath346 , @xmath374 on the other hand , using the mean value theorem , we obtain that @xmath375 \nonumber \\[-6pt]\\[-6pt ] & = & k|\xi| \max_{1\leq j\leq k}(|a_j|r_j(n , m , t)).\nonumber\end{aligned}\ ] ] hence , ( [ s4-p1 - 8 ] ) follows from ( [ s4-p1 - 10])([s4-p1 - 14 ] ) with @xmath376 for @xmath377 , we apply the mean value theorem again . then , as @xmath346 , @xmath378 \nonumber \\[3pt ] & \leq & k\xi\max_{1\leq j\leq k}{\mathbb{e}}\biggl[\biggl|a_j\int_0^t\!\!\!\int_{-t}^0\bigl(\phi_{t_j}(x , y)-q^{m , j}(x , j)\bigr)b(\mathrm{d}x,\mathrm{d}y)\biggr|\biggr ] \\[3pt ] & \leq & k\xi\max_{1\leq j\leq k}\biggl\{\biggl[\int_{0}^{t}\!\!\!\int_{-t}^{0}\bigl(\phi_{t_j}(x , y)-q^{m , j}(x , y)\bigr)^{2}\,\mathrm{d}x\,\mathrm{d}y\biggr]^{1/2}\biggr\}\to0 . \nonumber\end{aligned}\ ] ] from ( [ s4-p1 - 5])([s4-p1 - 8 ] ) and ( [ s4-p1 - 15 ] ) , we obtain that for any fixed @xmath357 , as @xmath331 , @xmath379 \(2 ) in a similar way as was used to prove ( [ s4-p1 - 8 ] ) , there exists some @xmath380 such that for @xmath381 , as @xmath382 , @xmath383 \(3 ) in a similar way as was used to prove ( [ s4-p1 - 15 ] ) , we obtain that there exists some @xmath384 such that as @xmath382 , @xmath385 therefore , combining ( [ s4-p1 - 2 ] ) and ( [ s4-p1 - 16])([s4-p1 - 18 ] ) , we can obtain that @xmath386 as @xmath331 , completing the proof of proposition [ s4-prop1 ] .
note that theorem [ s4-thhm1 ] is an immediate result of propositions [ s3-prop1 ] and [ s4-prop1 ] .
therefore , the proof of theorem [ s4-thhm1 ] is complete .
this work was done during our visit to the department of statistics and probability at michigan state university .
the authors thank professor yimin xiao for stimulating discussions and the department for its good working conditions .
this research was supported in part by the national natural science foundation of china ( no .
10901054 ) .
thanks is also due to the anonymous referees for their careful reading and detailed comments which improved the quality of the paper .
kaj , i. and taqqu , m.s .
convergence to fractional brownian motion and to the telecom process : the integral representation approach . in _ in and out of equilibrium 2_.
probab . _ * 60 * 383427 .
basel : birkhuser . | approximations of fractional brownian motion using poisson processes whose parameter sets have the same dimensions as the approximated processes have been studied in the literature . in this paper , a special approximation to the one - parameter fractional brownian motion is constructed using a two - parameter poisson process .
the proof involves the tightness and identification of finite - dimensional distributions . |
accreted stellar halos are a natural consequence of galaxy formation in a @xmath1cdm universe , in which the smallest structures collapse first and galaxy formation is thought to proceed bottom - up ( e.g. * ? ? ?
* ; * ? ? ?
halos , composed of both dark and baryonic matter , grow by merging with other halos . while the gas from mergers and accretions loses its energy through cooling and settles into a disk , the accreted stars and dark matter form a halo around the galaxy .
galactic stellar halos are thus predicted to be built from multiple accretion events starting from the first structures to collapse in the universe .
it is for this reason that much work has focused on using kinematically identified halo stars to look for fossil records of the galaxy s accretion history .
the possibility remains , however , that stars formed within the potential well of a galaxy can become displaced from the inner - most regions into a kinematic stellar halo . in order to draw conclusions about the accretion history of a galaxy based on halo stars
, one must know how abundant such in - situ stars are in the halo , and how they may dilute the signatures of accretions and mergers .
owing to the low surface brightness of halos , the only presently well studied stellar halos are those of the milky way and m31 .
current observational studies of the milky way s halo find strong evidence that its outer regions ( @xmath5kpc ) were at least partially assembled through accretion events .
numerous stellar streams believed to have originated from small accreted galaxies have been observed in the milky way s outer halo ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
multiple wraps of stars believed to have been tidally removed from the sagittarius dsph @xcite highlight the process of disrupting satellites in the milky way s potential , and serve as strong evidence of hierarchical merging . @xcite
have found , using f and g turn off stars in sdss , that the distribution of halo stars is highly structured , and the outer halo is consistent with being formed almost entirely from accreted subhalos .
m31 s stellar halo is observed to contain possibly even more substructure and streams than what is observed around the milky way ( e.g. * ? ? ?
? * ; * ? ? ?
* ) , suggesting its recent merger history may have been more active .
despite the plethora of streams observed around the milky way and m31 , both the fraction of halo that is accreted , as well as the number of distinct accretion events leading to the observed substructure , is unclear .
although the nearby halo of our galaxy has been studied more extensively than the outer halo , in some ways its properties are less well understood .
this is due in part to the short dynamical timescales close to the disk , where accreted stars would no longer betray their origin through spatial structures .
it is in these inner several tens of kpc of the halo that one might expect dissipative processes to contribute to the overall stellar population .
observational evidence of the contribution of accretion to the inner halo is slowly growing .
@xcite have detected two streams in the solar neighborhood by studying the angular momentum of stellar orbits .
@xcite have also recently shown that the angular momentum distribution of a sample of stars with a median height above the galactic plane of 2 kpc does not seem to be smooth , possibly indicative of an accretion origin .
a hybrid theory in which the milky way was assembled through a combination of hierarchical accretions and dissipative processes has been used by some to understand properties of field halo stars ( e.g. * ? ? ?
* ; * ? ? ?
@xcite , using a large set of halo stars selected with no kinematic bias , find a vertical gradient in the mean rotation of metal poor stars , a signature of dissipational processes .
@xcite , using number count maps , have shown that a large fraction of the halo of m31 appears to be rather smooth .
the authors interpret these observations to suggest that some of the halo of m31 may have formed through dissipative processes .
however , the degree to which dissipation is important in the formation of stellar halos , or one of their components , remains uncertain .
simulations have also been extensively used to investigate the formation of galactic halos .
the work of @xcite combines dark matter only n - body simulations with semi - analytic models to study the build up of stellar halos via accretions alone .
their work has been highly successful in explaining the discrepant chemical abundances of milky way satellite galaxies and field halo stars @xcite .
while this powerful and efficient tool is particularly suited to better understand progenitors and signatures of accretion events , it can not tell us about the possible relative importance of dissipative processes in the assembly of a galactic halo .
@xcite have used n - body + sph to simulate the formation of disk galaxies in a self consistent way , following the evolution of dark matter , gas and star formation .
they find that recently accreted satellites contribute stars to the halo which can be distinguished from an early formed halo through phase - space information .
@xcite , using a set of fully cosmological simulations , find that in - situ stars dominate the inner 20 kpc of their galaxies , while accreted stars make up the majority of the outer regions .
while the abadi work distinguishes between the accreted and in - situ populations in their galaxies , the work does not investigate the possible importance of in - situ stars in simulated stellar halos . in this paper , we use four high resolution cosmological sph + n - body simulations to investigate the origin of stellar halos .
these self - consistent simulations permit the study of the competing importance of in - situ star formation and accretion from subhalos in the formation of stellar halos in a @xmath1cdm universe .
furthermore , the supernova feedback and cosmic uv background implemented in our simulated galaxies results in a luminous satellite population that is more similar to the milky way s than those in previous n - body + sph studies .
this improvement is essential to model realistic stellar halos .
we quantify the degree to which the halos are accreted , and note the effect of accretion histories on the overall formation of the stellar halos .
while past numerical efforts have concentrated on the growth of stellar halos through pure accretion , we show that both accretion and in - situ star formation contribute to the inner regions of stellar halos .
the paper is organized as follows .
section 2 describes the details of the simulations used , and the methods used to determine the origin of halo stars . in section 3 we discuss the dual origin of the stellar halos of our simulated galaxies , and investigate how the properties of the in - situ and accreted populations in the stellar halos vary with merging history .
section 4 discusses possible numerical and resolution effects on the formation of the stellar halos in our simulations , as well as the effect of feedback on the accreted fraction .
section 5 discusses the connection to observational results and concludes .
lcccccc mw1hr&@xmath6 & @xmath7 & @xmath8&@xmath9&0.3 + mw1med&@xmath6&@xmath10&@xmath11&@xmath12&0.3 + gal1&@xmath13&@xmath14&@xmath15 & @xmath16&0.3 + h277&@xmath17&@xmath18&@xmath19 & @xmath20&0.35 + h285&@xmath21&@xmath22&@xmath19&@xmath20&0.35 +
the high resolution simulations used in this study were run to z = 0 , as part of an ongoing simulation project .
the two more massive galaxies ( mw1hr , gal1 ) are described in detail in ( * ? ? ?
* hereafter g07 ) , but have since been rerun at eight times the mass resolution and twice the spatial resolution .
the reader is referred to g07 for the full details of these simulations . the remaining two lower mass galaxies ( h277
, h285 ) are a continuation of these efforts .
h277 is described in @xcite .
we briefly summarize the salient features of these simulations here .
the halos used in this study were selected , based on their mass and merger history , from low resolution , dark matter only simulations run with gasoline @xcite , using wmap year 1 cosmology for mw1hr and gal1 , and wmap year 3 cosmology for h277 and h285 .
each halo was then resimulated at higher resolution using the volume renormalization technique @xcite .
this approach simulates only the region within a few virial radii at the highest resolution , while still simulating a large enough box to account for the effect the large scale tidal field has on the angular momentum of the halo .
we concentrate a large part of this work on mw1hr , the highest resolution n - body + sph simulation of a milky way mass galaxy run to @xmath23 published to date .
we single this simulation out for extensive study as it has a lower resolution counterpart , run with the identical initial conditions , but 1/8 the mass resolution of mw1hr , as well as several runs completed with different treatments of feedback .
the previous extensive study done on mw1hr ( see g07 ) permit the numerical and resolution tests necessary for this work .
mw1hr has 4.9 million particles ( dark matter + gas + stars ) within its virial radius , and a mass of @xmath24 .
we define the redshift of last major merger as the time at which a secondary galaxy , with a mass no less than 1/3 of mw1hr s mass , first enters the virial radius of mw1hr .
this occurs at a redshift of 4 , although the merger is not complete until @xmath25 .
the merger history of mw1hr is thus similar to that of the milky way , which is thought to have experienced its last major merger at z @xmath26 @xcite .
the mass of star particles in this simulated run is @xmath27 , high enough to resolve an in - falling dwarf galaxy of stellar mass @xmath28 with more than @xmath29 star particles .
a physically motivated recipe was used to describe star formation and supernovae feedback @xcite , with a uniform uv background that turns on at z=9 , mimicking cosmic reionization @xcite .
the properties of all the simulated galaxies used in this work are described in table 1 . all dark matter halos and subhalos in the simulation
are identified using amiga s halo finder ( ahf , * ? ? ?
each halo is identified using a minimum of 64 dark matter particle members , above which the halo mass function converges ( g07 , * ? ? ?
the properties of each simulation particle are output at z=6 , z=5 , z=4 , and every 320 myr after until z=0 , for a total of 43 time steps output . before we can study the mechanisms that built - up the stellar halos of our simulated galaxies , it is first necessary to identify the star particles which make up the galactic halos . in both our study , as well as in observational work , it is important that the stellar samples used to draw conclusions about halo properties and formation are not contaminated by a population of disk stars .
unless an observational sample only includes stars that reside several disk scale heights above a galaxy s mid - plane , the use of kinematic and metallicity information is necessary to obtain a clean sample of halo stars .
some observational works rely on samples of stars chosen based on a metallicity cut , where the calculated [ fe / h ] is several tenths of dex below that of disk stars ( for example * ? ? ?
* ; * ? ? ?
other observational studies have used the proper motions of stars to isolate those with halo kinematics @xcite , while other works combine both metallcities and kinematics @xcite .
we use the full simulation information to identify the most complete and uncontaminated set of halo stars possible .
we have chosen to define our halo sample using the full 3 dimensional phase space information available for all stars in the simulations .
rather than implementing a specific spatial cut , this approach allows us to study the detailed origin of all stars belonging to a halo .
if we had instead chosen to define the stellar halo of each simulated galaxy as the population of stars with distances greater than @xmath30 scale heights above the galactic disk , the trends we report in the following sections would be unchanged .
however , our results then would not have been applicable to the many detailed observational studies of the milky way s halo that are focused on stars within @xmath31 . in a future paper
, we will investigate the possible observational consequences of this predicted origin .
a kinematic analysis of the star particles in each galaxy was done to decompose them into disk , bulge , and halo kinematic components . to decompose the disk from the spheroid
, we first align the angular momentum vector of the disk with the z - axis .
this is done in order to place the disk in the x - y plane .
we then calculate the angular momentum of each star in the x - y plane , @xmath32 , as well as the momentum of the co - rotating circular orbit with similar orbital energy , @xmath33 . in this definition ,
stars with circular orbits in the plane of the disk will have @xmath34 .
disk stars are selected with a cut of @xmath35 , which is equivalent to an eccentricity cut of @xmath36 .
this condition matches the eccentricities observed of disk stars in the milky way @xcite .
once the disk population has been identified , we define a sphere centered on each galaxy , with a radius equal to that of the disk s . for stars within this inner sphere , the spheroid ( halo @xmath37 bulge ) of each galaxy is defined using a cut in @xmath38 such that the net rotation of the spheroidal population equals zero .
the spheroidal component of the galaxy is further decomposed into halo and bulge because a break in the mass density profile is observed .
stars whose total energy is calculated to be low , and hence are tightly bound to the galaxy , are classified as part of the bulge .
all other stars are placed in the halo category .
the energy cut between the halo and bulge is set in order to make a two component fit of the mass profile with bulge stars within the `` break '' and halo stars outside the `` break '' .
bulge contamination is not a concern because the energy cut also ensures that stars which reside close to the center of the galaxy , and are classified as belonging to the halo , have orbits which take them far from a classical bulge .
moreover , observational studies that rely on kinematics to select halo stars would have associated our halo stars with a halo , not bulge , component .
we show in section 3 that the stellar populations we associate with the halo extend out to regions far beyond what would be considered a bulge .
stars which reside beyond the inner sphere defined by the disk s radius are all classified as halo stars , regardless of their kinematics .
we have verified that the halo population has a rotational velocity with a gaussian distribution .
the @xmath38 cuts described above leave a set of stars whose kinematics do not make either the disk or spheroid cuts .
these stars are dominated by thick disk and pseudo bulge stars .
we note that using kinematic information for decomposing galactic components unavoidably leads to some slight overlap in populations . for the remainder of the paper
, we use the term `` halo stars '' to refer to stars identified using this kinematic definition .
there is no bias for our selected halo in - situ stars to be prograde in any galaxy .
this indicates that we have not mistakenly classified a significant number of disk stars ( in particular thick disk stars ) as halo stars in our analysis .
we trace each star particle located within the main galaxy at @xmath23 back in time .
we identify at each time step the dark matter ( dm ) halo to which each particle belonged using ahf .
stars are considered bound to a dm halo if they are identified as being a member of that halo for two consecutive timesteps .
we follow the particles histories back to @xmath39 , as well as follow the gas particles from which stars have formed .
this procedure results in four different classifications for stars origin : accreted , in - situ , ambiguous , and other .
these classifications are described in the following sections .
we use these stellar histories to distinguish between stars that were accreted and those that formed `` in - situ '' .
accreted stars formed inside of halos other than the primary dark matter halo , but through accretion and merging have become unbound to their progenitor and now belong to the primary halo . in -
situ stars formed from gas particles bound to the primary .
these stars formed within the primary potential well and will be referred to as the `` in - situ '' sample . in section 3.1 we show that while these in - situ stars reside in the halo at z=0 , they formed near the center of the galaxies potential wells , as part of their primordial disks
. the gas particle progenitors of these in - situ stars ended up in the primary halo from both smooth gas accretion , particularly in cold flows @xcite , as well as from subhalos that underwent gas stripping as they merged with the primary . all galaxies are traced back far enough in redshift , to @xmath39 , that we can accurately determine the star formation history of the vast majority ( @xmath40 ) of halo stars .
approximately 1% of the stars in these simulations formed before z=6 , and as such we can not determine their formation origin .
we can thus accurately distinguish the accreted stars from the in - situ stars for the vast majority of particles in each simulated run .
our simulations also contain stars whose formation origin is unclear due to the limited number of time steps output for each simulation .
for example , if a gas particle spawns a star particle at approximately the same time that it first becomes bound to the primary , it is uncertain whether that star formed in the primary or in its original subhalo .
such stars are referred to as `` ambiguous '' in the reminder of the paper .
these stars comprise less than @xmath41 of all the kinematically identified halo stars in all of the simulations ( see table 2 ) . to ensure that the in - situ stars are a robust feature of these simulations , we now investigate whether this population of halo stars were formed preferentially near the density and/or temperature thresholds for star formation , where spurious star formation may occur due to numerical effects .
the spatial resolution of these simulations is @xmath42 pc , much coarser than the scales on which star formation occurs in real galaxies .
it is therefore necessary to use a prescription for star formation that captures globally the observed trends in galaxies .
the prescription used here for star formation enforces a local gas density greater than 0.1 @xmath43 and a temperature colder than @xmath44 k at which stars form .
if a gas particle matches these criteria , there is some probability that it will form a star particle .
this probability is scaled to match the observed kennicutt - schmidt law for star formation ( see @xcite for further details ) .
we found that in - situ stars that formed at a look back time of 10 gyrs or more did not form near the density and temperature cutoffs .
they formed under the same conditions as stars with similar ages in the disk of the galaxy .
however , in - situ halo stars in the simulation younger than 10 gyr old formed at preferentially lower densities and higher temperatures than their young disk counterparts , near the implemented cutoff for density threshold . while we have no reason to believe these late forming in - situ stars are a numerical artifact , changing our prescription for star formation would possibly affect whether or not these in - situ stars had formed at all in the simulation .
@xcite have shown that accounting for hot halo gas instabilities may be important in understanding the total baryonic mass of a galaxy .
these authors show that the fragmentation of the hot halo gas results in hot low density gas which does not cool , and is able to survive in the halo of a galaxy .
this might explain the young in - situ stars in our simulation that formed from a low density hot gas . to be maximally conservative on our prediction of the extent to which in - situ stars compose the halo , we have not included in - situ stars that formed within the last 10 gyr in any of the following analysis .
the broad conclusions of this paper are not affected by whether or not we choose to ignore these intermediate and young in - situ stars .
we refer to these stars as `` other '' for the reminder of the paper , and only consider insitu stars formed more than 10 gyr ago when discussing `` in - situ '' stars .
the total mass contributed by other and in - situ stars to each simulated galaxy is listed in table 2 .
the presence of both in - situ and accreted stars is a generic feature of the kinematically defined halos of all four simulated galaxies .
each simulated stellar halo was found to be composed of a purely accreted outer halo , and an inner halo where both accreted and in - situ populations reside .
this is illustrated by figure [ fig_radial ] .
the red solid lines in this figure show the fractional contribution of in - situ stars to the stellar halos as a function of distance .
the blue dotted lines and green dash - dot lines show the same for stars accreted to the halo at lookback times of more and less than 9 gyr , respectively .
this figure shows that all four stellar halos have an in - situ component that primarily resides within their inner @xmath45 kpc , and that the accreted component of each stellar halo extends to more than 100 kpc from the center of each galaxy .
the outermost regions of the stellar halos of all four simulated galaxies are dominated by stars accreted within the last 9 gyr .
the central location of all four in - situ populations shows that their presence is most relevant for observational studies that are limited to the inner halo of the milky way .
table 2 lists the fractional contribution , as well as the total mass , of the in - situ and accreted populations in each of the four simulated halos in our sample .
we go on to study in detail the formation properties of in - situ stars in the halo of mw1hr .
as explained in section 2 , we focus on this particular simulated halo because of the extensive numerical and resolution testing which it has undergone .
we find that the in - situ stars of gal1 , h277 , and h285 formed and populated their stellar halos by the same qualitative process described below for mw1hr . in order to understand where the in - situ stars in the halo of mw1hr originated , we studied the gas particles that formed this stellar population .
the vast majority of the gas progenitors were brought into the primary galaxy in smooth gas flows .
fewer than @xmath46 of the in - situ stars were formed from gas stripped from accreted subhalos .
cold gas flow was found to also be the dominant contributor to the disk stars of mw1hr @xcite . to investigate how in - situ stars came to form part of the stellar halo of mw1hr , we first study the locations at which these stars formed relative to the center of the galaxy .
figure 2 shows the radial distribution of in - situ stars at the time of their formation relative to the center of the primary ( black solid line ) .
this shows that in - situ stars form near the very center of mw1hr s potential well .
approximately @xmath47 of the in - situ stars were formed by z @xmath48 . between @xmath49 ,
mw1hr experienced at least 3 significant mergers with mass ratios @xmath50 , defined at the time of virial radius crossing .
substantial mergers disrupt the orbits of stars in both the primary and merging galaxy . as two galaxies of approximately equal mass collide ,
their potentials change rapidly , causing the energies of their stars to change as well .
the violent relaxation process will cause some stars to gain energy , and alter their initial orbits .
we find that by z=2 ( red dotted line in figure 2 ) , the distribution of in - situ stars is no longer sharply peaked at the inner 1 kpc of the galaxy , but is instead peaked at 5 kpc from the galactic center .
this transformation from centrally located orbits to halo orbits was most likely due to the major mergers which mw1hr experienced between @xmath51 .
after z=2 , mw1hr does not undergo any more mergers with such small mass ratios , and so we expect that the radial distribution of in - situ stars will not change .
indeed , we find that their distribution at the present day ( blue dashed line in figure 2 ) is very similar to that at z=2 .
the origin of the halo in - situ stars in cold gas flows , and their formation in the innermost regions of mw1hr s potential well all highlight that this population of stars formed in a rapid inflow and collapse of gas that also formed the disk of mw1hr , but were displaced through merger events into the halo .
the process by which these insitu stars formed is reflected in that their relative importance in each stellar halo rises toward the center of each simulated galaxy .
much work has been done recently with dark matter only n - body simulations coupled with semi - analytic models to study the growth of milky way - like galactic halos via pure accretion of stars . in this section ,
we compare results from such numerical studies with our findings for mw1hr s accreted halo .
we focus here on mw1hr as it was chosen to have a total mass , and merger history similar to that of the milky way @xcite , making it a good candidate to compare with previous works which have also focused on simulating milky way - like stellar halos .
we find that the accreted halo of mw1hr displays properties which are typical in comparison with other numerical studies . as shown in the top left panel of figure 1 , accreted stars dominate the halo of mw1hr at all radii . while the innermost 20 kpc of the halo contains both accreted and in - situ stars , the halo beyond this
is highly dominated by accretion .
the earliest accreted stars , which became bound to the primary halo of mw1hr at a lookback time of @xmath52 gyr or more , dominate the accreted population of the halo out to 30 kpc from the center . in the halo s outer regions , however , the majority of stars were accreted more recently .
the accretion history of mw1hr s stellar halo agrees well with the numerical study of @xcite .
these authors found that @xmath53 of stars are in place in the inner 10 kpc of a milky way - like halo @xmath54 gyr ago , while the mass growth of the outer regions is more continuous .
the work of @xcite also found that nearly @xmath55 of the stars in the inner 20 kpc of simulated halos were in place at a lookback time of 9 gyr . @xcite
have shown in their numerical work that the inner regions of halos are typically assembled from a few satellites with stellar masses @xmath56 . in mw1hr , the majority ( @xmath57 by mass in stars ) of the accreted mass in the inner 50 kpc originated in two subhalos , each with total mass @xmath58 .
the accreted stars which reside in the halos of gal1 and h285 differ significantly from those found in the stellar halo of mw1hr .
unlike the accreted population of mw1hr , the majority of the accreted stars in the halos of gal1 and h285 became bound to their respective stellar halo more recently , as shown in figure 1 .
this is likely due to the fact that both gal1 and h285 have had a more active recent merger history , as described in the following section .
the accreted stellar halo of h277 was assembled at similar times to that of mw1hr , although the accreted population of this stellar halo contributes far less to the overall mass of the halo in comparison to mw1hr .
lcccc|c mw1hr & & 70 & 21 & 9 & + & @xmath59 & @xmath60 & @xmath61 & @xmath62 & @xmath63 + mw1med & & 66 & 25 & 9 & + & @xmath64 & @xmath65&@xmath66 & @xmath67 & @xmath68 + mw1thermal & & 85 & 12 & 3 & + & @xmath69&@xmath70&@xmath71&@xmath72&@xmath73 + gal1 & & 85 & 12 & 3 & + & @xmath74 & @xmath75 & @xmath76 & @xmath77 & @xmath78 + h277 & & 32 & 58 & 10 & + & @xmath79&@xmath80&@xmath81&@xmath82&@xmath83 + h285 & & 87 & 7 & 6 & + & @xmath84&@xmath85&@xmath86&@xmath87&@xmath88 + despite the universal presence of both in - situ and accreted halo stars in all of the simulations , the relative contributions of these dual components varies widely among the four stellar halos .
the fractional contributions of these stellar populations to the simulated stellar halos , as well as their total masses , are summarized in table 2 .
the fraction of accreted halo stars ranges from @xmath89 , while the fraction of in - situ halo stars ranges from less than 10@xmath90 to more than 50@xmath90 .
the high and low ends of these ranges do not sum to @xmath91 because of the small presence of the ambigious stars .
although the fractional contribution of in - situ stars covers a wide range , the total mass of these stars in all four simulations is substantial .
the mass of in - situ stars in both the stellar halos of h277 and gal1 is @xmath92 , and @xmath93 for mw1hr and h285 .
all four of these simulated galaxies have similar total masses , however they span a range of merger histories .
three of the galaxies , mw1hr , h277 , and h285 each host a stellar disk with a mass of @xmath94 , while the mass of gal1 s stellar disk is @xmath95
. the total halo mass of h277 and mw1hr is @xmath96 , and @xmath97 for gal1 and h285 .
the bulge to disk ratio ( b / d ) of each simulated galaxy was obtained using a two component sersic @xmath37 exponential fit to the one dimensional radial profile of the i band surface brightness maps made with the sunrise software package @xcite .
the b / d for mw1hr , gal1 , h277 , and h285 are 0.37 , 0.66 , 0.29 , and 1.2 , respectively ( brooks et al , in prep ) .
what role might merger history play in the range of accreted / in - situ fractions observed in these four simulations ?
there is not an obvious trend between epoch of last major merger and either in - situ fraction or average assembly time of the accreted component ; gal1 and mw1 had their last 3:1 merger at @xmath98 and h277 and h285 at @xmath99
. however , there appears to be a possible trend with the galaxies recent merger histories .
figure [ fig_radial ] illustrates one manifestation of this .
the accreted stellar halo components of mw1 and h277 , the two galaxies with the highest in - situ fraction , were largely assembled more than 9 gyr ago , with the innermost regions of these halos containing very few stars accreted in the last 9 gyr .
however , gal1 and h285 , which contain few in - situ halo stars , have accreted stellar halos that , on average , were assembled much more recently at all radii , with at least @xmath100 accreted within the last 9 gyr .
it is important to recall that we have defined accretion time as the time at which a star becomes unbound from its subhalo , and bound to the primary dark matter halo , which is a different , and invariably later , time than when the subhalo entered the primary s virial radius . in order to more explicitly study the merger histories of the simulated galaxies
, we traced the total number of luminous satellites that enter the virial radius at different redshifts .
in particular , we are interested in those satellites whose masses are in the range @xmath101 at the time of merging , as these are expected to be the building blocks of the accreted component of stellar halos @xcite .
table 3 shows the number of such merged satellites at redshifts smaller than @xmath102 .
the stellar halos of gal1 and h285 , the two galaxies with more recently assembled stellar halos , have continued to accrete a large number of satellites after z=2 .
the masses of these galaxies accreted stellar halos are comparatively large ( @xmath103 ) , overwhelming the population of halo in - situ stars .
in contrast , mw1hr and h277 , whose stellar halos were found to have assembled at more ancient times , have had a more quiet recent merger history .
these galaxies merged with fewer satellites in the studied mass and redshift range , and have the less massive accreted halos .
mw1hr and h277 also both have halos which host a comparatively larger in - situ population than gal1 and h285 .
lccccc mw1hr & 2 & 5 + gal1 & 2 & 10 + h277 & 1 & 2 + h285 & 7 & 8 +
owing to a dual origin , the detailed properties of a stellar halo are a function of both i ) the properties of the primary galaxy itself , and ii ) the properties of the protogalaxies that merged to form its accreted component .
accurate modeling of a galaxy s stellar halo thus requires an accurate treatment of the formation and evolution of that galaxy over a wide dynamical range . in this section
we discuss the effects of feedback , as well as numerical and resolution effects , on the overall formation of stellar halos in our simulated galaxies .
the four simulated galaxies used in this study have sizes , ages , and metallicities that match those observed at low redshift .
g07 and @xcite showed that disk galaxies naturally form in a cosmological context in simulations such as those studied here , and have sizes that place them on the i - band tully fisher relation ( tf , e.g. @xcite ) .
@xcite used a set of gasoline simulations ( including mw1 ) to show that galaxies with total @xmath104 0 halo masses in the range @xmath105 have ( o / h ) abundances that match both those measured by @xcite for more than 53,000 sdss galaxies locally , and those measured by erb et al.(2006 ) at @xmath104 2 .
@xcite demonstrated that galaxies simulated with gasoline ( including the progenitors of h277 and h285 ) are also a good match to the observed mass metallicity relationship at @xmath104 3 .
the success of these simulations in matching the observed mass metallicity relationship at varying redshifts @xcite demonstrates that these simulations have made substantial progress in overcoming the historic `` overcooling problem '' ( e.g. * ? ? ?
gas in simulations has traditionally cooled rapidly , forming stars quickly and early , and thus producing a mass metallicity relation that is too enriched at a given stellar mass , particularly at high @xmath106 . in the simulations examined here ,
supernova energy is deposited into the ism mimicking the blast - wave phase of a supernova ( following the sedov - taylor solution , details in @xcite ) .
cooling is turned off for gas particles within the supernova blastwave radius for a period of time after the supernova event .
supernova feedback regulates star formation efficiency as a function of halo mass , resulting in the mass metallicity relationship described above .
this regulation of star formation also leads to realistic trends in gas fractions , with our lowest mass galaxies being the most gas rich @xcite , and reproducing the observed incidence rate and column densities of damped lyman @xmath107 systems at @xmath104 3 @xcite .
feedback , with its role in regulating star formation , will also effect the luminosity function of galaxies in a simulation , which will in turn effect the properties of the accreted stellar halo .
for example , if the simulated dwarfs have too many stars relative to dwarf galaxies in the real universe , then we may overestimate the relative importance of accreted versus in - situ stars in the halos .
we therefore compare the luminosity function of satellites within the virial radius of our two medium resolution galaxies , mw1med and mw1thermal .
we use these two simulated runs because they have the same mass resolution , but different implemented feedback mechanisms .
mw1med adopts the blastwave model discussed above , and used for all of the high resolution runs studied here . in mw1thermal ,
supernova energy was deposited as thermal energy to the surrounding gas particles but cooling is not turned off in these particles .
these gas particles will then quickly radiate away the deposited energy , and this simulation will suffer from the overcooling problem . using the age and metallicity of each stellar particle in a satellite , we calculated the absolute v band magnitudes of the z = 0 satellites of mw1med and mw1thermal using the stellar population models of starburst99 @xcite . figure 3 shows the luminosity function of the satellites in mw1med and mw1thermal , as well as the observed luminosity function of milky way dwarfs , from @xcite .
the satellites of mw1med are similar in total number and luminosity to the milky way s , except at the bright end of the luminosity function
. the satellites of mw1thermal , however , are too numerous at all luminosities .
the discrepancy between the luminosity function of mw1med and mw1thermal is due to the overcooling problem typical of simulations with thermal feedback , which causes satellites to form which are highly centrally concentrated and contain too many stars . because the mw1thermal run produces over - luminous dwarf galaxies , we expect that its halo will be more dominated by accreted stars than the halo of mw1med . as shown by table 2 , which lists the accreted and in - situ fractions of both of these galaxies , the halo of the mw1thermal has @xmath108 more accreted stars in its halo relative to in - situ stars than mw1med ( @xmath109 vs @xmath110 ) and @xmath111 1/2 of the fraction of in - situ stars ( @xmath112 vs @xmath113 ) .
while the total mass of in - situ halo stars is about the same for both simulated runs ( @xmath114 ) , the total mass of accreted stars in the thermal run is @xmath115 , and only @xmath116 in the run with blast - wave feedback . in summary ,
several features of the simulations we analyze here contribute to bringing the number of luminous satellite galaxies into good agreement with those observed around the mw : our uniform uv background unbinds baryons from simulated halos with total masses below @xmath117 , creating `` dark '' halos and bringing down the number of luminous satellite galaxies .
our supernova feedback scheme helps to significantly reduce the number of stars produced in the remaining low mass satellite galaxies .
however , because our satellites are still slightly too bright ( ie , have too many stars ) , our accreted fractions are an upper limit .
this only serves to increase the potential relative contribution of in - situ stars in the inner halos of galaxies .
@xcite found that several thousand particles were required inside of the virial radius of a halo in order for the star formation history to converge .
for the high resolution simulations studied here , this convergence criterion is met in halos with total masses above a few @xmath118 .
for the medium resolution runs , the convergence limit is met in halos with total masses above @xmath119 .
resolution testing has shown that poorly resolved halos have smaller total stellar masses than halos which are fully resolved .
we therefore underestimate the contribution to the primary stellar halos from dwarf galaxies with total mass below this convergence limit .
however , we find that dwarf galaxies less massive than @xmath120 contribute less than 5% to the overall mass in accreted stars in the high resolution stellar halos . due to the resolution effects discussed above , this small fraction of accreted satellites will contribute even fewer stars at lower resolution than they would if they were fully resolved , decreasing their overall contribution to the accreted stellar halo of the medium resolution runs ( mw1med and mw1thermal ) .
this effect is minimal due to the tiny contribution of these halos .
our findings agree with the work of @xcite , who showed that satellites with total mass greater than @xmath121 contribute @xmath122 of the mass to the primary stellar halo .
because the in - situ stars are forming deep in the potential well of the high resolution main halo , we do not expect their formation properties to vary with resolution because the star formation history of the main halo is fully resolved . in order to test this assumption , we compared the in - situ formation in the high resolution mw1hr run with the lower resolution mw1med run .
we expect that if the formation of these stars was caused by a numerical resolution effect , then their fractional contribution to the simulated halos would change as the resolution at which the galaxy is simulated changed .
we find , however , that at both resolutions , in - situ stars make up a substantial fraction of the stellar halo of the milky way massed galaxies .
table 2 shows that the overall in - situ fraction of the stellar halos of mw1hr and mw1med are @xmath123 and @xmath113 , repectively .
the @xmath124 larger contribution of in - situ stars to the stellar halo of mw1med is likely due to the slight underestimating of accreted stars in this medium resolution run , as described above .
in this paper we have analyzed four sph + n - body simulations of approximately @xmath125 disk galaxies to investigate the formation of their stellar halos .
we have shown that the inner halos of all four galaxies contain both stars accreted from merged satellites , as well as in - situ stars formed within the primary galaxy . while these results are theoretical , they are backed by recent observational studies of the stellar halos of the milky way and m31 .
observational work has shown that the properties of milky way halo stars exhibit signatures of a possible double population . @xcite
have found that a sample of local halo stars in sdss can be thought of as two distinct , but overlapping , components , with different spatial distributions , orbits and metallicities .
they attribute , as others have , these different stellar properties to different formation mechanisms for the two halo components .
they conclude that the inner halo may have formed dissipatively from the gas rich merger of early sub - galactic halos , while the outer halo stars were accreted from subhalos which merged with the milky way s dark matter halo
. observations of m31 have revealed that this giant spiral galaxy is also surrounded by a large extended stellar population .
m31 has been shown to have a distinct stellar halo with a surface brightness profile similar to the milky way s , and a metallicity gradient with the inner regions of its halo ( r@xmath126 kpc ) more metal rich than the outer regions @xcite . using number count maps ,
@xcite find a large metal poor smooth component to the halo of m31 , with a more metal rich inner halo .
these works highlight the possible signatures that m31 s stellar halo may , like the milky way , have two underlying components , with different formation histories .
the results of such observational findings for the halos of the milky way and m31 , however , have previously lacked counterparts in simulations as most of the numerical studies done to date on stellar halos have concentrated on their build up through pure accretions . our work begins to bridge the gap by studying self - consistent simulations where the contribution of in - situ stars to the overall halo have been taken into account .
our main results are : in - situ stars are a generic feature of kinematically defined stellar halos .
these stars formed deep in the potential well of the primary galaxy and were later displaced to halo orbits as a result of mergers .
in - situ stars reside primarily in the inner few tens of kpc of the galaxies halos .
stars accreted off satellites dominate the stellar halo at all radii for three of the four galaxies studied .
the accreted stellar halo components were found to extend to out to a few hundred kpc .
the fractional contribution of in - situ stars to the simulated galaxies halos varies greatly , from @xmath127 .
the two galaxies in our sample whose halos had the smallest contributions from in - situ stars were also those with accreted halos that assembled less than 9 gyr ago .
this is likely because such galaxies have more active recent accretion histories , which lead to massive accreted halos .
the large contribution of recently accreted stars overwhelm the underlying in - situ populations in these halos .
the other two galaxies in our sample , with large contributions from in - situ stars , had a significant fraction of their accreted stars in place at times earlier than 9 gyr ago .
the relative importance of in - situ stars in these halos is due to the recent quiet merger history of these galaxies , which have led to less massive accreted halos .
different supernova feedbacks greatly impact the formation of simulated stellar halos .
this underscores the necessity of carefully implementing feedback to produce a realistic satellite luminosity function , and to hence conduct realistic studies of halo formation . the overall importance of in - situ stars to the stellar halos found in this study is likely a lower limit .
this is because i ) we have ignored a fraction of in - situ stars for numerical reasons , and ii ) the satellites of the simulated galaxies studied here are brighter than those observed in the milky way .
the calculated contributions of accreted stars to the stellar halos are thus an upper - limit .
the qualitative trends of our results are unchanged if we had instead used a spatially defined set of stellar halos .
our results suggest that galaxies with more active recent mergers , like m31 , may host halos where the relative importance of an in - situ population is low .
galaxies , such as the milky way , with little evidence for recent merger events , on the other hand , may have a larger relative contribution from an in - situ population in its inner regions , along with its accreted stars .
however , a larger set of simulations with a wide range in merger histories is needed to fully understand the complex relationship between halo formation and mergers .
vast observational resources are currently dedicated to the study of the milky way s and m31 s halos .
spatial , kinematic and chemical abundance information of halo stars being collected and used to attempt to reconstruct the accretion and formation history of the galaxies .
our results highlight that such data sets , particularly ones focused on local , and hence inner , halo stars , will contain a significant population of stars born in - situ in the galaxies .
these in - situ stars may dilute the signatures of accretion history in observations . given the short dynamical timescales in the inner halo
, one may not expect to be able to discriminate these two populations using their kinematics alone .
however , given the very different sites of their formation , the chemical abundances of in - situ and accreted stars may be quite distinct . in a future paper , we will investigate possible ways to observationally disentangle the two halo populations .
we thank the anonymous referee for suggestions which helped to extensively clarify the paper .
az thanks the institute of theory & computation at harvard - smithsonian s center for astrophysics for its hospitality during part of this work .
bw and az thank the smithsonian astrophysical observatory and the clay fellowship for partial financial support .
we thank joe cammisa at haverford for computing support .
az was partially supported by new york university s horizon fellowship .
ab acknowledges support from the sherman fairchild postdoctoral program in theoretical astrophysics .
dwh was partially supported by nasa grant nnx08aj48 g .
simulations were run at arsc , nasa ames and texas supercomputing center .
fg acknowledges support from a theodore dunham grant , hst go-1125 , nsf itr grant phy-0205413 ( also supporting tq ) , nsf grant ast-0607819 and nasa atp nnx08ag84 g .
cbb acknowledges the support of the uk s science & technology facilities council ( st / f002432/1 ) .
, e. f. , zucker , d. b. , belokurov , v. , sharma , s. , johnston , k. v. , bullock , j. s. , hogg , d. w. , jahnke , k. , de jong , j. t. a. , beers , t. c. , evans , n. w. , grebel , e. k. , ivezi , . ,
koposov , s. e. , rix , h .- w . ,
schneider , d. p. , steinmetz , m. , & zolotov , a. 2008 , , 680 , 295 , v. , zucker , d. b. , evans , n. w. , gilmore , g. , vidrih , s. , bramich , d. m. , newberg , h. j. , wyse , r. f. g. , irwin , m. j. , fellhauer , m. , hewett , p. c. , walton , n. a. , wilkinson , m. i. , cole , n. , yanny , b. , rockosi , c. m. , beers , t. c. , bell , e. f. , brinkmann , j. , ivezi , . , & lupton , r. 2006 , , 642 , l137 , r. , nagao , t. , grazian , a. , cocchia , f. , marconi , a. , mannucci , f. , cimatti , a. , pipino , a. , ballero , s. , calura , f. , chiappini , c. , fontana , a. , granato , g. l. , matteucci , f. , pastorini , g. , pentericci , l. , risaliti , g. , salvati , m. , & silva , l. 2008 , , 488 , 463 , h. j. , yanny , b. , rockosi , c. , grebel , e. k. , rix , h .- w . , brinkmann , j. , csabai , i. , hennessy , g. , hindsley , r. b. , ibata , r. , ivezi , z. , lamb , d. , nash , e. t. , odenkirchen , m. , rave , h. a. , schneider , d. p. , smith , j. a. , stolte , a. , & york , d. g. 2002 , , 569 , 245 , b. , newberg , h. j. , grebel , e. k. , kent , s. , odenkirchen , m. , rockosi , c. m. , schlegel , d. , subbarao , m. , brinkmann , j. , fukugita , m. , ivezic , . , lamb , d. q. , schneider , d. p. , & york , d. g. 2003 , , 588 , 824 | we investigate the formation of the stellar halos of four simulated disk galaxies using high resolution , cosmological sph + n - body simulations .
these simulations include a self - consistent treatment of all the major physical processes involved in galaxy formation .
the simulated galaxies presented here each have a total mass of @xmath0 , but span a range of merger histories .
these simulations allow us to study the competing importance of in - situ star formation ( stars formed in the primary galaxy ) and accretion of stars from subhalos in the building of stellar halos in a @xmath1cdm universe .
all four simulated galaxies are surrounded by a stellar halo , whose inner regions ( r@xmath2 kpc ) contain both accreted stars , and an in - situ stellar population .
the outer regions of the galaxies halos were assembled through pure accretion and disruption of satellites .
most of the in - situ halo stars formed at high redshift out of smoothly accreted cold gas in the inner 1 kpc of the galaxies potential wells , possibly as part of their primordial disks .
these stars were displaced from their central locations into the halos through a succession of major mergers .
we find that the two galaxies with recently quiescent merger histories have a higher fraction of in - situ stars ( @xmath3 ) in their inner halos than the two galaxies with many recent mergers ( @xmath4 in - situ fraction ) .
observational studies concentrating on stellar populations in the inner halo of the milky way will be the most affected by the presence of in - situ stars with halo kinematics , as we find that their existence in the inner few tens of kpc is a generic feature of galaxy formation . |
a self - shuffling word , a notion which was recently introduced by charlier et al .
@xcite , is an infinite word that can be reproduced by shuffling it with itself .
more formally , given infinite words @xmath4 over a finite alphabet @xmath5 we define @xmath6 to be the collection of all infinite words @xmath7 for which there exists a factorization @xmath8 with each @xmath9 and with @xmath10 , @xmath11 .
an infinite word @xmath12 is _ self - shuffling _ if @xmath13 .
various well - known words , e.g. , the thue - morse word or the fibonacci word , were shown to be self - shuffling .
harju @xcite studied shuffles of both finite and infinite square - free words , i.e. , words that have no factor of the form @xmath14 for some non - empty factor @xmath15 .
more results on square - free shuffles were obtained independently by harju and mller @xcite , and currie and saari @xcite .
however , the question about the existence of an infinite square - free self - shuffling word , posed in @xcite , remained open .
we give a positive answer to this question in sections 2 and 3 .
the _ shift orbit closure _ @xmath16 of an infinite word @xmath0 can be defined , e.g. , as the set of infinite words whose sets of factors are contained in the set of factors of @xmath0 . in @xcite it has been proved that each word has a non - self - shuffling word in its shift orbit closure , and the following question has been asked : does there exist a word for which no element of its shift orbit closure is self - shuffling ( question 7.2 ) ? in section 4 we provide a positive answer to the question . more generally , we show the existence of a word such that for any three words @xmath17 in its shift orbit closure , if @xmath18 is a shuffle of @xmath19 and @xmath7 , then the three words are pairwise different . on the other hand , we show that for any infinite word there exist three different words @xmath17 in its shift orbit closure such that @xmath20 ( see proposition [ prop_xyz ] ) .
apart from the usual concepts in combinatorics on words , which can be found for instance in the book of lothaire @xcite , we make use of the following notations : for every @xmath21 , we denote the alphabet @xmath22 by @xmath23 . for a word @xmath24
we say that @xmath15 is a _ prefix _ of @xmath0 , @xmath25 is a _ factor _ of @xmath0 , and @xmath7 is a _ suffix _ of @xmath0 .
we denote these prefix- and suffix relations by @xmath26 and @xmath27 , respectively . by @xmath28
$ ] we denote the factor of @xmath0 starting at position @xmath29 and ending after position @xmath30 .
note that we start numbering the positions with @xmath31 .
a _ prefix code _ is a set of words with the property that none of its elements is a prefix of another element .
similarly , a _ suffix code _ is a set of words where no element is a suffix of another one .
bifix code _ is a set that is both a prefix code and a suffix code .
a morphism @xmath32 is _ square - free _
if for all square - free words @xmath0 , the image @xmath33 is square - free .
let @xmath34 be the morphism defined as follows : @xmath35 we will show that the fixed point @xmath36 is square - free and self - shuffling in the following . note that @xmath37 is not a square - free morphism , that is , it does not preserve square - freeness , as @xmath38 contains the square @xmath39 .
[ no_3u1u3 ] the word @xmath36 contains no factor of the form @xmath40 for some @xmath41 .
we assume that there exists a factor of the form @xmath40 in @xmath0 , for some word @xmath41 . from the definition of @xmath37
, we observe that @xmath15 can not be empty .
furthermore , we see that every @xmath42 in @xmath0 is preceded by either @xmath31 or @xmath43 . if @xmath44 , then we had an occurrence of the factor @xmath45 in @xmath0 , which is not possible by the definition of @xmath37 , hence @xmath46 .
now , every @xmath42 is followed by either @xmath31 or @xmath47 in @xmath0 and @xmath48 is followed by either @xmath47 or @xmath42 .
since both @xmath49 and @xmath50 are factors of @xmath0 , we must have @xmath51 .
this means that the factor @xmath52 appears at the center of @xmath53 , which can only be followed by @xmath43 in @xmath0 , thus @xmath54 .
however , this results in the factor @xmath55 as a prefix of @xmath40 , which does not appear in @xmath0 , as seen from the definition of @xmath37 .
the word @xmath36 is square - free .
we first observe that @xmath56 is a bifix code .
furthermore , we can verify that there are no squares @xmath14 with @xmath57 in @xmath0 .
let us assume now , that the square @xmath14 appears in @xmath0 and that @xmath15 is the shortest word with this property .
if @xmath58 , then @xmath59 must hold , since @xmath60 appears only as a factor of @xmath61 , and thus @xmath14 is a suffix of the factor @xmath62 in @xmath0 . as @xmath63 , also the shorter square @xmath64 appears in @xmath0 , a contradiction .
the same desubstitution principle also leads to occurrences of shorter squares in @xmath0 if @xmath65 and @xmath66 . if @xmath67 then either @xmath68 or @xmath69 or @xmath70 , by the definition of @xmath37 . in the last case , that is when @xmath70 , we must have @xmath54 , which is covered by the previous paragraph .
if @xmath71 , then @xmath14 is followed by @xmath47 in @xmath0 and we can desubstitute to obtain the shorter square @xmath72 in @xmath0 . if @xmath67 and @xmath59 , and @xmath14 is preceded by @xmath73 or followed by @xmath47 in @xmath0 , we can desubstitute to @xmath74 or @xmath75 , respectively .
therefore , assume that @xmath76 and @xmath14 is preceded by @xmath77 and followed by @xmath60 in @xmath0 .
this however means that we can desubstitute to get an occurrence of the factor @xmath78 in @xmath0 , a contradiction to lemma [ no_3u1u3 ] .
we now show that @xmath36 can be written as @xmath79 with @xmath80 .
the word @xmath36 is self - shuffling . in what follows we use the notation
@xmath81 meaning that @xmath82 for finite words @xmath83 .
we are going to show that the self - shuffle is given by the following : @xmath84 now we verify that @xmath85 from which it follows that @xmath0 is self - shuffling .
it suffices to show that each of the above products is fixed by @xmath37 .
indeed , straightforward computations show that @xmath86 which is fixed by @xmath37 : @xmath87 hence @xmath88 is fixed by @xmath37 and thus @xmath89 . in a similar way
we show that @xmath90 .
we remark that we can immediately produce a square - free self - shuffling word over @xmath91 from @xmath92 : charlier et al .
@xcite noticed that the property of being self - shuffling is preserved by the application of a morphism .
furthermore , brandenburg @xcite showed that the morphism @xmath93 , defined by @xmath94 is square - free .
therefore , the word @xmath95 is a ternary square - free self - shuffling word , from which we can produce a multitude of others by applying square - free morphisms from @xmath96 to @xmath96 .
in this section we provide a positive answer to the question from @xcite whether there exists a word for which no element of its shift orbit closure is self - shuffling . the _ hall word _ @xmath97 is defined as the fixed point of the morphism @xmath98 .
sometimes it is referred to as a _
ternary thue - morse _ word .
it is well known that this word is square - free .
we show that no word in the shift orbit closure @xmath99 of the hall word is self - shuffling .
more generally , we show that if @xmath18 is a shuffle of @xmath19 and @xmath7 for @xmath100 , then they are pairwise different .
[ prop_xyy ] there are no words @xmath101 in the shift orbit closure of the hall word such that @xmath102 .
suppose the converse , i.e. , there exist words @xmath103 such that @xmath104 define the set @xmath105 of infinite words as follows : @xmath106 in other words , @xmath105 consists of words in @xmath99 which can be introduced as a shuffle of some word @xmath19 in @xmath99 with itself .
now suppose , for the sake of contradiction , that @xmath105 is non empty , and consider @xmath107 with the first block @xmath108 of the smallest possible positive length .
we remark that such @xmath18 and corresponding @xmath19 are not necessarily unique .
we can suppose without loss of generality that @xmath19 starts with @xmath31 or @xmath109 . otherwise , we exchange @xmath31 and @xmath47 , consider the morphism @xmath110 , and the argument is symmetric .
it is not hard to see from the properties of the morphism @xmath32 that removing every occurrence of @xmath43 from @xmath19 results in @xmath111 . hence the blocks in the factorizations of @xmath19 after removal of @xmath43 are of the form @xmath112 for some integer @xmath29 .
thus the first letter of each block @xmath2 and @xmath3 that is different from @xmath43 is @xmath31 , and the last letter different from @xmath43 is @xmath47
. then , @xmath2 and @xmath3 are images by the morphism @xmath32 of factors of the fixed point of @xmath32
. therefore , there are words @xmath113 such that @xmath114 and @xmath115 , @xmath116 .
notice that the first block @xmath108 can not be equal to @xmath43 . indeed , otherwise @xmath18 starts with @xmath45 , which is impossible , since @xmath45 is not a factor of the fixed point of @xmath32 .
clearly , taking the preimage decreases the lengths of blocks in the factorization ( except for those equal to @xmath43 ) , and since @xmath117 , the length of the first block in the preimage is smaller , i.e. , @xmath118 .
this is a contradiction with the minimality of @xmath119 .
there are no self - shuffling words in the shift orbit closure of @xmath120 . with a similar argument we can prove the following : [ prop_xxy ] there are no words @xmath101 in the shift orbit closure of @xmath120 such that @xmath121 .
first we introduce a notation @xmath122 , meaning that there exists a shuffle starting with the word @xmath7 ( i.e. , @xmath123 , @xmath124 ) .
next , @xmath125 implies that there exists @xmath7 in the same shift orbit closure such that @xmath126 .
indeed , one can remove the prefix @xmath108 of @xmath18 to get @xmath7 , i.e. , @xmath127 , and keep all the other blocks @xmath2 , @xmath3 in the shuffle product . define the set @xmath128 of infinite words as follows : @xmath129 in other words , @xmath128 consists of words in @xmath99 which can be introduced as a shuffle of some word @xmath19 in @xmath99 with @xmath7 starting with the block @xmath130 . now consider @xmath131 with the first block @xmath130 of the smallest possible length .
we remark that such @xmath7 and a corresponding @xmath19 are not necessarily unique . as in the proof of proposition [ prop_xyy ] ,
the shuffle can not start with a block of length @xmath43 .
again , if we remove every occurrence of @xmath43 in @xmath19 ( and in @xmath7 ) , we get @xmath111 or @xmath132 ; moreover , since @xmath130 contains letters different from @xmath43 , the first letter different from @xmath43 is the same in @xmath19 and @xmath7 .
so , without loss of generality we assume that both @xmath19 and @xmath7 without @xmath43 are @xmath111 , and the blocks @xmath2 and @xmath3 without @xmath43 are integer powers of @xmath60
. then , @xmath2 and @xmath3 are images by the morphism @xmath32 of factors of @xmath120
. therefore , there are words @xmath133 such that @xmath134 and @xmath135 , @xmath136 ( i.e. , @xmath137 ) . as in the proof of proposition [ prop_xyy ] , since @xmath138 , the length of the first block in the preimage is smaller , i.e. , @xmath139 .
this is again a contradiction with the minimality of @xmath140 .
so , we proved that if there are three words @xmath141 in the shift orbit closure of the fixed point of @xmath32 such that @xmath142 , then they should be pairwise distinct .
now we are going to prove that for any infinite word there exist three different words in its shift orbit closure such that @xmath20 .
an infinite word @xmath18 is called _ recurrent _ , if each its prefix occurs infinitely many times in it .
[ prop_xyz ] let @xmath18 be a recurrent infinite word .
then there exist two words @xmath143 in the shift orbit closure of @xmath18 such that @xmath144 .
we build the shuffle inductively .
start from any prefix @xmath108 of @xmath18 . since @xmath18 is recurrent , each of its prefixes occurs infinitely many times in it .
find another occurrence of @xmath108 in @xmath18 and denote its position by @xmath145 .
put @xmath146 $ ] . at step
@xmath147 , suppose that the shuffle of the prefix of @xmath18 is built : @xmath148=\prod_{l=0}^{k-1 } u_l v_l , \quad y[0 , \sigma_{l=0}^{k-1}|u_l|-1]=\prod_{l=0}^{k-1 } u_l , \quad z[0 , \sigma_{l=0}^{k-1 } |v_l|-1]=\prod_{l=0}^{k-1 } v_l,\ ] ] such that @xmath149 is the suffix of @xmath150=\prod_{i=0}^{k-1 } u_l v_l$ ] starting at position @xmath151 , and @xmath152 is the suffix of @xmath150=\prod_{i=0}^{k-1 } u_l v_l$ ] starting at position @xmath153 .
find another occurrence of @xmath152 in @xmath18 at some position @xmath154 .
we can do it since @xmath18 is recurrent .
put @xmath155 $ ] .
we note that @xmath156 is a factor of @xmath18 by the construction ; more precisely , it occurs at position @xmath151 .
find an occurrence of @xmath156 at some position @xmath157 , put @xmath158 is a factor of @xmath18 by the construction since it occurs at position @xmath153 .
moreover , both @xmath156 and @xmath159 are suffixes of @xmath160=\prod_{i=0}^{k } u_l v_l$ ] .
continuing this line of reasoning , we build the required factorization .
since each infinite word contains a recurrent ( actually , even a uniformly recurrent ) word in its shift orbit closure , we obtain the following corollary : [ col : xyz ] each infinite word @xmath0 contains words @xmath141 in its shift orbit closure such that @xmath20 . the following example shows that the recurrence condition in proposition [ prop_xyz ] can not be omitted : consider the word @xmath161 which is obtained from @xmath120 by adding a letter @xmath42 in the beginning .
then the shift orbit closure of @xmath162 consists of the shift orbit closure of @xmath120 and the word @xmath162 itself .
assuming @xmath162 is a shuffle of two words in its shift orbit closure , one of them is @xmath162 ( there are no other @xmath42 s ) and the other one is something in the shift orbit closure of @xmath120 , we let @xmath19 denote this other word . clearly , the shuffle starts with @xmath162 , and cutting the first letter @xmath42 , we get @xmath163 , a contradiction with proposition [ prop_xxy ] .
there also exist examples where each letter occurs infinitely many times : the following word : @xmath164 does not have two words @xmath165 in its shift orbit closure such that @xmath142 .
the idea of the proof is that the shift orbit closure consists of words of the following form : @xmath166 , @xmath167 , @xmath18 itself and all their right shifts .
shuffling any two words of those types , it is not hard to see that there exists a prefix of the shuffle which contains too many or too few occurrences of some letter compare to the prefix of @xmath18 .
we leave the details of the proof to the reader . by corollary [ col : xyz ] , there are @xmath17 in the shift orbit closure of @xmath120 such that @xmath168 .
to conclude this section , we give an explicit construction of two words in the shift orbit closure of @xmath120 which can be shuffled to give @xmath120 .
we remark though that this construction gives a shuffle different from the one given by corollary [ col : xyz ] .
let : @xmath169 by definition , the shift orbit closure of the hall word is closed under @xmath32 .
moreover this shift orbit closure is also closed under @xmath170 , since the factors of the hall word are closed under the morphism @xmath171 .
@xmath172 note that if @xmath0 is an infinite word , then @xmath173 and @xmath174 .
@xmath175 .
let @xmath176 and for every @xmath177 , @xmath178 let furthermore @xmath179 we show that @xmath180 , @xmath181 and @xmath182 .
note that @xmath183 , thus @xmath184 .
then we have @xmath185 moreover , @xmath186 since @xmath187 , the word @xmath188 is the fixed point @xmath189 of @xmath190 , and then @xmath191 is the fixed point @xmath192 .
thus @xmath193 . finally : @xmath194 applying the morphism @xmath32 to the second expression for @xmath0 , we get @xmath195 thus @xmath180 since @xmath32 is injective .
we showed that infinite square - free self - shuffling words exist .
the natural question that arises now is whether we can find infinite self - shuffling words subject to even stronger avoidability constraints : for this we recall the notion of _ repetition threshold _ @xmath196 , which is defined as the least real number such that an infinite word over @xmath23 exists , that does not contain repetitions of exponent greater than @xmath196 .
due to the collective effort of many researchers ( see @xcite and references therein ) , the repetition threshold for all alphabet sizes is known and characterized as follows : @xmath197 a word @xmath198 without factors of exponent greater than @xmath196 is called a _ dejean word_. charlier et al . showed that the thue - morse word , which is a binary dejean word , is self - shuffling @xcite . | in this paper we answer two recent questions from @xcite and @xcite about self - shuffling words .
an infinite word @xmath0 is called self - shuffling , if @xmath1 for some finite words @xmath2 , @xmath3 .
harju @xcite recently asked whether square - free self - shuffling words exist .
we answer this question affirmatively . besides that
, we build an infinite word such that no word in its shift orbit closure is self - shuffling , answering positively a question from @xcite .
* keywords : * infinite words , shuffling , square - free words , shift orbit closure , self - shuffling words |
a calculation of the partition function exactly in the thermodynamic limit is equivalent to knowledge of all equilibrium properties of a given system .
this task is usually beyond our horizon , except for some selected cases which we call `` trivial '' .
the ising model seems to be at the border of our possibilities since the beginning of the 20-th century . even for a finite system of @xmath6
spins our computational ability is limited by the number of the system configurations , which is @xmath7 .
here we describe a new algorithm , designed to improve the speed of the calculation of the partition function for a finite ising system . in the ising model
@xcite only two spin states are possible , say `` up '' and `` down '' ( @xmath8 ) .
the energy @xmath9 of a given spin configuration may be expressed in terms of the number @xmath10 of spins pointing `` up '' ( say @xmath11 ) and the number @xmath12 of anti - parallel bonds between the nearest neighbors @xmath13 : @xmath14 where @xmath15 denotes the exchange integral , @xmath16 is an external magnetic field , and @xmath17 the linear size of the lattice .
the partition function can be written as @xmath18,\ ] ] where @xmath19 is the number of lattice configurations with given numbers @xmath10 and @xmath12 , @xmath20 , @xmath21 is a boltzmann constant and @xmath22 is temperature .
then , the average value of any quantity @xmath23 may be calculated as @xmath24.\ ] ] the magnetic susceptibility per spin @xmath25 may be also expressed in the terms of @xmath10 , @xmath12 and @xmath17 : @xmath26=\beta[\langle ( 2n - l^2)^2\rangle-\langle 2n - l^2\rangle^2],\ ] ] where the index @xmath12 enters through the averaging procedure .
a computation of the partition function requires an evaluation of the histogram @xmath19 . tab .
[ tab_cpu_time ] shows the cpu time needed to check the numbers @xmath10 and @xmath12 of @xmath27 configurations of a @xmath1 large lattice on sgi 2800 machine .
a rough estimation of the cpu time necessary for full investigation of all of @xmath28 possible lattice configurations gives the value larger than four millions years what makes traditional / direct method practically useless . .
the cpu time necessary for investigation of @xmath27 different configurations of @xmath1 large 2d ising lattice on sgi 2800 machine . [ cols= " < ,
< , < , < , < , < , < " , ]
an effective way of generation @xmath29 is a successive merging of smaller lattices and their histograms , namely @xmath30 and @xmath31 .
however , the procedure requires storing information on the @xmath32 dependence not only on @xmath10 and @xmath12 , but also on @xmath33 the state of the @xmath34-sites - long lattice border . in fig .
[ fig_net](a ) a @xmath35-long bit - string equivalent to @xmath36 large array ( see fig .
[ fig_net](b ) ) is presented . for @xmath37
the bit - string corresponds to an integer number from the interval @xmath38 .
this correspondence allows to investigate all possible lattice configurations in a simple manner .
dark sites in fig .
[ fig_net](a ) correspond to the dark border of the lattice in fig .
[ fig_net](b ) , and they can be represented by an integer number @xmath39 .
the first four bits of @xmath40 ( marked as dark sites in fig .
[ fig_net](c ) ) allow to determine the additional number of anti - parallel bonds , created by merging two @xmath41 lattices together to get a @xmath42 lattice ( see fig . [
fig_net](f ) ) .
on the other hand , the last four digits of @xmath40 ( dark sites in fig .
[ fig_net](d ) ) are equivalent to the part of the border @xmath43 ( the dark sites in fig .
[ fig_net](g ) ) of @xmath42 lattice . in the next step ,
two @xmath44 lattices are merged to create the desired @xmath1 large lattice , as presented in fig . [ fig_net](h ) .
\(a ) described in sec .
[ sec_algorithm].,title="fig : " ] + ( b ) described in sec .
[ sec_algorithm].,title="fig : " ] ( c ) described in sec .
[ sec_algorithm].,title="fig : " ] ( d ) described in sec .
[ sec_algorithm].,title="fig : " ] + ( e ) described in sec . [ sec_algorithm].,title="fig : " ] ( f ) described in sec .
[ sec_algorithm].,title="fig : " ] ( g ) described in sec .
[ sec_algorithm].,title="fig : " ] + ( h ) described in sec .
[ sec_algorithm].,title="fig : " ] histograms ( @xmath45 and @xmath46 ) for the lattices larger than @xmath41 may also be easily computed basing on @xmath47 : @xmath48 where @xmath49 is the additional number of anti - parallel bonds in the darker part of fig . [ fig_net](f ) and @xmath50 is combined from @xmath51 and @xmath52 . the histogram @xmath29 may be constructed in a similar way : @xmath53 and again @xmath54 is the number of anti - parallel bonds in the darker part of fig . [ fig_net](h ) . this procedure on sgi 2800 machine takes _ only _ 22 hours of the machine time instead of a few million years in case of the usage of the traditional / direct method .
successive merging may be repeated recursively to obtain the partition function for larger lattices . to evaluate the results obtained for the @xmath1 lattice , we show also the data obtained by the monte carlo simulations , the results of the mean field model , and last but not least the experimental data .
standard monte carlo metropolis algorithm @xcite is applied to determine the magnetic spin susceptibility of a @xmath55 ising lattice . after getting equilibrium , each point of the plot
is obtained as the time average over a thousand of time steps .
the mean field model is a direct generalization of the case of a ferromagnet .
namely , we solve numerically a set of two equations for two sublattices @xmath56 and @xmath57 : @xmath58 where @xmath59 . in this model ,
the curie temperature @xmath60 and the susceptibility @xmath25 is found as @xmath61 for a small value of applied magnetic field , e.g. @xmath62 .
the experimental data are collected from ref .
they concern two - dimensional ising antiferromagnets rb@xmath63cof@xmath64 and k@xmath63cof@xmath64 where @xmath0 .
the van vleck susceptibility is subtracted to obtain a pure spin contribution .
in fig . [ fig_omega ] the histogram @xmath29 obtained by means of the method described in sec . [ sec_algorithm ] is presented .
the numbers are available at our web page @xcite . the symbols on @xmath10-@xmath12 plane ( see fig .
[ fig_omega](a ) ) indicate @xmath65 pairs with non - zero values of @xmath29 . in fig .
[ fig_omega](b ) the number of configuration @xmath29 is presented . for fixed @xmath10 different symbols
correspond to different @xmath12 . in fig .
[ fig_chi ] we show the thermal dependence of the antiferromagnetic ( @xmath59 ) susceptibility , defined in eq . .
the curie temperature is assumed to be at the inflexion point of the curve , which is at the half of the peak height . below the curie temperature , the values of @xmath66 obtained with our new algorithm fit well the results obtained with the monte carlo results and the experimental data . above @xmath67 ,
the agreement is only qualititative .
we hope that it can be also quantitative if periodic boundary conditions are applied . however , in this case the lattice border is longer , and so is the computation time . still , even with the present method the qualitative accordance is better than the results of the mean field theory .
it would be of interest to apply the finite size scaling to our results to evaluate the critical properties of the system .
then , the results could be compared with other methods , e.g. the finite size scaling renormalization group @xcite or the phenomenological renormalization @xcite . for this purpose , however , periodic boundary conditions seem to be more appropriate as a starting point of the calculations .
the function @xmath68 , once known , can be easily used for the calculation of all equilibrium thermodynamic properties , for ferro- and antiferromagnets , various values of temperature and magnetic field .
the summation over @xmath10 and @xmath12 is much faster , than the summation over @xmath69 spin configurations .
the results can be relevant also for other applications of the ising model , e.g. for the percolation problem .
as for our knowledge , the partition function has never been calculated exactly for the lattice as large as @xmath1 . in principle , the algorithm can be applied to larger lattices , with a cost of more time and memory .
* acknowledgements .
* the authors are grateful to dietrich stauffer for valuable comments .
the simulations were carried out in ack - cyfronet - agh .
the machine time on sgi 2800 is financed by the polish state committee for scientific research ( kbn ) with grant no .
kbn / sgi2800/agh/077/2002 .
reichl , _ a modern course in statistical physics _ , j. wiley ( new york , 1998 ) .
heermann , _
computer simulation methods in theoretical physics _ , springer - verlag ( berlin , 1990 ) .
breed , k. gilijamse and a.r .
miedema , physica * 45 * ( 1969 ) 205 .
de oliveira , europhys .
* 20 * ( 1992 ) 621 .
nightingale , physica * a83 * ( 1976 ) 561 . | a new and efficient algorithm is presented for the calculation of the partition function in the @xmath0 ising model . as an example , we use the algorithm to obtain the thermal dependence of the magnetic spin susceptibility of an ising antiferromagnet for a @xmath1 square lattice with open boundary conditions .
the results agree qualitatively with the prediction of the monte carlo simulations and with experimental data and they are better than the mean field approach results . for the @xmath1 lattice , the algorithm reduces the computation time by nine orders of magnitude . * new algorithm for the computation of the partition function for the ising model on a square lattice * + _
k. malarz@xmath2 , m.s .
magdo - maksymowicz@xmath3 , a.z .
maksymowicz@xmath4 , b. kawecka - magiera@xmath4 and k. kuakowski@xmath4 _ @xmath4 department of applied computer science , faculty of physics and nuclear techniques , university of mining and metallurgy ( agh ) + al .
mickiewicza 30 , pl-30059 krakw , poland + @xmath3 department of mathematical statistics , agriculture university + al .
mickiewicza 21 , pl-31120 krakw , poland e - mail : @[email protected] + march 17 , 2003 _ keywords : _ antiferromagnets ; ising model ; magnetic susceptibility ; mean field approximation ; world records . |
listening to `` folklore versions '' of quantum mechanics one may consider it as a key statement of quantum theory that there is no measurement without disturbing the measured system .
however , the fact that this is not true is well - known in modern quantum information theory and is , in some sense , the reason why classical information exists at all although our world is quantum . consider a two - level system , i.e. , a quantum system with hilbert space @xmath3 and denote its upper or lower state by @xmath4 and @xmath5 , respectively .
assume that we know by prior information that the system is not in a quantum superposition but only in one of the two states @xmath6 .
then the measurement with projections @xmath7 and @xmath8 show which state is present without disturbing it at all . here
we have used the two - level system as _ classical _ bit .
the situation changes if the two - level system is used as _
quantum _ bit ( `` qubit '' ) and is prepared in a quantum superposition @xmath9 where the complex coefficients @xmath10 and @xmath11 with @xmath12 are unknown to the person who measures
. then any von neumann measurement with projections @xmath13 and @xmath14 will on the one hand only provide some information about @xmath15 and will on the other hand disturb the unknown state @xmath15 since it `` collapses '' to the state @xmath16 with probability @xmath17 and to the orthogonal state @xmath18 with probability @xmath19 . from the point of view of the person who has prepared the state @xmath15 and
does not notice the measured result ( `` non - selective operation '' ) , the measurement process changes the density matrix of the system from @xmath20 to @xmath21 i.e. , the measurement causes an entropy increase of @xmath22 .
the general condition under which information about unknown quantum states can be gained without disturbing them is well - known and reads as follows @xcite : let @xmath23 be the unknown density matrix of a system . by prior information
one knows that @xmath23 is an element of a set @xmath24 of possible states .
then one can get some information on @xmath23 if and only if there is a projection @xmath25 commuting with all matrices in @xmath24 such that the value @xmath26 is not the same for all @xmath27 . as noted in @xcite
, this can never be the case if the set @xmath24 is the orbit @xmath28 of a hamiltonian system evolving according to @xmath29 .
this holds even if @xmath23 and @xmath30 act on an infinite dimensional hilbert space . in this sense ,
timing information is always to some extent quantum information that can not be read out without state disturbance .
it can only become classical information if either ( 1 ) prior information tells us that the time @xmath31 is an element of some discrete set @xmath32 ( see @xcite ) or ( 2 ) in the limit of infinite system energy @xcite . at first sight
the statement that classical timing information can only exist in one of these two cases seems to be disproved by the following dissipative `` quantum clock '' : let @xmath33 be the upper state of a two - level system .
let the system s time evolution for positive @xmath31 be described by the bloch relaxation ( see e.g. @xcite ) @xmath34 since all the states @xmath35 commute with the projections @xmath36 and @xmath37 one can certainly gain some information about @xmath31 by the measurement with projections @xmath38
. however , this situation is actually the infinite energy limit since semi - group dynamics of this form is generated by coupling the system to a heat bath of infinite size and infinite energy spectrum @xcite .
of course the fact that well - known derivations of relaxation dynamics require heat baths with infinite energy spectrum does not prove our claim that this is necessarily the case .
this claim is rather an implication of theorem [ mainth ] in section [ main ] .
the paper is organized as follows . in section
[ main ] we show quantitatively to what extent information on the ( non - discrete ) time is always quantum information as long as the clock is a system with limited energy bandwidth @xmath0 .
explicitly , we prove a lower bound on the entropy increase in the clock caused by von - neumann measurements , i.e. , measurements that are described by an orthogonal family of projections where the state change of the system is described by the projection postulate .
generalized measurement procedures are considered in section [ gen ] .
they do not necessarily increase the entropy of the measured clock since the process can include some kind of cooling mechanisms but will lead to an increase of entropy of the total system which includes the clock s environment .
this has implications for low power computation since it shows that the distribution of timing information inherent in a microscopic clock produces necessarily some phenomenological entropy . in section [ tight ]
we discuss whether our bound is tight . in section [ dec ]
we shall show that the result of section [ gen ] can be applied to the situation that a clock with limited energy controls the switching process of a classical bit . here a classical bit is understood as a two state quantum system on which decoherence takes place on a time scale that is in the same order as the switching time or on a smaller scale . for low power computation
, this proves which amount of dissipation is required whenever the autonomous dynamics of a microscopic device controls a classical output . in section [ group ]
we apply the bound to other one - parameter groups .
for the moment we will restrict our attention to von - neumann measurements .
for technical reasons we will assume them to have a finite set of possible outcomes .
hence the measurement is described by a family @xmath39 of mutual orthogonal projections on the system s hilbert space @xmath40 that may be infinite dimensional .
the state of the system is described by a density matrix , i.e. , a positive operator with trace @xmath41 acting on @xmath40 . according to the projection postulate any state @xmath42 changes to the post - measurement state @xmath43 if the _ unselected _ state is considered , i.e. , the outcome is ignored .
the post - measurement state @xmath44 coincides with @xmath42 if and only if @xmath42 commutes with all projections @xmath45 .
now we will compare the von - neumann entropy of @xmath42 and @xmath44 .
for any state @xmath42 the von - neumann entropy is defined as @xmath46 the following lemma shows that the measurement can never decrease the entropy : [ kulllemma ] let @xmath42 be an arbitrary density matrix on a hilbert space and @xmath39 be a family of orthogonal projections with @xmath47 .
set @xmath48 then we have @xmath49 where @xmath50 is the kullback - leibler distance ( or relative entropy ) between @xmath44 and @xmath42 .
it is defined by @xcite @xmath51 proof : according to the definition of entropy it is sufficient to show the equation @xmath52 .
the second term on the right in eq . ( [ kull ] ) is equal to @xmath53 note that @xmath54 commutes with each @xmath55 .
hence we get @xmath56 this completes the proof .
@xmath57 since the entropy increase is the kullback - leibler distance between the pre- and the post - measurement state we can obtain a lower bound on @xmath58 in terms of the trace - norm distance between them : [ delta ] for the entropy increase @xmath58 we obtain the lower bound @xmath59 where @xmath60 is the trace - norm of an arbitrary matrix @xmath61 . the proof is immediate using @xmath62 ( see @xcite ) .
now we consider the states @xmath35 on the orbit of the time evolution and show that the norm distance between @xmath35 and @xmath63 is large at every moment where the outcome probabilities of the measurement @xmath39 change quickly , i.e. , where the values @xmath64 \rho_t)$ ] are large .
this is the essential statement that is used to prove the information - disturbance trade - off relation .
but first we have to introduce the energy bandwidth of a system . for hamiltonians with discrete eigenvalues
it is just the difference between the greatest and smallest eigenvalues of the energy states with non - zero occupation probability .
the generalization to hamiltonians with continuous and discrete parts in the spectrum reads : let @xmath30 be the hamiltonian of a quantum system and @xmath65 be the spectral family corresponding to @xmath30 , i.e. , @xmath66 projects onto the hilbert space corresponding to energy values not greater than @xmath67 . then @xmath68 is the distribution function of a ( time - independent ) probability measure @xmath69 which is called the spectral measure of @xmath35 corresponding to @xmath30 .
let @xmath70 $ ] be the smallest interval supporting this spectral measure .
then @xmath71 is the energy bandwidth of the system in the state @xmath35 . by rescaling the hamiltonian it is easy to see that the time evolution of the state @xmath23 can equivalently be described by @xmath72 since @xmath73 .
clearly , @xmath74 .
the energy spread is decisive for our lower bound on the trace - norm distance between @xmath35 and @xmath63 .
[ distancegeschw ]
let @xmath75 be the probability of the measurement outcome @xmath76 at time @xmath31 .
let @xmath0 be the energy bandwidth of @xmath35 .
set @xmath77 then we have @xmath78 proof : for a specific moment @xmath31 define the operator @xmath79 by @xmath80 where @xmath81 if @xmath82 and @xmath83 if @xmath84 .
note that we have @xmath85 \rho_t)\,.\ ] ] furthermore easy computation shows that @xmath86 \tilde{\rho}_t)=0 $ ] .
we conclude @xmath87 ( \rho_t-\tilde{\rho}_t ) ) \leq 2\|h\| \,\|\rho_t-\tilde{\rho}_t\|_1\,,\ ] ] since the operator norm of @xmath79 is @xmath41 .
we assume @xmath88 without loss of generality and have @xmath89 @xmath57 now we are able to prove our main theorem .
[ mainth ] let the energy bandwidth @xmath0 of a quantum system be less than @xmath90 .
assume that the true time @xmath31 is in the interval @xmath91 $ ] for arbitrary @xmath92 .
let the prior probability for @xmath31 be the uniform distribution on @xmath91 $ ] .
let @xmath39 be the family of projections corresponding to a von - neumann measurement .
assume that the measurement has the time resolution @xmath1 in the following sense : for each @xmath93 $ ] there is a decision rule based on the measurement outcome that decides whether the state @xmath35 or @xmath94 is present with error probability at most @xmath95 .
then the mean entropy increase @xmath96 ( averaged over the interval @xmath91 $ ] ) caused by the measurement is at least @xmath97 note that our definition of time resolution is not the usual one since it does not require that the measurement distinguishes between @xmath35 and @xmath98 , for instance .
however , it is exactly the definition of time resolution that we need in the proof .
proof : each decision rule distinguishing between @xmath35 and @xmath94 that is based on the measurement outcome is of the following form : if the outcome is @xmath76 it decides for @xmath31 with probability @xmath99 .
if the true state is @xmath35 it decides for @xmath31 with probability @xmath100 and if the true state is @xmath94 it decides erroneously for @xmath31 with probability @xmath101 the difference between the probability to decide correctly for @xmath31 and to decide erroneously for @xmath31 is at least @xmath102 by assumption .
hence @xmath103 note that @xmath104 is in the interval @xmath105 $ ] for all @xmath76 .
therefore @xmath106 obviously we have @xmath107 since @xmath108 .
we conclude @xmath109 hence @xmath110 therefore we find that the average of @xmath111 over the interval @xmath112 $ ] is at least @xmath113 .
since this holds for every @xmath93 $ ] the average of @xmath114 over the whole interval @xmath91 $ ] is at least @xmath113 .
we denote this average by @xmath115 for the average entropy generation we have @xmath116 by lemma [ delta ] .
note that the average of the square is at least the square of the average .
we conclude @xmath117 due to lemma [ distancegeschw ]
we have @xmath118 as long as we measure the energy spread in units of @xmath119 . in physics , it is more common to use si - units which lead to @xmath120 @xmath57
the assumption that every measurement is described by projections @xmath39 and that the effect of the measurement is given by the projection postulate @xmath121 is not general enough . in general
, a measurement can be described by a positive operator valued measure ( povm ) ( see @xcite for the most general description of measurements ) .
if the set of possible outcomes is finite , a povm is defined by a family @xmath122 of positive operators with @xmath123 such that @xmath124 is the probability for the outcome @xmath76 .
furthermore the connection between the povm and the effect on the state is given by the following consistency condition : there exist completely positive maps @xmath125 such that @xmath126 and that , given the outcome @xmath76 , the post - measurement state is @xmath127 .
the unselected post - measurement state is hence given by @xmath128 clearly , in this general setting it is even possible that the measurement can decrease the entropy of the clock since the maps @xmath125 may include processes that are cooling mechanisms for the clock and transport entropy to the environment .
however , we will show that each time measurement leads unavoidably to an entropy increase in the total system consisting of the measurement apparatus and the clock .
let @xmath129 be the hilbert space of the measurement apparatus and be @xmath130 be the hamiltonian its free time evolution as long as no measurement interaction is active .
we assume the state @xmath42 of the apparatus to be invariant under its free evolution .
otherwise the measurement apparatus would be a clock in its own since it contained some information about the time @xmath31 .
then we switch on the measurement interaction which leads to a unitary transformation @xmath131 on the space @xmath132 ( the definition of the apparatus includes its environment such that the total evolution is unitary ) .
this process is often called a _ pre - measurement_. the different outcomes @xmath76 correspond to mutual orthogonal subspaces of @xmath129 ( `` pointer states '' ) .
let @xmath133 be the projections onto these subspaces .
now we assume that decoherence on the measurement apparatus pointer takes place @xcite .
the process given by the unitary pre - measurement process @xmath131 and the decoherence is described by @xmath134 the entropy of this state is clearly the same as the entropy of @xmath135 by defining @xmath136 we have reduced the problem to the problem in section [ main ] with the only generalization that @xmath39 is not a family of projections acting on the hilbert space @xmath40 of the clock but on the hilbert space of clock + measurement apparatus where the apparatus is at rest .
now we show that the stationarity of the apparatus state ensures that the bound in theorem [ mainth ] holds for the considered kind of generalized measurement as well .
formally , we claim : [ mainthgen ] let @xmath137 be a density matrix on a hilbert space @xmath138 .
let the time evolution of the composed system be given by the hamiltonian @xmath139 with @xmath140=0 $ ] .
let @xmath39 be a von - neumann measurement acting on the hilbert space @xmath138 .
let @xmath0 be the energy bandwidth of the left component of the composed system ( given by the state @xmath23 with hamiltonian @xmath30 ) .
let the measurement have the time resolution @xmath1 in the sense of theorem [ mainth ] .
then we have the following lower bound on the average entropy increase @xmath141 of the composed system caused by the measurement : @xmath142 proof : clearly , the hamiltonian @xmath130 is irrelevant for the evolution of the state @xmath143 due to @xmath144 hence we can treat the evolution as if it was implemented by the hamiltonian @xmath145 , which can assumed to be bounded as in section [ main ] .
@xmath57 . note that our arguments do not require that the time evolution @xmath131 takes place on a smaller time scale than @xmath1 .
the reason is that we have argued that @xmath146 may formally be considered as projections of a measurement performed _ before _ the interaction was switched on . our results in section [ main ] and [ gen ]
may be given an additional interpretation as information - disturbance trade - off relation .
lemma [ kulllemma ] shows that the entropy increase caused by the measurement is the kullback - leibler distance between pre- and post - measurement state .
information - disturbance trade - off relations are an important part of quantum information theory @xcite .
unfortunately , theorem [ mainth ] is restricted to von - neumann measurements .
theorem [ mainthgen ] extends the statement to general measurements as far as the disturbance of the total state @xmath147 of the `` clock '' and the ancilla system is considered .
however , this is interesting from the thermodynamical point of view taken in this article but not in the setting of information - disturbance trade - off . in the latter
setting , the disturbance of the ancilla state that is used to implement a non - von - neumann measurement is not of interest .
therfore we admit , that our bound on the state disturbance does only hold for von - neumann measurements .
the entropy increase predicted by our results can never be greater than @xmath102 .
this can be seen by the following argument . using the definitions of the proof of lemma [ distancegeschw ] we have @xmath148\rho_t ) \leq 2\|h\|\,\|r(t)\| = \delta e \|r(t)\|\\&= & \delta e\,.\end{aligned}\ ] ] note also the connection to the heisenberg uncertainty principle @xmath149 where @xmath150 is the energy spread of the system , i.e. , the standard deviation of the energy values and @xmath151 the standard deviation of the time estimation @xcite based on any measurement .
however , using the symbols @xmath152 instead of @xmath153 , we emphasize that these definitions do not agree with ours .
note that the energy spread @xmath154 is _ at most _ the energy bandwidth @xmath0 .
but @xmath151 can exceed our time resolution @xmath1 by an arbitrary large value .
this shows the following example : assume the clock to be a two - level system with period @xmath155 .
set @xmath156 with @xmath157 given the prior information that the time is in an interval @xmath91 $ ] with @xmath158 we can only estimate the time up to multiples of @xmath155 and obtain an error in the order @xmath159 .
in contrast , the time resolution @xmath1 in our sense is @xmath160 since we can certainly distinguish between the state at the time @xmath31 and at @xmath161 .
however , we conjecture that similar bounds as in theorem [ mainth ] on the entropy generation can be given for systems where the energy spectrum is _ essentially _ supported by an interval @xmath162 $ ] . therefore our results suggest the following interpretation : every measurement that allows to estimate the time up to an error that is not far away from the heisenberg limit produces a non - negligible amount of entropy .
the following example suggests that the necessary entropy generation may even be much above our lower bound if `` time measurements '' are used that are extremely close to the heisenberg limit .
consider the wave function of a free schrdinger particle moving on the real line .
a natural way to measure the time would be to measure its position .
we may use , for instance , von - neumann measurements that correspond to a partition of the real line into intervals of length @xmath163 .
assume one wants to improve the time accuracy by decreasing @xmath163 arbitrarily .
the advantage for the time estimation is small if @xmath163 is smaller than the actual position uncertainty of the particle .
however , if the state is pure it is easy to see that the generated entropy goes to infinity for @xmath164 .
the following example shows that the entropy generation in a time measurement can really go to zero when @xmath0 goes to infinity .
consider the hilbert space @xmath165 of square sumable functions over the set of integers .
let @xmath166 with @xmath167 be the canonical basis vectors .
let the hamiltonian be the diagonal operator @xmath168 consider the state @xmath169 by fourier transformation @xmath40 is isomorphic to the set @xmath170 of square integrable functions on the unit circle @xmath24 and the dynamical evolution is the cyclic shift on @xmath24 .
the period of the dynamics is @xmath171 . in this picture , @xmath15 is a wave package that has its maximum at the angle @xmath172 with an uncertainty @xmath173 in the order @xmath174 .
we assume to know that the true time is in the interval @xmath175 and construct a measurement that allows to decide for each @xmath176 whether the true time is @xmath31 or @xmath177 with confidence that is increasing with @xmath178 .
we use a measurement with projections @xmath179 projecting on the four sectors of the unit circle .
if the angle uncertainty is considerably smaller than @xmath180 this measurement can clearly distinguish between @xmath31 and @xmath181 with high confidence .
hence , for @xmath178 large enough , we get @xmath182 as time resolution of the measurement .
note that we have only non - negligible entropy generation at that moments where the main part of the wave packet crosses the border between the sectors , i.e. , when there are two sectors containing a non - negligible part of the wave function .
if , for instance , each sector contains about one half of the probability , we generate the entropy one bit , i.e. , the entropy @xmath183 in natural units . the probability that the measurement is performed at a time in which non - negligible entropy generation takes place is about @xmath184 .
the average entropy generation decreases therefore for increasing @xmath178 , i.e. , for increasing @xmath185 .
let @xmath186 for @xmath187 be the probabilities for the @xmath188 possible outcomes when the wave packet with energy spread @xmath189 is measured at the time instant @xmath31 .
there are four moments where the maximum of the wave packet is exactly on the border of the sectors .
the probability to meet these times is zero .
for all the other times @xmath31 there is one @xmath76 such that @xmath190 tends to zero with @xmath191 by standard fourier analysis arguments .
note that a measurement at time @xmath31 produces the entropy @xmath192 we conclude by elementary analysis that @xmath58 tends to zero with @xmath193 for all times @xmath31 except from @xmath188 irrelevant values @xmath31 and therefore the average entropy generation tends to zero with @xmath193 .
note that the decrease of the lower bound of theorem [ mainth ] is asymptotically a little bit faster since it is @xmath191 ( due to @xmath185 ) .
now we consider the system consisting of the clock , its environment and the classical bit . for the moment
we ignore the fact that the bit is quantum and claim that the two logical states @xmath194 and @xmath41 correspond to an orthogonal decomposition @xmath195 of subspaces of the hilbert space of the composed system . if @xmath196 and @xmath197 are non - isomorphic as hilbert spaces we extent the smaller space such that they are isomorphic .
then we can assume without loss of generality @xmath40 to be of the form @xmath198 without assuming that the bit is physically realized by a two - level quantum system .
the fact that the bit is classical will now be described by the fact that it is subjected to decoherence , i.e. , that all superpositions between @xmath5 and @xmath4 are destroyed and changed to mixtures on a time scale that is not larger than the switching time .
decoherence keeps the diagonal values of the density matrix whereas its non - diagonal entries decay .
if the process is a uniform time evolution , i.e. , given by a semi - group dynamics , decoherence of the two - level system is given by an exponential decay of both non - diagonal entries : @xmath199 where @xmath200 are the coefficients of the density matrix .
note that it is assumed that the effect of the environment does not cause any bit - flips in the system but only destroys coherence .
the parameter @xmath201 defines the decoherence rate .
let @xmath36 and @xmath37 be the projections onto the states @xmath202 and @xmath4 , respectively .
the decoherence process can be simulated by measurement processes that are performed at randomly chosen time instants .
this analogy is explicitly given as follows : let @xmath203 be the effect of the measurement that distinguishes the two logical states .
then we have @xmath204 the second expression provides the following intuitive approximation of the process : in each small time interval of length @xmath205 a measurement is performed with probability @xmath206 .
let @xmath207 be the extension of @xmath208 to the total system .
we assume that the dynamical evolution of the total system is generated by @xmath209 + ( g - id)\,,\ ] ] i.e. , the decoherence of the bit is the only contact of the system to its environment . define the switching time as the length of the time interval @xmath210 $ ] where the probability of one of the logical states changes from @xmath95 to @xmath211 .
the reason that we do not assume it to switch between @xmath194 and @xmath41 is that this is impossible within a finite time interval with a hamiltonian that has limited spectrum .
this can be seen by the fact that all expectation values with hamiltonians of lower ( ! ) bounded spectrum are analytical functions @xcite .
consider the case that the decoherence time is small compared to the switching time @xmath1 , i.e. , @xmath212 .
then the probability to have more than one measurement during the switching process is small ( of second order in @xmath213 ) and the probability that one measurement occurs is about @xmath214 .
given that it occurred , the probabilities of its time @xmath215 of occurrence is equally distributed in @xmath210 $ ] .
since entropy is convex the entropy generated by the switching process is at least @xmath214 times the entropy that is generated if a measurement has occurred during the switching process . also by convexity arguments ,
we conclude that the entropy generated by the process `` perform a measurement at a randomly chosen ( unknown ) time instant in @xmath210 $ ] '' is at least the average entropy generation if the time instant is known .
the latter situation meets exactly the assumptions of theorem [ mainth ] with the following parameters : set @xmath216 , i.e. , assume that the prior information is that the true time is in @xmath217 . furthermore the switching time @xmath1 coincides with the time resolution @xmath1 of theorem [ mainth ] since reading out the logical state can distinguish between the times @xmath194 and @xmath1 with error probability at most @xmath95 . taking into account that a measurement occurs only with probability @xmath214 we multiply the bound of theorem [ mainth ] by this factor and obtain : @xmath218 note that we do not speak about _ average _ entropy generation since we consider ( in contrast to theorem [ mainth ] ) the entropy of a single density matrix which is already an average over a set of density matrices which are obtained when the measurement occurs at different times .
consider now the case that the decoherence time is so small that more than one measurement during the switching process is likely .
then we have to worry about the post - measurement state and whether its energy spectrum is still bounded . if not , we can not apply our arguments .
if it is , we can use the bound ( [ coh ] ) nevertheless .
this is less obvious than it may seem at first sight .
the total dynamics is assumed to be the semi - group evolution @xmath219 with @xmath220 as in eq .
( [ fdef ] ) .
now we investigate the amount of entropy generated by a measurement at time @xmath31 .
note that we do not have a lower bound by theorem [ mainth ] since the dynamical evolution leading to the state @xmath35 was not hamiltonian .
however , the bound @xmath221 holds nevertheless .
this is seen by the observation that the generator @xmath222 is irrelevant if the proof of lemma [ distancegeschw ] should be converted to the situation here .
with @xmath223 we have @xmath224 ) \ ] ] due to @xmath225 for every state @xmath42 .
it should be emphasized that this argument would fail if the two - level system was not only subjected to decoherence but also to relaxation .
if @xmath226 is the generator of a relaxation process that causes directly transitions from the state @xmath4 to the state @xmath5 the equation @xmath227 does not hold , i.e. , @xmath226 would not be irrelevant for our proof .
this is consistent with the observation in section [ in ] that the `` relaxation clock '' can be read out without generating entropy _ by the measurement_. we conclude that the bound ( [ coh ] ) holds for arbitrarily high decoherence rate .
formally , the bound predicts infinite entropy production if @xmath1 and @xmath0 are constant and @xmath228 .
however , this asymptotics is meaningless due to the quantum zeno effect ( see e.g. @xcite ) .
infinite decoherence would stop the switching process completely .
consider a short time @xmath31 after a measurement has been performed on the state .
then the derivative of the probabilities @xmath229 corresponding to a second measurement is only of the order @xmath230 .
the essential consequence is that in particular for small @xmath0 when our bound would predict large entropy generation fast switching processes are impossible . in order to show how to apply our bound to realistic situations consider the following example : a light signal ( guided by an optical fiber , for instance ) is sent to an apparatus that contains a two - level system .
the incoming signal triggers the transition @xmath231 from the lower to the upper level ( see fig.1 ) .
the apparatus may be a huge physical system and we are only interested in the fact that the _ transition time _ is given by the incoming light signal , the energy may be provided by the apparatus itself .
assume the light signal consists of photons with frequency bandwidth @xmath232 .
the signal is assumed to contain at most @xmath178 photons . than the energy bandwidth of
the signal is at most @xmath233 .
for @xmath234 the time evolution of the total system is evolving approximatively as @xmath235 where @xmath42 is the ( stationary ) state of the apparatus .
when the signal meets the apparatus the time evolution is given by an unknown hamiltonian @xmath236 of the total system combined with the decoherence of the two - level system .
the energy bandwidth of the total system according to @xmath236 is unknown .
the `` measurement '' on the two - level system takes definitely place at a moment where the total evolution of the system is generated by the unknown hamiltonian @xmath236 .
nevertheless we can apply the bound of theorem [ mainthgen ] . in section [ gen ]
we have argued that theorem [ mainthgen ] does not assume that the measurement interaction was only switched on during a time interval that is small compared to the time resolution @xmath1 .
we observed that the measurement can formally be considered as a measurement that had been performed _ before _
the interaction had been switched on .
analogously , we argue that the unknown interaction between light field and apparatus is irrelevant : let @xmath237 be the state of the total system at the time @xmath31 .
let the switching process happen during the interval @xmath210 $ ] .
let @xmath238 be the time evolution of the total system implemented by a possibly unbounded hamiltonian .
the entropy generated by `` measurements '' @xmath45 performed on the state @xmath237 during this period due to the decoherence is the same as produced by @xmath239 performed on the state @xmath240 .
for @xmath241 the family of states @xmath242}$ ] evolve approximatively like the free evolution @xmath243 of the light field .
we conclude that the bound in eq .
( [ coh ] ) can be applied and the relevant energy bandwidth is the energy bandwith of the free light field .
the entropy generation is at least @xmath244 statements of this kind may be relevant in future computer technology when miniaturization reduces signal energies on the one hand and requires on the other hand reduction of power consumption .
our results show that reduction of signal energy down to the limit of the energy - time uncertainty principle leads unavoidably to heat generation as long as the signal control _
classical _ bits .
the entropy generation @xmath58 leads to an energy loss of @xmath245 ( where @xmath178 is boltzmann s constant and @xmath159 is the absolute temperature ) due to the second law of thermodynamics .
obviously , our results generalize to other one - parameter groups since the proofs do not rely on the interpretation of the unitary group @xmath246 as the system s _ time _ evolution .
consider for instance the case that the momentum of a schrdinger wave package is restricted to the interval @xmath247 $ ] .
let @xmath23 be the particle s density matrix and @xmath248 the state translated by @xmath249 .
then we conclude that every measurement that is suitable to distinguish between @xmath248 and @xmath250 in the sense of theorem [ mainth ] produces at least the entropy @xmath251 .
note that the measurement is not necessarily a position measurement , it has not even to be compatible with the position operator . in this sense
, our results may be interpreted as a kind of `` generalized uncertainty relation '' .
another very natural application is to consider the group of rotations around a specific axis .
consider a spin - k/2 particle and the group of rotations @xmath252 on its hilbert space @xmath253 , where @xmath254 is the operator of its angular momentum in @xmath255-direction .
we have @xmath256 and conclude the following : each measurement that distinguishes between @xmath257 and @xmath258 with error probability at most @xmath95 produces at least the entropy @xmath259 .
our bound on the entropy that is generated when information about the actual time is extracted from quantum system holds only for hamiltonian time evolution .
a simple counterexample in section [ in ] has shown that time readout without state disturbance is possible for some dissipative semi - group dynamics .
this leads to an interesting question : in physical systems , each dissipative semi - group dynamics that is induced by weak coupling to a reservoir in thermal equilibrium is unavoidably accompanied by some loss of free energy .
this shows that the loss of free energy caused by the time measurement can only be avoided by systems that loose energy during its autonomous evolution
. it would be desirable to know whether there is a general lower bound ( including dissipative dynamics ) on the total amount of free energy that is lost as soon as timing information is converted to classical information .
thanks to thomas decker for helpful comments . this work has been supported by grants of the dfg project `` komplexitt und energie '' of the `` schwerpunktprogramm verlustmarme informationsverarbeitung '' be 887/12 . | we consider hamiltonian quantum systems with energy bandwidth @xmath0 and show that each measurement that determines the time up to an error @xmath1 generates at least the entropy @xmath2 .
our result describes quantitatively to what extent all timing information is quantum information in systems with limited energy .
it provides a lower bound on the dissipated energy when timing information of microscopic systems is converted to classical information .
this is relevant for low power computation since it shows the amount of heat generated whenever a band limited signal controls a classical bit switch .
our result provides a general bound on the information - disturbance trade - off for von - neumann measurements that distinguish states on the orbits of continuous unitary one - parameter groups with bounded spectrum .
in contrast , information gain without disturbance is possible for some completely positive semi - groups .
this shows that readout of timing information can be possible without entropy generation if the autonomous dynamical evolution of the `` clock '' is dissipative itself . 2 |
the apache point observatory galactic evolution experiment ( apogee ) is one of 4 experiments that are part of the sloan digital sky survey iii ( sdss - iii ; eisenstein et al .
apogee is obtaining high - resolution ( r@xmath922,300 ) , high signal - to - noise ratio ( s / n@xmath10100 per pixel ) , h - band ( @xmath111.511.69 @xmath1 m ) spectra of evolved , late - type stars , with the goal being to measure chemical abundances of 15 elements per star .
coupled with radial velocities that are accurate to @xmath9100 m - s@xmath12 , apogee is creating the first high - precision spectroscopic and radial - velocity survey of all galactic stellar populations ( bulge , bar , disks , halo ) using a uniform set of stellar tracers and spectral diagnostics , with a plan to observe and derive these parameters for 100,000 red giants by the end of 2014 .
the physical and chemical parameters of the apogee stars are derived from a suite of software packages that , together , are called the apogee stellar parameters and chemical abundances pipeline ( aspcap ) ; the details of this software and its analysis techniques will be discussed by allende - prieto et al .
( 2012 , in preparation ) .
the first - generation output of aspcap , which will be based on 1-d stellar atmosphere analyses in lte , will consist of the stellar parameters of effective temperature , t@xmath13 , surface gravity ( noted as log g ) , and the 1-d microturbulence parameter ( @xmath14 ) , along with the abundances of up to 16 elements ( fe , @xmath4c , @xmath6n , @xmath7o , mg , al , si , k , ca , ti , v , cr , mn , co , ni , or cu ) and , for many stars , the @xmath5c isotopic abundance
. one of the central components of the aspcap machinery is a spectral linelist constructed to produce the synthetic spectra that are used to compare to the observed ones ; the apogee linelist will be presented and discussed in detail by shetrone et al .
( 2012 , in preparation ) .
the goal of this study is to analyze a small number nearby field red giants with well - defined stellar parameters and use the apogee linelist to derive detailed chemical abundances via standard `` manual '' abundance analysis techniques ( ramirez and allende prieto 2011 is a recent example using such techniques ) .
the derived stellar parameters and chemical abundances can be compared to the same quantities derived via aspcap using the same spectra , linelist , and model atmosphere grid .
this comparison will be an important test of aspcap .
this work will also present new abundances in these stars based on an extensive and up - to - date spectral linelist in the near - ir h - band .
abundances will be presented in this paper using two different types of nomenclature , with one abundance scale defined as a(x)=log(n(x)/n(h ) ) + 12.0 .
the other scale compares the element ratio of n(x)/n(q ) in a target star to the sun via [ x / q]=(a(x)@xmath15 a(x)@xmath16 ) ( a(q)@xmath15
a(q)@xmath16 ) .
the apogee high - resolution spectra cover the wavelength range from @xmath9@xmath1115100 to 16900 and to analyze bright red giants over this spectral interval , data were obtained from the archives of the fourier transform spectrometer ( fts hall et al .
1979 ) that operated at the coude focus of the kitt peak national observatory mayall 4-m reflector .
a more complete description of observing techniques and the original data is discussed in hinkle , wallace , & livingston ( 1995 ) , in particular for the spectral atlas of @xmath0 boo .
the originally reduced spectra of selected stars were recorded as flux versus wavenumber ( cm@xmath12 ) over different dates and runs and covered a broad region in the h - band .
the spectra analyzed here were resampled in wavelength space and restricted to a wavelength range of @xmath1115000 to 17000 .
the resolution of the spectra for different stars varies from r=45,000 to 100,000 , but were all higher than what is being observed for apogee .
the higher resolution allows for a detailed investigation of the limits of what can be derived high - quality ( i.e. , high resolution and high signal - to - noise ) red - gaint spectra when using the apogee linelist .
there are five stars analyzed here : the well - studied @xmath0 boo , a mildly metal - poor k1.5 giant , @xmath1 leo , the prototype metal - rich star ( spinrad & taylor 1969 ) , which is a k1 giant , two near - solar - metallicity m giants , @xmath2 and ( m0iii ) and @xmath3 oph ( m0iii ) , and hd199799 , an agb star of spectral type ms with a mild enhancement of @xmath4c and a slightly elevated ratio of c / o ( relative to solar ) due to third dredge - up ( smith & lambert 1990 ) .
figure 1 illustrates a small portion of the apogee wavelength coverage for four of the red giants here .
this region from 15560
15590 includes representative species and how they vary with spectral type ( or effective temperature ) and metallicity .
lines from the main cno - containing molecules , co , oh , and cn fall within this region , as noted in the figure . of particular interest is the metal - rich k - giant @xmath1 leo , which exhibits quite strong cn lines ; much of the variation in the spectrum of @xmath1 leo compared to the other giants result from the high line - density of cn .
this figure also points to the desirability of analyzing large numbers of red giants via spectrum synthesis techniques using a detailed linelist . with a significant fraction of its red - giant targets in the thin and thick disks , plus the inner bulge and bar , the five types of red giants studied here
span a range of temperature and metallicity that provides a good test for aspcap .
the apogee survey targets red giants with effective temperatures from t@xmath13=3400k5000k , having surface gravities from log g=3.0 to -0.5 , and metallicities from [ fe / h]= -5.0 to + 1.0 .
these nearby giants have well - measured photometry and parallaxes , so temperatures , luminosities , and masses are all rather well - constrained . in addition , four of the five stars have previous abundance analyses in the literature and thus provide comparison tests for the abundances derived from the apogee linelist .
deriving stellar chemical abundances from high - resolution spectra requires as basic input the effective temperature ( t@xmath13 ) , surface gravity ( characterized as log g with g as cm - s@xmath18 ) , and overall metallicity as the fundamental parameters of the model atmosphere .
in addition , high - resolution abundance analyses of cool stars that use static 1-d stellar atmospheres also require the derivation of a non - thermal doppler - like broadening term called microturbulence ( @xmath14 ) , which is determined as part of the spectral analysis .
the first step is the determination of t@xmath13 and , for these relatively nearby red giants , is set by the near - infrared magnitudes of j and k. this particular choice is used because the apogee targets are selected from the 2mass catalog ( strutskie et al .
2006 ) and therefore all have ( j k ) available .
the aim is to focus on utilizing near - ir data and then compare to published literature results .
all of the red giants here have parallaxes measured by hipparcos and thus have distances known to various levels of accuracy .
the parallax is used to set the distance and , thus , m@xmath19 .
bolometric corrections from bessell , castelli , & plez ( 1998 ) are used in conjunction with ( j k ) to determine m@xmath20 and consequently stellar luminosity .
basic data for the red giants are presented in table 1 .
the most important derived quantity affecting the overall shape of the spectrum is t@xmath13 , which here is based on an average of two t@xmath13-(j k ) calibrations ; one from gonzalez hernandez & bonifacio ( 2009 ) and the other from bessell et al .
the gonzalez hernandez & bonifacio calibration is defined in the 2mass photometric system of ( j k@xmath21 ) , so the original johonson ( 1965 ) ( j k ) colors and k magitudes for these bright red giants ( which have saturated magnitudes in the 2mass catalog ) were transformed to 2mass values using the prescription from carpenter ( 2001 ) .
the differences between the two calibrations , for a given metallicity , are less than 30 - 50k for t@xmath13@xmath224500k , but above this temperature the bessell et al .
( 1998 ) becomes cooler by about 100k for a given ( j k@xmath21 ) ( where we have transformed the bessell et al .
scale onto the 2mass system ) .
there is also a small metallicity effect on the ( j k@xmath21 ) calibration which , for a given ( j k@xmath21 ) can be @xmath9100k over the broad metallicity range of [ m / h]= -2.0 to + 0.0 ( where m is taken to represent the generic abundance of metals , i.e. those elements other than h and he )
. given differences between two independent t@xmath13 calibrations of less than 30 - 50k over the temperature range of the target stars , as well as their modest range in metallicity , and well - determined magnitudes with @xmath9 0.02 uncertainty in ( j k@xmath21 ) ( which translates to @xmath23t@xmath13=30k ) , the derived values of t@xmath13 have uncertainties of @xmath2450k or less .
given the stellar luminosity and effective temperature , these fundamental properties are then compared to stellar evolution models presented in bertelli et al .
( 2008 ; 2009 , with data from these papers taken from the website http://stev.oapd.inaf.it/yzvar ) to estimate stellar masses and , ultimately , surface gravities through : g / g@xmath16= ( m / m@xmath16 ) @xmath25 ( l@xmath16/l ) @xmath25 ( t@xmath13/t@xmath26)@xmath8 . determining the surface gravity from evolutionay tracks is an iterative process , because it requires knowledge of the stellar metallicity .
this is illustrated in figure 2 where the stellar luminosity and t@xmath13 are plotted along with evolutionary models from a range of masses , with each panel having a different metallicity : z=0.017 ( top panel ) , 0.008 ( middle panel ) , and 0.004 ( bottom panel ) , correpsonding to overall metallicities of [ m / h]= + 0.07 , -0.26 , and -0.56 , respectively , when taking z=0.0145 for the sun , as suggested by lodders ( 2010 ) .
the illustrated evolutionary tracks follow stellar evolution up to the tip of the first ascent red giant branch ( which we label rgb ) .
initial model metalliities for the stars studied here are based on previously published results , with @xmath2 and , @xmath3 oph , and hd199799 started with solar metallicity , @xmath0 boo with [ fe / h]= -0.5 , and @xmath1 leo with [ fe / h]= + 0.3 . as there are numerous fe i lines in this spectral region , a sample of fe i lines are used to determine the iron abundance .
given the limited range of metallicity spanned by the nearby red giants studied here , the fe abundance is a good proxy for the overall stellar metallicity . if the derived fe abundance differed by more than 0.1 dex from the assumed value , a new surface gravity was derived and a new model generated in order to reanalyze the fe i lines until convergence between model abundance and derived abundance . for these relatively well - studied red giants ,
no more than two iterations were required until convergence .
table 2 presents , for these red giants , the derived parameters t@xmath13 , log ( l / l@xmath16 ) , mass , log g , microturbulent velocity ( @xmath14 see section 4.2 ) , and overall metallicity , which is represented by [ fe / h ] as discussed in section 4.2 .
given the position of the stars in the log l versus t@xmath13 plane shown in figure 2 , it can be ascertained that 4 stars ( @xmath0 boo , @xmath1 leo , @xmath3 oph , and @xmath2 and ) are most likely first - ascent red giants of low to moderate mass .
based on its warmer temperature and somewhat lower luminoisty , it may be more likely that @xmath1 leo is in , or very near the core - he burning phase ( clump giant ) and has thus already ascended the rgb and experienced the core - he flash .
it is also possible that the other 3 stars already noted above may be post - he core - burning stars ascending the agb ; however , lifetimes on the two separate evolutionary sequences would favor them as being on the rgb .
all fall above the `` luminosity bump '' ( fusi pecci et al .
1990 ) and would have thus experienced any extra - mixing that may occur during that phase of stellar evolution .
the coolest and most luminous red giant in this sample , hd199799 , is an agb star that is experiencing the third dredge - up , as evidenced by the observation of tc i in its spectrum and its elevated s - process abundances , such as [ y / fe]@xmath9 + 0.5 or [ nd / fe]@xmath9 + 0.4 ( smith & lambert 1988 ; 1990 ) .
as with the initial abundances to be derived from apogee , the analysis here will rely on 1-d model atmospheres computed in lte . within apogee , the choice was made to base the initial abundance determinations on an extensive model atmosphere database generated using the atlas9 code ( kurucz 1993 ) , thus the model atmospheres used here are based on these apogee models .
atlas9 is widely used as a universal lte 1d plane - parallel atmosphere modeling code .
the details of constructing the atmosphere database for apogee using atlas9 are presented in meszaros et al .
( 2012 ) .
the individual elemental abundances themselves will also be derived under the assumption of lte with the spectrum synthesis code moog as the analysis tool ( sneden 1973 is the original reference , but updated information , data , and the most recent versions of the code can be found at http://www.as.utexas.edu/ chris / moog.html ) .
an extensive effort was made to produce as complete a spectral linelist as possible , so the analysis technique used here , as in apogee in general , is to utilize spectrum synthesis in a quantitative comparison between synthetic and observed spectra .
all abundances are determined for each spectral line , or group of closely spaced molecular lines , via minimizing the residuals between observed and synthetic spectra as a function of the abundance in question .
due to both its relatively large cosmic abundance and large number of energy levels , iron has historically been used as a diagnostic species for certain stellar parameters , as well as an overall metallicity indicator , so the results for fe i lines selected from the apogee linelist ( shetrone et al .
2012 ) are discussed here . in particular , fe i lines are well - suited for setting the microturbulent velocity , @xmath14 , that is required for 1-d model atmosphere abundance analyses . a large number of fe i lines fall within the apogee spectral window ( 15100 16900 ) , although there are no fe ii lines strong enough to be detected in these red giants .
since all abundance determinations are done via spectrum synthesis , the fe i lines were culled to include those lines which contain only contributions from fe i ( this was determined by preliminary synthesis of the red giant spectra , then setting the fe abundance to zero and ensuring that the spectral feature in question vanished ) ; since the fe i lines are so numerous , this stringent selection still results in an adequate list of lines .
table 3 presents the fe i lines used for the iron abundance determinations . as is true of most of the atomic lines in the h - band ,
the excitation energies are dominated by rather high - excitation lines , although for fe i there are two lower - excitation lines .
the fe i lines span a large enough range in line - strength that they can be used to set the microturbulent velocity .
this parameter is defined by the value of @xmath14 that produces no trend in the fe abundance as a function of line - strength . for a range of microturbulent velocities
fe abundances were determined from each fe i line and the adopted value of @xmath14 was one in which a linear regression between a(fe ) as a function of log(w@xmath27/@xmath11 ) yielded a slope of zero .
this procedure is illustrated graphically in figure 3 , where the abundances derived for each individual line is plotted versus @xmath14 ( the continuous lines ) ; the circle with errorbars shows where the slope of a(fe ) with line - strength goes to zero .
the microturbulent velocities set by the fe i lines , as well as the fe abundances are listed in table 2 as part of the derived set of stellar parameters .
the individual line - by - line abundances as determined for each fe i line are also listed in table 3 .
the fe abundances were used to set the final overall metallicities of the atlas9 model atmospheres for the analysis of additional elements via atomic or molecular lines .
abundances for @xmath4c , @xmath6n , @xmath7o , and the minor carbon isotope @xmath5c are derived from combinations of vibration - rotation ( v r ) lines of co ( x@xmath28@xmath29@xmath30 ) and oh ( x@xmath31@xmath32 ) , along with electronic transitions of cn ( a@xmath31@xmath32 x@xmath31@xmath29 ) . although the details of the ingredients that went into these molecular lines in the apogee linelist will be found in shetrone et al .
( 2012 in preparation ) , a few highlights are noted here .
the adopted dissociation energies ( d@xmath33 ) are 11.092 ev for co , 4.395 ev for oh , and 7.76 ev for cn .
the gf - values are from goorvitch ( 1994 ) for co , goldman et al .
( 1998 ) for oh , and the cn gf - values are taken from the baseline linelist of kurucz ( 1993 ) , with updates to these values from the prescription given in melendez & barbuy ( 1999 ) . the procedure used for the cno analysis , since all red giants studied here have
c / o @xmath22 1 , is to use co to set the carbon abundance . with that carbon abundance ,
oh provides an o - abundance , which , if different from the initial value ( taken to be the solar value scaled by the stellar value of [ fe / h ] ) , is used to re - analyze the co lines .
this process is repeated until the c and o abundances yield consistent abundances from co and oh . with these values of c and o , the cn lines are used to derive the abundance of nitrogen . in general , the abundance of n has little to no effect on the co and oh lines , however , the respective final c , n , and o abundances all provide self - consistent results from co , oh , and cn . all spectral features were synthesized and the molecular regions that were used to extract the abundances are listed in table 4 . in this case , if a range of wavelengths is given it indicates that a spectral window containing numerous molecular lines was used ( for co and oh ) , while for cn individual transitions were identified and synthesized . as with table 3 ,
the abundances derived for each feature are also listed in table 4 . in the case of @xmath5c ,
as has been customery for minor isotopes , the abundance is listed as the ratio of @xmath4c/@xmath5c , with a(@xmath4c ) being set by the mean abundance of carbon-12 . with the fe and c , n , and
o abundances in hand , additional elements are derived via atomic spectral lines in the wavelength interval 15100 16900 .
the apogee linelist was searched for suitable species , all neutral over the t@xmath13 range of the red giants here , and some 12 other elements were deemed detectable and able to be analyzed quantitatively : mg i , al i , si i , k i , ca i , ti i , v i , cr i , mn i , co i , ni i , and cu i. the additional atomic lines are listed in table 5 . as with the molecular lines , if a precise wavelength is shown , the transition in question is represented by a single line , however if the wavelength is approximated by an integer value it indicates multiple transitions due to hyperfine ( hfs ) splitting or isotopic splitting . in the case of isotopic components ,
solar - system isotopic fractions are assumed and , in general , this assumption has no significant effect on the derived total atomic abundances .
figure 4 illustrates the fitting procedure for an atomic line from mn i in @xmath3 oph , with the underlying hfs components illustrated , as well as nearby blending features , with several synthetic spectra plotted which have differing mn abundances . as with the previous tables 3 and 4 ,
individual abundances are provided for each line or feature .
as noted previously , the details of the generation of the apogee linelist will be found in shetrone et al .
( 2012 ) .
table 6 provides a summary table of all abundances listed as their mean values and standard deviations from the lines or spectral intervals shown in tables 3 , 4 , and 5 . in cases where only one line or region was observed ,
no standard deviation is given .
the abundance of @xmath5c is given as the isotopic ratio of @xmath4c/@xmath5c .
most of the abundances presented here are derived from more than one atomic or molecular line or features of lines , thus internal consistency in the average abundances are defined by the scatter from the individual features .
this scatter is characterized by the standard deviations of table 6 . because multiple features arise from differing line strengths ( weak versus strong ) , excitation potentials , or ionization fractions , these respective standard deviations provide some estimate as to how well the 1-d lte analysis can recover the details of the underlying stellar spectrum from the basic model atmosphere parameters t@xmath13
, log g , metallicity , and derived value of @xmath14 .
the typical standard deviations from table 6 are @[email protected] dex . of note , is the metal - rich k - giant @xmath1 leo , which exhibits exceptionally strong cn lines , that are numerous throughout this h - band window .
the background cn absorption , creates blending which lowers the overall signature of other abundances and increases the uncertainty in these other abundances ; this is reflected in the generally larger standard deviations found in the @xmath1 leo abundances . besides internal consistency from the various spectral lines of different species ,
it is useful to explore their sensitivities to stellar parameters in order to characterize the magnitudes of possible systematic effects due to uncertainties in the four primary parameters of effective temeprature , surface gravity , metallicity , and 1-d derived microturbulent velocity . for this discussion t@xmath13
is written as t , log g as g , model atmosphere metallicity ( [ m / h ] ) as m , and microturbulent velocity as @xmath14 . because the abundance for a given element , a = log[n(a)/n(h ) ] + 12 .
= a(t , g , m,@xmath14 ) , the incremental change in this abundance due to small variations in the parameters is given by da=(@xmath34)dt + ( @xmath35)dg + ( @xmath36)d@xmath14 + ( @xmath37)dm .
the scatter in the abundance caused by small changes in the stellar parameters is then defined as @xmath38@xmath31@xmath39=@xmath40da@xmath31@xmath41 and da@xmath31 can be written as ( da)@xmath31=[(@xmath34)dt]@xmath31 + [ ( @xmath35)dg]@xmath31 + [ ( @xmath36)d@xmath14]@xmath31 + [ ( @xmath37)dm]@xmath31 + ( @xmath34)(@xmath35)[dtdg + dgdt ] + ( @xmath34)(@xmath36)[dtd@xmath14 + d@xmath14dt ] + ( @xmath34)(@xmath37)[dtdm + dmdt ] + ( @xmath35)(@xmath36)[dgd@xmath14 + d@xmath14dg ] + ( @xmath35)(@xmath37)[dgdm + dmdg ] + ( @xmath36)(@xmath37)[d@xmath14 dm + dmd@xmath14 ] .
the partial derivative terms are calculated by changing each model parameter and determining the resulting change in logarithmic abundance , a. since the perturbation in stellar parameters is relatively small , the changes are approximated by linear trends . as this paper is a limited study of a few stars as a test of the apogee linelist ,
a test of abundance sensitivity to stellar parameters is carried out for a representative model taken with t@xmath13=4000k , log g=1.3 , [ m / h]=+0.0 , with @xmath14=2.0 km - s@xmath12 . predicted equivalent widths were generated for the lines used in the abundance analysis and the stellar parameters were then pertrubed by dt=+50k , dg=+0.2 dex , dm=+0.1 dex , and d@xmath14=+0.2 km - s@xmath12 , with new abundances derived for each separate perturbation , giving the change in abundance for each parameter change .
these coefficients , which reflect the change in abundance , are listed in table 7 and provide an overall view of how sensitive the derived abundances are to changes in stellar parameters that are representative of realistic errors for these particular red giants .
columns 6 and 7 in table 7 provide the values of da two cases : one using only the first four terms which contains the change in abundance due to changes in each primary parameter ( da in column 6 , which is valid if all parameters are independent of each other ) , while column 7 includes the covariant terms . within the analysis technique applied here , it was found that the microturbulent velocity did not measurably depend on the derived effective temperature or gravity , and the dt - d@xmath14 and dg - d@xmath14 cross terms are not included . as can be seen by the comparison of da and da , the covariant terms contribute significantly , in many species , to the uncertainty .
the largest covariant terms are typically those between t@xmath13 and log g , caused by the slope of the rgb in the t@xmath13log(l / l@xmath16 ) plane ( since l@xmath9t@xmath43/g ) .
the abundances derived here rely on the apogee h - band spectral window from @xmath1115100 to 16900 , along with parallaxes and photometry .
historically , most detailed abundance patterns obtained from high - resolution spectroscopy have tended to be based on wavelengths below @xmath1110000 so it is useful to compare the abundances based on the apogee linelist with those from other studies for stars in common . within the sample of red giants discussed here ,
previous high - resolution abundance analyses have been conducted on @xmath0 boo , @xmath1 leo , @xmath2 and , and hd 199799 .
being the nearest red giant , @xmath0 boo is often used as the standard star of choice for abundance comparisons .
ramirez & allende prieto ( 2012 ) have produced a recent and detailed high - resolution optical spectral analysis , revieiwng both fundamental stellar parameters for this star , as well as deriving an extensive set of elemental abundances .
they used the observed spectral energy distribution and compared to theoretical ones from the kurucz grid of no - overshoot model atmospheres with @xmath0-element enhanced compositions ( [ @xmath0/fe]=+0.4 ; castelli & kurucz 2003 ) in order to set the t@xmath13 for @xmath0 boo .
the flux from @xmath11 @xmath90.3 10 @xmath1 m was used with the resultant best - fit t@xmath13=4286@xmath2430k .
this value is entirely consistent with that obtained here ( t@xmath13=4275@xmath2450k ) based on the ( j k ) color and calibration used here from bessell et al .
( 1998 ) and gonzalez hernandez & bonifacio ( 2009 ) .
ramirez & allende prieto ( 2012 ) used stellar evolutionary model tracks to estimate the mass ( as well as age ) of @xmath0 boo , as the effective temperature and parallax are known to high precision .
the mass was set by comparison to yonsei - yale isochrones ( yi et al .
2001 ; kim et al . 2002 ) and was found to be [email protected]@xmath16 , which yields a surface gravity of log [email protected] . in this
study the t@xmath13 = 4275k value from the ( j k ) color is adopted , while the ramirez & allende prieto log g is used .
these stellar parameters were used to generate a model atmosphere as described in section 4 with the atlas9 code in the same manner as is being done for the apogee targets ( meszaros et al .
2012 ) .
the microturbulence , @xmath14 , derived by ramiriz & allende prieto was set by fe i lines and was found to be @xmath14=1.74 km - s@xmath12 .
the fe i lines used from the apogee linelist yielded a value of @[email protected] km - s@xmath12 : very close to that derived from optical fe i lines .
a direct comparison of the derived abundances here with those from ramirez & allende prieto ( 2012 ) using completely differsent sets of lines leads to the following mean difference and standard deviation for 13 elements in common of @xmath23a(x)(this paper ramirez & allende prieto)[email protected] dex .
there is no significant offset in the overall abundance patterns and the scatter is that expected given the uncertainties in the log gf scales and internal line - to - line scatter within each study .
this comparison suggests that abundances derived from the apogee linelist can be used for comparisons to those found from analyses at visual wavelengths when high - quality spectra are analyzed .
this k - giant is often cited as the `` prototype '' for metal - rich stars , having been noted as such by spinrad & taylor ( 1969 ) , and some of its spectral regions contain numerous and strong cn lines .
two older abundance studies in the literature ( gratton & sneden 1990 ; smith & ruck 2000 ) are used to compare to the abundances derived here , as well as a more recent analysis by fulbright , mcwilliam , & rich ( 2007fmr ) . as with the @xmath0
boo comparison in section 4.5.1 , both of these published analyses are based on high - resolution spectra in the visual . both studies used t@xmath13=4540k for @xmath1 leo , which is similar to the value 4550k adopted here .
the surface gravities were slightly different , with log g= [email protected] in gratton & sneden and [email protected] in smith & ruck ; both values agree with the value here , within the uncertainties , of log g=2.1 derived from the luminosity and estimate of stellar mass .
the value of @xmath14=1.8 km - s@xmath12 found from the h - band fe i lines is slightly higher than 1.2 km - s@xmath28 found by both gratton & sneden ( 1990 ) and smith & ruck ( 2000 ) , but with all studies having uncertainties of 0.3 - 0.5 km - s@xmath12 , this offset can not be claimed to be significant .
a comparison of abundances between here and gratton & sneden ( 1990 ) for 12 elements finds a mean @xmath23a(x)(this paper - gs)[email protected] dex , while the same comparsion with smith & ruck ( 2000 ) yields @xmath23a(x)[email protected] dex for the 3 elements analyzed by them : fe , mg , and ca .
the abundance offsets are not significant and the scatter is within what is expected .
the more recent abundance analysis of @xmath1 leo by fmr was part of their study of bulge red giants .
their derived stellar parameters were quite similar to those found here , with t@xmath13(fmr)=4520k , ( @xmath23t(this study - fmr)= + 30k ) , log g = 2.33 ( @xmath23log g(this study - fmr)= -0.23 ) , and @xmath14=1.50 km - s@xmath12 ( @xmath23@xmath14(this study - fmr)= + 0.3 km - s@xmath12 ) .
there are 7 elements in common ( fe , o , mg , al , si , ca and ti ) and the mean difference is @xmath23a(x)(this study - fmr)=-0.10 @xmath24 0.13 dex .
a comparison amongst the three @xmath1 leo studies points to rather small offsets in stellar parameters ( @xmath9 30k in t@xmath13 , 0.2 dex in log g , 0.3 - 0.5 km s@xmath44 in @xmath14 , and 0.1 dex in [ m / h ] ) with the conclusion that h - band spectroscopy can be compared reliably to optical analyses with reasonable accuracies , even at this higher metallicity .
this m - giant was analyzed previously by smith & lambert ( 1985 ) who derived t@xmath13=3800k ( from its ( v k ) color ) , log g=1.6 , and @xmath14=2.1 km - s@xmath12 .
the derived effective temperature here of 3825k is very close to smith & lambert , as is the microturbulence of 2.2 km - s@xmath12 .
the surface gravity from smith & lambert is larger by 0.7 dex , but their value relied on the distance ( as it does in this paper ) , for which they used a luminosity calibration for the ca ii k - line reversal that was tied to a pre - hipparcos hyades distance that was too small . making the estimated log g corrections to the smith & lambert ( 1985 ) abundances ( based on their published sensitivities of abundance with log g ) for @xmath4c , @xmath6n , @xmath7o , ti , fe , and ni we find a mean difference and standard deviation of @xmath23a(x ) ( this paper - smith & lambert)[email protected] dex .
beta and was also used as a comparison star by chou et al .
( 2010 ) in their study of sgr dwarf - galaxy tidal streams in the galactic halo .
chou et al . ( 2010 ) applied a similar technique as that employed here to estimate temperatures and gravities and found t@xmath13=3850@xmath2475k , log g=0.9 and @xmath14=1.96 km - s@xmath12 , giving differences of only 25k in t@xmath13 , 0.0 in log g , and + 0.24 km - s@xmath12 in @xmath14 .
their analysis used the spectral region around @xmath117500 containing a sample of fe i and ti i lines and determined a(fe)[email protected] and a(ti)[email protected] , which can be compared to values of [email protected] and [email protected] , respectively , here , yielding differences of + 0.11 dex in fe and + 0.17 dex in ti .
hd 199799 is classified as spectral type m2s , which means that it displayed enhanced bands of zro in classification spectra .
smith & lambert ( 1988 ) detected the radioactive s - process element tc i in its spectrum , identifying this star as an intrinsic tp - agb star undergoing third dredge - up .
a quantitative abundance analysis was then carried out by smith & lambert ( 1990 ) and they derived similar stellar parameters to those derived here : the same t@xmath13=3400k , with a slightly lower value of log g=+0.3 , compared to + 0.5 here . the microturbulent velocity determined here from the h - band spectra is @xmath14=2.4 km - s@xmath12 , which is 0.7 km - s@xmath12 larger than smith & lambert ( 1990 ) , who used fe i lines near @xmath117500 . taking into account the slightly lower gravity and smaller @xmath14 ,
the abundances from smith & lambert ( 1990 ) can be compared to those here for fe , @xmath4c , @xmath6n , @xmath7o , ti , cr , co , and ni , with the mean difference being @xmath23a(x ) ( us - sl90)= [email protected] .
the respective carbon isotopic ratios compare well , with smith & lambert ( 1990 ) finding @xmath4c/@xmath5c=28 , while a value of 27 is obtained here .
the derived c / o ratios are the same in the two studies , with @xmath4c/@xmath7o=0.68 .
the comparisons presented here are between studies that have possibly used somewhat different families of model atmospheres ( e.g. marcs or atlas ) , different spectral regions , and linelists derived from heterogeneous sources . given the variety of data and analysis techniques , the derived abundances , in all cases , compare well at approximately the 0.10 to 0.15 dex level .
a glance at table 7 reveals that such differences fall within this range of uncertainty for most elements given modest changes in fundamental model atmosphere parameters .
since the derived stellar parameters themselves rely to some degree on the type of analysis , the generally good agreement between the derived abundances from the various published sources , along with the expected uncertainties in this study , as presented in table 7 , is encouraging .
the gist of the comparisons , since many of the studies are based on optical spectra , is that h - band spectrscopy can be used to define a precise internal abundance scale that can also extended to abundances derived from other , more common spectral windows , such as in the optical .
the abundances from table 6 and an estimation of their respective uncertainties from table 7 , along with the stellar parameters from table 2 represent , to some extent , what aspcap will produce from the apogee spectra .
the following discussion highlights these abundances in light of internal red giant stellar evolution , as well as galactic chemical evolution and stellar populations .
these discussion topics anticipate what sorts of information will be gleaned from the apogee targets .
based on their luminosities and effective temperatures ( figure 2 ) , four of the red giants studied here are either first - ascent giants or , perhaps for @xmath1 leo , a clump red giant .
hd 199799 is in a more advanced stage of stellar evolution where it is undergoing thermal pulses and third dredge - up on the agb .
the combination of nucleosynthesis and convective mixing in evolved red giants should result in surface abundances of @xmath4c , @xmath5c , and @xmath6n that have been altered from their initial ( main sequence ) values due to cn - cycle h - burning and first dredge - up .
the result of cn - cycle mixing is seen mostly easily in the @xmath4c/@xmath5c ratios which are much higher in main - sequence stars , with the sun having a value of 89 .
all four of the rgb or clump giants have quite low values , ranging from @xmath4c/@xmath5c@xmath96 - 20 .
gilroy ( 1989 ) and many subsequent studies of both open and globular clusters ( e.g. , mikolatis et al . 2012 or
briley et al . 1995 ) have noted that the carbon isotopic ratios correlate with stellar mass in stars above the luminosity bump on the rgb or in the clump ; red giants more massive than [email protected]@xmath16 have values of @xmath4c/@xmath5c@xmath922 - 30 , while lower - mass red giants exhibit decreasing isotopic ratios with declining mass .
the mass estimates derived here for field stars from distance - luminosity and stellar evolutionary tracks are modestly well - defined and the top panel of figure 5 shows the @xmath4c/@xmath5c values versus the estimated stellar mass . also plotted are representative results for open clusters from gilroy ( 1989 ) and mikolaitis et al .
( 2012 ) , and the globular cluster m71 from briley et al .
the cluster red giants have recently evolved from the cluster turn - offs , which have a well - determined masses . for the lower - mass stars included in these cluster studies , evolution along
the rgb is rapid ( @xmath910@xmath45 yrs ) when compared to main sequence lifetimes ( @xmath910@xmath46 - 10@xmath47 yrs ) so the red giant mass is similar to the turn - off mass .
this can be demonstrated using the evolutionary tracks from bertelli et al .
( 2009 ) , where an isochrone with a turn - off mass of 4.0m@xmath16 has a red giant tip mass of 4.2m@xmath16 , or a ratio of rgb to to masses of 1.05 .
this ratio of masses becomes even closer to 1.0 for lower masses , which include all of the points in figure 5 .
the points from gilroy represent the mean values of @xmath4c/@xmath5c for 19 open clusters with a total of 55 red giants in her sample . the mean carbon isotopic ratios for two other open clusters ( with a total of 10 red giants from cr261 or ngc6253 ) from mikolaitis et al .
( 2012 ) are also shown ( open blue squares ) . a representative globular cluster , m71 , which has a lower main - sequence mass ( [email protected]@xmath16 ) from briley et al . ( 1995 )
is plotted as the open blue triangle and represents the mean of two of the cn - weak stars in this cluster .
the cn - strong giants that briley et al . also studied are not plotted , as they exhibit slightly lower isotopic ratios but almost certainly represent 2nd generation stars in m71 that were formed from large fractions of the ejecta of 1st generation intermediate - mass red giants and thus are not representative of single - star red giant evolution .
the curves in the top panel of figure 5 represent stellar evolutionary model results from charbonnel & lagarde ( 2010 ) ; the solid black line illustrates their predictions for standard rgb dredge - up ( no `` extra - mixing '' or rotation included ) .
clearly , standard rgb first dredge - up does not fit the observed trend of @xmath4c/@xmath5c for red giants with [email protected]@xmath16 .
this poor comparison between theory and observation has led to work in trying to identify other types of `` extra mixing '' mechanisms that would lead to lower carbon - isotopic ratios in the lower - mass red giants .
two such hypothesized mechanisms are thermohaline mixing or thermohaline mixing coupled with rotational mixing , as discussed recently by charbonnel & lagarde ( 2010 ) .
the dashed curves present results from charbonnel & lagarde also , with the short dashes for thermohaline mixing only and the long dashes for thermohaline mixing plus rotational mixing for models with initial equatorial rotational velocities of 110 km - s@xmath12 . the stellar models that incorporate these particular types of extra - mixing
provide fair agreement with the observations , although these are not the only types of processes that have been studied and are used here to merely illustrate the effect of extra mixing .
the trends found in the clusters also overlap well with the field stars analyzed here , except for hd 199799 , which is in a different evolutionary phase than the other red giants .
this star has enriched its surface in @xmath4c that was synthesized by he - burning as a tp - agb star . in the simplest case of pure @xmath4c dredge - up
, the surface @xmath4c/@xmath5c ratio would increase , which is where hd 199799 lies in the top panel of figure 5 relative to the other red giants .
such an effect has been noted by both lambert et al .
( 1986 ) and smith & lambert ( 1990 ) , where these studies find increasing values of @xmath4c/@xmath5c as tp - agb stars evolve from m to ms to s to c stars ( with increasing values of @xmath4c/@xmath7o ) .
the bottom panel of figure 5 plots the abundance by number ratio of n(c)/n(n ) versus stellar mass , where the carbon abundance is now the total of @xmath4c + @xmath5c : for reference the solar ratio is 3.80 .
nitrogen is also the total abundance , however in the case of the observations , @xmath48n is ignored as its abundance is very low ( and not detected in these red giants ) when compared to @xmath6n .
there is an observed decrease in c / n with increasing mass as predicted by both standard rgb dredge - up and dredge - up plus thermohaline mixing ( charbonnel & lagarde 2010 ) , which are shown as the solid and dashed curves , respectively , and are from the same models as in the top panel . in the case of the c
/ n ratios , the total carbon and nitrogen abundances are not as strongly influenced by non - standard mixing as the carbon isotopic ratios .
figure 6 combines the c / n ratios versus @xmath4c/@xmath5c ratios for both the observed red giants and the models from charbonnel & lagarde ( 2010 ) . when displayed this way , the different behavior between c / n versus @xmath4c/@xmath5c is striking and these two values are purely spectroscopically derived quantities .
the trend of c / n with @xmath4c/@xmath5 seems to provide a sensitive indicator of red giant mixing processes , as well as yields a mass estimate for stars on the rgb .
again , hd 199799 stands apart from the other red giants as it has internally enhanced its @xmath4c abundance by about a factor of two since it was a rgb star , based on its @xmath4c/@xmath5c and mass , relative to other rgb stars in the top panel of figure 5 .
the dashed - dotted line connects to the region in this diagram where hd 199799 would fall if its surface @xmath4c abundance were reduced by a factor of two , which is a simplified sketch of how tp - agb stars might evolve across such a diagram from the rgb .
the assumption of only cn - cycle mixing taking place in the red giants is tested in figure 7 , where the total carbon abundance ( @xmath4c + @xmath5c ) is plotted versus the nitrogen abundance ( here @xmath6n ) .
the solar values of c and n are also shown , along with the dashed line illustrating scaled solar - abundances for both c and n. the continuous blue curves represent decreasing carbon and increasing nitrogen abundances from the scaled solar lines , with the constraint that the number of c + n nuclei is conserved , as in the cn - cycle ( with no leakage into the on - cycle ) . each curve is scaled in its initial c and n abundances by the stellar value of [ fe / h ] for @xmath0 boo , @xmath2 and , @xmath3 oph , and @xmath1 leo .
cn - cycle mixing alone is sufficient to represent the derived abundances of c and n with initial abundances having [ c +
n / fe]@xmath90.0 for each rgb star . due to having undergone
third dredge - up and @xmath4c enrichment , hd 199799 will be offset from its cn curve , so no curve is plotted for it .
the dashed - dotted line simply shows the shift in a(c ) if hd 199799 has doubled its rgb @xmath4c abundance .
having nearly the same metallicity as @xmath2 and , hd 199799 might be expected to have been near the cn curve for @xmath2 and before it began its evolution as a tp - agb star .
the apogee h - band wavelength region presents good opportunities for probing many nuclear and mixing processes that occur over the various phases of red giant evolution .
other than the c and n isotopes , the vast majority of apogee red giant targets will not have altered their surface abundances of the other studied elements , so the remaining elements are dominated by chemical evolution within the various stellar populations of the milky way ( or its accreted systems ) .
the derived abundances of each of the other elements are discussed in the light of observed trends of these elements with a(fe ) from other studies in the literature .
the number of published abundances is now enormous ; however in this first exploratory study using the apogee linelist , a comparison with a small number of selected literature sources is sufficient . for comparison purposes in this paper a stellar sample representative of the milky way stellar populations
was constructed using reddy et al .
( 2003 ; thin disk with some thick disk and having 58 stars ) , reddy et al .
( 2006 ; thick disk with 176 stars ) , johnson ( 2002 ; halo with 23 stars ) , and fulbright ( 2002 ; halo with 178 stars ) .
the investigation of elements derived from the apogee linelist is via a number of figures , beginning with figure 8 , which presents values of o / fe , mg / fe , and al / fe versus a(fe ) ; the shorthand notation here of x / fe denotes the logarithmic number ratio of element x to iron , log[n(x)/n(fe ) ] , or a(x ) - a(fe ) .
this nomenclature is chosen over the [ x / fe ] in the plots as it presents a direct logarithmic number ratio with the solar abundance ratio being noted in each panel , thus no reference to the underlying solar abundance is needed , but the relative scales are equivalent to [ x / fe ] .
the top panel of figure 8 shows the behavior of o / fe and , as has been well - known for decades , values of o / fe are elevated in the thick disk and halo stars . among the field red giants here ,
the slightly metal - poor objects @xmath0 boo and @xmath2 and both have o / fe enhanced by @xmath9 + 0.2 + 0.3 .
the near - solar metallicity m - giant @xmath3 oph has a near - solar ratio of o / fe , while the metal - rich giant @xmath1 leo also exhibits a near - solar value o / fe .
the behavior of mg / fe is shown in the middle panel of figure 8 and , like oxygen , mg / fe ratios are typically elevated in the thick disk and halo populations .
the four red giants closest to solar in a(fe ) all exhibit basically solar mg / fe ratios , with @xmath0 boo having a slight enhancement of@xmath9 + 0.1 dex , which overlaps the thick disk populations .
aluminum abundances are plotted in the bottom panel of figure 8 and the behavior of al / fe versus a(fe ) is not so well - defined , with relatively large scatter .
mu leo is found to be modestly enhanced in al / fe by about + 0.10+0.15 dex , as is @xmath0 boo , with an enhancement of @xmath9 + 0.2 .
the abundances for the rest of the elements studied here are illustrated in figure 9 ( for si , k , and ca ) , figure 10 ( for ti , v , and cr ) , figure 11 ( for mn , co , and ni ) , and figure 12 ( for cu ) . in all elements , the general trends found in the general galactic populations are recovered in the small sample of field red giants included here .
the one exception is ti in @xmath3 oph and @xmath1 leo , both of which are found to exhibit enhanced values of ti / fe of + 0.2 dex .
there is no obvious explanation for this , however the continuing analysis of apogee spectra with this linelist wuill reveal if this possible effect appears in significant numbers of other red giants .
the results presented and discussed here indicate that h - band spectra analyzed in lte with the apogee linelist and 1d model atmospheres can provide accurate results for chemical abundances .
archival high - resolution fts spectra from @xmath111510016900 are analyzed to derive detailed chemical abundance distributions for five field red giants using the apogee linelist and model atmosphere grid .
the red giants span a range in t@xmath13 from 3400k to 4540k and log g from 0.5 to 2.1 , covering a significant part of the effective temperature - surface gravity range being observed in the apogee survey .
chemical abundances are determined for 16 elements , along with the minor carbon isotope , @xmath5c .
comparisons of the abundances derived here with those from previously published studies of four of the target stars here , mostly from visual - wavelength high - resolution spectra , find abundance differences of @xmath90.1 dex .
the apogee wavelength window with its derived linelist and model atmosphere grid , coupled to spectrum synthesis modelling , can be used to analyze high - resolution h - band spectra of red giants to probe both galactic chemical evolution in stellar populations , as well as internal red giant nucleosynthesis and mixing .
we thank ken hinkle , for helping to make the archival fts spectral available and to luan ghezzi for useful comments on an initial draft of this paper . vs
acknowledges partial support for this research from the national science foundation ( ast1109888 ) and capes , brazil .
this publication makes use of 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 / california institute of technology , funded by the national aeronautics and space administration and the national science foundation .
this research has made use of the simbad database , operated at cds , strasbourg , france .
funding for sdss - iii has been provided by the alfred p. sloan foundation , the participating institutions , the national science foundation , and the u.s .
department of energy office of science .
the sdss - iii web site is http://www.sdss3.org/. sdss - iii is managed by the astrophysical research consortium for the participating institutions of the sdss - iii collaboration including the university of ari- zona , the brazilian participation group , brookhavenna- tional laboratory , university of cambridge , carnegie mellon university , university of florida , the french participation group , the german participation group , harvard university , the instituto de astrofisica de ca- narias , the michigan state / notre dame / jina participa- tion group , johns hopkins university , lawrence berke- ley national laboratory , max planck institute for astro- physics , new mexico state university , new york univer- sity , ohio state university , pennsylvania state univer- sity , university of portsmouth , princeton university , the spanish participation group , university of tokyo , uni- versity of utah , vanderbilt university , university of vir- ginia , university of washington , and yale university .
allende prietro , c. and aspcap team 2012 , in preparation bertelli , g. , girardi , l. , marigo , p. , nasi , e. 2008 , , 484 , 815 bertelli , g. , nasi , e. , girardi , l. , marigo , p. 2009
, , 508 , 355 bessell , m. s. , castelli , f. , plez , b. 1998 , , 333 , 321 briley , m. m. , smith , v. v. , & lambert , d. l. 1995 , , 424 , l119 carpenter , j.m .
2001 , , 121 , 285 castelli , f. & kurucz , r. l. 2003 , new grids of atlas9 model atmospheres , iaus , 210 , 20pd charbonnel , c. & lagarde , n. 2010 , a&a 522 , 10 chou , m - y .
majewski , s. r. , cunha , k. , smith , v. v. , patterson , r. j. , & martnez - delgado , d. 2010 , apj , 720 l5 eisenstein , d. j. , weinberg , d. h. , agol , e. , hiroaki , a. , allende prieto , c. , and 239 coauthors 2011 , , 142 , 72 fulbright , j. p. 2002 , aj , 123 , 404 fulbright , j. p. , mcwilliam , a. , rich , r. m. 2007 , apj , 661 , 1152 fusi pecci , f. , ferraro , f. r. , crocker , d. a. , rood , r. t. , buonanno , r. 1990 , , 238 , 95 gilroy , k. k. 1989 , apj , 347 , 835 goldman , a. , shoenfeld , w. g. , goorvitch , d. , chackerian , c. , jr . , dothe , h. , melen , f. , abrams , m. c. , & selby , j. e. a. 1998 , j. quant . spectrosc . radiat .
transfer , 59 , 453 gonzalez hernandez , j.i .
, & bonifacio , p. 2009
, , 497 , 497 goorvitch , d. 1994 , apjs , 95 , 535 gratton , r. g. , & sneden , c. 1990 , , 234 , 366 hall , d. n. b. , ridgway , s. t. , bell , e. a. , yarborough , j. m. 1979 , proc .
spie , 172 , 121 hinkle , k. , wallace , l. , & livingston , w. 1995 , , 107 , 1042 johnson , h. l. 1965 , comm .
lunar & planetary lab .
53 johnson , j. a. 2002 , , 139 , 219 kim , y. , demarque , p. , yi , s. k. & alexander , d. r. 2002 , apjs , 143 , 499 kurucz , r. l. 1993 , atlas9 stellar atmosphere programs and 2 km s1 grid .
kurucz cd - rom no .
cambridge , mass.:smithsonian astrophysical observatory , 1993 , 13 lambert , d. l. , gustafsson , b. , eriksson , k. , hinkle , k. h. 1986 , apjs , 62 , 373 lodders , k. 2010 , in principles and perspectives in cosmochemistry , astrophysics & space science proceedings , part 3 ( springer - verlag : berlin heidelberg ) , a. goswami & b.e .
reddy , eds .
, p.379 melendez , j. , & barbuy , b. 1999 , , 124 , 527 meszaros , s. , allende prieto , c. , edvardsson , b. , castelli , f. , garcia perez , a. e. , gustafsson , b. , majewski , s. r. , plez , b. , schiavon , r. , shetrone , m. , de vicente , a. 2012 , , 144 , 120 mikolaitis , s. , tautvaisiene , g. , gratton , r. , bragaglia , a. , & carretta , e. 2012 , a&a 541 , 137 ramirez , i. & allende prieto c. 2011 , apj , 743 , 135 reddy , b. e. , tomkin , j. , lambert , d. l. , & allende prieto , c. 2003 , mnras , 340 , 304 reddy , b. e. , lambert , d. l. , & allende prieto , c. 2006 , 367 , 1239 shetrone , m. and aspcap team , in preparation smith , g. , & ruck , m. j. 2000 , , 356 , 570 smith , v. v. , & lambert , d. l. 1985 , , 294 , 326 smith , v. v. , & lambert , d. l. 1988 , , 333 , 219 smith , v. v. , & lambert , d. l. 1990 , , 72 , 387 sneden , c. 1973 , apj , 184 , 839 spinrad , h. , & taylor , b. j. 1969 , , 157 , 1279 skrutskie , m.f . ,
cutri , r.m . ,
stiening , r. , weinberg , m.d . , schneider , s.,carpenter , j.m .
, beichman , c. , capps , r. , chester , t. , elias , j. , huchra , j. , liebert , j. , lonsdale , c. , monet , d.g . ,
price , s. , seitzer , p. , jarrett , t. , kirkpatrick , j.d . , gizis , j. , howard , e. , evans , t. , fowler , j. , fullmer , l. , hurt , r. , light , r. , kopan , e.l . , marsh , k.a .
, mccallon , h.l . , tam , r. , van dyk , s. , wheelock , s. 2006 , , 131 , 1163 van leeuwen , f. 2007 , , 474 , 653 yi , s. ; demarque , p. ; kim , y. , lee , y. , ree , c. h. ; lejeune , t. & barnes , s. 2001 , apjs , 136 , 417 ccccccc @xmath2 and & 337 & m0iii & [email protected] & 61@xmath243 & 0.98 & [email protected] + @xmath1 leo & 3905 & k2iii & [email protected] & 38@xmath241 & 0.68 & [email protected] + @xmath0 boo & 5340 & k2iii & [email protected] & [email protected] & 0.79 & [email protected] + @xmath3 oph & 6056 & m0iii & [email protected] & [email protected] & 0.97 & [email protected] + hd 199799 & ... & m2s & [email protected] & 386@xmath2492 & 1.25 & [email protected] + ccccccc @xmath2 and & [email protected] & [email protected] & 3825@xmath2475 & [email protected] & [email protected] & -0.2 + @xmath1 leo & [email protected] & [email protected] & 4550@xmath2450 & [email protected] & [email protected] & + 0.3 + @xmath0 boo & [email protected] & [email protected] & 4275@xmath2450 & [email protected] & [email protected] & -0.4 + @xmath3 oph & [email protected] & [email protected] & 3850@xmath2475 & [email protected] & [email protected] & + 0.0 + hd 199799 & [email protected] & [email protected] & 3400@xmath2475 & [email protected] & [email protected] & -0.2 + cccccccc 15194.492 & 2.223 & -4.779 & 7.04 & 7.22 & 7.37 & 7.83 & 7.05 + 15207.526 & 5.385 & + 0.080 & 7.04 & 7.29 & 7.40 & bl@xmath49 & 7.20 + 15395.718 & 5.620 & -0.341 & 6.96 & 7.27 & bl@xmath49 & 7.88 & 7.42 + 15490.339 & 2.198 & -4.807 & 6.99 & 7.23 & 7.55 & 7.76 & 7.50 + 15648.510 & 5.426 & -0.701 & 6.97 & 7.20 & 7.35 & 7.75 & 7.03 + 15964.867 & 5.921 & -0.128 & 6.93 & 7.22 & 7.35 & 7.65 & 7.33 + 16040.657 & 5.874 & + 0.066 & 6.94 & 7.20 & bl@xmath49 & 7.69 & 7.14 + 16153.247 & 5.351 & -0.743 & 6.94 & 7.22 & 7.53 & 7.75 & 7.32 + 16165.032 & 6.319 & + 0.723 & 6.98 & 7.24 & 7.56 & 7.73 & 7.30 + ccccccc & & & & & & + ( 3 - 0 ) v - r & 15578 15586 & 7.86 & 8.05 & 8.27 & w@xmath49 & 8.51 + ( 4 - 1 ) v - r & 15774 15787 & 7.97 & 8.11 & 8.24 & 8.46 & 8.51 + ( 5 - 2 ) v - r & 15976 16000 & 7.90 & 8.07 & 8.23 & 8.56 & 8.38 + ( 6 - 3 ) v - r & 16183 16196 & 7.90 & 8.03 & 8.22 & 8.54 & 8.43 + * @xmath4c/@xmath5c ratios from @xmath5c@xmath7o and @xmath5c@xmath6n lines * & & & & & & + @xmath5c@xmath7o(3 - 0 ) v - r & 15922 15926 & 7 & 17 & 14 & w@xmath49 & 30 + @xmath5c@xmath7o(4 - 1 ) v - r & 16120 16125 & 6 & 15 & 11 & w@xmath49 & 32 + @xmath5c@xmath7o(6 - 3 ) v - r & 16740 16747 & 6 & 13 & 11 & w@xmath49 & 20 + @xmath5c@xmath6n & 15314 15315 & w@xmath49 & w@xmath49 & w@xmath49 & 22 & w@xmath49 + @xmath5c@xmath6n & 15354 15356 & w@xmath49 & w@xmath49 & w@xmath49 & 17 & w@xmath49 + * @xmath7o from @xmath7oh lines * & & & & & & + ( 2 - 0 ) p@xmath50 9.5 & 15277 15282 & 8.55 & 8.83 & 8.78 & 9.10 & 8.80 + ( 3 - 1 ) p@xmath51 3.5 & 15390 15392 & 8.54 & 8.78 & 8.76 & 8.93 & 8.60 + ( 3 - 1 ) p@xmath51 4.5 & 15504 15507 & 8.48 & 8.71 & 8.73 & 9.15 & 8.58 + ( 2 - 0 ) p@xmath51 11.5 & 15568 15573 & 8.51 & 8.80 & 8.78 & 8.98 & 8.52 + ( 3 - 1 ) p@xmath51 9.5 & 16189 16193 & 8.47 & 8.78 & 8.81 & 9.07 & 8.63 + * @xmath6n from @xmath4c@xmath6n lines * & & & & & & + ( 1 - 2 ) q2 41.5 & 15260 .
& 7.61 & 8.10 & 8.23 & 8.83 & 8.31 + ( 1 - 2 ) p2 34.5 & 15322 . & 7.68 & 8.07 & 8.21 & 8.88 & 8.35 + ( 1 - 2 ) r2 56.5 & 15397 .
& 7.65 & 8.06 & 8.26 & 8.63 & 8.28 + ( 0 - 1 ) r1 68.5 & 15332 .
& 7.63 & 7.95 & bl@xmath52 & 8.68 & 8.35 + ( 0 - 1 ) p2 49.5 & 15410 . & 7.66 & 8.03 & 8.13 & 8.80 & bl@xmath52 + ( 0 - 1 ) q2 59.5 & 15447 . & 7.64 & 7.98 & 8.28 & 8.73 & bl@xmath52 + ( 0 - 1 ) q1 60.5 & 15466 . & 7.65 & 8.06 & 8.23 & 8.51 & bl@xmath52 + ( 1 - 2 ) p2 38.5 & 15472 . & 7.71 & 8.13 & 8.11 & 8.68 & 8.25 + ( 0 - 1 ) p1 51.5 & 15482 . & 7.56 & 8.03 & 8.15 & 8.68 & 8.25 + ccccccccc & 15740.716 & 5.931 & -0.262 & 7.15 & 7.25 & 7.53 & 7.80 & 7.29 + & 15748.9 & 5.932 & + 0.276 & 7.09 & s@xmath49 & 7.57 & s@xmath49 & 7.23 + & 15765.8 & 5.933 & + 0.504 & 7.04 & s@xmath49 & s@xmath49 & 7.87 & bl@xmath52 + & 15879.5 & 5.946 & -1.248 & 7.13 & 7.20 & 7.45 & 7.82 & bl@xmath52 + & 15886.2 & 5.946 & -1.555 & 7.22 & 7.21 & 7.53 & bl@xmath52 & 7.18 + & 15889.485 & 5.946 & -2.013 & bl@xmath52 & 7.30 & 7.55 & 7.90 & w@xmath53 + & 15954.477 & 6.588 & -0.807 & 7.25 & 7.36 & 7.61 & bl@xmath52 & 7.45 + * al i * & 16718.957 & 4.085 & + 0.290 & 6.15 & 6.15 & s@xmath49 & 6.85 & bl@xmath52 + & 16763.359 & 4.087 & -0.524 & 6.16 & 6.09 & 6.45 & 6.94 & 6.10 + * si i * & 15361.161 & 5.954 & -1.925 & 6.99 & bl@xmath52 & bl@xmath52 & bl@xmath52 & w@xmath53 + & 15376.831 & 6.222 & -0.649 & 7.18 & 7.17 & 7.57 & 7.66 & w@xmath53 + & 15833.602 & 6.222 & -0.168 & 6.93 & 6.95 & 7.28 & 7.72 & 7.20 + & 15960.063 & 5.984 & + 0.107 & 7.07 & 7.14 & 7.43 & 7.88 & w@xmath53 + & 16060.009 & 5.954 & -0.566 & 7.21 & 7.25 & 7.59 & 7.74 & 7.33 + & 16094.787 & 5.964 & -0.168 & 7.11 & 7.06 & 7.46 & 7.66 & 7.32 + & 16215.670 & 5.964 & -0.665 & 7.15 & 7.23 & 7.62 & bl@xmath52 & bl@xmath52 + & 16680.770 & 5.984 & -0.140 & 7.11 & 7.25 & 7.62 & 7.83 & 7.35 + & 16828.159 & 5.984 & -1.102 & 7.13 & 7.17 & 7.45 & 7.85 & bl@xmath52 + * k i * & 15163.067 & 2.670 & + 0.524 & 4.81 & 4.82 & 5.18 & 5.63 & 5.20 + & 15168.376 & 2.670 & + 0.347 & 4.77 & 4.89 & bl@xmath52 & bl@xmath52 & bl@xmath52 + * ca i * & 16136.823 & 4.531 & -0.552 & 5.75 & 6.09 & 6.23 & 6.69 & 6.06 + & 16150.763 & 4.532 & -0.229 & 5.88 & 5.95 & 6.21 & 6.53 & 6.07 + & 16155.236 & 4.532 & -0.619 & 5.82 & 6.02 & 6.30 & 6.61 & 6.16 + & 16157.364 & 4.554 & -0.208 & 5.90 & 6.02 & 6.23 & 6.63 & 6.16 + * ti
i * & 15543.756 & 1.879 & -1.120 & 4.53 & 4.74 & 5.05 & 5.58 & 4.74 + & 15602.842 & 2.267 & -1.643 & 4.66 & 4.74 & 5.03 & 5.43 & bl@xmath52 + & 15698.979 & 1.887 & -2.060 & 4.52 & 4.69 & 5.03 & 5.13 & bl@xmath52 + & 15715.573 & 1.873 & -1.250 & 4.57 & 4.71 & 5.14 & bl@xmath52 & 4.70 + & 16635.161 & 2.345 & -1.807 & 4.65 & bl@xmath52 & 5.10 & 5.46 & 4.90 + * v i * & 15924 .
& 2.138 & -1.175 & 3.61 & 3.66 & 3.86 & 4.18 & bl@xmath52 + * cr i * & 15680.063 & 4.697 & + 0.179 & 5.08 & 5.38 & 5.62 & 6.11 & 5.20 + & 15860.214 & 4.697 & -0.012 & 5.14 & 5.39 & 5.57 & 6.16 & 5.54 + * mn i * & 15159 .
& 4.889 & + 0.544 & 4.88 & 5.19 & bl@xmath52 & 5.82 & bl@xmath52 + & 15217 .
& 4.889 & + 0.414 & 4.91 & 5.26 & 5.47 & 5.83 & 5.32 + & 15262 .
& 4.889 & + 0.345 & 4.78 & 5.08 & 5.21 & 5.71 & 5.30 + * co i * & 16757.7 & 3.409 & -1.230 & 4.50 & 4.73 & 4.85 & 5.23 & 4.50 + * ni
i * & 15605.680 & 5.299 & -0.376 & 5.76 & 6.08 & 6.18 & 6.69 & 6.05 + & 15632.654 & 5.305 & -0.106 & 5.82 & 5.99 & 6.20 & 6.65 & 6.02 + & 16584.439 & 5.299 & -0.528 & 5.71 & 6.00 & 6.21 & 6.58 & 6.12 + & 16589.295 & 5.469 & -0.600 & 5.81 & 6.03 & 6.23 & 6.72 & 5.98 + & 16673.711 & 6.029 & + 0.317 & 5.72 & bl@xmath52 & bl@xmath52 & 6.39 & 6.12 + & 16815.471 & 5.305 & -0.606 & 5.79 & 5.94 & 6.06 & 6.58 & 5.95 + & 16818.760 & 6.039 & + 0.311 & 5.77 & bl@xmath52 & 6.22 & bl@xmath52 & 5.95 + * cu i * & 16005.7 & 5.348 & -0.131 & 3.55 & 4.01 & 4.18 & 4.41 & 4.20 + cccccc fe & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + @xmath4c & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + @xmath4c/@xmath5c & [email protected] & 15@xmath242 & 12@xmath242 & 20@xmath243 & 27@xmath246 + @xmath6n & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + @xmath7o & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + mg & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + al & [email protected] & [email protected] & 6.45 & [email protected] & 6.10 + si & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + k & [email protected] & [email protected] & 5.18 & 5.63 & 5.20 + ca & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + ti & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + v & 3.61 & 3.66 & 3.86 & 4.18 & ... + cr & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + mn & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + co & 4.50 & 4.73 & 4.85 & 5.23 & 4.50 + ni & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + cu & 3.55 & 4.01 & 4.18 & 4.41 & 4.20 + ccccccc fe i & + 0.015 & + 0.003 & -0.056 & + 0.032 & 0.066 & 0.087 + co & + 0.032 & + 0.072 & -0.011 & + 0.041 & 0.089 & 0.167 + cn & -0.018 & + 0.064 & -0.006 & + 0.038 & 0.077 & 0.127 + oh & + 0.124 & -0.022 & -0.079 & + 0.111 & 0.186 & 0.363
+ mg i & + 0.017 & -0.031 & -0.042 & + 0.029 & 0.062 & 0.098 +
al i & + 0.055 & -0.077 & -0.052 & + 0.027 & 0.111 & 0.207
+ si i & + 0.003 & -0.013 & -0.039 & + 0.038 & 0.056 & 0.065 + k i & + 0.042 & -0.046 & -0.048 & + 0.014 & 0.081 & 0.143 + ca i & + 0.045 & -0.046 & -0.032 & + 0.017 & 0.074 & 0.143 + ti i & + 0.082 & -0.004 & -0.111 & + 0.015 & 0.139 & 0.168 + v i & + 0.076 & + 0.009 & -0.054 & + 0.011 & 0.094 & 0.132 + cr i & + 0.038 & -0.018 & -0.031 & + 0.016 & 0.055 & 0.100 + mn
i & + 0.032 & -0.057 & -0.056 & + 0.041 & 0.095 & 0.163 + co
i & + 0.026 & + 0.033 & -0.075 & + 0.034 & 0.092 & 0.136
+ ni i & -0.005 & + 0.017 & -0.043 & + 0.031 & 0.056 & 0.068 + cu i & + 0.001 & + 0.032 & -0.013 & + 0.019 & 0.039 & 0.044 + | high - resolution h - band spectra of five bright field k , m , and ms giants , obtained from the archives of the kitt peak national observatory ( kpno ) fourier transform spectrometer ( fts ) , are analyzed to determine chemical abundances of 16 elements . the abundances were derived via spectrum synthesis using the detailed linelist prepared for the sdss iii apache point galactic evolution experiment ( apogee ) , which is a high - resolution near - infrared spectroscopic survey to derive detailed chemical abundance distributions and precise radial velocities for 100,000 red giants sampling all galactic stellar populations .
the red giant sample studied here was chosen to probe which chemical elements can be derived reliably from the h - band apogee spectral region .
these red giants consist of two k - giants ( @xmath0 boo and @xmath1 leo ) , two m - giants ( @xmath2 and and @xmath3 oph ) , and one thermally - pulsing asymptotic giant branch ( tp - agb ) star of spectral type ms ( hd 199799 ) .
measured chemical abundances include the cosmochemically important isotopes @xmath4c , @xmath5c , @xmath6n , and @xmath7o , along with mg , al , si , k , ca , ti , v , cr , mn , fe , co , ni , and cu .
the k and m giants exhibit the abundance signature of the first dredge - up of cn - cycle material , while the tp - agb star shows clear evidence of the addition of @xmath4c synthesized during @xmath8he - burning thermal pulses and subsequent third dredge - up .
a comparison of the abundances derived here with published values for these stars reveals consistent results to @xmath90.1 dex .
the apogee spectral region and linelist is , thus , well - suited for probing both galactic chemical evolution , as well as internal nucleosynthesis and mixing in populations of red giants via high - resolution spectroscopy . |
during recent years supersymmetric field theories have been investigated using d - branes on a singular manifold such as orbifolds and a conifold . in this
setup field theories arise as world volume theories of the d - branes .
aspects of field theories in question are encoded in geometric information of the singularity . on the other hand
there has been another approach of construction of supersymmetric field theories using branes which appear in string theory . in this framework quantities of field theories
are determined by configurations of branes .
this approach has an advantage that aspects of field theories can be visualized .
for some cases relations between the two approaches have been discussed , and investigations on brane configurations are also helpful to the study of geometries around singularities probed by d - branes .
in this paper we study brane realizations of supersymmetric field theories corresponding to d - branes on an orbifold @xmath0 with @xmath1 a nonabelian finite subgroup of @xmath2 .
finite subgroups of @xmath2 are classified into ade like series just like the finite subgroups of @xmath8 . for a - type subgroups , which are abelian ,
d - branes on the orbifold were investigated in @xcite and brane configurations corresponding to this case are known as brane box models @xcite . for nonabelian cases ,
gauge groups and field content were investigated in @xcite .
brane configurations for d - type subgroups @xmath9 and @xmath10 were also proposed in @xcite .
the crucial point of the construction of the configuration is the correspondence between quiver diagrams of @xmath1 and brane configurations for @xmath11 . according to the fact that the quiver diagram of @xmath12 ( @xmath10 ) is a @xmath13 ( @xmath14 ) quotient of the quiver diagram of @xmath15 , the configuration for d - branes on @xmath16 ( @xmath17 ) is obtained from the configuration for d - branes on @xmath18 by taking @xmath13 ( @xmath14 ) quotient . in order
that the @xmath13 quotient can be defined , brane configuration for @xmath7 must have @xmath13 symmetry . in @xcite ,
the @xmath13 symmetry was realized by a web of @xmath19 5-branes .
it is interesting to study whether similar approach can be applied to the cases in which @xmath1 are exceptional type subgroups of @xmath2 .
we can see that quiver diagrams of such subgroups can be obtained from that of @xmath20 by certain quotients : for instance , the quiver diagram of an exceptional type subgroup @xmath21 is obtained by a quotient by the tetrahedral group . from the correspondence between quiver diagrams and brane configurations , we expect that a brane configuration for @xmath22 is a quotient of the configuration for @xmath23 by the tetrahedral group .
it is however difficult to realize such a quotient on the brane configuration for @xmath23 since it has essentially two - dimensional structure and it is not clear how the group acts on it .
the situation is similar to the case of the brane configuration for @xmath5 . in that case
, we constructed a brane configuration for @xmath7 with a @xmath13 symmetry in order that we can define a @xmath13 quotient . in the present case
, it would be natural that the brane configuration for @xmath23 on which the tetrahedral group acts has three - dimensional structure and manifest symmetry under the tetrahedral group . in this paper , as a step to the construction of brane configurations corresponding to the e - type subgroups , we propose brane configuraions for @xmath16 and @xmath17 with three - dimensional structure .
the point of the construction is to lift the quiver diagram of @xmath24 , which has essentially two - dimensional structure , into three - dimensions .
the configuration corresponding to this quiver diagram consists of three kinds of ns5-branes intersecting each other and d4-branes whose three directions are bounded by ns5-branes .
naively , the configuration seems to be inappropriate since the number of supersymmetries of the configuration is half of the number the field theory should have : the configuration have two supercharges while field theories of d - branes on an orbifold @xmath11 with @xmath25 have four supercharges .
we argue , however , that the supersymmetry of the field theory on d - branes is enhanced and the number of supersymmetries is the same as that of d - branes on @xmath11 with @xmath25 . by taking quotients on the configuration , brane configurations for @xmath16 and @xmath17 can be constructed .
the organization of this paper is as follows . in section 2
, we present three - dimensional realization of the quiver diagrams of @xmath26 and @xmath10 . in section 3 , we first review relations between quiver diagrams and brane configurations . based on the relations , we construct brane configuration corresponding to d1-branes on orbifolds @xmath7 , @xmath5 and @xmath6 .
we also argue the enhancement of supersymmetries of the field theory on d - branes . in section 4 ,
we comment on a possibility of constructing configurations for @xmath11 with @xmath1 an e - type subgroup of @xmath2 based on the configuration for @xmath23 constructed in section 3 .
in this section , we draw quiver diagrams of finite subgroups of @xmath2 in three - dimensional form .
a quiver diagram of a finite group @xmath1 , which consists of nodes and arrows , represents algebraic structure of irreducible representations of @xmath1 , @xmath27 here @xmath28 denotes irreducible representations of @xmath1 and @xmath29 is a faithful three - dimensional representation of @xmath1 .
each node of the quiver diagram represents each irreducible representation of @xmath1 and the coefficient @xmath30 is the number of arrows starting from the @xmath31-th node and ending at the @xmath32-th node .
we first present the quiver diagram of an abelian subgroup @xmath24 of @xmath2 .
irreducible representations of @xmath15 consist of @xmath33 one - dimensional representations @xmath34 with @xmath35 .
we assign an irreducible representation @xmath34 to points of a two - dimensional lattice @xmath36 represented as @xmath37 if we choose the three - dimensional representation to be @xmath38 the decomposition of the product @xmath29 and @xmath34 becomes @xmath39 this equation implies that there are three arrows which start from the node @xmath40 ; the three end points are @xmath41 , @xmath42 and @xmath43 .
the quiver diagram of the group @xmath24 is depicted in figure [ fig : quiver9 ] .
now we redraw the quiver diagram of figure [ fig : quiver9 ] in three - dimensional form .
the idea is that we consider the quiver diagram of figure [ fig : quiver9 ] as a projection of a certain quiver diagram in three - dimensional space onto a certain plane . concretely speaking
, we consider a part of the quiver diagram in figure [ fig : quiverunit](a ) as a projection of a cube in figure [ fig : quiverunit](b ) along a diagonal direction of the cube . under this consideration , the assignment ( [ eq : assign ] ) of the irreducible representations to a two - dimensional lattice @xmath44
is translated into the assignment of @xmath34 to points on a three - dimensional lattice @xmath45 represented as @xmath46 where @xmath47 is an integer .
as indicated in figure [ fig : quiverunit](b ) , three arrows going from the nodes @xmath48 have end points at @xmath49 , @xmath50 and @xmath51 .
an example of the quiver diagram is depicted in figure [ fig : quiver9 - 3 ] .
note that the quiver diagram is uniform along the direction @xmath52 , which becomes a key point in the discussion of supersymmetry enhancement . as discussed in @xcite , the quiver diagram of the group @xmath12
is obtained by a @xmath13 quotient of the quiver diagram of @xmath24 .
the @xmath13 acts on the label of irreducible representations as @xmath54 if the three nodes related by the @xmath13 action correspond to different representations , the @xmath13 quotient means that the three nodes must be identified . for instance , the three nodes labeled by ( 3,1 ) , ( 3,2 ) and ( 2,1 ) must be identified if @xmath55 .
such nodes represent a three - dimensional irreducible representation of @xmath12 . for representations invariant under the @xmath13 action such as @xmath56
, the @xmath13 quotient must be understood as a split of the node into three nodes , each of which represents a one - dimensional irreducible representation of @xmath12 .
the quiver diagram of the group @xmath12 is depicted in figure [ fig : quiver48](a ) .
we now consider a three - dimensional realization of the quiver diagram .
the idea is the same as the two - dimensional realization of the quiver diagram .
that is , the quiver diagram of the group @xmath12 is obtained by a @xmath13 quotient of the quiver diagram of @xmath24 given in figure [ fig : quiver9 - 3 ] .
the action of @xmath13 on the lattice points of @xmath45 is defined as follows .
@xmath57 it means that @xmath13 acts on the three - dimensional space as a @xmath58 rotation along the line specified by @xmath52 . combining with the assignment ( [ eq : assign-3 ] ) of the irreducible representations of @xmath24 on the lattice @xmath45
, one can see that the action ( [ eq : z3 - 3 ] ) is equivalent to the action of @xmath13 given in ( [ eq : z3 ] ) .
three - dimensional version of the quiver diagram is depicted in figure [ fig : quiver48](b ) .
now we comment on the speciality when @xmath59 is an integer . when @xmath59 is not an integer , the only fixed node under the @xmath13 action is @xmath60 . on the other hand ,
when @xmath59 is an integer , there are additional fixed nodes @xmath61 and @xmath62 as one can see from figure [ fig : quiver27 - 3 ] .
the quiver diagram of the group @xmath10 is obtained by a further @xmath64 quotient on the quiver diagram of @xmath12 .
the @xmath64 acts on the label of irreducible representations as @xmath65 it means that @xmath64 acts as a reflection with respect to the line extending a diagonal direction @xmath66 . if the two nodes related by the @xmath64 action correspond to different representations , they must be identified .
for instance , the nodes ( 1,2 ) and ( 2,1 ) , both of which are three - dimensional irreducible representations of @xmath12 for @xmath55 , must be identified .
such nodes represent six - dimensional irreducible representation of @xmath10 . for nodes
invariant under the @xmath64 action , @xmath64 acts as a split the node into a certain set of nodes . for details on the irreducible representations of @xmath10 ,
see @xcite .
the quiver diagram of the group @xmath10 is depicted in figure [ fig : quiver96](a ) . in the three - dimensional version of the quiver diagram ,
@xmath64 acts on a three - dimensional lattice @xmath45 as @xmath67 it means that @xmath64 acts as a reflection with respect to the plane specified by @xmath66 .
combining with the @xmath13 action ( [ eq : z3 - 3 ] ) , it is equivalent to the reflections with respect to the planes @xmath68 and @xmath69 . the three - dimensional version of the quiver diagram of the group @xmath10 is depicted in figure [ fig : quiver96](b ) .
as discussed in @xcite , brane box type realization of d - brane gauge theory on an orbifold @xmath11 has a direct correspondence to the quiver diagram of @xmath1 . in the d - brane gauge theory on the orbifold @xmath11
, each node in the quiver diagram corresponds to a gauge group factor @xmath70 where @xmath71 is the dimension of the irreducible representation . in the brane box type realization of the gauge theory
, it corresponds to a box with @xmath71 d - branes .
arrows in the quiver diagram represent matter contents of the d - brane gauge theory on @xmath11 .
that is , an arrow from the node @xmath31 to the node @xmath32 represents a bifundamental matter transforming as @xmath72 under @xmath73 . in the brane box type configuration , it comes from an oriented open string starting from d - branes on the @xmath31-th box and ending on d - branes on the @xmath32-th box .
it implies that the @xmath31-th box and @xmath32-th box must adjoin each other .
these two boxes are separated by another brane , for example , ns 5-branes or @xmath19 5-branes . due to the orientation of such branes at the boundary ,
only one orientation of open strings are allowed and it induces a particular set of bifundamental matters @xcite .
thus the arrows in the quiver diagram indicate how to connect boxes with d - branes .
we summurize the correspondence in table [ table : correspondence ] .
.the correspondence among representation theory , gauge theory , quiver diagram and brane box model .
[ cols="^,^,^ " , ] it is also important that the configuration provides the same supersymmetry as that of the d - brane gauge theory on the orbifold .
several brane configurations satisfying these reqirement were constructed . in @xcite ,
a brane configuration is constructed for @xmath74 by using d5-branes and two kinds of ns5-branes . in @xcite ,
brane configurations were constructed for the nonabelian groups @xmath12 and @xmath10 as well as the abelian group @xmath24 using @xmath19 5-branes and d3-branes . in this section , we construct another kind of brane configurations for @xmath75 , @xmath12 and @xmath10 based on the three - dimensional version of the quiver diagrams given in the last section . in this subsection
, we construct brane configurations for @xmath7 based on the three - dimensional version of the quiver diagram given in figure [ fig : quiver9 - 3 ] .
as the nodes in the quiver diagram lie at the lattice points of @xmath45 , boxes also lie at the lattice points of @xmath45 .
the node at @xmath48 is connected to the nodes at @xmath49 , @xmath50 and @xmath51 in the quiver diagram by three outgoing arrows , so the box at @xmath48 must adjoin boxes at @xmath49 , @xmath50 and @xmath51 .
a natural brane configuration satisfying these requirement is depicted in figure [ fig : brane9 - 3a ] .
note that the cube at @xmath48 has the same label as the cube at @xmath76 .
the configuration consists of the following branes : * ns5-branes located along 012345 directions . *
ns5-branes located along 012367 directions . * ns5-branes located along 014567 directions . *
d4-branes located along 01246 directions .
d4-branes are bounded in the direction 6 by the ns5-branes , in the direction 4 by the ns5-branes , and in the direction 2 by the ns5-branes .
thus the non - compact directions of the d4-branes are 0 and 1 , and hence the low energy theory becomes two - dimensional .
the numbers written in the boxes are the labels of the irreducible representations of @xmath24 . due to the orientation of ns5-branes , only one orientation of open strings
is allowed and it gives a bifundamental matter corresponding to the arrow of the quiver diagram .
there is however a subtlety on the number of supersymmetries .
the configuration consists of four kinds of branes , each of which breaks 1/2 supersymmetries .
thus the brane cube configuration has two supercharges .
since the number of supersymmetries of the gauge theory of d - branes on @xmath11 with @xmath25 is four , it seems that the two gauge theories have different numbers of supersymmetries .
as we will discuss , however , the low energy field theory of d - branes obtained from the brane cube configuration has an enough number of supersymmetries due to a certain enhancement of supersymmetry .
to explain the supersymmetry enhancement , we derive the brane cube configuration from different point of view .
it is obtained from d1-branes on @xmath77 by performing t - duality three times . in fact , a similar but different configuration were discussed in @xcite .
both of them have a property that the configuration is uniform along one direction .
it is the key point of the supersymmetry enhancement .
we start with the d1-brane gauge theory on @xmath77 with @xmath1 an abelian subgroup of @xmath78 @xcite .
we consider @xmath79 d1-branes on @xmath80 and let the d1-branes extend along 01 directions .
the low energy effective action of the d1-brane is given by the dimensional reduction of ten - dimensional @xmath81 supersymmetric @xmath82 yang - mills theory to two dimensions .
the theory contains gauge field @xmath83 , four complex scalar fields @xmath84 ( @xmath85 ) , eight left - handed fermions @xmath86 and eight right - handed fermions @xmath87 .
these fields transforms in the adjoint of @xmath82 .
then we project the theory onto @xmath1 invariant states . to perform the projection
, one must define how @xmath1 acts on @xmath80 .
this is specified by a four - dimensional representation @xmath88 .
we denote the decomposition of @xmath88 into irreducible representations @xmath28 ( @xmath89 ) of @xmath1 as @xmath90 ( as @xmath1 is abelian , its irreducible representations are one - dimensional . ) each irreducible representation @xmath91 acts on the coordinates @xmath92 of @xmath80 as @xmath93 .
the requirement that @xmath1 is a subgroup of @xmath78 can be stated as @xmath94 where @xmath95 is the trivial representation .
if we write @xmath96 , the condition is represented as @xmath97 . to define d - branes on @xmath77
, one must also determine how @xmath1 acts on the chan - paton indices .
we define the action of @xmath1 on the chan - paton indices as the regular representation of @xmath1 .
note that the theory obtained by the projection is a two - dimensional ( 0,2 ) supersymmetric theory with the gauge group @xmath98 , where @xmath99 represents the gauge group @xmath100 corresponding to the representation @xmath28 .
the field content surviving the projection is as follows : * a gauge field @xmath101 of @xmath99 , * four complex bosons @xmath102 which transform in the @xmath103 of @xmath104 , * a left - handed dirac fermion @xmath105 which transform in the adjoint of @xmath99 , * three left - handed dirac fermions @xmath106 which transforms in the @xmath103 of @xmath107 , * four right - handed dirac fermions @xmath108 which transforms in the @xmath103 of @xmath104 , where @xmath31 runs from 1 to @xmath79 .
here @xmath106 is antisymmetric in @xmath109 , so there are six fields to each @xmath31 , three of which are independent .
there is an ambiguity to determine the three independent left - handed dirac fermions .
we take three fields @xmath110 ( @xmath111 ) be independent .
these fields form ( 0,2 ) multiplets as follows : * the gauge field @xmath101 and the left - handed fermion @xmath105 and an auxiliary field @xmath112 form a ( 0,2 ) gauge multiplet @xmath113 , @xmath114 * four complex bosons @xmath102 and four right - handed fermions @xmath108 form four ( 0,2 ) chiral multiplets
@xmath115 , @xmath116 * three left - handed fermions @xmath110 and auxiliary fields @xmath117 form three ( 0,2 ) fermi multiplets @xmath118 , @xmath119 where @xmath120 is the supersymmetric derivative and @xmath121 is a function of chiral superfields which will be defined below .
the two - dimensional ( 0,2 ) supersymmetric gauge theories are described by a lagrangian of the form , @xmath122 @xmath123 @xmath124 @xmath125 @xmath126 @xmath127 where @xmath128 is the field strength of the superspace gauge field @xmath113 , @xmath129 is the gauge covariant derivative and @xmath130 represents a fayet - iliopoulos parameter and a theta parameter of @xmath99 .
the interactions of the theory are completely defined by functions @xmath131 and @xmath132 . in the present case ,
@xmath131 and @xmath132 take the following form , @xmath133 @xmath134 they satisfy the following relation @xmath135 to realize the gauge theory in terms of brane configurations , we perform t - dualities three times along @xmath100 orbits associated with three complex planes . by a manifold with the same singularity but different asymptotics to make the radius of the @xmath100 orbits finite at infinity .
] there is an ambiguity to choose three directions out of four complex coordinates .
if we t - dualize along three compact directions associated with @xmath136 , @xmath137 and @xmath138 , we obtain a brane cube configuration depicted in figure [ fig : branec4a ] .
it consists of three kinds of ns5-branes and d4-branes .
the ( 0,2 ) chiral multiplet @xmath115 transforming in the @xmath139 comes from open strings extending from the box labeled by @xmath31 to the box labeled by @xmath140 .
we can represent the four chiral multiplets @xmath141 , @xmath142 , @xmath143 and @xmath144 by arrows with components @xmath145 , @xmath146 , @xmath147 and @xmath148 .
the ( 0,2 ) fermi multiplet @xmath149 transforming in the @xmath150 comes from open strings from the box @xmath31 to the box @xmath151 .
we can represent the three fermi multiplets @xmath152 , @xmath153 and @xmath154 by arrows with components @xmath155 , @xmath156 and @xmath157 .
the arrows are depicted in figure [ fig : quiverc4a ] .
now we restrict the model to the cases with @xmath158 .
then the condition for @xmath159 in ( [ eq : su(4 ) ] ) becomes @xmath160 it means that the orbifold @xmath77 is @xmath161 where @xmath1 is a subgroup of @xmath2 . in this case ,
the supersymmetry of the two - dimensional theory is ( 2,2 ) instead of ( 0,2 ) , and the ( 0,2 ) multiplets are combined into ( 2,2 ) multiplets as follows : * the ( 0,2 ) gauge multiplet @xmath113 and the adjoint ( 0,2 ) chiral multiplet @xmath144 are combined to form ( 2,2 ) vector multiplet @xmath162 .
( @xmath144 transforms in the adjoint of @xmath99 due to @xmath158 . ) @xmath163 * the ( 0,2 ) chiral multiplets @xmath164 and the ( 0,2 ) fermi multiplets @xmath118 form three ( 2,2 ) chiral multiplets @xmath165 .
( @xmath118 transforms in the @xmath166 due to @xmath158 . )
@xmath167 where @xmath168 and @xmath169 . in the brane cube configuration in figure [ fig : branec4a ] , the cube at @xmath48 and the cube at @xmath76 are equivalent due to the condition @xmath158 .
this is nothing but the configuration considered in figure [ fig : brane9 - 3a ] if we take @xmath170 , @xmath171 , @xmath172 and @xmath173 .
the ( 2,2 ) chiral multiplets @xmath141 , @xmath142 and @xmath143 can be represented by arrows with components @xmath145 , @xmath146 and @xmath147 .
the ( 2,2 ) vector multiplet @xmath113 can be represented by an arrow with components @xmath148 .
the actions @xmath174 ( [ eq:02gauge ] ) and @xmath175 part of @xmath176 ( [ eq:02ch ] ) of the ( 0,2 ) supersymmetric theory are combined into @xmath174 of the ( 2,2 ) supersymmetric theory @xmath177 with some interaction terms . here
@xmath178 is the field strength of the ( 2,2 ) superspace gauge field @xmath162 .
remaining part of @xmath176 ( [ eq:02ch ] ) and @xmath179 ( [ eq:02f ] ) become @xmath176 of the ( 2,2 ) supersymmetric theory , @xmath180 @xmath181 ( [ eq:02j ] ) corresponds to the superpotential term @xmath182 with @xmath183 , @xmath184 three arrows representing the three chiral multiplets appearing in the superpotential form a triangle up to the identification along the diagonal direction .
note that the superpotential @xmath185 is related to @xmath131 as @xmath186 and the equation ( [ eq : je ] ) implies the gauge invariance of the superpotential . combining with the d - term part @xmath187 one obtain two - dimensional ( 2,2 ) supersymmetric gauge theory .
thus although the brane cube configuration has only two supercharges the two - dimensional field theory on d - branes has an enough number of supersymmetries .
to understand the reason for the supersymmetry enhancement , it is useful to compare the configuration of figure [ fig : brane9 - 3a ] with a brane configuration obtained by another t - duality . as noted earlier , there is an ambiguity to choose three directions to perform t - duality out of four complex coordinates .
if we t - dualize along three compact directions associated with @xmath136 , @xmath137 and @xmath188 , we obtain a brane cube configuration depicted in figure [ fig : branec4b ] .
the chiral multiplet @xmath115 transforming in the @xmath139 comes from open strings from the box @xmath31 to the box @xmath140 .
we can represent the four chiral multiplets @xmath141 , @xmath142 , @xmath143 and @xmath144 by arrows with components @xmath145 , @xmath146 , @xmath148 and @xmath147 .
the fermi multiplet @xmath149 transforming in the @xmath150 comes from open strings from the box @xmath31 to the box @xmath151 .
we can represent the three fermi multiplets @xmath152 , @xmath153 and @xmath154 by arrows with components @xmath189 , @xmath190 and @xmath157 .
the arrows are depicted in figure [ fig : quiverc4b ] .
we would like to emphasize that the two configurations in figure [ fig : branec4a ] and figure [ fig : branec4b ] give the same field theory since both of them are related to the d - branes on @xmath77 by t - dualities and hence the two configurations are related by t - duality . in the brane cube configuration given in figure [ fig : branec4b ] , the cube at @xmath48 and the cube at @xmath51 are equivalent .
if we set @xmath158 , the ( 2,2 ) chiral multiplets @xmath141 , @xmath142 and @xmath143 can be represented by arrows with components @xmath145 , @xmath146 and @xmath148 , while the ( 2,2 ) vector multiplet @xmath113 can be represented by an arrow with components @xmath147 .
this is the model considered in section 3.2 of @xcite .
that is , if we take @xmath170 , @xmath171 , @xmath172 and @xmath173 , we obtain the configuration in figure [ fig : brane9 - 3b ] . in the configuration of figure [ fig : brane9 - 3b ]
, we can remove ns5-branes extending along 012345 directions without changing the matter contents and gauge groups since the configuration of figure [ fig : brane9 - 3b ] is trivial along the direction 8 .
the resulting configuration is equivalent ( up to a certain t - duality ) to the usual brane box model @xcite with two kinds of ns5-branes .
therefore the field theory obtained from the configuration in figure [ fig : brane9 - 3b ] is equivalent to the field theory obtained from the usual brane box model , so the field theory has four supercharges .
as stated above , the configuration of figure [ fig : brane9 - 3a ] is related to figure [ fig : brane9 - 3b ] t - duality , the field theory realized by the configuration has the enough number of supersymmetries .
note that the configuration of figure [ fig : brane9 - 3a ] is more natural than that of figure [ fig : brane9 - 3b ] in the sense that the three coordinates of @xmath191 ( three chiral multiplets @xmath192 ) are treated equivalently .
this is the key point of the construction of the configuration for @xmath5 . due to the symmetric assignment of the multiplets
, however , enhancement of supersymmetry becomes rather nontrivial .
the brane configuration for the orbifold @xmath5 is obtained from that for @xmath7 by a @xmath13 quotient .
the @xmath13 acts as a @xmath193 rotation along the line @xmath194 .
the brane configuration is given in figure [ fig : brane27 - 3 ] .
the reason that the configuration gives the structure of the irreducible representations of @xmath12 is the same as in @xcite .
the gauge group coming from each box depend on whether the box include fixed points of the @xmath13 action .
when @xmath59 is not an integer , only the boxes with index ( 0,0 ) include fixed line of the @xmath13 action .
when @xmath59 is an integer , the boxes with indices @xmath61 and @xmath62 also include fixed lines in addition to ( 0,0 ) . for such boxes , we must take a @xmath13 quotient , which leads the gauge group to be @xmath195 since the gauge field on the box takes the following form , @xmath196 it implies that such boxes correspond to a sum of three one - dimensional representations of @xmath12 . on the other hand , for the boxs which do not include the fixed lines of the @xmath13 action , three d - branes simply pile up and give the gauge group @xmath197 .
it implies that such boxes correspond to three - dimensional irreducible representations of @xmath12 .
one can see that the quotienting procedure precicely reproduces the structure of the irreducible representations of @xmath12 .
we can also verify that matter contents obtained after the @xmath13 quotient coincide with those specified by the quiver diagram of @xmath12 .
the brane configuration for the orbifold @xmath6 is obtained from that for @xmath5 by a @xmath64 quotient .
the @xmath64 acts as a reflection with respect to the plane @xmath198 as indicated in the quiver diagram of @xmath10 in figure [ fig : quiver96](b ) . combining with the @xmath13 identification for the configuration corresponding to @xmath5 , it is equivalent to the reflections with respect to the planes @xmath199 and @xmath200 .
the brane configuration is given in figure [ fig : brane54 - 3 ] .
one can verify that the configuration reproduces the structure of the quiver diagrams of @xmath10 .
for details on the structure of the irreducible representations , see @xcite .
in this paper , we have proposed brane cube configurations corresponding to d1-branes on nonabelian orbifolds @xmath0 with @xmath25 . this is based on the three - dimensional version of the quiver diagrams of @xmath1 .
the configuration consists of three kinds of ns5-branes and d4-branes .
since the d4-branes are bounded by ns5-branes along three directions , the low energy theory becomes two - dimensional . due to the fact that the configuration is uniform along a diagonal direction , supersymmetry of the field theory
is enhanced twice as many as naively expected . the original motivation to consider configurations with three - dimensional structure comes from the study of configurations for an orbifold @xmath11 with @xmath1 an e - type subgroup of @xmath2 .
as mentioned in introduction , the quiver diagram of @xmath21 is a quotient of the quiver diagram of @xmath20 by the tetrahedral group . according to the correspondence between quiver diagrams and brane configurations for @xmath11
, we expect that a brane configuration for @xmath22 is obtained from the configuration for @xmath23 given in figure [ fig : brane27 - 3 ] by taking a quotient by the tetrahedral group .
in fact , the configuration in figure [ fig : brane27 - 3 ] has structure of the tetrahedral group : the configuration consists of cubes , which are in a sence dual to tetrahedra , so we can define an action of the tetrahedral group on the configuration . it is now under investigation whether such a quotient actually reproduces required properties of the gauge theories .
finally we would like to compare the configurations in this paper with those given in @xcite .
the configuration given in @xcite consists of a web of @xmath19 5-branes and d3-branes . as the d3-branes are bounded along two - directions by @xmath19 5-branes , they provide two - dimensional field theory , which coincides with the brane cube configurations .
thus it seems that we should start with _
d1_-branes on @xmath201 to realize @xmath13 symmetry of brane configurations . from this viewpoint , the fact that the brane box model gives four - dimensional theory is owing to the speciality of the abelian case in which an explicit @xmath13 symmetry is not necessary .
as we have discussed , the configurations given in @xcite have esentially two - dimensional structure , while the brane cube configurations are three - dimensional .
it is interesting to see whether there is a relation like duality between the two types of configurations .
if such a relation can be found , it may be possible to construct brane configurations corresponding to @xmath11 with @xmath1 an e - type subgroup based on the brane configurations made out of a web of @xmath19 5-branes . in @xcite
, it was also argued that the brane configurations are dual to toric diagrams of @xmath11 and that three - dimensional mckay correspondence @xcite-@xcite may be understood as t - duality .
we hope that the three - dimensional version of the quiver diagrams and the brane cube configurations provide some hints on investigations along these lines . | we study d - branes on three - dimensional orbifolds @xmath0 where @xmath1 are finite subgroups of @xmath2 .
the quiver diagram of @xmath3 can be expressed in three - dimensional form . according to the correspondence between quiver diagrams and brane configurations , we construct a brane configuration for @xmath4 which has essentially three - dimensional structrue .
brane configurations for nonabelian orbifolds @xmath5 and @xmath6 are obtained from that for @xmath7 by certain quotienting procedure .
22.5cm16.3cm-.1 cm -.4 cm # 1#2 # 1#2 ' '' '' height#2pt ' '' '' height#2pt # 1#2 # 1#2 |
quantum theory has the remarkable feature that certain dynamical degrees of freedom may become `` slaved '' and thus lose their autonomous status .
the origin of this phenomenon is a separation , both in space and time , into slow and fast degrees of freedom .
the fast modes quickly adapt to the slow modes which in turn are governed by a suitable effective hamiltonian .
this mechanism is called adiabatic decoupling . as paradigm we mention the motion of nuclei .
the electronic degrees of freedom rapidly adjust to the state of lowest energy at given positions of the nuclei and the electronic energy band serves as effective potential in the hamiltonian for the nuclei .
this born - oppenheimer approximation is the basis for the dynamics of molecules and , as a consequence , also for the microscopic theory of classical fluids .
there are many other examples of a similar structure .
a very widely studied case are electrons moving in the periodic crystal potential , which defines the short scale .
the envelope of the electronic wave function is governed by an effective hamiltonian obtained from the peierls substitution , in which the band energy is taken as effective kinetic energy . for an electron coupled to the quantized radiation field
the photons are the fast degrees of freedom and the dynamics of the electron is governed by an effective hamiltonian accounting for spin precession . these and other systems have been studied extensively by model specific approximate methods without realizing that they share a common structure .
in fact , all the examples given can be molded into the generic form @xmath5 here @xmath1 is an operator - valued function on the classical phase space @xmath6 of the slow degrees of freedom with @xmath7 a position like and @xmath8 a momentum like variable .
@xmath1 is self - adjoint and acts on the hilbert space @xmath9 of internal " fast degrees of freedom .
after quantization @xmath10 becomes the hamiltonian of our system . to properly define it one has to specify an ordering of the operators @xmath11 , for which we will adopt the weyl quantization rule as to be explained in full detail in the appendix
@xmath12 is a wave function on @xmath13 with values in @xmath14 .
thus the quantum mechanical hilbert space of states is @xmath15 .
finally @xmath3 is a dimensionless parameter which controls the scale separation , @xmath16 .
examples will be given in due course and we only remark that , in general , ( [ a ] ) is already the result of a proper identification of the slow degrees of freedom .
for example , in nonrelativistic qed @xmath7 stands for the position of the electron , whereas @xmath8 is the total momentum of the electron and photons . for an electron in a periodic potential ( [ a ] )
is the hamiltonian in the crystal momentum basis with @xmath17 standing for the bloch momentum @xmath18 in the first brillouin zone , see @xcite .
our goal is to construct approximate solutions to ( [ a ] ) , a task which we divide into four steps . *
( i ) almost invariant subspace . *
the adiabatic decoupling can be traced to a spectral property of @xmath1 .
it is assumed , and it can be proved in many particular cases , that for each @xmath19 the spectrum , @xmath20 , of @xmath1 can be decomposed into a
relevant " part @xmath21 and a remainder in such a way that the relevant energy band @xmath22 is separated from the remainder @xmath23 by a gap , i.e. by a corridor of finite width . in many cases of interest
the relevant part of the spectrum consists of a single , possibly degenerate eigenvalue and the relevant energy band is the graph of a smooth function .
of course , a given @xmath1 could have several such relevant energy bands and we suppose one of them to be singled out , e.g. , through the initial condition . to the relevant energy band of @xmath24 one associates a subspace @xmath25 of @xmath26 with the property that if @xmath27 , then @xmath28 up to an error which is smaller than any power of @xmath3 . for this reason @xmath25
is called an almost invariant subspace . if @xmath29 denotes the weyl quantization of @xmath24 , then @xmath30={\mathcal{o}}({\varepsilon}^\infty)$ ] , where @xmath31 is the orthogonal projector onto the closed subspace @xmath25 . of course
, in practice , approximations to @xmath31 are constructed , for which at order @xmath32 the error is @xmath33 . *
( ii ) reference hilbert space . * clearly , on @xmath25 the dynamics is generated by the diagonal hamiltonian @xmath34 , up to @xmath35 . while properly defined , for a further analysis of the effective dynamics inside @xmath25 a representation in terms of adapted coordinates for the slow degrees of freedom
is more powerful .
we call the corresponding hilbert space @xmath36 the reference space . by definition @xmath37
is independent of @xmath3 .
the next task is thus to construct a linear map @xmath38 , which is unitary in the sense that @xmath39 and @xmath40 . *
( iii ) effective hamiltonian . *
the effective hamiltonian is defined through @xmath41 as operator on @xmath37 .
it is unitarily equivalent to the diagonal hamiltonian , but acts on a simpler space .
while @xmath42 is still a rather abstract object , it can be expanded in powers of @xmath3 , which is the adiabatic perturbation theory of the title .
already the lowest order approximations provide a wealth of information on the motion of the slow degrees of freedom . *
( iv ) semiclassical limit .
* in many applications of interest the effective hamiltonian is of a form which allows for a semiclassical analysis .
the simplest situation is a relevant band which consist only of a single , possibly degenerate , eigenvalue .
then the effective hamiltonian has a scalar principle symbol and the semiclassical analysis is straightforward .
we are certainly not the first ones to investigate approximate solutions to ( [ a ] ) and we have to explain which parts of the program outlined above have been achieved before and which parts are our novel contribution .
in addition there have been other approaches to ( [ a ] ) on which we briefly comment below . to our knowledge , the notion `` almost invariant subspace '' was first coined by nenciu @xcite in the context of gauge invariant perturbation theory . in the context of space - adiabatic problems brummelhuis and nourrigat @xcite
construct @xmath31 for the particular case of the dirac equation and martinez and sordoni @xcite based on @xcite consider born - oppenheimer type hamiltonians .
the general scheme for the construction of @xmath31 is sketched in nenciu and sordoni @xcite and applied to the matrix - valued klein - gordon equation .
our construction is based on the one in @xcite , but differs in a few technical details . as in the case of @xmath31
we construct the unitary @xmath43 in several steps .
first we compute order by order the formal symbol of @xmath43 , which , after quantization , gives rise to an `` almost unitary '' operator .
finally the almost unitary is modified to yield a true unitary exactly intertwining @xmath25 and @xmath37 .
our method is specifically designed to deal also with problems as the dirac equation and the bloch electron with external magnetic fields , where the projector @xmath31 has no limit for @xmath44 , see remark [ nagyrem ] . while the specific application and the proof are new , the general idea to construct a pseudodifferential operator which is almost unitary and diagonalizes a given pseudodifferential operator has a long tradition , @xcite section 7 and references therein , @xcite,@xcite .
the method of successive diagonalization is also prominent in the physics literature , for example @xcite in the derivation of the pauli equation and its corrections , @xcite for periodic schrdinger operators , @xcite for the dirac equation , @xcite for born - oppenheimer type hamiltonians .
our central result is the expansion of the effective hamiltonian @xmath42 .
we provide a scheme which is applicable in general and work out explicitly the expansion as @xmath45 . in particular , for a relevant band consisting of a single eigenvalue , @xmath46 is the peierls substitution and @xmath47 contains among other things the information on geometric phases . since @xmath42 has , in general , a matrix - valued symbol , we discuss and apply some results on the semiclassical limit for matrix - pseudodifferential operators with scalar principal symbol .
we include this part to make the paper self - contained and to demonstrate that in many cases of interest the motion of the slow degrees of freedom can be approximately described through the appropriate classical hamiltonian flow .
surely the reader will have noticed that our program makes no mention of the initial conditions for ( [ a ] ) .
the reason is simply that our estimates are uniform on @xmath25 , respectively on @xmath37 .
physically this partial independence from the initial conditions is most welcome , because in general they can not be controlled so easily .
on the other hand , in the approach of hagedorn and joye @xcite one constructs for a specific @xmath3-dependent @xmath48 the wave packet @xmath49 , which solves ( [ a ] ) up to @xmath35 or even @xmath50 .
hagedorn and joye study born - oppenheimer hamiltonians .
periodic schrdinger operators are considered in @xcite , who employ a related multi - scale analysis .
whereas in the other approaches the adiabatic and the semiclassical limit are taken simultaneously , both limits are clearly separated in our results .
we consider this an important conceptual advantage . to give a brief outline of our paper .
in section 2 we explain the precise assumptions on the hamiltonian @xmath24 and construct the almost invariant subspace @xmath25 .
the reference space @xmath37 and the unitary intertwining it with @xmath25 are explained section 3 .
the central part of our paper is the expansion of the effective hamiltonian in section 4 . in particular , we work out the expansion for born - oppenheimer hamiltonians including order @xmath51 . we also show that by the howland trick the standard time - adiabatic theory can be subsumed under ( [ a ] ) .
the semiclassical analysis for the effective hamiltonian is summarized in section 5 .
last but not least , in section 6 we discuss the dirac equation with slowly varying external vector potentials , since it is the simplest hamiltonian for which the full generality of our approach is needed and yields interesting physical results . in this case @xmath52 and the classical symbol has two two - fold degenerate energy bands , one for the electron and one for the positron .
thus the reference hilbert space e.g. for the electron band is @xmath53 .
the effective hamiltonian is determined including order @xmath3 .
the principal part @xmath46 describes the translational motion through the peierls substitution and the subprincipal part @xmath47 yields the spin precession as governed by the bmt equation .
since the external vector potential appears through minimal coupling , the projection @xmath54 on the electron subspace depends nontrivially on both coordinates , in contrast to born - oppenheimer and time - adiabatic theory , where the relevant projection @xmath54 depends only on one of the canonical coordinates .
we end the paper with some concluding remarks in section 7 and with an appendix reviewing some results on pseudodifferential calculus with operator - valued symbols .
* acknowledgments : * we are grateful to andr martinez and gheorghe nenciu for helpful discussions , which in part initiated our work .
we thank berndt thaller for his remarks on the problem of observables for the dirac equation and patrick grard for pointing out to us reference @xcite . s.t .
also thanks raymond brummelhuis , george hagedorn , and alain joye for valuable comments and their interest in this work .
is grateful to fabio pioli for a useful remark about fiber bundles .
space - adiabatic perturbation theory deals with quantum systems in which it is possible to distinguish between fast and slow degrees of freedom .
in particular we assume that the hilbert space @xmath26 admits a natural decomposition as @xmath55 @xmath56 , where @xmath57 is the state space for the slow degrees of freedom and @xmath56 is the state space for the fast degrees of freedom . as the second structural ingredient
we require that the hamiltonian is given as the quantization of a @xmath58-valued function on the classical phase space @xmath59 of the slow degrees of freedom .
hence we need to consider the generalization of the usual quantization rules to the case of @xmath58-valued functions on @xmath59 .
this theory is well covered in the literature , see for example @xcite .
still , for the convenience of the reader , we provide a self - contained review of the basic results in the appendix , where we also introduce the relevant notation .
we now state the general assumptions on which the adiabatic perturbation theory will be based in the following .
let @xmath2 be a separable hilbert space , the state space for the fast degrees of freedom , and @xmath60 .
the hamiltonian @xmath29 of the full system is given as the weyl quantization of a semiclassical symbol @xmath61 .
we assume that @xmath29 is essentially self - adjoint on @xmath62 . a point in the classical phase space @xmath59
is denoted by @xmath63 .
the adiabatic decoupling relies on a gap condition for the principal symbol @xmath64 of @xmath24 . * * condition ( gap)**@xmath65 for any @xmath66 the spectrum @xmath67 of @xmath68 contains a relevant subset @xmath69 which is uniformly separated from its complement @xmath70 by a gap . more precisely there are two continuous functions @xmath71 @xmath72 ( with @xmath73 ) such that : 1 . for every @xmath74 the spectral component @xmath69 is entirely contained in the interval @xmath75 $ ]
the distance between @xmath76 and the interval @xmath77 is uniformly bounded away from zero and increasing for large momenta , i.e. @xmath78 3 .
the width of the interval @xmath77 is uniformly bounded , i.e. @xmath79 we denote the spectral projector corresponding to @xmath69 by @xmath80 .
as explained in the introduction , one expects interband transitions to be suppressed for small @xmath3 . to prove such a property we need either one of the following assumptions to be satisfied . * * condition of increasing gap ( ig)**@xmath81*. * + let @xmath24 be an _ hermitian _
symbol in @xmath82 ( with @xmath83 and @xmath84 ) such that the principal symbol @xmath85 satisfies condition ( gap)@xmath86 with @xmath87 .
* condition of constant gap ( cg ) .
* + let @xmath24 be an _
symbol in @xmath88 such that the principal symbol @xmath85 satisfies condition ( gap)@xmath86 with @xmath89 .
note that for the case @xmath90 one can show that @xmath91 the weyl quantization of @xmath24 is essentially self - adjoint on the domain @xmath92 .
the proof is postponed to an appendix to the space - adiabatic theorem . in analogy with the usual time - adiabatic theorem of quantum mechanics ,
see section [ sta ] , we baptize the following result as _ space - adiabatic theorem_. it establishes that there are almost invariant subspaces associated with isolated energy bands . in spirit
the result is not new .
however , to our knowledge it appears in this explicit form only recently in the literature .
brummelhuis and nourrigat @xcite gave a proof for the dirac equation , martinez and sordoni @xcite considered born - oppenheimer type hamiltonians ( cf . section [ sbo ] ) based on results from @xcite , and nenciu and sordoni @xcite sketched the general scheme and applied it to a matrix - valued klein - gordon type problem . assume either ( ig)@xmath81 or ( cg ) .
let @xmath91 be the weyl quantization of @xmath24 .
then there exists an orthogonal projector @xmath93 such that @xmath94=\mathcal{o}_{0}(\varepsilon^{\infty } ) \ ] ] and @xmath95 , where @xmath96 is the weyl quantization of a semiclassical symbol @xmath97 whose principal part @xmath80 is the spectral projector of @xmath98 corresponding to @xmath69 .
[ th invariant subspace ] the subspace @xmath99 is thus an almost invariant subspace for the dynamics generated by the hamiltonian @xmath91 , i.e. @xmath100=\mathcal{o}_0({\varepsilon}^\infty|t|)$ ] , and it is associated with the spectral band @xmath69 . the terminology was introduced in @xcite .
note , however , that ran@xmath31 is , in general , _ not _ an almost invariant subspace in the sense of @xcite , since @xmath31 need not have a limit as @xmath44 .
_ note that the growth condition on the gap in ( ig)@xmath101 is stronger than one would expect from the analysis in @xcite or @xcite .
indeed , in both examples a gap which is bounded globally over phase space suffices to prove uniform adiabatic decoupling also in the presence of a hamiltonian with principal symbol increasing linearly in momentum .
more general , uniform adiabatic decoupling should hold whenever ( ig)@xmath101 is satisfied with @xmath102 . indeed
( [ tiscom ] ) follows from the following proof with slight modifications under this weaker condition on the growth of the gap .
however , this modified proof does not give @xmath103 , a fact we will make use of in the following . to avoid further complications in the presentation , we decided to state only the stronger result for the stronger growth condition .
_ * proof.*we decompose the proof into two steps .
* step i. construction of the moyal projector * in general @xmath104 is not a projector in the moyal algebra , i.e. @xmath105 . the following lemma shows that @xmath104 can be corrected , order by order in @xmath106 , so to obtain a true moyal projector @xmath107 which moyal commutes with @xmath24 .
similar constructions appeared in the context of the schrdinger equation several times in the literature @xcite .
our proof was strongly influenced by the one in @xcite , but differs in relevant details , since we consider different symbol classes .
it relies on the construction of the _ local _ moyal resolvent of @xmath108 .
the construction of the global inverse of an elliptic symbol , often called the parametrix , is well known @xcite .
[ prop moyal proj ] assume either ( ig)@xmath81 or ( cg ) .
then there exists a _
unique _ formal symbol @xmath109 such that @xmath80 is the spectral projector of @xmath98 corresponding to @xmath110 with the following properties : 1 .
@xmath111 2 .
@xmath112 3 .
@xmath113_{\#}:= h\ , \#\ , \pi -\pi\ , \#\ , h=0.$ ] * proof .
* we give the proof under the assumption ( ig)@xmath81 .
the proof under assumption ( cg ) is simpler , since all the symbols which appear belong to @xmath114 .
we first provide a constructive scheme for the special case where @xmath115 is an eigenvalue , which , at the same time , proves uniqueness of @xmath116 in the general case .
it follows basically the construction as given in @xcite .
the reason for including this scheme is that the aim of adiabatic perturbation theory is , in particular , to give an as simple as possible recipe for explicitly computing the relevant quantities .
the inductive scheme for constructing @xmath116 in the special case @xmath115 is much better suited for explicit computations than the general construction which will follow later on .
note that @xmath117 and @xmath118_\#=\mathcal{o}({\varepsilon})$ ] and proceed by induction .
assume that we found @xmath119 such that @xmath120 where , in particular , ( [ picon1 ] ) defines @xmath121 .
thus the next order term in the expansion @xmath122 must satisfy @xmath123 which uniquely determines the diagonal part of @xmath122 to be @xmath124 since @xmath125 follows from the fact that @xmath121 is the principal symbol of @xmath126 , @xmath127 indeed satisfies ( i ) up to an error of order @xmath128 . by induction assumption
we also have that @xmath129_\#=\mathcal{o}({\varepsilon}^{n+1})$ ] and thus @xmath130_\#={\varepsilon}^{n+1}f_{n+1 } + \mathcal{o}({\varepsilon}^{n+2})\,.\ ] ] hence , the diagonal part of @xmath122 being fixed already , the off - diagonal part of @xmath122 must satisfy @xmath131 = - f_{n+1}$ ] .
in particular , @xmath132 for all @xmath66 .
we first show that if ( [ picon4 ] ) has a solution @xmath133 , it is unique , i.e. that the kernel of the map @xmath134 $ ] restricted to ran@xmath135 contains only zero .
to see this let @xmath136 and note that , due to the gap condition , @xmath137 is invertible on ran@xmath135 with @xmath138 .
hence @xmath139 = 0 \leftrightarrow [ h_0(z)-\overline\sigma_{\rm r}(z ) , \pi_{n+1}^{od1}(z ) ] = 0 } \\ & \leftrightarrow & \pi_{n+1}^{od1}(z ) = ( h_0(z)-\overline\sigma_{\rm r}(z ) ) \pi_{n+1}^{od1}(z ) ( h_0(z)-\overline\sigma_{\rm r}(z))^{-1}\end{aligned}\ ] ] and therefore @xmath140 with @xmath141 . hence @xmath142 and
we conclude that @xmath122 is unique when it exists . in the special case that @xmath115 , ( [ picon4 ] ) can be solved , and one finds @xmath143 using that @xmath144 is the principal symbol of @xmath145_\#$ ] , that @xmath146 is the principal symbol of @xmath147 and that @xmath147 satisfies ( i ) up to @xmath128 , one finds that @xmath148 and thus that @xmath149 defined through ( [ picon2 ] ) and ( [ picon5 ] ) satisfies ( i ) and ( iii ) up to @xmath128 .
we conclude that by induction we have uniqueness of @xmath116 in the general case , and an explicit construction for @xmath116 when @xmath115 .
the latter one involves four steps at each order : [ a ] evaluation of @xmath121 as in ( [ picon1 ] ) , [ b ] computation of @xmath150 as in ( [ picon2 ] ) , [ c ] evaluation of @xmath144 as in ( [ picon3 ] ) , [ d ] computation of @xmath151 as in ( [ picon5 ] ) .
we now turn to the construction of @xmath116 in the general case .
since the moyal product is a local operation ( it depends only on the pointwise value of the symbols and their derivatives ) it suffices to construct @xmath116 locally in phase space and then uniqueness will liberate us from gluing the local results together .
let us fix a point @xmath152 . from the continuity of the map @xmath153 and the gap condition
it follows that there exists a neighborhood @xmath154 of @xmath155 such that for every @xmath156 the set @xmath69 can be enclosed in a positively - oriented complex circle @xmath157 ( independent of @xmath158 ) in such a way that @xmath157 is symmetric with respect to the real axis , @xmath159 and @xmath160 where @xmath161 is the radius of the complex circle @xmath162 .
the constant @xmath163 in ( [ cgcon ] ) is the same as in ( [ gap ] ) and the existence of a constant @xmath164 _ independent _ of @xmath165 such that ( [ crcon ] ) is satisfied follows from assumption ( g3 ) .
we keep @xmath166 in the notation as a bookkeeping device , in order to distinguish the contributions related to the gap , although @xmath167 .
let us choose any @xmath168 and restrict all the following expressions to @xmath169 .
there exist a formal symbol @xmath170 the _ local moyal resolvent _ of @xmath24 such that @xmath171 the symbol @xmath170 can be explicitly constructed .
we abbreviate @xmath172 where the inverse is understood in the @xmath173-sense and exists according to ( [ cgcon ] ) . by induction , suppose that @xmath174 satisfies the first equality in ( [ moyal resolvent ] ) up to @xmath175-terms , i.e.@xmath176 by choosing @xmath177 , we obtain that @xmath178 satisfies the same equality up to @xmath179-terms .
then the formal symbol @xmath180 satisfies the first equality in ( [ moyal resolvent ] ) which by the associativity of the moyal product implies the second one .
equation ( [ moyal resolvent ] ) implies that @xmath170 satisfies the _ resolvent equation _
@xmath181 from the resolvent equation
it follows by using an argument similar to the standard one in operator theory @xcite that the symbol @xmath182 defined by @xmath183 is a moyal projector such that @xmath184_{\#}=0 $ ] on @xmath154 . indeed , for every fixed @xmath185 and @xmath186 , the map @xmath187 is holomorphic in a neighborhood of the circle @xmath157
. then @xmath157 can be expanded to a slightly larger circle @xmath188 without changing the left hand side of ( [ riesz formula ] ) and we obtain @xmath189_{j } \nonumber \\ & = &
\frac{i}{2\pi } \int_{\gamma } r_{j}(\zeta ) \ d\zeta = \pi_{j } \nonumber\end{aligned}\ ] ] where ( [ resolvent eq ] ) has been used .
the first equality in ( [ pi square ] ) follows by noticing that for every @xmath190 @xmath191 and by expanding the moyal product order by order in @xmath192 .
since the circle @xmath193 is symmetric with respect to the real axis one immediately concludes that @xmath194 , since @xmath195 as a consequence of ( [ moyal resolvent ] ) . from ( [ riesz formula ] )
it follows that @xmath107 moyal - commutes with @xmath196 for any @xmath197 .
then , by multiplying @xmath198 by @xmath199 on both sides , one obtains that @xmath200 .
finally we have to show that @xmath201 for every @xmath186 . from the riesz formula ( [ riesz formula ] ) it follows that for every @xmath190 one has @xmath202according to ( [ crcon ] ) we are left to prove that @xmath203 where @xmath204 must _ not _ depend on @xmath165 . as for @xmath205
, we notice that according to ( [ cgcon ] ) one has @xmath206 and moreover , @xmath207 where the last bound follows from the fact that @xmath208 ( recall that @xmath87 ) . by induction
one controls higher order derivatives and ( [ sufficient cond ] ) follows for @xmath209 .
again by induction , assume that @xmath210 satisfy the bound ( [ sufficient cond ] ) .
then , by writing out @xmath211 and using ( [ composition ] ) , one concludes that @xmath212 satisfies ( [ sufficient cond ] ) with @xmath87 . @xmath213
* step ii .
quantization * first of all , by resummation ( prop .
[ prop resummation ] ) we obtain a semiclassical symbol @xmath214 @xmath58 whose asymptotic expansion is given by @xmath215 .
then , by weyl quantization , one gets a bounded operator @xmath216 ( see prop .
[ prop caldvaill ] ) which is an almost - projector , in the sense that 1 .
@xmath217 2 .
@xmath218 3 .
@xmath219=\mathcal{o}_{-\infty } ( \varepsilon ^{\infty } ) $ ] notice that the assumption @xmath83 is crucial in order to obtain ( iii ) for an unbounded @xmath91 .
in order to get a true projector we follow the idea of @xcite and notice that @xmath220 and the spectral mapping theorem for self - adjoint operators imply that for each @xmath221 there is a @xmath222 such that @xmath223\cup[1-c_n{\varepsilon}^n,1+c_n{\varepsilon}^n ] = : \sigma^{\varepsilon}_{0 } \cup \sigma^{\varepsilon}_{1 } \,.\ ] ] hence one can define for @xmath224 @xmath225 then @xmath226 follows and we claim that @xmath95 .
indeed , @xmath227 where @xmath228 is the projection valued measure of @xmath229 .
finally notice that @xmath230 & = & \frac{i}{2\pi } \int_{|\zeta-1|=\frac{1}{2 } } [ \widehat h , ( \widehat\pi-\zeta)^{-1 } ] d\zeta\\ & = & -\,\frac{i}{2\pi } \int_{|\zeta-1|=\frac{1}{2 } } ( \widehat\pi-\zeta)^{-1 } [ \,\widehat h,\,\widehat \pi\ , ] ( \widehat\pi-\zeta)^{-1 } d\zeta\,,\end{aligned}\ ] ] which implies that @xmath231\big\|_{\mathcal{b(h ) } } \leq c \big\| [ \widehat h , \widehat\pi]\big\|_{\mathcal{b(h ) } } = \mathcal{o}_0({\varepsilon}^\infty)\,.\ ] ] this concludes the proof of the theorem .
@xmath213 * essential self - adjointness of * @xmath91**. * * since @xmath24 is an _
symbol its weyl quantization @xmath91 is symmetric on the invariant domain @xmath232 . if @xmath24 belongs to @xmath233 then @xmath91 is a bounded operator , and there is nothing to prove . in order to prove essential self - adjointness in the case @xmath234 , we use an argument of @xcite .
the proof does not exploit the smallness of @xmath192 and we therefore consider any @xmath235 . for @xmath236 let @xmath237 which , according to proposition [ prop point inversion ] , belongs to @xmath238
. moreover @xmath239 where @xmath240 , since @xmath241 and @xmath242 .
after weyl quantization we obtain that @xmath243 the latter bound following ( for @xmath244 large enough ) from proposition [ prop caldvaill ] and from estimating the frchet semi - norms of @xmath245 .
essential self - adjointness of @xmath29 on the domain @xmath246 follows , if we can show that ker@xmath247 for some @xmath236 .
for this let @xmath248 ker@xmath249 and @xmath250 .
using @xmath251 , we obtain @xmath252 since @xmath253 for @xmath244 large enough , @xmath254 is dense in @xmath255 and hence @xmath256 follows .
the fact that the subspace associated with an isolated energy band decouples from its orthogonal complement up to small errors in @xmath3 leads immediately to the following question .
is there a natural way to describe the dynamics of the system inside the almost invariant subspace ran@xmath31 ?
the main obstruction for such a simple description is the fact that the subspace ran@xmath31 depends on @xmath3 and is not easily accessible .
even worse , in general the limit @xmath257 does not exist , meaning that ran@xmath31 is not even close to an @xmath3-independent subspace . in order to obtain a useful description of the effective intraband dynamics we thus need to map ran@xmath31 to an easily accessible and @xmath3-independent _ reference _ subspace .
> from the continuity of @xmath153 and the gap condition it follows that there is a subspace @xmath258 independent of @xmath259 such that the subspaces ran@xmath260 are all isomorphic to @xmath261 .
let @xmath262 be the projection on @xmath263 , then @xmath264 @xmath265 will serve as the projector on the _ reference subspace _ @xmath266 ran@xmath267 .
of course @xmath261 is _ highly _ non - unique and a convenient choice must be made in concrete applications .
once the reference hilbert space is fixed we next chose a unitary operator valued smooth function @xmath268 which pointwise in phase space intertwines @xmath269 and @xmath262 , i.e.@xmath270 the existence of such a _ smooth _ map follows from a bundle - theoretic argument given at the end of this section . again
@xmath271 is not unique and must be chosen conveniently .
we will see in section [ sdirac ] that there is an optimal choice for @xmath271 , which reflects the physics of the problem .
we can not prove that it is possible to choose @xmath272 in @xmath273 .
indeed , relation ( [ intertwining ] ) does not imply any bound at infinity on the derivatives of @xmath272 , as can be seen by multiplying @xmath272 with a highly oscillating phase .
hence we _ assume _ that @xmath272 is in @xmath273 , as will be the case in the physical examples
. in the following @xmath274 will denote the group of unitary operators over @xmath26 .
[ th unitaries ] assume either ( ig)@xmath81 or ( cg ) and that there exists a @xmath275-valued map @xmath276 which satisfies ( [ intertwining ] ) .
then there exist a unitary operator @xmath277 such that @xmath278 and @xmath279 , where @xmath280 in @xmath281 with principal symbol @xmath272 . _ [ nagyrem ] in @xcite the nagy transformation ( [ nagy ] ) is used in order to map ran@xmath31 to the @xmath3-independent subspace ran@xmath282 .
this is possible because in their application the symbol @xmath146 depends only on @xmath7 and , as a consequence , @xmath282 is a projector satisfying @xmath283 . however , in general @xmath146 depends on @xmath7 and @xmath8 , see section 6 and @xcite for relevant examples , and the mapping to the reference space becomes more subtle .
_ * proof . *
* step i. construction of the moyal unitaries
. * again @xmath272 fails to be a moyal unitary ( i.e. @xmath284 ) and to intertwine @xmath116 and @xmath285 .
however , the following lemma shows that @xmath272 can be corrected order by order to reach this goal .
the idea of constructing a pseudodifferential operator which is almost unitary and diagonalizes a given pseudodifferential operator has a long tradition , cf .
@xcite section 7 and references therein , and was applied in different settings many times , e.g. @xcite .
assume either ( ig)@xmath81 or ( cg ) and that there exists a @xmath275-valued map @xmath276 which satisfies ( [ intertwining ] ) .
then there is a formal symbol @xmath286 , with @xmath287 , such that 1 . @xmath288 and @xmath289 , 2 .
@xmath290 , where @xmath116 is the moyal projector constructed in lemma [ prop moyal proj ] .
[ prop moyal unitaries ] [ nonunirem ] _ we emphasize that as opposed to the moyal projector @xmath116 appearing in lemma [ prop moyal proj ] the moyal unitary @xmath291 is highly non - unique even for fixed @xmath292 .
as it will follow from the proof , all the possible choices of moyal unitaries intertwining @xmath116 and @xmath285 with prescribed principal symbol @xmath292 are parametrized by the antihermitian moyal symbols which are diagonal in the @xmath285-splitting . _ * proof of lemma [ prop moyal unitaries ] . *
observe that @xmath292 satisfies ( i ) and ( ii ) on the principal symbol level .
we proceed by induction and assume that we found @xmath293 satisfying ( i ) and ( ii ) up to @xmath294 .
we will construct @xmath295 such that @xmath296 satisfies ( i ) and ( ii ) up to @xmath128 .
to this end we write without restriction @xmath297 with @xmath298 hermitian and @xmath299 anti - hermitian . by induction assumption
we have @xmath300 u^{(n)*}\ , \#\ , u^{(n)}-1 & = & \varepsilon^{n+1}\tilde{a}_{n+1}+ \mathcal{o}(\varepsilon^{n+2})\ , .
\]]thus @xmath295 has to solve @xmath301 u_{n+1}^*\ , u_{0 } + u_{0}^*\ , u_{n+1 } & = & -\tilde{a}_{n+1}\ , .
\end{array } \label{un2}\ ] ] the first equation in ( [ un2 ] ) fixes @xmath302 , since @xmath303 is hermitian as it is the principal symbol of @xmath304 .
the second equation in ( [ un2 ] ) is then also satisfied , since the compatibility equation @xmath305 follows from @xmath306 by noticing that @xmath307 ( resp .
@xmath308 ) is the principal symbol of the l.h.s ( resp .
r.h.s ) .
note that ( [ un2 ] ) puts no constraint on @xmath299 and we are left to determine it using ( ii ) .
let @xmath309 , then by induction assumption @xmath310and thus @xmath311\right ) + \mathcal{o } ( \varepsilon^{n+2})\,.\ ] ] hence we need to find an anti - hermitian @xmath299 satisfying @xmath312=0\ , , \]]which is given by @xmath313\ , , \label{un6}\ ] ] provided that @xmath314 is hermitian and off - diagonal in the @xmath285-splitting , i.e. @xmath315 and @xmath316 vanish .
this follows by noticing that @xmath314 is the principal symbol of @xmath317 and then @xmath318 where for the last equality we inserted @xmath319 and used that @xmath320 solves ( i ) up to @xmath128 and that @xmath116 is a moyal projector .
a similar argument shows that @xmath315 vanishes too .
note also that ( [ un6 ] ) fixes only the off - diagonal part of @xmath299 and one is free to choose the diagonal part of @xmath299 arbitrarily , which is exactly the non - uniqueness mentioned in remark [ nonunirem ] .
it remains to show that the assumption @xmath321 implies that @xmath322 belongs to @xmath323 .
assume by induction that @xmath324 .
then the formula @xmath325 shows that @xmath298 belongs to @xmath326 as it is the @xmath327-th term of an element of @xmath328 . by proposition [ prop point product ] , @xmath329 as well .
analogously we have that @xmath330 by induction assumption , therefore @xmath331 and thus @xmath332 , which finally gives @xmath333 . @xmath213 * step ii .
quantization * now let @xmath291 denote a resummation of the formal power series @xmath286 in @xmath281 ( see prop .
[ prop resummation ] ) . then , by weyl quantization , one gets a bounded operator @xmath334 ( see prop . [ prop caldvaill ] ) such that : 1 .
@xmath335and@xmath336 2 .
@xmath337 . as a first step we modify @xmath338 by an @xmath339-term in order to get a true unitary operator @xmath340 ( which , in general , does not correspond to the weyl quantization of any semiclassical symbol ) .
let @xmath341 notice that @xmath342 is a self - adjoint positive operator which is @xmath343-close to the identity operator .
then @xmath344 is well - defined and again @xmath343-close to the identity operator .
hence ( [ unitarization ] ) defines a unitary operator which moreover is @xmath343-close to @xmath338 .
finally we modify @xmath345 in order to obtain a unitary which exactly intertwines @xmath267 and @xmath31 . since @xmath346 for @xmath192 sufficiently small , the nagy formula as used in @xcite @xmath347 ^{-\frac{1}{2}}\left [ \widetilde u\,\pi\,\widetilde u^{\,*}\,{\pi_{\rm r}}+(1-\widetilde u\,\pi\,\widetilde u^{\,*})(1- { \pi_{\rm r}})\right]\ ] ] defines a unitary operator @xmath348 such that @xmath349 and @xmath350 .
thus by defining @xmath351 one obtains ( [ intertwines ] ) , with the desired properties .
@xmath213 _ we sketch how to prove the existence of a smooth map @xmath272 satisfying ( [ intertwining ] ) .
given @xmath352 the map @xmath353 @xmath354 defines a fibration of hilbert spaces over the base space @xmath59 . _ _ the fibration is locally trivial .
indeed for any @xmath355 there exists a neighborhood @xmath154 such that @xmath356 for any @xmath185 , so that the nagy formula @xmath357^{-\frac { 1}{2}}\left [ \pi_{0}(z)\pi_{0}(z_{0})+(1-\pi_{0}(z))(1-\pi _ { 0}(z_{0}))\right]\ ] ] locally defines a unitary operator @xmath358 such that @xmath359 .
a local trivialization of the fibration is then explicitly given by @xmath360 where we use the fact that there exists a unitary operator @xmath361 .
the existence of @xmath362 follows from the fact that the dimension of @xmath363 is independent of @xmath155 , but the map @xmath364 may be _ a priori _ even discontinuous .
_ _ moreover one can check that any two such trivializations are @xmath365-compatible , and the previous data define a linear @xmath365-bundle . _ _ since the base space is contractible , the bundle is trivial and the associated principal @xmath365-bundle ( i.e. the bundle of the orthonormal frames ) admits a global _ smooth _ section .
this implies the existence of a _ smooth _ map @xmath366 @xmath367 such that ( [ intertwining ] ) holds true.[rem bundle ] _
in the previous section we constructed a unitary @xmath43 on @xmath26 which exactly intertwines the almost invariant subspace ran@xmath368 and the reference subspace @xmath369 ran@xmath267 . @xmath43 and @xmath31 are @xmath370-close to pseudodifferential operators with symbols @xmath291 and @xmath116 both in @xmath371 .
we define the _ effective hamiltonian _ @xmath372 as the quantization of a resummation @xmath42 of the formal symbol @xmath373 recall that we do not distinguish semiclassical symbols and formal symbols in the notation .
the following theorem is the basis for the adiabatic perturbation theory , as it relates the unitary time - evolution generated by the original hamiltonian @xmath29 to the one generated by the effective hamiltonian @xmath372 .
[ hthm ] under the assumptions of theorem [ th unitaries ] , one has that @xmath374 and @xmath372 is essentially self - adjoint on @xmath246
. furthermore @xmath375=0\,,\ ] ] @xmath376 and @xmath377 * proof . *
since @xmath378 and @xmath379 , the composition rule for semiclassical operators ( see prop . [
prop moyal product ] ) yields @xmath380 and thus @xmath381 .
let @xmath382 .
since @xmath383 is bounded with bounded inverse , one finds , by checking definitions , that @xmath384 is self - adjoint on @xmath385 and that @xmath384 is essentially self - adjoint on @xmath386 . according to equation ( 8.10 ) in @xcite , which generalizes to @xmath387-valued symbols , @xmath388 and thus @xmath389 .
hence @xmath62 is a core for @xmath384 and , since @xmath390 , the same conclusions hold for @xmath372 .
next observe that , by construction , @xmath391=0 $ ] for all @xmath186 and thus @xmath392_{\widetilde{\#}}=0 $ ] because @xmath285 does not depend on @xmath393 . hence @xmath394=0 $ ] and
thus ( [ hcommu ] ) follows . for ( [ groupd1 ] ) observe that @xmath395 since , by construction , @xmath396 .
finally ( [ groupd2 ] ) follows from ( [ groupd1 ] ) using @xmath397 . @xmath213
_ it might seem more natural to define the effective hamiltonian as @xmath398 clearly one should have @xmath399 in some sense .
however , if @xmath29 is unbounded , this closeness does not follow in the norm of bounded operators from our results , since @xmath43 need not be a semiclassical operator . as a consequence no asymptotic expansion of @xmath400 in the norm of bounded operators would be available . _
in the remainder of this section we will study the finite order asymptotic approximations @xmath401 to the effective hamiltonian @xmath402 . by virtue of ( [ hcommu ] ) , we can , whenever appropriate , restrict our attention to the reduced hilbert space @xmath403 .
furthermore we define @xmath404 and obtain a finite order expansion of the unitary @xmath43 as @xmath405 .
our main interest are approximations to the solution of the time - dependent schrdinger equation @xmath406over times of order @xmath407 , where @xmath408 does not depend on @xmath192 and @xmath409 is arbitrary . starting with ( [ groupd1 ] ) on the almost invariant subspace we obtain @xmath410
is assumed in order to have @xmath411 .
hence , given the level of precision @xmath412 and the time scale @xmath413 , the expansion of @xmath402 must be computed up to order @xmath414 and the expansion of @xmath43 up to order @xmath415 . put differently , in order to improve the error , a better approximation to the unitary transformation is necessary . on the other hand , in order to enlarge the time - scale of validity for the space - adiabatic approximation , only the effective hamiltonian @xmath402 must be computed to higher orders .
specializing ( [ ua ] ) to @xmath416 and @xmath417 , one obtains the leading order solution of the schrdinger equation as @xmath418 where @xmath419 . here
the choice of @xmath417 corresponds to the macroscopic or semiclassical time - scale @xmath420 . on this time - scale the effective dynamics @xmath421 on the reference subspace
is expected to have a nice semiclassical limit , under suitable conditions on @xmath402 .
note that one can replace in ( [ ua ] ) and analogously in ( [ u1 ] ) @xmath408 by @xmath422 and obtains @xmath423 thus one can enlarge the time - span for which the approximation holds without the need to compute further terms in the expansion .
the price to be paid is a larger error , of course .
we emphasize that ( [ ua ] ) and ( [ u1 ] ) are purely space - adiabatic expansions with _ no semiclassical approximation _ invoked yet . as a consequence one obtains uniform results and a simple bound on the growth of the error with time .
note in particular that the space - adiabatic approximation holds on time - scales far beyond the ehrenfest time - scale , the maximal time - scale for which semiclassical approximations are expected to hold .
for some particular cases semiclassical expansions of the full propagator @xmath424 have been derived directly , e.g. in the context of the dirac equation @xcite .
these expansions hold , in general , only for short times , in the sense that they must be modified each time a caustic in the corresponding classical flow is encountered . more important , the clear separation of the space - adiabatic and the semiclassical expansion is not maintained , which is a severe drawback , since in many physical situations the space - adiabatic approximation is valid to high accuracy , while the semiclassical approximation is not , cf . section [ sdirac ] . on the other hand ,
a semiclassical expansion of the right hand side of ( [ u1 ] ) is straightforward in many interesting cases , as will be discussed in section [ ssemi ] .
in parentheses we remark that the space - adiabatic approximation can be used also in the time - independent setting , i.e. to estimate spectral properties of @xmath91 .
if one is able to compute eigenvalues of @xmath425 up to errors of order @xmath426 , @xmath427 it follows that @xmath428 if , in addition , one knows from some a priori arguments that @xmath91 has pure point spectrum near @xmath429 , it follows that @xmath91 has an eigenvalue @xmath426-close to @xmath429 .
otherwise one can at least conclude that there is a `` resonance '' in the sense of a quasi bound state @xmath426-close to @xmath429 .
we stress that no explicit knowledge of @xmath43 is needed as long as the interest is in approximate eigenvalues only . for example , the scheme just described can be applied to the time - independent born - oppenheimer theory , where one is interested in the low lying spectrum of a molecule .
the standard approaches to the time - independent born - oppenheimer approximation @xcite yield in some respects mathematically stronger results .
however , our scheme suffices for estimating asymptotic expansions of eigenvalues and is simpler to handle , in general .
we turn to the explicit determination of the leading order terms @xmath430 in the expansion of @xmath402 using ( [ heffdef ] ) . of course
, in concrete applications only @xmath24 and @xmath272 are given explicitly , while the higher order terms in the expansion of @xmath291 must be calculated using the construction from section [ suni ] .
for a general hamiltonian @xmath91 such a program is feasible only for the terms @xmath431 , @xmath432 and possibly @xmath433 , which will be our concern in the following .
the principal symbol of @xmath42 is given by @xmath434higher order terms can be obtained using ( [ heffdef ] ) .
the double moyal product becomes rather awkward to handle , and alternatively we proceed inductively by observing that @xmath435 with the subprincipal symbol on the left hand side being @xmath436 recall the notation @xmath437 for the expansion of the moyal product , see the appendix . combining ( [ h11 ] ) and ( [ h12 ] ) one obtains @xmath438 the expression ( [ h1 ] ) further simplifies if one specializes to the case where @xmath439 consists of a single eigenvalue of @xmath440 and one projects on the relevant subspace , @xmath441 the right hand side has the nice property to be independent of @xmath442 and thus to depend only on known quantities .
along the same lines and under the same condition on @xmath443 , one computes @xmath444 again , ( [ h21 ] ) does not depend on @xmath445 for the special case under consideration , but it does depend on @xmath442 , which must now be computed using the construction from section [ suni ] .
although ( [ h21 ] ) looks still rather innocent , in general , it requires some work to compute it explicitly .
this is partly because the second order expansion of the moyal product in ( [ h21 ] ) tends to become rather tedious to obtain .
but , in general , also the determination of @xmath442 is nontrivial . to convince the reader , we state without details that the construction from sections [ spro ] and [ suni ] yields @xmath446 + \frac{i}{4}\big [ \big ( \{u_0,\pi_0\}u_0^*+u_0\{\pi_0,u_0^*\}\big),\pi_{\rm r}\big]\big)\,,\ ] ] with @xmath447 where we used that @xmath448 .
recall the definition ( [ poisson ] ) of the poisson bracket @xmath449 . to compute @xmath450 from the given quantities
one has to use the construction explained in section [ spro ] .
one finds @xmath451 where @xmath452 is uniformly bounded because of the gap condition . for sake of completeness
we mention that @xmath453 in this case . for the higher orders in the expansion of @xmath42 we only remark
that , in general , @xmath454 depends on @xmath455 , @xmath456 and @xmath457 . in the special , but interesting case of an isolated eigenvalue @xmath458
, @xmath454 depends only on @xmath459 , @xmath456 and @xmath457 and is thus considerably easier to obtain . _
note that in the case of @xmath439 , not only the principal symbol @xmath460 , but also the subprincipal symbol @xmath461 as given by ( [ h1d ] ) is well defined regardless of the gap condition , provided that the spectral projection @xmath260 is sufficiently regular .
indeed , it can be shown , at least in some special cases , that there is still adiabatic decoupling to leading order and an effective dynamics generated by @xmath462 without a gap condition @xcite , @xcite . _ to get even more explicit formulas for @xmath432 and @xmath433 , note that in most applications one has no naturally given transformation @xmath272 .
instead one chooses a suitable basis @xmath463 of ran@xmath464 and defines @xmath465 , where the vectors @xmath466 form a basis for ran@xmath285 and @xmath467 is some arbitrary unitary intertwining ran@xmath468 and ran@xmath469 .
@xmath470 is independent of the choice of the unitary @xmath467 for all @xmath471 .
we remark that such a basis @xmath463 of global smooth sections of the bundle over @xmath59 defined by @xmath472 always exists , since @xmath59 is contractible ( see remark [ rem bundle ] ) . however , we are not aware of a proof which insures @xmath473 .
the situation changes completely , once one considers local domains in the base space which are not contractible .
then it might become necessary to chose as reference space the space of sections of a globally nontrivial bundle . assuming that @xmath439 consists of a single eigenvalue of @xmath440 of multiplicity @xmath474 ( including @xmath475 )
, we obtain the @xmath476-matrix @xmath477 as @xmath478 with @xmath479 the indices @xmath480 and @xmath481 are matrix - indices , both running from @xmath482 to @xmath474 .
equations ( [ hs ] ) and ( [ h1w ] ) are one of our central results .
they are still of a simple form and mostly suffice to compute the basic physics . the first term in ( [ hs ] ) is referred to as peierls substitution and the first order correction carries information on the intraband spinor evolution .
e.g. , as will be discussed in section [ sdirac ] , for the dirac equation @xmath47 governs the spin precession .
the reason for the particular splitting of the terms in ( [ h1w ] ) will be discussed in section [ ssemi ] .
here we only remark that the second term in ( [ h1w ] ) is related to a `` generalized '' berry connection .
we omit the analogous formula for @xmath483 , since it is too complicated to be helpful .
an instructive example to which formula ( [ h1w ] ) applies are born - oppenheimer type hamiltonians of the form @xmath484 @xmath485 , with an electronic energy band @xmath486 of constant multiplicity @xmath474 , i.e. @xmath487 .
adiabatic decoupling for born - oppenheimer type hamiltonians is established with exponentially small errors by martinez and sordoni @xcite , see also @xcite .
their result partly triggered our interest to develop a general theory .
exponentially accurate coherent state solutions for born - oppenheimer type hamiltonians have been constructed by hagedorn and joye in @xcite .
note that the quadratic growth of @xmath488 as a function of @xmath8 prevents from applying the general results directly .
as to be discussed in section [ scut ] , energy cutoffs need to be introduced . for the moment
we ignore this problem and proceed by working out the perturbative scheme formally .
we fix arbitrarily an orthonormal basis @xmath489 of ran@xmath490 depending smoothly on @xmath7 which then satisfies @xmath491 with @xmath492 for @xmath493 . only the second term of our formula ( [ h1w ] ) contributes and yields @xmath494 which is well known in the case of a nondegenerate eigenvalue , @xcite . @xmath495 has the geometrical meaning of a gauge potential , i.e. coefficients of a connection on the trivial bundle @xmath496 , the so called berry connection . as mentioned already , a more detailed discussion of the origin of the berry connection will be given in section [ ssemi ] .
for the born - oppenheimer hamiltonian the calculation of @xmath483 is still feasible without much effort and the result is @xmath497 recall the definition of @xmath498 , which reduces to @xmath499 in the present case .
although we omit the details of the computation leading to ( [ h2bo ] ) , we shortly describe how ( [ h21 ] ) relates to ( [ h2bo ] ) . since @xmath500 and @xmath501 the corresponding terms in ( [ h21 ] ) do not contribute .
since @xmath272 and @xmath104 are functions of @xmath7 only , the second term in ( [ u1 ] ) is the only one contributing to @xmath442 , and thus the third term in ( [ h21 ] ) also vanishes after projecting with the @xmath285 s from outside the brackets .
the last two terms in ( [ h21 ] ) cancel each other .
the seventh term in ( [ h21 ] ) yields the first term in ( [ h2bo ] ) and the fourth and sixth term in ( [ h21 ] ) combine to the second and third term in ( [ h2bo ] ) .
in particular the calculation yields for the symbol of the unitary @xmath502 thus the symbol of the second order effective born - oppenheimer hamiltonian reads @xmath503 where the first term from ( [ h2bo ] ) nicely completes the square to the first term in ( [ h3bo ] ) .
note that the third term on the right side of ( [ h3bo ] ) depends on @xmath7 only and was interpreted in @xcite as a geometric electric potential in analogy to the geometric vector potential @xmath504 . in the special case of a nondegenerate eigenvalue @xmath505 and a matrix - valued hamiltonian @xmath24 , ( [ h3bo ] )
reduces to the expression obtained by littlejohn and weigert @xcite .
they also remark that the previous studies @xcite of the expansion of the effective born - oppenheimer hamiltonian missed the last term in ( [ h3bo ] ) .
this strengthens our point of the usefulness of a general and systematic space - adiabatic perturbation theory .
the full power of our scheme is in force in cases where ran@xmath146 is degenerate and depends both on @xmath7 and @xmath8 , since then the known techniques @xcite can not be applied .
the simplest example of this kind is the one - particle dirac equation with slowly varying electric and magnetic potentials , which will be discussed in section [ sdirac ] . with little additional effort
our scheme can be applied even to the time - adiabatic setup . as for notation
, we replace the phase space @xmath506 by @xmath507 in the following . given a hilbert space @xmath26 and family @xmath508 , @xmath509 of self - adjoint operators such that @xmath510 , the solutions of the equations @xmath511 define a _
unitary propagator_. a unitary propagator is a unitary operator - valued map @xmath512 strongly continuous in @xmath513 and @xmath244 jointly , such that @xmath514 for any @xmath515 .
in particular we have that @xmath516 solves the time - dependent schrdinger equation @xmath517 for any @xmath518 .
it is assumed in addition that @xmath519 , the principal symbol of @xmath508 , has a relevant part @xmath520 of its spectrum , which is separated by a gap from the remainder uniformly for @xmath521 . as before we denote the spectral projection on @xmath520 by @xmath522 .
the following theorem is a variant of the _ time - adiabatic theorem of quantum mechanics _
@xcite , however formulated in the language of adiabatic perturbation theory .
sjstrand first recognized the usefulness of pseudodifferential calculus in this context @xcite and we are grateful to g. nenciu for pointing this out to us .
we remark that the proof below can be adapted to the case of a time - dependent operator - valued classical symbol @xmath523 , as for example the dirac hamiltonian or the pauli - fierz hamiltonian with _
slowly _ varying time - dependent external potentials .
let @xmath524 and @xmath520 be as above . 1 .
decoupled subspace .
there exists a family of orthogonal projectors @xmath525 such that @xmath526 , @xmath527 and @xmath528 uniformly for @xmath529 .
whenever @xmath530 for some @xmath509 and all @xmath531 , then @xmath532 .
2 . intertwining unitaries .
there exists a family of unitaries @xmath533 with @xmath534 and a family of unitaries @xmath535 such that @xmath536 3 .
effective dynamics .
there exists a family of self - adjoint operators @xmath537 , @xmath538 , such that @xmath539 = 0\qquad \mbox{for all}\qquad t\in\mathbb{r}\ ] ] and the solution of the initial value problem @xmath540satisfies @xmath541 the asymptotic expansion of @xmath537 in @xmath542 reads @xmath543 where @xmath544 is the asymptotic expansion of @xmath524 in @xmath542 and @xmath545 is the asymptotic expansion of @xmath546 in @xmath542 .
before we turn to the proof we remark that , for @xmath547 and @xmath548 an orthonormal basis of ran@xmath522 , the effective hamiltonian including second order reads @xmath549 where @xmath550 . for the unitary @xmath546 one
finds @xmath551 * proof.*in order to apply the general scheme developed in the previous sections it is convenient in analogy with the extended configuration space in classical mechanics to introduce the extended space @xmath552 and to define the extended hamiltonian @xmath553 which is self - adjoint on the domain @xmath554 . by following howland @xcite
, we notice that the unitary group @xmath555 , @xmath556 , is related to the unitary propagator ( [ unit propag ] ) through @xmath557 moreover , the unitary group @xmath558 can now be studied by means of the techniques developed in the previous sections , since @xmath559 is nothing but the weyl quantization of the operator - valued function @xmath560 , and @xmath561 belongs to @xmath562 . by assumption
@xmath563 satisfies assumption ( gap)@xmath564 with @xmath89 .
however , because of the simple dependence of @xmath565 on @xmath566 , the conclusion of theorem [ th invariant subspace ] and [ th unitaries ] hold still true in a sense to be made precise . indeed , by following
the proof of lemma [ prop moyal proj ] one obtains a semiclassical symbol @xmath567 , depending on @xmath513 only , such that @xmath568_{\tilde{\#}}\asymp 0 $ ] in @xmath569 . on the other hand , @xmath570_{\tilde{\#}}=[h,\pi ] _
{ \tilde{\#}}+[\eta , \pi ] _ { \tilde{\ # } } = [ h,\pi ] _ { \tilde{\#}}-i\varepsilon \left ( \partial_{t}\pi \right ) \]]where the last equality follows from the fact that @xmath571_{\tilde{\#}}$ ] is the symbol of @xmath572 $ ] @xmath573 .
since both @xmath184_{\tilde{\#}}$ ] and @xmath574 belong to @xmath575 , one concludes that the asymptotic expansion @xmath576_{\tilde{\#}}\asymp 0\ $ ] holds true in @xmath577 , and hence @xmath578=\mathcal{o}_{0 } ( \varepsilon ^{\infty } ) $ ] .
finally one defines @xmath579 and finds @xmath580 , @xmath581 and @xmath582=\mathcal{o}_0({\varepsilon}^\infty|\sigma|)$ ] as in section [ spro ] .
together with ( [ howlandu ] ) this implies @xmath583 however , since @xmath525 and @xmath512 are continuous functions of @xmath513 , the pointwise statement ( [ tasubspace ] ) follows . for @xmath584 one can use for example
kato s construction @xcite and define @xmath584 as the solution of the initial value problem @xmath585 \,u_0^*(t)\ , , \quad u_0^*(0 ) = { \bf 1}\,.\ ] ] clearly @xmath584 belongs to @xmath586 .
notice that the same construction does not work in the multidimensional case , since the evolutions in different directions do not commute .
@xmath587 can be obtained as in section [ suni ] , where the fact that @xmath588 and @xmath584 both depend on @xmath513 only and not on @xmath589 simplifies the construction considerably and yields , in particular , a fibered unitary @xmath546 . as in the general
setting let the effective hamiltonian be defined as a resummation of @xmath590 with the explicit expansion ( [ kexpansion ] ) . according to theorem [ hthm ] we then have @xmath591 the pointwise statement ( [ kcompare ] ) follows again from the continuous dependence on @xmath513 of all involved expressions .
@xmath213 the born - oppenheimer type hamiltonians as well as many other physically relevant hamiltonians do not satisfy the general assumptions we imposed in sections [ spro ] and [ suni ] .
this is so for two reasons .
first of all they are quantizations of symbols taking values in the unbounded operators . secondly , the gap does not increase as fast as the hamiltonian for large momenta , e.g. quadratically in the born - oppenheimer setting .
the first problem is purely technical and the domain questions which arise have to be dealt with case by case .
the second problem causes a qualitative change in the sense that the adiabatic decoupling is no longer uniform , as can be seen from the construction of the almost invariant subspace in section [ spro ] . to deal with the second problem one
therefore needs a cutoff for large momenta .
there are basically two ways to implement such a cutoff .
one possibility is to directly cut off large momenta as was done in @xcite , but then one needs to control the times for which no momenta exceeding the cutoff are produced under the dynamics .
however , for a large class of hamiltonians including the born - oppenheimer type hamiltonian ( [ bohamiltonian ] ) , cutting off high energies is equivalent to cutting high momenta .
then conservation of energy immediately ensures that no momenta exceeding the cutoff are produced over time .
this idea was developped in @xcite and also used in @xcite .
we will briefly indicate an alternative way on how to implement such an energy cutoff in order to fit the born - oppenheimer and similar settings into our general assumptions .
let @xmath592 be elliptic and positive , i.e. there is a constant @xmath593 such that @xmath594 .
for example the born - oppenheimer hamiltonian as defined in ( [ bohamiltonian ] ) satisfies @xmath595 and it is elliptic provided that @xmath596 is positive ( otherwise just add a constant to @xmath64 since @xmath597 ) .
then we can prove adiabatic decoupling uniformly for energies below any @xmath598 , i.e. on ran@xmath599}(\widehat{h}_{0})$ ] .
let @xmath600 , then bounding the total energy by @xmath601 essentially corresponds to confining the slow degrees of freedom to the region @xmath602 in phase space .
more precisely , let @xmath603 such that @xmath604 } = 1 $ ] and @xmath605 for some @xmath606 , then @xmath607 is a semiclassical operator .
furthermore , its symbol @xmath608 has an asymptotic expansion which is identically equal to @xmath482 on @xmath602 , i.e. @xmath609 and identically equal to @xmath610 on the set where @xmath611 .
the statements about @xmath612 and its symbol follow from the functional calculus for semiclassical operators as developped e.g.in @xcite , theorem 8.7 .
next we assume that one can define an auxiliary hamiltonian @xmath613 such that 1 .
@xmath614 for all @xmath615 , 2 .
@xmath616 for all @xmath617 and @xmath618 3 .
and @xmath619 satisfies the global gap condition ( gap)@xmath620 .
this can be easily achieved e.g. in the born - oppenheimer setting by replacing @xmath621 by an appropriate bounded function .
it follows from the previous discussion that @xmath622 and that @xmath623 . using @xmath624 } ( \widehat h_0 ) = \1i_{(-\infty,\lambda ] } ( \widehat h_0)\,,\ ] ] one finds , in particular , that @xmath625 } ( \widehat h_0 ) = \mathcal{o}_0(\varepsilon^\infty)\ ] ] in the norm of bounded operators and thus also @xmath626 } ( \widehat h_0 ) } \\ & & \hspace{-10pt}=\,-\ , i e^{-i\widehat h_{\rm aux}t } \int_0^t \,ds\ , e^{i\widehat h_{\rm aux}s } \left ( \widehat h_{\rm aux } - \widehat h_{0 } \right ) e^{-i\widehat h_{0}s } \ , \1i_{(-\infty,\lambda ] } ( \widehat h_0 ) = \mathcal{o}_0(\varepsilon^\infty|t|)\,.\nonumber\end{aligned}\ ] ] now the scheme of sections [ spro ] , [ suni ] and [ slop ] can be applied to @xmath627 and by virtue of ( [ h0haux ] ) and ( [ h0haux2 ] ) all results are valid for @xmath64 up to @xmath628 if one restricts to energies below @xmath601 .
in particular one finds that for @xmath629 the leading order symbols of @xmath630 are given by the formulas obtained in section [ sfoe ] using the symbol @xmath631 .
the results of the previous sections are genuine quantum mechanical : semiclassical symbols have been used _ only as a tool _ in order to construct ( and , eventually , to approximate ) @xmath368 and @xmath43 , but no semiclassical limit has been performed .
indeed , the adiabatic decoupling of energy bands is a purely quantum phenomenon , which is , in general , independent from the semiclassical limit .
however , under the assumption that @xmath439 consists of a single eigenvalue of @xmath440 of necessarily constant multiplicity @xmath474 , the principal symbol of @xmath402 is a scalar multiple of the identity , i.e. @xmath632 , and a semiclassical analysis of @xmath372 can be done in a standard way . in particular , the dynamics of quantum observables can be approximated by quantities constructed using only the classical flow @xmath633 generated by the ( classical , scalar ) hamiltonian @xmath458 .
this results in a generalized egorov s theorem , see theorem [ egorovt ] .
we emphasize that for more general energy bands @xmath443 one can not expect a simple semiclassical limit , at least not in the usual sense .
* egorov s theorem .
* for the moment , we identify @xmath263 with @xmath634 and @xmath42 with @xmath635 , an @xmath476-matrix - valued formal symbol .
at least formally , egorov s theorem is obtained through an expansion of the heisenberg equations of motion for semiclassical observables : let @xmath636 , then the quantum mechanical time evolution of @xmath637 is given by @xmath638 and satisfies @xmath639\ , .
\label{heis1}\ ] ] expanding both sides of ( [ heis1 ] ) on the level of symbols and using @xmath640\equiv 0 $ ] , @xmath641 , one obtains the following hierarchy of equations : @xmath642 \label{eg1 } \\
\frac{d\,a_{1}(t)}{dt } & = & \{{e_{\rm r}}\mathbf{1},a_{1}(t)\}+i[h_{1},a_{1}(t)]- \frac{1}{2}\big ( \{h_{1},a_{0}(t)\}-\{a_{0}(t),h_{1}\}\big ) \nonumber \\ & & + \,i[h_{2},a_{0}(t)]\label{eg15 } \\
\frac{d\,a_{2}(t)}{dt } & = & \{{e_{\rm r}}\mathbf{1},a_{2}(t)\}+i[h_{1},a_{2}(t)]+ \ldots \ , .
\label{egg2}\end{aligned}\ ] ] since @xmath643 does not depend on higher orders , the equations can be solved iteratively .
the solution of ( [ eg1 ] ) with initial condition @xmath644 is given through @xmath645 where @xmath646 is the solution flow corresponding to the scalar hamiltonian @xmath458 .
more precisely , @xmath647 , where @xmath648 is the solution of the classical equations of motion @xmath649 with initial condition @xmath650 .
@xmath651 is the solution of @xmath652 with initial condition @xmath653 .
one can think of ( [ deq ] ) for fixed @xmath393 as an equation for the schrdinger - like unitary evolution induced by the time - dependent hamiltonian @xmath654 on the hilbert space @xmath634 . since @xmath461 is self - adjoint for all @xmath393 , the solution @xmath651 of ( [ deq ] ) is unitary for all @xmath655 . to see that ( [ a0 ] ) is indeed the solution of ( [ eg1 ] ) , note that the mappings @xmath656 defined through ( [ a0 ] ) for @xmath509 , i.e.@xmath657 form a one - parameter group of linear automorphisms on the banach space @xmath658 @xmath659 , since @xmath660 here the group structures of @xmath661 and of the solutions of ( [ deq ] ) are used .
hence @xmath546 is a group and it suffices to check that ( [ a0 ] ) solves ( [ eg1 ] ) at time @xmath662 , which is easy to see . the physical interpretation becomes simpler when translated to states : a `` classical '' particle which started at time @xmath610 at the phase space point @xmath259 with spinor @xmath663 , is at time @xmath513 located at the phase space point @xmath664 with spinor @xmath665 .
hence ( [ deq ] ) implies that @xmath666 one can also think of @xmath546 as being the action on observables of a `` classical '' flow @xmath667 on phase space @xmath668 defined as @xmath669 turning to the higher order corrections ( [ eg15 ] ) , ( [ egg2 ] ) etc . , they are of the form @xmath670 + i_{n}(a_0(t ) , \ldots , a_{n-1}(t))\ ] ] with an inhomogeneity @xmath671 depending only on the known functions @xmath672 @xmath673 .
thus , assuming @xmath674 , one finds @xmath675 in order to solve equation ( [ eg15 ] ) for the subprincipal symbol one needs to know @xmath433 .
however , if one is interested in semiclassical observables with a principal symbol which is a scalar multiple of the identity , e.g. in the position @xmath676 , the last term in ( [ eg15 ] ) vanishes at all times , since , according to ( [ a0 ] ) , @xmath677 is a scalar multiple of the identity for all times . in section [ sdirac ]
the back reaction of the spin of an electron on its translational motion will be discussed on the basis of ( [ eg15 ] ) .
we summarize the preceding discussion on egorov s theorem .
[ egorovt ] let @xmath24 satisfy either ( ig)@xmath81 for @xmath678 and @xmath679 or ( cg ) with @xmath680 .
let @xmath439 be an eigenvalue of @xmath440 of finite multiplicity @xmath474 .
then the classical flow @xmath633 generated by @xmath458 and the solution of ( [ deq ] ) with initial condition @xmath653 exist globally in time . for @xmath681 , @xmath677 given by ( [ a0 ] )
is a solution of ( [ eg1 ] ) and @xmath682 for all @xmath513 . for each @xmath683
there is a constant @xmath684 such that for all @xmath685 $ ] @xmath686 where @xmath687
. * proof . * up to the modifications discussed before , the proof follows easily along the lines of egorov s theorem for scalar valued observables ( cf .
@xcite ) : to make the expansion of the heisenberg equation ( [ heis1 ] ) rigorous , note that @xmath688 with @xmath678 and thus the corresponding hamiltonian vector field is smooth and bounded .
it follows by standard ode techniques @xcite that @xmath689 and hence also @xmath690 , where @xmath691 is given by ( [ a0 ] ) .
thus one can interchange quantization and differentiation with respect to time and obtains @xmath692 -\mathcal{w}_{\varepsilon } \left ( \frac{da_{0}}{dt}(t - s)\right ) \right ) e^{-i\widehat{h}s/\varepsilon } \,.\end{aligned}\ ] ] now , by construction , @xmath693 -\mathcal{w}_{\varepsilon } \left ( \frac{da_{0}}{dt}(t - s)\right ) $ ] is a semiclassical operator in op@xmath694 with vanishing principal symbol .
hence the integrand is really @xmath695 as a bounded operator and ( [ egeq ] ) follows .
@xmath213 this matrix - valued version of egorov s theorem has been discussed several times in the literature @xcite .
* berry connection . * with this preparation
we explain the motivation behind the particular splitting of the terms in ( [ h1w ] ) .
it is of geometrical origin and related to the berry connection . recall that in the born - oppenheimer setting @xmath696 and thus @xmath495 acts as a gauge potential of a connection on the trivial bundle @xmath496 .
its origin is purely geometrical , since it comes from the connection which the trivial connection on the trivial bundle @xmath697 induces on the subbundle defined by @xmath698 . if one assumes that ran@xmath698 is @xmath482-dimensional , the internal rotations along classical trajectories are just phase changes , the so called berry phases , and are due to parallel transport with respect to the berry connection @xcite . in the general case
the second term of @xmath699 in ( [ h1w ] ) , which we denote by @xmath700 corresponds exactly to this parallel transport along the generalized berry connection .
more precisely , the trivial connection on the trivial bundle @xmath701 induces a @xmath702-connection on the subbundle defined by @xmath464 . after unitary rotation @xmath703 the coefficients of this connection on the bundle @xmath704 are @xmath705 in the sense that a section @xmath706 is parallel if @xmath707 .
it is parallel along some curve @xmath708 in @xmath59 if @xmath709 for classical trajectories , where @xmath710 , this condition becomes @xmath711 if @xmath712 , ( [ parallel ] ) is exactly equation ( [ spintransport ] ) for the rotation of the spinor @xmath713 along the trajectory of the particle .
this means if @xmath712 , the spin dynamics corresponds to parallel transport with respect to the berry connection along classical trajectories .
emmrich and weinstein @xcite give a geometric meaning also to the remaining terms in their analog of @xmath432 .
while this is a natural venture in the context of geometric wkb approximation , it seems to be less natural in our approach , since we work in a fixed basis in order to obtain simple analytic expressions .
* wigner function approach . * the previous results on the time - evolution of semiclassical observables translate , by the duality expressed through @xmath714 to the time - evolution of the wigner transform @xmath715 as @xmath716 transport equations for matrix - valued wigner measures were derived in @xcite and applied to the dirac equation in @xcite .
* semiclassical propagator .
* often one is not only interested in the semiclassical propagation of observables , but more directly in a semiclassical expansion of the kernel @xmath717 of the unitary group @xmath718 as in the case of egorov s theorem , generalizing the known results for hamiltonians with scalar symbols to the case of operator - valued symbols is straightforward , whenever the principal symbol @xmath431 of @xmath42 is a scalar multiple of the identity . as in the scalar case ,
@xcite , one makes an ansatz of the form @xmath719where @xmath720 is real valued and the @xmath721 s take values in the bounded linear operators on @xmath634 . demanding ( [ kdef ] ) at time @xmath662 , i.e. @xmath722 , imposes the following initial conditions on @xmath723 and @xmath724 : @xmath725 for later times the coefficients are determined by formally expanding the schrdinger equation for @xmath726 @xmath727 in orders of @xmath192 . at
leading order only @xmath728 contributes and one obtains as in the scalar case @xmath729 the hamilton - jacobi equation for the symbol @xmath431 .
the next to leading order equation is the so called transport equation for @xmath730 : @xmath731 the differential operator @xmath732 is the same as in the scalar case , see @xcite for an explicit formula .
here we just want to point out that the known techniques from the scalar case apply with one modification : as in ( [ eg1 ] ) , also in ( [ transport ] ) @xmath432 contributes as an additional rotation in the transport equation for the leading order term .
since the solution of ( [ hj ] ) exists only until a caustic is reached , the approximation ( [ hj ] ) , ( [ transport ] ) to the propagator is a short time result only .
the extension to arbitrary times is a complicated task , in general @xcite .
ultimately the goal is to approximate expectation values of observables in in the original hilbert space @xmath733 rather than in @xmath734 . before stating a theorem an obvious , but important observation should be made , which seems to have been overlooked , or at least not stressed sufficiently , in related discussions , e.g. , @xcite : we proved that in the case @xmath439 the effective hamiltonian @xmath402 projected on the subspace @xmath735 has a semiclassical limit in the sense of a generalized egorov theorem , in principle , to any order in @xmath192 .
however , the variables @xmath7 and @xmath8 in the rotated representation are _ not _ the canonical variables of the slow degrees of freedom in the original problem .
more precisely , let @xmath736 and @xmath737 be the position and momentum operators of the slow degrees of freedom acting on @xmath26 and let @xmath738 and @xmath739 be the same operators acting on @xmath740
. then @xmath741 and @xmath742 , with a , in general , nonvanishing @xmath192-correction .
physically this means that the quantities which behave like position and momentum in the semiclassical limit are only close to the position and momentum of the slow degrees of freedom , but not equal .
this phenomenon is well known in the case of the nonrelativistic limit of the dirac equation . the newton - wigner position operator and not the standard position operator goes over to the position operator in the pauli equation .
the standard position operator has neither a nice nonrelativistic limit nor , as we will see , a nice semiclassical limit , because of the zitterbewegung .
switching to the newton - wigner position operator corresponds to averaging over the zitterbewegung , or , in our language , to use the position operator @xmath743 in the rotated representation .
we remark that in the born - oppenheimer case , and more generally whenever @xmath104 depends on @xmath7 only , one has @xmath744 .
with this warning we exploit that semiclassical observables do not change after unitary rotation in leading order and state the egorov theorem for the observables in the original representation .
[ egorov2]let @xmath24 satisfy either ( ig)@xmath81 with @xmath678 and @xmath745 or ( cg ) with @xmath680 and let @xmath439 consist of a single eigenvalue of @xmath440 of finite multiplicity @xmath474 .
let @xmath746 such that @xmath747=0 $ ] and @xmath748 .
let @xmath749 and define @xmath677 is in ( [ a0 ] ) .
then for each @xmath683 there is a constant @xmath750 such that for all @xmath685 $ ] @xmath751 for @xmath752 , with @xmath753 , one obtains as a special case of ( [ eg2 ] ) that @xmath754 corollary [ egorov2 ] follows from theorem [ egorovt ] and a straightforward expansion in @xmath106 of the terms to be estimated after rotation with @xmath43 .
we apply the adiabatic perturbation theory to the one - particle dirac equation with slowly varying external potentials , i.e. to @xmath755acting on @xmath756 . here
@xmath757 is the vector potential of an external magnetic field @xmath758 and @xmath759 the potential of an external electric field @xmath760 . for the dirac matrices
@xmath480 , @xmath481 we make the standard choice @xmath761where @xmath762 denotes the vector of the pauli spin matrices .
the small parameter @xmath235 controls the variation of the external potentials . to keep track of the size of the error terms , in this section all physical constants , including @xmath763 , are displayed .
transforming to the macroscopic space - scale @xmath764 one obtains the dirac hamiltonian @xmath765 and we are interested in the solution of the time - dependent dirac equation for times of order @xmath766 , i.e. in solutions of @xmath767 for @xmath768 .
the solutions of ( [ tdde ] ) for small @xmath106 approximately describe the dynamics of electrons , resp.positrons , in weak fields , as in storage rings , accelerators , or cloud chambers , for example .
@xmath769 is the weyl quantization of the matrix - valued function @xmath770on phase space @xmath771 , where now weyl quantization is in the sense of @xmath772 , i.e. on the right hand side of ( [ weyl def ] ) @xmath192 must be replaced by @xmath773 .
@xmath763 appears here for dimensional reasons and is a fixed physical constant .
the small parameter of the space - adiabatic expansion is @xmath192 .
@xmath774 has two two - fold degenerate eigenvalues @xmath775 with the corresponding eigenprojections @xmath776 where @xmath777 .
obviously @xmath778 whenever @xmath779 is uniformly bounded .
therefore the corresponding subspaces are adiabatically decoupled and the effective dynamics on each of them can be computed using our general scheme . assuming @xmath780 and @xmath781 , one finds that @xmath782 and thus the assumptions from section [ spro ] are satisfied .
in particular , @xmath769 is essentially self - adjoint on @xmath783 and @xmath784 on @xmath785 . to be consistent with the notation from the previous sections ,
let @xmath260 @xmath786 be the projector on the electron band .
the reference subspace for the electrons is @xmath787 and it is convenient to define it as the range of @xmath788 in @xmath789 .
the only choice left is the one of @xmath703 or , equivalently , of a basis @xmath790 of ran@xmath464 .
since the degeneracy of ran@xmath464 is related to the spin of the electron , a natural choice is the @xmath791-representation with respect to the `` mean''-spin @xmath792 which commutes with @xmath774 @xcite .
the eigenvectors @xmath793 of the operator @xmath794 in ran@xmath464 are @xmath795 we abbreviated @xmath796 for the velocity . the relevant part of @xmath272 for the analysis of the electron band is thus given by @xmath797 @xmath798 with @xmath799 .
of course the positron part indicated by @xmath800 s would be given through charge conjugation . in our construction
we want to emphasize , however , that no specification is needed in order to determine the expansion of the effective electron hamiltonian @xmath801 up to arbitrary order .
an alternative way to arrive at the same @xmath802 is to note that the foldy - wouthuysen transformation @xmath803 , c.f .
@xcite , diagonalizes the free dirac hamiltonian @xmath804 , i.e. @xmath805 with @xmath806 . including the fields @xmath807 then diagonalizes @xmath774 . for the principal symbol of @xmath808 one finds of course @xmath809for the subprincipal symbol after a lengthy but straightforward calculation our basic formula ( [ h1w ] ) yields @xmath810 note that the factor @xmath811 comes from the fact that the @xmath32th term in the space - adiabatic expansion carries a prefactor @xmath812 . defining @xmath813 one
concludes that @xmath814 we remark that the second term in ( [ h1w ] ) , the `` berry term '' , does not coincide with any of the terms in ( [ odef ] ) .
indeed , the compact expression ( [ odef ] ) is obtained only through cancellations in more complicated expressions coming from both terms contributing in ( [ h1w ] ) .
we summarize our results on the adiabatic decoupling and the effective dynamics for the dirac equation in the following [ diractheorem ] let @xmath815 and @xmath816 .
then there exist orthogonal projectors @xmath817 with @xmath818 such that @xmath819=\mathcal{o}_{0}(\varepsilon^{\infty } ) $ ] , and there exists a unitary @xmath43 and @xmath820 with @xmath821 such that @xmath822 here @xmath823 and @xmath824 are semiclassical operators on @xmath825 with @xmath826 and @xmath827 for all @xmath828 , where @xmath829 is constructed as in section [ suni ] .
in particular , @xmath830 is given by ( [ h1d ] ) and thus @xmath831 analogous results hold for @xmath832 .
the errors in ( [ hdexp ] ) and ( [ hel ] ) are in the norm of bounded operators on @xmath833 , resp . on @xmath825 .
_ _ according to the effective hamiltonian ( [ hel ] ) the @xmath834-factor of the electron equals @xmath835 .
there would be no problem to add to the dirac hamiltonian the standard subprincipal symbol @xcite , which accounts for the slightly larger @xmath834-factor of real electrons .
blount @xcite computes the second order effective hamiltonian @xmath836 , which he finds to be proportional to @xmath837 .
@xmath836 is a sum of terms allowed by dimensional reasoning , i.e.proportional to @xmath838 , @xmath839 , @xmath840 , @xmath841 , @xmath842 .
second order corrections seem to be of interest for the dynamics of electrons in storage rings . ignoring the contribution @xcite , nonrigorous expansions are @xcite and @xcite . equipped with @xmath843 and @xmath844
we can apply the general results of section [ ssemi ] on the semiclassical limit to the dirac equation .
let @xmath845 be the hamiltonian flows generated by @xmath846 on phase space @xmath771 and let @xmath847 , @xmath848 , be a semiclassical observable in the unrotated hilbert space which does not depend on spin . from corollary [ egorov2 ] we conclude for each @xmath683 the existence of a constant @xmath849 such that for all @xmath685 $ ] @xmath850 where @xmath851 .
hence , to leading order , states in the range of @xmath852 behave like classical relativistic electrons and states in the range of @xmath853 like classical relativistic positrons .
we emphasize that , in general , @xmath817 are not spectral projections of @xmath769 , since the variation of @xmath854 can be larger than the mass gap @xmath855 .
hence in the limit of slowly varying potentials a natural characterization of `` electronic '' and `` positronic '' subspaces is obtained which does _ not _ come from spectral projections of the free or full dirac hamiltonian .
next we discuss the leading order spin dynamics , which in the first place requires to figure out which operator represents the spin of the electron .
there has been a considerable discussion on this point , cf .
@xcite , with no general consensus reached .
we suspect that the problem is void .
the wave function is spinor valued and what is observed is the spatial splitting of different spinor components in inhomogeneous magnetic fields .
hence we should pick the `` spin observable '' @xmath856 such that the splitting can nicely be attributed to it .
e.g. , in a magnetic field with gradient along the @xmath158-direction the eigenvectors of @xmath857 should have the property that their spatial support goes either parallel to @xmath858 or to @xmath859 , but should not split . in view of ( [ hel ] )
a natural choice is to take as spin operator the vector of pauli - matrices @xmath860 in the rotated electronic subspace . in the original hilbert space
this amounts to @xmath861 where @xmath792 is the `` mean '' spin defined before .
the leading order semiclassical approximation for @xmath862 follows from theorem [ egorovt ] . for each @xmath863
there is a constant @xmath864 such that for @xmath685 $ ] @xmath865 where @xmath866 , @xmath867 , is obtained as the solution of @xmath868\ ] ] with initial condition @xmath869 .
this follows from the equations ( [ a0 ] ) and ( [ deq ] ) by setting @xmath870 .
to solve equation ( [ bmt0 ] ) one makes an ansatz @xmath871 with @xmath872 . using @xmath873=2\,i\,\varepsilon_{nmk}\,\sigma_k$ ]
, one finds that the spin- or `` magnetization''-vector @xmath874 is given as the solution of @xmath875 ( [ bmt1 ] ) is the bmt - equation @xcite on the level of observables .
it was derived by bargmann , michel and telegdi in 1959 on purely classical grounds as the simplest lorentz invariant equation for the spin dynamics of a classical relativistic particle .
the semiclassical limit of the dirac equation has been discussed repeatedly and we mention only some recent work .
yajima @xcite considers time - dependent external fields and proves directly a semiclassical expansion for the corresponding propagator . as mentioned already at the end of section [ ssemi
] , this program is mathematically rather involved , since one faces the problem of caustics in the classical flow , and different expansions have to be glued together in order to obtain results valid for all macroscopic times . based on the same approach bolte and keppeler @xcite derive a gutzwiller type trace formula .
since @xmath876 and @xmath877 are isospectral and since ( [ hdexp ] ) holds , a trace formula for the eigenvalue statistics of @xmath769 could as well be derived from the semiclassical propagator of @xmath878 . as argued in section [ ssemi ] ,
the latter is somewhat easier to obtain . in @xcite the semiclassical limit of the dirac equation
is discussed using matrix - valued wigner functions .
their results hold for an arbitrary macroscopic time interval , but fuse , as does the wkb approach , adiabatic and semiclassical limit .
no higher order corrections seem to be accessible and the results are weaker than ours in the sense that the approximations do not hold uniformly in the states .
this leads us to the next natural question : what can be said about higher order corrections ? while in general one would need @xmath836 , according to ( [ eg15 ] ) the semiclassical limit of observables of the type @xmath879 , @xmath880 , can be determined without this explicit information .
for such a scalar symbol the principal symbol @xmath881 , i.e. the solution to ( [ eg1 ] ) , will remain scalar and thus its commutator with @xmath882 in ( [ eg15 ] ) vanishes identically for all times .
the solution @xmath883 of ( [ eg15 ] ) with initial condition @xmath884 , is not scalar , in general .
hence , at this order there is back reaction of the spin dynamics on the translational motion .
we illustrate this point for the position operator @xmath885 and refer to @xcite for a general analysis of the higher order effects in the semiclassical dynamics of dirac particles .
now @xmath886 and @xmath887 is obtained , according to equation ( [ eg15 ] ) , as the solution of @xmath888 - \{h_{\mathrm{e,1}},x_0(t)\}\ ] ] with initial condition @xmath889 .
the homogeneous part of this equation is just the classical translational and spin motion and the inhomogeneity is @xmath890 which is not scalar and thus responsible for the splitting of trajectories of electrons with distinct spin orientation .
hence , as in ( [ hoe ] ) , @xmath891where @xmath546 is the `` classical flow '' defined through ( [ uddef ] ) . without claim of rigor , we observe in ( [ odef ] ) that for small velocities @xmath892 one has @xmath893 let us further assume that @xmath894 , then @xmath895 and thus according to ( [ x1 ] ) , ( [ 84 ] ) the correction to the velocity is proportional to @xmath513 , corresponding to a constant force with absolute value @xmath896 , as expected for a spin-@xmath897 particle .
the basic formulae ( [ h1d ] ) , ( [ h21 ] ) can be applied , in essence in a mechanical fashion , to any concrete quantum problem with two provisos .
first of all the problem has to be cast into the general form ( [ a ] ) and secondly one must have sufficient information on the principal symbol @xmath631 . depending on @xmath64 considerable simplifications of ( [ h1d ] ) , ( [ h21 ] ) may be in force , one example being the effective hamiltonian of the time - adiabatic theorem studied in section [ sta ] . as a net result ,
if the conditions of the space - adiabatic theorem [ th invariant subspace ] are satisfied , the full schrdinger equation is approximated by an effective schrdinger equation referring to a specific relevant energy band .
the errors are estimated and , in general , the time scale of validity is much larger than the one which can be reached within a semiclassical approximation .
we focused our interest on a single relevant energy band .
no information on the complement is needed except for global quantities like the resolvent @xmath898 . in previous investigations
@xcite all energy bands are treated simultaneously .
an example which would not fall under such a scheme is nonrelativistic qed , which governs electrons coupled to the quantized radiation field . in this case
the principal symbol has a two - fold degenerate eigenvalue at the bottom of the spectrum separated by a gap from the continuous spectrum , provided @xmath899 is sufficiently small and there is a suitable infrared cutoff @xcite .
the main restriction of our work is the gap condition of section [ spro ] .
there are two standard mechanisms of how this condition is violated .
( 1 ) there are two ( or possibly more ) locally isolated energy bands of constant multiplicity which cross on a lower dimensional submanifold . away from the crossing region
the wave function in one band is governed by the effective hamiltonian discussed before .
if the wave function comes close to the crossing manifold , there is a certain probability to make a transition to the other band . in rather specific model systems
such transitions have been studied in considerable detail @xcite .
( 2 ) @xmath64 has a smooth band of constant multiplicity bordering the continuous spectrum _ without _ gap .
this is the rule in models from nonrelativistic qed with massless photons .
results for the massless nelson model @xcite indicate that smoothness of @xmath260 suffices also in general for adiabatic decoupling at leading order with intraband dynamics generated by @xmath900 as defined by ( [ hs ] ) .
however , the expansion stops at this stage .
physically , the electron looses energy through radiation , which means that the next order correction must be dissipative .
> from the physics point of view the dynamics of molecules and the dynamics of electrons in a solid are the two most prominent areas of application for the space - adiabatic perturbation theory .
the former has been discussed already in section [ sbo ] .
bloch electrons do not quite fall into our scheme , since the classical phase space is @xmath901 , @xmath902 a flat @xmath903-dimensional torus .
this requires substantial changes which are discussed in @xcite .
pseudodifferential operators with operator - valued symbols have been widely discussed in the literature .
the results presented in this appendix can be found in @xcite .
we start with some notation .
let @xmath904 be a banach space , then @xmath905 denote the space of @xmath906-valued continuous functions on @xmath907 . in the same spirit
we will employ the notation @xmath908 , @xmath909 , with the obvious meaning .
note that , in the special case where @xmath910 is an hilbert space , one has @xmath911 .
the space of the bounded operators on @xmath904 will be denoted as @xmath912 .
let @xmath779 be a @xmath58-valued rapidly decreasing smooth function on @xmath59 , i.e. @xmath913 .
if we denote by @xmath914 the fourier transform of @xmath779 then , by fourier inversion formula , @xmath915 where the integral is a bochner integral for @xmath916-valued functions .
this suggest to define an operator @xmath917 , called the * weyl quantization * of @xmath779 , by substituting @xmath918 with @xmath919 where @xmath920 is multiplication by @xmath17 and @xmath921 in @xmath922 .
the exponential is defined by using the spectral theorem and it is explicitly given by @xmath923 thus @xmath924 and , in particular , @xmath925 which implies that @xmath926 belongs to @xmath927 provided the fourier transform of @xmath779 belongs to @xmath928 @xmath929 .
we will also use the notation @xmath930 in order to emphasize the @xmath106 -dependence .
a function @xmath935 belongs to the symbol class @xmath936 @xmath937with @xmath938 and @xmath939 if for every @xmath940 there exists a positive constant @xmath941 such that @xmath942 for every @xmath943 , where @xmath944 .
it is convenient to introduce a special notation for such classes of operators acting on @xmath952 , called pseudodifferential operators , @xmath953 in the following we will sometimes denote @xmath945 simply as @xmath954 and we will use the shorthand @xmath955 . notice that @xmath956 for any @xmath957
. * notation . * denote by @xmath959 the space of @xmath904-valued , @xmath18 times continuously differentiable functions on @xmath960 such that all the derivatives up to the order @xmath18 are bounded . equipped with the norm
@xmath961 it is a banach space .
assume that @xmath779 @xmath975 is a normal symbol which is elliptic , in the sense that there exists a constant @xmath976 such that @xmath977 then the pointwise inverse @xmath978 exists and belongs to @xmath979 .
[ prop point inversion ] * proof .
* as a consequence of the spectral theorem ( for bounded normal operators ) one has @xmath980 similar bounds on derivatives can be obtained by noticing that @xmath981 and applying the chain rule.@xmath213 the crucial result for pseudodiferential calculus is the following .
one can define an associative product in the space of classical symbols which corresponds to the composition of the operators . given @xmath982 and @xmath983 we know that @xmath926 and @xmath984 map @xmath985 into itself .
then @xmath986 is still an operator on @xmath987 and one can show that there exists a unique @xmath192-dependent symbol @xmath988 such that @xmath989 the symbol @xmath990 is called the * weyl product * ( or the twisted product ) of the symbols @xmath779 and @xmath991 . for the proof of the following proposition in the operator valued case
we refer again to @xcite .
in particular , @xmath995 and @xmath996 are algebras with respect to the weyl product @xmath997 .
since the product @xmath998 depends on @xmath192 by construction , one can expand ( [ weyl product ] ) in orders of @xmath106 . to this end , it is convenient to define suitable classes of @xmath192-dependent symbols , called * semiclassical symbols * , which roughly speaking are close to a power series in @xmath192 of classical symbols with nicer and nicer behavior at infinity .
our definition is a special case of the standard ones ( see @xcite ) .
a map @xmath999 is called a semiclassical symbol of order @xmath1000 and weight @xmath1001 if there exists a sequence @xmath1002 with @xmath1003 such that for every @xmath1004 one has that @xmath1005 belongs to @xmath1006 _ uniformly _ in @xmath192 , in the following sense : for any @xmath409 there exists a constant @xmath1007 such that for any @xmath1008 one has @xmath1009 where @xmath1010 is the @xmath18-th frchet semi - norm in @xmath1011 introduced in ( [ seminorms ] ) .
one calls @xmath1012 and @xmath1013 the * principal symbol * and the * subprincipal symbol * of @xmath1014 .
the space of semiclassical symbols of order @xmath1000 and weight @xmath1015 will be denoted as @xmath1016 .
if condition ( [ semi symb ] ) is fulfilled , one writes @xmath1017and one says that @xmath779 is asymptotically equivalent to the series @xmath1018 in @xmath1016 . if @xmath779 is asymptotically equivalent to the series in which @xmath1019 for every @xmath1020 , we write @xmath1021 . to be precise ,
we should write @xmath1021 in @xmath1022 , but the latter specification is omitted whenever it is unambiguous from the context . let be @xmath1002 an arbitrary sequence such that @xmath1024 then there exists @xmath1025 such that @xmath1026 in @xmath1022 and @xmath779 is unique up to @xmath1027 , in the sense that the difference of two such symbols is @xmath1028 in @xmath1016 .
the semiclassical symbol @xmath779 is called a * resummation * of the formal symbol @xmath1023.[prop resummation ] for example @xmath1036 is simply given by the pointwise product @xmath1037 and @xmath1038 where @xmath449 denotes the poisson bracket on @xmath1039 , defined through @xmath1040 notice that , in general , @xmath1041 since operator - valued derivatives do not commute , in particular @xmath1042 .
the usual poisson algebra is recovered in the special case in which one of the two arguments is a multiple of the identity , i.e. @xmath1043 . as a consequence of the previous result
, it is convenient to introduce the space of the formal power series with coefficients in @xmath1039 . this space , equipped with the associative product given by ( [ composition ] ) and with the involution defined by taking the adjoint of every coefficient , will be called the * algebra of formal symbols * over @xmath58 .
in particular we will denote as @xmath1044 the subspace of the formal power series with a resummation in @xmath1022 , i.e.@xmath1045 in the context of formal power series , the product defined by ( [ composition ] ) will be called the * moyal product * and denoted simply as @xmath1046 .
notice that @xmath1046 defines a map from @xmath1047 to @xmath1048 .
the moyal product can also be regarded as a map from @xmath1049 to @xmath1050 , where in ( [ composition ] ) the operator @xmath779 and its derivatives act on the vector @xmath991 and its derivatives . to sum up the previous discussion ,
we wish to point out that one can prove statements on three levels : formal symbols ( i.e. formal power series ) , semiclassical symbols , and operators on @xmath1051 .
a simple example illustrates the interplay between these levels .
suppose that two formal symbols @xmath1052 and @xmath1053 moyal commute , i.e. @xmath1054_{\#}=a\#b - b\#a=0 $ ] .
let @xmath1055 and @xmath1056 be any two resummations of @xmath779 and , respectively , @xmath991 . since we know _ a priori _ ( by prop .
[ prop composition ] ) that the weyl product @xmath1057 belongs to @xmath1058 it follows that the weyl commutator @xmath1059_{\tilde{\#}}$ ] is asymptotically close to zero in @xmath1058 , which can be rephrased in the following way : for any @xmath1060 there exists a constant @xmath1007 such that for any @xmath1008 one has @xmath1061_{\tilde{\#}}\right\| _ { k}^{(m_{1}+m_{2}-n\rho ) } \leq c_{n , k}\ \varepsilon^{n}\,.\ ] ] if @xmath83 we obtain that definitely @xmath1062 for some @xmath1004 and then prop .
[ prop caldvaill ] assures that the operator commutator @xmath1063 $ ] can be bounded in the @xmath927-norm . moreover , for @xmath1064 , we can conclude that @xmath1063 $ ] is a smoothing operator ( i.e. it belongs to @xmath1065 ) and in particular one can prove that it is a `` small '' bounded operator between the sobolev spaces @xmath1066 and @xmath1067 for any @xmath1068 . to be precise , for any @xmath1069 there
exist a constant @xmath1070 such that @xmath1071\right\|_{\mathcal{b}(h^{q},h^{q+r } ) } \leq c_{n , q , r}\,\varepsilon^{n}\ ] ] for any @xmath1008 , where @xmath1066 stands for @xmath1072 .
notice that for @xmath1073 and @xmath1074 it is not possible to conclude from @xmath1054_\#=0 $ ] that @xmath1063 $ ] is a bounded operator , since it could happen for example that @xmath1059_{\tilde{\#}}=e^{-\frac{1}{\varepsilon } } p^{m},$ ] which is asymptotically close to zero in @xmath1075 .
in the following we will use the same symbol for an element in @xmath1076 and its expansion in @xmath1077 . as suggested by the preceding discussion
, we introduce the following synthetic notation . with a little abuse , we will employ the same notation for pseudodifferential operators too , i.e. we write @xmath1080 if @xmath1078 . as noticed above this
is a strong concept of closeness , since it implies that @xmath1081 is a smoothing operator .
compare with the following weaker concept .
* notation .
* let be @xmath1082 and @xmath723 two ( @xmath106-dependent ) operators on @xmath26 .
we will say that @xmath1083 if for every @xmath1004 there exists a constant @xmath1084 such that @xmath1085 for every @xmath1008 .
in such a case we will say that @xmath1082 is @xmath343-close to @xmath723 .
k. heinemann and d. p. barber .
_ the semiclassical foldy - wouthuysen transformation and the derivation of the bloch equation for spin-@xmath897 polarized beams using wigner functions _ , preprint arxiv.org physics/9901044 ( 1999 ) . | we study approximate solutions to the time - dependent schrdinger equation @xmath0 with the hamiltonian given as the weyl quantization of the symbol @xmath1 taking values in the space of bounded operators on the hilbert space @xmath2 of fast `` internal '' degrees of freedom . by assumption @xmath1
has an isolated energy band . using a method of nenciu and sordoni @xcite
we prove that interband transitions are suppressed to any order in @xmath3 . as a consequence , associated to that energy band there exists a subspace of @xmath4 almost invariant under the unitary time evolution .
we develop a systematic perturbation scheme for the computation of effective hamiltonians which govern approximately the intraband time evolution . as examples for the general perturbation scheme
we discuss the dirac and born - oppenheimer type hamiltonians and we reconsider also the time - adiabatic theory .
math - ph/0201055 |
this work is based on our previous paper @xcite where we set the general dynamical - system formalism for accretion of perfect fluids onto static spherically symmetric black holes .
we keep using the same notation for the thermodynamic functions of the fluid and the hamiltonian @xmath0 .
so , @xmath1 , @xmath2 , @xmath3 , @xmath4 , @xmath5 , @xmath6 , and @xmath7 are the baryon number density , specific enthalpy ( enthalpy per particle ) , energy density , specific entropy , pressure , temperature , and four - velocity vector , respectively . in a locally inertial frame ,
the three - dimensional speed of sound @xmath8 is given by @xmath9 .
when the entropy @xmath4 is constant , which is the case for accretion of perfect fluids onto static spherically symmetric black holes , this reduces to @xmath10 .
the aim of this short work is two - fold : ( 1 ) generalize the dynamical - system formalism for accretion of perfect fluids to metrics of the form @xmath11 this will allow us to generalize the properties of the accreting fluid intended for future use .
( 2 ) obtain new interesting solutions not discussed so far in the scientistic literature . in , ( @xmath12 )
are any functions of the radial coordinate @xmath13 assumed to be well - defined and positive - definite in the regions where the killing vector @xmath14 is timelike and their ratio @xmath15 is positive - definite on the horizons too .
for the metric to describe a black hole solution , the equations @xmath16 and @xmath17 should have the same set of solutions with same multiplicities . in our applications
we restrict ourselves to the cases where the global structure of the spacetime is well - determined by ( @xmath12 ) . to keep the analysis general
, we do not assume asymptotic flatness of the metric as we intend to apply it to the de sitter and anti - de sitter - like black holes .
this metric form generalizes the one used in refs .
@xcite ; in ref . @xcite we restricted ourselves to the case @xmath18 and @xmath19 .
it is easy to show that equations ( 5 ) , ( 7 ) , ( 11 ) , ( 23 ) , ( 24 ) , ( 25 ) of ref .
@xcite generalize respectively to @xmath20 where ( @xmath21 ) are constants of motion , @xmath22 , and @xmath23 is the three - velocity of a fluid element as measured by a locally static observer .
the second line in expresses the law of particle conservation , @xmath24 , and @xmath25 is the constant of motion @xmath26 [ @xmath14 is a timelike killing vector ] .
this constant is the inertial - equivalent generalization of the energy conservation equation @xmath27 @xcite .
the constant @xmath28 in can be written as @xmath29 where 0 " denotes any reference point ( @xmath30 ) from the phase portrait ; this could be a cp , if there is any , spatial infinity ( @xmath31 ) , or any other reference point .
we can thus write @xmath32 the equation of state ( eos ) of the fluid may be given in the form @xmath33 or equivalently in the form @xmath34 @xcite .
it has been shown @xcite that ( @xmath35 ) , the specific enthalpy , and the three - dimensional speed of sound satisfy @xmath36 where the prime denotes derivative with respect to @xmath1 . equations and imply that the specific enthalpy @xmath2 depends explicitly on ( @xmath37 ) only .
there is no explicit dependence on @xmath38 .
as we shall see in the next subsection , this implies that the hamiltonian @xmath0 , which defines the dynamical system , will also depend explicitly on ( @xmath37 ) only .
this fact will have consequences on the location of the critical points ( cp s ) .
if the fluid had further properties , say , being isothermal or polytropic , @xmath2 and @xmath0 can be expressed explicitly in terms of ( @xmath39 ) , as this is done in the following section .
horizons @xmath40 are defined by @xmath41 and @xmath42 or simply by @xmath42 since the equations @xmath41 and @xmath42 are assumed to have the same set of solutions with same multiplicities .
the zeros of @xmath42 determine the regions in three - space where the fluid flow takes place : these are the regions where @xmath14 is timelike . note that the case @xmath43 , corresponding either to @xmath44 ( no fluid ) or to @xmath45 and any @xmath1 ( no flow ) , is not interesting .
so , we assume @xmath46 . as the fluid approaches , or emanates from , any horizon ( @xmath47 ) , @xmath48 approaches 0 .
three cases emerge from : 1 .
@xmath49 and @xmath1 may converge or diverge there ( @xmath47 ) ; 2 .
@xmath50 and @xmath1 diverges there .
the fluid cumulates near the horizon resulting in a divergent pressure which repulses the fluid backwards @xcite ; 3 .
@xmath51 assumes any value between 0 and 1 there .
this yields a divergent @xmath1 as @xmath47 . since @xmath52 there is no reason that the fluid cumulates near the horizon : the flow continues until all fluid particles have crossed the horizon .
we rule out this case for it is not physical .
this conclusion , due to the law of particle conservation , is general and it does not depend on the fluid characteristics . since a non - perfect simple fluid ( containing a single particle species ) also obeys the law of particle conservation , these conclusions remain valid for real fluids too .
if the fluid had a uniform pressure , that is , if the fluid were not subject to acceleration , the specific enthalpy @xmath2 reduces to the particle mass @xmath53 and the first equation in reduces to @xmath54 along the fluidlines .
this is the well know energy conservation law which stems from the fact that the fluid flow is in this case geodesic . now ,
if the pressure throughout the fluid is not uniform , acceleration develops through the fluid and the fluid flow becomes non - geodesic ; the energy conservation equation @xmath55 , which is no longer valid , generalizes to its relativistic equivalent @xcite @xmath56 as expressed in the first equation in .
let the hamiltonian @xmath0 of the dynamical system be proportional to @xmath57 , which is a constant of motion .
substituting the first equation in into the first equation in yields @xmath58 with @xmath0 given by , the dynamical system reads @xmath59 ( here the dot denotes the @xmath60 derivative where @xmath60 is the time variable of the hamiltonian dynamical system ) . in it
is understood that @xmath13 is kept constant when performing the partial differentiation with respect to @xmath61 in @xmath62 and that @xmath61 is kept constant when performing the partial differentiation with respect to @xmath13 in @xmath63 .
the critical points ( cps ) of the dynamical system are the points ( @xmath64 ) where the rhs s in are zero .
to take advantage of the calculations made in @xcite we introduce the radial coordinate @xmath65 and the notation @xmath66 defined by @xmath67 the hamiltonian takes the form @xmath68 the derivative @xmath69 has been evaluated in ref
. @xcite by [ eq .
( 40 ) of ref .
@xcite ] : @xmath70.\ ] ] using @xmath71\\ = \frac{h^2}{1-v^2}\big[\frac{{{\rm d}}a}{{{\rm d}}r}-2a^2a~\frac{{{\rm d}}\ln(\sqrt{a}c)}{{{\rm d}}r}\big],\end{gathered}\ ] ] we obtain @xmath72.\end{aligned}\ ] ] introducing the notation @xmath73 and @xmath74 where @xmath75 is any function of @xmath13 , the following equations provide a set of cps that are solutions to @xmath76 and @xmath77 : @xmath78 where @xmath79 is the three - dimensional speed of sound evaluated at the cp .
the first equation states that at a cp the three - velocity of the fluid equals the speed of sound .
the second equation determines @xmath80 once the eos , @xmath33 or @xmath34 , is known . from the set of equations we see that the metric function @xmath38 does not enter explicitly in the determination of the cp s ; however , it does that implicitly via the successive dependence of @xmath2 on @xmath1 , of @xmath1 on @xmath61 , and of @xmath61 on @xmath81 . other sets of cps , solutions to @xmath76 and @xmath77 , may exist too .
for instance , we may have ( 1 ) @xmath82 and @xmath83 which , by and , yield @xmath76 and @xmath77 without having to impose the constraint @xmath84 at the cp .
this corresponds to a double - root horizon of an extremal black hole .
when this is the case , the accretion becomes transonic well before the fluid reaches the cp , which is the horizon itself ( recall that the three - velocity @xmath61 , as the fluid approaches the horizon , tends to @xmath85 ) .
however , extremal black holes are unstable and whatever accretes onto the hole modifies its mass making it non - extremal so that @xmath83 no longer holds
. we may also have ( 2 ) @xmath86 at some point where @xmath14 is timelike , which may hold only for non - ordinary ( dark , phantom , or else ) accreting matter .
the isothermal eos is of the form @xmath87 with @xmath88 where @xmath89 , the so - called state parameter , obeys the constraints @xmath90 .
the differential equation reads @xmath91 yielding @xmath92 where we have used . using this and in we obtain @xmath93 where
all the constant factors have been absorbed into the redefinition of the time @xmath60 and the hamiltonian @xmath0 .
the polytropic equation of state is @xmath94 where @xmath95 and @xmath96 are constants . inserting into the differential equation
, it is easy to determine the specific enthalpy by integration @xcite @xmath97 where we have introduced the baryonic mass @xmath53 . introducing the constant @xmath98 then using
, @xmath2 takes the form @xmath99.\ ] ] finally , the hamiltonian reduces to @xmath100 ^ 2,\ ] ] where @xmath101 has been absorbed into a re - definition of ( @xmath102 ) .
the three - dimensional speed of sound is obtained from @xmath103 since @xmath104 , this implies @xmath105 and , particularly , @xmath106 if there are cps to the hamiltonian system .
equation bears a striking similarity with eq .
( 2.249 ) on page 119 of ref .
@xcite . using the second equation in and ,
we rewrite @xmath107 as @xmath108 substituting into , we arrive at @xmath109 this equation along with the second line in take the following expressions at the cps @xmath110 for a given value of the positive constant @xmath111 , the resolution of this system of equations in ( @xmath64 ) provides all the cps , if there are any . in ref .
@xcite we worked with @xmath112 and @xmath19 reducing to @xmath113 which is the correct expression of eq .
( 112 ) of ref .
@xcite . in both equations ( 111 ) and ( 112 ) of sec .
vi of ref .
@xcite , one should replace the constant factor @xmath114 by 1 .
the presence of this extra factor did not affect the results , solutions , and conclusions made in ref .
@xcite ; however , some new interesting solutions have been missed in its sec .
vi on polytropic fluids .
besides the results and conclusions we have discussed so far in the two first sections of this work , we aim ( 1 ) to re - derive the same solutions derived in sec .
vi of ref .
@xcite , using the correct expression , and ( 2 ) to construct new solutions .
@xcite , we considered three models of f(r ) gravity @xcite . for the model of ref . @xcite the black hole solution is of the form @xmath112 and @xmath19 with @xmath115 following the notation of ref .
@xcite , we employ in this work the symbol @xmath66 " for the metric component , @xmath116 , and the symbol f " for the function f(r ) defining the f(r)-gravity model . the solutions shown in fig . 5 of ref
. @xcite , which depict the accretion of a polytropic perfect fluid onto an anti - de sitter - like @xmath117 black hole , are re - derived using the same values of the parameters : @xmath118 , @xmath119 , @xmath120 , @xmath121 , and @xmath122 .
the re - derived solutions using the correct expressions and are plotted in fig .
[ fig1 ] of this work .
+ + the first thing we note is that the re - derived solution corresponding to @xmath123 , right panel of fig .
[ fig1 ] , has no cps .
apart from this , the re - derived solutions have the same characteristics as those shown in fig . 5 of ref .
@xcite . + the solutions depicting the accretion of a polytropic perfect fluid onto a de sitter - like @xmath117 black hole , which are shown in fig .
[ fig2 ] of this work , have been constructed using the same values of the parameters used in fig .
6 of ref .
@xcite : @xmath118 , @xmath119 , @xmath124 , @xmath125 , and @xmath126 .
the magenta and blue solutions were discovered in ref .
the new solutions are the semi - cyclic critical black plots of fig .
the first semi - cyclic solution represents a supersonic accretion from the cosmological horizon , where the initial three - velocity is almost @xmath85 , then it becomes subsonic passing the cp , and it vanishes on the event horizon .
the accretion is followed by a flowout back to the cosmological horizon reversing all the details .
the second semi - cyclic solution is a flowout from the event horizon with an initial three - velocity in the vicinity of @xmath127 , which decreases gradually until it is sonic at the cp then zero at the cosmological horizon .
this flowout is then followed by an accretion back to the event horizon .
notice that on the horizons , @xmath128 ( event horizon ) or @xmath129 ( cosmological horizon ) , the pressure of the fluid diverges as @xcite @xmath130 if there @xmath45 .
this explains why the fluid , once it reaches any horizon with vanishing three - velocity , it is repulsed backward under the effect of its own pressure . as the value of the hamiltonian exceeds the critical value @xmath131 , cyclic flows between the two horizons form .
these flows are sandwiched by two separate branches corresponding to supersonic accretion and flowout , as depicted by the magenta and blue plots of fig .
the separation between these supersonic branches increases with the value of the hamiltonian resulting in faster accretion and flowout while the cyclic flow tends to become more and more nonrelativistic . for the f(r ) model of ref .
@xcite we constructed the constant - curvature black hole solution @xcite @xmath132r^2}-\frac{r_0}{12}~r^2,\ ] ] where @xmath133 here @xmath134 , ( @xmath135 ) are proportional constants @xcite @xmath136 and the mass scale @xmath137 at the present epoch
@xcite @xmath138 taking @xmath139 , we found @xcite @xmath140 the physical solution corresponds to @xmath141 . using the correct equation and eq . , which takes the form @xmath142},\ ] ]
we construct the new solutions , shown in fig .
[ fig3 ] , using the same values of the parameters used in fig . 7 of ref .
@xcite : @xmath118 , @xmath143 , @xmath144 , @xmath125 , @xmath126 , @xmath145 , @xmath146 , and @xmath147 and @xmath148 are given by taking the lower sign corresponding to the physical solution @xmath141 .
these new solutions have the same characteristics of those depicted in fig . [ fig2 ] .
only the semi - cyclic critical black plots represent new solutions : the solution depicted by the magenta plot of fig .
[ fig3 ] is not new and was discovered in ref .
in this addendum we have first generalized the dynamical - system procedure describing the accretion / flowout of perfect fluids to all black holes endowed with spherical symmetry .
this is needed for many future investigations @xcite . in our dynamical - system procedure
we took the radial coordinate and the three - velocity as dynamical variables of the hamiltonian , which is proportional to the square of the constant of motion @xmath26 [ @xmath14 is a timelike killing vector ] .
this constant is the relativisitc equivalent generalization of the energy conservation equation @xmath27 @xcite .
we have shown that the de sitter - like black holes , of different f(r)-gravity models , present cyclic non - critical flows and semi - cyclic critical flows all characterized by a vanishing three - velocity on either horizon or a luminal three - velocity there : no situations where the fluid reaches , or emanates from , either horizon with intermediate three - velocity occur .
this is due to the law of particle conservation and it is not related to the nature of the fluid .
this conclusion remains valid for real fluids too as they approach any horizon from within a region where @xmath14 is timelike . | we present new accretion solutions of a polytropic perfect fluid onto an f(r)-gravity de sitter - like black hole .
we consider two f(r)-gravity models and obtain finite - period cyclic flows oscillating between the event and cosmological horizons as well as semi - cyclic critical flows executing a two - way motion from and back to the same horizon . besides the generalizations and new solutions presented in this work , a corrigendum to eur .
phys .
j. c ( 2016 ) 76:280 is provided . |
is the most luminous star in our part of the galaxy and indeed is one of the most mysterious known to us ( davidson & humphreys 1997 ) .
it is surrounded by complex ejecta created in several outbursts of the star and an outgoing massive wind with periodic events every 5.54 years ( damineli 1996 ) .
the largest mass ejection occurred during the so - called great eruption observed in the 1840s which the star survived . in that event , released about 10@xmath2 ergs of kinetic energy contained in more than 10 m@xmath3 of material ( smith et al .
2003 ) , which has since expanded to form the bipolar homunculus nebula we see today ( gaviola 1950 ; smith & gehrz 1998 ; morse et al .
2001 ) . despite the hot and very luminous central star ,
the circumstellar nebula is cool enough to produce a remarkably rich emission spectrum of lines from neutral and low ionization species like fe@xmath4 and ni@xmath4 ( davidson et al . 2001 ;
zethson 2001 ; zethson et al.2001 ) , line - of - sight absorption from several neutral and singly ionized atoms ( gull et al .
2005 ) , and even emission from h@xmath5 and very cool dust ( smith 2002a ; smith et al .
in addition to being dense and cool , the ejecta of are nitrogen rich , representing ashes of the cno cycle ( davidson et al .
1982 , 1986 ; smith & morse 2004 ; verner et al .
it is the high density , low - excitation , and cno abundances of the ejecta that make the homunculus a unique laboratory among circumstellar nebulae , having a profound effect on the results discussed in this paper . in february 1999
a series of stis observations of was made of a region about 1.5 " northwest of the star ( zethson et al.2001 ) .
the spectrum from this region was significantly different from the nebular emission spectra elsewhere in the homunculus , with two of the lines being identified as forbidden emission of singly - ionized strontium ( [ 2 ] ) .
the measured velocity of the filament ( @xmath6 km / s ) and its distance from the central star are consistent with the gas having been ejected from the same star during the great eruption .
further spectroscopic observations and a complete list of lines identified in the filament were reported by hartman et al .
they show that lines of fe i , v ii , and 2 are strong in the filament , whereas only a few 2 lines are detected .
other lines in the spectrum at about the same velocity come from c i , mg i , al ii , ca i , ca ii , sc ii , cr ii , mn ii , co ii , and 2 .
curiously no h , he , s , o , or n lines are present in this region .
the cosmic abundance of strontium is lower than that of iron and nickel by factors of 30,000 and 3,000 respectively .
thus , strontium emission is normally undetectable in nebular spectra .
the fact that [ 2 ] emission was detected with strengths comparable to those of [ 2 ] and [ 2 ] lines indicates that either strontium is locally overabundant in by orders of magnitude , which is inconsistent with the evidence of cno - cycle - processes ejecta , or very unusual conditions exist in this region of the nebula . in a recent study of the 2 and [ 2 ] spectra of the filament ( bautista et al.2002 ) we found that [ 2 ] emission may be strongly enhanced by continuum fluorescence , which partially explains the strength of these lines .
although it was not possible at that time to establish the precise abundance of sr , this result did imply that sr might not be as overabundant as previously thought .
this point opened up new questions .
if [ 2 ] lines can be efficiently excited by fluorescence , why are the lines seen only in this structure , but not elsewhere in the homunculus , and not in any other emission line nebulae ?
the observed spatial and velocity distribution of the sr - filament suggests that it is located in the equatorial region between the hourglass - shaped bipolar homunculus ( davidson et al 2001 ; smith 2002a ; ishibashi et al , 2003 ) .
the sr - filament is located at a projected distance of 10 lightdays ( at 2300 pc distance ) , where it should receive intense uv radiation from .
the spectra of the weigelt blobs b and d , located a few light days from , require the spectral flux distribution of an o star with te=35,000k during the 5-year broad maximum ( verner et al.2005 ) .
the sr - filament should receive a diluted amount of this same uv radiation , but its emission line spectrum indicates a much lower ionization condition .
the sr - filament appears to be shielded not only by neutral hydrogen ( 13.6 ev ) , but also singly - ionized iron ( 7.9 ev ) which requires a substantial column density of 2 between and the sr - filament .
one possibility is that the sr - filament is located at the outer side of an apparent optically thick , dusty , torus seen in the mid - ir ( smith et al .
2003 ) . in order to demonstrate this scenario
, we must determine the nature of a hot star continuum filtered out by an 2 blocking column and predict the physical conditions within the filament that would sustain such a low ionization region .
then we can determine the actual abundances of all species observed in this structure .
in this paper we study the emission spectra of two prominent ions in the filament : 2 and 2 .
2 is studied for the very first time as the relevant atomic data has never before been obtained .
2 emission lines are rarely seen in emission nebulae as its ionization potential , 13.58 ev , leads to ti iii in most h ii regions .
2 , often seen in nebular emission spectroscopy , is much better understood and herein serves as a reference ion .
the remainder of the paper is organized as follows : in section 2 we present the atomic models built for the 2 and 2 systems , with particular detail in the calculations of atomic parameters that had to be performed for 2 . in section 3 we show the results of the nebular diagnostics . in section 4
we derive the relative abundances of 2 and 2 in gas phase , while in section 5 we discuss possible scenarios of ni and ti depletion in the nebula .
finally , our discussion and conclusions are given in section 6 .
for the present analysis of [ 2 ] spectra we employ a 76-level model for the ion that includes continuum fluorescence excitation as in bautista et al.(1996 ) and improved collisional data reported by bautista ( 2004 ) .
this new collisional data , that results from very extensive calculations , is expected to be more accurate than previous data . for 2 , neither complete
transitions rates for forbidden transitions nor collision strengths had been reported , thus large calculations of atomic data for this ion were carried out within the present work .
l spectroscopic configurations + @xmath7 , @xmath8 , @xmath9 , @xmath10 correlation configurations @xmath11 , @xmath12 , @xmath13 , @xmath14 , @xmath15 , @xmath16 , @xmath17 , @xmath18 , @xmath19 , @xmath20 , @xmath21 , @xmath22 , @xmath23 , @xmath24@xmath25 , @xmath26 , @xmath27 , @xmath28 , @xmath29 @xmath30 , @xmath31 , @xmath32 , @[email protected](1s ) , 1.12290(2s ) , 1.06650(2p ) , 1.09690(3s ) , 1.08430(3p),1.12590(3d ) , 1.27470(4s ) , 1.22660(4p ) , -0.79940(@xmath35 )
-0.64020(@xmath36 ) , 2.99520(@xmath37 ) , 1.90030(@xmath38 ) , 1.88930(@xmath39 ) .energy terms in rydberg for the ti ii ion relative to the ground state .
the table compares experimental energies with _ ab initio _ calculated energies ( theo . ) and energies with term energy corrections ( theo .- tec ) [ cols=">,<,<,>,>,>",options="header " , ] we have computed electron impact collision strengths for transitions among the lowest 82 energy levels of 2 that belong to the configurations @xmath40 , @xmath41 , @xmath42 , and @xmath43 .
we employ the breit - pauli r - matrix method and the bprm set of codes developed in the framework of the iron project ( hummer et al.1993 ; berrington et al .
partial wave contributions are included from 60 @xmath44 total symmetries with angular momentum @xmath45 , total multiplicities @xmath46 , and parities even and odd .
then , the collision strengths were `` topped up '' with estimates of contributions from higher partial waves based on the coulomb - bethe approximation ( burgess 1974 ) .
collision strengths for the fine - structure levels were obtained by algebraic recoupling of the ls reactance matrices ( hummer et al.1993 ) .
the collision strengths were calculated at @xmath47 energy points from 0 to 6 ry , with 90% of the points in the region with resonances .
this number of energy points was found to provide sufficiently good resolution for accurate calculations of the contributions of autoionizing resonances to the effective collision strengths at temperatures between 5000 k and 20000 k. figure 1 shows the collision strengths for some of the dominant transitions .
we compute maxwellian - averaged effective collision strengths , defined as : @xmath48 where @xmath49 is the collision strength for the transition @xmath50 to @xmath51 , and @xmath52 is the energy of the outgoing electron .
the @xmath53 s were computed for various temperatures between 5000 and 20 000k to be used in the calculation of excitation rates .
we built an 82-level excitation equilibrium model for 2 which considers both electron impact excitation and continuum fluorescence excitation .
a model of this size is large enough to account for collisionally excited emission at the temperatures of interest ( @xmath54 k ) , and the dominant photoexcitation channels from the ground level .
if more levels were included in our models they would tend to increase the efficiency of continuum fluorescence excitation , although it has been shown that the contribution of individual levels decreases rapidly with the level energy ( bautista et al .
we assume that the radiation field density , @xmath55 , at photon energies below the lyman ionization limit ( 13.6 ev ) can be approximated by a blackbody with temperature @xmath56 times a geometrical dilution factor @xmath57 , i.e. @xmath58 for the present calculations we adopt a blackbody temperature of 35 000k ( see verner et al.2005 ) while the dilution factor is diagnosed from the observed spectra .
a different choice in the blackbody temperature would lead to a change in the diagnosed dilution factor , but would have negligible effect on the modeled spectrum .
2 presents a diagram of the dominant photo - excitation channels for two [ 2 ] lines arising from quartets and doublets terms .
the main photoexcitation channels are indicated by dashed lines and the radiative deexcitation transitions are marked by solid lines .
such dipole allowed radiative decays lead to emission lines , observable in the spectra , just as they populate the levels that give rise to forbidden lines .
it is worth noticing that in the absence of collisional excitation , the strength of forbidden lines would be tightly bound to the intensity of the dipole transitions from the pumping mechanism .
on the other hand , the contribution of collisional excitations to the forbidden lines introduces a complex temperature and density dependence in the relative strengths of dipole and forbidden transitions , which can be used for spectral diagnostics as illustrated in the next section . fig .
3 presents the line emissivity per ion of the 2 lines [ @xmath59 ( @xmath60 ) and [ @xmath61 ( @xmath62 ) and the blend [ @xmath63 ( @xmath64 + @xmath65 ) vs. @xmath66 for an electron temperature of 6000 k , a blackbody temperature of 35 000 k and four values of the dilution factor , @xmath67 0 , @xmath68 , @xmath69 , and @xmath70 . for electron densities
less than @xmath713 continuum fluorescence dominates the strength of most lines . on the other hand , for electron densities around @xmath71 , as previously diagnosed for the strontium filament ( bautista et al .
2002 ) , collisional excitation dominates the [ @xmath59 and [ @xmath63 features , while fluorescence still makes a very significant contribution to the excitation of the @xmath72 level , responsible for the [ @xmath61 line .
this is because collisional transition rates from the ground levels to the @xmath73 multiplet are rather small .
we use the [ 2 ] and [ 2 ] emission line strengths to diagnose the physical conditions of the filament on each of the three hst / stis observations : march 2000 , april 2001 , and november 2001 ( hereafter mar00 , apr01 , nov01 respectively ; bautista et al.2002 ; hartman et al.2004 ) .
previous diagnostics were reported in bautista et al .
( 2002 ) from the [ sr ii ] lines .
there we estimated an electron density of the order of @xmath0 3 and an upper limit to the dilution factor for photoexcitation of @xmath74 . for the present work
, we use the deep spectra extending from 1640 to 10100 analyzed by hartman et al .
the original data was recorded during three visits with the hubble space telescope / space telescope imaging spectrograph ( hst / stis ) under several different observational programs . for the mar00 observation the long 52``x0.2 '' aperture
was aligned close to the axis of the homunculus ( position angle -41deg ) , whereas the aperture was aligned at 22 deg in apr01 and -130 deg in nov01 .
the different angles were defined primarily by hst sun - angle restrictions , but were used to study the extent of the strontium filament in different directions .
the intersection of these apertures ( see fig . 1 in hartman et al.2004 ) is on the apparent peak of the strontium filament as first noticed by the discovery observations ( zethson et al.2001 ) .
the line fluxes were integrated across the line profiles of the components centered at @xmath75 - 100 km / s at specific positions along the slit .
the intersection position of the different visits is the primary position of interest .
however as noted below , several additional positions were sampled along the aperture oriented radially to ( p.a.=-41@xmath76 ) and tangentially ( -130@xmath76 ) to understand the spatial variation of specific line ratios and therein , to study the excitation and ionization structure of the filament .
the measured line intensities vary along the aperture and between visits .
the structure is complex as other nebular structures are in line of sight : namely the homunculus and the little homunculus . at some positions , hartman et al.(2004 )
noted a second velocity structure in many lines .
these two velocity systems are likely related to the two disks noted by davidson et al.(2001 ) located between the bipolar lobes .
the two velocity components correspond to material in the same equatorial disk but with two different ages , as noted by davidson et al .
the more prominent feature is associated with the parts of the skirt ejected along with the homunculus in the 1840 s .
the fainter and more blueshifted feature is also in the equator , but was ejected in the 1890 outburst along with the little homunculus ( ishibashi et al .
2003 ; smith 2005 ) and the weigelt knots ( smith et al . 2004 ; weigelt et al .
1995 ) . in this study
we focus on the structure ejected in the 1840s . as with any spectroscopic analysis
the observed line intensities must be corrected for extinction .
none of the lines that form the classical line ratios ( e.g. hydrogen recombination lines ) are detected .
we use two relatively well known [ 2 ] transitions that arise from a common upper level , i.e. @xmath77@xmath78 ) and @xmath79@xmath80 ) , and the standard interstellar extinction curve of cardelli et al.(1989 ) . while these lines are relatively close in wavelength , thus far they are the best candidates for estimating extinction as all other lines of interest are much weaker or their predicted line ratios are less certain .
the observed line ratios are @xmath81 in mar00 , @xmath82 in apr01 , and @xmath83 in nov01 .
thus , there are no significant variations of the line ratio between the three observations . from these observations
we derive an extinction magnitude , @xmath84 , as required to correct the ratios to the theoretical value of 1.6 , with an estimated uncertainty of @xmath7510% .
the relatively large uncertainty in @xmath85 comes from the fact that the two lines used in the diagnostic are close in wavelength .
this value for the extinction magnitude is consistent with the value measured , @xmath86 , from photometry of stars of the tr 16 cluster in the carinae nebula ( tapia et al.2003 ) .
the actual extinction magnitude to the filament could be somewhat higher than to the carinae stars due to the dust within the nebula .
smith ( 2002b ) studied local extinction within the larger carina nebula , and found a total - to - selective extinction of r=@xmath87 , which is different than the normal ism value of 3.1 .
furthermore , the extinction around itself is likely to be very gray because of the large grain size ( smith et al .
however , thermal - ir maps of the dust column density in the homunculus reveal a local minimum near the position of the sr filament ( smith et al .
2003 ) , so the unusual reddening there may not be a severe problem .
we thus adopt for the remaining of this paper a extinction law with @xmath88 and the interstellar @xmath89 .
clearly , the remaining uncertainty on this value can be important in trying to compare lines well separated in wavelength , thus all diagnostics should be carried out with lines close to each other .
tables 5 and 6 present the complete lists of [ 2 ] and [ 2 ] lines used for the present study and their intensities corrected for extinction .
lcrrrrrr @xmath90 & transition & & + & & & + & & mar00 & apr01 & nov01 & mar00 & apr01 & nov01 7379.970&@xmath91&286.00 & 211.00 & 239.00 & 1088.0 & 802.7 & 909.2 8303.270&@xmath92 & & & 11.80 & & & 34.6 & & & & & & & 6668.670&@xmath93 & 64.60 & 41.60 & 51.30 & 304.3 & 195.9 & 241.6 7413.690&@xmath94&142.0 & 80.60 & 106.0 & 534.8 & 303.5 & 399.2 & & & & & & & 4327.453&@xmath95 & & 18.10 & 27.00 & & 246.8 & 368.1 7257.860&@xmath96 & 23.00 & 12.80 & 10.70 & 90.8 & 50.5 & 42.2 & & & & & & & 6815.450&@xmath97 & 15.00 & 5.71 & 7.32 & 67.6 & 25.7 & 33.0 & & & & & & & 6793.350&@xmath98 & 10.40 & 3.50 & 19.70 & 47.2 & 15.9 & 89.4 & & & & lcrrrrrr @xmath90 & transition & & + & & & + & & mar00 & apr01 & nov01 & mar00 & apr01 & nov01 4918.247&@xmath99&13.60 & 8.62 & 11.50 & 124.8 & 61.5 & 105.6 4984.205&@xmath100&2.64 & 1.32 & 2.00 & 23.3 & 11.7 & 17.7 6153.623&@xmath101 & & & 3.76 & & & 24.0 & & & & & & & 4927.288&@xmath102&14.70 & 13.2b&13.40 & 134.2 & 120.5&122.46079.567&@xmath103 & & & 2.86 & & & 16.1 & & & & & & & 6126.311&@xmath104 & & & 11.8 & & & 65.5 6228.990&@xmath105 & & & 2.03 & & & 10.9 & & & & & & & 6148.896&@xmath106 & & & 3.20 & & & 17.6 6252.341&@xmath107 & & & 18.8 & & & 100.3 & & & & & & & 6652.610&@xmath108&3.36 & 1.97 & 2.32 & 15.9 & 9.3 & 11.0 6727.670&@xmath109&5.12 & 4.78 & 5.38 & 23.7 & 22.1 & 24.9 9652.716&@xmath110&0.71 & & 1.39 & 1.7 & & 3.3 9408.356&@xmath111&2.62 & & 1.58 & 6.3 & & 3.8 & & & & & & & 6648.980&@xmath112&1.87 & 1.67 & 1.42 & 8.9 & 7.9 & 6.7 9645.089&@xmath113&1.02 & & 0.60 & 2.4 & & 1.4 & & & & & & & 7918.452&@xmath114 & & & 2.62 & & & 8.5 7977.897&@xmath115 & & & 6.03b & & & 19.3 8531.906&@xmath116 & & & 11.00 & & & 30.5 8587.494&@xmath117 & & & 5.78 & & & 15.8 8664.452&@xmath118 & & & 53.8b & & & & & & & & & & 7979.330&@xmath119 & & & 6.03b & & & 8568.400&@xmath120 & & & 6.44 & & & 17.78663.807&@xmath121 & & & 53.80b & & & 150.2 & & & & & & 8335.590&@xmath122 & & & 3.74 & & & 10.98526.997&@xmath123 & & & 13.90 & & & 38.6 & & & & & & & 8412.354&@xmath124 & & & 7.72 & & & 22.0 10069.74b&@xmath125&1.70 & & 3.57 & 3.2 & & 8.3 & @xmath126 & & & & & & & & & & & & & 10128.770&@xmath127&5.82 & & 7.84 & 13.6 & & 18.3 & & & & & & & the combined contributions of collisional excitation and fluorescence make spectral diagnostics much more difficult than with traditional collisionally excited lines
. most line ratios of interest become simultaneously sensitive to electron density , temperature , and the continuum radiation field ( i.e. dilution factor ) , thus many lines have to be studied as a whole to derive reliable conclusions about the physical conditions of the emitting plasma . in the case of [ 2 ] lines , only a few diagnostic line ratios can be used because out of the seven dipole forbidden lines measured in the spectrum several share the same upper level .
4 shows the line ratio @xmath128)/i([\lambda 7379])$ ] , @xmath129)/i([\lambda 7379])$ ] , and @xmath129)/i([\lambda\lambda 6668 + 7413])$ ] against the logarithm of the dilution factor .
the ratios were computed for three values of the electron density , @xmath130 @xmath131 , @xmath0 , and @xmath132 displayed in separated panels from left to right in the figure . in each panel
we present the line ratios as computed for temperatures of 5000 , 7000 , and 9000 k. the observed lines ratios are indicated in the plots by horizontal lines for each observation .
two conclusions can be derived from these plots in regards to the dilution factor in the filament .
first , for the nov01 observation the value of @xmath133 is in between -9.2 and -8.5 .
there is evidence for different dilution factors for each of the three observations , with a difference as large as 1 dex between the mar00 and the apr01 observations , with the higher value corresponding to the mar00 observation .
the dilution factor from the nov01 observation lies between the values for the other two observations .
the spread in line ratios , and consequently on derived physical conditions , in the three observations are the result of different orientations of the spectrograph slit with respect to the star ( see hartman et al .
for the mar00 observations the slit was centered on with the fiducial , f2 , blocking the starlight .
this provided a spatial sampling of the sr - emitting region extending radially from the central source . that this spectrum yields the highest value for the dilution factor is due to finding the radial distance for which both apply .
the nov01 spectrum was taken at position angle -130 degrees and provided a slice of the sr - filament that is tangential to the distance from the central source .
as the position of the slit is more distant than the optimal position for highest dilution factor , we get a lower value for the dilution factor along this slit .
the dilution factor changes by an order of magnitude between the optimal radial position and the tangential sample because these observations sample slightly different spatial regions .
photons with energies below 7.9 ev encounter increasing optical depths with distance from the central source . for these energies
the main source of opacity is free - free opacity ( bremsstrahlung ) .
assuming an electron temperature of the order of @xmath134
k , density @xmath135 3 , and a difference in the path length within the filament between the two observations of @xmath136 cm , one gets an optical depth , @xmath137 @xmath138 , which is consistent with the observed variation in the dilution factor .
the large number of 2 and [ 2 ] lines present in the spectra allows us to constrain the physical conditions of the filament as discussed earlier , the combined contributions of collisional excitation and fluorescence lead to complex temperature and density dependence in the relative strengths of dipole to forbidden emission lines .
nonetheless , such ratios are much more sensitive to the strength of the dilution factor than to other parameters , so they can be used to diagnose the value of the dilution factor . in fig .
5 we plot 2 line ratios between dipole allowed and forbidden lines against the dilution factor , together with the observed ratios .
these line ratios are very sensitive to the value of the dilution factor , but depend relatively little on @xmath139 and @xmath66 . from this figure ,
the line ratios for the nov01 observation are consistent with @xmath133 between -10.0 and -8.6 , while the ratios from the mar00 spectrum indicate a higher dilution factor , by as much as 0.5 dex .
these results are in agreement with those obtained with the [ 2 ] lines . considering all possible line ratios among forbidden lines we found one that is primarily sensitive to the dilution factor , and depends little on the temperature and density .
this ratio is shown in fig .
6 , as computed for @xmath139= 5000 , 6000 , and 7000 k and @xmath130 @xmath131 , @xmath0 , and @xmath140 3 .
once again the observed line ratios lay in the range of @xmath133 between -9.1 and -8.6 . from the agreement between various 2 and 2 diagnostics we find the value of the dilution factor to be @xmath141 . with the dilution factor constrained
, one can diagnose other parameters of the emitting region .
presents line ratios among 2 forbidden lines vs. the electron temperature .
the line ratios are shown in various panels for @xmath130 @xmath131 , @xmath0 , and @xmath140 3 . in each case
we show the line ratios for @xmath142 , -8.5 , and -9.5 .
the observed line ratios from the nov01 spectrum are also indicated .
inspection of these plot reveals that the best agreement for all the lines observed is obtained for @xmath143 k. this relatively low temperature is consistent with the observed ionization structure of fe and ni in the filament .
the observed iron emission arises mostly from neutral iron , indicating that this dominates over 2 .
because of the iron shielding of the filament it is expected that there would be few photons capable of ionizing fe i. but , even in the absence of photoionization collisional ionization drives most of the iron into 2 whenever electron temperatures exceed @xmath144 k ( see arnaud and raymond 1992 ) . below this temperature neutral iron
may be the dominant ionization stage .
on the other hand , under collisional ionization conditions the fraction of ionized to neutral nickel is more than twice the fraction for iron at the same temperature .
thus , 2 is expected to remain the dominant ionization stage for temperatures as low as @xmath145 k. finally , we can determine the electron density .
8 depicts four line ratios as function of the electron density for a fixed @xmath133=-9.0 and temperatures of @xmath146 k. the electron density is readily diagnosed at around @xmath0 3 .
given the physical conditions diagnosed above we compare the overall agreement between the computed spectrum and observations .
this comparison for the nov01 spectrum is shown in table 7 .
the agreement is reasonably good considering the remaining uncertainties in the atomic data and the shape of the pumping radiation continuum .
the 2 @xmath147)/i([\lambda 4918])$ ] ratio is large across most of the spectral slit , which allows us to further investigate the spatial variations of the dilution factor near the sr - filament . to do this we integrated the intensity over 0.13 `` wide regions of the spectrum from mar00 and nov01 .
the measured line ratios , corrected for extinction , and the derived dilution factors from them are shown in fig .
the mar00 observations , when the f2 fiducial ( 0.85 '' long ) was centered on with the large aperture extending radially from the star at the middle of the filament ( 1.3 " from ) , clearly show a linear decay of the dilution factor with distance from the central source . consistently with this picture , the nov01 slit that was centered on the sr - filament and perpendicular to the radial direction from the central source , shows a maximum of the dilution factor at the middle of the filament and symmetric drop offs toward both sides of this position .
rrr @xmath148 & & obs . & theo .
4918.26 & 15.7 & 6.38 4927.29 & 18.2 & 18.3 4984.20 & 2.6 & 1.01 6079.57 & 2.4 & 0.91 6126.31 & 9.7 & 7.23 6148.90 & 2.6 & 1.78 6153.62 & 3.1 & 0.25 6228.99 & 1.6 & 0.91 6252.34 & 14.9 & 11.70 6648.98 & 1.0 & 1.0 6652.61 & 1.6 & 1.49 6727.67 & 3.7 & 3.72 7918.45 & 1.3 & 0.51 7977.90 & 2.9 & 0.84 8335.59 & 1.6 & 0.80 8412.35 & 3.3 & 2.46 8527.00 & 5.7 & 4.50 8531.91 & 4.5 & 4.39 8568.40 & 2.6 & 8.98 8587.49 & 2.4 & 8.04 9645.09 & 0.2 & 0.98 9652.72 & 0.5 & 0.74 10069 b & 2.2 & 2.12 10128.77 & 2.7 & 3.26
once the physical conditions of the sr - filament have been determined we proceed to compute the relative abundances of 2 and 2 .
we use the nov01 spectrum because : ( 1 ) much more spectral interval was covered during that visit and the greatly increased number of lines , which permits a better statistical determination , and ( 2 ) the long aperture likely sampled similar regions for the 2 and 2 due to the tangential slit orientation .
nonetheless , we have confirmed that the computed relative abundances from all other observations also agree , within the uncertainties .
10 depicts the @xmath149 as derived from the [ 2 ] @xmath150 and [ 2 ] @xmath151 lines .
the calculated abundance ratio is plotted against the assumed value for the dilution factor and for temperatures of @xmath146 k as diagnosed from the spectrum .
given the fact that the lines observed from both species have similar excitation energies and arise from levels with the same multiplicity and parity , they respond similarly to changes in the dilution factor .
so we see that the uncertainty in the abundance ratio is negligible for the mean temperature of 6000 k , and uncertainty in the derived abundance does not exceed 20% within the whole range of dilution parameters and temperatures .
thus , we conclude that @xmath152 , with a conservative uncertainty that may not exceed a factor of three , including the dispersion from all the lines which contains non - systematic uncertainties in the atomic data and the treatment of the pumping continuum field . to estimate the total @xmath153 abundance ratio we notice that only neutral and singly ionized ions are present in the spectra .
it is expected that the ionic ratios ti i / ti
ii and ni i / ni ii are similar since ti i and ni i have approximately the same ionization potential . moreover , from observations one can conclude that ni and ti are both singly ionized , as there are no lines from the neutral stages of these ions in the spectra . thus , @xmath154 in gas phase .
this value is significantly higher than the best solar abundance ratio of @xmath750.05 .
in other words , the gas phase @xmath153 abundance ratio in the sr - filament is nearly two orders of magnitude greater than solar value .
further , we have checked on the @xmath155 abundance ratio in the filament .
this calculation is complicated by the fact that the [ 2 ] emission comes from a more extended region in velocity space .
thus , we only consider the integrated flux of [ 2 ] lines within the range in velocity space that coincides the [ 2 ] emission we find a value for the @xmath155 ratio roughly consistent with the solar @xmath156 ratio , assuming that the ionic fractions of 2 and 2 are tied to each other by charge exchange and are thus within a factor of two of each other
. it must be noted , however , that the filament exhibits prominent emission from fe i , suggesting a large abundance of this ion , but no emission from ni i or ti i within the sensitivity of the observations .
= 9 cm = 8 cm the observed large overabundance of titanium with respect to both nickel and iron in the gas phase is quite striking .
ti / fe abundance ratios higher than solar by up to 0.4 dex are known to occur in metal - poor and extremely metal - poor stars ( ryan et al . 1996 ) , but to the best of our knowledge there are no reports of such a high overabundance as we have analyzed here .
it is likely that the abnormal relative abundances found here are only representative of the gas phase , rather than the total absolute abundances . under the conditions of the sr - filament nearly 90% of fe and ni
are condensed into grains .
thus , a large ti / ni gas phase abundance ratio would be attained if some kind of destruction mechanism of ti - bearing condensates acted in the filament .
although we do not have a clear explanation for such a mechanism we devote this section of the paper to study the chemistry of the ti , ni , and fe under the conditions of the sr - filament .
condensation temperatures of ti , fe , and ni are somewhat different and ti is generally more refractory than fe and ni during condensation from a solar composition gas ( e.g. , lodders 2003 ) .
as described below , ti also forms a more refractory condensate than fe and ni from the overall composition relevant for the ejecta here .
condensation temperatures depend on bulk composition and total pressure . in a solar composition gas with c / o = 0.5
, titanium condenses as calcium titanate ( e.g. , perovskite , catio@xmath157 ) , followed by an iron and nickel metal alloy at somewhat lower temperatures ( lodders 2003 ) .
trace elements such as sr condense into solid solution with major element host phases , and condensation of sr as `` srtio@xmath157 '' into catio@xmath157 proceeds at temperatures between that of catio@xmath157 and metal alloy formation .
as long as the overall composition of s ejecta remained near - solar , solar - type condensates should also have formed . however , in the sr - filament and in the rest of the emission ejecta within the homunculus , the little homunculus and the weigelt blobs , very little oxygen has ever been detected . moreover , in the weigelt blobs where [ o i ] @xmath158 emission is seen , oxygen seems to be underabundant by at least two orders of magnitude ( e.g. thackeray 1967 , zethson 2001 , verner et al .
2005 ) . in the outer ejecta carbon and oxygen
are severely depleted relative to solar abundances ( davidson et al .
1982 ; dufour et al 1997 ; smith and morse 2004 ) .
hydrogen is also depleted whereas n and he are enriched . in order to explore the condensation chemistry , we adopt a solar composition gas in which solar c and o abundances are both reduced by a factor of 100 . the n abundance is increased by the amount removed from the solar c and o abundances so that the sum of solar cno nuclei is preserved .
the he abundance is taken as solar plus a quarter of the solar h abundance , and the h abundance is set to zero .
note that the chemistry of ti , fe , ni , and si is relatively insensitive to the total h abundance because there are no major h - bearing gases for those elements .
even if h were fairly abundant , the following condensation results would change little .
descriptions of the thermochemical computation methods are given in lodders ( 2003 ) and lodders & fegley ( 1997 ) .
the results of this calculation serve as a good illustration of the dust chemistry , and should remain qualitatively correct despite variations of the precise relative abundances .
the condensates ( fig .
11 ) for this `` cno '' processed composition are first tic , and at about 100 k lower temperature , an iron - nickel silicide alloy , ( fe , ni)si . below total pressures of @xmath159 bar , sic condensation temperatures are within a few degrees of those of the silicide alloy .
the condensates from the `` cno '' processed composition are the same as appear when c and o abundances of an otherwise solar composition gas are only slightly ( i.e. orders of factor 2 ) modified so that c / o @xmath160 1 ( lodders & fegley 1995 , 1999 ) . here
the c / o ratio remains solar but the large decrease in absolute c and o abundances is responsible for the appearance of the carbides and silicide because there is no oxygen for sio and silicate or oxide formation .
the elemental composition here has enough c for condensation of all ti as tic ( the total ti / c is @xmath75 0.03 ) but only a little sic can form because si is more abundant than c ( si / c @xmath75 14 ) and si condenses also into ( fe , ni)si .
the overall c abundance is so low that graphite does not condense at the relatively high temperatures where tic and ( fe , ni)si form .
if carbon were completely absent here , titanium nitride ( tin , osbornite ) would condense instead of tic at about the same temperature , sic would be completely absent , and condensation of ( fe , ni)si would remain unchanged .
condensation temperatures describe when a compound starts to form ( i.e. , where the saturation ratio reaches unity ) but lower temperatures are needed to fully remove an element from the gas into a condensate , as shown for a constant total pressure in fig
complete titanium removal from the gas requires a drop of about 100 k below the condensation temperature of tic and all ti is condensed at temperatures when ( fe , ni)si condensation begins to proceed . given the nature of the ti- , ni- , and fe - bearings described here we suggest that selective photoevaporation of ti - bearings is a possible explanation for the gas phase ti / ni abundance found in the filament .
two different scenarios may be suggested : ( 1 ) there was some spatial separation between the tic and(ni , fe)si condensates , related to their different condensation temperatures , that left the tic rich gas more exposed to photoevaporation ; ( 2 ) the tic grains are generally smaller than the ( ni , fe)si grains , and so the tic condensates were much less resistant to photoevaporation .
these scenarios sketched out here seem promising to explain the observations and need to be studied in more detail .
further , it is important to determine the abundance ratios of other elements of different volatility .
we have studied the 2 and 2 spectra of the sr - filament in , as taken in three different epochs with the hst . in doing this
we build spectral models for the two ions using the best published data for 2 , while we carry out extensive calculations of radiative and electron impact rates for 2 .
the spectral models were employed for diagnosing the physical conditions of the emitting region .
the results are generally consistent with previous determinations using [ 2 ] and 2 lines .
moreover , the [ 2 ] lines allow us to constrain the conditions for observations with the aperture across the s central source to @xmath161 3 , @xmath162 k , and dilution factor @xmath163 for an assumed blackbody continuum radiation with temperature of 35 000 k. for observations with the slit perpendicular to the radial direction the dilution factor seems to be somewhat lower .
study of the observed [ 2 ] emission reveals a very large overabundance of ti in gas phase with respect to ni and fe relative to the cosmic values .
we study the chemistry of ti , ni , and fe in cno - cycled gas and find that ti condenses as tic , or alternatively as tin , while ni and fe are locked into ( ni , fe)si . moreover , the tic condenses at higher temperatures than ni- and fe - bearings .
later , we suggest that the tic would be selectively destroyed by the stellar uv radiation leading to a relative overabundance of ti in gas phase .
this explanation for the unusual ti / ni ratios in the sr - filament requires that significant dust and molecules had formed in this material shortly after the great eruption , but have since been largely photoevporated here . in this context
, it is very interesting that near - ir emission from molecular hydrogen is seen throughout the polar lobes of the homunculus nebula ( smith 2002a ) , but this same h@xmath5 emission is not seen at comparable levels from the equatorial skirt in the vicinity of the sr - filament ( smith 2002a , 2004 ) .
today , the gas densities we infer in the sr filament ( @xmath7510@xmath164 @xmath1 ) are very high compared to many other circumstellar nebulae around massive stars , and are similar to the extremely high densities seen in the walls of the homunculus ( e.g. , gull et al . 2005
; nielsen et al . 2005 ) .
a weaker radiation field at larger radii is not the only explanation for this difference : h@xmath5 emission is seen both in the polar caps of the homunculus at larger radii from the star than the sr - filament , and in the side walls of the polar lobes near the equator , even _ closer _ to the star than the sr filament ( smith 2004 ) . in the past when the sr - filament ejecta were even denser , it is conceivable that h@xmath5 or other molecules may have existed , but have since been destroyed . if true , this would seem to support the scenario of unusual chemistry resulting from selective photoevaporation and cno processed ejecta , as we have suggested here .
the observations of the strontium filament were made with the space telescope imaging spectrograph on the nasa / esa hubble space telescope and were obtained by the space telescope science institute , which is operated by the association of universities for research in astronomy , incorporated , under nasa contract nas5 - 26555 .
support for this research was provided in part by nasa through the stis gto grant and grants from the space telescope science institute .
h. hartman is supported by a grant from the swedish national space board .
n. smith was supported by nasa through grant hf-01166.01a from stsci .
work by kl supported in part by grant nng04g157a from the nasa astrobiology institute .
data for this analysis was gathered under go programs headed by kris davidson and go and gto programs headed by theodore gull . | we study the nature of the [ 2 ] and [ 2 ] emission from the so - called _ strontium filament _ found in the ejecta of . to this purpose
we employ multilevel models of the 2 and 2 systems which are used to investigate the physical condition of the filament and the excitation mechanisms of the observed lines .
for the 2 ion , for which no atomic data was previously available , we carry out _ ab initio _ calculations of radiative transition rates and electron impact excitation rate coefficients .
it is found that the observed spectrum is consistent with the lines being excited in a mostly neutral region with an electron density of the order of @xmath0 @xmath1 and a temperature around 6000 k. in analyzing three observations with different slit orientations recorded between march 2000 and november 2001 we find line ratios that change among various observations , in a way consistent with changes of up to an order of magnitude in the strength of the continuum radiation field .
these changes result from different samplings of the extended filament , due to the different slit orientations used for each observation , and yield clues on the spatial extent and optical depth of the filament . the observed emission indicates a large ti / ni abundance ratio relative to solar abundances .
it is suggested that the observed high ti / ni ratio in gas is caused by dust - gas fractionation processes and does not reflect the absolute ti / ni ratio in the ejecta of .
we study the condensation chemistry of ti , ni and fe within the filament and suggest that the observed gas phase overabundance of ti is likely the result of selective photo - evaporation of ti - bearing grains .
some mechanisms for such a scenario are proposed .
= = = = = = = = # 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 = = = = = = = = # 1 2ti ii 2ni ii 2fe ii 2sr ii 3@xmath1 [ firstpage ] atomic data atomic processes line : formation stars : abundances stars : eta carinae |
since the pioneering work of simkin ( 1974 ) , major advances in the measurement of stellar orbits in early - type galaxies have been made . the first quantities to be measured were the line - of - sight rotation velocity ( v@xmath3 ; which can be considered a measure of ordered motion within a galaxy ) and velocity dispersion ( @xmath4 ; a measure of the random motions ) . by combining these two measures with photometric ellipticity ( @xmath5 ) it was possible to probe the anisotropy of stellar orbits in galaxies ( binney 1978 ) .
further progress was made in this field when , in the 1980s , bender ( 1988 ) showed that although the isophotal shapes of early - type galaxies were well described by ellipses , small variations from ellipticity could be used to categorise them as either ` boxy ' or ` discy ' , and that these variations relate directly to the stellar orbital isotropy .
the combination of surface photometry and detailed kinematics were therefore demonstrated to provide significant insights into the global structure of galaxies .
more recently , higher order moments of the line - of - sight velocity distribution ( losvd ) have been measured ( e.g. bender 1990 ; rix & white 1992 ; gerhard 1993 ; van der marel & franx 1993 ) .
these characterise deviations from a pure gaussian shape , e.g. skewed and symmetric deviations , denoted h@xmath0 and h@xmath1 respectively .
such parameters can indicate kinematic substructures , such as embedded stellar discs .
the development of integral field units ( ifus ) , such as the sauron instrument , have given new impetus to this field , as these provide 2-d maps of the losvd .
this allows the definition and estimation of a number of important new parameters that can be averaged over the surface of a galaxy in a self - consistent way .
perhaps the most important of these is the ` spin ' parameter ( @xmath6 ) which is a proxy for the projected angular momentum per unit mass .
applied to the sauron galaxy sample of 48 early - type galaxies , it was used to show that a dichotomy exists between ` slow- ' and ` fast- ' rotators ( emsellem et al . 2007 ) .
trends between photometric and kinematic axis ratios and position angles have also been shown to vary with @xmath6 ; with slow - rotators often exhibiting strong misalignments .
these observations provide new clues to the formation mechanisms of these differing galaxy types ( krajnovi et al .
2008 ; cappellari et al . 2007 ; thomas et al . 2007 ) .
statler & smecker - hane ( 1999 ) used a different approach to building 2-d maps of the losvd in ngc 3379 . using long - slit spectroscopy along four position angles
, they were able to reconstruct a model of the 2-d kinematics by making suitable interpolations between their observed axes . however , such studies are rare due to the large amount of telescope time required to observe multiple position angles out to large radii . despite these advances ,
most observations typically reach to less than one effective radius , with only a handful of studies extending this coverage up to 23 r@xmath7 ( but see saglia et al .
1993 ; graham et al . 1998 ; mehlert et al .
2000 ; proctor et al .
2005 ; sanchez - blazquez et al . 2006 ; forestell & gebhardt 2008 ) .
however , such outer regions potentially provide key information regarding the physical properties of galaxies .
for example , radii within 1 r@xmath7 only sample 50% of the baryonic mass and @xmath2 10% of the angular momentum of an idealised galaxy with an r@xmath8 like light profile and a flat rotation profile .
the situation is even worse when considering the distribution of total galaxy mass ( including dark matter ) .
for instance , the simulations of dekel et al .
( 2005 ) show that only @xmath25% of the total mass is contained within 1 r@xmath7 ( see also mamon & lokas 2005 ) .
furthermore , the signatures of some formation mechanisms only become apparent at large galacto
( e.g. mcmillan , athanassoula & dehnen 2007 ; weil & hernquist 1996 ) .
a means to probe the kinematics in the outer regions of early - type galaxies , and hence include a significant fraction of a galaxies angular momentum and mass , is therefore highly desirable .
in this paper we measure stellar losvds from deep multi - object spectroscopic observations and develop a new technique for their analysis .
specifically , we measure rotation velocity , velocity dispersion and the gauss - hermite coefficients h@xmath0 and h@xmath1 , out to @xmath23 r@xmath7 in five nearby early - type galaxies . by extending the region studied to such radii we encompass more than 85% of the baryonic mass , 30% of the angular momentum and 15% of the total mass , a significant increase in the fractions so surveyed in galaxies . to achieve this we exploit the large aperture of the keck telescope , the high sensitivity of the deimos instrument at 8500 @xmath9 and the strong calcium triplet absorption features at these wavelengths .
even with this configuration , exposures of @xmath2 2 hours are required for our analysis .
+ the paper is organised as follows : in section [ odr ] we outline the observations and data reductions , detailing the technique by which we extract the host galaxy spectra . in section
[ anal ] we outline our new kinematic analysis technique .
results are presented in section [ results ] .
we discuss our findings in section [ discuss ] , and present our conclusions in section [ concs ] .
the galaxies in our sample were selected with the intention of obtaining spectra of their globular clusters ( gcs ) .
the galaxies are all nearby , early - types with moderately rich gc systems .
the selection also had the requirement that recession velocities were such that the calcium triplet lines would not generally be adversely affected by strong sky lines .
the main science goal was to obtain gc kinematics , and hence probe the mass profile of their host galaxy halos .
initial results for ngc 1407 from the 2006 nov . run
have recently been reported in romanowsky et al .
( 2008 ) .
the data used in this work are from three observing runs using deimos on the keck-2 telescope .
the first observing run was carried out in 2006 nov . using the 1200 l / mm grating centered at @xmath27500 @xmath9 and a 1 arcsec slit .
this allows coverage of the calcium triplet at 8498 , 8542 and 8662 @xmath9 at a resolution of @xmath21.5 @xmath9 .
the second run was carried out in 2007 nov . and the third in 2008 april , both using the same instrument setup as the first run .
seeing was @xmath20.5 to 0.8 arcsec on all nights .
a total of five galaxies were observed ( ngc 821 , ngc 1400 , ngc 1407 , ngc 2768 and ngc 4494 ) . between two and six independent masks
were observed per galaxy and the 16 x 5 arcmin field - of - view allowed the observation of @xmath280 - 120 spectra in each mask .
however , only @xmath220% of these were at radii from which the signal of the background host galaxy was sufficient for the extraction of spectra ( see table [ fits ] ) .
[ thumbnails ] shows the positions of slits from which host galaxy spectra were extracted .
we note that it was only possible to measure recession velocities in some of the outermost slits , i.e. velocity dispersion and higher order moments are not presented for some of the outermost positions . | we detail an innovative new technique for measuring the 2-d velocity moments ( rotation velocity , velocity dispersion and gauss - hermite coefficients h@xmath0 and h@xmath1 ) of the stellar populations of galaxy halos using spectra from keck deimos multi - object spectroscopic observations .
the data are used to reconstruct 2-d rotation velocity maps . here
we present data for five nearby early - type galaxies to @xmath23 effective radii .
we provide significant insights into the global kinematic structure of these galaxies , and challenge the accepted morphological classification in several cases .
we show that between 13 effective radii the velocity dispersion declines very slowly , if at all , in all five galaxies .
for the two galaxies with velocity dispersion profiles available from planetary nebulae data we find very good agreement with our stellar profiles . we find a variety of rotation profiles beyond 1 effective radius , i.e rotation speed remaining constant , decreasing _ and _ increasing with radius .
these results are of particular importance to studies which attempt to classify galaxies by their kinematic structure within one effective radius , such as the recent definition of fast- and slow- rotator classes by the sauron project .
our data suggests that the rotator class may change when larger galacto - centric radii are probed .
this has important implications for dynamical modeling of early - type galaxies .
the data from this study are available on - line . -0.5
cm latexl-.36em.3ex-.15em t-.1667em.7ex-.125emx [ section ] [ firstpage ] galaxies : general , galaxies : individual ; ( ngc 821 ; ngc 1400 ; ngc 1407 ; ngc 2768 ; ngc 4494 ) , galaxies : kinematics and dynamics , galaxies ; structure , stellar dynamics , techniques ; spectroscopic , |
the turku meeting was the third in an historic series of conferences on bl lac objects , so it seems appropriate to begin from an historical perspective .
the first bl lac meeting was held in pittsburgh in 1978 and the second 10 years later , in como , italy . over this 20-year period
, there has been enormous progress in understanding bl lac objects , or blazars , using the collective term for bl lac objects and flat - radio - spectrum , highly polarized quasars .
this occurred despite the scarcity of sources in 20 years the number of catalogued bl lacs increased only by a factor of a few , roughly 1/10 the growth in catalogued quasars .
( we have at least reached the point where the number of bl lac objects is greater than the number of conference participants ! )
the field has changed a great deal in 20 years . while many of the 84 pittsburgh participants have gone on to illustrious careers in astrophysics ( this is the good news , for students who wonder whether blazars are a good career move ) , almost none are still actively working on blazars ( the bad news ? ) .
for example , only 8 people from the pittsburgh meeting were at the 1988 como meeting , and only 1 person came to both como and turku ( answer at the end ) .
this represents real turnover , as the number of blazar researchers has not increased dramatically roughly the same number of people came to como as pittsburgh . in turku
the attendance was 50% higher , in large part because of the welcome influx of gamma - ray astronomers .
altogether , the demographics of these three conferences indicates considerable evolution in our understanding and renewed interest in the physics of blazars . in his overview at pittsburgh ,
wayne stein was reliably prescient .
he suggested tongue - in - cheek that bl lac objects might be defined as `` totally uninteresting astronomical sources [ to the spectroscopist ] because they may exhibit neither absorption nor emission lines . ''
( this surely resonates with many optical observers who have trouble getting telescope time for spectroscopy ! )
he also pointed out , more seriously , the reason for the high level of interest in this topic , saying , `` bl lac objects are our most direct observable link with the ultimate energy source of the quasi - stellar objects . '' indeed , we now recognize that bl lac objects and other blazars are the best probe of jet formation and evolution in agn because , although they are rare , unification tells us they are the same as all radio - loud agn , and at the same time , their jet emission is conveniently enhanced by relativistic beaming .
thus blazars are critical to understanding jets in all radio - loud agn what was discussed at turku about bl lac objects has near - universal relevance for agn .
we can look back at what have we learned since pittsburgh , what we knew then , what have been the major problems and solutions , and what are the most promising avenues for progress in the future .
over 10-year intervals , not surprisingly , real progress does happen !
but it s also humbling to recognize how much was already known in 1978 .
in the most quoted paper of any conference proceedings ever , blandford and rees ( 1978 ) already outlined a fair fraction of the underlying physics of bl lac objects : they understood that the spectral energy distributions ( seds ) were dominated by synchrotron and probably inverse compton emission ; they realized the emission region(s ) had to be relativistically expanding ; and they even knew that the selection effects would favor the discovery of bl lac objects even though they were rare objects .
it is also interesting to observe the evolution of topics under discussion at pittsburgh , como , and turku : * _ observational properties at various wavelengths
_ there was the usual breakdown into wavelengths radio , optical , x - ray but the mix changes .
in turku there was more discussion of vlbi results , and certainly more about gamma - rays .
the discovery of strong gamma - ray emission from blazars has completely changed our understanding of their energy output . * _ multiwavelength monitoring _ the topic of multiwavelength monitoring first came up in como and was discussed much more in turku .
spectral variations are a very powerful tool for understanding the physics of bl lac objects . * _ spectroscopy : line strengths and redshifts _ in pittsburgh , there was a great deal of interest in spectral lines and distances .
these are still important , despite being less discussed today . in one of our most important bl lac samples , the 1 jy radio - selected sample , about 1/4 of the redshifts are still missing . *
_ models _ theorists continue to keep up , and often to be ahead of the observations .
we now understand a lot about the physics of blazars and can even speculate on the similar physics in galactic black hole candidates and perhaps gamma - ray burst sources . * _ surveys and population statistics _
the small number of well - defined blazar samples indicates the critical importance of surveys . in como
, the first flux - limited samples were presented ; these were essential for understanding the true nature of bl lac objects . in pittsburgh
, there was one talk on number counts ( setti 1978 ) ; in como , there were more , plus speculation about luminosity functions , evolution , and unification .
since then , unification has been put on a firm statistical basis using the new flux - limited samples ( see padovani , this volume ) . *
_ gamma - rays
_ another significant advance since como was the discovery of blazars as bright gamma - ray sources , first at gev energies with cgro egret and more recently at tev energies with whipple and subsequent ground - based cerenkov detectors . a large fraction of blazar light had previously been missed .
the highly variable , strong gamma - ray emission confirms that blazars are relativistically beamed and offers strong constraints on the physics of the jet . * _ host galaxies _ there is considerable new data on host galaxies of bl lac objects , from hst and ground - based observatories . in pittsburgh
there was speculation about whether bl lacs sit in galaxies ; it is now clear they are the active nuclei of normal , fairly luminous elliptical galaxies with intrinsically low radio powers . in this talk
i review in broad terms our current understanding of bl lac objects , and point out where progress is being made or where it can be made . most of these topics are discussed in more detail throughout this volume .
the spectral energy distributions ( seds ) of blazars have a universal two - bump structure ( in @xmath0 , fig .
1 ) , almost certainly caused by synchrotron radiation peaking at low energies , and at gamma - ray energies , possibly inverse compton - scattering of soft photons radiation ( although there are other possibilities ; see below ) .
while i may refer to the gamma - ray component as the `` compton component '' and discuss the implications of inverse compton scattering models , bear in mind that this model is not yet established .
the characteristics of blazar seds , and particularly their spectral variability , offer clues to the underlying physics in blazar jets .
( here `` jet '' means the unresolved , inner jets of radio - loud agn , which dominate the observed sed .
the longer - wavelength radio emission comes from more extended regions of the jet , which are sometimes resolved in radio or even optical images . ) for more than a decade , bl lac objects have been sorted into two types according to their broad - band spectral properties ( e.g. , @xmath1 vs. @xmath2 ; ledden & odell 1985 , stocke et al . 1985 ) ; these can be described as `` red '' bl lacs , technically defined as lbl ( low - frequency peaked bl lacs ) and found primarily in radio - selected samples , and `` blue '' bl lacs , a.k.a .
hbl ( high - frequency peaked bl lacs ) , found primarily in x - ray - selected samples .
( strong emission - line blazars [ quasars ] are very similar to red bl lacs . )
it is now becoming clear that the seds , luminosities , emission line and other properties of blazars form a continuous distribution , and that the original red - blue dichotomy was caused by selection biases at high flux limits ( see talks by laurent - muehleisen , padovani , ghisellini , and others , this volume ) .
blazars vary most at frequencies above the peak in both sed components , and intensity variations are approximately correlated between the two peaks ( fig . 2 ; also , ulrich , maraschi & urry 1997 ) .
the lags at high energies are short typically less than a day and depend on wavelength .
the variability amplitudes are usually larger at higher energies , increasing greatly above the sed peak frequencies .
the correlated variability suggests that the same electrons could be radiating both components .
it has long been noted that blue bl lacs are much less variable in the optical than red bl lacs . this has been taken to mean that they are somehow less active . instead , from an sed - based perspective , the difference is an accident of where the sed peak occurs relative to the optical band : in red bl lacs , optical frequencies lie above the synchrotron peak , the region of maximum variability , whereas in blue bl lacs this is well below the synchrotron peak frequency , where variability is typically small .
in fact , the blue bl lacs are at least as variable , in terms of large amplitude and short time scale , as red bl lacs , but at the corresponding part of the sed , i.e. , at x - ray energies . by the same argument , one would predict high euv / x - ray polarization in blue bl lacs , comparable to the optical polarization in red bl lacs .
many speakers address inverse compton scattering ( ghisellini , kirk , marscher , levinson , this volume ) .
the various models share several basic characteristics , including a region of energetic particles moving relativistically , often in an expanding jet geometry commensurate with the unification scheme for blazars ( urry & padovani 1995 ) .
these particles radiate synchrotron radiation and also inverse compton scatter ambient photons to gamma - ray energies .
the ambient photons can be synchrotron photons in the jet ( the synchrotron self - compton , or ssc , model ) , or uv / x - ray photons from the disk , or reprocessed uv ( or even ir ) photons from the broad - line region or obscuring region beyond . which of these soft photon sources dominates depends on their local energy density near the electrons of interest .
because of the bulk relativistic motion of the electrons , photons impinging on the jet from the side or front are relativistically boosted in the frame of the electrons , enhancing their apparent intensity .
this means strong - emission - line sources are almost certainly dominated by the inverse compton scattering of external photons ( the external compton , or ec , mechanism ) , at least at gev energies . in the weak - lined objects , the bl lacs , probably the ssc mechanism dominates at all gamma - ray energies .
a few general points about synchrotron and compton scattering scenarios should be noted .
first , the two sed peaks reflect a characteristic electron energy , perhaps where the cooling time , which decreases with frequency , balances the time - independent escape time . below the synchrotron peak , then , escape could be significant , which could explain the very flat infrared spectra sometimes observed , at wavelengths where the source should be optically thin .
( flat radio spectra , at longer wavelengths , are instead caused by optical depth effects . )
a second interesting point is that , roughly speaking , the ratio of peak frequencies in both red and blue blazars , @xmath3/@xmath4 , appear similar .
( bear in mind , however , that @xmath3 can be affected by klein - nishina effects in the blue blazars , and that gamma - ray data are very sparse . ) assuming this is physically meaningful , the interpretation depends on how the gamma - rays are produced at the second peak .
if ssc dominates in all blazars , then @xmath5 ( where @xmath6 is the characteristic energy of electrons radiating at the synchrotron peak ) , and one would conclude that the peak electron energies are independent of blazar luminosity . if instead the gamma - rays are produced by the ec mechanism in all blazars , then @xmath7 , where @xmath8 is the typical frequency of the soft seed photon and b is the magnetic field .
this would imply the ratio of @xmath9 , a sort of ratio of accretion energy to magnetic energy , is roughly constant in all blazars .
more likely , ssc dominates the gamma - ray peak in low - luminosity blazars and ec the peak in high - luminosity blazars ( see [ ssec : trends ] ) . in this case @xmath9 in luminous blazars must approximately `` match '' @xmath10 in less luminous blazars .
this might indicate that electron acceleration , which determines @xmath6 when cooling is less important , is related to the magnetic field strength and overall source ( accretion ) luminosity , in a kind of energy balance constraint .
there remains the question of whether blazar jets are homogeneous or inhomogeneous .
( of course , jets are unlikely to have uniform electron densities and magnetic fields on all scales , but at high energies the dominant emission region could be approximately homogeneous . )
perhaps surprisingly , much of the existing multiwavelength data are still consistent with a homogeneous model .
for example , the lags between soft and hard x - rays can be explained by energy - dependent cooling in a homogeneous volume ( see [ sec : blue ] ) .
however , inhomogeneities may be needed to explain variability amplitudes even at the highest energies .
for example , at x - ray energies in blue blazars , we typically see variations by factors of 2 , not factors of 10 or 10% .
it is natural to get very large amplitude variations from a single acceleration event the high - energy emission should be `` on '' or `` off '' because the loss time scales are very short . at the other extreme , small amplitude variations occur naturally from the superposition of a large number of strong flares . to get the observed
, intermediate amplitude variability is unnatural in the homogeneous case , and implies instead a quiescent flux , at a level comparable to the flare peak intensity .
this implies that flares are on top of some quiescent level , and that the flaring region varies with large amplitude , which suggests that jets are at least partially inhomogeneous ( sikora , madejski & begelman 1997 ) .
if the dominant emission region is homogeneous , there must be additional electron acceleration above the characteristic @xmath6 , for at least two reasons .
first , a sharp cutoff in electron energy should produce much larger amplitude variations than are observed .
second , the tev spectra of mrk 421 and mrk 501 extend to 10 tev , perhaps even to 30 tev in some bright states ( krawczynski , this volume ) ; if there were a sharp cutoff at @xmath6 , the gamma - ray spectrum should cutoff sharply as well .
thus the acceleration process(es ) must produce electrons of a characteristic energy @xmath6 _ and _ produce a power - law distribution of electron energies above @xmath6 .
this might occur because electrons are universally accelerated to @xmath6 in a blast wave or bulk acceleration process then further accelerated by shocks , or because they are injected at high energies and cool to @xmath6 ( below @xmath6 cooling produces a flat electron energy distribution , flatter still if escape is important ; thus electrons at @xmath6 dominate the emission ) .
more fundamental issues are how energy is extracted from the central black hole and how it is transported to the radiative zone . much of the extracted energy is eventually manifested as kinetic energy in the jet , some fraction of which is then radiated in the observed sed ( the rest feeds the extended radio lobes ) .
very near the black hole , the radiation and particle densities in blazars would be high enough to efficiently generate pairs , even taking relativistic beaming into account .
the resulting cascades would trap high - energy radiation in an optically thick zone , contrary to observation .
this means the observed emission must originate in a radiative zone a fair distance from the black hole , and the energy transport process is effectively dissipationless .
sikora ( 1997 ) has argued that cold protons ( @xmath11 ) are unlikely to carry much energy in blazar jets , assuming the energy distributions of electrons and protons are similar .
this is because cold electrons , which would have an effective lorentz factor @xmath12 due to the bulk relativistic outflow , should comptonize ambient uv radiation up to @xmath13 kev .
the absence of this so - called `` bulk compton '' bump indicates there are too few cold electrons to transport much energy along the jet .
instead , there must be a dissipationless transport of energy , such as poynting flux . observed blazar seds change shape systematically with luminosity , as noted by rita sambruna in her ph.d .
thesis ( sambruna et al .
1996 , sambruna 1997 ) and by fossati et al .
both synchrotron and gamma - ray sed peaks decrease in frequency with increasing luminosity .
at the same time , the ratio between the two sed components , @xmath14 often called the `` compton dominance '' increases , from @xmath15 or even less for the blue bl lacs to as much as 1001000 for some luminous quasars .
the luminosity in emission lines also increases with bolometric luminosity ; the distinction between bl lacs and quasars , at an equivalent width of 5 , is essentially an arbitrary one ( scarpa & falomo 1997 ) .
this is not to say that bl lacs and emission - line blazars are identical , or even that their properties overlap strongly , simply that there is no clear gap between the populations .
a possible exception is the vlbi polarization properties , which appear to differ discontinuously between bl lacs and quasars ( gabuzda , this volume ) .
this apparent dichotomy could be affected , however , by differences in intrinsic physical scale sampled at fixed vlbi resolution .
apart from this uncertain issue , the distinction between quasars and bl lacs occurs at an arbitrary point ( i.e. , 5 equivalent width ) , spanned by a continuous distribution of physical properties .
lest this cause confusion , i repeat : _ of course the average properties of quasars and bl lacs are different , and indeed the extremes of each class are grossly different ! but the point at which one defines a bl lac is arbitrary because the properties are continuous .
_ the observed trends of blazar properties with luminosity , if they reflect an intrinsic physical reality , support a self - consistent picture of equilibrium between electron acceleration and cooling ( ghisellini et al .
1998 ; ghisellini , this volume ) .
the basic paradigm is that electrons are accelerated to similar high energies in all jets , in an approximate power - law distribution , and then they cool on the ambient radiation field .
because the local photon densities are higher in more luminous sources , there is more cooling , and the equilibrium values of @xmath6 are lower , which is reflected in lower frequencies of the synchrotron sed peaks ( see also georganopoulos & marscher 1998 ; marscher , this volume ) .
this is an attractive and viable paradigm but is not yet well tested .
the multiwavelength data for existing samples are sparse and not uniform , particularly at gamma - ray energies . to confirm that the two sed components are produced by the same electrons
requires much better multiwavelength monitoring data , to find correlated variations and to `` identify '' the electrons radiating at each wavelength .
furthermore , the luminosity - dependent paradigm is based on blazars for which there are gamma - ray detections , and those samples may be biased in the direction of the observed sed trends , either through the extra beaming in ec sources ( dermer 1995 ) or through variability ( hartman and wagner , this volume ) .
that is , the gamma - ray - detected blazars may not be representative of blazars as a class , nor representative of the average state of the particular blazar .
because they are very bright at optical / uv / x ray wavelengths , mrk 421 , mrk 501 and pks 2155304 are the best - monitored blazars .
each has been the target of multiple well - sampled , long - duration monitoring campaigns . for mrk 421 , both the high - energy synchrotron emission and the gamma - ray component
have been well monitored because this blazar is consistently bright at tev energies .
mrk 501 has more recently been well monitored , following a dramatic tev and x - ray flare in april 1997 .
pks 2155304 has been monitored for longer and with better sampling at optical through x - ray wavelengths but without simultaneous tev monitoring ( cf .
chadwick et al . 1998 ) . in all three of these blue blazars ,
the synchrotron component dominates out to at least a few kev . in both mrk 421 and
pks 2155304 , the hard x - rays ( @xmath16 kev ) lag soft x - rays ( @xmath17 kev ) by 1 hour ( takahashi et al .
1996 , urry et al .
1997 ; also treves , this volume ) , roughly as expected for a homogeneous source , where the lag time is roughly the radiative time scale at the lower energy . of great interest is the correlated variability between synchrotron and compton components at corresponding points on the sed ; for these blue blazars , the spectral peaks are in the x - ray and tev bands .
mrk 421 appears to flare roughly simultaneously in x - rays and tev but the lag is very poorly constrained ( macomb et al .
1995 , buckley et al .
more recent multiwavelength monitoring of mrk 421 in april 1998 , for a full two weeks with much denser x - ray sampling , should provide better constraints ( takahashi et al .
, in prep . ) . during a bright flare in 1997 ( catanese et al .
1997 , pian et al .
1998 ) , mrk 501 was easily detected at tev energies and with the rxte x - ray all - sky monitor ( asm ) .
the cross - correlation of the two well - sampled light curves ( with at least daily sampling , over several months ) has a highly significant peak at zero lag , with an uncertainty of @xmath18 day ( krawcynski , this volume ) .
this is the first well - measured x - ray / tev correlation .
the close correlation of x - rays and tev emission in blue blazars is strong evidence for a common production mechanism for both synchrotron and gamma - ray sed components . measuring the lags accurately between the real sed peaks , a good determination of which requires detailed monitoring over a large wavelength range
, should distinguish between compton scattering models and alternatives like proton - induced cascades ( pic ) .
in the pic model , protons and electrons are accelerated to extremely high energies ; the primary electrons radiate the low - energy sed component via synchrotron ; protons interact with ambient photons to induce pair cascades ; and a subsequent ( third ? ) generation of pairs radiates the high - energy sed component , also via synchrotron radiation .
the low - energy radiation should appear first , followed by the high - energy component , and the relative amplitude of variability at the two peaks is essentially free .
in contrast , the compton scattering picture predicts true simultaneity and larger amplitude in the gamma - rays , at least for flares caused by electron acceleration .
variability amplitudes are very informative , although the observational data at present are somewhat limited .
this is because , for most gev blazars , egret lacks the dynamic range to measure variations by more than a factor of 10 or so , and also because the corresponding synchrotron peaks are poorly monitored , being in the far infrared . for the tev blazars
, the synchrotron peaks can be well observed but there are very few sources detected in gamma - rays .
that said , if the multi - epoch sed of 3c279 is typical ( fig .
2 ) , the gamma - ray component is generally much more variable than the synchrotron component .
the egret data are near the gamma - ray sed peak in 3c279 , which varies in intensity by more than a factor of 30 ; above the synchrotron peak , in the optical / uv , the simultaneous intensity varied by less than a factor of 2 ( wehrle et al . 1998 ) . indeed
, the gamma - ray variation is more than the square of the apparent synchrotron variation ( with the caveat that we do not actually observe the infrared peak ) .
it should go roughly linearly in the external compton model and roughly as the square in the ssc model , for simple electron acceleration events .
clearly additional complexity is present , such as enhanced beaming in the external compton case ( dermer 1995 ) or possibly a change in the doppler factor ( dermer & chiang 1998 ) the synchrotron peak frequencies in individual objects can change dramatically during a flare , increasing by a factor of @xmath19 during the strong mrk 501 flare ( pian et al . 1998 , catanese et al .
1997 ; also pian , this volume ) .
a corresponding shift in the gamma - ray peak was not seen , probably because klein - nishina effects dominate at the peak .
such strong spectral hardening in blazars had not previously been observed , and it is important to look for these events now that bepposax provides sufficient hard x - ray sensitivity .
( it remains difficult to observe similar spectral hardening in the red blazars because their peaks are often in the far infrared spectral region , where there are few monitoring data , perhaps a handful of iso results slowly coming out . ) +
a prerequisite for understanding jet formation is understanding the demographics of jets in radio - loud agn .
it is now recognized that blazar jets , as evidenced by their range of seds , encompass a range of physical states .
to first order , it appears that the low luminosity , low kinetic power jets of blue blazars have systematically higher frequency synchrotron cutoffs and therefore higher electron energies , while the high luminosity , high kinetic power jets of red blazars have lower electron energies .
however jets are formed , nature produces a distribution of intrinsic properties .
selection effects are strong in blazar samples . not only is their jet emission strongly beamed , but the principal selection bands for existing flux - limited samples , radio and x - ray , are at well separated points on the sed .
thus it is no surprise that current blazar samples are highly biased .
nonetheless , people seem surprised to learn that the extrapolation from those samples , to infer the relative numbers of low - power ( blue ) and high - power ( red ) jets , is uncertain _ by a factor of @xmath19 _ , depending on assumptions .
this is illustrated ( for bl lacs ) by the comparison of integral number counts for radio and x - ray samples ( fig .
the left panel shows the number counts ( urry , padovani & stickel 1991 ) from five major x - ray samples ( open symbols ) , which are primarily blue bl lacs , and the `` bivariate '' x - ray counts for the 1 jy radio sample ( filled triangles ) which are primarily red bl lacs . under the assumption that the x - ray surveys are an unbiased probe of blazars ( i.e. , unaffected by their red / blue nature )
, one would conclude that blue bl lacs are ten times more numerous than red bl lacs .
however , under the assumption that the 1 jy survey is instead unbiased , one comes to the opposite conclusion ( giommi & padovani 1994 , padovani & giommi 1995 ) .
the right hand panel in figure 3 shows the integral radio counts for the 1 jy sample , combined with the `` bivariate '' radio counts for the x - ray samples , i.e. , the x - ray - derived surface density of blue bl lacs at the observed radio fluxes for those samples . the solid line shows the extrapolation to lower radio fluxes of the expected counts for the best - fit unification of radio galaxies and bl lac objects ( urry et al .
if the radio survey were equally sensitive to red and blue bl lacs , then one would conclude that the powerful red objects are 10 times more numerous than blue .
note that the padovani and giommi calculation is approximately symmetric with respect to assumed sed .
they assume a distribution of spectral cutoffs toward higher frequency , with the radio survey being unbiased .
one could equally well assume the x - ray surveys are unbiased , and that there is a distribution of spectral cutoffs toward lower frequency .
the ensuing calculation does indeed reproduce the observed radio counts ( fossati et al .
_ thus the relative numbers of red and blue blazars remain indeterminate at present . _ because blazar sed shapes strongly affect surveys at both radio and x - ray wavelengths , the truth probably lies somewhere in between ( see padovani , this volume ) . an attempt to estimate the bolometric luminosity function , thus correcting in a primitive way for the luminosity effect ( but still a `` bivariate '' approach in the sense that missed objects are still missing ) , implied a much more similar space density in the region of overlapping luminosity ( urry & padovani 1995 ) .
but the fact remains that the question of relative space density is still largely an open one , and requires deeper surveys to solve .
the most promising avenue is to look at deeper flat - spectrum radio samples ; the flux limit of the 1 jy sample is so high that going even a factor of a few deeper should reveal differences between the extreme x - ray - dominant and radio - dominant scenarios .
i have described how , at present , the distribution of jets in nature , high luminosity red jets versus low power blue jets , is not well constrained .
obviously nature makes both types , and their ratio may change in cosmic time .
in particular , if blazar power / luminosity decreases in cosmic time , as suggested by global evolutionary properties of agn , there should be more blue bl lacs locally and more red ones at high redshift , more or less consistent with the different evolutionary properties of the two types .
the x - ray and radio number counts for blue ( open symbols ) and red ( filled symbols ) bl lac objects illustrate the ambiguity over which population is more numerous .
_ left _ observed x - ray counts for x - ray - selected samples are @xmath12 times higher than the `` bivariate '' x - ray counts for radio - selected samples ( radio - derived surface density combined with observed x - ray fluxes ; urry et al .
1991 ) .
_ right _ beaming model extrapolation ( solid line ) of observed radio counts for radio - selected samples ( urry & padovani 1995 ) is @xmath12 times higher than the `` bivariate '' radio counts for x - ray - selected samples ( x - ray - derived surface density combined with the observed radio fluxes ) .
[ fig : counts ] figure 4 .
simulated soft x - ray spectra for three viable models of circumnuclear absorption in blazars , at typical grating resolution .
current low - resolution ccd data have detected broad absorption clearly but can not unambiguously determine its origin , as shown by the residuals to a simple power law fit to asca data ( bottom panel ) .
axaf , xmm and/or astro - e will clearly distinguish among the possibilities .
[ fig : axaf ]
very little is known about the sub - parsec structure of bl lac objects . according to the agn paradigm ,
unless structures change drastically at low luminosity ( and there are few if any hallmarks of such abrupt change ) , the centers of bl lacs should contain the same broad emission line regions and accretion disks as conventional quasars .
clearly at least some bl lacs have a broad - line region ( e.g. , corbett et al .
1996 , vermeulen et al .
1995 , stickel et al .
1991 ) , as was evident already at the pittsburgh conference ( mgii lines seen in bl lacs ; miller 1978 ) .
these properties have not been surveyed systematically , the blue bump and broad lines being hidden in large part by the relativistically boosted jet continuum , so the distribution of line or accretion disk luminosities is only poorly known .
it is extremely important to survey the `` thermal '' properties of blazars .
one clue comes from x - ray spectroscopy . at present , the state of the art ccd spectroscopy with resolving power of @xmath12 in the soft x - ray energy range suggests there is highly ionized circumnuclear material in bl lac objects , producing absorption near 0.6 kev ( fig .
at least three quite different models for the absorber an outflowing wind , a line blend , and a so - called `` warm absorber '' fit the low - resolution ccd data equally well .
with the much higher spectral resolution soon available with axaf , xmm , and astro - e , identifying the origin of the x - ray absorbing gas in bl lacs , and perhaps mapping it crudely via variability , should be relatively easy .
such data can be used to map the innermost structure of bl lac objects .
host galaxies are easily detected in bl lac objects , as found in our large hst survey ( falomo et al .
1997 , urry et al .
1999a , b ; see also falomo , this volume ) .
bl lac host galaxies are large , round , smooth ellipticals , roughly 1 mag brighter than a typical ( @xmath20 ) elliptical ( wurtz et al . 1996 ; urry et al .
1999b ) , fitting well the @xmath21-@xmath22 projection of the fundamental plane for elliptical galaxies , at the high luminosity end ( urry et al .
1999b ) .
according to unified schemes , the unbeamed properties of bl lac objects , such as host galaxies , should be identical to those of their parent population , nominally fri radio galaxies .
indeed , bl lac host galaxy magnitudes are typical of fris in the same redshift range .
_ because flux - limited samples of bl lacs usually span higher redshifts than fris , as do friis , simple comparisons of host - galaxy magnitude distribution , ignoring redshift , can lead to incorrect conclusions .
another common mistake is to compare , in a ks sense , the observed distribution of some property of blazars and radio galaxies , such as host galaxy magnitude . because of selection effects related to beaming and/or the redshift - luminosity correlation in flux - limited samples , the shape of the distributions will not necessarily match , even in intrinsically identical agn ; rather , the parameter values `` match '' if they span the same range . ) furthermore , both the host galaxy magnitude and the extended radio power of bl lacs must match those of the parent population , since the division between fri and frii galaxies appears sharp only in this two - dimensional parameter space ( owen & ledlow 1994 ) .
figure 5 shows clearly that the extended radio powers and host galaxy magnitudes of bl lac objects overlap those of fri radio galaxies and are unlike friis .
the naive projection of these data in optical magnitude alone , independent of radio luminosity , has led some to the incorrect conclusion that friis are a better match for the parent population .
this mistake happens in part because there are no detections of bl lac host galaxies as luminous as the very brightest fris , typically the central dominant galaxies in rich clusters .
indeed , bl lacs may avoid the very richest clusters ( owen , ledlow & keel 1996 ) , but as figure 5 illustrates , this must be a second - order issue in the largely correct unification of fris and bl lacs .
comparison of unbeamed properties extended radio power versus host galaxy magnitude of bl lac objects ( filled circles and squares ) and radio galaxies shows that fris ( labeled `` 1 '' ) are a much more likely parent population than friis ( labeled `` 2 '' ) .
the bl lacs were taken from complete radio- or x - ray flux limited samples and host galaxy determinations were made using hst data ( urry et al .
the fri / ii samples were from the 2 jy sample ( wall & peacock 1985 ) , selected in a similar way to the bl lac radio samples .
[ fig : radopt ]
now for some speculation about how studies of bl lac objects might inform our larger understanding of agn .
a critical issue is the radio - loud / radio - quiet dichotomy , which essentially translates to why only some agn appear to form relativistic jets .
blazar studies have indicated the importance of selection biases in samples flux - limited at wavelengths located in different parts of the sed . by analogy , is it possible that the distribution of @xmath1 is not intrinsically bimodal , i.e. , that there is no sharp physical distinction between radio - strong and radio - weak agn ?
the radio flux distribution for pg quasars is not so strongly bimodal ( kellermann et al .
1989 ) ; perhaps beaming effects ( as yet unquantified ) could enhance the radio relative to the optical emission in the so - called radio - loud objects .
there do exist `` transition '' objects with intermediate values of @xmath1 , which have been interpreted as radio - quiet quasars with significantly beamed cores ( falcke , patnaik & sherwood 1996 ) . in deeper radio samples ,
intermediate values of @xmath1 should be further filled in ; possibly this is seen in the first radio survey ( helfand , priv .
comm . )
alternatively , sharp transitions ( intrinsic bimodality ) in radio loudness could result from a critical dependence on jet kinetic power and host galaxy mass , analogous to the fri / ii transition ( bicknell 1995 ) . at very high jet powers
, one gets an frii type source ( an emission line blazar when aligned ) ; at lower jet power , there is a sharp transition to diffuse fri - type jets ( a bl lac object when aligned ) ; perhaps at some still lower power , the jet can not form or is quenched so early that one gets a radio - quiet object .
flux - limited samples can be corrected in a straightforward way for objects that have been missed because of their lower fluxes .
thus we have confidence that they are representative .
but such corrections work only _ when the excluded objects are intrinsically identical to the discovered objects_. if the missing objects instead have a different spectral energy distribution , then their number and properties depend on assumptions about their numbers and seds basically , the correction is a circular ( model dependent ) problem . as far as i can see , this problem has not been addressed explicitly for the radio - loud / radio - quiet issue .
the future in blazar research is very rich . here is a personal list of the key questions ahead : * what is the energy production and balance in blazar jets ? * to understand the extraction of energy from the black hole , we have first to quantify the energy distribution among particles , magnetic fields , and photons .
a practical near - term issue is identifying how the gamma - rays are produced .
several years without gev data loom ahead , so for the moment progress means detecting more tev sources ; the prospects are good , as tev telescopes rapidly improve and extend to southern latitudes .
tev sources should be monitored also in x - rays , to look for correlated variability in the two spectral components ; ditto for red objects at optical and gev energies with glast , possibly sooner with agile - type gamma - ray satellites or ground - based gamma - ray telescopes like magic .
the goal is to identify the electrons radiating in each component , and to see if / how they are related .
pic models predict more independent variations than ec or ssc models . *
what kind of jets does nature make ? * quantifying the distribution of jet powers the intrinsic luminosity function of jets , in a global sense is an important constraint on jet formation scenarios .
the present uncertainty in the relative numbers of blue and red jets can be resolved with deeper samples ( some reported here ; e.g. , perlman , laurent - muehleisen , this volume ) .
given the range of seds in blazars , selection effects are strong and samples flux - limited in any one band offer a necessarily biased and incomplete picture .
* how do radio - loud agn form and evolve ? * the study of host galaxies is a direct probe of the relative evolution of nuclei ( black holes ) and normal galaxies .
the high - redshift blazars turning up in deeper samples ( e.g. , dxrbs ) are essential for studying the evolution of low - luminosity jets ( which must be beamed to be seen at high redshift ) . the new and upcoming large telescopes with infrared sensitivity , including ngst , the vlt , magellan , gemini , etc .
, are ideally suited to these studies .
further study of blazar environments is also important . with its high spatial resolution and excellent sensitivity
, axaf will be a powerful tool for detecting clusters around bl lac objects .
this is not only easier than counting galaxies in the optical and constraining cluster membership with multi - object spectroscopy , but it probes the diffuse gas , which contains most of the mass , so classification of richness is more meaningful . for their hospitality during my sabbatical , i thank the center for astrophysical sciences at the johns hopkins university and the brera observatory of milan .
i am grateful to my blazar colleagues in italy and elsewhere for stimulating discussions and collaborations , and i thank the organizers for hosting an excellent conference in turku . this work was supported in part by nasa grants nag5 - 2538 and go-06363.01 - 95a , and was based in part on observations made with the nasa / esa hubble space telescope , obtained at the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas 5 - 26555
the die - hard blazar man is dick miller , the only person to attend all three bl lac conferences . _
bicknell , g. v. 1995 , , 101 , 29 blandford r. d. & rees m. j. 1978 , in pittsburgh conference on bl lac objects , ed .
a. n. wolfe ( pittsburgh university press ) , p. 328
buckley , j. h. , et al .
1996 , , 472 , l9 catanese m. , bradbury s. m. , breslin a. c. et al .
1997 , apjl , 487 , 143 chadwick , p. m. , et al .
1998 , , in press ( astroph 9810209 ) corbett , e. a. , et al .
1996 , , 281 , 737 dermer , c. d. 1995 , , 446 , l63 dermer , c. d. , & chiang , j. 1998 , new astronomy , 3 , 157 falcke , h. , patnaik , a. , & sherwood , w. 1996 , apj , 473 , l13 falomo , r. , urry , c. m. , pesce , j. e. , scarpa , r. , giavalisco , m. , & treves , a. 1997 , , 476 , 113 fossati g. , celotti a. , ghisellini g. & maraschi l. 1997 , mnras , 289 , 136 georganopoulos m. & marscher a. p. 1998 , apj , 506 , 621 ghisellini g. haardt f. & svensson r. 1998 , mnras , 297 , 348 giommi p. & padovani p. 1994 , mnras , 268 , l51 kellermann , k. i. , sramek , r. , schmidt , m. , shaffer , d. b. , & green , r. 1989 , , 98 , 1195 ledden j. e. & odell s. l. 1985 , apj , 298 , 630 macomb , d. j. , et al .
1995 , , 449 , l99 miller , j. s. , french , h. b. , & hawley , s. a. 1978 , in pittsburgh conference on bl lac objects , ed .
a. n. wolfe ( pittsburgh university press ) , p. 176
owen , f. n. , & ledlow , m. j. 1994 , in the physics of agn , ed .
g. v. bicknell , m. a. dopita , & p. j. quinn ( asp ) , p. 319
owen , f. n. , ledlow , m. j. , & keel , w. c. 1996 , , 111 , 530 padovani p. & giommi
p. 1995 , apj , 444 , 567 pian e. , vacanti g. , tagliaferri g. , et al .
1998 , apjl , 492 , 17 sambruna r. , maraschi l. & urry c. m. 1996 , apj , 463 , 444 sambruna r. 1997 , apj , 487 , 536 scarpa r. & falomo r. 1997 , a&a , 325 , 109 setti g. 1978 , in pittsburgh conference on bl lac objects , ed .
a. n. wolfe ( pittsburgh university press ) , p. 385
sikora , m. 1997 , in proceedings of the fourth compton symposium , ed . c. d. dermer , m. s. strickman , j. d. kurfess ( williamsburg : aip ) v. 410 , 494 sikora , m. , madejski , g. m. , & begelman , m. c. 1997 , in relativistic jets in agn , ed .
ostrowski et al .
( krakow : astr . obs . jag . ) , p. 287
stickel , m. , fried , j. w. , khr , h. , padovani , p. , & urry , c. m. 1991 , , 374 , 431 stocke j. t. , liebert j. , schmidt g. , et al .
1985 , apj , 298 , 619 takahashi , t. , et al .
1996 , , 470 , 89 ulrich m. h. , maraschi l. & urry c. m. 1997 , ara&a , 35 445 urry , c. m. , et al .
1997 , , 486 , 799 urry , c. m. , falomo , r. , scarpa , r. , pesce , j. e. , treves , a. , & giavalisco , m. 1999a , , 512 , in press ( astroph 9809030 ) urry c. m. & padovani p. 1995 , pasp , 107 , 803 urry c. m. , padovani p. & stickel m. 1991 , apj , 382 , 501 urry , c. m. , et al .
1999b , in preparation vermeulen , r. c. , et al .
1995 , , 452 , l5 wehrle a. e. , pian e. , urry c. m. , et al .
1998 , apj , 497 , 178 wurtz r. , stocke j. t. & yee h. k. c. 1996 , apjs , 103 , 109 wall j. v. & peacock j. a. 1985 , mnras , 216 , 173 | the past 20 years have seen phenomenal progress in our understanding of bl lac objects .
they form part of the blazar class , which are radio - loud agn whose relativistic jets are aligned along our line of sight .
several critical milestones have helped establish this picture , first proposed at the pittsburgh bl lac meeting 20 years ago , most recently the egret and tev detections of beamed gamma - ray emission .
the spectral energy distributions are double peaked and follow a self - similar sequence in luminosity , which can be explained by electron cooling on ambient photons .
this simple paradigm has yet to be tested , and further questions remain , notably about physical conditions in blazar jets the kinetic power , magnetic energy density , acceleration time scales , proton content , etc . and how this energy is transported in the innermost regions .
some clues are available from multiwavelength monitoring campaigns although better sampling over longer periods is clearly called for . recent work on the host galaxies of bl lac objects supports their unification with low - power , aligned fri radio galaxies .
nature makes jets with a large range of kinetic powers , but the distribution of such jets i.e. , the relative number densities of low - luminosity ( `` blue '' bl lacs ) or high - luminosity ( `` red '' bl lacs and fsrq ) blazars is highly uncertain .
deeper radio - selected samples of flat - spectrum sources may resolve the degeneracy in the present demographic analyses . since according to unified schemes blazars are representative of all radio - loud agn , their jet properties have broad implications .
future euv / x - ray observations will illuminate the circumnuclear structure , especially the hot , highly ionized , high velocity gas on sub - parsec scales , which could play a role in jet dynamics and could possibly affect the formation of fri vs. frii type jets .
the study of blazars may also help us eventually understand the difference between radio - quiet and radio - loud agn . |
ferromagnetic thin films exhibiting a magnetic domain structure in the form of thin parallel stripes have been the subject of intense research in the last few decades , both experimentallysaito , hehn , gehanno2,okamoto , buteramfm , fega and theoretically.kooy,murayama,sukstanskii,brucas,hubert this kind of structure is observed in films that present an out of plane anisotropy component ( due to stress , crystalline texture , interfacial or other effects ) and in a simplified picture it can be described as a periodic pattern of parallel in - plane magnetized regions in which the magnetization has a relatively smallcomponent that points alternatively in the two directions that are normal to the film plane .
a stripe ( or bubble ) pattern is generally observed for all film thicknesses when the perpendicular anisotropy energy constant , @xmath5 is larger than the demagnetizing shape energy , @xmath6 , ( @xmath7 is the saturation magnetization ) but can also be found below a critical thickness @xmath8 when @xmath9 is smaller than one .
the transition from planar to stripe domains at @xmath10 @xmath11 is due to the minimization of the total magnetic energy which can include the contribution of anisotropy , demagnetizing , domain wall and zeeman ( for @xmath12 ) terms .
the critical thickness depends on the material properties such as the effective anisotropy , the saturation magnetization and the exchange stiffness constant , and also on the external field .
there are several models for the calculation of @xmath11 see for example refs .
[ ] , that predict larger values of @xmath13 in materials with a large saturation magnetization , a large exchange , or a small anisotropy .
the value of the critical thickness is in the range of 20 - 30 nm for co,@xcite partially ordered fepdgehanno2 or disordered fept filmsbuteramfm , butera1,butera2,buterafmr , buterassw , jonas , buterafmr2 , and can take larger values ( of the order of 200 nm ) in films with lower anisotropy such as permalloy.@xcite films with stripe domains have characteristic @xmath14 vs. @xmath15 in - plane loops in which the following features are often observed:@xcite @xmath16 the low field part of the curve increases almost linearly from remanence until the saturation field is reached .
this in - plane saturation field was shown to increase with film thickness following approximately the relationship @xmath17 , $ ] with @xmath18 @xmath19 due to the formation of the stripe structure the in - plane coercivity increases abruptly and the remanence decreases considerably above @xmath20 @xmath21 for @xmath22 rotatable anisotropy , _
i.e. _ the alignment of the stripe structure at remanence in the direction of a previously applied field , is observed .
the magnitude of this anisotropy also increases with film thickness and is usually characterized by a field @xmath23 .
@xmath24 the period of the stripe structure increases approximately as the square root of the film thickness , @xmath25 the study of the magnetic interactions present in films in which a crossover from a planar to a striped magnetic domain structure is observed can then give a deeper insight to understand this behavior .
both henkel plotsspratt and delta@xmath26 ( @xmath27 curves,@xcite together with the magnetic viscosity , @xmath28@xcite can be used to estimate the sign of the magnetic interactions and the magnetic reversal volumes in the samples .
magnetic interactions have been widely studied in small particles,@xcite thin continuous films,@xcite granular systems@xcite and nanostructured films@xcite using magnetic remanence measurements .
the @xmath2 curve is defined as the difference between two remanence curves:@xmath29where the @xmath30 curve ( also known as the isothermal remanent magnetization , irm ) is obtained by starting from a state of zero remanence , erased following a well defined protocol , and then measuring the magnetization at zero field after applying fields of increasing magnitude . the @xmath31 ( or dc demagnetization , dcd ) curve
is obtained by saturating the sample in a negative field and then repeating the same procedure as for the @xmath32 curves .
these two curves are usually normalized to the remanence saturation value ( @xmath33 ) and labeled as @xmath34 and @xmath35 . in the case of a noninteracting system wohlfarth predicted@xcite that the two remanence curves should be identical and hence @xmath36 . if @xmath37 the effects of magnetic interactions can been accounted for using a phenomenological model@xcite for the effective interaction field , @xmath38 that takes into consideration dipolar - like ( demagnetizing ) and exchange - like ( magnetizing ) interactions . in this model @xmath39 which means that the interaction field has a linear dependence with @xmath40 ( which can be both @xmath34 or @xmath35 ) with a slope of magnitude @xmath41 . this parameter can be either positive or negative depending on the dominant type of interaction , exchange - like or dipolar - like , respectively .
the term with the parameter @xmath42 accounts for first order interaction field fluctuations from the mean field .
a numerical method to calculate @xmath41 and @xmath42 is described in ref . [ ] , but they can be more easily obtained from the experimental data following the procedure of ref . [ ] @xmath43 in the above expression @xmath44 is the applied field normalized to the remanent coercivity @xmath45 ( defined as the reverse negative field that , after saturation in the positive direction , produces a zero magnetization at zero field ) and @xmath46 is the remanent magnetization at the point where @xmath2 curves cross zero . in order to get a deeper insight in the magnetic behavior ,
remanence measurements are often complemented with magnetic relaxation experiments .
when a sample is magnetized in a negative saturating field and after that a positive field is applied , the magnitude of @xmath14 often varies linearly with the logarithm of time @xmath47 .
changes in @xmath14 are due to thermally assisted processes that provide the necessary energy to overcome the barrier energy of magnitude @xmath48 .
the proportionality parameter is the magnetic viscosity @xmath49 and the relationship is often written as:@xcite@xmath50with @xmath51 the initial time and @xmath52 the initial value of @xmath14 at @xmath53 for a given @xmath54 the viscosity @xmath49 can be shown to depend on temperature , @xmath55 , saturation magnetization , @xmath7 , and the distribution of activation energies , @xmath56 , in the following way@xcite@xmath57the magnetic viscosity depends also on the forward applied field , through the dependence of @xmath56 on @xmath58 and is generally maximum for an applied field @xmath59 which is close to the macroscopic coercive field @xmath60 .
viscosity and remanence measurements can be related using the field derivative of the dcd curve , known as the irreversible susceptibilitygaunt2 @xmath61the variation of the activation energy with the magnetic field can be related to the so called activation volume , @xmath62 , where @xmath63 is a constant of the order of unity and its value depends on the kind of system that is under consideration .
simple calculations@xcite for monodomain particles or strong domain - wall pinning give @xmath64 while for weak domain - wall pinning @xmath65 if demagnetizing effects are considered,@xcite @xmath66 for strong pinning and @xmath67 for weak pinning . using eqs .
[ [ s ] ] and [ [ chi ] ] the activation volume can be written as:@xmath68 in the case of thin films in which the magnetization changes by a process of domain wall motion , the activation volume can be interpreted as the volume swept by a single jump between pinning centers .
this volume is usually related with the fluctuation field , @xmath69 , defined as:@xcite@xmath70 magnetic interactions in fept have been investigated in different systems , including continuous films,@xcite annealed multilayers,@xcite exchange - coupled bilayers,@xcite granular films,@xcite and nanoparticles,@xcite all in the atomically ordered l1@xmath71 phase .
negative interparticle interactions were reported in the cases of films , annealed multilayers and nanoparticles , when the external field was applied parallel to the in - plane direction ( these films show in - plane anisotropy ) . on the other hand , continuous films exhibiting out of plane anisotropyjeong , wei present positive @xmath2 curves when remanence curves are measured perpendicular to the film plane .
magnetic relaxation has been reported in the case of exchange - coupled fe / fept bilayers,@xcite annealed fe / pt multilayers,@xcite and polycrystalline thin filmsjeong all of them in the hard magnetic phase . for a single layer of 10 nm of fept with an average grain size of @xmath72 nm , the authors in ref .
[ ] reported @xmath73 12500 nm@xmath74 in the second case the authors estimated @xmath73 1200 nm@xmath75 for a multilayer with a total thickness of 15 nm . in the last case an activation volume @xmath73 400 nm@xmath75
was estimated for a film 5 nm thick with a crystalline grain size of 10 nm .
this last sample presented a maze structure of magnetic domains at remanence , consisting of irregular elongated regions magnetized perpendicular to the film plane with a length of several micrometers and a width of 100 - 150 nm . assuming spherical reversal volumes , the corresponding `` activation diameters '' are @xmath76 , 13 , and 9 nm , respectively .
note that if the activation volume is divided by the film thickness , and cylindrical domains are assumed , the resulting `` activation length '' is in the range of 40 nm for the first sample and 9 nm in the last two systems .
as far as we know , magnetic interactions and time dependent effects in fept films in the a1 disordered phase have not been yet characterized . the possibility to tune the domain structure by varying the film thickness
can be used to study how these effects are affected by the way in which the magnetic domains order . in the following sections we present a detailed experimental study in a series of as - made fept thin films of different thicknesses in which the magnetic interactions have been investigated by means of dcd - irm , delta-@xmath14 plots and magnetic viscosity measurements
fept films have been fabricated by dc magnetron sputtering on naturally oxidized si ( 100 ) substrates .
a detailed description of the preparation and the structural characterization can be found in ref .
the samples were deposited from an fept alloy target with a nominal atomic composition of 50/50 .
we sputtered eight films with thicknesses of 9 , 19 , 28 , 35 , 42 , 49 , 56 and 94 nm .
the samples were studied using x - ray diffraction , transmission electron microscopy ( tem ) and energy - dispersive x - ray spectroscopy ( eds ) techniques .
the x - ray diffractograms showed that the samples grow in the fcc a1 crystalline phase , without traces of the ordered l1@xmath71 structure .
a [ 111 ] texture normal to the film plane was observed and comparison with stress released films revealed that as - made samples were also subjected to an in - plane compressive stress .
an average crystallite grain diameter of 4 nm was obtained from tem micrographs .
the photoemission spectra indicated that the fe / pt atomic ratio of the films was approximately @xmath77/55 .
stress effects are the main contribution to an effective magnetic anisotropy perpendicular to the film plane of magnitude @xmath78 erg / cm@xmath75 , which gives rise to a magnetic domain structure in the form of stripes for @xmath80 nm . as we have already shown in ref .
[ ] using magnetic force microscopy ( mfm ) techniques , the half period of the stripe pattern scales with the square root of the film thickness starting at @xmath81 nm for @xmath82 nm and reaching @xmath83 nm for @xmath84 nm . for @xmath1 an in - plane planar domain structure is observed . in both domain
regimes a strong correlation between the domain configuration and the shape of the hysteresis loops was found .
the dc demagnetization ( dcd ) , isothermal remanent magnetization ( irm ) and viscosity data were measured using a lakeshore model 7300 vsm , capable of a maximum field of 10000 oe . for the dcd measurements we used the following sequence of applied fields @xmath85 in this case a negative saturation field @xmath86
is applied before each data point is acquired at @xmath87 after applying a field @xmath88 in most cases we set @xmath89 oe and @xmath90 oe , depending on the coercivity of the sample .
a waiting time of 5 seconds was used before measuring the remanent magnetization .
there is an alternative field sequence for performing dcd experiementswang @xmath91 in which the saturation field is applied only at the beginning of the experiment . in principle , this method should be less sensitive to the waiting time and the field step @xmath92 , and differences between the dcd and irm curves due to viscosity effects are minimized . in our case
we did not observe significant differences between both dcd sequences and decided to use the first method .
the irm curve is obtained by starting from a demagnetized state and measuring the magnetization at zero field following the sequence @xmath93the ideal demagnetized remanent state is the one obtained by heating the sample above the curie temperature , @xmath94 , and then cooling in zero field . because of the appearance of irreversible effects in the magnetic response,@xcite our films can not be heated to @xmath94 @xmath95 k , so we adopted two different protocols to demagnetize the samples .
the `` linear '' demagnetization routine is the usual procedure in which the sample is saturated in one direction and a sequence of decreasing fields is applied in both senses , until zero field is reached .
films can be also demagnetized in a slowly decreasing field ( from saturation to zero ) while they are quickly rotated around an axis perpendicular to the magnetic field .
the `` rotating '' demagnetization routine usually gives a remanent state that is more disordered and isotropic in the film plane than in the linear case , resembling the state that can be obtained by cooling the sample from above @xmath94 . in the case of magnetic relaxation measurements films
were saturated in a negative field of 5000 oe , a positive field was then applied and kept constant during the whole experiment while the magnetization was measured in intervals of 10 seconds during approximately 30 minutes .
we calculated the viscosity from the linear fit of the time variation of @xmath14 ( eq . viscosity ) .
the same routine was repeated for several fields in the vicinity of @xmath60 from which the magnetic viscosity @xmath96 is obtained .
in all films we have measured the irm curves using the two demagnetizing sequences mentioned in the previous section . for films with @xmath97 nm additional care
must be taken in order to reach a truly demagnetized state , because the magnetization switching at @xmath98 occurs in a very narrow field range of only a few oe . the differences between `` rotating '' and `` linear '' demagnetizing routines are more pronounced in thicker films . in fig .
[ figdemag ] ( a ) we show the upper right quadrant of the hysteresis loop for the film with @xmath84 nm , together with the virgin curve obtained after demagnetizing the film using the rotating routine .
it can be observed that there is a field region in which the virgin curve is not within the hysteresis loop .
this effect is almost absent when the sample is demagnetized using the linear sequence and to explain it one must consider that the remanent state obtained when the sample is demagnetized using the rotating routine consists of an array of randomly oriented stripes.buteramfm on the other hand , in the case of the linear protocol almost all stripes are already aligned at remanence in the direction of the demagnetizing field .
when the sample is saturated , rotational anisotropy imposes an easy magnetization axis along the field direction and the stripes are always aligned in that direction . taking into account these effects
one can understand why in the linear case the virgin curve stays inside the loop , while after the rotating cycle larger fields are needed to reach the same magnetization value because part of the field energy is used in aligning the stripes in the direction of the applied field .
the same differences are observed in the irm curves , as can be seen in fig .
[ figdemag ] ( b ) . in this case
starting from an initially more disordered and isotropic state ( rotating routine ) , makes more difficult the magnetization of the sample in the direction of the applied field .
note that the magnetization process occurs in several steps . in the low field region ,
a relatively fast initial increase of @xmath34 ( from @xmath99 to @xmath100 occurs for fields @xmath101 @xmath102 oe , for @xmath84 nm . we associate these changes to domain wall movement in the small fraction of regions which were already aligned in the direction of the applied field .
then @xmath34 stays relatively constant until @xmath103 oe , which is more or less coincident with a kink in the virgin magnetization curve or the beginning of the reversible part of the @xmath104 loop .
these features were assigned in ref .
[ ] to the rotational anisotropy field @xmath23 , the field necessary to rotate the in - plane easy axis of the stripes in the direction of the applied field .
once the stripes are aligned they can be more easily moved by the mechanism of domain wall displacement and a very large increase in @xmath34 ( from @xmath105 to @xmath106 occurs in the range @xmath107 oe .
comparison with the hysteresis loop suggests that for @xmath108 oe and until @xmath109 oe irreversible changes in @xmath34 are probably due to the rotation of regions that are magnetized perpendicular to the film plane .
the linearly demagnetized irm curve shows similar characteristics , but the irreversible changes at low fields ( @xmath110 ) are considerably larger , with @xmath34 reaching almost 50% of the saturation value . above this field a rapid increase and then a more gradual approach to saturation is observed , with the same overall behavior already described for the rotating routine .
in the rest of the films there are still differences between both demagnetizing protocols , but they tend to disappear as the films become thinner . for @xmath111
nm both irm curves are almost identical . in fig .
[ figirm - dcd ] we show the normalized dcd and irm ( starting from a rotating demagnetizing cycle ) curves for all films .
we have plotted @xmath35 and @xmath112 in order to compare both measurements .
the most significant feature that can be observed is that for @xmath113 nm the irm is above the dcd curve and the relationship is inverted for @xmath114 nm . as can be deduced from eq .
[ deltam ] this implies a change in sign in the @xmath2 curve that is indicating a change in the dominant magnetic interactions .
the fact that @xmath115 @xmath35 in the case of thinner films is telling us that in these samples the saturated state can be reached more easily , i.e. the magnetic interactions favor a magnetized state . in thicker films the irm is always below the dcd curve , which reflects that dipolar - like interactions are dominant in these samples . from the field where @xmath35 and @xmath112 curves cross zero
we can extract the remanent coercivity @xmath45 and the irm half reversal field , @xmath116 , respectively .
we will show later that the normalized difference @xmath117 may be used as a very good estimation of the sign and magnitude of the magnetic interactions .
this quantity is very similar to the so - called interaction field factor , iff@xmath118 , differing only in the normalization variable . in fig .
[ fighrem ] we plotted these two fields , together with the coercivity @xmath60 , and the field @xmath23 obtained from ref .
this field is a measure of the average magnetic field needed to overcome the rotational anisotropy . for a stoner - wohlfarth system the two remanence fields should have the same value as @xmath119 which is relatively small for the thinner films , increases considerably when the stripe structure is formed , has a maximum at @xmath120 nm and levels off at @xmath121 oe for larger thicknesses .
this behavior is approximately followed by @xmath122 although as expected @xmath123 , but is definitely not true for @xmath116 . the irmreversal field increases continuously with film thickness giving another indication of the change in the magnetic interactions when the stripe structure is formed .
as already discussed in the case of the @xmath84 nm film the irm curve is a fingerprint of the field necessary for gradually aligning the domains that are not parallel to @xmath15 in the direction of the applied field .
it is then expected that @xmath116 values follow closely the thickness dependence of @xmath23 , the field needed to overcome the rotational anisotropy . as can be seen in fig .
[ fighrem ] both fields follow a similar trend , with the differences in the absolute values arising from the different remagnetizing mechanism that @xmath116 and @xmath124 describe .
one of the methods to characterize qualitatively the magnetic interactions is by using the @xmath2 plots ( see eq . [ deltam ] ) , which reflect the deviations from the stoner - wohlfarth behavior .
as already mentioned , if the irm is above the dcd curve the @xmath2 plot is positive and the interactions tend to be of the exchange type , favoring a magnetized state .
dipolar - like interactions are more important when @xmath2 is negative . in fig .
[ figdeltam ] we show the @xmath2 plots for all the studied samples as a function of the applied field ( normalized by the remanent coercivity , @xmath45 ) .
we have used full symbols to indicate @xmath125 plots obtained from an irm curve that was isotropically demagnetized ( rotational routine ) and open symbols for the case of a linearly demagnetized sample .
note that there are differences between both @xmath125 curves in the case of thicker films that tend to decrease gradually as the thickness is decreased . for @xmath113
nm the two curves are almost coincident .
these results show again explicitly that the dominant interaction changes from magnetizing to demagnetizing when the stripe structure starts to develop at @xmath126 nm and they also give additional evidence of the effects of the rotational anisotropy on the irm remanence curves .
we have already shown in figs .
[ figirm - dcd ] and [ fighrem ] that the fields where the dcd and irm curves cross zero are more separated in the case of thicker films .
this difference explains the shift in the minimum in the @xmath2 plots from @xmath127 to at least twice this value for @xmath84 nm . an estimation of the strength of the magnetic interactions can be obtained from eq .
[ alpha ] , which gives the interaction parameter @xmath41 of the che and bertram model.@xcite the integral of the @xmath2 plots as a function of @xmath128 is presented in fig .
[ figalpha ] .
again we show values of @xmath41 obtained with both demagnetizing routines .
we have plotted in the same figure the quantity @xmath129 which may be also used to estimate the magnetic interactions . in the case of perfectly square @xmath130 and @xmath112 curves , the values of @xmath41 and @xmath131 should be the same because the @xmath2 plot is rectangular with an area @xmath132 due to the different distribution of switching fields in the irm and dcd curves , the values of @xmath133 differ from this simple estimation but , as can be observed in fig .
[ figalpha ] , the values and the shape of the curves of @xmath41 and @xmath134 as a function of film thickness are very similar , confirming that @xmath131 is also a very reasonable parameter for the estimation of the magnetic interactions .
a dimensional analysis of eq . [ alpha ] reveals that the interaction parameter @xmath41 may be associated to a normalized energy and hence could be correlated with the dominant energy contribution to the magnetic domain configuration . in the case of domains formed by parallel slabs of size
@xmath135 magnetized perpendicular to the film plane ( see the sketch in fig .
[ fig6 ] ) it is possible to calculate@xcite the magnetostatic energy per unit surface area as @xmath136=0.374m_{\perp } ^{2}l.$ ] we can estimate this energy for the different films that show a stripe structure by identifying the thickness of the slabs with the values of the half period of the stripe structure ( @xmath137 and estimating the component of the magnetization perpendicular to the film plane as @xmath138^{2}}.$ ] the thickness dependence of @xmath139 and @xmath140 is shown in the inset of [ fig6 ] . in the last formula @xmath141 is the remanence in the direction of the applied field ( obtained from the saturation value of dcd or irm measurements ) and was normalized by @xmath142 nm@xmath143 instead of @xmath7 to consider that there is always a small component of @xmath14 that is neither parallel to the anisotropy axis induced by @xmath15 nor parallel to the film normal . in the main panel of fig .
[ fig6 ] we plotted the interaction parameter @xmath41 as a function of @xmath144 for @xmath145 nm and found that there is a good linear correlation between both magnitudes , indicating that for films with @xmath146 it is energetically favorable to form a stripe structure which has a magnetostatic energy that increases with the stripe period ( and the film thickness ) .
magnetic relaxation measurements were also performed in the whole set of samples . for films with @xmath147
nm we found that eq .
[ viscosity ] is closely obeyed ( see fig .
[ figrelaxation ] ( a ) ) while in the case of thinner films ( 9 nm and 19 nm ) the relaxation of the magnetization follows a nonlogarithm behavior or occurs in discrete steps , as can be observed in fig .
[ figrelaxation ] ( b ) .
this last behavior has been only detected for fields very close to @xmath60 and is an indication of the very narrow distribution of energy barriers ( or switching field distribution ) in the thinner films .
for the relaxation measurements in these samples we took data every 0.2 oe which is almost equal to the stability limit ( 0.1 oe ) of the electromagnet power supply .
possible fluctuations in the applied field can switch the magnetization and it is then difficult to conclude that in this case the reversal of the magnetization is only due to thermal effects . the film with @xmath148 nm was at the limit where a reasonably linear fit could be obtained and was included in the viscosity data , although with a larger uncertainty in the determination of @xmath49 .
the viscosity parameter , obtained from the slope of curves similar to fig .
[ figrelaxation ] ( a ) , is plotted in fig .
[ figsvsh ] for the different films as a function of the applied field . in all cases
we observed a maximum value of viscosity , @xmath149 , at a field @xmath59 which is close , but always smaller , than @xmath60 ( see fig . [
fighrem])@xmath150 the distribution of viscosity values around @xmath149 has a field width at half maximum height ( fwhm ) characterized by @xmath151 which is very narrow for @xmath148 nm ( @xmath152 oe ) , increases to an average value @xmath153 oe for @xmath154 nm and increases again to @xmath155 oe for @xmath84 nm . as already discussed in section [ introduction ] , the field dependence of @xmath49 is a measure of the distribution of energy barriers ( see eq .
[ s ] ) and should correlate closely with the irreversible susceptibility obtained from the derivative of the dcd curves . in fig .
[ figsmax ] ( a ) we present the thickness dependence of the maxima in the magnetic viscosity and the irreversible susceptibility , @xmath149 and @xmath156 , obtained from figs .
[ figsvsh ] and figirm - dcd , respectively , and in the lower panel of the same figure we can observe the fwhm value of the field distribution of both magnitudes . as expected , the same overall behavior of @xmath149 and @xmath156 is found for all samples with the exception of @xmath148 nm which has been indicated with an open symbol in fig .
[ figsmax ] ( a ) . as we already mentioned
this film is at the limit in which a logarithm time decay of @xmath14 is found and , as can be seen in fig .
[ figsvsh ] , it has a very narrow field distribution which complicates the precise determination of @xmath149 .
it is then quite possible that the real value of the maximum viscosity for @xmath148 nm be considerably larger than the reported value , that should then be considered as a lower limit of @xmath157 discarding this value of viscosity , it is observed that @xmath149 decreases with film thickness , indicating that the magnetic relaxation in thinner films is faster than in thicker samples . as expected from eqs .
[ s ] and [ chi ] and observed in fig .
[ figsmax ] ( b ) the field distribution of both @xmath149 and @xmath158 has the same thickness dependence , which indicates that the distribution of activation energies @xmath56 tends to be considerably narrower for films with @xmath1 .
the sharpness of @xmath159 peaks ( _ i.e. _ smaller @xmath160 values ) has been argued@xcite to be an indication of strong exchange interactions between neighbor grains , consistent with our findings from @xmath2 curves .
the activation volume can be calculated from eq .
[ vac ] using the ratio between the maximum values of @xmath49 and @xmath159 or by averaging different values of @xmath161 in the vicinity of the coercive field . to estimate the parameter @xmath63 entering in eq .
[ vac ] we need to know the reversal mechanism present in our films .
we have measured the out of plane angular variation of the coercive field and found that @xmath60 increases when the field is applied at increasing angles with respect to the film plane , an indication that reversal is due to the displacement of domain walls . for this case
there is a criterion given by gaunt@xcite for the determination of the pinning regime .
he defined a parameter @xmath162 where @xmath163 is the maximum restoring force a pin can exert on a wall , @xmath164 is the wall energy and @xmath165 the wall width . for @xmath166
the domain walls are in the weak pinning regime while for @xmath167 the strong pinning situation occurs . a crude estimation for the pinning force
is given by @xmath168 ( @xmath169 is the radius of the pinning centers or inclusions ) and the wall energy can be written as @xmath170 so that we can write : @xmath171 in our films we have@xcite @xmath172 and an average grain size of 4 nm , which may be used as an estimation for the size of the pinning inclusions .
the wall width can be obtained@xcite from @xmath173 nm ( @xmath174 erg / cm is the exchange stiffness constant@xcite ) giving @xmath175 for the studied films , which as an indication of weak pinning .
we have then used @xmath176 in eq .
[ vac ] and plotted the values of @xmath177 as a function of film thickness in fig .
[ figvac ] .
we can observe that , within the experimental error , there are no significant differences in the two approaches used for calculating @xmath177 .
even more , the activation volume seems to be rather constant for the different samples , with an average value @xmath178 nm@xmath75 which , for spherical volumes , is equivalent to an average activation diameter @xmath179 nm . for the studied samples with @xmath180 the activation diameter @xmath181 is larger than the grain size , which implies that although the predominant interactions for @xmath180 are dipolar - like , there seems to be a positive intergranular exchange coupling which contributes to the collective reversal of volumes larger than the grain size .
this is consistent with the fact that the interaction parameter @xmath41 is positive for @xmath82 nm , the first sample for which the stripe structure is observed , and may also explain the positive ordinate in the @xmath41 vs. @xmath182 curve of fig .
note also that this value of the activation diameter is of the same order than the film thickness or the stripe half period when @xmath183 , but is much shorter than the stripe length ( which is of the order of tens of micrometers ) , implying that the magnetic volumes that reverse by thermal effects are considerably smaller than the physical volume of the stripes .
our activation diameters are larger than those reported in refs .
[ ] by a factor of two or more ( considering that the parameter @xmath63 appearing in eq .
[ vac ] was taken as @xmath184 in those previously published papers ) .
this is not surprising due to the totally different microstructure between chemically ordered and disordered samples and the larger exchange length in our magnetically soft films.@xcite . the activation volume obtained by the procedure described above
may be compared with the theoretical approach in the case of weak domain wall pinning . in this case the activation energy to overcome the barrier depends linearly@xcite on the magnetic field @xmath58@xmath185with @xmath186 the pinning field at zero temperature .
since the activation volume is related to the field derivative of the activation energy , we can write@xmath187 in the last formula we have used @xmath188 and @xmath189 with this equation it is possible to calculate the activation volume if the coercive field at @xmath190 is known .
we have discussed in ref . [ ] that at low temperatures there is an unexpected decrease in @xmath60 because interface stress effects hinder the formation of stripes , so that a reduction in @xmath60 occurs at low temperatures and a value for @xmath186 is not experimentally accessible .
however , we can still take the maximum value of @xmath191 as a lower bound estimation for @xmath192 using the data from fig .
[ fighrem ] and eq .
[ vacgaunt ] we calculated @xmath193 for the set of samples with @xmath145 nm and show the results in fig .
we can observe that the calculated values of @xmath177 are approximately independent of film thickness , with the exception of @xmath82 nm , a case that should be taken with extra care because a maximum in @xmath194 was not observed in the studied temperature range .
this form of calculating @xmath177 gave in all cases larger values than those obtained using eq .
[ vac ] , approximately by a factor of two .
the difference may be due to the underestimation of @xmath186 or to an overestimation of the wall width @xmath165 .
apart from this relatively small discrepancy , the observed experimental behavior is weakly dependent on film thickness , in accordance with the prediction of eq .
[ vacgaunt ] .
another experimental procedure for the estimation of the activation volume , which does not need the explicit measurement of @xmath159 , is the so - called `` waiting time method''.@xcite this method is based on time relaxation measurements of @xmath14 at different fields close to @xmath59 , the same curves that are used for the determination of @xmath195 the model is based on the assumption that both @xmath49 and @xmath159 are relatively constant for fields around @xmath196 when @xmath197 curves are plotted together as a function of @xmath198 it can be shown that the following relation is obeyed:@xmath199if a horizontal line of constant @xmath14 is drawn , @xmath200 represents the field distance between intersection points , and @xmath201 the time of intersection . a plot of @xmath200 as a function of @xmath202 has a slope @xmath69 from which @xmath177 can be obtained using eq .
[ hf ] . in fig .
[ fighf ] we plotted the fluctuation fields obtained from the previously calculated values of @xmath177 and added the data deduced using the waiting time method .
eventhough the error bars are relatively large , it can be seen that these new values of @xmath69 are of the same order of magnitude and relatively constant in the studied range of thicknesses , consistent with those previously estimated using the remanence and viscosity measurements .
following ref .
[ ] we have tried to correlate the values of @xmath69 with the coercivity @xmath60 .
according to wohlfarth there should be a power law relationship between both parameters , @xmath203 , with @xmath204 in the range 0.5 - 1 depending on the microstructure and the type of domain wall pinning of the system .
although our data points fall close to those shown in fig . 1 of ref .
[ ] it was not possible to fit them using a power law due to the reduced span of the coercivity and the fluctuation field values .
we have studied the role of magnetic interactions and thermally activated processes in fept alloy films as a function of film thickness .
we have found that when @xmath128 is larger than the critical thickness for the formation of a structure of stripes with an antiparallel out of plane component of the magnetization the interactions tend to be dipolar - like , while for @xmath1 positive values of @xmath41 are obtained@xmath150 this change is probably due to the larger relative weight of the dipolar field present in the films with stripe domains which arrange in a flux closure configuration that tends to favor a demagnetized state . we have found that the large differences between @xmath45 and @xmath116 are mostly due to the rotational anisotropy generated when the stripe structure is present .
the interaction parameter @xmath41 becomes more negative with increasing thickness which again is a consequence of the predominance of magnetostatic demagnetizing effects for larger values of @xmath205 we have shown that this parameter is in close correlation with the surface demagnetizing energy , confirming that dipolar interactions are predominant above the critical thickness .
magnetic viscosity was also found to depend strongly on the domain configuration . in thinner films relaxation
seems to occur in discrete steps while for @xmath180 the usual logarithm behavior is found .
@xmath96 and @xmath206 curves are a good estimation for the distribution of energy barriers and also have a strong variation in the field width depending on the domain structure .
we finally estimated the values of the activation volumes that reverse the magnetization assisted by thermal effects and found that they are approximately independent of film thickness .
the value of @xmath207 is almost an order of magnitude larger than the grain size , evidencing that a relatively large number of grains is coupled by the exchange interaction , but @xmath208 is considerably smaller than the length of the stripes ( which are several micrometers long ) , indicating that the reversal occurs in small regions compared to the size of the domains .
different methods of calculating the activation volumes and the fluctuation fields yielded approximately the same results , supporting the procedure used for the estimation of these parameters . as far as we know , this is the first time that this kind of magnetic measurements have been performed in chemically disordered fept films in which a transition in the domain structure occurs at a critical thickness . we have clearly evidenced that strong changes in most variables accompany the switch of the magnetic configuration from planar domains to parallel stripes and gave an interpretation of the observed results . this was supported in part by conicet under grant pip 112 - 200801 - 00245 , anpcyt grant pme # 1070 , and u.n .
cuyo grant 06/c235 , all from argentina .
we would like to acknowledge very fruitful discussions with dr .
emilio de biasi .
m. barturen , b. rache salles , p. schio , j. milano , a. butera , s. bustingorry , c. ramos , a. j. a. de oliveira , m. eddrief , e. lacaze , f. gendron , v. h. etgens , and m. marangolo , appl .
* 101 * , 092404 ( 2012 ) .
a. butera and j. a. barnard in high - density magnetic recording and integrated magneto - optics : materials and devices editors : k. rubin , j.a .
bain , t. nolan , d. bogy , b.j.h .
stadler , m. levy , j.p .
lorenzo , m. mansuripur , y.okamura , r. wolfe , materials research society ( mrs ) symposium proceedings ( san fransisco , ca , spring meeting 1998 ) vol . : 517 , p. 349 | we have investigated the relationship between the domain structure and the magnetic interactions in a series of fept ferromagnetic thin films of varying thickness .
as - made films grow in the magnetically soft and chemically disordered a1 phase that may have two distinct domain structures . above a critical thickness @xmath0
nm the presence of an out of plane anisotropy induces the formation of stripes , while for @xmath1 planar domains occur .
magnetic interactions have been characterized using the well known dcd - irm remanence protocols , @xmath2 plots , and magnetic viscosity measurements . we have observed a strong correlation between the domain configuration and the sign of the magnetic interactions .
planar domains are associated with positive exchange - like interactions , while stripe domains have a strong negative dipolar - like contribution . in this last case
we have found a close correlation between the interaction parameter and the surface dipolar energy of the stripe domain structure . using time dependent magnetic viscosity measurements ,
we have also estimated an average activation volume for magnetic reversal , @xmath3 nm@xmath4 which is approximately independent of the film thickness or the stripe period . |
we study a reaction - diffusion equation with a nonlocal reaction term : @xmath0 this problem , introduced by bnichou , calvez , meunier , and voituriez @xcite , models a population structured by a space variable @xmath1 and a motility trait @xmath2 .
our analysis is focused on the case where the trait space @xmath3 is a bounded subset of @xmath4 .
the parameters @xmath5 and @xmath6 are positive and represent , respectively , the rate of mutation and the net reproduction rate in the absence of competition .
the problem ( [ n ] ) is of fisher - kpp type . the classical fisher - kpp equation
, @xmath7 describes the growth and spread of a population structured by a space variable @xmath1 .
the diffusion coefficient @xmath8 is a positive constant .
the behavior of solutions to ( [ kpp ] ) has been widely studied , starting from its introduction in @xcite . the most important difference between (
[ n ] ) and the fisher - kpp equation is that the reaction term in ( [ n ] ) is nonlocal in the trait variable .
this is because competition for resources , which is represented by the reaction term , occurs between individuals of _ all _ traits that are present in a certain location .
another key feature of ( [ n ] ) is that the trait @xmath9 affects how fast an individual moves this is why the coefficient of the spacial diffusion in ( [ n ] ) depends on @xmath9 .
in addition , the trait is subject to mutation , which is modeled by the diffusion term in @xmath9 .
thus , ( [ n ] ) describes the interaction between dispersion of a population and the evolution of the motility trait .
we further discuss the biological motivation for ( [ n ] ) and review the relevant literature in more detail later on in the introduction . throughout our paper , we assume that the trait space @xmath3 is a bounded interval @xmath10 where @xmath11 and @xmath12 are positive constants .
we study classical solutions of ( [ n ] ) with initial condition that is non - negative and regular enough . " we state these assumptions precisely as ( a[asump c2 ] ) , ( a[asm n0 ] ) and ( a[asm theta ] ) in section [ sec : asm ] .
a significant challenge for us is that ( [ n ] ) does not enjoy the maximum principle .
this is due to the presence of the nonlocal reaction term .
nevertheless , we are able to establish a global upper bound for solutions of ( [ n ] ) .
we prove : [ sup bd intro ] suppose @xmath13 is nonnegative , twice differentiable on @xmath14 and satisfies ( [ n ] ) in the classical sense , with initial condition @xmath15 that satisfies ( a[asm n0 ] ) .
there exists a constant @xmath16 such that @xmath17 theorem [ sup bd intro ] is a key element in the proof of our second main result .
previous work on the fisher - kpp equation @xcite and formal computations concerning ( [ n ] ) @xcite suggest that solutions to ( [ n ] ) should converge to a travelling front in @xmath1 and @xmath18 . in order to study the motion of the front
, we perform the rescaling @xmath19 .
this rescaling leads us to consider solutions @xmath20 of the @xmath21-dependent problem , @xmath22 ( we point out that ( [ nep ] ) is _ not _ the rescaled version of the @xmath21-independent problem ( [ n ] ) , as we consider initial data @xmath23 , instead of @xmath24 , in ( [ nep ] ) .
please see remark [ remark : relationship ] for more about this . )
we study the limit of the @xmath20 as @xmath25 .
we find that there exists a set on which the sequence @xmath20 converges locally uniformly to zero , and another set on which a certain limit of @xmath26 stays strictly positive .
these sets are determined by two viscosity solutions of the hamilton - jacobi equation , @xmath27 the function @xmath28 arises from the eigenvalue problem ( [ spectral ] ) and is determined by @xmath29 , @xmath11 , @xmath6 , and @xmath5 ( see proposition [ prop : spectral ] ) .
one may view the function @xmath30 as encoding the effect of the motility trait on the limiting behavior of the @xmath20 .
our main result , theorem ( [ result on n ] ) , says that the hamilton - jacobi equation ( [ hj ] ) describes the motion of an interface which separates areas with and without individuals .
viscosity solutions of ( [ hj ] ) with infinite initial data play a key role in our analysis and we provide a short appendix where we discuss the relevant known results . for the purposes of the introduction
, we state the following lemma : [ lem intro ] for any @xmath31 , there exists a unique continuous function @xmath32 that is a viscosity solution of ( [ hj ] ) in @xmath33 and satisfies the initial condition @xmath34 in addition , we have @xmath35 for all @xmath36 and @xmath37 . we are interested in @xmath38 for two sets @xmath39 determined by the initial data @xmath15 .
we define these two sets , @xmath40 and @xmath41 , by @xmath42 we see that @xmath1 belongs to @xmath40 if initially there is at least _ some _ individual living at @xmath40 , and @xmath1 belongs to @xmath41 if individuals with _ all _ traits are present at @xmath1 .
our main result says that the limiting behavior of the @xmath20 is determined by @xmath43 and @xmath44 : [ result on n ] assume ( a[asump c2 ] ) , ( a[asm n0 ] ) and ( a[asm theta ] ) .
let @xmath43 and @xmath44 be the functions given by lemma [ lem intro ] .
then , @xmath45 and @xmath46 let us remark on a special case of theorem [ result on n ] .
suppose the initial data @xmath15 is such that the two sets @xmath40 and @xmath41 are equal ( this occurs if , for example , @xmath15 is independent of @xmath9 ) .
in this case , @xmath47 and so @xmath48 , which means that theorem [ result on n ] gives information about the limiting behavior of @xmath20 _ almost everywhere _ on @xmath49 , for all times @xmath18 .
we present the following corollary : [ corc*informal ] assume ( a[asump c2 ] ) , ( a[asm n0 ] ) and ( a[asm theta ] ) .
let us also suppose that each of @xmath40 and @xmath41 is a bounded interval
. there exists a positive constant @xmath50 , which depends only on @xmath5 , @xmath6 , @xmath29 and @xmath11 , such that * if @xmath51 , then @xmath52 for all @xmath2 ; and , * if @xmath53 , then @xmath54 .
the proof of corollary [ corc*informal ] is in subsection [ subsect : pf of cors ] .
it uses the work of majda and souganidis @xcite concerning a class of equations of the form ( [ hj ] ) but with more general hamiltonians .
[ remarkbc ] corollary [ corc*informal ] directly connects our result to the main result of bouin and calvez @xcite .
indeed , theorem 3 of @xcite says that there exists a travelling wave solution of ( [ n ] ) of speed @xmath50 , for the same @xmath50 as in corollary [ corc*informal ] . while it is not known whether solutions of ( [ n ] ) converge to a travelling wave , corollary [ corc*informal ] is a result in this direction it says that , in the limit as @xmath25 , the regions where the @xmath20 is positive and zero travel with speed @xmath50 .
the biological question is , as time goes on , which territory will be occupied by the species and which will be left empty ? to answer this question , it is enough to determine where the functions @xmath43 and @xmath44 are zero . in fact , corollary [ corc*informal ] gives information about the limit of the @xmath20 simply in terms of the sets @xmath40 , @xmath41 , and a constant @xmath50 .
indeed , we see that at time @xmath18 , if we stand a point that is far " from @xmath40 , then there are no individuals at @xmath1 . on the other hand , if we stand at a point @xmath1 that is pretty close " to @xmath41 , then there are some individuals living near @xmath1 .
we formulate another corollary : [ cor : onn ] assume ( a[asump c2 ] ) , ( a[asm n0 ] ) , ( a[asm theta ] ) and that @xmath40 and @xmath41 are bounded intervals . then : * if @xmath55 , then @xmath52 for all @xmath2 ; and , * if @xmath56 , then @xmath54 .
we remind the reader of the fact , due to aronson and weinberger @xcite , that @xmath57 is the asymptotic speed of propagation of fronts for ( [ kpp ] ) .
thus , a consequence of corollary [ cor : onn ] is that , in the limit , the population we re considering spreads slower than one with constant motility @xmath11 and faster than one with constant motility @xmath29 .
we give the proof of corollary [ cor : onn ] in subsection [ subsect : pf of cors ] .
in addition , please see remark [ rem : interpret ] for further comments on the biological implications of our results .
biologists are interested in the interplay between traits present in a species and how the species interacts with its environment in other words , between evolution and ecology @xcite .
it has been observed , for example in butterflies in britain @xcite , that an expansion of the territory that a species occupies may coincide with changes in a certain trait the butterflies that spread to new territory were able to lay eggs on a larger variety of plants than the butterflies in previous generations .
the phenotypical trait in this case is related to adaptation to a fragmented habitat .
some biologists have focused specifically on the interaction between ecology and traits that affect motility .
phillips et al @xcite recently discovered a species of cane toads whose territory has , over the past 70 years , spread with a speed that _ increases _ in time .
this is very interesting because this is contrary to what is predicted by the fisher - kpp equation @xcite and has previously been observed empirically @xcite .
spacial sorting was also observed the toads that arrive first in the new areas have longer legs than those in the areas that have been occupied for a long time .
in addition , it was discovered that toads with longer legs are able to travel further than toads with shorter legs .
it is hypothesised that the presence of this trait length of legs is responsible for both the front acceleration and the spacial sorting .
similar phenomena were observed in crickets in britain over a shorter time period @xcite . in that case , the motility trait was wingspan .
the cases we describe demonstrate the need to understand the influence of a trait in particular , a motility trait on the dynamics of a population .
the fisher - kpp equation has been extensively studied , and we refer the reader to @xcite for an introduction .
hamilton - jacobi equations similar to ( [ hj ] ) are known to arise in the analysis of the long - time and long - range behavior of ( [ kpp ] ) , other reaction - diffusion pde , and systems of such equations
see , for example , friedlin @xcite , evans and souganidis @xcite and barles , evans and souganidis @xcite and fleming and souganidis @xcite .
the methods of @xcite are a key part of our analysis of ( [ nep ] ) .
as we previously mentioned , ( [ kpp ] ) describes populations structured by space alone , while there is a need to study the interaction of dispersion and phenotypical traits ( in particular , motility traits ) .
most models of populations structured by space and trait either consider a trait that does not affect motility or do not consider the effect of mutations .
champagnat and mlard @xcite start with an individual - based model of such a population and derive a pde that describes its dynamics .
in the case that the trait affects only the growth rate and not the motility , alfaro , coville and raoul @xcite study this pde , which is a reaction - diffusion equation with constant diffusion coefficient : @xmath58 the population modeled by ( [ acr ] ) has a preferred trait that varies in space .
berestycki , jin and silvestre @xcite analyze an equation similar to ( [ acr ] ) , but with a different kernel @xmath41 and growth term @xmath6 that represent the existence of a trait that is favorable for all individuals .
the aims and methods of @xcite are quite different from those in this paper .
the main result of @xcite is the existence of traveling wave solutions of ( [ acr ] ) for speeds above a critical threshold . in @xcite
, the authors establish the existence and uniqueness of travelling wave solutions and prove an asymptotic speed of propagation result for the equation that they consider .
we also mention that a local version of ( [ acr ] ) was investigated by berestycki and chapuisat @xcite .
desvillettes , ferriere and prvost @xcite and arnold , desvillettes and prvost @xcite study a model in which the dispersal rate does depend on the trait and the trait is subject to mutation , but the mutations are represented by a nonlocal linear term , not a diffusive term .
there has also been analysis of traveling waves and steady states for equations of the form @xmath59 where the reaction term @xmath60 is the convolution of @xmath61 with some kernel @xmath62 .
we refer the reader to berestycki , nadin , perthame and ryzhik @xcite , hamel and ryzhik @xcite , fang and zhao @xcite , alfaro and coville @xcite and the references therein .
an important difference between ( [ kernel ] ) and ( [ n ] ) is that the reaction term of ( [ n ] ) is local in the space variable and nonlocal in the trait variable , while the reaction term in ( [ kernel ] ) is fully nonlocal .
the long time behavior of solutions to ( [ kernel ] ) is studied in @xcite .
in addition , ( * ? ? ? * theorem 1.2 ) establishes a supremum bound for solutions of ( [ kernel ] ) .
bouin and mirrahimi @xcite analyze the reaction - diffusion equation @xmath63 with neumann conditions on @xmath61 on the boundary of @xmath3 .
the main difference between ( [ eqnbm ] ) and ( [ n ] ) is that the coefficient of spacial diffusion in ( [ eqnbm ] ) is constant , which means that ( [ eqnbm ] ) models a population where the trait does not affect motility .
the methods we use here are similar to those of @xcite ( and , in turn , both ours and those of @xcite are similar to those used in @xcite to study ( [ kpp ] ) ) .
however , in general it is easier to obtain certain bounds for solutions of ( [ eqnbm ] ) than for solutions of ( [ n ] ) .
for example , because the coefficient of @xmath64 in ( [ eqnbm ] ) is constant , integrating ( [ eqnbm ] ) in @xmath9 implies that @xmath65 is a subsolution of a local equation in @xmath1 and @xmath18 that enjoys the maximum principle .
this immediately implies that @xmath65 is globally bounded ( * ? ? ?
* lemma 2 ) .
this strategy does not work for ( [ n ] ) .
indeed , a serious challenge in studying ( [ n ] ) , as opposed to ( [ kpp ] ) or nonlocal reaction diffusion equations with constant diffusion coefficient such as ( [ eqnbm ] ) , is obtaining a global supremum bound for solutions of ( [ n ] ) .
another challenge that arises in our situation but not in @xcite is in establishing certain gradient estimates in @xmath9 ( see remark [ rem : gradbdbm ] ) . in addition
, we compare our main result , theorem [ result on n ] , with that of @xcite in remark [ remrho ] .
let us discuss the literature that directly concerns ( [ n ] ) and ( [ nep ] ) .
the problem ( [ n ] ) was introduced in @xcite .
the rescaling leading to ( [ nep ] ) was suggested by bouin , calvez , meunier , mirrahimi , perthame , raoul , and voituriez in @xcite .
in addition , formal results about the asymptotic behavior of solutions to ( [ n ] ) were obtained in @xcite . in particular , part ( i ) of proposition [ main result ] of our paper was predicted in ( * ? ? ?
* section 2 ) .
( we briefly remark on the term motility . "
it was used heavily in @xcite , which is where we first learned of the problem ( [ n ] ) . however , one of the referees of our paper pointed out that this term applies mainly to unicellular organisms . )
bouin and calvez @xcite also study ( [ n ] ) .
they prove that there exist traveling wave solutions to ( [ n ] ) but do not analyze whether solutions converge to a traveling wave .
in fact , to our knowledge , there are no previous rigorous results about the asymptotic behavior of solutions of ( [ n ] ) or the limiting behavior of solutions to ( [ nep ] ) .
the main difficulty is the lack of comparison principle for ( [ n ] )
. please see corollary [ corc*informal ] , remark [ remarkbc ] , as well as the fourth point below , for further discussion of the connections between the results of our paper and those of @xcite .
* to the best of our knowledge , theorem [ sup bd intro ] is the first global supremum bound for a fisher - kpp type equation with a nonlocal reaction term and non - constant diffusion . *
theorem [ result on n ] completes the program that was proposed in @xcite for analyzing the asymptotic behavior of the model ( [ n ] ) in the case where the trait space @xmath3 is bounded .
* we view our main result , theorem [ result on n ] , as evidence that the presence of a motility trait does affect the limiting behavior of populations .
* corollary [ corc*informal ] provides a direct connection between our work and the main result of @xcite . indeed , ( * ? ? ?
* theorem 3 ) states that there exist travelling wave solutions to ( [ n ] ) of a certain speed , while corollary [ corc*informal ] shows that this exact speed characterizes the limiting behavior of the @xmath20 .
* we hope that our work is a step towards analyzing ( [ n ] ) in the case where the trait space @xmath3 is unbounded .
it is in this case that the phenomena of accelerating fronts is predicted to occur @xcite .
the proof of theorem [ sup bd intro ] is quite involved .
the difficulty comes from the combination of the nonlocal reaction term and non - constant diffusion .
our proof of theorem [ sup bd intro ] uses regularity estimates for solutions of elliptic pde , a heat kernel estimate , and an averaging technique similar to that of ( * ? ? ?
* theorem 1.2 ) .
we believe this combination of methods is new and may be useful in other contexts .
we include a detailed outline in subsection [ subsec : outline ] .
to analyze the limit of the @xmath20 , we preform the transformation @xmath66 such a transformation is used in @xcite .
we prove locally uniform estimates on @xmath67 : [ bd on utheta ] assume ( a[asump c2 ] ) , ( a[asm n0 ] ) and ( a[asm theta ] ) and let @xmath67 be given by ( [ def : up ] ) .
suppose @xmath68 is compactly contained in @xmath33 .
there exists a constant @xmath16 that depends on @xmath68 , @xmath5 , @xmath6 , @xmath29 and @xmath11 such that for all @xmath69 and for all @xmath70 such that @xmath71 and @xmath72 , we have @xmath73 and @xmath74 we define the half - relaxed limits @xmath75 and @xmath76 of @xmath67 by , @xmath77 the supremum estimates of proposition [ bd on utheta ] imply that @xmath76 and @xmath75 are finite everywhere on @xmath78 .
moreover , the gradient estimate of proposition [ bd on utheta ] implies @xmath67 becomes independent of @xmath9 as @xmath21 approaches zero .
thus , it is natural that @xmath75 and @xmath76 should be independent of @xmath9 .
there is also a connection to homogenization theory we can think of @xmath9 as the fast " variable , which disappears in the limit .
it is the hamilton - jacobi equation ( [ hj ] ) , which arises in the @xmath25 limit , that captures the effect of the fast " variable .
we use a perturbed test function argument ( evans @xcite ) and techniques similar to the proofs of ( * ? ? ?
* theorem 1.1 ) , ( * ? ? ?
* propositions 3.1 and 3.2 ) , and ( * ? ? ?
* proposition 1 ) to establish : [ main result ] assume ( a[asump c2 ] ) , ( a[asm n0 ] ) and ( a[asm theta ] ) and let @xmath75 and @xmath76 be given by ( [ halfrelaxed ] ) .
then : @xmath75 is a viscosity subsolution and @xmath76 is a viscosity supersolution of ( [ hj ] ) in @xmath79 ; and we have @xmath80 and @xmath81 part ( i ) of proposition [ main result ] was predicted via formal arguments in ( * ? ? ?
* section 2 ) .
theorem [ result on n ] follows easily from proposition [ main result ] by arguments similar to those in the proofs of ( * ? ? ?
* theorem 1.1 ) and ( * ? ? ? * theorem 1 ) .
[ remrho ] an interesting question is whether theorem [ result on n ] can be refined to obtain better information about the limit of the @xmath82 in the interior of the set @xmath83 .
for instance , is @xmath84 bounded from below in this set ?
when the diffusion coefficient is constant , which is the situation studied in @xcite , the answer is yes . indeed , ( * ? ? ?
* theorem 1 ) provides a lower bound on @xmath85 , where @xmath86 is a rescaling of the solution @xmath61 of ( [ eqnbm ] ) .
this lower bound of ( * ? ? ?
* theorem 1 ) is obtained using an argument that relies on the diffusion coefficient of @xmath64 being constant , and thus does not work in our case . in the next subsection
, we state our assumptions , give notation , and provide the definition of @xmath30 .
the rather lengthy section [ sec : sup bd ] is devoted to the proof of theorem [ sup bd intro ] .
this section is self - contained . in section [ sec : bdutheta ]
we prove proposition [ bd on utheta ] .
the proof of proposition [ main result ] is in section [ sec : limitsuep ] .
the proof of theorem [ result on n ] is given in section [ sec : result on n ] , and the proofs of corollaries [ corc*informal ] and [ cor : onn ] are in subsection [ subsect : pf of cors ] .
we also provide an appendix with a discussion of results on existence , uniqueness , and comparison for hamilton - jacobi equations with infinite initial data .
we have organized our paper so that a reader who is interested mainly in our proof of theorem [ sup bd intro ] may only read section [ sec : sup bd ] . on the other hand ,
a reader who is interested in our results about the limit of the @xmath20 , and not in the proofs of the supremum bound on @xmath20 and the estimates on @xmath67 , may skip ahead to sections [ sec : limitsuep ] and [ sec : result on n ] after finishing the introduction .
our results hold under the following assumptions : 1 .
[ asump c2 ] @xmath20 is a non - negative classical solution of ( [ nep ] ) .
[ asm n0 ] @xmath87 for some @xmath88 , where @xmath89 $ ] is defined by ( [ a ] ) .
[ asm theta ] @xmath90 , where @xmath91 .
the assumption ( a[asm n0 ] ) implies that @xmath23 is contained in @xmath92 and satisfies @xmath93 for all @xmath36 .
we use condition ( a[asm n0 ] ) in our proof of theorem [ sup bd intro ] . [ remark : relationship ] we remark on the relationship between ( [ n ] ) and ( [ nep ] ) .
our results in theorem [ result on n ] concern the limiting behavior of @xmath20 , and we emphasize that we do not make any precise statements concerning the asymptotic behavior of solutions to ( [ n ] ) . as we mentioned ealier ,
this is due to our assumption that the initial data of ( [ nep ] ) is independent of @xmath21 .
indeed , let us suppose @xmath13 satisfies ( [ n ] ) and _ define _ @xmath94 .
we find that @xmath20 satisfies ( [ nep ] ) , but with initial data @xmath95 .
thus , our assumption that the initial data of ( [ nep ] ) does not depend on @xmath21 is essentially saying that we study the problem ( [ n ] ) with initial data that is invariant under the rescaling @xmath96 .
one example of such data is @xmath97}(x)f(\theta)$ ] , where @xmath98}$ ] is the indicator function of @xmath99 $ ] .
however , this example does not satisfy the regularity assumption ( a[asm n0 ] ) .
we leave for future work the possibility of establishing a supremum bound on @xmath13 that does not require such regular initial data .
let us also explain why we consider the problem ( [ nep ] ) with initial data independent of @xmath21 .
a key element of our study is the transformation @xmath100 .
we know that @xmath101 , but it is not necessarily true that @xmath102 .
it is therefore possible that @xmath67 , @xmath76 , or @xmath103 take on the value @xmath104 " . in order for our analysis to make sense ,
we need to eliminate this possibility for times @xmath105 . to do this , we can make one of two assumptions regarding the initial data @xmath106 .
one option is to assume that @xmath107 holds everywhere ( this is essentially the assumption made in ( * ? ? ?
* line ( 1.5 ) ) ) .
another option is to assume that the initial data does not depend on @xmath21 .
this is the assumption made in @xcite , as well as in this paper .
we find that @xmath67 is bounded from below on compact sets ( see proposition [ bd on utheta ] ) .
such a bound is proven by a barrier argument similar to those in @xcite , but this argument does not work when the initial data depends on @xmath21 .
we will slightly abuse notation in the following way .
if @xmath68 is a subset of @xmath33 , then we will use @xmath108 to denote the set of @xmath109 such that @xmath71 and @xmath2 .
we record this as : @xmath110 next we state the spectral problem ( [ spectral ] ) of ( * ? ? ?
* proposition 5 ) , which describes the speed @xmath50 and allows us to define the hamiltonian @xmath30 .
[ prop : spectralbc ] for all @xmath111 , there exists a unique solution @xmath112 of the spectral problem @xmath113 the map @xmath114 is continuous for @xmath111 , and satisfies , for all @xmath111 , @xmath115 in addition , the function @xmath114 achieves its minimum , which we denote @xmath50 , at some @xmath116 . in the next proposition
, we define the hamiltonian @xmath30 and list the properties of @xmath30 that we will use .
[ prop : spectral ] we define the hamiltonian @xmath28 by @xmath117 for all @xmath118 , there exists a unique solution @xmath119 of the spectral problem @xmath120 moreover , we have that the map @xmath121 is continuous , convex , and satisfies , for all @xmath122 , @xmath123 we have , for all @xmath118 , @xmath124 where @xmath50 is as in proposition [ prop : spectralbc ] .
finally , we have @xmath125 first , let us verify that @xmath30 is continuous .
since the bound ( [ bdoncbc ] ) holds for all @xmath111 , we have @xmath126 and in particular this limit is finite , so that @xmath121 is continuous on all of @xmath49 .
we also note that ( [ bdonc ] ) does indeed hold for all @xmath122 .
we remark that @xmath30 is the even reflection of the map @xmath127 . for @xmath128
, we define @xmath129 to be @xmath130 , where the latter is the solution of ( [ spectralbc ] ) corresponding to @xmath131 .
if @xmath132 , we see that @xmath133 solves ( [ spectral ] ) .
hence , ( [ spectral ] ) holds for all @xmath122 . in addition , according to the proof of proposition [ spectralbc ] , the map @xmath134 is given by , @xmath135 where @xmath136 is the eigenvalue of a certain eigenvalue problem with parameter @xmath122 . from this characterization , one can check that @xmath136 is convex in @xmath122 .
hence @xmath134 is also convex in @xmath122 .
the even reflection of a convex function is not in general convex , due to a possible issue at 0 .
however , the fact that @xmath30 satisfies ( [ bdonc ] ) allows us to deduce that @xmath30 is indeed convex .
finally , we verify ( [ hc * ] ) .
according to the definition of @xmath30 we have , @xmath137 according to proposition [ prop : spectralbc ] , @xmath138 achieves its minimum , @xmath50 , at some @xmath116 .
thus , the infimum in the previous line is achieved at @xmath139 .
hence we have established ( [ hc * ] ) . to verify ( [ boundsonc * ] ) , we use ( [ bdoncbc ] ) to find
, @xmath140 hence the minimum value of @xmath141 must be between the minimum values of the functions on the left - hand side and the right - hand side .
these are exactly @xmath142 and @xmath143 , respectively .
this establishes ( [ boundsonc * ] ) and completes the proof of the proposition .
[ rem : interpret ] we give a further ( slightly informal ) interpretation of theorem [ result on n ] .
according to ( * ? ? ?
* theorem 1.1 ) , the behavior of solutions to ( [ kpp ] ) is characterized by a solution @xmath61 of the hamilton - jacobi equation @xmath144 ( we remark that this is the negative of the equation that appears in @xcite ; we write it this way to be consistent with the signs employed in the rest of this paper . )
let us compare ( [ forkpp ] ) and ( [ hj ] ) .
we see that both are of the form @xmath145 , but for different hamiltonians @xmath30 .
it is this difference that captures the effect of the trait . indeed , according to proposition [ prop : spectral ] , the term @xmath146 satisfies ( [ bdonc ] ) .
thus , we see that the term @xmath147 in ( [ hj ] ) is like a quadratic " , but of different size than the quadratic in ( [ forkpp ] ) . in addition , according to ( [ candgamma ] ) and the definition of @xmath30 , we have @xmath148 where @xmath149 is _ positive _ for all @xmath122 .
hence the hamiltonian in ( [ hj ] ) is never exactly a quadratic , and hence must be different from that of ( [ forkpp ] ) .
since ( [ hj ] ) and ( [ forkpp ] ) characterize the behavior of populations with and without a motility trait , respectively , we interpret our results as evidence that the presence of the trait does affect the asymptotic behavior of a population .
this section is devoted to the proof of : [ prop : supbd ] suppose @xmath13 is nonnegative , twice differentiable on @xmath14 and satisfies ( [ n ] ) in the classical sense .
assume that @xmath3 satisfies ( a[asm theta ] ) and the initial condition @xmath15 satisfies ( a[asm n0 ] ) .
there exists a constant @xmath16 that depends only on @xmath29 , @xmath11 , @xmath150 , @xmath6 and @xmath151 such that @xmath152 we briefly explain notation for norms and seminorms , which we will be using only in this section . for @xmath153
we denote : @xmath154_{\eta , u } } } = \sup_{(x , t ) , ( y , s)\in u}\frac{|u(x , t)-u(y , s)|}{(|x - y|+|s - t|^{1/2})^{\eta } } , \ \ \
{ \ensuremath{|u|_{\eta , u}}}= { \ensuremath{||u||_{l^{\infty}(u)}}}+{\ensuremath{[u]_{\eta , u}}},\ ] ] and @xmath155 we present the supremum bound for solutions of ( [ nep ] ) : [ supbd : cor ] assume ( a[asump c2 ] ) , ( a[asm n0 ] ) and ( a[asm theta ] ) .
there exists a constant @xmath16 that depends only on @xmath29 , @xmath11 , @xmath150 , @xmath6 , @xmath151 , and @xmath156 such that for all @xmath69 , @xmath157 let us explain how corollary [ supbd : cor ] follows from theorem [ prop : supbd ] .
let us fix some @xmath69 and suppose @xmath20 satisfies ( [ nep ] ) in the classical sense and @xmath15 satisfies ( a[asm n0 ] ) .
let us define @xmath158
. then @xmath13 satisfies ( [ n ] ) with initial data @xmath159 .
we have that @xmath160 satisfies ( a[asm n0 ] ) .
therefore , according to theorem [ prop : supbd ] , we have that the estimate ( [ supbdconclusion ] ) holds , with @xmath160 instead of @xmath15 on the right - hand side .
since we have @xmath161 and @xmath162 , we obtain the conclusion of corollary [ supbd : cor ] .
we introduce the following piecewise function @xmath89 $ ] . for any @xmath163 let @xmath164 be such that @xmath165 .
we define @xmath166 by @xmath167 we use @xmath168 to extend @xmath169 to a function @xmath170 defined for all @xmath163 .
this is equivalent to extending @xmath169 by even reflection across @xmath171 ( since @xmath13 satisfies neumann boundary conditions , this gives a periodic function on @xmath172 with derivative zero on the boundary ) and then by periodicity to the rest of @xmath49
. please see figure [ fig : aandn ] . if we use @xmath168 to extend an _
twice differentiable function on @xmath3 with derivative zero at the boundary to all of @xmath49 , the result would still have continuous first derivative but would not necessarily be twice differentiable in @xmath9 .
however , it turns out that since @xmath13 solves the equation ( [ n ] ) , the extension @xmath173 is in @xmath174 .
moreover , the hlder norm of the second derivatives of @xmath175 is bounded in terms of the supremum of @xmath13 .
we state this precisely in proposition [ deriv bd in terms of m ] in subsection [ subsec : holder ] . instead of working directly with @xmath175 ,
we work with its averages in @xmath9 on small intervals : we define the function @xmath61 by @xmath176 our idea to use these averages came from reading the proof of ( * ? ? ?
* theorem 1.2 ) .
we have that @xmath61 is twice differentiable on @xmath177 and satisfies @xmath178 for all @xmath36 , @xmath179 , and @xmath37 . the first step in our proof of the supremum bound on @xmath13
is to obtain a bound on @xmath180 for some @xmath181 .
we then use the bound on @xmath61 to establish the supremum bound on @xmath13 itself . in other words , we use an @xmath182 bound for @xmath13 to obtain an @xmath183 bound . for this
, we formulate proposition [ prop : bdfroml1 ] in subsection [ subsec : l1linf ] .
the proof of proposition [ prop : bdfroml1 ] uses certain estimates for the heat kernel associated to the operator @xmath184 .
these estimates say that this kernel is like the kernel for the heat equation and were established by aronson @xcite
. we will use the following result on the solvability in @xmath174 of parabolic equations : [ classical ] let us take @xmath185 and suppose the coefficients @xmath186 are uniformly elliptic , @xmath187 , @xmath188 , and @xmath189 .
then the initial value problem @xmath190 has a unique solution @xmath191
. moreover , there exists a constant @xmath16 that depends on @xmath151 , the ellipticity constant of @xmath186 , and @xmath192 such that @xmath193}}}\leq c({\ensuremath{|f|_{\eta , { \ensuremath{\mathbb{r}}}^2\times [ 0,t ] } } } + { \ensuremath{|u|_{\eta,{\ensuremath{\mathbb{r}}}^2\times[0,t]}}}+{\ensuremath{|u_0|_{2+\eta , { \ensuremath{\mathbb{r}}}^2}}}).\ ] ] that there exists a unique solution @xmath194 is part of the statement of ladyzenskaja et al ( * ? ? ?
* chapter iv theorem 5.1 ) .
the estimate ( [ estimateclassprop ] ) follows from krylov ( * ? ? ?
* theorem 9.2.2 ) , which asserts that @xmath195 obeys the estimate , @xmath196}}}\leq c_1({\ensuremath{| a^{ij}u_{x_ix_j } - u_t - u|_{\eta , { \ensuremath{\mathbb{r}}}^2\times [ 0,t ] } } } + { \ensuremath{|u_0|_{2+\eta , { \ensuremath{\mathbb{r}}}^2}}}),\ ] ] where the constant @xmath197 depends only on @xmath151 , the ellipticity constant of @xmath186 and @xmath192 .
we use the equation that @xmath195 satisfies to bound from above the first term on the right - hand side of ( [ estpreclass ] ) and find , @xmath198 } } } = { \ensuremath{| f - u|_{\eta , { \ensuremath{\mathbb{r}}}^2\times [ 0,t]}}}\leq { \ensuremath{|f|_{\eta , { \ensuremath{\mathbb{r}}}^2\times [ 0,t ] } } } + { \ensuremath{|u|_{\eta,{\ensuremath{\mathbb{r}}}^2\times[0,t]}}}.\ ] ] using the previous line to bound from above the first term on the right - hand of ( [ estpreclass ] ) yields the estimate ( [ estimateclassprop ] ) .
in addition , we need the following lemma : [ lem : unique ] assume @xmath3 satisfies ( a[asm theta ] ) . given @xmath199 and continuous and non - negative functions @xmath200 and @xmath201 , the solution @xmath195 of @xmath202 is unique .
we point out that here @xmath203 is a _ given _ function and the equation ( [ eqn : ulem ] ) for @xmath195 is _
local_. because of this , the proof of the lemma is standard and we omit it .
we are now ready to state and prove the hlder estimate for @xmath13 : [ deriv bd in terms of m ] assume ( a[asump c2 ] ) , ( a[asm n0 ] ) , and ( a[asm theta ] ) .
we define the extension @xmath204 by @xmath205 1 .
[ item : bneqn ] we have that @xmath175 is twice differentiable on @xmath206 and satisfies @xmath207 2 .
[ item : bdbn ] let @xmath199 and assume @xmath208 } n = m\geq 1 $ ] .
there exists a positive constant @xmath209 that depends on @xmath151 , @xmath150 , @xmath6 @xmath29 and @xmath11 so that @xmath210 } } } \leq\tilde { c } ( m^{2+\eta/2}+{\ensuremath{|n_0(x , a(\theta))|_{2+\eta , { \ensuremath{\mathbb{r}}}\times{\ensuremath{\mathbb{r}}}}}}).\ ] ] we apply proposition [ classical ] with right - hand side @xmath211 , initial condition @xmath212 , and the matrix of diffusion coefficients being the diagonal matrix with entries @xmath166 and @xmath5 .
the assumption ( a[asm n0 ] ) on @xmath15 says exactly that the assumption of proposition [ classical ] on the initial condition is satisfied . by proposition [ classical ]
, there exists a unique solution @xmath213 of @xmath214 and we have the estimate @xmath215}}}\leq c({\ensuremath{|r \bar{n}(1-\rho)|_{\eta , { \ensuremath{\mathbb{r}}}^2\times [ 0,t ] } } } + { \ensuremath{|{\ensuremath{\tilde{u}}}|_{\eta , { \ensuremath{\mathbb{r}}}^2\times [ 0,t]}}}+{\ensuremath{|{\ensuremath{\tilde{u}}}_0|_{2+\eta , { \ensuremath{\mathbb{r}}}^2}}}),\ ] ] where @xmath16 depends on @xmath29 , @xmath11 and @xmath5 .
let us remark that the maps @xmath216 and @xmath217 satisfy @xmath218 together with the fact that the solution @xmath219 is _ unique _ , this implies @xmath219 has this symmetry as well .
thus , @xmath220 for all @xmath1 and for all @xmath105 .
therefore , @xmath221 with @xmath2 satisfies @xmath222 since @xmath13 also satisfies this equation , lemma [ lem : unique ] implies @xmath223 for all @xmath36 , all @xmath2 , and all @xmath224 $ ] .
therefore , @xmath225 .
in particular , we have that item ( [ item : bneqn ] ) of the proposition holds .
in addition , the estimate ( [ estimateclass ] ) holds for @xmath175 and reads , @xmath226 } } } \leq c ( { \ensuremath{|{\ensuremath{\bar{n}}}(1-\rho)|_{\eta , { \ensuremath{\mathbb{r}}}\times{\ensuremath{\mathbb{r}}}\times [ 0,t]}}}+{\ensuremath{|{\ensuremath{\bar{n}}}|_{\eta,{\ensuremath{\mathbb{r}}}\times{\ensuremath{\mathbb{r}}}\times[0,t]}}}+{\ensuremath{|n_0(x , a(\theta))|_{2+\eta , { \ensuremath{\mathbb{r}}}\times{\ensuremath{\mathbb{r}}}}}}),\ ] ] where the constant @xmath16 depends on @xmath151 , @xmath5 , @xmath6 , @xmath29 and @xmath11 . we will now establish item ( [ item : bdbn ] ) of the proposition .
let us recall that we assume @xmath227 } n = m\ ] ] and @xmath228 .
we have @xmath208 } n = \sup_{{\ensuremath{\mathbb{r}}}\times{\ensuremath{\mathbb{r}}}\times [ 0,t ] } { \ensuremath{\bar{n}}}$ ] . for the remainder of the proof of this proposition , we drop writing the domains in the semi - norms and norms ( it is always @xmath229 $ ] ) .
we will now bound the right - hand side of ( [ firstneta ] ) from above . to this end , we first recall the definition of @xmath230 : @xmath231_{\eta}}}+||{\ensuremath{\bar{n}}}(1-\rho)||_{\infty}.\end{aligned}\ ] ] for functions @xmath232 and @xmath233 we have the elementary estimate @xmath234_{\eta}\leq ||f||_{\infty}[g]_{\eta}+||g||_{\infty}[f]_{\eta}.\ ] ] we apply this with @xmath235 and @xmath236 to obtain a bound from above on the first term of the right - hand side of ( [ usedefnorm ] ) : @xmath237_{\eta}}}||(1-\rho)||_{\infty}+{\ensuremath{[1-\rho]_{\eta}}}||{\ensuremath{\bar{n}}}||_{\infty}+ ||{\ensuremath{\bar{n}}}(1-\rho)||_{\infty}.\end{aligned}\ ] ] in addition , the definition of @xmath203 implies @xmath238_{\eta } } } \leq ( \theta_m-\theta_m ) { \ensuremath{[n]_{\eta}}}$ ] and @xmath239 .
we use this to estimate the right - hand side of the previous line and find @xmath240_{\eta}}}+m+m^2.\end{aligned}\ ] ] similarly we estimate the second term on the right - hand side of ( [ firstneta ] ) : @xmath241_{\eta}}}+||{\ensuremath{\bar{n}}}||_{\infty}\leq { \ensuremath{[{\ensuremath{\bar{n}}}]_{\eta}}}+m.\ ] ] we use ( [ estsemifirstterm ] ) and the previous line to bound from above the first and second terms , respectively , on the right - hand side of ( [ firstneta ] ) and obtain , @xmath242_{\eta}}}+m+m^2)+{\ensuremath{[{\ensuremath{\bar{n}}}]_{\eta } } } + m+{\ensuremath{|n_0(x , a(\theta))|_{2+\eta } } } ) .\ ] ] we now use @xmath228 and obtain , @xmath243_{\eta } } } + m^2+{\ensuremath{|n_0(x , a(\theta))|_{2+\eta}}}).\ ] ] by interpolation estimates for the seminorms @xmath244 $ ] ( for example , ( * ? ? ?
* theorem 8.8.1 ) ) , we have , for any @xmath245 , @xmath246_{\eta}}}\leq c({\ensuremath{\varepsilon}}{\ensuremath{|{\ensuremath{\bar{n}}}|_{2+\eta } } } + { \ensuremath{\varepsilon}}^{-\eta/2}m).\ ] ] we use the estimate ( [ semin ] ) in the right - hand side of the bound ( [ n2eta ] ) for @xmath247 and obtain @xmath248 choosing @xmath249 , we obtain @xmath250 and @xmath251 . rearranging the above
we thus obtain , @xmath252 where the last inequality follows since @xmath228 . we will employ certain @xmath183 estimates on the heat kernel , which we state in the following lemma .
[ lem : kernel ] there exists a kernel @xmath41 such that if @xmath253 is non - negative and @xmath213 is a weak solution of @xmath254 then @xmath255 moreover , there exist constants @xmath256 and @xmath257 that depend only on @xmath5 , @xmath29 and @xmath11 such that @xmath258 for all @xmath1 , @xmath259 , @xmath9,@xmath260 in @xmath49 , and for all @xmath105 .
we apply two theorems of aronson @xcite .
that we have @xmath261 is the content of ( * ? ? ?
* theorem 11 ) . the bound on @xmath41
is exactly ( * ? ? ?
* item ( ii ) , theorem 10 ) .
aronson s results apply to any parabolic equation @xmath262 in @xmath263 with bounded coefficients and with @xmath264 regular enough . in our case , @xmath265 and @xmath266
( so , in particular , it satisfies aronson s hypotheses ) and @xmath267 is the diagonal matrix with entries @xmath166 and @xmath5 . in the general case ,
the constants @xmath256 and @xmath257 depend on the ellipticity constant of @xmath268 and the @xmath183 norm of @xmath268 .
since these depend only on @xmath5 , @xmath29 and @xmath11 , we have that @xmath256 and @xmath257 depend only on @xmath5 , @xmath29 and @xmath11 as well . for the proof of proposition [ prop : bdfroml1 ]
we also need the following lemma .
its proof is elementary and we omit it . [ lem : elem ] for @xmath269 we have @xmath270 .
we now state and prove the main proposition of this subsection .
[ prop : bdfroml1 ] suppose @xmath219 is a non - negative classical solution of @xmath271 where @xmath272 satisfies , for some @xmath273 and for all @xmath36 and @xmath179 , @xmath274 there exists a constant @xmath275 that depends only on @xmath150 , @xmath29 and @xmath11 such that , for all @xmath1 and @xmath9 , @xmath276 according to lemma [ lem : kernel ] , we have , for all @xmath36 , all @xmath163 , and all @xmath105 , @xmath277 the second inequality follows from the upper bound on the kernel @xmath41 given by lemma [ lem : kernel ] , where we write @xmath278 instead of @xmath257 to simplify notation .
let us take @xmath279 to obtain the following bound for @xmath280 : @xmath281 we will now split up the integral in @xmath260 into a sum of integrals over intervals of size @xmath282 .
we have @xmath283 for @xmath284 we have @xmath285 we use this to bound each of the integrals in @xmath260 and find , @xmath286 for each @xmath287 , we can take @xmath288 in assumption ( [ asm bu0 ] ) of this proposition to obtain , for all @xmath287 , @xmath289 therefore , @xmath290 where the last inequality follows from lemma [ lem : elem ] applied with @xmath291 .
we use @xmath292 ( and so @xmath293 ) to bound the right - hand side of the previous line from above and obtain , @xmath294 we now use this bound on the integral in @xmath260 in the estimate ( [ eq : bdbu1 ] ) for @xmath280 and obtain @xmath295 where @xmath16 depends only on @xmath29 , @xmath11 and @xmath5 .
this holds for all @xmath1 and @xmath9 , so the proof is complete .
we have now established the two auxillary results that we need , and are ready to proceed with : let @xmath275 be the constant from proposition [ prop : bdfroml1 ] and let @xmath209 be the constant from proposition [ deriv bd in terms of m ] .
define the constants @xmath296 and @xmath297 by @xmath298 and @xmath299 we claim @xmath300 we proceed by contradiction : let us assume @xmath301 .
let us now fix a number @xmath302 with @xmath303 and @xmath304 throughout the rest of the proof of this proposition , @xmath16 denotes a positive constant that may change from line to line and depends only on @xmath29 , @xmath11 , @xmath6 , @xmath150 and @xmath151 ( in particular , @xmath16 does not depend on @xmath302 ) .
let us consider the map @xmath305 that takes @xmath18 to the supremum of @xmath13 at time @xmath18 , in other words : @xmath306 the map @xmath307 is continuous .
in addition , since @xmath13 satisfies ( [ n ] ) , and @xmath13 and @xmath203 are non - negative , we have that @xmath308 is a subsolution of @xmath309 the equation ( [ eqn : ufromzero ] ) satisfies the comparison principle .
therefore , we have the following bound on @xmath310 from above : for all @xmath36 and @xmath2 , @xmath311 taking supremum in @xmath1 and @xmath9 and multiplying by @xmath312 gives the bound @xmath313 for all @xmath105 .
in particular , taking supremum over @xmath314 $ ] , we have @xmath315 } s(t)\leq e\sup_{x\in { \ensuremath{\mathbb{r}}},\theta\in \theta } n_0(x,\theta)\leq m_0<m,\ ] ] where the second inequality follows from the definition of @xmath297 . line ( [ bdbn ] ) implies @xmath316 .
since @xmath307 is continuous and @xmath317 for @xmath318 , there exists a first time @xmath319 for which @xmath320 .
so , we have @xmath227 } n = m \text { and } \sup_{{\ensuremath{\mathbb{r}}}\times\theta } n(\cdot , \cdot , t ) = m.\ ] ] we will now work with the extension @xmath321 defined in proposition [ deriv bd in terms of m ] . by the previous line , we have @xmath322 } { \ensuremath{\bar{n}}}=m \text { and } \sup_{{\ensuremath{\mathbb{r}}}\times{\ensuremath{\mathbb{r } } } } { \ensuremath{\bar{n}}}(\cdot , \cdot , t ) = m.\ ] ] we apply proposition [ deriv bd in terms of m ] to @xmath175 .
part [ item : bneqn ] implies that @xmath175 satisfies equation ( [ eqnbn ] ) .
part ( [ item : bdbn ] ) gives us the estimate @xmath210 } } } \leq \tilde{c } ( m^{2+\eta/2}+{\ensuremath{|n_0(x , a(\theta))|_{2+\eta , { \ensuremath{\mathbb{r}}}\times{\ensuremath{\mathbb{r}}}}}}).\ ] ] since we have @xmath228 , the second term on the right - hand side of the previous line is smaller than @xmath323 so we find , @xmath210 } } } \leq \tilde{c}(m^{2+\eta/2}+{\ensuremath{|n_0(x , a(\theta))|_{2+\eta , { \ensuremath{\mathbb{r}}}\times{\ensuremath{\mathbb{r}}}}}}m^{2+\eta/2}).\ ] ] we use our choice of @xmath296 in ( [ barc ] ) to bound the right - hand side from the previous line from above and obtain , @xmath324 } } } \leq \bar{c}m^{2+\eta/2}.\ ] ] let us take @xmath325 and define @xmath326 by @xmath176 * first step : * we will prove @xmath327 since @xmath13 satisfies ( [ n ] ) , we have that @xmath61 satisfies @xmath328 let us explain why we may assume , without loss of generality , that the supremum of @xmath61 on @xmath329 is achieved at some @xmath330 . since @xmath331 is periodic of period @xmath332 , and we are considering times @xmath18 in the bounded interval @xmath333 , we know that there exist @xmath334 , @xmath335 $ ] and a sequence @xmath336 with @xmath337 as @xmath338 . for each @xmath339 , we define the translated functions @xmath340 @xmath341 and @xmath342 we similarly define the translates of the first and second derivatives of @xmath61 .
we have @xmath343 , and according to ( [ seminm ] ) , @xmath344}}}\leq \bar{c}m^{2+\eta/2}.\ ] ] therefore , @xmath345 , @xmath346 , @xmath347 , and the translates of the first and second derivatives of @xmath61 are uniformly bounded and uniformly equicontinuous on @xmath348 $ ] .
hence , there exists a subsequence ( still denoted by @xmath339 ) and functions @xmath349 , @xmath350 , and @xmath351 such that @xmath346 , @xmath352 and @xmath347 converge locally uniformly to @xmath349 , @xmath350 and @xmath351 , respectively ; the derivatives of @xmath347 converge locally uniformly to those of @xmath351 ; and @xmath353 .
moreover , @xmath354 satisfies @xmath355 on @xmath356 and we have , for all @xmath36 , all @xmath357 , and all @xmath358 , @xmath359 in other words , @xmath351 achieves its supremum on @xmath360 .
we now drop the superscript @xmath361 . at the point @xmath330 where @xmath61 achieves its supremum , we have @xmath362 and @xmath363 we point out that @xmath364 does not appear in ( [ eqnv ] ) , the equation that @xmath61 satisfies .
we will bound from above the corresponding term that does appear in ( [ eqnv ] ) .
this term is : @xmath365 the inequality ( [ vxx ] ) implies that there exists @xmath366 with @xmath367 .
in addition , let @xmath368 $ ] be so that @xmath369 } a(\theta ) = a(\tilde{\theta}).\ ] ] since @xmath168 is positive , we multiply ( [ vxx ] ) by @xmath370 and find @xmath371 adding the term ( [ term ] ) that we re interested in to both sides of this inequality , we find @xmath372 let us recall that @xmath373 is non - negative on @xmath374 $ ] .
we thus use the estimate ( [ seminm ] ) on the seminorm of @xmath375 and the fact that @xmath367 to estimate the right - hand side of the previous line from above and obtain @xmath376 since @xmath366 , we have @xmath377 . in addition , @xmath168 is lipschitz with lipschitz constant @xmath378 , so we have @xmath379 .
we use these two inequalities to bound the right - hand side of the previous line from above and find @xmath380 the last inequality follows by an elementary calculus computation that relies on the fact that @xmath381 is contained in @xmath374 $ ] . using this estimate together with the information ( [ derivsv ] ) about the other derivatives of @xmath61 at @xmath382 in the equation ( [ eqnv ] ) that @xmath61 satisfies ,
we obtain @xmath383 since @xmath384 we may bound @xmath385 from below by @xmath386 : @xmath387 we use the previous estimate to bound the right - hand side of ( [ vatmax ] ) from above and obtain @xmath388 upon rearranging we find , @xmath389 by our choice of @xmath282 , we have @xmath390 .
we use this to bound the right - hand side of the previous line and find @xmath391 so @xmath392 since @xmath61 achieved its supremum on @xmath360 at @xmath382 , we conclude @xmath393 * second step : * we will now deduce a supremum bound on @xmath175 from the bound on @xmath61 .
let us fix any @xmath394 .
let @xmath195 be the solution of @xmath395 we have that @xmath396 is a subsolution of the equation for @xmath195 for @xmath397 .
since they are equal at @xmath398 , the comparison principle for the equation for @xmath195 implies the bound @xmath399 for all @xmath400 , and @xmath397 .
in particular , we evaluate the above at @xmath279 and take supremum in @xmath1 and @xmath9 to find @xmath401 we will now apply proposition [ prop : bdfroml1 ] to @xmath195 .
the supremum bound ( [ bdv ] ) on @xmath61 says exactly that assumption ( [ asm bu0 ] ) is satisfied , with @xmath402 .
therefore , proposition [ prop : bdfroml1 ] implies @xmath403 therefore , we may bound @xmath195 by @xmath404 on the right - hand side of ( [ bdnbyu ] ) and find , @xmath405 this holds for any @xmath406 , so in particular at @xmath407 .
according to line ( [ supn ] ) , we have @xmath408 , so we obtain @xmath409 recall that we chose @xmath410 . if @xmath411 , then we find @xmath412 ; if @xmath413 , then @xmath414 ; and if @xmath415 we obtain @xmath416 which implies , since @xmath417 , the following bound for @xmath302 : @xmath418 therefore , @xmath419 but we had taken @xmath303 . recalling our choice of @xmath297 in line ( [ choicem0 ] ) yields the desired contradiction and hence the proof is complete .
[ remark : choice of nonlinearity ] we remark on other possible choices of the nonlinear term @xmath420 , specifically as related to the proof of theorem [ prop : supbd ] .
( for the purposes of this remark , we use @xmath16 to denote any constant that does not depend on @xmath302 . )
the exponent @xmath421 in the right - hand side of the estimate of item ( [ item : bdbn ] ) of proposition [ deriv bd in terms of m ] is very important .
let us suppose the estimate read , @xmath210 } } } \leq c ( m^{p}+1),\ ] ] for some @xmath422 .
the key difference in step one of the proof of theorem [ prop : supbd ] is that line ( [ explanationsigma ] ) would have the term @xmath423 instead of @xmath424 .
thus , we would need to choose @xmath425 in order to proceed with the estimate on @xmath61 .
then , in line ( [ explanationmbound ] ) in the second step of the proof , we would obtain @xmath426 hence , to deduce a bound on @xmath302 we need @xmath427 to satisfy @xmath428 .
the value of @xmath427 in the estimate of item ( [ item : bdbn ] ) of proposition [ deriv bd in terms of m ] is determined by the stucture of the nonlinear term @xmath420 .
for example , analyzing the proof of proposition [ deriv bd in terms of m ] yields that if the nonlinear term were @xmath429 , for some @xmath430 , the conclusion would hold with @xmath431 . since our method requires @xmath428 , it would work for @xmath432 ( and , indeed , @xmath433 satisfies this ) .
we leave further considerations regarding different choices of nonlinearities for future work .
in this section we give the proof of proposition [ bd on utheta ] .
we believe that the arguments in this section are the most technical ones of the paper .
let us point out that @xmath434 satisfies @xmath435 in subsection [ subsec : lowbd ] we establish the upper and lower bounds on @xmath67 .
the upper bound follows directly from corollary [ supbd : cor ] , while the lower bound needs a barrier argument that is quite similar to ( * ? ? ?
* lemma 2.1 ) and ( * ? ? ?
* lemma 1.2 item ( i ) ) .
we skip a few details in the proof because it is quite similar to the cited ones . in subsection
[ subsec : grad ] we prove the gradient estimate by an application of the so - called bernstein method " ( see ( * ? ? ?
* lemma 2.1 ) and ( * ? ? ?
* lemma 2.2 ) ) .
this argument is quite lengthy and has a some rather technical parts .
we provide all the details . in order to construct the barrier necessary for the proof of the lower bound
, we need a lemma .
the proof of the lemma is an elementary computation which we include for the sake of completeness .
[ lem : barrier ] given @xmath436 we define the function @xmath437 for @xmath438 by , @xmath439 there exists a constant @xmath16 that depends only on @xmath440 such that for all @xmath441 and for all @xmath69 , @xmath442 we remark that @xmath443 is negative .
computing the derivatives of @xmath444 , substituting them into @xmath445 and rearranging yields , for @xmath438 , @xmath446 we have @xmath447 , so that the term in the inner - most parentheses on the right - hand side of the previous line is bounded from above by @xmath448 .
consequently we find , for @xmath438 , @xmath449 we now consider @xmath443 such that @xmath450 , in which case we have @xmath451
. therefore , @xmath452 where we have again used that @xmath453 .
hence the term in the parenthesis on the right - hand side of ( [ psieqn ] ) is bounded from above by @xmath454 .
therefore , the right - hand side of ( [ psieqn ] ) is negative for @xmath450 .
thus it is left to consider @xmath443 such that @xmath455 .
but in this case , we see @xmath456 . using this ,
together with the fact that the term in the parenthesis on the right - hand side of ( [ psieqn ] ) is bounded from above by 1 , and find , @xmath457 since @xmath458 is non - positive for @xmath438 , we have , for @xmath69 and @xmath438 , @xmath459 so that , @xmath460 this completes the proof of the lemma . the proof of the lower bound proceeds in two steps .
first we fix a cube @xmath461 on which @xmath462 is strictly positive .
thus , @xmath463 is bounded from below on this cube .
we use this , together with the barrier @xmath444 , to obtain a lower bound on @xmath464 for @xmath465 in a smaller cube @xmath466 and for times @xmath467 $ ] .
the second step is to use this lower bound to build a barrier for @xmath67 on the remainder of @xmath468 $ ] .
this barrier function is infinite at @xmath469 , hence the bound we obtain does not reach the initial time @xmath470 . due to corollary [ supbd : cor ] ,
there exists a positive constant @xmath302 that depends only on @xmath29 , @xmath11 , @xmath5 and @xmath6 such that @xmath67 is a supersolution of @xmath471 * step one : * let us fix some @xmath472 and @xmath473 with @xmath474 , and @xmath475 small so that @xmath476 for all @xmath465 in the cube @xmath477 . without loss of generality we may take @xmath436 . since @xmath15 is continuous and positive in @xmath478 , there exists a constant @xmath479 such that @xmath480 .
therefore , if @xmath465 is contained in @xmath478 , we have @xmath481 let @xmath16 be the constant given by lemma [ lem : barrier ] , and define @xmath482 .
we define the function @xmath62 by , @xmath483 using lemma [ lem : barrier ] , it is easy to check that @xmath62 is a subsolution of ( [ eqn : uepsuper ] ) in @xmath484 .
moreover , @xmath62 lies below @xmath67 on the parabolic boundary of @xmath485 .
indeed , for points @xmath486 , we have @xmath487 . and ,
if @xmath465 is contained in @xmath488 , then @xmath489 for all @xmath105
. therefore , the comparison principle for equation ( [ eqn : uepsuper ] ) implies @xmath490 let us fix some @xmath199 .
by ( [ ugeqphi ] ) and the definition of @xmath62 , we have , @xmath491.\ ] ] * step two : * we will now use this lower bound to build a barrier for @xmath67 on @xmath492 .
we define the function @xmath493 by , @xmath494 we see that for @xmath495 and @xmath278 properly chosen and depending only on @xmath496 , @xmath440 , @xmath29 , @xmath11 , @xmath5 and @xmath6 , we have that @xmath493 is a subsolution of ( [ eqn : uepsuper ] ) on @xmath497
. moreover , we have @xmath498 for all @xmath70 .
in particular , if @xmath499 and @xmath467 $ ] , then , according to ( [ deftau ] ) , we have @xmath500 since @xmath501 and @xmath502 for all @xmath465 , a comparison principle argument similar to that of ( * ? ? ?
* lemma 2.1 ) implies @xmath503 together with ( [ ugeqphi ] ) , the previous estimate implies that if @xmath504 , then there exists a constant @xmath16 that depends on @xmath68 , @xmath5 , @xmath6 , @xmath29 and @xmath11 such that for all @xmath69 and for all @xmath505 and all @xmath506 , we have @xmath507 according to corollary [ supbd : cor ] , the estimate @xmath508 holds for a constant @xmath16 depending only on @xmath29 , @xmath11 , @xmath5 and @xmath6 .
the upper bound on @xmath67 follows since , by definition we have @xmath509 .
the proof of the gradient bound proceeds via a bernstein argument in other words , we use that the derivatives of @xmath195 also satisfy certain pdes in order to obtain estimates on them .
we are most interested in obtaining an estimate on the derivative in @xmath9 . to this end , let us denote @xmath510 and differentiate the equation ( [ eqn : uep ] ) for @xmath67 in @xmath9 to find that @xmath443 satisfies , @xmath511 we bring the reader s attention to the last term , which involves a _
second _ derivative of @xmath67 in the space variable @xmath1 .
this means that we can not ignore " the space variable and must try to estimate @xmath512 and @xmath513 at the same time . in particular , the essense of our strategy is to consider the pde satisfied not by @xmath443 but by @xmath514 , and obtain estimates on this quantity .
our proof is similar to that of ( * ? ? ?
* lemma 2.4 ) and ( * ? ? ?
* lemma 2.1 ) , which use a bernstein argument to obtain gradient bounds for hamilton - jacobi equations with variable coefficients . however , our situation is more delicate because of the different scaling in the space and trait variable .
[ rem : gradbdbm ] the term @xmath515 in ( [ eq : demo ] ) only arises because the diffusion coefficient is not constant in @xmath9 .
let us compare our method to that of the proof of lemma 2 of @xcite .
we recall that the authors of @xcite consider the pde ( [ eqnbm ] ) , which , after a rescaling and an exponential transformation , yields an equation similar to our equation ( [ eqn : uep ] ) for @xmath67 , but with constant diffusion coefficient .
thus , they are able to carry out the strategy we first describe essentially , they differentiate their pde in @xmath9 and , because extraneous second derivative terms do not appear , they are able to use this equation to obtain the desired estimate .
most of this subsection is devoted to the proof of : [ prop : gradbd ] fix @xmath69 .
we denote @xmath465 by @xmath259 .
suppose @xmath516 satisfies @xmath517 where the coefficients @xmath268 and @xmath264 are given by @xmath518 and @xmath232 is independent of @xmath519 and satisfies @xmath520 and @xmath521 given @xmath522 , there exists a constant @xmath16 that depends only on @xmath68 , @xmath523 , @xmath524 , @xmath29 , @xmath11 , @xmath5 , @xmath6 and @xmath302 such that @xmath525 and @xmath526 we remark that we only use the estimate ( [ eq : lemuy2 ] ) in the remainder of the paper .
nevertheless , our proof automatically " yields estimate ( [ eq : lemuy1 ] ) as well , so we state it for the sake of completeness .
the gradient bound of proposition [ bd on utheta ] follows from proposition [ prop : gradbd ] once we verify its hypotheses : we apply proposition [ prop : gradbd ] to @xmath67 with @xmath527 . according to the supremum bound of proposition [ bd on utheta ] , which we have just established ,
we have that @xmath67 is bounded on @xmath528 , uniformly in @xmath21 . according to corollary [ supbd : cor ] we have that @xmath20 , and hence @xmath82 ,
is uniformly bounded from above .
in addition , @xmath82 is non - negative .
hence the hypothesis ( [ linff ] ) is satisfied .
now let us demonstrate that ( [ fyasump ] ) holds : since we have @xmath527 and @xmath529 , this amounts to establishing @xmath530 for some constant @xmath302 .
let us define @xmath531 . since @xmath20 satisfies ( [ nep ] ) and @xmath15 satisfies ( a[asm n0 ] ) , we have that @xmath13 satisfies ( [ n ] ) with initial data that satisfies ( a[asm n0 ] ) .
thus , according to proposition [ deriv bd in terms of m ] and theorem [ prop : supbd ] , @xmath13 is bounded in @xmath174
. therefore , there exists a constant @xmath16 so that , for all @xmath70 , @xmath532 integrating in @xmath9 yields the desired estimate for @xmath533 . for the proof of proposition [ prop : gradbd ] we will need the following elementary facts .
we summarize them as a lemma and omit the proof .
[ matrixlem ] for any diagonal @xmath534 positive definite matrix @xmath302 , any @xmath534 matrices @xmath535 , @xmath536 , and any @xmath537 , we have latexmath:[\[\label{matrixineqlem } in addition , if @xmath535 is symmetric then @xmath539 .
let @xmath540 be a cutoff function supported on @xmath523 and identically @xmath378 inside @xmath68 .
let us define @xmath541 and @xmath542 we record for future use the following facts : @xmath543 @xmath544 where we denote @xmath545 . throughout the remainder of the argument we will use @xmath16 to denote constants that may depend on @xmath5 , @xmath6 , @xmath546 , @xmath29 , @xmath11 , @xmath68 and @xmath302 and that may change from line to line .
we also define the constant @xmath122 by @xmath547 we remark that @xmath122 is of the form , @xmath548 we define the function @xmath549 and we claim @xmath550 * why ( [ locmaxz ] ) implies the estimates ( [ eq : lemuy2 ] ) and ( [ eq : lemuy1 ] ) .
* let us assume ( [ locmaxz ] ) holds , and we will establish ( [ eq : lemuy2 ] ) . * first step : * we will first prove the following upper bound on @xmath551 holds : @xmath552 to this end , we use ( [ defz ] ) and the fact that @xmath553 is the maximum of @xmath443 to obtain , @xmath554 next we consider two cases : the first is @xmath555 , and the second is @xmath556 .
in the first case , we have @xmath557 , so we find @xmath558 .
therefore , the estimate ( [ thzy ] ) says , @xmath559 since we have @xmath560 , this establishes the estimate ( [ thatmax ] ) in the case @xmath555 .
now let us suppose @xmath556 , so that , according to ( [ locmaxz ] ) , we have @xmath561 we now use ( [ defz ] ) to express the right - hand side of ( [ thzy ] ) in terms of @xmath551 and @xmath195 : @xmath562 we use that @xmath563 and the estimate ( [ hy0t0 ] ) to bound the first term on the right - hand side of the previous line of the above by @xmath16 , and that @xmath564 to bound the last two terms .
this yields ( [ thatmax ] ) .
* second step : * next we will use ( [ thatmax ] ) to verify the following upper bound on the derivatives of @xmath195 in terms of @xmath16 and @xmath122 : @xmath565 to this end , we use the definition of @xmath551 and the fact that @xmath260 is identically @xmath378 inside @xmath68 to obtain , @xmath566 we use the estimate ( [ thatmax ] ) to bound the first term on the right - hand side of the previous line from above by @xmath567 and obtain the estimate ( [ eq : lemalmost ] ) . *
third step : * there are two possible values for @xmath122 , and we will show that , together with the estimate ( [ eq : lemalmost ] ) that we just established , either one implies the desired estimate ( [ eq : lemuy2 ] ) . let us first suppose @xmath568 . upon substituting this on the right - hand side of ( [ eq : lemalmost ] ) we find ,
@xmath569 which yields the desired estimates ( [ eq : lemuy2 ] ) and ( [ eq : lemuy1 ] ) by multiplying both sides by @xmath570 .
now let us suppose that @xmath122 takes on the other possible value , so that @xmath571 .
we use that @xmath572 , the definition of @xmath551 , the line ( [ thatmax ] ) and the value of @xmath122 to obtain , for any @xmath573 , @xmath574 from this we deduce @xmath575 and hence we have @xmath576 ( and , since @xmath577 on @xmath68 , ( [ eq : lemuy1 ] ) holds ) .
using this on the right - hand side of ( [ eq : lemalmost ] ) implies that ( [ eq : lemuy2 ] ) holds . *
the proof of ( [ locmaxz ] ) .
* let us suppose @xmath553 is the maximum of @xmath443 on @xmath528 and @xmath556 .
there are two cases to consider :
the first is that @xmath553 is in the interior of @xmath578 , and the second is that @xmath579 .
let us tackle the first case , so that we have , @xmath580 we seek to establish @xmath581 as we have just shown , once we establish ( [ wanth ] ) the proof of the proposition will be complete . to this end , we compute , @xmath582 where @xmath583 is the sum of the left - over terms from @xmath584 and @xmath585 : @xmath586.\ ] ] ( throughout , we use @xmath587 to denote the derivative in @xmath259 . )
we use that @xmath195 satisfies ( [ eqnu ] ) to write the first term on the right - hand side of ( [ eqforz ] ) as @xmath588 where we have used ( [ linff ] ) to obtain the inequality .
now let us look at the second term on the right - hand side of ( [ eqforz ] ) .
we recognize that @xmath589 is almost " the derivative of @xmath590 , up to a term that involves a derivative of @xmath268 .
in addition , rearranging the equation that @xmath195 solves implies @xmath591 .
we find : @xmath592 multiplying by @xmath260 we obtain , @xmath593 so far ,
our computations hold on all of @xmath528 .
now we will specialize to @xmath553 and obtain an alternate expression for the first term on the right - hand side of the previous line .
we recall the derivative of @xmath443 is zero at @xmath553 , so that , @xmath594 which , upon rearranging becomes , @xmath595 we substitute the right - hand side of the previous line for the first term of the right - hand side of ( [ secondterm ] ) to obtain , at @xmath553 , @xmath596 we take dot product with @xmath597 and use the previous line to find : @xmath598 where @xmath599 is the sum of the left - over terms : @xmath600 next , according to ( [ hpdu ] ) , we have @xmath601 .
we use this on the right - hand side of ( [ secondtermuse ] ) and find , @xmath602 let us now consider ( [ eqforz ] ) evaluated at @xmath553 . according to ( [ atmaxz ] ) , the left - hand side of ( [ eqforz ] ) is non - negative
. we use ( [ firstterm ] ) to estimate the first term on the right - hand side of ( [ eqforz ] ) , and we use ( [ secondtermagain ] ) for the second term , and find , @xmath603 we now claim that the sum of the leftover terms @xmath583 and @xmath599 is bounded : @xmath604 we point out that once the bound ( [ esterror ] ) is established , the proof of ( [ wanth ] ) , and hence of the proposition , will be complete . indeed , using ( [ esterror ] ) to estimate the right - hand side of ( [ handdph ] ) yields @xmath605 which , upon rearranging and dividing by @xmath111 yields ( [ wanth ] ) .
* proof of bound ( [ esterror ] ) on error terms .
* let us start with @xmath583 .
we will prove , @xmath606 + \frac{\lambda}{4 } g(du , y).\ ] ] we use the expressions ( [ thpp ] ) for @xmath607 and @xmath608 to rewrite @xmath583 as @xmath609 - \frac{4\psi}{{\ensuremath{\varepsilon}}}\operatorname{tr}\left[a^2\cdot du\otimes d\psi d^2u \right ] -4\zeta\operatorname{tr}\left [ a w d^2u \right ] -\operatorname{tr}\left[a{\ensuremath{\tilde{g}}}_{yy}\right ] .
\end{split}\ ] ] let us bound from above the third term in @xmath583 by applying the inequality of lemma [ matrixlem ] with @xmath610 , @xmath611 , @xmath612 and @xmath613 .
we obtain : @xmath614\leq \frac{8}{{\ensuremath{\varepsilon } } } \operatorname{tr}(a^2 ( du\otimes d\psi)(du\otimes d\psi)^t)+\frac{\psi^2}{2{\ensuremath{\varepsilon}}}\operatorname{tr}(a^2 d^2u d^2u).\ ] ] let us apply lemma [ matrixlem ] again , this time with @xmath615 , @xmath616 , @xmath617 and @xmath613 .
we obtain the following upper bound for the fourth term in @xmath583 : @xmath618\leq 8 \zeta \operatorname{tr}\left(ww^t\right)+\frac{\zeta}{2}\operatorname{tr}(a^2 d^2u d^2u ) = 8\zeta u_1 ^ 2 + \frac{\zeta}{2}\operatorname{tr}(a^2 d^2u d^2u),\ ] ] where we have also used the second statement in lemma [ matrixlem ] .
we use ( [ estwithd2u ] ) and the previous line to bound from above the third and fourth terms , respectively , in on the right - hand side of ( [ traceinstar ] ) . notice that the terms involving @xmath619 in the previous line and in ( [ estwithd2u ] ) will be absorbed " by the second term in @xmath583 ( we are using that @xmath620 ) .
we obtain : @xmath621 + \frac{8}{{\ensuremath{\varepsilon}}}\operatorname{tr}(a^2 ( du\otimes d\psi)(du\otimes d\psi)^t ) + 8\zeta u_{y_1}^2 -\operatorname{tr}\left[a{\ensuremath{\tilde{g}}}_{yy}(du , y)\right].\ ] ] the first term on the right - hand side of ( [ bdforrhstracealmost ] ) is simply @xmath622 , which is less than @xmath623 .
in addition , we have @xmath624 .
we use this to bound the right - hand side of ( [ bdforrhstracealmost ] ) from above and find , @xmath625 + \frac{8}{{\ensuremath{\varepsilon}}}\operatorname{tr}(a^2 ( du\otimes d\psi)^2 ) -\operatorname{tr}\left[a{\ensuremath{\tilde{g}}}_{yy}(du , y)\right].\ ] ] we will show now show that each of the last two terms on the right - hand side of the previous line is less than @xmath626 .
once we show this , the estimate ( [ estonstar ] ) will be established . to this end
, we use that @xmath493 is independent of @xmath519 , to compute @xmath627 multiplying by @xmath628 and using the definitions of @xmath264 and @xmath122 we obtain , @xmath629 let us estimate the last term of ( [ bdforrhstrace ] ) .
we compute @xmath630 in terms of the derivatives of @xmath264 and @xmath260 and find , @xmath631 we multiply both sides of the previous line by @xmath632 and take trace .
the first term gives zero .
the second is simply @xmath633 .
we summarize this as , @xmath634 thus we have proved ( [ estonstar ] ) .
now for @xmath599 .
we have @xmath635 since since @xmath636 , the first term in @xmath599 is simply @xmath637 for the second term in @xmath599 , we use that @xmath638 , the cauchy - schwarz inequality , the assumption ( [ fyasump ] ) , and the definitions of @xmath264 and @xmath122 to find , @xmath639 finally we will bound the last term in @xmath599 .
we have @xmath640 , so the last term is simply @xmath641 where the first inequality follows by applying cauchy schwarz , the second from the definitions of @xmath268 and @xmath264 , and the third from our choice of @xmath122 .
we therefore find , @xmath642 adding our upper bound ( [ estonstar ] ) on @xmath583 to the previous line we obtain , @xmath643 as desired .
thus our analysis of the case that @xmath553 is in the interior of @xmath578 is complete .
_ completing the proof of ( [ locmaxz ] ) : analysis of the second case .
_ we now tackle the second case , and suppose @xmath579 .
we do not have that ( [ atmaxz ] ) holds ; let us first determine the replacement " for ( [ atmaxz ] ) in this case .
since the @xmath644 and @xmath18 coordinates of the maximum @xmath553 of @xmath443 are interior , we have @xmath645 , @xmath646 , and @xmath647 at @xmath553 .
we now need to consider @xmath648 and @xmath649 . to this end , we use the definitions of @xmath443 and @xmath264 to compute , @xmath650 since @xmath651 in @xmath652 , we have @xmath653 on @xmath652 as well . hence , on @xmath654 , we find @xmath655 taking another derivative of @xmath443 in @xmath519 and using that @xmath653 on @xmath652 , we obtain @xmath656 putting everything together , we find that instead of ( [ atmaxz ] ) we have , @xmath657 we recall that we used ( [ atmaxz ] ) in two places .
the first was to find that the left - hand side of ( [ eqforz ] ) is non - negative at @xmath553 , which allowed us to obtain the lower bound of @xmath470 for the left - hand side of ( [ handdph ] ) .
we see that , since ( [ atmaxzsecond ] ) holds , we still have this lower bound . the second place where we used ( [ atmaxz ] ) was line ( [ dz0 ] ) , to say @xmath658 .
now , we instead have @xmath659 .
let us proceed as in the proof of the first case ( we will see that , because we will soon take dot product with a vector of the form @xmath660 , this extra term will not affect the argument ) . line ( [ dz0 ] ) no longer has @xmath470 on the left - hand side but becomes , @xmath661 we rearrange and find , @xmath662 the next step is to substitute the right - hand side of the previous line for the first term of the right - hand side of ( [ secondterm ] ) . doing this we obtain , at @xmath553 , @xmath663 we take dot product with @xmath597 .
however , the key here is that @xmath597 is simpler than before : we use the definition of @xmath264 to compute , @xmath664 where the second inequality follows since @xmath665 , and @xmath651 on @xmath652 .
hence , the term @xmath666 disappears upon taking dot product with @xmath597 , and therefore ( [ secondtermuse ] ) remains unchanged .
thus the remainder of the argument goes through as in the first case , and the proof of proposition [ prop : gradbd ] is complete .
this section is devoted to studying the half - relaxed limits @xmath75 and @xmath76 of the @xmath67 that we mentioned in the introduction . for the convenience of the reader , we recall their definitions here
: @xmath667 and @xmath668 we also summarize the results of the previous sections .
we have established corollary [ supbd : cor ] , which says that there exists a constant @xmath16 so that @xmath669 .
we have also proved proposition [ bd on utheta ] , which says that if @xmath670 , then there exists a constant @xmath16 that depends on @xmath68 such that for @xmath71 and @xmath2 , we have @xmath671 and @xmath74 in particular , these results imply that @xmath75 and @xmath76 are finite everywhere on @xmath33 ( although they may be infinite at time @xmath398 , as we will demonstrate in the proof of proposition [ main result ] ) .
in addition , we obtain that @xmath75 and @xmath76 are non - positive : indeed , since @xmath669 , we have @xmath672 for any @xmath673 . taking @xmath674 or @xmath675 as @xmath25 implies , @xmath676 for the reader s convenience , we recall what it means for @xmath195 to be a sub- or super- viscosity solution of the constrained hamilton - jacobi equation ( [ hj ] ) .
we say @xmath195 is a sub- ( resp .
super- ) viscosity solution of @xmath677 if , for any point @xmath678 and any smooth test function @xmath62 such that @xmath679 has a local maximum ( resp .
minimum ) at @xmath678 , we have @xmath680 @xmath681 in subsection [ sec : pf main prop ] we prove that @xmath75 and @xmath76 are , respectively , a sub- and super- viscosity solution of the hamilton - jacobi equation ( [ hj ] ) .
this is part part ( i ) of proposition [ main result ] .
we use a perturbed test function argument @xcite and some techniques similar to the proofs of ( * ? ? ?
* theorem 1.1 ) , ( * ? ? ?
* propositions 3.1 and 3.2 ) , and ( * ? ? ?
* proposition 1 ) .
we will perturb our test functions by @xmath682 , where @xmath68 satisfies the spectral problem ( [ spectral ] ) for an appropriate @xmath122 . from first glance
this is slightly different from the perturbations in , say , @xcite , due to the presence of @xmath683 ; however , this is natural seeing as we have defined @xmath684 . in subsection
[ sec : at time 0 ] we study the behavior of @xmath75 and @xmath76 at @xmath398 and establish part ( ii ) of proposition [ main result ] .
we follow the strategy of ( * ? ? ?
* propositions 3.1 , 3.2 ) . throughout this section
we employ the notational convention we mentioned in section [ notation ] : if @xmath685 , then we will use @xmath108 to denote , @xmath110 we formulate the following lemma , which is similar to ( * ? ? ?
* ; * ? ? ?
* lemma 6.1 ) .
[ lem : u * ] let @xmath686 and @xmath687 be smooth functions .
suppose @xmath688 ( resp .
@xmath689 ) has a local minimum ( resp .
maximum ) at @xmath690 .
define @xmath691 .
then there exists a subsequence @xmath692 such that 1 .
[ itemlimit ] @xmath693 , 2 .
[ itemmin ] @xmath694 has a local minimum ( resp .
maximum ) at @xmath695 for some @xmath696 , and 3 .
[ itemu ] @xmath697 ( resp .
@xmath698 ) . we postpone the proof of the lemma until the end of subsection [ sec : pf main prop ] .
we place one computation into a separate lemma .
[ lem : for main prop ] suppose @xmath686 is a smooth function that satisfies , for some @xmath690 , @xmath699 we set @xmath700 and take @xmath687 to be the solution of the spectral problem ( [ spectral ] ) corresponding to this @xmath122 .
we define @xmath701 there exist positive constants @xmath702 , @xmath703 such that for all @xmath704 , all @xmath506 and all @xmath705 we have @xmath706 we postpone the proof of the lemma and proceed with : let us fix a point @xmath707 such that @xmath708 . by proposition [ bd on utheta ] , @xmath709 is finite , so we denote @xmath710 .
we aim to prove that @xmath76 is a supersolution of ( [ hj ] ) in the viscosity sense , so let us suppose @xmath688 has a local minimum at @xmath690 for some smooth function @xmath61 .
we shall show @xmath711 we proceed by contradiction and assume that ( [ supersol ] ) does not hold .
therefore , there exists @xmath712 such that @xmath61 satisfies ( [ v at x0t0 ] ) .
we define the perturbed test function @xmath713 by ( [ def vep ] ) . according to lemma [ lem : for main prop ] ,
there exist @xmath714 and @xmath715 such that @xmath86 satisfies ( [ eqn for vep ] ) on @xmath716 for all @xmath717 . by proposition
[ bd on utheta ] , there exists some positive constants @xmath278 , @xmath16 , @xmath718 such that for all @xmath719 and for all @xmath720 , latexmath:[\[\label{utheta specific } let @xmath695 be the sequence of points given by lemma [ lem : u * ] .
according to item ( [ itemu ] ) of lemma [ lem : u * ] , we have @xmath722 therefore , for all @xmath13 large enough , we find , @xmath723 together with the estimate ( [ utheta specific ] ) on @xmath724 , this implies that there exists @xmath725 such that for @xmath726 and for all @xmath2 , @xmath727 the previous estimate implies that @xmath728 is bounded from above , uniformly in @xmath13 .
indeed , we use the definition of @xmath729 , the relationship @xmath730 and the previous line to obtain @xmath731 thus there exists @xmath732 such that for all @xmath733 , we have @xmath734 since @xmath694 has a local minimum at @xmath695 @xmath735 and @xmath736 we remark that ( [ thetaderivs ] ) holds both if @xmath696 is an interior point of @xmath3 and if @xmath696 is a boundary point of @xmath3 . indeed ,
let us suppose that @xmath737 . by the definition of @xmath86 in ( [ def vep ] ) , we have @xmath738 , where @xmath68 satisfies ( [ spectral ] ) . since @xmath68 is positive and satisfies neumann boundary conditions , we have that @xmath739 is also zero on the boundary of @xmath3 . in particular
, we find @xmath740 since @xmath741 has a local minimum at @xmath695 and the previous line holds , we also deduce that @xmath742 holds at @xmath695 , as desired . using ( [ txderivs ] ) ,
( [ thetaderivs ] ) , and that @xmath743 satisfies ( [ eqn : uep ] ) , we obtain , at @xmath695 , @xmath744 according to lemma [ lem : for main prop ] , @xmath61 satisfies ( [ eqn for vep ] ) in @xmath716 . since @xmath745
, we have @xmath746 for @xmath13 large enough .
hence , at the point @xmath695 , we have , @xmath747 subtracting the previous line from ( [ vepgeq ] ) yields , @xmath748 but this contradicts ( [ rho tiny ] ) .
we have reached the desired contradiction and conclude that @xmath76 is indeed a supersolution of ( [ hj ] ) .
the proof of lemma [ lem : for main prop ] is a simple computation .
we include it for the sake of completeness .
we observe @xmath749 so that we have , @xmath750 thus we have , for some @xmath702 small , and for all @xmath751 and all @xmath21 small enough , @xmath752 where the last equality follows since @xmath68 satisfies ( [ spectral ] ) .
this proof is less involved than that for @xmath76 .
according to ( [ sign ] ) we have @xmath753 .
next , let us suppose that @xmath754 has a local maximum at @xmath690 .
we proceed as in the proof for @xmath76 : for contradiction , we assume @xmath755 for some @xmath712 .
we set @xmath700 and take @xmath687 to be the solution of the spectral problem ( [ spectral ] ) corresponding to this @xmath122 . as in the previous proof , we define @xmath86 by ( [ def vep ] ) and find that for some @xmath702 small , for all @xmath756 , and for all @xmath21 small enough , @xmath757 according to lemma [ lem : u * ] there exists a subsequence @xmath758 and points @xmath759 such that @xmath760 has a local maximum at @xmath759 , and @xmath761 .
since @xmath743 satisfies ( [ eqn : uep ] ) , we obtain , at @xmath759 , @xmath762 ( the previous line holds even if @xmath696 is a boundary point of @xmath3 , by an argument similar to that in the previous part of the proof of this proposition ) . for @xmath13 large enough ,
@xmath763 , so both ( [ eqn : vepsuper ] ) and ( [ eqn : vepsuper2 ] ) hold at @xmath759 .
subtracting ( [ eqn : vepsuper ] ) from ( [ eqn : vepsuper2 ] ) yields @xmath764 which is impossible since @xmath765 .
now we present the proof of lemma [ lem : u * ]
. we will give the proof of the case that @xmath688 has a local minimum at @xmath690 .
the proof of the other case is similar . without loss of generality
we assume that @xmath688 has a strict local minimum at @xmath690 in @xmath766 for some @xmath767 .
let @xmath768 be a local minimum of @xmath694 in @xmath769 .
by the definition of @xmath709 , there exists a sequence @xmath770 with @xmath771 and such that @xmath772 .
we proceed with the proof of item ( [ itemlimit ] ) of the lemma .
let @xmath773 be any subsequence of @xmath774 with @xmath775 .
for @xmath13 large enough we have @xmath776 .
since @xmath777 is a local minimum of @xmath694 on @xmath769 , we obtain , @xmath778 we take @xmath779 of both sides of ( [ eq : uepveplocmin ] ) .
the definition of @xmath770 implies that the first term on the right - hand side converges to @xmath709 .
in addition , since @xmath61 is continuous and the sequences @xmath780 and @xmath781 converge to @xmath782 and @xmath690 , respectively , we find , @xmath783 since @xmath784 , the definition of @xmath785 implies , @xmath786 we use the previous line to estimate the left - hand side of ( [ afterlimsup ] ) from below and obtain @xmath787 . since @xmath690 is a strict local maximum , we see @xmath788 .
this completes the proof of items ( [ itemlimit ] ) and ( [ itemmin ] ) .
next let us take @xmath789 of both sides of ( [ eq : uepveplocmin ] ) . since we now know @xmath790 and @xmath791 , the terms with @xmath61 are equal .
in addition , the definition of @xmath770 implies that the first term on the right - hand side converges to @xmath709 .
thus we find @xmath792 since @xmath790 , the definition of @xmath709 implies that equality holds in the above .
this completes the proof of the lemma . in the previous subsection we showed that @xmath75 and @xmath76 are a supersolution and subsolution , respectively , of ( [ hj ] ) on @xmath33 . in this section
we study the behavior of @xmath75 and @xmath76 at time @xmath398 .
first we show that if @xmath793 then @xmath794 . indeed ,
if @xmath793 then there exists a point @xmath795 such that @xmath796 . hence @xmath797 and
, according to our observation ( [ sign ] ) we have @xmath798 , so we obtain @xmath799 . in the remainder of the proof we will analyze the behavior of @xmath800 for @xmath801 . to this end , we observe that , by the definition of @xmath40 , @xmath802 we fix a constant @xmath803 and a cutoff function @xmath804 that satisfies @xmath805 * first step : * we claim that @xmath75 is a viscosity subsolution of @xmath806 by which we mean if @xmath689 has a local maximum at @xmath807 for some test function @xmath61 , then either @xmath808 or @xmath809 for @xmath793 , we have @xmath810 and @xmath811 , so we find ( [ ugeq0 ] ) holds . now let us suppose @xmath812 , ( [ ugeq0 ] ) does nt hold , and @xmath813 has a local maximum at @xmath807 for some smooth @xmath61 .
we proceed as in the proof of part ( i ) of proposition [ main result ] : let us assume for contradiction that ( [ eqnvx0 ] ) also does not holds , so that @xmath814 for some @xmath712 .
we set @xmath815 and take @xmath687 to be the solution of the spectral problem ( [ spectral ] ) corresponding to this @xmath122 .
we define @xmath86 by ( [ def vep ] ) , and find that there exist points @xmath759 such that @xmath761 as @xmath816 , @xmath760 has a local maximum at @xmath759 , and @xmath817
. we claim @xmath818 if ( [ tep ] ) holds , then @xmath743 satisfies the equation ( [ eqn : uep ] ) at @xmath759 .
hence the argument in part ( i ) of the proof of proposition [ main result ] applies in this situation as well and leads to a contradiction ( namely , we will find that ( [ eqn : vepsuper ] ) , ( [ eqn : vepsuper2 ] ) both hold , which again yields ( [ rhoimpossible ] ) , and the latter can not hold ) . thus
, once we show ( [ tep ] ) , we will find that @xmath61 satisfies ( [ eqnvx0 ] ) and so we will have established ( [ var ineq at 0 ] ) .
we proceed by contradiction and assume that ( [ tep ] ) does not hold .
thus , there exists a subsequence of @xmath758 , also denoted @xmath758 , with @xmath819 . since @xmath812 , there exists @xmath714 such that @xmath820 .
we have that for all @xmath13 large enough , @xmath821 and hence @xmath822 for all @xmath13 large enough .
since , by definition , we have @xmath823 , we obtain , @xmath824 but we had assumed ( [ ugeq0 ] ) does nt hold , so that @xmath825 .
we have reached the desired contradiction , hence ( [ tep ] ) must hold , and so we conclude that @xmath75 is a viscosity subsolution of ( [ var ineq at 0 ] ) . *
second step : * we will now use that @xmath75 is a viscosity subsolution of ( [ var ineq at 0 ] ) to prove @xmath826 on @xmath827 . to this end , let us fix any @xmath812 and assume for contradiction @xmath828 for @xmath829 , let us define the test functions @xmath830 where we use @xmath831 to denote @xmath832 since @xmath75 is upper - semicontinuous , there exist @xmath833 such that @xmath834 has a maximum at @xmath833 in @xmath835 . in particular , we have @xmath836 where the equalities follow since @xmath837 and from ( [ um ] ) .
we now use that @xmath838 and the definition of @xmath839 to estimate the left - hand side of the previous line from above by @xmath840 .
thus we obtain an estimate on the distance between @xmath841 and @xmath842 : @xmath843 we will now establish the inequality @xmath844 indeed , according to proposition [ prop : spectral ] , we have @xmath845 for all @xmath122 .
we apply this with @xmath846 and obtain , @xmath847 where the second inequality follows from ( [ estxdelx0 ] ) .
let us add @xmath831 to both sides of the previous line .
the left - hand side becomes exactly the left - hand side of ( [ estonquantity ] ) .
the right - hand side becomes @xmath848 , due to our choice of @xmath849 in ( [ choicemu ] ) .
thus we find ( [ estonquantity ] ) holds .
we now recall that @xmath834 has a maximum at @xmath833 in @xmath835 .
let us suppose that @xmath850 for some @xmath851 . according to part ( i ) of proposition [ main result ] , @xmath75 is a subsolution of ( [ hj ] ) in @xmath33 , which implies @xmath852 but this is impossible , as we have just established ( [ estonquantity ] ) .
therefore , we must have @xmath853 for all @xmath851 .
but we also know that @xmath75 is a subsolution of ( [ var ineq at 0 ] ) on @xmath854 .
therefore , we have @xmath855 but , again according to ( [ estonquantity ] ) , we have that the first term inside the @xmath856 must be strictly positive .
therefore , in order for the previous line to hold , the second term inside the @xmath856 must be non - positive .
thus we have , @xmath857 since @xmath858 is a local maximum of @xmath834 , we have @xmath859 because @xmath860 and @xmath837 , we have that the left - hand side of the previous line is exactly @xmath861 . in addition , we use that @xmath839 is non - negative and the estimate ( [ bumuzeta ] ) to bound the right - hand side of the previous line from above .
we find , @xmath862 since @xmath863 as @xmath864 and @xmath260 is continuous , we obtain @xmath865 , which is impossible since @xmath866 and @xmath867 is arbitrary .
we have obtained the desired contradiction and thus ( [ um ] ) can not hold .
we conclude @xmath868 , and hence the proof is complete .
this proof is similar to the one for @xmath75 .
first we show that if @xmath869 , then @xmath870 . indeed ,
since @xmath869 , there exists @xmath871 with @xmath872 .
therefore , @xmath873 we will now prove that @xmath874 for @xmath875 .
* first step : * we prove that @xmath76 is a supersolution of @xmath876 to this end , let us suppose @xmath875 , @xmath877 and @xmath688 has a local minimum at @xmath807 for some test function @xmath61 .
we proceed as in the proof of part ( i ) of proposition [ main result ] : we define @xmath86 by ( [ def vep ] ) , and find that there exist points @xmath759 such that @xmath761 as @xmath816 , @xmath760 has a local minimum at @xmath759 , and @xmath878
. we claim @xmath879 if ( [ tep ] ) holds , then the argument in the proof of part ( i ) of proposition [ main result ] applies in this situation as well .
thus , once we show ( [ tep for uunder ] ) , we will find that @xmath61 satisfies @xmath880 at @xmath807 and so we will have established that ( [ var ineqn underu ] ) in the viscosity sense .
we proceed by contradiction and assume that ( [ tep for uunder ] ) does not hold .
thus , there exists a subsequence , also denoted @xmath758 , with @xmath819 . since @xmath875 , there exists @xmath714 such that @xmath881 . since @xmath882 , we have that for all @xmath13 large enough , @xmath821
. therefore , there exists @xmath883 so that @xmath884 for all @xmath13 large enough .
hence we find , @xmath885 where the first equality follows from the definition of the sequence @xmath759 .
but the previous line contradicts our assumption ( [ uux0 ] ) .
therefore , ( [ tep for uunder ] ) must hold , and we conclude that @xmath76 is a viscosity supersolution of ( [ var ineqn underu ] ) . * second step : * let us fix @xmath875 .
we will prove @xmath886 , which , together with ( [ sign ] ) , implies @xmath874 .
let us suppose for contradiction @xmath887 we point out that @xmath888 is finite .
indeed , if @xmath870 then @xmath688 has a minimum at @xmath807 for all @xmath61 , and since we know @xmath76 is a supersolution of ( [ var ineqn underu ] ) , we find @xmath889 at @xmath807 for all @xmath61 , which is of course impossible .
since @xmath76 is lower - semicontinuous and finite at @xmath807 , there exists a neighborhood @xmath68 of @xmath807 and some finite @xmath890 such that if @xmath71 , then @xmath891 .
we define , for @xmath829 , the test functions @xmath892 where we define @xmath849 by ( [ choicemu ] ) as in the previous proof .
there exists a sequence @xmath893 such that @xmath894 has a local minimum at @xmath895 .
we find , as in the previous proof , that the upper bound ( [ estxdelx0 ] ) on @xmath896 holds here as well .
we use ( [ estxdelx0 ] ) , the definition of @xmath831 and the properties of @xmath30 given in proposition [ prop : spectral ] to obtain , for all @xmath851 , @xmath897 let us suppose @xmath850 .
since @xmath894 has a minimum at @xmath833 and , according to part ( i ) of proposition [ main result ] , @xmath76 is a supersolution of ( [ hj ] ) in @xmath33 , we find that @xmath898 must hold in [ negubar ] ) .
but this is impossible , since we have already established ( [ negubar ] ) . therefore , we find @xmath853 for all @xmath851 .
according to ( [ var ineqn underu ] ) , we have that @xmath76 is a supersolution of ( [ var ineqn underu ] ) on @xmath899 , so we find @xmath900 according to ( [ negubar ] ) we see that the second term in the @xmath901 is strictly negative . hence we obtain @xmath902 for all @xmath851 .
together with @xmath837 , the fact that @xmath858 is a minimum of @xmath894 , and @xmath903 , this implies , @xmath904 but this contradicts our assumption ( [ undux0 ] ) .
thus we find @xmath905 and the proof is complete .
in this short section we use the results that we ve established in the rest of the paper to give the proof of our main result , theorem [ result on n ] .
the arguments are similar to those in the proofs of ( * ? ? ?
* theorem 1.1 ) and ( * ? ? ? * theorem 1 ) .
since ( [ hj ] ) satisfies the comparison principle ( see proposition [ prop : prelem ] ) , proposition [ main result ] and the definitions of @xmath43 and @xmath44 imply @xmath906 and @xmath907 on @xmath33 .
let us fix some @xmath908 .
since @xmath43 is continuous , there exists @xmath712 such that @xmath909 for all @xmath910 . since @xmath906
, we have @xmath911 for all @xmath910 as well
. therefore , @xmath912 for all @xmath910 , for all @xmath2 and for all @xmath21 small enough .
thus , for all @xmath910 and for all @xmath9 , we have @xmath913 so we find that @xmath914 uniformly at an exponential rate on @xmath915 .
now let us suppose @xmath690 is a point in the interior of @xmath83 , so that @xmath916 for some @xmath714 . according to ( [ sign ] ) , we have @xmath917 on @xmath78 . since @xmath907
, we have @xmath918 on @xmath919 , so we see @xmath920 on @xmath919 .
we define the test function @xmath921 since @xmath920 on @xmath919 , we find that @xmath922 has a local minimum at @xmath690
. therefore there exists a sequence @xmath923 with @xmath924 such that @xmath925 has a local minimum at @xmath923 . since @xmath67 is satisfies ( [ eqn : uep ] ) in the viscosity sense , we find , at @xmath923 , @xmath926 using the definition of @xmath62 and rearranging yields , @xmath927 taking the limit @xmath25 of the previous line yields , @xmath928 recalling the definition of @xmath929 completes the proof .
corollary [ corc*informal ] follows easily from our main result , theorem [ result on n ] , and the following lemma , which is essentially a restatement of a result of majda and sougandis @xcite . before stating the lemma
, we introduce one more bit of notation . given a bounded interval @xmath930 , we use @xmath931 to denote a @xmath932 bounded function that is positive on @xmath39 , negative on @xmath933 , with a maximum at some @xmath934 , strictly increasing to the left of @xmath935 and strictly decreasing to the right of @xmath935 .
[ lemfornewcor ] assume @xmath930 is a bounded interval .
let @xmath936 be the unique viscosity solution of @xmath937 let @xmath38 be given by lemma 1.1 .
then : 1 .
[ itemms ] @xmath938 and @xmath939 ; and , 2 .
[ itemchar ] @xmath940 and @xmath941 . in @xcite ,
the authors analyze the properties of the front @xmath942 , where @xmath195 solves an equation of the form ( [ hj ] ) .
it turns out that , for a general hamiltonian @xmath30 that depends on space and time , this front may not be very regular .
however , our situation is rather simple and no degeneracy appears indeed , item ( [ itemms ] ) says that our front is geometric . item ( [ itemms ] ) is a special case of propositions 2.3 and 2.4 of @xcite .
indeed , what we denote @xmath39 , @xmath38 , @xmath936 are denoted @xmath943 , @xmath944 , @xmath195 in @xcite . and ,
if we use @xmath945 to denote the hamiltonian @xmath30 of @xcite , reserving @xmath30 itself for our hamiltonian , then we have the correspondence @xmath946 between the two notations . in addition , the velocity field @xmath947 of @xcite is @xmath470 in our situation , so the hypotheses of both propositions are satisfied .
finally , we may use ( [ hc * ] ) to find that the general nonlinearity @xmath948 in line ( 2.5 ) of @xcite is , in our situation , simply given by @xmath949 .
( for the benefit of the reader , we also point out two typos in section 2.2 of @xcite . first , ( 1.6 ) " should read ( 1.9 ) " throughout that section .
second , @xmath950 " should read @xmath951 " in the conclusion of proposition 2.3 . ) the proof of item ( [ itemchar ] ) is based on the method of characteristics .
the characteristics for equations of the form @xmath952 are analyzed in detail in barles @xcite .
taking @xmath953 in order to study ( [ eqn : wom ] ) yields that the characteristic emanating from @xmath842 is @xmath954 ( where @xmath955 if @xmath956 , @xmath957 if @xmath958 , and @xmath470 if @xmath959 ) , and , @xmath960 let us use the properties of @xmath961 to continue our analysis .
we recall that @xmath961 has a maximum at @xmath935 , is strictly increasing to the left of @xmath935 and strictly decreasing to the right .
hence , if @xmath842 is to the left of @xmath935 , then the characteristic has constant slope @xmath962 in the @xmath678 plane ; while , if @xmath842 is to the right of @xmath935 , then the slope is @xmath963 .
we remark that the characteristics never cross . and
, @xmath936 is constant along these characteristics , so , in particular , the sign of the initial condition is preserved along them .
now , let @xmath678 be a point with @xmath964
. we shall show @xmath965 .
let us suppose @xmath1 lies to the right of @xmath935 ( the other case is similar )
. then @xmath678 must lie on a characteristic with slope @xmath963 , so that @xmath966 for some @xmath842 .
we compute , @xmath967 where the equality follows since @xmath966 , and the second inequality since @xmath964 .
since @xmath842 lies outside @xmath39 , @xmath961 is negative there . because @xmath968 is constant along characterisics , we conclude @xmath969 .
we omit the similar arguments that establish the remainder of the proposition .
we now proceed with : let @xmath43 and @xmath44 be as in lemma [ lem intro ] and let @xmath970 and @xmath971 be as in lemma [ lemfornewcor ] . combining the two parts of lemma [ lem intro ] yields , @xmath972 on the other hand , theorem [ result on n ] implies , * if @xmath973 , then @xmath974 for all @xmath2 . * if @xmath678 is in the interior of @xmath975 , then @xmath54 . combining this with the previous line yields the corollary .
we now use corollary [ corc*informal ] and the bounds ( [ boundsonc * ] ) on @xmath50 to present : to establish the first statement of the corollary , let @xmath678 be such that @xmath55 . according to the second inequality in ( [ boundsonc * ] ) , we also have @xmath51 . using corollary [ corc*informal ] completes the proof of the first statement ; the proof of the second is analogous .
[ [ appendix ] ] in this appendix , we state an existence result and a comparison principle for ( [ hj ] ) with infinite initial data ( proposition [ prop : prelem ] ) .
these were established in ( * ? ? ?
* section 4 ) but not explicitely stated there , and we could not locate another reference in the literature . because of this , we carefully explain how to obtain them from ( * ? ? ?
* section 4 ) .
in addition to @xcite , we also refer the reader to crandall , lions and souganidis @xcite for more about hamilton - jacobi equations with infinite initial data .
[ prop : prelem ] there exists a viscosity solution @xmath968 of ( [ hj ] ) on @xmath976 with with initial data ( [ infinite ] ) . moreover ,
suppose @xmath977 and @xmath978 are , respectively , a viscosity subsolution and a viscosity supersolution of ( [ hj ] ) on @xmath976 , and both with the initial data ( [ infinite ] ) .
then we have @xmath979 in particular , the solution to ( [ hj ] ) with initial data ( [ infinite ] ) is unique .
we explain why proposition [ prop : prelem ] is exactly what was proven in ( * ? ? ?
* section 4 ) . since both we and @xcite use @xmath30 as notation for the hamiltonian , for the purposes of this proof we use @xmath945 to denote the hamiltonian of @xcite .
there is a difference in sign between our paper and ( * ? ? ?
* section 4 ) , which we now address .
we have that @xmath980 , @xmath981 and @xmath982 are , respectively , a viscosity solution , supersolution , and subsolution of @xmath983 with initial data @xmath984 thus we see that , if we take @xmath985 and @xmath986 , then we are exactly in the situation of ( * ? ? ?
* section 4 ) .
we may take @xmath987 , @xmath988 , and @xmath989 .
thus we ve established the existence of a solution . plus , the conclusion of the comparison of the arguments in ( * ? ? ? * section 4 ) ( specifically , lines ( 4.2 ) and ( 4.5 ) ) is @xmath990 translating back to our notation , we see ( [ ws ] ) holds .
the author thanks her thesis advisor , takis souganidis , for his guidance and encouragement .
the author is grateful to vincent calvez and sepideh mirrahimi for reading earlier drafts of this paper very thoroughly .
their remarks were invaluable .
in addition , the author thanks benoit perthame for a stimulating discussion that led to the developement of corollary [ corc*informal ] .
the author also thanks the anonymous referees .
their comments have greatly helped improve the exposition and raised several interesting questions ; for example , remarks [ remark : relationship ] and [ remark : choice of nonlinearity ] were motivated by their reports .
m. alfaro , j. coville , and g. raoul .
traveling waves in a nonlocal equation as a model for a population structured by a space variable and a phenotypical trait . comm .
partial differential equations ( cpde ) .
volume 38 , issue 12 , 2013 a. arnold , l. desvillettes and c. prevost .
existence of nontrivial steady states for populations structured with respect to space and a continuous trait .
pure appl .
11 ( 2012 ) , no . 1 , 83 - 96 .
bouin , emeric ; calvez , vincent ; meunier , nicolas ; mirrahimi , sepideh ; perthame , benot ; raoul , gal ; voituriez , raphal . invasion fronts with variable motility : phenotype selection , spatial sorting and wave acceleration . c. r. math .
paris 350 ( 2012 ) , no .
15 - 16 , 761 - 766 .
crandall , michael g. ; lions , pierre - louis ; souganidis , panagiotis e. maximal solutions and universal bounds for some partial differential equations of evolution . arch .
105 ( 1989 ) , no . 2 , 163 - 190 .
a. kolmogorov , i. petrovskii , and n. piscounov .
a study of the diffusion equation with increase in the amount of substance , and its application to a biological problem . in v. m. tikhomirov ,
editor , selected works of a. n. kolmogorov i , pages 248 - 270 .
kluwer 1991 , isbn 90 - 277 - 2796 - 1 .
translated by v. m. volosov from bull .
moscow univ . , math .
mech . 1 , 1 - 25 , 1937 thomas , c. d. , bodsworth , e. j. , wilson , r. j. , simmons , a. d. , davies , z. g. , musche , m. , conradt , l. ecological and evolutionary processes at expanding range margins .
nature 411 , 577 - 581 ( 2001 ) . | we study a reaction - diffusion equation with a nonlocal reaction term that models a population with variable motility .
we establish a global supremum bound for solutions of the equation .
we investigate the asymptotic ( long - time and long - range ) behavior of the population .
we perform a certain rescaling and prove that solutions of the rescaled problem converge locally uniformly to zero in a certain region and stay positive ( in some sense ) in another region .
these regions are determined by two viscosity solutions of a related hamilton - jacobi equation . |
the study of pion double charge exchange ( dcx ) has been pursued with the aim of elucidating the mechanisms of multiple pion interactions in nuclei . whatever the details of the mechanism , this reaction must at least result in the charges of two nucleons being changed .
indeed , it has been found that the simplest picture of the reaction , in which it proceeds by successive single charge exchanges ( see fig . 1 ) , provides a qualitative explanation of the systematics that are observed @xcite .
serious investigation of the inclusive dcx reaction began in 1980 , with the first measurement using modern equipment of the doubly differential cross section for @xmath3o@xmath4x at three angles with 240 mev pions performed at sin ( now psi ) @xcite .
subsequently , an extensive series of measurements of inclusive double charge exchange cross sections was carried out at the los alamos meson physics facility @xcite .
the @xmath5 reaction was studied in @xmath6he , and the @xmath7 reactions in nine nuclei : @xmath8he , @xmath0li , @xmath1be , @xmath2c , @xmath3o , @xmath9ca , @xmath10rh , and @xmath11pb .
doubly differential cross sections were measured for each target over the outgoing pion energy range from 10 mev up to the kinematic limit , at three to five angles for incident pion energies in the range from 120 to 270 mev .
some 300 pion energy spectra have been obtained with good statistics and absolute accuracies of 10 - 15% .
the systematics of the reaction in @xmath3o , @xmath9ca , @xmath10rh , and @xmath11pb revealed by these experiments has been previously discussed @xcite .
briefly , it is found that only in @xmath3o do the doubly differential cross sections much resemble the most likely phase space distribution ( that containing a pion , two nucleons and a residual nucleus ) or , in fact , any other reasonable distribution generated by assuming that the mechanism is governed by a constant matrix element . in the heavier nuclei
the distribution of outgoing pions is strongly peaked toward lower energies and depleted at higher energies when compared to the phase space distribution .
the doubly differential cross sections for dcx in @xmath6he and @xmath8he exhibited striking double peaked outgoing pion energy spectra at forward angles ( 25@xmath12 , 50@xmath12 ) that evolved into a single peak distribution as the angle of observation reached 130@xmath12 .
this feature was most pronounced at an incident energy of 240 mev , but persisted at incident energies down to 150 mev in @xmath8he @xcite and 120 mev in @xmath6he @xcite . in this paper
we report measurements of the doubly differential cross section for inclusive dcx in the light @xmath13-shell
nuclei @xmath0li , @xmath1be , and @xmath2c .
these observations will serve to trace the evolution of the double peaked structure into a nearly featureless spectrum as the complexity of the target nucleus is increased .
a second objective of the present work was to investigate , by comparing the @xmath14 reactions in @xmath15li and @xmath16li , the effect on the dcx cross section of a single extra neutron in the target nucleus .
only fragmentary data for dcx in these nuclei existed prior to the present measurement .
_ have reported a few measurements at 0@xmath12 for @xmath4 in li and be @xcite and @xmath17 in @xmath16li , @xmath1be , and @xmath2c @xcite , at incident energies in the 200 mev region .
measured total dcx cross sections for 80 mev @xmath18 on be and al and 140 mev @xmath19 on be and c @xcite . using a propane bubble chamber , massue _ et al . _
measured total cross sections for @xmath4 in @xmath2c at 138 , 204 , and 228 mev @xcite .
et al . _ measured a @xmath18 spectrum at 30@xmath12 for 107 mev @xmath19 on @xmath16li , but with poor statistical accuracy @xcite .
subsequent to the measurements discussed in the present paper , forward - angle cross sections for the @xmath20 reaction on @xmath0li and @xmath2c at incident energies 0.59 , 0.75 , and 1.1 gev have been reported @xcite .
in the present experiment , the inclusive dcx processes @xmath7 have been investigated by measuring the doubly differential cross sections , @xmath21 , at three to five angles in the range 25@xmath12130@xmath12 , for incident pion energies between 120 and 240 mev . at each angle , the range of outgoing pion energies from 10 mev up to the kinematic limit for each reaction was covered .
preliminary results of this experiment were reported in ref .
@xcite .
the experiment was performed with incident @xmath22 beams from the high - energy pion channel ( `` p@xmath6 '' ) at the los alamos meson physics facility .
the outgoing pions were detected using a 180@xmath12 vertical bend , double focusing magnetic spectrometer , with an effective solid angle of about 16 msr and a momentum bite of about 8% , as shown in fig .
the experimental apparatus and procedure have been discussed in detail by wood _
et al . _
@xcite , yuly _ et al . _
@xcite , and kinney _ et al .
_ @xcite and will be treated only briefly here .
the detector system had five components : a wire chamber ( wc0 ) placed in the mid - plane of the spectrometer , two wire chambers ( wc1 and wc2 ) near the focal plane that were used to reconstruct trajectories , a 1.6 mm scintillator ( s1 ) behind wc1 and wc2 that provided an accurate time reference for the trigger as well as pulse height information , and a fluorocarbon ( fc-88 ) erenkov detector that distinguished electrons from pions .
the trigger was a fourfold coincidence among the three wire chambers and the scintillator .
inclusion of wc0 guaranteed that a particle had passed through the spectrometer .
the timing information available from this chamber was also used to distinguish pions from very slow protons .
all of the targets ( @xmath15li , @xmath16li , @xmath1be , @xmath2c ) in this experiment were solid rectangular slabs .
in addition , a polyethylene ( ch@xmath23 ) target was employed to simultaneously normalize both the beam flux and spectrometer acceptance by observing elastic @xmath24 scattering .
the targets were mounted in aluminum frames on a ladder that could be raised or lowered ( to place the target of interest in the incident beam ) and rotated about the vertical axis ( to orient the plane of the target at an appropriate angle with respect to the beam ) . before reaching the target ,
the pion beam passed through an ionization chamber that was used to determine the relative flux . since this device was sensitive to all charged particles in the beam , it was necessary to normalize the flux measurement each time the beam transport elements were adjusted .
downstream of the target the pion beam impinged on a ch@xmath23 slab from which pions scattered at 90@xmath12 on each side of the beam were detected by scintillator telescopes , in order to monitor the position of the incident beam and to provide a cross - check of the ionization chamber . to measure the doubly differential dcx cross sections each target
was exposed to an incident pion beam and events were recorded at a series of spectrometer momentum settings until the measured statistical uncertainties in the number of detected pions were approximately 5% .
data were collected at 10 mev intervals in outgoing pion energy .
in addition , background observations were made at a selected set of spectrometer settings using an empty target frame .
each time the incident pion charge or energy was changed , a series of ch@xmath23 normalizations was performed .
the goal of this experiment was to measure a doubly differential cross section for each reaction , incident pion energy , scattering angle , and outgoing pion energy .
the doubly differential cross section is related to observable quantities as follows : @xmath25 where @xmath26 is the number of pions detected , @xmath27 is the number of incident pions , @xmath28 is the effective thickness of the target , @xmath29 is the target density , @xmath30 is the effective solid angular acceptance , @xmath31 is the range of outgoing pion energy , @xmath32 is the correction of the spectrometer acceptance due to multiple scattering and energy loss , @xmath33 is the correction due to pion decay , and @xmath34 is the dead - time correction . the analysis procedure is described in detail in ref .
@xcite and will only be summarized here .
( see also refs .
@xcite . )
calibration constants relating time differences of the signals from the wire chambers to positions on the focal plane were established by placing a collimated @xmath35fe source at precisely measured positions in front of the chambers . in the data analysis ,
the position information from wc1 and wc2 for each event was used to reconstruct the particle trajectory back to the focal plane of the spectrometer .
reconstructed trajectories were tested for conformity with the distribution in phase space of particles that could have been transmitted by the spectrometer from the target to the focal plane .
positive pions and protons were generally distinguished on an event - by - event basis by means of their different pulse heights in the scintillator , or , in some cases , by their time - of - flight between wc0 and s1 .
since pions and electrons both emit erenkov radiation in fc-88 at spectrometer settings greater than 180 mev / c ( @xmath36 mev ) , it was not possible to use the erenkov detector to distinguish electrons from pions on an event - by - event basis . instead , a statistical method was used , which is described in detail in ref .
@xcite and is very similar to that described in ref .
@xcite .
the momentum acceptance , @xmath37 , and dispersion were determined by changing the spectrometer magnetic field to move a @xmath24 elastic scattering peak across the spectrometer focal plane . the relative acceptance as a function of focal plane position obtained by this method
was used to correct @xmath38 .
the effective total momentum acceptance was determined to be about 8% by integration across the entire focal plane . the absolute number of incident pions , @xmath27 in eq .
( [ eqn : xsect ] ) , was obtained by comparison of @xmath24 elastic scattering measurements with known cross sections determined from the energy - dependent phase shift program scatpi @xcite .
the effective total solid angle , @xmath30 , was included in this calibration .
this procedure was repeated at several scattering angles for each setting of the beam transport system in order to improve the accuracy of the normalization , as well as to check for the presence of angle - dependent systematic errors . a linear interpolation between the spectrometer momentum settings at which empty - frame ( background ) measurements were made was used to obtain the background contribution which
was then subtracted from the raw dcx spectra .
the largest backgrounds occurred at the lowest measured outgoing pion energy of 10 mev and were about 25% of the foreground .
corrections were made to account for other effects that would change the shape and magnitude of the measured cross sections .
the number of pions detected was reduced by decay as they traveled from the target to the detectors .
approximately 80% of the pions survived at 200 mev , while at 10 mev only 30% survived . some of the decay muons traversed the spectrometer , while others did not .
moreover , it was possible for pions emerging from the target that would not have traversed the spectrometer to decay into muons that could do so .
since the detector system could not separate the muons from pions , these effects were accounted for by monte carlo methods which produced the factor @xmath33 in eq .
( [ eqn : xsect ] ) .
this procedure is described in detail in refs.@xcite .
the fraction of muon contamination ranges from 1% at 10 mev to 20% at 200 mev outgoing pion kinetic energy .
energy loss and multiple scattering in the target and in wc0 changed the effective acceptance of the spectrometer ; these were corrected for by the factor @xmath32 , which was determined by simulation to be different from unity only for outgoing pion energies less than 30 mev .
the maximum value of @xmath32 was about 1.3 .
integration of the doubly differential cross sections over outgoing pion energy , using the trapezoid rule , yielded an angular distribution .
the angular distributions were fitted with sums of legendre polynomials , which were integrated to determine a quantity which we term the `` total reaction cross section . '' the uncertainties in these procedures are discussed in ref .
@xcite and were less than , or comparable to , the systematic uncertainties discussed in the next section .
the uncertainty in the normalization of the cross section contains contributions from uncertainties in the thicknesses of the targets ( 23% ) and in the phase shift prediction of the free @xmath24 cross section ( 2% ) @xcite . the uncertainty in the determination of a particle s position in the spectrometer focal plane and of the spectrometer acceptance function are each estimated to be 1% .
for the measurements at 180 mev and 240 mev there is an additional 4% uncertainty in a correction for trigger inefficiency .
the uncertainties due to the determination of electronics deadtime , spectrometer dispersion , and wire chamber efficiency were insignificant compared to those listed above . in addition , there is an uncertainty due to slowly varying changes in the response of the ionization chamber used to monitor the incident beam flux .
these variations were monitored by comparison with the primary proton beam monitor and the downstream scattering monitor , and were observed to occur on roughly the same time scale as changes of spectrometer angle . the resulting uncertainty is thus treated as an angle - dependent uncertainty .
there are also uncertainties that depend on the outgoing pion energy .
these include contributions from the electron - pion separation procedure , which were between 2% and 5% , from estimates of the corrections for energy loss and multiple scattering in the target and in the mid - spectrometer wire chamber , and from effects of pion decay and muon contamination , which contributed an additional 5% .
table i lists the systematic uncertainties for each incident beam . for incident
pion energies above about 170 mev , detection of an outgoing pion of charge opposite to that of the incident pion no longer provides a unique signature of dcx , as the processes @xmath39 or @xmath40 can occur on one of the target nucleons .
although the contribution of pion - induced pion production ( pipp ) is negligible ( see ref .
@xcite ) for incident energy 180 mev , this effect must be taken into account at 240 mev . to estimate the contribution of pipp
, we follow the procedure used in ref .
@xcite . assuming that pipp in a nucleus occurs as a quasi - free process on a single nucleon , the effects of absorption and other competing @xmath41-nucleus reactions and of the nuclear medium are taken into account by multiplying the free nucleon pipp cross section by an `` effective '' number of nucleons , @xmath42 .
an extensive study of @xmath41-nucleus reactions was reported by ashery _
et al._@xcite , from which values of @xmath42 were extracted .
these authors found that @xmath42 for positive ( negative ) incident pions exhibited a universal , energy - independent power - law dependence on @xmath43 ( @xmath44 ) .
moreover , the incident energy dependence of @xmath42 was found to be very similar for all target nuclei .
these two results allow us to obtain , by interpolation , values of @xmath42 for @xmath0li , @xmath1be , @xmath2c , and @xmath3o for incident @xmath45 energy 240 mev .
precise measurements of the @xmath46 cross section have been reported by bjork _
et al._@xcite .
the observed@xcite equality of the @xmath47 and @xmath48 cross sections , and of the @xmath49 and @xmath46 cross sections , justifies using the results of bjork _
et al._@xcite for both positive and negative incident pions .
this procedure yields the values of pipp cross sections given in table ii . although extensive studies of the differential cross sections for pipp in nuclei have been carried out , _
e.g. _ , most recently by the chaos collaboration @xcite , few total cross sections have been quoted .
_ @xcite observed the @xmath2c@xmath50 reaction at 292 mev , and grion _ et al .
_ @xcite observed the @xmath3o@xmath51 reaction at 280 mev .
unfortunately , these results are inconsistent when rescaled using a reasonable energy and @xmath52-dependence ( see ref .
_ speculate that the discrepancy may arise from the methods used to extrapolate the data to obtain a total reaction cross section .
they point out that the fraction of phase space covered in their work is 43 times larger than that covered in the experiment of grion _
et al._. if we use the energy dependence of the free nucleon pipp cross section @xcite to rescale rahav _
s result for @xmath2c to 240 mev , we obtain @xmath53b , in fair agreement with the value obtained using @xmath42 given in table ii .
we note that pipp can only affect the low energy portion of the outgoing pion energy spectra .
the estimated contributions will be shown when the results are presented in the next section .
a sample of the results of the present investigation will be displayed in this section and some qualitative and semi - quantitative aspects of the measured cross sections will be discussed .
the data will be compared with the predictions of two theoretical models in the next section .
doubly differential cross sections for the @xmath15li@xmath14 process are shown in figs . 3 , 4 , and 5 for incident energies 120 , 180 , and 240 mev , respectively , and angles of observation between 25@xmath12 and 130@xmath12 .
the @xmath15li@xmath20 cross sections ( see ref .
@xcite ) are essentially identical in shape and magnitude and are not shown .
the solid curves represent the distribution of events in four - body ( outgoing pion , two nucleons , and residual nucleus ) phase space and are normalized in each case to the area under the experimental spectrum .
the phase space curves are seen to provide a good representation of the data at incident pion energy 120 mev and a poorer representation at the higher incident energies .
the dashed curves in fig . 5 indicate the estimated contribution of pipp , assuming a phase - space distribution ( as was found by rahav _
the angular dependence of the pipp cross section was obtained from the measurements of yuly _ et al .
_ @xcite and the normalization determined by the total cross section given in table iii .
figure 6 shows the doubly differential cross sections for @xmath16li@xmath54 ( solid symbols ) and @xmath16li@xmath20 ( open symbols ) at 240 mev incident energy .
the dotted and dot - dashed curves indicate the estimated contributions of pipp to the @xmath14 and @xmath55 cross sections , respectively . comparing figs . 5 and 6 , it is seen that the shapes of the energy spectra are very similar at 240 mev for the two li isotopes .
this similarity persists at the lower incident energies @xcite .
although the @xmath16li@xmath14 and @xmath20 cross sections are essentially identical in shape , the former is seen to be approximately a factor of three larger than the latter .
this difference in magnitude will be discussed later in this section .
it is also seen that the pipp contribution is nearly identical for both reactions , and thus is relatively larger for @xmath16li@xmath20 .
figure 7 shows the doubly differential cross sections for @xmath1be@xmath54 ( solid symbols ) and @xmath1be@xmath20 ( open symbols ) at 180 mev incident energy .
the @xmath14 cross section is seen to be approximately a factor of three larger than that for @xmath20 . as was stated earlier , one motivation for the study of inclusive dcx in light @xmath13-shell nuclei was to trace the @xmath52-dependence of the cross section in particular the forward - angle cross section from the double - peaked structure seen in @xmath56he to the single peak seen in @xmath3o and heavier nuclei .
figure 8 shows the doubly differential cross sections at 25@xmath12 for 240 mev ( fig .
8(a ) ) and 180 mev ( fig .
8(b ) ) positive pions incident on @xmath8he , @xmath15li , @xmath16li , @xmath1be , @xmath2c , and @xmath3o .
indeed , one observes a smooth progression in the shape of the spectra as @xmath52 increases .- dependence is observed in the @xmath20 cross sections .
] at 240 mev , the double - peaked structure is prominent in @xmath8he , apparent in the li isotopes , disappearing for @xmath1be , and is absent in @xmath2c and @xmath3o . at 180 mev the double - peaked structure is only seen in the @xmath8he spectrum , although the data for @xmath0li and @xmath1b exhibit strength at high energy that does not appear in the @xmath2c and @xmath3o spectra .
we note that in @xmath8he the double peaks are still visible at 150 mev @xcite , and in @xmath6he at all energies down to 120 mev @xcite .
it is seen that the relative size of the pipp contribution , shown as the dotted curves in fig .
8(a ) , decreases with increasing @xmath52 .
the origin of the double - peaked structure has been discussed previously ( see , _
e.g. _ , refs .
@xcite ) , and has been argued to be consistent with a sequential single charge exchange mechanism . in the @xmath57 resonance region
the @xmath41-nucleon cross section is strongly forward- and backward - peaked .
two forward scatterings or two backward scatterings , which are favored over scattering at intermediate angles , can yield an outgoing dcx pion at forward angles : in the former case the pion emerges with high energy ; in the latter case it emerges with lower energy . a theoretical model based on these ideas
was developed by kinney and successfully applied to dcx data on @xmath8he @xcite and , subsequently , @xmath6he @xcite .
given the prominent double - peaked structure in the @xmath15li cross section , we have extended the model to this nucleus , and will show the results in the next section .
alqadi and gibbs @xcite have presented calculations for dcx in @xmath8he which they compare with the data given in ref .
these authors found that including pipp in their intranuclear cascade model was essential to produce reasonable agreement with the data at 240 mev , and suggest that this mechanism is largely responsible for the low - energy peak in the pion energy spectra .
the contribution of pipp for @xmath58 , obtained as discussed previously , is shown by the dashed curves in fig .
8(a ) .
as described in the previous section , the doubly differential cross sections were integrated to yield angular distributions ( see ref .
a typical set of results , those for @xmath15li , is shown in fig .
9 . for this nucleus and
the other nuclei investigated , the angular distribution is seen to be forward peaked at 240 mev and to become more isotropic as the incident energy decreases .
no significant difference is seen in the angular dependences of the @xmath4 and @xmath17 cross sections in those nuclei for which both were measured .
although the angular distributions are relatively featureless , their shape does exhibit a consistent @xmath52-dependence , becoming less forward - peaked as @xmath52 increases . as a measure of the forward - peaking , the ratios of the cross sections measured at @xmath59 and @xmath60 for the @xmath61 reactions at 240 mev for @xmath62 were calculated and are displayed in table iii .
the observed trend is consistent with the increased probability of absorption of forward - going pions as @xmath52 increases ; the dcx pion has a greater chance of escaping from the `` rear '' of a heavy nucleus .
the unusually small ratio seen for @xmath11pb@xmath17 may be attributable to the large excess of neutrons , with which the incoming @xmath19 preferentially interacts , in this nucleus .
total dcx reaction cross sections were determined as described in the previous section ; the results for the light @xmath13-shell nuclei are given in table iv .
the quoted errors include statistics , systematics , and the uncertainties arising from the integration procedures .
the numbers in parentheses for 240 mev incident pions are the cross sections after the subtraction of the estimated contributions of pipp . for a self - conjugate nucleus , the @xmath4 and @xmath17 cross
sections should be equal , apart from differences attributable to the different binding energies of the final state nuclei , which would affect primarily the high - energy end of the outgoing pion energy spectrum .
the total cross sections for dcx in @xmath3o and @xmath9ca were found to be equal within the experimental uncertainties @xcite , although the @xmath17 cross section slightly exceeded the @xmath4 cross section for @xmath9ca .
table iv shows the two cross sections for @xmath15li to be essentially equal within errors .
starting from the notion that double charge exchange proceeds by sequential single charge exchange scatterings competing with more probable reactions that either interrupt the sequence or scatter pions out of the nucleus before it starts , a `` scaling rule''was found which organizes the @xmath52-dependence of the total cross section for both the @xmath4 and @xmath17 reactions @xcite . according to this rule the total cross section for dcx should vary as @xmath63 where @xmath64 is the number of nucleons ( @xmath44 or @xmath43 ) on which dcx occurs for a given incident pion charge ( positive or negative ) ; the factor @xmath65 is proportional to the projected area of the nucleus .
total dcx reaction cross sections in nuclei ranging in mass from @xmath15li to @xmath11pb are found substantially to obey this rule ; only the isotopes of he violate it @xcite .
given that the scaling rule is based on the classical transport of pions through nuclear matter , it is perhaps surprising that the agreement is so good , especially for the light nuclei .
the validity of the rule has recently been discussed by buss _
_ @xcite .
let us examine the predictions of the scaling rule in more detail for the light @xmath13-shell nuclei .
by comparing the @xmath7 cross sections in @xmath15li and @xmath16li , one can examine the effect of adding one neutron . for @xmath4
one is adding a target nucleon ; for @xmath17 one is adding a source of competing reactions and the number of target nucleons is unchanged .
( 2 ) predicts the ratios @xmath66 and @xmath67 .
note that in the absence of competing reactions @xmath68 would be equal to unity .
the observed ratios are given in table v. the agreement between experiment and prediction for @xmath68 is excellent at both 180 and 240 mev ; the observed values of @xmath69 slightly exceed those predicted .
we can further examine the rule by comparing the @xmath4 and @xmath17 reactions in the @xmath70 nuclei . in @xmath16li and @xmath1be ,
the extra neutron should enhance the @xmath4 reaction by providing more targets for the charge - exchange reactions and suppress the @xmath17 reaction by providing more targets for the competing reactions .
( 2 ) predicts ratios of @xmath4 to @xmath17 of 2.67 in @xmath16li and 2.08 in @xmath1be .
the experimental ratios of the total cross sections , given in table vi , are seen to agree well with these predictions at 240 mev but to exceed them at 180 mev . to avoid additional uncertainties and assumptions , the ratios in tables v and vi
were obtained from the measured cross sections without the subtraction of pipp .
it was found that performing this subtraction does not alter the ratios outside their uncertainties and thus does not alter the conclusions drawn .
in the intranuclear cascade calculation of oset and collaborators @xcite , the probabilities for pion quasi - elastic scattering and absorption are computed using a microscopic model in which the pion is propagated classically through the nucleus .
quasi - elastic single charge exchange as well as non - charge exchange interactions are allowed , and thus this model can be employed to generate dcx cross sections based on the sequential single charge exchange mechanism .
the calculation is performed in infinite nuclear matter , invoking the local density approximation to evaluate the reaction probabilities .
further details of the calculation may be found in ref .
results of this calculation have previously been compared with dcx data on @xmath3o , @xmath9ca , @xmath10rh , and @xmath11pb @xcite .
@xcite also contains some comparisons with results of the present experiment for @xmath1be and @xmath2c .
the predictions of this model are compared with the present data for @xmath15li@xmath4 and @xmath16li@xmath4 at 240 mev in figs .
10 and 11 , respectively .
a persistent feature of the calculations is a low - energy peak that is not seen in the data .
a hint of the double - peaked structure seen in the forward - angle spectra does appear in the calculation , but the high energy peak is at too high an energy and of insufficient amplitude to match that observed .
comparison of this calculation with data on @xmath1be shows qualitatively similar results
see fig .
6 of ref . @xcite .
if one attributes the measured strength in the low - energy region of the forward angle spectra to pipp , as suggested by the work of alqadi and gibbs @xcite , then the disagreement between the measured and calculated doubly - differential cross sections is even more striking . given the disparities in the measured and calculated doubly differential cross sections , comparison of the integrated cross sections may not be very meaningful .
however , it can be seen by inspection of the energy spectra that the magnitude of the cross section is better reproduced by the theory for @xmath16li than for @xmath15li .
12 shows the theory and measurement for @xmath16li@xmath17 at 240 mev .
the comparison of theory and experiment is qualitatively similar to that for @xmath16li@xmath4 .
calculations have also been performed @xcite at 180 mev and 120 mev .
although the calculation reproduces the approximate isotropy of the cross sections at these lower energies , quantitative comparisons are on the whole no more successful than at the higher energy .
but perhaps it is not surprising that a calculation designed for heavy nuclei can not predict the details of the dcx process in these very light nuclei .
vicente - vacas and oset @xcite comment that a major weakness of their calculation is the fact that the model @xcite predicts narrower quasi - elastic scattering peaks than are observed , and that this might have a bearing on the discrepancy seen between theory and measurement for light nuclei .
we next compare our data with the predictions of a model developed by kinney @xcite based on the sequential single charge exchange mechanism .
this treatment is an extension of some of the work of van loon @xcite and relies on the formalism developed by thies @xcite to describe inelastic , multistep reactions with quantum - mechanical transport theory .
the essential features of the calculation are discussed in ref .
@xcite ; an extended description is contained in ref . @xcite . in this model
the incident pion interacts sequentially with two like - charge nucleons only , and thus only the leading , or `` double scattering '' , term in the transition matrix is used , _
i.e. _ , @xmath71 where the @xmath72 are the in - medium transition operators ( @xmath73-matrices ) for scattering from the @xmath74th nucleon , and @xmath75 is the in - medium pion propagator .
what distinguishes this calculation from simpler folding models @xcite is the non - static treatment of the @xmath76 interaction , which is important because in the energy region of the @xmath57 resonance , the interaction between the pion and the nucleon varies strongly with pion - nucleon relative energy .
also , the binding of the nucleons is taken into account via an approximate prescription .
the predictions of this model have been compared with dcx data on @xmath6he @xcite and @xmath8he @xcite .
here we extend the calculations to @xmath15li , and accordingly must modify the initial state wave function to take into account the presence of the @xmath13-shell proton and neutron .
we describe the @xmath15li nucleus by a product of @xmath77 and @xmath78 harmonic oscillator wave functions : @xmath79 .
the dcx amplitude is summed over the three possible sequences of first and second interactions : @xmath80 , @xmath81 , and @xmath82 .
the @xmath80 sequence is seen to contribute about 25% and the combined @xmath81 and @xmath82 sequences about 75% to the total amplitude @xcite . in fig .
13 we compare the results of this calculation with the measured doubly differential cross section at five angles for incident pion energy 240 mev , with two assumptions for the value of the average nuclear potentials @xmath83 and @xmath84 at the first and second scattering , respectively .
the solid curves ( lower branches ) represent the calculation for @xmath85 mev ; the dashed curves ( lower branches ) that for @xmath86 .
the sequential scattering model does indeed predict a high energy peak in the forward angle cross sections that is reasonably close , both in magnitude and in position in energy , to that seen in the data ; the evolution of this peak to larger angles is also quite well reproduced .
whereas for @xmath6he and @xmath8he this calculation predicted a double - peaked spectrum at forward angles that agreed fairly well with the data , the inclusion of a @xmath13-shell nucleon seems to greatly diminish the low - energy peak , compared with the data .
the question arises as to whether the primary origin of the observed low - energy peaks at 240 mev in the light @xmath13-shell dcx data is in fact pipp , as was suggested by alqadi and gibbs @xcite for dcx in @xmath8he .
the upper branches of the solid and dashed curves were obtained by adding the estimated pipp cross sections from fig . 5 ( which assume a phase - space distribution ) to the calculated dcx cross sections .
the agreement with the data is seen to be surprisingly good , and may be fortuitous , given the simplicity of the assumptions made for both dcx and pipp .
the dcx calculations in fig . 13 show very little dependence on the nuclear potential .
however , when the calculation is compared to the data at 25@xmath12 for incident energies 120 , 180 , and 240 mev ( fig .
14 ) , we see that the choice of the potential has a significant effect . whereas the calculation for @xmath86 provides a fairly good representation of the magnitude of the cross section at all three energies , that for @xmath87 mev produces a prediction that is approximately three times too large at 180 mev and five times too large at 120 mev .
the same phenomenon was seen in the comparison of this model with data on @xmath6he and @xmath8he : only by setting both potentials to zero could the magnitude of the predicted cross section be brought close to the data at 120 and 180 mev .
a systematic study of the inclusive dcx process in light @xmath13-shell nuclei has been performed for incident pion energies between 120 mev and 240 mev .
the relative magnitudes of the cross sections were found to be roughly consistent with a `` scaling rule''@xcite derived by assuming dcx to proceed via two sequential single charge exchange ( sscx ) reactions in competition with other processes .
the data are compared with the predictions of two theoretical models : an intranuclear cascade model of oset and collaborators @xcite and a model developed by kinney @xcite based on the sscx mechanism .
although one would expect the model of oset , which allows multiple interactions , to provide a better reproduction of the data , this does not appear to be the case .
in particular , the strength at low outgoing pion energies , which one would expect to arise from multiple interactions , is calculated to be much larger than that seen in the data . either the effect of these interactions has been overestimated , or one or more of the approximations in this calculation has produced a distortion of the spectrum .
the calculation of kinney @xcite provides an overall better representation of the data for @xmath15li , particularly if pipp is included .
although a double - peaked structure in the forward angle spectra is expected if sscx is the dominant mechanism , the structure seen for @xmath15li and @xmath16li at 240 mev probably contains a contribution from pipp .
we note that the data at 180 mev for these nuclei do not exhibit a low - energy peak .
however , pipp can not provide the complete explanation , since for @xmath6he and @xmath8he the low - energy peak persists to energies well below the pipp threshold .
we suggest that both sscx and pipp mechanisms are at work to produce pions of charge opposite to that of the incident beam .
the double peak signature of sscx appears clearly in @xmath6he and @xmath8he owing to the simplicity of these nuclei .
as @xmath52 increases , the presence of additional nucleons affords the possibility of more than two interactions , and `` washes out '' this structure . in conclusion ,
the work reported in this paper completes a systematic study of the @xmath52-dependence of the inclusive double charge exchange process at intermediate energies .
results for lighter @xcite and heavier @xcite nuclei have been published previously .
although many qualitative features of the inclusive dcx process are now fairly well understood , a quantitative theoretical description of the measured cross sections has yet to be achieved .
this work was supported in part by funds provided by the u.s .
department of energy .
we acknowledge the contributions of d. cors , s. hibrten , p. mansky , and c. schermer to the data - taking phase of this experiment .
we wish to thank e. oset for making his code available to us and for guidance in its use .
we are grateful to m. yuly for insightful discussions of the data analysis procedures and for a careful reading of the final manuscript .
t. akdogan provided valuable assistance in the preparation of the figures .
b. m. abramov , l. alvarez - ruso , yu .
a. borodin , s. a. bulychjov , m. j. vicente vacas , i. a. dukhovskoy , a. p. krutenkova , v. v. kulikov , m. a. matsyuk , and e. n. turdakina , yadernaya fizika * 68 * , 1336 ( 2005 ) [ physics of atomic nuclei * 68 * , 1283 ( 2005 ) ] .
e. piasetzky , p. a. m. gram , d. w. macarthur , g. a. rebka , jr .
, c. a. bordner , s. hoibraten , e. r. kinney , j. l. matthews , s. a. wood , and j. lichtenstadt , phys .
lett . * 53 * , 540 ( 1984 ) ; j. lichtenstadt , d. ashery , s. a. wood , e. piasetzky , p. a. m. gram , d. w. macarthur , r. s. bhalerao , l. c. liu , g. a. rebka , jr . and d. roberts , phys .
c * 33 * , 655 ( 1986 ) .
m. vicente - vacas and e. oset , in _ second lampf international workshop on pion - nucleus double charge exchange _ , los alamos , 1989 , edited by w. r. gibbs and m. j. leitch ( world scientific , singapore , 1990 ) , p. 120 .
c|ccc + incident beam & normalization & angle - dependent@xmath88 & overall@xmath89 + 120 mev @xmath18 & 4.8 & 1.3 & 5.0 + 120 mev @xmath19 & 5.1 & 5.3 & 7.4 + 180 mev @xmath18 & 6.5 & 5.1 & 8.3 + 180 mev @xmath19 & 6.4 & 1.0 & 6.4 + 240 mev @xmath18 & 6.2 & 1.9 & 6.5 + 240 mev @xmath19 & 6.8 & 1.6 & 7.0 + .estimated cross sections ( @xmath90b ) for pion - induced pion production by 240 mev incident pions .
the result for @xmath8he is taken from ref .
the quoted uncertainties reflect only the uncertainty in the @xmath91 cross section , except in the case of @xmath16li@xmath92 where there is an additional uncertainty arising from the determination of @xmath42 . [
cols="^,^,^",options="header " , ] | we report the results of a series of measurements of the differential cross sections for inclusive pion double charge exchange in @xmath0li , @xmath1be , and @xmath2c for positive and negative incident pions of energies 120 , 180 , and 240 mev .
the data are compared with the predictions of an intranuclear cascade model and a model based on two sequential single charge exchange processes .
= 9.0 true in = 6.5 true in |
our understanding of the structural and evolutionary properties of close - in extrasolar planets ( radius , mass , density ) is continuously improved by new detections of transiting planets .
fourteen such systems are known to datepont / transits.htm . ]
. the accelerated rate at which such systems have been discovered of late suggests the prospects are bright for transit - search projects , as well as for the possibility of critically testing physical models of hot jupiters in the near future based on statistically significant ensemble properties of transiting planet systems ( for a review see @xcite @xcite ) .
the accurate determination of the physical properties of transiting exoplanets depends critically upon our knowledge of a number of basic parameters of the parent stars .
in particular , the mass and radius of a planet , which are of fundamental importance for testing theoretical predictions of planetary structure ( e.g. , @xcite @xcite ; @xcite @xcite ; @xcite @xcite ; @xcite @xcite . for a review
see for example @xcite @xcite and @xcite @xcite , and references therein ) , depend rather directly on the mass and radius of the parent star , placing strict demands on the accuracy of the latter .
evidence for correlations between transiting planet properties and stellar characteristics such as metallicity ( @xcite @xcite ; @xcite @xcite ) , and their implications for competing giant planet formation models ( e.g. , @xcite @xcite , @xcite ; @xcite @xcite ; @xcite @xcite ; @xcite @xcite , @xcite ; @xcite @xcite ) , relies in turn on the accurate determination of chemical abundances of the host stars .
attempts to constrain the amount of mass loss experienced by hot and very - hot jupiters ( @xcite @xcite ) , to refine our knowledge of their relative frequencies ( e.g. , @xcite @xcite ) as well as to compare observations with theoretical evaporation rates of insolated giant planets ( @xcite @xcite ; @xcite @xcite ; @xcite @xcite ; @xcite @xcite ) are severely affected by large uncertainties in the determination of stellar ages , which can become pathological for field stars , depending on spectral type .
finally , the accurate determination of parent star parameters requires particular attention in cases in which a direct distance estimate to the system ( trigonometric parallax ) is unavailable . for about 2/3 of the presently known nearby transiting systems ( @xmath13 pc ) , and for over 3/4 of the full sample , such measurements are not available at the present time , and will only be made possible by future high - precision astrometric observatories , both on the ground and in space ( e.g. , @xcite @xcite , and references therein ) . among the recently discovered transiting extrasolar planets , ( @xcite @xcite )
is the first detected in the field of view of the _ kepler _ mission ( @xcite @xcite ) .
it has the largest impact parameter , and is the most massive planet , among the currently known nearby transiting systems ( @xcite @xcite ) . in this work
we report on a detailed spectroscopic determination of the properties of the parent star of , including the effective temperature and surface gravity , as well as the chemical abundances of iron and lithium .
we also measure the chromospheric activity and provide constraints on the age of the system from this and other indicators . because of the importance of the effective temperature for deriving other stellar characteristics , we have made an effort to provide several external checks on its accuracy .
we then use these properties along with other constraints from the light curve analysis of holman et al .
@xcite to infer the mass and radius of the star with realistic uncertainties .
in particular , we show how the use of the _ stellar density _ obtained directly from the light curve fit is in this case a much better proxy for luminosity than the spectroscopic surface gravity , typically used in cases such as this in which the parallax is unknown . our new stellar parameters in turn lead to improved values for the mass and radius of the planet over those reported by odonovan et al .
we conclude by providing a summary of our results and by revisiting some of the evidence connecting the properties of close - in extrasolar planets to the characteristics of their parent stars .
the spectroscopic observations used here are the same as described previously by odonovan et al .
briefly , they consist of twelve echelle spectra obtained with the hires spectrograph on the keck 1 telescope ( @xcite @xcite ) during the summer of 2006 , with a nominal resolving power @xmath14 .
eleven of these spectra were obtained with an i@xmath15 cell placed in front of the slit to provide a wavelength fiducial for high - precision velocity determinations ( see , e.g. , @xcite @xcite ) , with typical exposure times of 15 min resulting in an average @xmath16 pixel@xmath17 . one additional spectrum was obtained without the cell for use as a template .
the effective wavelength coverage provided by the three - ccd array of hires is @xmath1832008800 .
additionally we used 5 echelle spectra obtained with the center for astrophysics ( cfa ) digital speedometers ( @xcite @xcite ) , which cover 45centered at 5187 at a resolving power @xmath19 , and have @xmath20 ratios ranging from 10 to 15 per resolution element .
a detailed analysis of the template spectrum obtained with the keck telescope was carried out following the same procedures described in detail by sozzetti et al .
( @xcite , @xcite , and references therein ) in order to determine the effective temperature ( @xmath21 ) , surface gravity ( @xmath4 ) , and iron abundance [ fe / h ] of .
a set of 30 relatively weak lines of and 4 of were selected , and equivalent widths ( ews ) were measured using the splot task in iraf .
metal abundances are derived under the assumption of local thermodynamic equilibrium ( lte ) , using the 2002 version of the spectral synthesis code moog ( @xcite @xcite ) , a grid of kurucz atlas plane - parallel model stellar atmospheres ( @xcite @xcite ) , and imposing excitation and ionization equilibrium .
we obtained @xmath22 k , @xmath23 , @xmath24 km s@xmath17 , and [ fe / h ] = @xmath25 .
the uncertainties in the first three parameters were estimated following the prescriptions of neuforge & magain @xcite and gonzalez & vanture @xcite , and were rounded off to the nearest 25 k in @xmath3 , 0.1 dex in @xmath4 , and 0.05 km s@xmath17 in @xmath26 . for [ fe / h ]
the uncertainty given corresponds to the scatter obtained from the lines rather than the formal error of the mean , since we consider the latter to be unrealistically small in this case .
no significant departures from lte are expected for a star with the temperature and metallicity of ( e.g. , @xcite @xcite ) , so for the purpose of this study we have not included non - lte effects in our spectroscopic analysis .
we also quantified the sensitivity of our iron abundance determination to variations of @xmath27 with respect to the nominal @xmath3 , @xmath4 , and @xmath26 values , and found changes in [ fe / h ] of at most 0.06 dex , below the adopted uncertainty of 0.1 dex .
finally , we determined also the projected rotational velocity as @xmath28 km s@xmath17 , based on the synthesis of a set of unblended lines , following gonzalez @xcite .
the new values for the stellar parameters are consistent with the spectral type implied by the colors ( @xcite @xcite ) , and are in generally good agreement with those presented by those authors .
we note , however , that our @xmath3 value is somewhat lower , possibly due to the fact that odonovan et al .
@xcite assumed solar metallicity in their study , whereas our analysis indicates a slightly metal - deficient composition .
given the importance of the temperature determination for establishing the absolute mass and radius of the parent star of , we present in this section a number of other consistency checks that illustrate the reliability and accuracy of our estimate above .
cross - correlation of our cfa spectra against a large library of synthetic templates in the manner described by odonovan et al .
@xcite provides an independent estimate of the photospheric properties of the star . by testing all combinations of the four main parameters of these templates ( @xmath3 , @xmath29 , metallicity
[ m / h ] , and @xmath30 ) we seek to maximize the correlation averaged over the 5 available spectra . in principle
this allows the determination of the four parameters , although in practice the narrow wavelength coverage of only 45 results in strong correlations between some of those properties .
we therefore determined the first two of these quantities along with @xmath30 for fixed metallicities of [ m / h ] = 0.0 ( solar ) and [ m / h ] @xmath31 , which bracket our determination in
[ sec : atmospheric ] , and then interpolated to [ m / h ] @xmath32 .
the results are @xmath33 k , @xmath34 , and @xmath35 km s@xmath17 , which supersede the values given by odonovan et al .
@xcite that were based on a preliminary analysis .
the temperature is only 60 k lower than our determination in [ sec : atmospheric ] , well within the errors .
there is also excellent agreement in @xmath4 and @xmath30 . among the many diagnostics available for effective temperature estimation , the ratio of the depths of two spectral lines having different sensitivity to temperature is among the most powerful .
indeed , the line - depth ratio ( ldr ) technique allows the measurement of temperature _ differences _ of the order of a few kelvin in favorable cases ( @xcite @xcite ; @xcite @xcite ; @xcite @xcite ; @xcite @xcite ; @xcite @xcite ; @xcite @xcite ; @xcite @xcite ; @xcite @xcite ) , a much higher accuracy than currently possible with other methods that seek to determine temperatures on an absolute scale .
the latter typically have uncertainties of at least 50100 k. absolute temperature determinations with the ldr method still require translation to an absolute scale , usually through color / temperature calibrations since the ldr measurements are typically calibrated first against color indices , which are directly observable . in order to provide a further check of our spectroscopic @xmath3 determination , we selected in our keck template spectrum 9 pairs of temperature - sensitive lines from the list of biazzo et al .
@xcite in the red part of the spectrum ( @xmath366280 ) .
these authors provide calibrations between ldrs and temperature for a range of rotational velocities ( those for zero rotation were adopted here ) , which include corrections for surface gravity and are valid in the range 3800 k @xmath37 k. metallicity effects are negligible for stars near the solar abundance ( @xcite @xcite ; @xcite @xcite ) . the absolute temperatures derived in this way
rely implicitly on an intermediate calibration between the @xmath38 color and @xmath3 adopted from gray @xcite . in order to provide consistency with the calibrations we use below , we have converted each of our 9 temperature estimates back into @xmath38 and then into temperature again adopting the calibrations by ramrez & melndez @xcite , which have been compared carefully against absolute temperature determinations using the infrared flux method .
the average temperature we obtain in this way is @xmath39 k , in agreement with our estimates above .
the strong sensitivity of the wings of the h@xmath40 line profile to temperature variations , as well as the relatively weak sensitivity to changes in surface gravity and metal abundance , make this feature a very useful temperature indicator for solar - type dwarfs ( e.g. , @xcite @xcite ; @xcite @xcite ) .
the core of the line , however , is formed higher up in the atmosphere under conditions that violate lte , so is not useful here . as an additional consistency check on @xmath3
we therefore compared the h@xmath40 line profile outside of the core in our keck template spectrum against synthetic profiles for solar - metallicity dwarfs ( [ m / h ] = 0.0 , @xmath41 ) from the kurucz database . in figure [ halpha ]
we show the normalized flux in a 10 region centered on h@xmath40 , and four calculated profiles for different values of @xmath3 .
temperatures significantly hotter than 6000 k or significantly cooler than 5750 k appear inconsistent with the observed profile , and suggest the optimal value is somewhere in between . because of the difficulties in the placement of the continuum for such a broad line in an echelle spectrum , we view this comparison only as a rough check .
nevertheless , it agrees once again with our previous estimates .
photometric measurements for are available in the johnson , cousins , _ tycho _ , and 2mass systems , as listed by odonovan et al .
@xcite . based on these data and assuming zero reddening we have computed 7 different color indices ( not all completely independent of each other ) and applied the calibrations of ramrez & melndez @xcite for a fixed metallicity of [ fe / h ] @xmath32 to obtain effective temperatures .
there is excellent agreement between these estimates , and the average @xmath3 is @xmath42 k. this is 170 k cooler than our spectroscopic determination in
[ sec : atmospheric ] , a difference that is significantly larger than allowed by the combined uncertainties .
however , in view of the distance to the object of @xmath18230 pc ( @xcite @xcite ; see also below ) , a small amount of reddening would not be entirely unexpected , and was also suspected by those authors .
we find that if we applied a correction to the individual color indices corresponding to @xmath43 mag , the average temperature would agree exactly with our spectroscopic value .
indirect support for this amount of interstellar material is provided by the reddening maps of burstein & heiles @xcite and schlegel et al .
@xcite , which indicate total @xmath44 values along the line of sight to of @xmath180.05 mag and @xmath180.07 mag , respectively .
for the distance of and using the drimmel & spergel @xcite model of galactic dust distribution these numbers are reduced to @xmath45 and 0.03 mag , respectively .
further support is given by the comparison between the observed colors and those predicted from stellar evolution models described below , which gives @xmath46 . from the good agreement between our spectroscopic temperature determination in
[ sec : atmospheric ] and the estimates from the cfa spectra and the ldrs we conclude that the @xmath3 of is accurately established and we make use of it in [ sec : physics ] to infer the mass and radius of the star .
age determination for individual stars in the field is a difficult task .
a variety of indicators is available , such as h@xmath40 emission , x - ray activity , lithium abundance , h & k emission , asteroseismology , rotation , and galactic space motion , some more constraining than others depending on the particular case . while the relative agreement among multiple methods allows in principle for fairly reliable dating of stars with ages comparable to the age of the hyades or younger , constraints on ages for individual stars with @xmath47 gyr are usually quite weak .
here we have used the activity indicator and the lithium abundance as measured in our hires spectra of in an attempt to inform the theoretical models used for the determination of the stellar mass and radius ( see next section ) . in the top panel of figure [ cali ]
we show a region of the hires template spectrum centered on the h line .
no significant emission feature is present , and the same is true for the other keck spectra we collected using the i@xmath15 cell ( in which the iodine lines do not interfere because there are none shortward of @xmath185000 ) . following the procedure outlined in sozzetti et al .
@xcite , we have measured the chromospheric activity index s ( e.g. , duncan et al .
1991 ) from the h and k lines in our spectra , and then converted it into the chromospheric emission ratio @xmath48 , corrected for the photospheric contribution . for , the mount wilson index , averaged over all our spectra , is @xmath49 , and the resulting @xmath50 ( with formal uncertainties calculated from the scatter of individual measurements ) , suggesting a quite inactive star .
the resulting chromospheric age estimate , using the relations summarized in wright et al .
@xcite , is @xmath51 gyr .
we point out , however , that this estimate should not be taken at face value for a number of reasons .
firstly , it has been shown ( e.g. , @xcite @xcite ) that chromospheric activity can only be considered a reliable age estimator up to @xmath52 gyr , and that for stars showing low activity levels the ages derived are only lower limits . secondly , because stars have activity cycles like the sun , the proper measure of the chromospheric flux to use in determining the age of an individual star is an average over the entire magnetic cycle rather than a quasi - instantaneous value such as is available to us , to avoid the possibility of finding a star in a maunder minimum phase ( @xcite @xcite ; @xcite @xcite ) .
finally , there are hints ( @xcite @xcite ) that chromospheric age estimates tend to be systematically older than those derived with other methods , suggesting perhaps the need for re - calibration of the activity - based ages .
therefore , all we can claim here is that the lower limit for the chromospheric age of is a few gyr . displays a significant li @xmath536707.8 absorption feature .
we have carried out a detailed spectral synthesis of a 10 region of the keck template spectrum centered on this line , using the atmospheric parameters derived from the fe - line analysis and the line list of reddy et al . @xcite . in the bottom panel of figure [ cali ]
we show the comparison of the spectrum of with three synthetic spectra , each differing only in the li abundance assumed .
the best - fit model results in an abundance of @xmath54 . as shown in figure [ licomp ]
, the li abundance we obtain does not appear out of the ordinary in relation to those of other planet host stars of similar temperature , as measured by israelian et al .
@xcite ( once typical uncertainties of the order of 50 - 100 k on the @xmath3 determinations are considered ) . by comparison with li abundance curves as a function of effective temperature for clusters of different ages ( @xcite @xcite ) we infer for an age of about 12 gyr . this value would point to a somewhat younger age than the one inferred from the measurements .
however , it is not uncommon to observe a large spread in li abundance among stars in the same cluster that appear otherwise identical ( @xcite @xcite ) , so also in this case it seems safer to simply report a lower limit to the age of of 12 gyr .
, as well as the apparent discrepancy between the youth indicators .
the measurement of statistically significant trends of element abundance with condensation temperature @xmath55 ( e.g. , sozzetti et al .
2006 ; gonzalez 2005 , 2006 , and references therein ) or detection of the @xmath56li isotope ( e.g. , israelian et al .
2001 , 2003 ; gonzalez 2006 , and references therein ) in the atmosphere of tres-2 would be strong evidence in support of the self - enrichment scenario for this star .
further spectroscopic measurements of tres-2 are thus clearly encouraged ] finally , another argument for the star not being particularly young is given by the small projected rotational velocity we measure here ( see [ sec : atmospheric ] and [ sec : cfa ] ) .
this would indicate once again an age of @xmath47 gyr ( @xcite @xcite ; @xcite @xcite ) .
a common procedure for deriving the absolute mass and radius of planet host stars , needed to infer those of the transiting object , is to compare the measured stellar properties such as temperature and luminosity with stellar evolution models in the h - r diagram , or in some equivalent parameter space .
because the distance to is not precisely known ( it was not observed during the _ hipparcos _ mission ) , we do not have direct access to its luminosity .
an alternative measure of intrinsic brightness ( or evolution ) that has been used in the past is the spectroscopically determined value of @xmath4 ( see , e.g. , @xcite @xcite , @xcite , @xcite ; @xcite @xcite ; @xcite @xcite ; @xcite @xcite ) .
surface gravities are typically very difficult to determine accurately in this way , and as a result the constraint on the stellar radius is relatively weak . here we explore in detail the possibility of using other information available in transiting systems such as , that provide much tighter constraints on the luminosity , as also noted recently by pont et al .
we focus in particular on the quantities obtainable by fitting the transit light curves .
the three main adjustable parameters ( see , e.g. , @xcite @xcite ) are often taken to be the relative radius of the planet ( @xmath57 ) , the impact parameter ( @xmath58 ) , and the normalized separation between the star and the planet ( @xmath5 ) , where @xmath59 is the semimajor axis of the relative orbit and @xmath60 is the inclination to the line of sight .
these are largely independent of the stellar properties , except for a weak dependence on the limb - darkening coefficients ( a second - order effect ) that are typically a function of effective temperature , surface gravity , and composition .
one of these , @xmath5 , contains information intrinsic to the star : using kepler s third law ( as revised by newton ) it can be shown that @xmath61 ( see also @xcite @xcite ) where all quantities are expressed in _
cgs _ units and @xmath62 is the newtonian gravitational constant .
the left - hand side corresponds essentially to the stellar density , @xmath63 .
note that the first term on the right is entirely determined from measurable quantities : the orbital period ( @xmath64 days ; @xcite @xcite ) , and @xmath5 from the light curve fit .
the second term on the right , on the other hand , involves the planetary mass ( which is unknown until the stellar mass is determined ) as well as @xmath65 ( also unknown ) .
however , the size of this second term is typically two to three orders of magnitude smaller than the first for most transiting exoplanets including , so it can safely be ignored for all practical purposes .
thus the density of the star is determined _ directly _ from the observations , with no additional assumptions . in this particular case
the accuracy of the @xmath5 determination ( @xmath66 ; @xcite @xcite ) is very high : the uncertainty is only 1.6% .
the stellar density is a sensitive measure of evolution or luminosity , and as such it provides a very useful constraint on the size of the star .
we use it along with the effective temperature and the measured metallicity to establish the absolute mass and radius of .
we proceed by comparing @xmath63 ( or a closely related quantity ; see below ) and @xmath3 with current stellar evolution models from the yonsei - yale series by yi et al .
@xcite ( see also @xcite @xcite ) .
we explored the agreement with model isochrones calculated over a wide a range of uniformly spaced ages ( 0.19 gyr ) spanning the full range of metallicities allowed by our spectroscopic determination ( [ fe / h ] @xmath67 ) . along each isochrone we computed the theoretical stellar properties using a fine step in mass , and at each of these points we compared those properties with the observations , and recorded all cases yielding a match within the observational errors . in this way we established the range of permitted values of the stellar mass and radius .
all these matching models were assigned the same likelihood for this application . in practice ,
we have chosen to compare the models with the observations directly in the observational plane ( @xmath5 versus @xmath3 , rather than @xmath63 versus @xmath3 ) . therefore , instead of computing the stellar density along the isochrones and comparing it with the measured value of @xmath63 , we computed the theoretical values of @xmath5 ( which is essentially the cube root of the density ) from an expression obtained by rearranging eq .
[ eq : density ] , @xmath68 and compared them with the value resulting from the light - curve fit , @xmath69 ( @xcite @xcite ) . as discussed above , we have ignored here the small contribution from the mass of the planet , @xmath70 . to compute the predicted values of @xmath5 . ] the best match with the models ( which produces virtually perfect agreement with the measured parameters @xmath5 and @xmath3 ) is achieved for a stellar mass of @xmath71 , a radius of @xmath72 , and an age of @xmath8 gyr .
the uncertainties listed reflect the spread allowed by the observational errors in @xmath3 , @xmath5 , and [ fe / h ] , and exclude any systematics in the models themselves , which are difficult to quantify .
the surface gravity of the star for the best fit is @xmath73 , in excellent agreement with our spectroscopically measured value in [ sec : atmospheric ] , and the corresponding metal abundance for this best fit is [ fe / h ] @xmath74 , also virtually the same as the measured value .
the models indicate for the star an age similar to the sun , which is entirely consistent with the lower limits discussed earlier based on the activity indicators and li . the absolute visual magnitude predicted by the models for the adopted mass and age and our measured composition is @xmath75 , and
the color expected is @xmath76 .
the latter , compared with the measured index of @xmath77 ( odonovan et al .
2006 ) suggests a small amount of reddening of @xmath78 mag , not inconsistent with several other estimates discussed in [
sec : photometry ] .
this corresponds to a visual extinction @xmath79 mag .
accounting for this we derive a distance to of 220 pc with an estimated uncertainty of 10 pc . with the mean radial velocity of @xmath80 km s@xmath17 as reported by odonovan et al .
( 2006 ) , and ucac2 ( @xcite @xcite ) proper motion components [ @xmath81 , @xmath82 = [ 4.45 , @xmath833.40 ] mas yr@xmath17 , the galactic space motion vector of the star is [ @xmath84 , @xmath85 , @xmath86 = [ @xmath831.81 , + 0.88 , @xmath835.51 ] km s@xmath17 ( where @xmath84 is taken to be positive toward the galactic anti - center ) .
we collect these results along with other properties derived previously in table [ tab : star ] .
our stellar mass is @xmath1810% smaller than the value of @xmath87 @xmath88 adopted by odonovan et al .
@xcite , which is in part due to our using a lower heavy element abundance , derived from our detailed chemical analysis , as opposed to adopting the solar composition .
the stellar radius , on the other hand , is the same .
the fairly tight constraints we have obtained on the radius of the star ( errors less than 4% ) are the result of using the information on @xmath5 derived from the light - curve fit of holman et al .
@xcite . had we used the spectroscopically determined surface gravity instead , we estimate the constraint would have been some 5 times weaker .
this is illustrated in figure [ fig : iso ] , where the top panel shows yonsei - yale isochrones for ages of 1 gyr to 9 gyr and a heavy element abundance equal to the measured value of [ fe / h ] .
the shaded error box represents the uncertainties in the measured @xmath4 and @xmath3 of .
as seen , the error in gravity is so large as to span the full range of ages shown here , thus providing essentially no useful constraint on age and a weak one on @xmath65 . in the lower panel
we have re - mapped the vertical axis to @xmath5 ( using the measured orbital period @xmath89 ; see eq .
[ eq : ar ] ) .
the error box in this case is significantly smaller , making @xmath5 a much better measure of the luminosity than surface gravity .
we propose that the same procedure should be used in other transiting planets in which the quality of the light curves is sufficient to provide a superior constraint compared to surface gravity .
depending on the case , the accuracy of @xmath5 could be high enough that it may even compare favorably with the constraint afforded by a direct knowledge of the parallax .
the improved knowledge of the mass and radius of the parent star has a direct impact on the accuracy of the planetary parameters of .
we have combined the stellar properties in table [ tab : star ] with the mass function from the spectroscopic orbit of odonovan et al .
@xcite , @xmath90^{2/3 } m_\mathrm{jup}$ ] , and the fitted light - curve parameters from the new photometric analysis of holman et al .
@xcite , which are @xmath91 , @xmath92 , and @xmath69 .
we obtain for the planet @xmath93 and @xmath94 these are some 6% and 1% smaller than the values reported by odonovan et al .
@xcite , respectively .
an important but generally overlooked property of the spectroscopic and photometric solutions for transiting planets is the fact that the surface gravity of the planet is directly measurable from the observations , with no need to know the mass or radius of the parent star ( see , e.g. , @xcite @xcite ) .
the same was pointed out by @xcite ( @xcite ) in the context of mass and radius determinations for small stars in single - lined eclipsing binaries , and also by southworth et al .
( 2004 , 2007 ) .
this derives from the quadratic relation between @xmath70 and @xmath95 that can easily be obtained from the definition of the spectroscopic mass function and kepler s third law : @xmath96 ^ 2 } } \left({a / r_\star \over r_p / r_\star}\right)^2 r_p^2~.\ ] ] in this expression @xmath97 represents the velocity semi - amplitude of the star in response to the pull from the planet ( @xmath98 m s@xmath17 for ; @xcite @xcite ) , and @xmath99 the eccentricity of the orbit , which is usually found to be very close to zero for transiting planets .
the quantities @xmath100 , @xmath5 , and @xmath57 are obtained directly from the light curve analysis , often with very high precision as in our case .
the surface gravity of the planet follows immediately as : @xmath101 ^ 2\right ) + 2 \log\left({a / r_\star \over r_p / r_\star}\right ) + { 1\over 2}\log\left(1-e^2\right)~.\ ] ] the numerical constant is such that the gravity is in _ cgs _ units when @xmath89 and @xmath97 are expressed in their usual units of days and m s@xmath17 . for we obtain @xmath102 , in which the uncertainty includes all contributions from measured quantities .
we call the reader s attention to the very high precision of this determination , which rivals that of the best - known double - lined eclipsing binaries ( see , e.g. , @xcite @xcite ) .
while the planetary masses and radii in transiting systems have typically been the main focus of investigators in this field , and with good reason , those quantities depend critically on the mass and radius of the parent star , which are often the weak link in the chain and usually rely on stellar evolution models .
surface gravities are much closer to the observations , are model - independent , and can often be obtained with very high precision as in the case of . accurately determined surface gravities of planets are potentially important for constraining theoretical calculations of the spectra of extrasolar planets .
these have now begun to be tested through infrared photometry of the secondary eclipses as well as transmission spectroscopy in several cases ( see for example @xcite @xcite , and references therein ) .
our high - resolution , high-@xmath20 spectra from keck / hires have allowed us to derive new and accurate values of the stellar atmospheric parameters of the parent star of the transiting planet , principally the effective temperature and metallicity , which have in turn yielded improved parameters for the star .
the g0v main - sequence dwarf appears to have a metal abundance very similar to the average of the solar neighborhood ( [ fe / h]@xmath103 , see for example @xcite @xcite ) , making it nominally the most metal - deficient case in the current sample of transiting planets ( e.g. , @xcite @xcite ) .
the reliability of our temperature estimate is strongly supported by the results of several independent checks we have carried out ( some more constraining than others ) , all of which are in good agreement ( additional spectroscopy , line depth ratios , and h@xmath40 line profiles ) . the old age ( similar to the sun ) we infer for
is also supported by results from the measurements of the activity level , the lithium abundance , and rotation , which rule out a very young age for the system .
this is consistent with the notion that the planet s inferred mass escape rate ( @xmath104 g s@xmath17 , using the model of @xcite @xcite ) is not very high .
the inferred lifetime ( tens of gyr , well outside the evaporation - forbidden region indicated by lecavelier des etangs @xcite ) is long enough that very efficient evaporation scenarios ( @xcite @xcite ) are not likely to hold in the case of the system .
we have shown that the best constraint on the radius of the parent star comes not from the spectroscopically determined surface gravity ( the quantity most often used for this purpose when a trigonometric parallax is unavailable ) , but from the photometrically determined quantity @xmath5 , which is closely related to the stellar density .
this quantity is directly measured from the transit light curve , and depends only very weakly ( or in some cases , not at all ) on any assumed stellar properties . in this particular case
the gain from using this new constraint is about a factor of five in terms of the precision in @xmath65 .
similar arguments for using @xmath5 instead of @xmath4 were made by pont et al .
@xcite regarding the faint parent star of the transiting planet ogle - tr-10b , although they appear not to have actually applied the method in arriving at the mass and radius of the star reported in their work . with the significant improvements seen recently in the quality of the light curves of several of the known transiting planets ( @xcite @xcite , @xcite ; @xcite @xcite ; @xcite @xcite ; @xcite @xcite , @xcite , @xcite ; @xcite @xcite , @xcite , @xcite ; @xcite @xcite ; @xcite @xcite ; @xcite @xcite ; @xcite @xcite ) , the measured @xmath5 values are likely to be much better now than in the original discovery papers .
thus , it may pay to revisit the determination of stellar parameters of many of these systems along the lines of what we have done here , since this should result in significant improvements in the absolute mass and radius estimates of the attending exoplanets as well .
such a study is underway by a subset of the present authors . our improved stellar parameters yielding more precise values for
the planet mass and radius confirm that is the most massive among the currently known nearby ( @xmath11 pc ) transiting hot jupiters .
although the surface gravity of transiting planets , @xmath105 , has not usually attracted much attention , we point out here that the little - used quadratic relation between the mass and radius of a transiting planet allows the determination of this property of the object purely from observations , free from assumptions about the mass and radius of the parent star .
the high accuracy with which @xmath106 can be determined in and other cases makes it a potentially useful constraint to theory . among the fourteen extrasolar giant planets known to transit their parent stars , seems to belong to an increasing family of objects ( hd 209458b , hat - p-1b , wasp-1b ) whose measured radii apparently disagree with published theoretical models , in that they are larger than expected . in the past , this anomaly was explained invoking a source of internal heat ( e.g. , @xcite @xcite , @xcite ; @xcite @xcite ; @xcite
@xcite ) , either through eccentricity pumping by more distant companions or through persisting obliquity tides . however , these scenarios have some difficulties ( @xcite @xcite ; @xcite @xcite ) .
more recent work
( @xcite @xcite ) indicates that the discrepancy may be due instead to super - solar metallicities and opacities that naturally retain internal heat , thus helping to keep a hot jupiter s radius larger for longer times .
the smaller - than - expected radii of other transiting planets ( e.g. , hd 149026b ) can instead be explained by the presence of dense rocky cores , and a correlation has been proposed between inner core masses and host star metallicities ( @xcite @xcite ; @xcite @xcite ) . in this respect ,
the very small inferred core mass for ( @xcite @xcite ) agrees well with the lower - than - solar iron abundance measured for the parent star . considering the rapidly growing sample of transiting giant planet systems as an ensemble , other suggestive , albeit preliminary
, trends between different planet and/or stellar properties have begun to emerge from the data .
some are easier to explain within the context of the present theoretical framework , while others still await explanation .
for example , the apparent trend ( with two outliers ) seen in figure [ corr ] , left panel , of increasing planet radius with increasing mass of the parent star could be due in part to the fact that close - in planets orbiting more massive stars are more strongly irradiated ( @xcite @xcite ) . on the other hand , the trend of decreasing planet mass with increasing orbital period ( figure [ corr ] , right panel ) , first highlighted by mazeh et al .
@xcite , appears to be less well understood .
we note , however , that in this case the fact that the host sample is composed of relatively bright , nearby dwarfs targeted by wide - field transit surveys as well as fainter , more distant ogle targets may play some role ( see figure caption for details ) .
the above relations suggest there is a strong interplay between planet properties and host star characteristics , which is hardly unexpected , but it is also clear that the parameter space of properties to be investigated is quite large .
observations of transiting planet systems can best inform structural and evolutionary models when they yield accurate determinations of both planet and stellar properties , through high - precision photometric as well as spectroscopic measurements such as those presented here for the system .
we thank k. biazzo ( catania astrophysical observatory ) for providing the ldr calibrations in advance of publication , and a. burrows and a. spagna for helpful discussions .
gt acknowledges partial support for this work from nasa origins grant nng04lg89 g .
dc is supported in part by nasa origins grant nng05gj29 g . as gratefully acknowledges the kepler mission for partial support under nasa cooperative agreement ncc 2 - 1390 .
jbl gratefully acknowledges support from nsf grant ast-0307340 .
some of the data presented herein were obtained at the w.m .
keck observatory , which is operated as a scientific partnership among the california institute of technology , the university of california and the national aeronautics and space administration .
the observatory was made possible by the generous financial support of the w.m .
keck foundation .
the authors wish to recognize and acknowledge the very significant cultural role and reverence that the summit of mauna kea has always had within the indigenous hawaiian community .
we are most fortunate to have the opportunity to conduct observations from this mountain .
this research has made use of nasa s astrophysics data system abstract service and of the simbad database , operated at cds , strasbourg , france .
alibert , y. , mordasini , c. , benz , w. , & winisdoerffer , c. 2005 , , 434 , 343 andersen , j. 1991 , a&arv , 3 , 91 bakos , g. . , et al .
2006 , , 650 , 1160 baraffe , i. , chabrier , g. , barman , t. , allard , f. , & hauschildt , p. h. 2003 , , 402 , 701 baraffe , i. , selsis , f. , chabrier , g. , barman , t. , allard , f. , hauschildt , p. h. , & lammer , h. 2004 , , 419 , l16 barklem , p. s. , stempels , h. c. , allende prieto , c. , kochukhov , o. p. , piskunov , n. , & omara , b. j. 2002 , , 385 , 951 bodenheimer , p. , lin , d. n. c. , & mardling , r. a. 2001 , , 548 , 466 bodenheimer , p. , laughlin , g. , & lin , d. n. c. 2003 , , 592 , 555 borucki , w. j. , et al .
2003 , , 4854 , 129 boss , a. p. 2000 , , 536
, l101 boss , a. p. 2002
, , 567 , l149 bouchy , f. , pont , f. , melo , c. , santos , n. c. , mayor , m. , queloz , d. , & udry , s. 2005 , , 431 , 1105 bouvier , j. 1997 , mmsait , 68 , 881 burrows , a. 2005 , , 433 , 261 burrows , a. , hubeny , i. , budaj , j. , & hubbard , w. b. 2007 , apj , submitted ( astro - ph/0612703 ) burstein , d. , & heiles , c. 1982 , , 87 , 1165 butler , r. p. , marcy , g. w. , williams , e. , mccarthy , c. , dosanjh , p. , & vogt , s. s. 1996 , , 108 , 500 caccin , b. , penza , v. , & gomez , m. t. 2002 , , 386 , 286 catalano , s. , biazzo , k. , frasca , a. , & marilli , e. 2002 , , 394 , 1009 chabrier , g. , barman , t. , baraffe , i. , allard , f. , & hauschildt , p. h. 2004 , , 603 , l53 charbonneau , d. , et al .
2006 , , 636 , 445 charbonneau , d. , brown , t. m. , burrows , a. , & laughlin , g. 2007a , in protostars and planets v , b. reipurth , d. jewitt , and k. keil ( eds . ) , university of arizona press ( tucson ) , 701 charbonneau , d. , winn , j. n. , everett , m. e. , latham , d. w. , holman , m. j. , esquerdo , g. a. , & odonovan , f. t. 2007b , apj , in press ( astro - ph/0610589 ) demarque , p. , woo , j .- h .
, kim , y .- c . , &
yi , s. k. 2004 , , 155 , 667 diaz , r. f. , et al .
2007 , , accepted ( astro - ph/0701345 ) drimmel , r. , & spergel , d. n. 2001 , , 556 , 181 duncan , d. k. , et al .
1991 , , 76 , 383 fuhrmann , k. , axer , m. , & gehren , t. 1993 , , 271 , 451 gaudi , b. s. , seager , s. , & mall ' en - ornelas , g. 2005 , , 623 , 472 gaudi , b. s. 2005 , , 628 , l73 gaudi , b. s. 2007 , in transiting extrasolar planets workshop , c. afonso , d. weldrake , & t. henning ( eds . ) , asp conf . ser . in press ( astro - ph/0612141 ) gillon , m. , pont , f. , moutou , c. , bouchy , f. , courbin , f. , sohy , s. , & magain , p. 2006
, , 459 , 249 gonzalez , g. 1998 , , 334 , 221 gonzalez , g. , & vanture , a. d. 1998 , , 339 , l29 gonzalez , g. 2005 , , 367 , l37 gonzalez , g. 2006 , , 118 , 1494 gould , a. , dorsher , s. , gaudi , b. s. , & udalski , a. 2006 , aca , 56 , 1 gray , d. f. , & johanson , h. l. 1991 , , 103 , 439 gray , d. f. 1994 , , 106 , 1248 gray , d. f. , & brown , k. 2001 , , 113 , 723 gray , d. f. 2005 , the observation and analysis of stellar photospheres , 3rd edition , cambridge , uk : cambridge university press guillot , t. , & showman , a. p. 2002
, , 385 , 156 guillot , t. 2005 , annual review of earth and planetary sciences , 33 , 493 guillot , t. , santos , n. c. , pont , f. , iro , n. , melo , c. , & ribas , i. 2006 , , 453 , l21 henry , t. j. , soderblom , d. r. , donahue , r. a. , & baliunas , s. l. 1996 , , 111 , 439 holman , m. j. , et al .
2006 , , 652 , 1715 holman , m. j. , et al .
2007a , , 655 , 1103 holman , m. j. , et al . 2007b , , submitted ida s. , & lin d. n. c. , 2004a , , 604 , 388 ida s. , & lin d. n. c. , 2004b , , 616 , 567 israelian , g. , santos , n. c. , mayor , m. , & rebolo , r. 2001 , , 411 , 163 israelian , g. , santos , n. c. , mayor , m. , & rebolo , r. 2003 , , 405 , 753 israelian , g. , santos , n. c. , mayor , m. , & rebolo , r. 2004 , , 414 , 601 knutson , h. a. , charbonneau , d. , noyes , r. w. , brown , t. m. , & gilliland , r. l. 2007 , , 655 , 564 konacki , m. , torres , g. , jha , s. , & sasselov , d. d. 2003 , , 421 , 507 konacki , m. , et al .
2004 , , 609 , l37 konacki , m. , torres , g. , sasselov , d. d. , & jha , s. 2005 , , 624 , 372 kornet , k. , bodenheimer , p. , rzyczka , m. , & stepinski , t. f. 2005 , , 430 , 1133 kovtyukh , v. v. , soubiran , c. , belik , s. i. , & gorlova , n. i. 2003 , , 411 , 559 kurucz , r. l. 1993 , atlas9 stellar atmosphere programs and 2 km / s grid cdrom , vol .
13 , smithsonian astrophysical observatory lammer , h. , selsis , f. , ribas , i. , guinan , e. f. , bauer , s. j. , & weiss , w. w. 2003 , , 598 , l121 latham , d. w. 1992 , in iau colloquium 135 : complementary approaches to double and multiple star research , h.a .
mcalister & w.i .
hartkopf eds , asp conf .
ser . , 32 , 110 .
laughlin , g. , et al .
2005 , , 629 , l121 lecavelier des etangs , a. , vidal - madjar , a. , mcconnell , j. c. , & hbrard , g. 2004 , , 418 , l1 lecavelier des etangs , a. 2006 , , 461 , 1185 levrard , b. , correia , a. c. m. , chabrier , g. , baraffe , i. , selsis , f. , & laskar , j. 2007 , , 463 , l5 mandel , k. , & agol , e. 2002 , , 580 , l171 mayer l. , quinn t. , wadsley j. , & stadel j. , 2004 , , 609 , 1045 mazeh , t. , zucker , s. , & pont , f. 2005 , , 356 , 995 melo , c. , santos , n. c. , pont , f. , guillot , t. , israelian , g. , mayor , m. , queloz , d. , & udry , s. 2006 , , 460 , 251 minniti , d. , et al .
2007 , , accepted ( astro - ph/0701356 ) moutou , c. , pont , f. , bouchy , f. , & mayor , m. 2004 , , 424 , l31 neuforge - verheecke , c. , & magain , p. 1997
, , 328 , 261 nordstrm , b. , et al .
2004 , a&a , 418 , 989 odonovan , f. t. , et al .
2006 , , 651 , l61 pace , g. , & pasquini , l. 2004 , , 426 , 1021 pont , f. , bouchy , f. , queloz , d. , santos , n. c. , melo , c. , mayor , m. , & udry , s. 2004 , , 426 , l15 pont , f. , et al .
2007 , , submitted ( astro - ph/0610827 ) ramrez , i. , & melndez , j. 2005 , , 626 , 446 randich , s. , sestito , p. , primas , f. , pallavicini , r. , & pasquini , l. 2006 , , 450 , 557 reddy , b. e. , lambert , d. l. , laws , c. , gonzalez , g. , & covey , k. 2002 , , 335 , 1005 santos , n. c. , ecuvillon , a. , israelian , g. , mayor , m. , melo , c. , queloz , d. , udry , s. , ribeiro , j. p. , & jorge , s. 2006 , , 458 , 997 schlegel , d. j. , finkbeiner , d. p. , & davis , m. 1998 , , 500 , 525 seager , s. , & malln - ornelas , g. 2003 , , 585 , 1038 sestito , p. , & randich , s. 2005 , , 442 , 615 sneden , c. a. 1973 , ph.d .
thesis , the university of texas at austin song , i. , zuckerman , b. , & bessell , m. s. 2004 , , 614 , l125 southworth , j. , zucker , s. , maxted , p. f. l. , & smalley , b. 2004 , , 355 , 986 southworth , j. , wheatley , p. j. , & sams , g. 2007 , , in press ( arxiv:0704.1570 ) sozzetti , a. , yong , d. , torres , g. , charbonneau , d. , latham , d. w. , allende prieto , c. , brown , t. m. , carney , b. w. , & laird , j. b. 2004 , , 616 , l167 sozzetti , a. 2005 , , 117 , 1021 sozzetti , a. , yong , d. , carney , b. w. , laird , j. b. , latham , d. w. , & torres , g. 2006 , , 131 , 2274 strassmeier , k. g. , & schordan , p. 2000 , an , 321 , 277 vogt , s. s. , et al . , 1994 , in instrumentation in astronomy viii , d. l. crawford & e. r. craine eds . ,
spie , 2198 , 362 winn , j. n. , & holman , m. j. 2005 , , 628 , l159 winn , j. n. , holman , m. j. , & roussanova , a. 2007a , , in press ( astro - ph/0611404 ) winn , j. n. , et al .
2007b , , accepted (
astro - ph/0612224 ) winn , j. n. , holman , m. j. , & fuentes , c. i. 2007c , , 133 , 11 wright , j. t. , marcy , g. w. , butler , r. p. , & vogt , s. s. 2004 , , 152 , 261 yi , s. k. , demarque , p. , kim , y .- c . , lee , y .- w . , ree , c. h. , lejeune , t. , & barnes , s. 2001 , , 136 , 417 yong , d. , lambert , d. l. , allende prieto , c. , & paulson , d. b. 2004 , , 603 , 697 zacharias , n. , urban , s. e. , zacharias , m. i. , wycoff , g. l. , hall , d. m. , monet , d. g. , & rafferty t. j. 2004 , , 127 , 3043 lc @xmath3 ( k ) & 5850 @xmath107 50 + @xmath4 & 4.4 @xmath107 0.1 + @xmath4 & @xmath108 + @xmath30 ( km s@xmath17 ) & 2 @xmath107 1 + @xmath26 ( km s@xmath17 ) & 1.00 @xmath107 0.05 + @xmath109fe / h@xmath110 $ ] & @xmath830.15 @xmath107 0.10 + @xmath111 & @xmath112 + @xmath113 & 2.65 + @xmath63 ( g @xmath114 ) & 1.375 @xmath107 0.065 + @xmath115 ( @xmath88 ) & @xmath116 + @xmath65 ( @xmath117 ) & @xmath118 + age ( gyr ) & @xmath8 + @xmath119 ( mag ) & @xmath120 + distance ( pc ) & @xmath121 + @xmath84 , @xmath85 , @xmath122 ( km s@xmath17 ) & [ @xmath831.81 , + 0.88 , @xmath835.51 ] + @xmath32 , shown with the observational constraints . _ top : _ the measurement on the vertical axis is the spectroscopically determined value of @xmath4 , which provides only a weak handle on the stellar radius and no useful constraint on age . _
bottom : _ the use of the photometrically determined value of @xmath5 from the light curve analysis of holman et al .
@xcite instead of surface gravity provides a much stronger constraint on the age and radius ( by about a factor of 5 ) .
[ fig : iso],scaledwidth=60.0% ] | we report on a spectroscopic determination of the atmospheric parameters and chemical abundance of the parent star of the recently discovered transiting planet .
a detailed lte analysis of a set of and lines from our keck spectra yields @xmath0 k , @xmath1 , and [ fe / h ] @xmath2 .
several independent checks ( e.g. , additional spectroscopy , line - depth ratios ) confirm the reliability of our spectroscopic @xmath3 estimate .
the mass and radius of the star , needed to determine the properties of the planet , are traditionally inferred by comparison with stellar evolution models using @xmath3 and some measure of the stellar luminosity , such as the spectroscopic surface gravity ( when a trigonometric parallax is unavailable , as in this case ) .
we apply here a new method in which we use instead of @xmath4 the normalized separation @xmath5 ( related to the stellar density ) , which can be determined directly from the light curves of transiting planets with much greater precision . with the @xmath5 value from the light curve analysis of holman et al .
@xcite and our @xmath3 estimate we obtain @xmath6 and @xmath7 , and an evolutionary age of @xmath8 gyr , in good agreement with other constraints based on the strength of the emission in the h & k line cores , the lithium abundance , and rotation .
the new stellar parameters yield improved values for the planetary mass and radius of @xmath9 and @xmath10 , confirming that is the most massive among the currently known nearby ( @xmath11 pc ) transiting hot jupiters .
the surface gravity of the planet , @xmath12 , can be derived independently of the knowledge of the stellar parameters ( i.e. , directly from observations ) , and with a very high precision rivaling that of the best known double - lined eclipsing binaries . |
in 1971 , martinet @xcite showed that every 3-manifold admits a contact structure .
but in the subsequent twenty years , through the work of bennequin @xcite and eliashberg @xcite , it became apparent that not all contact structures are created equal in dimension 3 .
specifically , contact structures fall into one of two classes : tight or overtwisted . in this new light ,
what martinet actually showed was that every 3-manifold admits an overtwisted contact structure . in @xcite eliashberg classified overtwisted contact structures on closed 3-manifolds by proving the weak homotopy equivalence of the space of overtwisted contact structures up to isotopy and the space of 2-plane fields up to homotopy hence overtwisted contact structures could now be understood via homotopy theory . on the other hand ,
it has become apparent that tight contact structures have surprising and deep connections to the topology of 3-manifolds , not limited to homotopy theory .
for example , rudolph @xcite and lisca and mati @xcite found connections with slice knots and slice genus , kronheimer and mrowka @xcite found connections with seiberg - witten theory , and eliashberg and thurston @xcite found connections with foliation theory .
thus , whether or not every 3-manifold admits a tight contact structure became a central question in 3-dimensional contact topology . the first candidate for a 3-manifold without a tight contact structure was the poincar homology sphere @xmath0 with reverse orientation .
the difficulty of constructing a holomorphically fillable contact structure on @xmath1 was highlighted in gompf s paper @xcite .
subsequently lisca @xcite , using techniques from seiberg - witten theory , proved that @xmath1 has no symplectically semi - fillable contact structure . in this paper
we prove the following nonexistence result .
[ thm : main ] there exist no positive tight contact structures on the poincar homology sphere @xmath2 with reverse orientation .
this is the first example of a closed 3-manifold which does not carry a positive tight contact structure .
let @xmath1 be the poincar homology sphere with reverse orientation .
then the connect sum @xmath3 , where @xmath4 is @xmath1 with the opposite orientation , does not carry any tight contact structure , positive or negative .
this follows from theorem [ thm : main ] , since a tight structures on a reducible manifold may be decomposed into tight structures on its summands @xcite
we assume the reader is familiar with the basics ideas of contact topology in dimension 3 ( see for example @xcite ) . a thorough understanding of @xcite would be helpful but we include a brief summary of the ideas and terminology
. the reader might also find @xcite useful for various parts of this section . in this paper the ambient manifold @xmath1 will be an oriented 3-manifold , and
the contact structure @xmath5 will be _ positive _ , _ i.e. _ , given by a 1-form @xmath6 with @xmath7 . throughout this section
we only consider @xmath8 _ tight_. also , when we refer to _ legendrian curves _ we mean closed curves , in contrast to _ legendrian arcs_.
recall an oriented embedded surface @xmath9 in @xmath8 is called if there is a vector field @xmath10 transverse to @xmath9 whose flow preserves @xmath11 perhaps the most important feature of convex surfaces is the _ dividing set_. if @xmath12 is a singular foliation on @xmath9 then a disjoint union of ( properly ) embedded curves @xmath13 is said to @xmath12 if @xmath13 divides @xmath9 into two parts @xmath14 @xmath13 is transverse to @xmath15 and there is a vector field @xmath16 directing @xmath12 and a volume form @xmath17 on @xmath9 such that @xmath18 on @xmath19 and @xmath16 points transversely out of @xmath20 if @xmath6 is a contact 1-form for @xmath5 then the zeros of @xmath21 provide dividing curves for the characteristic foliation @xmath22 it is sometimes useful to keep in mind that the dividing curves are where @xmath10 is tangent to @xmath11 an isotopy @xmath23\to m$ ] of @xmath9 is called if @xmath24 is transversal to @xmath10 for all @xmath25 an important result concerning convex surfaces we will ( implicitly ) be using through out this paper is : [ giroux : main ] let @xmath13 be the dividing set for @xmath26 and @xmath12 another singular foliation on @xmath9 divided by @xmath27 then there is an admissible isotopy @xmath28 of @xmath9 such that @xmath29 @xmath30 and the isotopy is fixed on @xmath27 roughly speaking , this says that the dividing set @xmath13 dictates the geometry of @xmath9 , not the precise characteristic foliation
. we will let @xmath31 denote the number of connected components of @xmath13 .
if there is any ambiguity , we will also write @xmath32 instead of @xmath13 to denote the dividing set of @xmath9 . let @xmath33 and @xmath34 be compact convex surfaces with legendrian boundary , which intersect transversely along a common boundary legendrian curve @xmath35 . the neighborhood of the common boundary legendrian is locally isomorphic to the neighborhood @xmath36 of @xmath37 with coordinates @xmath38 and contact 1-form @xmath39 , for some @xmath40 .
let @xmath41 , @xmath42 , be an annular collar of the boundary component @xmath35 .
we may choose our local model so that @xmath43 and @xmath44 ( or the same with @xmath45 and @xmath46 switched ) .
assuming the former , if we join @xmath33 and @xmath34 along @xmath47 and round the common edge , the resulting surface is convex , and the dividing curve @xmath48 on @xmath33 will connect to the dividing curve @xmath49 on @xmath34 , where @xmath50 .
we now introduce the main idea on which this work is based .
let @xmath35 be a legendrian arc .
a half - disk @xmath51 is called a @xmath35 if @xmath52 is legendrian and consists of two arcs @xmath53 and @xmath54 such that , if we orient @xmath51 , @xmath55 are both positive ( negative ) elliptic singular points in @xmath56 there is a negative ( positive ) elliptic point along @xmath57 , and the singular points along @xmath54 are all positive ( negative ) and alternate between elliptic and hyperbolic .
see figure [ bypass ] .
= 1.3 in we also allow for the degenerate case when @xmath35 is a closed curve and the endpoints of @xmath57 are the same ( _ i.e. _ , @xmath58 ) .
we refer to such a bypass as a .
once an orientation on @xmath51 is fixed , the of the bypass is the sign of the elliptic point on the interior of @xmath59 the reason for the name ` bypass ' is that , instead of traveling along the legendrian arc @xmath35 , we may go around and drive through @xmath60 this has the effect of increasing the twisting of @xmath5 along the legendrian curve .
( here we are using the convention that left twists are negative . )
one can then show : [ thm : bypass ] assume @xmath1 has a convex boundary @xmath9 and there exists a bypass @xmath51 along the legendrian curve @xmath61 then we can find a neighborhood @xmath62 of @xmath63 with @xmath64 , and @xmath65 is related to @xmath32 as shown in figure [ bypassmove ] .
= 1.5 in consider a convex torus @xmath66 . if @xmath5 is tight , then one can show that no dividing curve for @xmath66 bounds a disk .
thus @xmath67 consists of @xmath68 parallel dividing curves with @xmath69 .
assume for now that the dividing curves are all horizontal .
using theorem [ giroux : main ] we can isotop @xmath66 so that there is one closed curve of singularities in @xmath70 in each region of @xmath71 these are called and are parallel to the dividing curves .
we can further assume all the other leaves of @xmath70 form a 1-parameter family of parallel closed curves transverse to the legendrian divides .
these can have any slope not equal to the slope of the dividing curves .
a convex torus @xmath66 is said to be in _ standard form _
if @xmath70 has this nongeneric form , consisting of legendrian divides and legendrian ruling curves .
suppose @xmath72 and we find a bypass for one of the ruling curves for @xmath66 then we may isotop @xmath66 across the bypass as in theorem [ thm : bypass ] and reduce the number @xmath73 by one . if @xmath74 we have a configuration change as follows : let @xmath66 be a convex torus with @xmath75 and in some basis for @xmath66 the slope of the dividing curves is 0 .
if we find a bypass for a ruling curve of slope between @xmath76 and @xmath77 , @xmath78 then after pushing @xmath66 across the bypass the new torus has two dividing curves of slope @xmath79 let @xmath80 be a legendrian curve in @xmath81 we can always find a neighborhood @xmath62 of @xmath80 whose boundary is a convex torus with two dividing curves .
the linking of a dividing curve on @xmath82 with @xmath80 is the thurston - bennequin invariant of @xmath83 we call the slope of these dividing curves the _ boundary slope _ of @xmath84 if @xmath80 is not null - homologous , then the framing induced on @xmath80 relative to some pre - assigned trivialization of the normal bundle of @xmath80 will be called the of @xmath83 in contrast to transverse curves , neighborhoods of legendrian curves have a quantifiable thickness no matter how small a ( nice ) neighborhood of a legendrian curve one takes , the boundary slope of the neighborhood is fixed . if @xmath62 is the neighborhood of a legendrian curve with twisting number @xmath85 ( relative to a framing that is already fixed ) , then the slopes of the dividing curves on @xmath86 are @xmath87 . on the other hand , any tight contact structure on the solid torus @xmath88 with boundary slope
@xmath87 is contact isotopic to a neighborhood of a legendrian curve with twisting number @xmath89 see @xcite .
it is easy to decrease the twisting number inside @xmath62 by adding ` zigzags ' ( see below for an explanation of terminology ) .
hence , increasing the twisting number is one way of thickening @xmath62 .
we now show how to use bypasses to increase the twisting number of @xmath83 let @xmath80 be a legendrian curve in @xmath1 with a fixed framing .
let @xmath62 be a standard neighborhood of @xmath80 and @xmath73 the twisting number of @xmath83 if there exists a bypass attached to a legendrian ruling curve of @xmath82 of slope @xmath90 and @xmath91 then there exists a legendrian curve with larger twisting number isotopic to @xmath83 this lemma is a very useful formulation of the observation that if one has a bypass for a legendrian knot then one can increase its twisting number .
now that we see bypasses are useful let us consider how to find them .
let @xmath9 be a convex surface with legendrian boundary @xmath92 if the thurston - bennequin invariant of @xmath35 ( or let us say twisting number with respect to @xmath9 ) is negative then we can arrange that all the singularities of @xmath26 along @xmath35 are ( half)-elliptic . moreover we can assume the singular foliation @xmath26 is morse - smale . if @xmath73 is the twisting number of @xmath35 , then @xmath32 intersects @xmath35 at @xmath68 points . in this situation we can use the flow of @xmath16 ( the vector field directing @xmath26 in the definition of the dividing set ) to flow any dividing curve to a legendrian arc on @xmath9 , all of whose singularities are of the same sign .
now , starting with a dividing curve @xmath93 with endpoints along @xmath35 isotopic to the boundary of @xmath9 , we push @xmath93 ` back into @xmath9 along @xmath16 ' and eventually arrive at a legendrian arc which cuts a bypass for @xmath35 off of @xmath94 thus to find bypasses we need only look for these @xmath95-compressible dividing curves .
we now explore the neighborhood of a legendrian knot in more detail . to this end
let @xmath62 be a standard neighborhood of a legendrian knot @xmath80 with a fixed framing so that @xmath96 and @xmath80 has twisting number @xmath97 in this framing .
recall this means that @xmath82 is convex with two dividing curves ( and two legendrian divides ) of slope @xmath98 inside of @xmath62 we can isotop @xmath80 ( but not legendrian isotop ) to @xmath99 so as to decrease the twisting number by one .
there are actually two different ways to do this .
if @xmath80 were a knot in @xmath100 with the standard tight contact structure then this process corresponds to stabilizing the knot ( reducing the thurston - bennequin invariant ) by adding zigzags , and one can do this while increasing or decreasing the rotation number by one . to see how to detect this difference in @xmath62
we do the following : first , fix an orientation on @xmath62 and orient @xmath80 so that @xmath80 and @xmath101 intersect positively .
let @xmath102 be a standard neighborhood of @xmath99 .
then consider the layer @xmath103=n\backslash n'$ ] , where we set @xmath104 and @xmath105 we assume the legendrian ruling curves are vertical so that @xmath106 $ ] is an annulus with legendrian boundary .
note that @xmath107 intersects the dividing curves on @xmath108 @xmath109 times and @xmath110 intersects the dividing curves on @xmath111 @xmath68 times .
so there will be a bypass ( in fact just one bypass ) on @xmath112 for @xmath113 if @xmath112 is oriented so that the orientation on @xmath107 agrees with the one chosen above on @xmath80 , then the sign of the bypass is what distinguishes the two possible @xmath99 s .
if @xmath93 is a curve on @xmath107 whose slope is not @xmath114 or @xmath115 then we can make the legendrian ruling curves on @xmath116 parallel to @xmath117 later it will be useful to know what the dividing curves on @xmath118 $ ] will look like
. the relative euler class in the next paragraph can be useful for this purpose .
let @xmath5 be any tight structure on a toric annulus @xmath119 $ ] ( not necessarily the same one as in the above paragraphs ) .
assume the boundary is convex and in standard form .
let @xmath120 be a section of @xmath121 which is transverse to and twists ( with @xmath5 ) along the legendrian ruling curves .
we also take @xmath120 to be tangent to the legendrian divides .
we may now form the relative euler class @xmath122 in @xmath123 first note that @xmath122 is unchanged if we perform a @xmath124-small isotopy of @xmath116 so as to alter the slopes of the ruling curve .
now given an oriented curve @xmath93 on @xmath125 we can assume the annulus @xmath126 $ ] has legendrian boundary and is also convex .
we orient @xmath112 so that it induces the correct orientation on @xmath117 now we have @xmath127 where @xmath128 are the positive and negative regions into which the dividing curves cut @xmath129 this formula follows from proposition 6.6 in @xcite once one observes that @xmath120 may be homotoped in @xmath121 so as to be tangent to and define the correct orientation on @xmath130 we end this section by making precise the notion of _ twisting _ along @xmath131
let @xmath132 be the kernel of @xmath133 on @xmath134 with coordinates @xmath135 .
this contact structure is tight .
the characteristic foliation on @xmath136 is by lines of a fixed slope .
the slope `` decreases '' as @xmath137 moves around @xmath138 in the positive direction .
let @xmath139 $ ] be an interval in @xmath140 $ ] so that at one end @xmath141 has slope @xmath114 and at the other end @xmath142 has slope @xmath143 we can @xmath144-perturb @xmath145 and @xmath146 so that they are convex and each has two dividing curves of slope @xmath114 and @xmath147 respectively . let @xmath148 $ ] be the layer between @xmath102 and @xmath62 as in [ layering ] .
it is possible to show that @xmath149 $ ] is contactomorphic to @xmath150 with the contact structure constructed above ( see @xcite ) .
the slopes of the characteristic foliations on @xmath151 vary from @xmath114 to @xmath115 in @xmath149 $ ] so they do likewise in @xmath152 from this one can conclude that if @xmath62 is a standard neighborhood of a legendrian knot @xmath80 with twist number @xmath73 ( with respect to some fixed framing ) , then for any slope in @xmath153 one can find a torus around @xmath80 whose characteristic foliation has this slope ( note if @xmath154 then @xmath153 means @xmath155 $ ] where @xmath156 is identified with @xmath157 ) . more generally ,
if @xmath131 $ ] has boundary slopes @xmath158 for @xmath159 , then we may find convex tori parallel to @xmath159 with any slope @xmath160 in @xmath161 $ ] ( if @xmath162 then this means @xmath163\cup
[ -\infty , s_0]$ ] ) .
this follows from the classification of tight contact structures on @xmath164 in the proof we layer @xmath164 into ` thin ' toric annuli , each of which is isomorphic to @xmath149 $ ] above ( see @xcite ) .
consider the seifert fibered space @xmath1 with 3 singular fibers over @xmath165 .
@xmath1 is described by the seifert invariants @xmath166 .
let @xmath167 , @xmath168 , be the neighborhoods of the singular fibers @xmath169 , isomorphic to @xmath88 and identify @xmath170 with @xmath171 where @xmath172 is a sphere with three punctures .
then @xmath173 is given by @xmath174 .
we identify @xmath175 , by choosing @xmath176 as the meridional direction , and @xmath177 as the longitudinal direction with respect to the product structure on @xmath167 .
we identify @xmath178 by letting @xmath177 be the direction of an @xmath138-fiber , and @xmath176 be the direction given by @xmath179 .
let @xmath1 be the poincar homology sphere @xmath2 with reverse orientation .
it is a seifert fibered space over @xmath165 with seifert invariants @xmath180 . in the case
@xmath181 , we choose @xmath182 .
notice there we has some freedom to choose @xmath183 , @xmath184 , since we could have postmultiplied by @xmath185 if we changed our framing for @xmath186 .
similarly , let @xmath187 , and @xmath188 .
now let @xmath5 be a positive contact structure on @xmath1 .
assume @xmath5 is tight .
the goal of the paper is to obtain a contradiction by finding overtwisted disks inside @xmath1 . in the first stage of the proof
, we will try to thicken neighborhoods @xmath167 of the singular fibers @xmath169 ( we may assume the singular fibers are legendrian after isotopy )
. this is done by maximizing the twisting number @xmath189 among legendrian curves isotopic to @xmath169 , subject to the condition that all three legendrian curves be simultaneously isotopic to @xmath190 .
let @xmath167 be a standard tubular neighborhood of @xmath169 with minimal convex boundary and boundary slope @xmath191 ( note this is negative ) .
it is useful to note how the dividing curves map under @xmath192 .
@xmath193 , @xmath194 , and @xmath195 .
therefore , the boundary slopes are @xmath196 , @xmath197 and @xmath198 , when viewed on @xmath199
. .1 in * warning : * we will often call the same surface by different names , such as @xmath200 and @xmath199 . although the surfaces themselves are the same , their _ identifications _ with @xmath201 are not . therefore ,
when we refer to slopes on @xmath200 , we implicitly invoke the identification of @xmath200 with @xmath201 given in the first paragraph of this section .
[ increase ] we can increase @xmath189 so that @xmath202 , @xmath203 , and thicken @xmath167 to @xmath204 so that the slopes of @xmath205 are all infinite . we may modify the legendrian rulings on both @xmath206 and @xmath207 so that they are vertical . take a vertical annulus @xmath208 spanning from a vertical legendrian ruling curve on @xmath206 to a vertical legendrian ruling curve on @xmath207 . here ` vertical ' means ` in the direction of the @xmath138-fibers ' .
note @xmath208 intersects the dividing curves on @xmath209 and @xmath210 @xmath211 and @xmath212 times respectively .
assume @xmath213 .
if @xmath214 , then there exists a bypass ( due to the imbalance ) , attached along a vertical legendrian curve of @xmath206 or @xmath207 .
we transform @xmath177 via @xmath215 to use the twist number lemma .
@xmath216 , and @xmath217 .
the legendrian rulings will have slope @xmath218 and @xmath219 , and , therefore , we may increase the twisting number by 1 if the twisting number is @xmath220 . next ,
assume @xmath221 and there are no bypasses on the vertical annulus .
then we may take @xmath208 to be standard , with vertical rulings , and parallel legendrian divides spanning from @xmath222 to @xmath223 . cutting along @xmath208 and rounding the corners , we obtain the torus boundary of @xmath224 ; if we identify this torus with @xmath201 in the same way as @xmath225 , then the boundary slope is @xmath226 .
when @xmath227 , then the slope is @xmath228 , and any legendrian divide gives rise to an overtwisted disk this is because @xmath229 , which corresponds to a slope of @xmath228 on @xmath225 .
when @xmath230 , then the slope is @xmath231 , which implies that on the @xmath232 , the twisting in the radial direction is almost @xmath233 .
in particular , there exists a convex torus with slope @xmath157 inside @xmath88 , and hence a vertical legendrian curve with zero twisting , obtained as a legendrian divide of the convex torus . any vertical annulus taken from this vertical legendrian to @xmath234 ( or @xmath235 )
will give bypasses .
we can now assume @xmath203 .
the boundary slopes of @xmath236 and @xmath237 are @xmath228 and @xmath238 , respectively .
again look at the vertical annulus @xmath208 spanning from @xmath234 to @xmath235 , with legendrian boundary .
there are three possibilities :
( 1 ) there are no bypasses along @xmath107 ( on the @xmath234 side ) in this case , we can cut along the annulus as before , and get boundary component with boundary slope @xmath228 , contradicting tightness .
( 2 ) there is one bypass along @xmath107 cutting along the annulus again , we find that the boundary slope is @xmath239 .
this means that the twisting of @xmath240 is large , and that we have a convex torus in standard form with legendrian divides of infinite slope as before .
( 3 ) there are two bypasses . in any case , there exist vertical legendrian curves with twisting number @xmath241 with respect to the framing from the fibers . since @xmath242 , we have bypasses for @xmath186 as well , and we can increase to @xmath202 using the twist number lemma . next , taking a vertical annulus from @xmath234 to one such vertical legendrian curve , we obtain two bypasses for @xmath234 and , similarly , we obtain four bypasses for @xmath235 . by attaching these two bypasses
, we obtain a thickening @xmath204 of @xmath167 , @xmath168 , so that @xmath243 has two vertical dividing curves .
in the section we use the structure of @xmath244 as a punctured torus bundle over @xmath138 to complete the proof of our main theorem .
the strategy is as follows : first we use the bundle structure to increase @xmath245 to 0 by finding a bypass along the boundary of the punctured torus fiber .
we then show how to increase @xmath245 to 1 by making the boundary of the punctured torus fiber legendrian with twist number 0 .
when @xmath246 the corresponding neighborhood of @xmath247 almost contains an overtwisted disk . by increasing it a little further it does contain one .
this will complete the proof of our main theorem .
we now describe the fibration as a punctured torus bundle over @xmath138 . in the previous section ,
@xmath167 was the neighborhood of a legendrian curve @xmath169 with twisting number @xmath189 , and @xmath248 its thickening .
write @xmath249 , where @xmath9 is a 3-holed sphere .
let @xmath80 be an embedded arc on @xmath9 connecting @xmath250 to @xmath251 , @xmath252 an annulus connecting @xmath250 to @xmath251 , and @xmath253 a neighborhood of @xmath112 in @xmath254 .
define @xmath255 , which is @xmath256 with a @xmath164 layer removed from the boundary .
let @xmath257 and @xmath258 be meridional disks for @xmath251 and @xmath259 respectively .
the slope of @xmath257 on @xmath260 is @xmath228 and the slope of @xmath258 on @xmath261 is @xmath262 thus we can take two copies of @xmath258 and three copies of @xmath257 and glue them together with six copies of @xmath263 to obtain a punctured torus @xmath66 in @xmath264 with boundary on @xmath265 ( see figure [ goodfiber ] . ) parallel copies of @xmath66 will fiber @xmath264 ( and @xmath244 ) as a punctured torus bundle over @xmath266 the slope of @xmath267 on @xmath268 is @xmath269 thus the slope on @xmath270 is 1 , so @xmath66 provides a seifert surface for @xmath271 we can show ( for example by using kirby calculus ) that @xmath272 represented in figure [ fig : surgery](a ) , is diffeomorphic to figure [ fig : surgery](b ) , and that @xmath273 is diffeomorphic to the complement of the right - handed trefoil knot in @xmath274 thus we may identify the monodromy of the punctured torus bundle as @xmath275 we begin by showing that either we can find a bypass for @xmath267 or arrange that the dividing curves on @xmath66 consist of exactly two parallel arcs .
if there are no bypasses , then the dividing set must consist of two parallel arcs and an even number , say @xmath279 of closed parallel curves .
consider @xmath280 $ ] with @xmath66 identified with @xmath281 here we are viewing @xmath264 as @xmath273 .
let @xmath6 be a closed legendrian curve on @xmath66 parallel to the dividing curves .
then consider @xmath282 $ ] which we may assume is convex with legendrian boundary .
the dividing curves do not intersect @xmath283 , but they intersect @xmath284 at least @xmath285 times .
thus we may find a bypass @xmath51 for @xmath286 the inner boundary @xmath287 of a neighborhood of @xmath288 is a convex torus with either a bypass for @xmath289 or two fewer dividing curves than @xmath290 repeating this argument @xmath85 times will result in a bypass for @xmath267 or a convex torus whose dividing set consist of exactly two parallel arcs .
now suppose @xmath67 consists of two parallel arcs , since otherwise we are done .
then @xmath291 consists of two arcs and @xmath292 consists of the image of these two arcs under the monodromy map for the bundle .
let @xmath6 be a closed curve on @xmath293 parallel to the dividing curves , and @xmath294.$ ] we may arrange for @xmath295 to be legendrian and for @xmath296 to be convex . now the dividing curves on @xmath296 do not intersect @xmath283 but intersect @xmath284 at least two times .
thus we have a bypass @xmath51 for the dividing curves on @xmath286 assume first that the intersection number is exactly two
. then the annulus @xmath296 may be split into two annuli , one of which , call it @xmath297 intersects @xmath298 and contains @xmath51 , which is of degenerate type .
the boundary of a small neighborhood of @xmath299 in @xmath300 is an annulus @xmath301 and has dividing curves as shown in figure [ fig : annulus ] .
the boundary of @xmath301 sits on @xmath286 if we cut out the annulus in @xmath298 that cobounds a solid torus with @xmath302 glue in @xmath301 , and smooth corners , then the dividing curves on the new @xmath298 are shown in figure [ fig : annulus ] .
if the intersection number is greater than two , then simply attach the bypass @xmath51 onto @xmath298 either the new @xmath298 has a bypass , in which case we are done , or we can take a new @xmath296 which has fewer ( but @xmath304 intersections ) .
we continue until the intersection number becomes two .
[ lem : twist ] we may increase @xmath245 to @xmath241 while keeping @xmath202 and @xmath305 moreover , @xmath234 may be further thickened to @xmath306 so that @xmath307 has dividing curves of slope @xmath277 since we have a bypass on @xmath267 we may apply the twist number lemma to increase the twist number of @xmath247 to @xmath308 so we have two vertical dividing curves on @xmath270 and on @xmath309 they have slope 0 . now repeating the argument in lemma [ increase ]
we see that we can arrange @xmath202 and @xmath305 note the dividing curves on @xmath310 and @xmath311 have slope @xmath228 and 0 , respectively . thus taking a vertical annulus between @xmath312 and @xmath209
, we find a vertical bypass along @xmath313 with this bypass we can thicken @xmath234 to @xmath306 whose dividing curves have slope @xmath277 note if we had another bypass along @xmath267 we could increase @xmath245 to 1 and from here we could then find an overtwisted disk ( see below ) .
however , we know of no direct way to prove this bypass exists
. therefore we use the following strategy , which can be called as ` thinning before thickening ' .
notice we have thickened @xmath235 so that @xmath246 ; we will now backtrack by having @xmath235 relinquish some of its thickness ( we peel off a toric annulus from @xmath235 and reattach to @xmath186 and @xmath234 ) , and then thicken again to obtain a contradiction .
let @xmath317 be the meridional disk to @xmath167 for @xmath318 if we arrange that the legendrian rulings on @xmath200 are horizontal and the @xmath317 are convex , then @xmath319 and @xmath320 .
the dividing sets divide @xmath257 into two regions , one positive and one negative , and @xmath258 into three regions without loss of generality we can assume that , two regions are positive and one is negative .
note that the signs of these regions depend on an orientation on the disks .
pick an orientation for the fibers of the seifert fibration , and the disks will be oriented so that their intersection with a fiber is positive .
note we can write @xmath321 , where @xmath322 is a solid torus with convex boundary and dividing curves of slope @xmath239 ( on @xmath323 ) , @xmath103 $ ] , @xmath324 , and @xmath325 is convex with vertical dividing curves .
we do this by stabilizing ( reducing the twist number ) @xmath247 with @xmath315 in @xmath326 in @xmath150 we can find a @xmath327 , say @xmath328 with dividing curves of slope @xmath219 ( which correspond to vertical dividing curves from the point of view of @xmath329 ) . by performing the correct stabilization we can arrange that on a convex annulus in @xmath330 $ ] of slope @xmath331 with legendrian boundary on @xmath332 and @xmath333 there is a negative bypass along the boundary component on @xmath334 we can use this bypass to thicken @xmath167 , @xmath42 , to @xmath204 so that @xmath335 have vertical slopes . using the relative euler class
we can easily see that each meridional disk @xmath336 in @xmath248 has an extra positive region . specifically ,
if @xmath337 is the relative euler class on @xmath338 ( as defined in section [ section : convexity ] ) and @xmath339 and @xmath340 are the horizontal and vertical curves on @xmath341 then one may easily compute that @xmath342)= 1 $ ] and @xmath343)=0.$ ] thus @xmath344)=1 $ ] where @xmath93 is the boundary of the meridional disk in @xmath345 from this and equation [ eqn : formula ] , we can conclude that the dividing curves are as in figure [ anndivides ] ( a ) , and hence a convex meridional disk in @xmath251 will have one negative and two positive regions in the complement of the dividing curves .
similarly , a convex meridional disk in @xmath250 will have one negative and three positive regions since a relative euler class argument will show its intersection with @xmath346 is shown in figure [ anndivides ] ( b ) we would now like to piece these meridional disks together to form a punctured torus fiber for @xmath348 , but to do this we first need to understand the complement of the singular fibers . recall all the boundary slopes of @xmath349 are infinite .
take @xmath350 , which we assume is convex with legendrian boundary .
all the boundary components of @xmath9 have exactly two half - elliptic points .
we argue that if there is a @xmath95-compressible dividing arc ( as in ( b ) , ( c ) , ( d ) ) on @xmath9 , then @xmath1 is overtwisted .
the @xmath95-compressible dividing arc implies the existence of a bypass ( of degenerate type ) along some @xmath351 .
hence , there exists a layer @xmath352 with one boundary component @xmath351 and boundary slopes @xmath241 and @xmath157 .
however , since there are vertical legendrian curves with twisting number 0 outside of @xmath353 , we obtain another layer @xmath354 , this time with boundary slopes @xmath157 and @xmath241 .
therefore , @xmath355 has too much radial twisting , and is overtwisted .
the only possible configuration without a @xmath95-compressible dividing arc is ( a ) .
we cut @xmath347 along @xmath9 , round the edges , and examine the dividing curves on the boundary of the resulting genus 2 handlebody .
we then cut along the meridional disks of the handlebody , which we may assume have legendrian boundary , and eventually obtain a 3-ball .
since each meridional disk of the handlebody intersects the dividing set only twice , the configuration of dividing curves on the meridional disks is unique .
therefore the initial configuration of dividing curves on @xmath9 uniquely determines the tight structure on @xmath347 .
the lemma implies that the tight structure on @xmath347 is a translation - invariant tight structure on @xmath164 with infinite boundary slopes , with the standard neighborhood of a vertical legendrian curve with zero twisting removed .
we view this @xmath164 ( minus @xmath88 ) as the region between @xmath359 and @xmath360 ( minus @xmath358 ) .
we may think of the @xmath361-factor as being quite small , and then isotop one of the legendrian divides on @xmath359 to one of the legendrian divides on @xmath360 and finally identify small neighborhoods ( in the tori ) of these divides .
we may then isotop @xmath362 , @xmath42 , away from these neighborhoods so that the meridional disk @xmath363 in @xmath204 has legendrian boundary .
forming the fiber @xmath66 in the fibration of @xmath364 from three copies of @xmath257 and two copies of @xmath258 , we can arrange that @xmath267 is legendrian and the dividing curves on @xmath66 are as in figure [ goodfiber ] .
thus there are six bypasses for @xmath303 we can use these to increase the twisting number of @xmath267 to 0 ( _ i.e. _ , @xmath267 will be a legendrian divide on some thickening of @xmath270 ) which corresponds to an increase of @xmath245 to 1 .
thus the slope of the dividing curves on @xmath365 is @xmath366 .
now repeat the argument in lemma [ increase ] take a vertical annulus from @xmath367 to @xmath368 , and start with @xmath369 small . since the denominator of @xmath366 is never equal to @xmath211
, we eventually arrive at @xmath305 this implies the existence of a vertical bypass for @xmath368 , with which we can increase the slope of the dividing curves on @xmath365 to @xmath370 .
thus the legendrian divide on our thickened solid torus bounds a meridional disk in @xmath235 , so we have found an overtwisted disk .
this completes the proof of the main theorem . | we exhibit a 3-manifold which admits no tight contact structure . |
let @xmath0 and @xmath9 be two graphs of order @xmath83 and @xmath84 respectively .
the corona product of @xmath0 and @xmath9 is defined as the graph obtained from @xmath0 and @xmath9 by taking one copy of @xmath0 and @xmath83 copies of @xmath9 and joining by an edge each vertex from the @xmath85-copy of @xmath9 with the @xmath85-vertex of @xmath0 .
we will denote by @xmath86 , the set of vertices of @xmath0 and by @xmath87 , the @xmath73-th copy of @xmath9 , where @xmath88 , such that @xmath89 for every @xmath90 .
note that the subgraph of @xmath91 induced by @xmath92 is @xmath93 and the corona graph @xmath94 is isomorphic to the join graph @xmath95 . for any integer @xmath96 , we define the graph @xmath97 recursively from @xmath7 as @xmath98 . it is also noted that @xmath99 and @xmath100 we call the copies of @xmath9 in @xmath7 as the copies of @xmath9 in @xmath101-corona , the newly added copies of @xmath9 in @xmath7 to obtain @xmath102 as the copies of @xmath9 in @xmath103-corona and generally the newly added copies of @xmath9 in @xmath104 to obtain @xmath97 as the copies of @xmath9 in @xmath105-corona . in @xmath97 for any positive integer @xmath106 , we name the vertices in @xmath104 as the root vertices of the copies of @xmath9 in @xmath105-corona , that are joined to these vertices in @xmath107 .
as one can color the vertices of @xmath97 in more than one ways , but in this paper for a disconnected graph @xmath9(containing isolated vertices ) of order at least two , we will consider the zero forcing set of @xmath97 , that contains only the vertices of @xmath9 but not the vertices of @xmath0 . in figure [ new123 ] ,
the graph with grey vertices is @xmath108 , the copies of @xmath9 with black vertices are the copies of @xmath9 in first corona and with white vertices are the copies of @xmath9 in 2nd corona .
the black and grey vertices are the root vertices of the corresponding copies of @xmath9 with white vertices . ,
width=302 ] we first recall the useful result obtained in @xcite .
[ manha ] @xcite let @xmath0 be a connected graph of order @xmath109 .
then + @xmath110 @xmath111 if and only if @xmath112 + @xmath113 @xmath114 if and only if @xmath115 [ u ] note that for a connected graph @xmath0 of order @xmath35 , we have @xmath116 let @xmath0 be a connected graph of order @xmath117 and let @xmath9 be a graph of order at least two .
let @xmath118 be the subgraph of @xmath119 corresponding to the @xmath85-copy of @xmath9 .
+ @xmath120 if @xmath2 is a zero forcing set of @xmath91 , then @xmath121 for every @xmath122 + @xmath123 if @xmath9 is a connected graph and @xmath2 is a zero forcing set of @xmath91 , then for every @xmath124 , @xmath125 is a zero forcing set of @xmath118 . @xmath120
suppose @xmath126 , for some @xmath73 .
then the vertex @xmath127 , initially black or forced to black in the zero forcing process , will not force any vertex in @xmath128 to turn black because it has more than one white neighbors , a contradiction .
+ @xmath123 suppose contrary that @xmath129 is not a zero forcing set for @xmath118
. then there exists a black vertex say @xmath130 , that has more than one white neighbors in graph @xmath118 , so no forcing situation can occur . note that the vertex @xmath130 also has more than one white neighbors in @xmath91 , a contradiction .
, width=302 ] here we introduce some terminology related to the following theorem .
we assume that @xmath131 is a forcing basis for @xmath0 and by using @xmath132 one can color all the other vertices of @xmath0 by a sequence of forces in the following order : @xmath133 , with appropriate indexing of vertices .
we denote the vertices of @xmath73-th copies of @xmath9 in @xmath134-th corona ( @xmath135 ) by @xmath136 , where @xmath137 and @xmath138 for each @xmath139 .
assume that @xmath140 and a forcing basis for @xmath85-copy @xmath118 of @xmath9 in @xmath101-corona is denoted by @xmath141 and forcing basis for @xmath73-th copies of @xmath9 in @xmath142-corona(@xmath143 ) by @xmath144 , where @xmath145 .
we denote the collection of forcing basis of all copies of @xmath9 in first corona by @xmath146 i.e @xmath147 , similarly the collection of forcing basis of all copies of @xmath9 in @xmath134-th corona by @xmath148 i.e @xmath149 and @xmath150 .
figure [ must ] helps in understanding the indices as mentioned .
[ api ] let @xmath0 and @xmath9 be connected graphs of order @xmath83 and @xmath84 respectively , then @xmath151 we prove the result by mathematical induction .
[ me ] for @xmath152 we have to show that@xmath153 + first , we show that@xmath154we define @xmath155 and @xmath156 .
we claim that @xmath157 is a zero forcing set of @xmath91 .
to prove the claim , first assume that @xmath157 is initially colored black and we color all the vertices of @xmath118 with @xmath158 which are associated with the vertices of @xmath132 using the corresponding sets @xmath159 .
now all the vertices in @xmath118 associated with @xmath63 , @xmath158 are colored black .
note that there is a vertex @xmath63 belonging to @xmath132 that has only one white neighbor @xmath160 .
. then we color all the vertices in @xmath162 by using the black vertices in @xmath163 .
continuing this process , we can color all the vertices of @xmath164 .
note that the degree of each vertex of @xmath0 is increased by @xmath84 and the degree of each vertex of @xmath9 is increased by @xmath62 in @xmath7 .
let @xmath165 and consider the corresponding copy @xmath118 of @xmath9 .
note that at least @xmath166 vertices are required as initially colored black to start the zero forcing process in each of these @xmath118 s , @xmath167 .
then @xmath168 and to continue the process at least @xmath169 more vertices are required in @xmath162 as initially colored black .
continue the process until all the vertices are turned black .
hence , @xmath170by ( [ 1 ] ) and ( [ 2 ] ) , ( [ k=1 ] ) holds .
+ suppose that the result is true for @xmath171 , i.e @xmath172now we have to show that the result is true for @xmath106 , i.e @xmath173we define @xmath174 and @xmath175 then @xmath176 is a zero forcing set of @xmath104 by ( [ if ] ) .
therefore , we color all the vertices of @xmath104 using @xmath176 .
now all the vertices from @xmath62st - corona to @xmath177-corona in @xmath178 are colored black .
it suffices to show that @xmath179 vertices in the copies of @xmath9 in @xmath105-corona in @xmath178 will be colored black by taking @xmath180 more vertices as initially colored black .
+ first , we show that @xmath181we define @xmath182 where @xmath183 .
we claim that @xmath157 is a zero forcing set of @xmath97 .
let @xmath157 is initially colored black .
note that the degree of each vertex in @xmath118 in @xmath105-corona in @xmath178 is increased by one .
we color all the vertices of copies of @xmath9 in @xmath105-corona by using @xmath184 and the corresponding root vertex in @xmath177-corona .
we obtain the derived set of all black vertices in @xmath178 resulting from repeatedly applying the color - change rule . hence , ( [ m ] ) holds .
note that in each copy of @xmath9 in @xmath105-corona at least @xmath169 vertices are required as initially colored black to continue the zero forcing process .
hence , @xmath185by ( [ m ] ) and ( [ e ] ) , ( [ v ] ) holds .
hence , the result is true for any positive integer @xmath106 . by using theorem [ api ] and proposition [ manha ]
, we have the following immediate corollaries : let @xmath0 and @xmath9 be connected graphs of order @xmath83 , @xmath186 respectively .
then @xmath187 if and only if @xmath188 .
let @xmath0 and @xmath9 be connected graphs of order @xmath83 , @xmath186 respectively . then @xmath189 if and only if @xmath190 .
the wheel graph of order @xmath191 is defined as @xmath192 , where @xmath193 is the singleton graph .
any three pairwise adjacent vertices of a wheel form a zero forcing set of @xmath194 .
let @xmath194 , @xmath195 , be a wheel graph .
then @xmath196 the fan graph @xmath197 is defined as the join graph @xmath198 .
the case @xmath199 corresponds to the usual fan graph @xmath200 . note that @xmath201 , where @xmath193 is the singleton graph .
two adjacent vertices of @xmath202 where one must be an end vertex form a zero forcing set of @xmath200 .
let @xmath203 , @xmath204 , be a fan graph .
then @xmath205 let @xmath0 be a connected graph of order @xmath206 and let @xmath9 be a graph of order @xmath207 .
then @xmath208 we denote by @xmath209 the subgraph of @xmath7 , obtained by joining the vertex @xmath127 with all the vertices of @xmath118 .
let @xmath210 be a forcing basis for @xmath209 and @xmath211 with @xmath212 .
we show that @xmath51 is a zero forcing set of @xmath7 .
now there are two cases : + case @xmath62 : @xmath9 contains no isolated vertex .
we have that @xmath213 .
so @xmath51 contains all the vertices of @xmath0 .
note that any black vertex in @xmath118 has only one white neighbor , so after finite many applications of the color - change rule all the vertices in @xmath93 for each @xmath214 are turned black .
now all the vertices in @xmath7 are colored black . + case
@xmath81 : @xmath9 contains isolated vertices , so there exists at least one vertex @xmath215 such that @xmath216 , for all @xmath217 .
we have that @xmath63 does not belong to the zero forcing set of minimum cardinality of @xmath209 .
so @xmath51 does not contain any vertex from @xmath0 .
+ subcase @xmath218 : @xmath9 has only one isolated vertex .
let @xmath219 is the isolated vertex of @xmath118 also @xmath220 and @xmath221 for any @xmath222 .
thus @xmath223 , @xmath224 .
now all the vertices of @xmath0 are forced to black .
note that any black vertex in @xmath118 has only one white neighbor , so after finite many applications of the color - change rule all the vertices in @xmath93 for each @xmath214 are turned black .
now all the vertices in @xmath7 are colored black .
+ subcase @xmath225 : @xmath9 has more than one isolated vertices , then all isolated vertices in @xmath118 belong to @xmath210 except one , say @xmath226 does not belong to @xmath210 , then @xmath63 will be forced by any isolated black vertex in @xmath118 , @xmath227 .
now all the vertices of @xmath0 are forced to black and after finite iterative applications of the color - change rule for connected subgraph of @xmath118 all the vertices in these graphs are turned black for each @xmath214 , and then @xmath228 hence @xmath229 therefore , the result follows .
let @xmath0 be a connected graph of order @xmath230 and @xmath9 be a disconnected graph of order @xmath207
. then @xmath231 if and only if @xmath232 .
[ q ] suppose @xmath233 . for @xmath234 , we have to show that @xmath235 .
we define @xmath236 , for any @xmath237 , and for each @xmath214 and @xmath238 .
we claim that @xmath157 is a zero forcing set of @xmath7 with @xmath239 .
to prove the claim , we first assume that @xmath157 is initially colored black .
note that every initial black vertex of @xmath118 has single white neighbor @xmath63 , so @xmath240 .
now there is only one white vertex @xmath241 in each @xmath128 and this vertex is the single white neighbor of the corresponding vertex @xmath63 , so @xmath242 for each @xmath243 .
consider @xmath127 and the corresponding copy @xmath118 of @xmath9 for any @xmath73 , @xmath227 .
note that at least @xmath244 vertices of @xmath118 are required to start the zero forcing process .
hence , @xmath245 .
therefore , the result follows . on the other hand
, @xmath246 implies @xmath233 .
suppose @xmath247 and let @xmath248 , for @xmath227 .
we define @xmath249 , @xmath250 and @xmath251 .
let @xmath252 with @xmath253 .
we show that @xmath157 is a zero forcing set of @xmath7 .
assume that @xmath157 is initially colored black .
note that each isolated vertex in each @xmath118 , @xmath227 , has only one white neighbor @xmath127 , so @xmath63 will be forced to black for each @xmath243 .
now all the vertices of @xmath0 are colored black . since @xmath254 has only one white neighbor @xmath255 for each @xmath243 , so @xmath256 , @xmath227 .
note that @xmath257 has only one white neighbor @xmath258 , so @xmath259 .
now all the vertices are colored black .
so @xmath157 is a zero forcing set of cardinality @xmath260 , a contradiction .
therefore , the result follows .
the following definitions are introduced in @xcite .
fix a graph @xmath261 .
a vertex of degree at least three is called a major vertex .
an end vertex @xmath28 is called a terminal vertex of a major vertex @xmath24 if @xmath262 for every other major vertex @xmath263 .
the terminal degree of a major vertex @xmath24 in @xmath261 , denoted by @xmath264 , is the number of terminal vertices of @xmath24 .
a major vertex @xmath24 is an exterior major vertex ( emv ) if it has positive terminal degree .
let @xmath265 denote the sum of terminal degrees of all major vertices of @xmath0 and let @xmath266 denote the number of emvs of @xmath0 .
we further define an exterior degree two vertex to be a vertex of degree two that lies on a path from a terminal vertex to its major vertex and an interior degree two vertex to be a vertex of degree two such that the shortest path to any terminal vertex includes a major vertex .
@xcite if @xmath261 is a tree that is not a path , then @xmath267 .
[ n ] @xcite for any tree @xmath261 , we have @xmath268 iff @xmath261 has no interior degree two vertices and each major vertex @xmath24 of @xmath261 satisfies @xmath269 .
[ r ] let @xmath261 be a tree of order @xmath195 , that has no interior degree two vertices and each major vertex @xmath24 of @xmath261 satisfies @xmath269 , then + @xmath270 + since @xmath271 is a tree , with no interior degree two vertex and each major vertex @xmath24 satisfies @xmath269 .
now for @xmath234 , @xmath272 and @xmath273 , @xmath274 by theorem [ n ] and theorem [ r ] we obtain the result . since we have @xmath275 , @xmath276 , so we obtain the result for @xmath277 by theorem [ n ] and theorem [ r ] , @xmath278 .
let @xmath279 be the number of connected components of a graph @xmath9 .
let us denote the connected components of @xmath9 by @xmath280 , where @xmath281 .
let @xmath0 be a connected graph of order @xmath83 and @xmath9 be a graph of order @xmath84 .
let @xmath279 be the number of connected components of @xmath9 of order greater than one and let @xmath282 be the number of isolated vertices of @xmath9 . then + @xmath283 + we define @xmath284 , @xmath281 , be a forcing basis for connected component @xmath285 of @xmath118 , @xmath227 .
we suppose @xmath286 .
we define @xmath287 to be the set of vertices of @xmath7 formed by all but one of the isolated vertices of @xmath118 , @xmath227 .
let us show that @xmath288 is a zero forcing set of @xmath7 with @xmath289 .
let @xmath157 is initially colored black .
note that @xmath290 has only one white neighbor @xmath127 , so @xmath291 for each @xmath73 , @xmath227 .
now all the vertices of @xmath0 are colored black .
we color all the vertices of connected components @xmath285 , @xmath281 , of @xmath118 using @xmath284 and the corresponding vertex @xmath127 , @xmath227 .
note that the vertex @xmath63 has only one white neighbor @xmath292 , the isolated vertex of @xmath118 not belonging to @xmath287 , so @xmath293 , @xmath227 , and we have the derived set of all black vertices in @xmath7 . as a consequence , @xmath294 .
therefore , the result follows .
now suppose @xmath295 , and @xmath296 .
let @xmath297 , where @xmath298 is a forcing basis for @xmath0 and by using @xmath132 one can color all the other vertices of @xmath0 by a sequence of forces in the following order : @xmath299 , with appropriate indexing of vertices .
we show that @xmath157 is a zero forcing set of @xmath7 .
consider iterative applications of the color - change rule with initial black set @xmath157 .
we color all the vertices of @xmath118 with @xmath300 which are associated with the vertices of @xmath132 using the corresponding sets @xmath284 , @xmath301 .
now all the vertices in @xmath118 associated with @xmath63 , @xmath158 are colored black .
note that there is a vertex @xmath63 belonging to @xmath132 that has only one white neighbor @xmath302 .
. then we color all the vertices in @xmath162 using the black vertices in @xmath303 , @xmath281 . continuing this process , we can color all the vertices of @xmath164 .
so @xmath304 .
therefore , the result follows .
now suppose @xmath295 , @xmath305 .
let @xmath306 be the set of all isolated vertices in each @xmath118 , @xmath227 , except one say @xmath307 .
we define @xmath308 .
we show that @xmath157 is a zero forcing set of @xmath7 .
let @xmath157 is initially colored black .
note that the vertex @xmath309 , @xmath250 , has only one white neighbor @xmath127 , so @xmath310 , @xmath311 .
we color all the vertices of @xmath118 , @xmath312 , which are associated with @xmath63 , @xmath312 using the corresponding sets @xmath284 , @xmath281 .
now all the vertices in @xmath118 associated with @xmath63 , @xmath312 are colored black .
now @xmath313 has only one white neighbor @xmath257 , so @xmath314 .
we color @xmath315 in @xmath316 using @xmath317 , @xmath318 .
now @xmath257 has only one white neighbor @xmath319 , so @xmath320 and we have the derived set of all black vertices .
so @xmath321 .
therefore , the result follows .
now suppose @xmath322 , @xmath323 . here
@xmath9 is an empty graph so the result followed by corollary [ q ] .
let @xmath0 and @xmath9 be two graphs .
the lexicographic product of @xmath0 and @xmath9 , denoted by @xmath8 , is the graph with vertex set @xmath324 = @xmath325 and @xmath326 , where @xmath327 is adjacent to @xmath328 whenever @xmath329 or @xmath330 and @xmath331 . for any vertex @xmath332 and @xmath333
, we define the vertex set @xmath334 = @xmath335 and @xmath336 = @xmath337 .
it is clear that the graph induced by @xmath334 , called a layer @xmath334 , is isomorphic to @xmath9 and the graph induced by @xmath336 , called a layer @xmath336 , is isomorphic to @xmath0 , denoted by @xmath338 and @xmath339 respectively .
we write @xmath340 when each vertex of @xmath334 is adjacent to all vertices of @xmath341 and vice versa , and @xmath342 means that no vertex of @xmath334 is adjacent to any vertex of @xmath341 and vice versa .
let @xmath0 be a connected graph and @xmath9 be a non - trivial graph containing @xmath343 components @xmath344 with @xmath345 for each @xmath346 .
for any vertex @xmath332 and @xmath347 , we define the vertex set @xmath348 = @xmath349 . let @xmath350 , @xmath347 . from the definition of @xmath351 ,
it is clear that for every @xmath352 , @xmath353 . if @xmath0 is a disconnected graph having @xmath354 components @xmath355 , @xmath356 , , @xmath357 ,
then @xmath8 is also a disconnected graph having @xmath106 components such that @xmath358 and each component @xmath359 is the lexicographic product of connected component @xmath360 of @xmath0 with @xmath9 , therefore throughout this section , we will assume @xmath0 to be connected .
first , we give a general lower bound on the zero forcing number of lexicographic product of graphs .
note that given any connected graph @xmath0 , then @xmath111 if and only if @xmath362 , @xmath14 .
so , if @xmath363 for some graph @xmath9 , then clearly @xmath364 is a path graph , i.e @xmath0 is the trivial graph @xmath193 and @xmath9 is a path or viceversa .
so , we have the following result : let @xmath368 . if @xmath369 we have nothing to prove .
suppose that @xmath370 . since @xmath0 is connected and @xmath371 and also @xmath24 is initially colored black so by equation ( [ eqz ] ) , there exists at least one white vertex @xmath372 such that @xmath373 is a forcing basis for @xmath366 .
[ thm1 ] let @xmath0 be a connected graph and @xmath9 be an arbitrary graph containing @xmath343 components @xmath374 and @xmath375 .
let @xmath75 be a zero forcing set of @xmath8 .
for any vertex @xmath332 , if @xmath376 for every @xmath377 , then @xmath378 . moreover , if @xmath379 is a forcing basis for @xmath118 , then @xmath380
. suppose that for some @xmath381 there exists a vertex @xmath332 such that @xmath382 .
then , by observation [ obs1 ] any vertex in @xmath348 can not be forced by any vertex in @xmath383 , @xmath384 for any @xmath385 , a contradiction .
now suppose that @xmath386 and @xmath387 for some forcing basis @xmath379 of @xmath118 , where @xmath388 .
then , each black vertex in @xmath348 has more than one white neighbors and no vertex of @xmath348 can be forced by any vertex in @xmath389 for any @xmath390 , @xmath384 .
hence , @xmath380 .
let @xmath0 be a connected graph and @xmath9 be an arbitrary graph containing @xmath343 components @xmath374 and @xmath375 .
let @xmath391 for @xmath392 .
then @xmath393 is a zero forcing set of @xmath334 .
let @xmath0 be a connected graph and @xmath9 be an arbitrary graph containing @xmath394 components @xmath344 and @xmath395 .
let @xmath332 and @xmath75 be a forcing basis for @xmath8 . if @xmath396 and @xmath397 . then @xmath398 for any @xmath399 , @xmath400 .
since @xmath0 is connected so for any @xmath401 , there exist at least one vertex @xmath390 such that @xmath402 and @xmath403 .
to start the zero forcing process at least all the vertices of @xmath404 along with @xmath405 vertices are initially colored black . since @xmath75 is a forcing basis for @xmath8 .
hence , @xmath406 and @xmath407 for any @xmath332 .
the projection of @xmath410 onto @xmath0 , denoted by @xmath411 , is the set of vertices @xmath332 for which there exists a vertex @xmath412 .
similarly , the projection of @xmath410 onto @xmath9 , @xmath413 , is the set of vertices @xmath414 for which there exists a vertex @xmath412 .
[ lem 3 ] let @xmath0 be a connected graph of order @xmath35 and @xmath9 be an arbitrary graph containing @xmath394 components @xmath344 and @xmath395 .
let @xmath75 be a forcing basis for @xmath364 and @xmath415 , where @xmath416 is the induced subgraph of @xmath364
. then @xmath417 .
let @xmath418 and @xmath419 for @xmath420 .
suppose @xmath421 , i.e there exists a vertex @xmath422 such that @xmath423 .
this implies @xmath424 for any @xmath425 for @xmath426 .
hence , @xmath427 , a contradiction by theorem [ thm1 ] .
let @xmath418 and @xmath419 for @xmath420 .
we define @xmath429 and @xmath430 .
we claim that @xmath75 is a zero forcing set of @xmath8 .
to prove the claim , assume that @xmath75 is initially colored black . since for any @xmath384 , @xmath431 in @xmath9 hence @xmath432 , @xmath433 , in @xmath8 .
since for any @xmath73 , @xmath118 is connected , so there exists at least one vertex @xmath434 such that @xmath435 in @xmath93 and hence @xmath436 , @xmath433 , in @xmath8 .
therefore , @xmath437 for @xmath433 .
hence , @xmath438 let @xmath418 and @xmath446 for @xmath347 .
we define , for @xmath447 , @xmath448 with @xmath449 . since @xmath450 so @xmath451 .
similarly , @xmath452 .
now all the vertices in @xmath8 are colored black .
therefore , the result follows .
now we study the zero forcing number of lexicographic product of graphs for some specific families of graphs and @xmath9 contains one component only .
note that @xmath453 , so @xmath454 . therefore , from now on we consider the graphs when at most one of the factors of the product is a complete graph .
note that for any @xmath457 , @xmath340 for all @xmath458 .
therefore all the vertices in @xmath341 for all @xmath459 are initially colored black . now
@xmath338 so @xmath169 vertices are required as initially colored black to complete the zero forcing process . by corollary [ cor1 ] ,
there exists at least one vertex @xmath332 such that @xmath467 . since @xmath468 ,
therefore for @xmath469 layers @xmath404 at least @xmath470 vertices from each layer are required as initially colored black to color all the vertices of @xmath465 .
hence @xmath471 .
now we study the zero forcing number of @xmath472 , for @xmath195 and a connected graph @xmath9 .
suppose @xmath473 . since @xmath474 , @xmath475 .
note that to color the vertices of @xmath476 , @xmath169 vertices in @xmath476 and all the vertices in @xmath477 are required as initially colored black and to color the vertices of @xmath478 , @xmath169 vertices in @xmath478 and all the vertices in @xmath479 are required as initially colored black .
continue the process untill all the vertices in @xmath472 are turned black .
now we study the zero forcing number of @xmath483 , for @xmath484 and a connected graph @xmath9 .
suppose @xmath485 . since @xmath474 , @xmath486 and @xmath487 .
note that to color the vertices of @xmath476 , @xmath169 vertices in @xmath476 and all the vertices in @xmath477 and @xmath488 are required as initially colored black and to color the vertices of @xmath478 , @xmath169 vertices in @xmath478 and all the vertices in @xmath479 are required as initially colored black .
continue the process untill all the vertices in @xmath483 are turned black .
99 aim minimum rank - special graphs work group ( f. barioli , w. barrett , s. butler , s. m. cioab , d. cvetkovi , s. m. fallat , c. godsil , w. haemers , l. hogben , r. mikkelson , s. narayan , o. pryporova , i. sciriha , w. so , d. stevanovi , h. van der holst , k. vander meulen and a. w. wehe ) .
zero forcing sets and the minimum rank of graphs , linear algebra appl .
, @xmath492 @xmath493 @xmath494 .
[ 3 ] f. barioli , w. barrett , s. m. fallat , h. t. hall , l. hogben , b. shader , p. van den driessche and h. van der holst , zero forcing parameters and minimum rank problems , linear algebra appl . , @xmath495 @xmath496 @xmath497 .
[ iqra ] a. berman , s. friedland , l. hogben , u. g. rothblum and b. shader , an upper bound for the minimum rank of a graph , linear algebra appl . , @xmath498 @xmath493 @xmath499 .
[ 4 ] g. chartrand , l. eroh , m. a. johnson and o. r. oellermann , resolvability in graphs and the metric dimension of a graph , discrete appl .
, @xmath500 @xmath501 @xmath502.[5 ] k. chilakamarri , n. dean , c. x. kang and e. yi , iteration index of a zero forcing set in a graph , bull .
appl . , @xmath503 @xmath504 @xmath505 .
[ 6 ] c. j. edholm , l. hogben , m. hyunh , j. lagrange and d. d. row , vertex and edge spread of zero forcing number , maximum nullity , and minimum rank of a graph , linear algebra appl .
, @xmath506 @xmath504 @xmath507 .
[ 7 ] l. eroh , c. x. kang and e. yi , on zero forcing number of graphs and their complements , discrete math .
appl . , @xmath508 @xmath509 doi : @xmath510 .
l. eroh , c. x. kang and e. yi , a comparison between the metric dimension and zero forcing number of trees and unicylic graphs , arxiv:@xmath511 [ math.co ] .
[ 13 ] s. m. fallat and l. hogben , the minimum rank of symmetric matrices described by a graph : a survey , linear algebra appl .
, @xmath512 @xmath513 @xmath514 .
[ 9 ] s. m. fallet and l. hogben , variants on the minimum rank problem : a survey ii , arxiv:@xmath515 [ math.co ] . [ 10 ] f. harary and r. a. melter , on the metric dimension of a graph , ars combinatoria . , @xmath516 . [ 17 ] l. hogben , m. huynh , n. kingsley , s. meyer , s. walker and m. young , propagation time for zero forcing on a graph , discrete appl .
, @xmath517 @xmath504 @xmath518 .
[ 11 ] p. j. slater , leaves of trees , congr .
, @xmath519 @xmath520 @xmath521 .
f. a. taklimi , zero forcing sets for graphs , ph.d thesis , university of regina , august @xmath522 . | the zero forcing number of a graph @xmath0 , denoted by @xmath1 , is the minimum cardinality of a set @xmath2 of black vertices ( where vertices in @xmath3 are colored white ) such that @xmath4 is turned black after finitely many applications of @xmath5the color change rule@xmath6 : a white vertex is turned black if it is the only white neighbor of a black vertex . in this paper , we study the zero forcing number of corona product , @xmath7 and lexicographic product , @xmath8 of two graphs @xmath0 and @xmath9 . it is shown that if @xmath0 and @xmath9 are connected graphs of order @xmath10 and @xmath11 respectively , then @xmath12 , where @xmath13 .
also , it is shown that for a connected graph @xmath0 of order @xmath14 and an arbitrary graph @xmath9 containing @xmath15 components @xmath16 with @xmath17 , @xmath18 , @xmath19 .
let @xmath20 be a simple , undirected , connected graph with @xmath21 .
the number of vertices and edges of @xmath0 are called the order and the size of @xmath0 respectively .
the degree of a vertex @xmath22 , denoted by @xmath23 , is the number of edges incident to the vertex @xmath24 in @xmath0 .
if there is no ambiguity , we will use the notation @xmath25 instead of @xmath23 .
an end vertex is a vertex of degree one .
given @xmath26 , @xmath27 means that @xmath28 and @xmath24 are adjacent vertices and @xmath29 means that @xmath28 and @xmath24 are not adjacent .
we define the open neighborhood of a vertex @xmath24 in @xmath0 , @xmath30 and the closed neighborhood of @xmath24 , @xmath31=n_{g}(v)\cup\{v\}$ ] . if there is no ambiguity , we will simply write @xmath32 or @xmath33 $ ] . if @xmath34 then @xmath28 is said to be a neighbor of @xmath24 .
we denote a path , cycle , complete graph and empty graph on @xmath35 vertices by @xmath36 , @xmath37 , @xmath38 and @xmath39 respectively .
all graphs considered in this paper are non trivial unless otherwise stated .
the notion of a zero forcing set , as well as the associated zero forcing number , of a simple graph was introduced in @xcite to bound the minimum rank of associated matrices for numerous families of graphs .
let each vertex of a graph @xmath0 be given one of two colors , @xmath40 and @xmath41 by convention .
let @xmath2 denote the initial set of black vertices of @xmath0 .
the color - change rule converts the color of a vertex @xmath42 from white to black if the white vertex @xmath42 is the only white neighbor of a black vertex @xmath43 ; we say that @xmath43 forces @xmath42 , which we denote by @xmath44 . and
a sequence , @xmath45 obtained through iterative applications of the color - change rule is called a forcing chain .
the set @xmath2 is said to be a zero forcing set of @xmath0 if all the vertices of @xmath0 will be turned black after finitely many applications of the color - change rule .
the zero forcing number of @xmath0 , denoted by @xmath1 , is the minimum of @xmath46 over all zero forcing sets @xmath47 .
a zero forcing set of cardinality @xmath1 is called a forcing basis for @xmath0 . for surveys on the zero forcing parameter ,
see @xcite . for more on the zero forcing parameter in graphs ,
see @xcite .
if @xmath48 is a field , @xmath49 denotes the set of all @xmath50 matrices over f. an @xmath35-square matrix @xmath51 is said to be a symmetric matrix if @xmath52 . the set of all real symmetric @xmath35-square matrices
is denoted by @xmath53 . to a given graph @xmath0 with vertex set @xmath54
, we associate a class of real , symmetric matrices as follows : @xmath55|a \in s_{n},\ , \mbox{for}\,\,\ , i\neq j , a_{ij}\neq0\leftrightarrow ij \in e(g)\}.\ ] ] note that there is no restriction on the value of @xmath56 with @xmath57 and the adjacency matrix @xmath58 belongs to @xmath59 , where the adjacency matrix of a graph @xmath0 is a square @xmath60-matrix of size @xmath35 , whose @xmath61-th entry is @xmath62 if and only if @xmath63 is adjacent to @xmath64 , since there are no loops in the graph , the diagonal entries of the adjacency matrix are zero . on the other hand ,
the graph of an @xmath35-square symmetric matrix @xmath51 , denoted by @xmath65 , is the graph with vertices @xmath54 and the edge set @xmath66 the minimum rank of @xmath0 is defined to be @xmath67 while the maximum nullity of @xmath0 is defined as @xmath68 we have @xmath69 the underlying idea for the zero forcing set of a graph is that a black vertex is associated with a coordinate in a vector that is required to be zero , while a white vertex indicates a coordinate that can be either zero or nonzero .
changing a vertex from white to black is essentially noting that the corresponding coordinate is forced to be zero if the vector is in the kernel of a matrix in @xmath59 and all black vertices indicate coordinates assumed to be or previously forced to be @xmath70 . hence the use of the term zero forcing set " , see @xcite .
the support of a vector @xmath71 , denoted by @xmath72 , is the set of indices @xmath73 such that @xmath74 .
let @xmath75 be a zero forcing set of @xmath0 and @xmath76 . if @xmath77 and @xmath78 , then @xmath79 , stated in @xcite .
also from @xcite , we have @xmath80 for a graph @xmath0 . in this paper
, we consider corona product and lexicographic product of graphs in the context of zero forcing number .
this paper consists of three sections .
section @xmath62 includes introduction .
sections @xmath81 and @xmath82 include several results related to the zero forcing number of corona and lexicographic product of graphs , respectively . |
regarding the production of strange baryons , the existing extensive studies on the production of @xmath4 and @xmath5 hyperons , which have strangeness @xmath6 , have accumulated many data for various physical quantities that are relevant for describing the production mechanisms of @xmath6 baryons and the parameters of nucleon resonances . by contrast , the production of multi - strangeness @xmath0 ( @xmath7 ) and @xmath1 ( @xmath8 ) baryons has not been studied very often .
one reason is that the processes for producing multi - strangeness baryons from non - strangeness initial states have very small cross sections . as a result ,
no significant information on multi - strangeness baryons has been added during the past two decades @xcite .
recently , however , the interest in multi - strangeness physics is increasing .
for example , the major research programs at j - parc , which has started its operation recently , include strangeness physics @xcite .
since the j - parc facility can provide energetic kaon beams , the production of @xmath2 should be possible .
furthermore , the planned upgrade of the kaon - beam energy will allow the production of @xmath0 resonances , thus providing a new turning point in the study of multi - strangeness baryons .
furthermore , the @xmath9anda collaboration has a plan to investigate the @xmath10 reaction at fair @xcite , and the clas collaboration at jlab has established the feasibility of investigating @xmath0 spectroscopy via photoproduction reactions like @xmath11 and @xmath12 @xcite .
the reported data for the total and differential cross sections as well as the @xmath13 and @xmath14 invariant mass distributions for the @xmath11 reaction @xcite are the first data measured for the exclusive photoproduction of the @xmath0 .
recent theoretical investigations of this reaction can be found in refs .
@xcite .
the investigation of multi - strangeness baryons will shed light on our understanding of baryon structure by allowing us to distinguish various hadron models .
( see , for example , ref .
@xcite and references therein . )
however , this requires knowledge of the quantum numbers of the baryons since the spin and parity quantum numbers heavily depend on the internal structure of the baryons and the underlying dynamics .
it is of crucial importance , therefore , to identify the quantum numbers of the observed @xmath0 baryons .
however , identifying the spin and parity quantum numbers of multi - strangeness baryons is very challenging .
for example , the fact that the @xmath15 baryon has spin-3/2 was only recently confirmed by experiment @xcite , more than 40 years after the discovery of the @xmath16 baryon @xcite . for the cascade baryons , only for three states , @xmath2 , @xmath17 , and @xmath18 , of the eleven cascade states reported in pdg @xcite , the spin and parity quantum numbers are known . in a recent report @xcite , the babar collaboration claimed that the @xmath19 has @xmath20 .
what is surprising is that the parity of the cascade ground state @xmath2 has never actually been measured ; the positive parity quoted by the pdg @xcite is taken from quark - model calculations . in this article
, we describe a novel , _ model - independent _ way to determine the parity of spin-1/2 @xmath0 baryons . to this end
, we consider the reaction , @xmath21 , where the arguments @xmath22 , @xmath23 , @xmath24 , and @xmath25 stand for the four - momenta of the respective particles . although the production mechanisms are highly model - dependent @xcite , our method is based on bohr s theorem @xcite and thus a direct consequence of the rotation and parity - inversion symmetries of the reaction amplitudes @xcite , and , therefore , model - independent .
furthermore , the cross section for this reaction is large enough to be measured experimentally .
when the produced @xmath0 has spin-1/2 , the most general spin structure of the reaction amplitude constrained by symmetry principles for the reaction of @xmath3 can be written as @xmath26 where @xmath27 and @xmath28 , respectively , apply to @xmath0 having positive or negative parity .
the amplitude functions @xmath29 are functions of mandelstam variables .
the unit vectors @xmath30 and @xmath31 are defined as @xmath32 , respectively . without loss of generality
, we may choose the coordinate systems such that @xmath33 is the unit vector along the positive @xmath34-axis and @xmath31 along the positive @xmath35-axis so that @xmath30 is the unit vector along the positive @xmath36-axis .
the reaction plane is then defined by the two vectors @xmath37 and @xmath30 .
this enables us to write the reaction amplitudes in a generic manner as @xmath38 where @xmath39 is the @xmath40 unit matrix . depending on the parity of @xmath0 , either @xmath41 or @xmath42 .
the unpolarized cross section is then given by @xmath43 the most interesting spin observable related to the parity of the @xmath0 is the ( diagonal ) spin - transfer coefficient @xmath44 defined by @xmath45 for @xmath46 . in terms of the cross sections ,
this corresponds to @xmath47 - \left[d\sigma_i^{}(+-)+d\sigma_i^{}(-+)\right ] } { \left[d\sigma_i^{}(++ ) + d\sigma_i^{}(--)\right ] + \left[d\sigma_i^{}(+-)+d\sigma_i^{}(-+)\right ] } , \label{eq:4a}\ ] ] where @xmath48 stands for the differential cross section with the polarization of the target nucleon and of the produced @xmath0 along the @xmath49-direction .
the first and second @xmath50 arguments of @xmath48 indicate the parallel ( @xmath51 ) or anti - parallel ( @xmath52 ) spin - alignment along the @xmath49-direction of the target nucleon and produced @xmath0 , respectively . the spin - transfer coefficient @xmath44 should be a function of the energy and scattering angle , in general .
however , because of the spin structure of the amplitude , we can see that @xmath53 is constant and @xmath54 where @xmath55 is the parity of the produced @xmath0 .
this is a direct consequence of the spin structures of the reaction amplitudes as exhibited in eq .
( [ eq:1 ] ) , which is a realization of reflection symmetry in the reaction plane .
therefore , the measurement of @xmath53 will directly determine the parity of the @xmath0 . another way to determine
the parity of spin-1/2 @xmath0 baryon is to use two single - polarization observables , namely , the target - nucleon asymmetry @xmath56 and the recoil-@xmath0 asymmetry @xmath57 , which are defined by @xmath58 + 2\,\mbox{im } \left[m_j m^*_k \right ] , \\ \frac{d\sigma}{d\omega}p_i & \equiv & \frac{1}{2}\,\mbox{tr}\left ( m m^\dagger \sigma_i \right ) = 2\,\mbox{re}\left[m_0m^*_i \right ] - 2\,\mbox{im}\left[m_j m^*_k \right],\end{aligned}\ ] ] where the subscripts @xmath59 run cyclically . then the spin structures of the amplitudes of eq .
( [ eq:1 ] ) immediately give @xmath60 , \qquad \frac{d\sigma}{d\omega}\left ( t_y - p_y\right ) = 0 , \end{aligned}\ ] ] for positive - parity @xmath0 and @xmath61,\end{aligned}\ ] ] for negative - parity @xmath0 .
this can be used to determine the parity of the @xmath2 unless @xmath62 .
in summary , we have studied the spin structure of the production amplitudes in the reaction of @xmath3 .
we found that the spin - transfer coefficient @xmath53 , which is a double - polarization observable , can give a direct measurement of the parity of the @xmath0 baryon .
furthermore , the combinations of the two single - polarization observables , the target asymmetry @xmath63 and the recoil asymmetry @xmath64 , can give another way to check the parity of the produced @xmath0 . since the @xmath0 is self - analyzing
, these asymmetries can be measured by preparing polarized nucleon targets at the current facilities like the j - parc .
although we considered the parity of spin-1/2 @xmath0 baryons in this article , our discussion can be generalized to the @xmath0 of higher spin and to the polarization observables in @xmath0 photoproduction @xcite .
similar relations can be found for the determination of the parity of the @xmath1 baryons as well .
this work was partly supported by the national research foundation of korea funded by the korean government ( grant no .
nrf-2011 - 220-c00011 ) . the work of k.n .
was also supported partly by the ffe grant no .
41788390 ( cosy-058 ) . | to understand the structure of hadrons and their production mechanisms , it is very important to be able to identify their quantum numbers . for investigating the spin and parity quantum numbers of @xmath0 and @xmath1 baryons , in particular
, it is very important to understand strong interactions in the strangeness sector . at present ,
such quantum numbers are known for only a few of the @xmath0 baryons , and even the parity of the ground state @xmath2 has never been measured . in this article , we present a novel , model - independent way to determine the parity of @xmath0 baryons solely based on symmetry considerations for the hadronic reaction @xmath3 . |
all stable elements heavier than lithium were forged in stars .
big bang nucleosynthesis produced helium efficiently , but it was halted by the expansion of the universe before it could go much further ( walker et al .
on the other hand , all stars observed to date have substantial mass fractions of carbon , silicon , iron and other elements that are the products of the final stages of stellar evolution .
in fact , even the most pristine stars observed ( cayrel et al . 2004 ; frebel et al . 2008 ; caffau et al .
2011 ) have been polluted by this material .
the earliest stellar generation , referred to as population iii ( pop iii ) , is missing . while this absence could be due to the formation of an extremely small number of metal - free stars , detailed theoretical studies suggest that it is more likely that these stars were too massive to survive to the present day ( scannapieco 2006 ; brook 2007 ) .
in fact , the absence of heavy elements drastically decreases the cooling rates in collapsing star - forming gas , such that that primordial gas clouds would have been much less susceptible to fragmentation , perhaps forming @xmath17 - 10@xmath18 solar mass star forming clumps ( hutchins 1976 ; abel , bryan , & norman 2000 ; bromm , coppi , & larson 2002 ; bromm & loeb 2003 ) .
furthermore , the strong accretion rates onto the central protostellar cores in such pristine clumps can not be arrested by radiation pressure , bipolar outflows , or rotation ( ferrara 2001 ) , meaning that these regions may have formed stars with masses hundreds of times greater than the sun . on the other hand ,
recent work suggests that physical processes such as enhanced hd cooling in shock - compressed primordial gas ( johnson & bromm 2006 ) , a lack of magnetic fields in primordial turbulent clouds ( padoan 2007 ) , photoionization of turbulent primordial clouds ( clark 2011a ) , fragmentation of the protostellar accretion disks ( stacy 2010 , clark et al .
2011b ) , early termination of accretion ( mckee & tan 2008 , hosokawa 2011 ) , and gravitational torques ( greif 2012 ) , may have led to primordial stars with masses @xmath19 in single or perhaps binary systems ( turk , abel , & oshea 2009 ) . even at these comparatively low masses , direct detections of primordial stars
would be possible only through observations of the high - redshift universe , relying on what is likely to have been an extended transition between metal free and pre - enriched ( population ii / i ) star formation ( scannapieco 2003 ; jimenez & haiman 2006 ; trenti & stiavelli 2007 ; 2009 ; maio 2010 ) , as well as the unusual observable signatures of metal free stars . during their lifetimes
, for example , the lack of heavy elements in pop iii stars drastically reduces their opacities , resulting in much higher surface temperatures and strong ultraviolet spectroscopic features that would distinguish them from current stellar populations ( schaerer 2002 ; nagao 2008 ) . alternatively ,
if they were extremely massive , pop iii stars could be detectable as they ended their lives as tremendously powerful pair - production supernovae ( bond 1984 ; heger & woosley 2002 ; scannapieco 2005 ; whalen 2012 ) .
furthermore , the reliance on h@xmath20 + hd cooling in primordial gas may have effects beyond masses of stars , such as affecting the phase nature of the interstellar medium and the star formation rate ( norman & spaans 1997 ) .
whatever the detection method , when and where metal free gas condensed into stars is a question of fundamental importance in planning searches for this remarkable early generation of stars . on cosmological scales ,
the key issue is the time it takes for heavy elements to propagate from one galaxy to another .
as shown in scannapieco et al .
( 2003 ) , the distances between these oases of early star - formation are so vast that for several 100 million years , the universe was divided into two regions : one in which galaxies formed out of material that was already polluted with heavy elements well above the minimum
critical mass fraction , " @xmath0 at which normal stellar evolution occurs ( schneider et al .
2003 ; bromm & loeb 2003 ; omukai et al .
2005 ) , and one in which galaxies were formed from initially pristine material .
the evolution of initially - pristine galaxies is especially interesting , as it depends on two important theoretical issues .
the first of these is the uncertain value of @xmath0 which is expected to lie in the range from @xmath21 to @xmath22 , ( or from @xmath23 to @xmath24 times solar metallicity ) , depending on whether the cooling is dominated by dust grains ( omukai et al .
2005 ) or by the fine structure lines of carbon and oxygen ( bromm and loeb 2003 ) .
the second important issue is the rate at which the gas within the galaxy can be polluted above this critical value by the turbulent mixing of heavy elements ( pan & scalo 2007 ) . within a given galaxy , the key quantity to characterize
the transition is the fraction , @xmath25 , of the interstellar gas with metal concentration @xmath26 below @xmath4 as a function of time .
the temporal behavior of this fraction depends not only on the rate at which new sources of metals are released to the interstellar gas , but , more importantly , on the transport and mixing of the metals in the interstellar gas .
for example , a high mixing efficiency would result in a rapid decrease in @xmath25 , and hence a sharp transition as the average concentration exceeds the threshold @xmath4 . on the other hand
, a low mixing efficiency would lead to a gradual transition .
the interstellar medium is known to be turbulent and highly compressible , and the turbulent motions are likely to be supersonic .
therefore , understanding mixing in supersonic turbulence is crucial to understanding the evolution of primordial gas in early galaxies .
the evolution of the pristine fraction was considered by oey ( 2000 , 2003 ) in the context of the sequential enrichment model , which , however , does not reflect or correctly capture the physics of mixing in interstellar turbulence . in pan ,
scannapieco , & scalo ( 2012 ; hereafter pss ) , we developed a theoretical approach to model the evolution of the pristine fraction in statistically homogeneous turbulence .
the starting point of our theoretical model was the pdf method for turbulent mixing , since the pristine fraction @xmath25 corresponds to the far left tail of the metallicity pdf .
the pdf equation for passive scalars can not be solved exactly , and we adopted several closure models from the literature and derived predictions for the evolution of the pristine fraction . using numerical simulations , we showed that a class of pdf closure models , called self - convolution models , provided successful fitting functions to the evolution of @xmath25 for a limited range of flow mach number and pollution properties .
these models are based on the physical picture of turbulence stretching pollutants and causing a cascade of concentration structures toward small scales ( pan & scalo 2007 ) , a picture that is generally valid in turbulent flows at all mach numbers ( pan & scannapieco 2010 ; hereafter ps10 ) .
mixing occurs as the scale of the structures becomes sufficiently small for molecular diffusivity to efficiently operate , and the homogenization between neighboring structures can be described as a convolution of the concentration pdf .
as discussed in more detail below , the models are dependent on two major parameters : @xmath27 which sets the characteristic timescale for convolution of the metal abundance pdf through turbulent stretching of concentration structures , and @xmath7 , which quantifies the degree of spatial locality of the pdf convolution process .
here we use a suite of numerical simulations to expand these results and show that the generalized self - convolution model provides good fits for all the turbulence and pollutant conditions relevant for primordial star formation . note that , besides affecting the temperature directly , h@xmath20 + hd cooling without heavy elements makes neutral primordial gas less compressible ( scalo & biswas 2002 ; spaans & silk 2005 ) , emphasizing the importance of studying the mach number dependence of the pollution processes . by comparing this model to simulations with mach numbers
@xmath28 2.1 , 3.5 , & 6.2 , in which pollutants are added at different length scales , and 4 different initial values of @xmath16 we are able to obtain detailed fits of @xmath27 and @xmath7 over all mach numbers , initial pollution fraction , and pollution scales of interest .
we tabulate and obtain empirical fits to our results , and show how they can be used both to make simple one - zone estimates of the evolution of @xmath5 within a single high - redhshift galaxy , as well as to construct a subgrid model for numerical simulations , which tracks the evolution of the primordial fraction below the resolution scale .
the model is expected to improve the prediction for the evolution of the primordial gas fraction in early galaxies , and should also be applicable to any physical problem in which the unresolved , unmixed fraction needs to be tracked throughout a simulation .
a realistic simulation for the pollution of the pristine gas in early galaxies needs to properly specify the driving mechanism of the interstellar turbulence .
a variety of physical processes may contribute to turbulent motions in the interstellar medium .
turbulence can be produced at galactic scales , e.g , by the gas infall from the halo or by merger events during the assembly of the galaxy .
supernova explosions by the first stars also contribute to the turbulent energy . in the present work ,
we adopt a solenoidal driving force in the simulations , which may be a good approximation if the interstellar turbulence is mainly driven by large - scale motions or instabilities , associated with galaxy formation , mergers or interactions . on the other hand ,
if the primary source for turbulent energy is stellar winds and supernova explosions , the driving force may be compressive rather than solenoidal .
in that case , several questions need to be addressed by future studies .
first , as shown by federrath et al .
( 2010 ) , a compressively driven supersonic turbulent flow shows significantly different statistics than the solenoidal case .
the amplitude of the density fluctuations is much larger due to stronger compressions , and the intermittency of the velocity field is significantly higher .
we will speculate how these features may qualitatively affect the parameters in our convolution models .
a quantitative understanding may be obtained by a suite of simulations using driving forces at different degrees of compressibility .
second , if supernovae are the primary energy source of turbulence , the injection of heavy elements is highly correlated with the turbulent driving force .
this is not accounted for in our simulations either .
these issues are complicated and require a systematic investigation .
this is , however , beyond of the scope of the current paper , which is essentially an initial step that provides the theoretical framework and useful subgrid methodology for the modeling of the pristine gas pollution in early galaxies .
the structure of this work is as follows . in
2 we introduce the pdf formulation for turbulent mixing , and discuss the fundamental mixing physics . in 3 , we describe the self- convolution closure models , and show how they can be used to predict the evolution of the pristine fraction in statistically homogeneous turbulence . in 4 we describe our suite of numerical simulations , and in 5 we use them to test and constrain the self - convolution models over the full range of turbulence and pollution conditions necessary to model primordial star formation .
having fixed the parameters in the theoretical models with our simulation results we show how they can then be applied to one zone models of high - redshift galaxies in 6 , and in 7 we show how they can be used to construct subgrid models for the unresolved primordial fraction in large , numerical simulations .
a summary of our conclusions is given in 8 .
the mixing of heavy elements in the interstellar medium can be studied by tracking the evolution of the concentration field , @xmath29 , defined as the ratio of the local density of these elements to the total gas density .
the concentration field obeys the advection - diffusion equation , @xmath30 where @xmath31 and @xmath32 denote the density and velocity fields in interstellar turbulence , @xmath33 is the molecular diffusivity , and the term @xmath34 represents continuing sources of heavy elements or pollutants .
the concentration field , @xmath29 , could also represent the local abundance of a specific element , but here we are interested in the mass fraction of all metals at a given location .
the pristine mass fraction in a flow corresponds to the low tail of the probability distribution function ( pdf ) of the pollutant concentration , so we adopt a pdf approach for turbulent mixing .
this approach was first established for the turbulent velocity field ( monin 1967 and lundgren1967 ) and was later extended to mixing of passive or reactive species in turbulent flows ( ievlev 1973 ; dopazo and obrien 1974 ; pope 1976 ; obrien 1980 ; pope 1985 ; kollemann 1990 ; dopazo et al .
it has been particularly successful in the field of reacting turbulent flows ( e.g. , haworth 2010 ) .
however , most of the work on the pdf modeling of turbulent mixing has been dedicated to incompressible or weakly compressible turbulence . in order to apply the method to the interstellar media of galaxies , where strong density fluctuations exist ,
pss generalized the pdf formulation to mixing in highly compressible turbulent flows at large mach numbers , emphasizing the importance of using a density weighting scheme . as in pss , herewe
use a statistical ensemble to define a density - weighted concentration pdf , @xmath35 , where @xmath36 is the fine - grained pdf in a single realization , @xmath37 denotes the ensemble average , and @xmath26 is the sampling variable .
the probability distribution defined here at a given position and time is an average over many independent realizations in the statistical ensemble .
the ensemble average density , @xmath38 , is in general a function of @xmath39 and @xmath40 . in pss
, @xmath38 was implicitly assumed to be constant .
below we consider the more general case that accounts for the spatial and temporal variations of @xmath38 .
an important motivation for using a density - weighting factor is that , when studying mixing of primordial gas in early galaxies , it is appropriate to consider the mass fraction , rather than the volume fraction , of the interstellar gas with @xmath41 .
an equation for @xmath42 can be derived using the advection - diffusion equation ( [ advection ] ) , for @xmath43 and the continuity equation , for @xmath31 . applying the same method as in appendix a of pss and accounting for the spatial and temporal dependence of @xmath44 , we find , @xmath45 where @xmath46 denotes the conditional ensemble average with density weighting .
for any physical quantity , @xmath47 , this ensemble average is defined as , @xmath48 where the conditional average @xmath49 without density - weighting is evaluated by selecting and counting only those realizations satisfying the constraint that the concentration @xmath29 at @xmath39 and @xmath40 is equal to @xmath26 .
setting @xmath38 to be constant reduces eq .
( [ pdfeq ] ) to eq .
( 2.2 ) in pss .
derivations of analogous pdf equations for passive and reacting scalar turbulence in the incompressible case can be found in e.g. , pope ( 2000 ) and fox ( 2003 ) .
the pdf equation is essentially a liouville equation for the conservation of the concentration probability in phase space . in analogy to the liouville equation in kinetic theory , the concentration field corresponds to the particle momentum and the advection - diffusion equation corresponds to the particle equation of motion ( pss ) .
although the pdf equation is derived in the context of a statistical ensemble , it is also useful for the study of scalar statistics in a single realization , i.e. , the real flow
. if the flow and the scalar statistics are spatially homogeneous , @xmath50 and @xmath51 , are independent of @xmath39 , and the ergodic theorem indicates that the statistics over an ensemble is equivalent to that over the spatial domain of a single realization .
this means that @xmath51 is equal to the pdf , @xmath52 , computed from the density and concentration fluctuations over the entire volume , @xmath53 , of the real flow domain . at galactic length scales ,
the assumption of statistical homogeneity is likely invalid .
for example , coherent mean flows , such as galactic rotation , infall or outflow , may exist at large scales in the interstellar medium .
also a large - scale metallicity gradient may develop if the star formation rate has a radial dependence , and the metallicity statistics can vary substantially from region to region . however , the ensemble - defined pdf @xmath54 may still be used to study this spatial dependence if it is understood as corresponding to the concentration pdf for local fluctuations in a region of considerable size in the real flow .
if the size is selected to be large enough to allow sufficient statistics , but small enough for local statistical homogeneity to be restored ( i.e. , considerably smaller than the characteristic scale for the mean flow or mean concentration gradient ) , then the ergodic theorem will apply locally , and @xmath51 will represent the pdf in the region around @xmath39 .
in fact , in an attempt to build a subgrid model for the pollution of pristine interstellar gas , a concentrationpdf characterizing the fluctuations in local regions is defined by applying a spatial filter to the real flow ( see 7 and appendix a ) , and the equation derived for the filtered pdf is identical to eq .
( [ pdfeq ] ) for @xmath54 over an ensemble .
this confirms the equivalence of @xmath54 to the local concentration pdf in the real flow and the applicability of eq .
( [ pdfeq ] ) to the study of mixing in a statistically inhomogeneous setting .
the last three terms in eq .
( [ pdfeq ] ) correspond to turbulent advection , molecular diffusivity and source terms in the advection - diffusion equation .
we give a brief general discussion for each term below , and refer the interested reader to pss for more details . the second term in eq .
( [ pdfeq ] ) is the advection term .
as it takes a divergence form , it conserves the global pdf with density - weighting , i.e. , the integral of @xmath55 over the entire flow domain .
this corresponds to a fundamental issue in mixing physics : the turbulent velocity field does not homogenize at all by itself .
intuitively , a velocity field moves , stretches and redistributes the concentration field , but it does not change the _ mass _ fraction of the fluid elements at a given concentration level , and thus does not truly mix . on the other hand , without density - weighting ,
the advection term in the pdf equation would not be a divergence term if the flow is compressible , but instead be a term representing the effect of expansions and compressions on the volume fraction of fluid elements at a given concentration .
this effect is different from mixing , and makes the pdf modeling more complicated .
therefore , in addition to the practical reasons mentioned above , adopting a density - weighting scheme is also strongly motivated on a theoretical basis .
the advection term vanishes and need not be considered in a flow that is statistically homogeneous . in case of statistical inhomogeneity
, the term corresponds to the transport of the concentration pdf by the velocity field : turbulent advection causes changes in the local pdf as it moves the fluid elements around .
if one is interested in the concentration fluctuations in a local region , this transport effect must be accounted for carefully .
a similar advection term exists in the equation for the filtered pdf of the local concentration fluctuations , which is used to build a subgrid model for the pollution of pristine gas in early galaxies in 7 . in that case , a proper treatment of the advection term is essential , as it is responsible for the flux of pristine mass fraction into or out of a computation cell due to the velocity field ( see 7 ) .
however , an exact treatment of the advection term is impossible due the usual difficulty in turbulence theory known as the closure problem ( see , e.g. , pope 2000 ) . a similar problem exists for the diffusivity term , which is discussed in more details in 2.2 .
we will adopt the commonly - used eddy - diffusivity approximation to model the advection term in 7 . as shown in pan & scalo ( 2007 ) and pss , the molecular diffusivity term in eq .
( [ pdfeq ] ) ,
i.e. , the first term on the right hand side , is the only term responsible for the homogenization of the concentration fluctuations .
the term can be rewritten as ( see appendix a ) , @xmath56 where the first term on the right hand side is a spatial diffusion of the fine - grained concentration pdf ( defined in 2 ) by the molecular diffusivity . as a divergence term , it conserves the global pdf and does not contribute to true homogenization .
the last term can be thought of as an anti - diffusion process in concentration space , as the coefficient is negative definite .
it continuously narrows the pdf toward the mean value .
unfortunately , this diffusivity term does not have an exact or closed form . as it involves concentration gradients , it is nonlocal and dependent on the two - point concentration pdf .
deriving an equation for the two - point pdf gives rise to terms that require knowledge of three - point statistics , and so on , leading to a chain of multipoint pdf equations similar to the bbgky hierarchy in kinetic theory ( lundgren 1967 ; dopazo and obrein 1974 ) .
assumptions must be made to truncate the hierarchy to obtain a closed set of equations .
this is the so - called closure problem . in pss , we considered a number of closure models and showed that a class of models based on the convolution of the concentration pdf are particularly successful in fitting the simulation results for the pollution of pristine material in turbulent flows .
one of these convolution models for closure of the diffusivity term was used in our earlier modeling of the primordial fraction ( pan & scalo 2007 ) .
we will summarize these models in 3 .
although the diffusivity term lacks an apparent dependence on the velocity field , an efficient homogenization of the concentration field does rely on the existence of turbulent motions .
the action of the diffusivity term is very slow at large length scales where the pollutants are injected , because the molecular diffusivity @xmath33 is usually tiny in most natural environments including the interstellar medium .
for example , pan & scalo ( 2007 ) estimated @xmath33 for the galaxy s ism , weighted by residence time in different phases of the neutral gas , of about @xmath57 @xmath58 s@xmath59 , with a corresponding diffusivity scale of @xmath60 pc , where @xmath61 is the galactic turbulence integral scale in units of 100 pc , and @xmath62 is the turbulent rms velocity in units of 10 km s@xmath59 .
this is the scale to which turbulence must stretch the pollutants in order for mixing to occur .
therefore , mixing by molecular diffusivity is negligible in the absence of a velocity field .
a turbulent velocity can act as a catalyst and significantly accelerate the mixing process .
this implicit role of turbulence on scalar homogenization is through the dependence of the diffusivity term on the concentration gradients ( see eq .
( [ pdfeq ] ) ) . by continuously stretching the pollutants , turbulence produces structures at smaller and smaller scales , resulting in an enormous increase in the concentration gradients .
once the structures reach a small scale called the diffusion scale , where molecular diffusivity operates faster than turbulent stretching , they are homogenized efficiently by the diffusivity term .
this suggests that the mixing timescale is essentially determined by turbulent stretching , even though the velocity itself does not truly mix .
it is the cooperation of molecular diffusivity and turbulent motions that gives a significant mixing efficiency . the last term in eq .
( [ pdfeq ] ) is the source term , corresponding to the injection of new pollutants into the turbulent flow . in general , pollutants are any source materials with a composition pattern different from that in the existing flow .
thus for mixing of heavy elements in the interstellar media of galaxies , the source term would include both ejecta from supernova and stellar winds and , if it exists , infall of low - metallicity or primordial gas .
to evaluate the source term , it is actually not necessary to compute its conditional average form in eq .
( [ pdfeq ] ) . instead
, it can be estimated by directly considering the rate at which the pollutants are injected and how they affect the concentration pdf in the flow .
for example , assuming that the supernova ejecta are nearly pure metals , the source term for the supernova contribution would take a delta function form at @xmath63 with a coefficient depending on the supernova rate , ejecta mass etc ( see 6 and pan and scalo 2007 ) .
on the other hand , a primordial infall would give a delta function at @xmath64 .
therefore , the effect of continuous sources of primordial gas and new metals is to force spikes in the concentration pdf at small and large concentration values , respectively .
the primary goal of this study is to investigate how the diffusivity term in the pdf equation , representing the homogenization by molecular diffusivity catalyzed by turbulent motions , reduces the fraction of pristine material in a turbulent flow . to understand the fundamental physics
, we consider an idealized problem : mixing of decaying scalars ( i.e. , @xmath65 = 0 ) in statistically stationary and homogeneous turbulence .
the initial scalar field is also assumed to be statistically homogeneous .
clearly , this idealized problem is much simpler than in a realistic galactic environment .
however , the simplified setting is extremely useful for understanding the underlying physics .
as discussed in 2 , under the assumption of statistical homogeneity , the advection term vanishes and the pdf , @xmath42 , is independent of @xmath39 and is equivalent to that computed from the spatial fluctuations in a single realization . with these simplifications ,
the pdf equation becomes , @xmath66 where we used eq .
( [ ensemblediffusivityterm ] ) .
the diffusivity term is the only term in the simplified pdf equation . in analogy with the mixing of primordial gas in early galaxies , we set the initial condition of the decaying scalar to be bimodal , consisting of pure pollutants ( @xmath63 ) and completely unpolluted flow ( @xmath64 ) .
this corresponds to a double - delta function form for the initial concentration pdf , @xmath67 where @xmath68 and @xmath69 are the initial probabilities / fractions of pristine gas and pollutants , respectively , and @xmath70 from normalization . before introducing closure models for the diffusivity term
, we discuss the evolution of the concentration variance , which helps reveal the general physics of turbulent mixing . in terms of the density - weighted pdf ,
the average concentration with density weighting is written as @xmath71 , which is equal to @xmath72 .
similarly , the density - weighted variance is expressed as @xmath73 , which is equivalent to @xmath74 with @xmath75 being the fluctuating part of the concentration field .
taking the 2nd - order moment of eq .
( [ simplepdfeq ] ) yields @xmath76 , which can also be derived directly from the advection diffusion equation using the assumption of statistically homogeneity ( see ps10 ) . we therefore have , @xmath77 the mixing timescale , @xmath78 , is the ratio of the concentration variance to its dissipation rate , @xmath79 clearly , @xmath80 is the timescale for the variance decay , and thus also characterizes the rate at which the diffusivity term reduces the pdf width .
as discussed in 2 , the mixing timescale , @xmath78 , depends on the rate of turbulent stretching , which produces concentration structures at small scales and feeds molecular diffusivity with large concentration gradients . in the classical phenomenology for turbulent mixing ,
the continuous production of small - scale structures is described as a cascade , in the sense that the process proceeds progressively faster toward smaller scales .
the picture is similar to the cascade of kinetic energy .
it predicts that the mixing timescale is determined mainly by the eddy turnover time at the scale where the pollutants are injected , but insensitive to the small diffusion scale , where the molecular diffusivity acts to homogenize .
the prediction has been confirmed by pp10 using simulated supersonic turbulent flows with solenoidal driving force .
they found that , at all mach numbers explored , the mixing time was close to the eddy turnover time at the pollutant injection scale , suggesting that the cascade picture , originally proposed for mixing in incompressible flows , is valid also for highly compressible turbulence .
ps10 also found that compressible modes in solenoidally - driven supersonic turbulence do not make a significant contribution to the cascade of concentration structures to small scales .
compressible modes consist of both expansions and compressions .
as passive scalars simply follow the flow motions , the compression events would decrease the length scale of the concentration structures , or equivalently increase the concentration gradients .
this makes a contribution to enhance the mixing rate . on the other hand
, the expansion events would cause the mixing process to slow down .
the two opposite effects tend to counteract each other . however , they do not exactly cancel out , and the effect of compressions appear to win slightly . using the density fluctuations as a measure for the strength of the compression events , pp10 found that , in solenoidally driven flows , the net contribution of compressible modes to the enhancement of the concentration gradients is much smaller than the solenoidal modes ( see 5 of pp10 ) .
a limitation in the effect of compressible modes on mixing is that the squeezing effect by compressions is not continuous due to the gas pressure .
it is likely a compressed region would expand before being squeezed by a second compression event .
such limitation does not exist in the stretching by solenoidal modes , which operates continuous and unlimited by the gas pressure .
the stretching effect by incompressible modes appears to be the primary mixer " in the simulated flows with solenoidal driving at all mach numbers . as a consequence , a useful measure for the mixing efficiency would be the fraction of energy contained in solenoidal modes in the inertial range of the flow , which is responsible for the cascade of passive scalars toward the diffusion scale .
a statistical analysis of the simulated velocity fields by pp10 showed that the solenoidal energy fraction in the inertial range decreases with @xmath11 for @xmath81 and then saturates at an equipartition value of 2/3 at @xmath82 .
this provides a satisfactory explanation for the behavior of the mixing timescale normalized to the flow dynamical time as a function of @xmath11 . the normalized mixing timescale increases with @xmath11 for @xmath81 and saturates at larger @xmath11 .
this finding supports our argument above that compressible modes are less efficient at enhancing mixing in solenoidally driven flows , and , with a larger fraction of compressible energy in the inertial range , the mixing is slower . it remains to be checked if the normalized timescale as a function of @xmath11 has the same behavior in supersonic turbulence with completely compressive driving .
one issue is that , quantitatively , the energy fraction of solenoidal or compressible modes at inertial - range scales as a function of @xmath11 in fully - developed flows with compressive driving may be different from the solenoidal case .
second , as shown in federrath et al .
( 2010 ) , at the same @xmath11 , the density fluctuations in a compressively driven flow are much stronger .
this implies that the net effect of compressible modes on amplifying the concentration gradients would be more efficient than in the solenoidal driving case .
if the contribution from compressible modes to the mixing efficiency in highly supersonic compressively driven flows is comparable to or even faster than the solenoidal modes , the behavior of @xmath78 as function of @xmath11 may be qualitatively different , with the normalized timescale decreasing with @xmath11 at sufficiently high @xmath11 .
on the other hand , if compressible modes in supersonic flows with compressive driving are still less efficient at enhancing mixing than solenoidal modes , one may expect a similar behavior for the normalized mixing timescale .
we will investigate these possibilities in a future work .
a variety of closure models have been developed for the diffusivity term in the pdf equation .
pss considered several existing models from the literature , including the mapping closure model ( chen et al .
1989 ) , based on an approximation for the exact but unclosed form of the diffusivity term , and a class of models , referred to as self - convolution models by pss , based largely on a physical picture of the turbulent mixing process ( curl 1963 , dopazo 1979 , janicka et al .
1979 , venaille and sommeria 2007 , villermaux and duplat 2003 , duplat and villermaux 2008 ) .
one of the self - convolution models was used in the initial study of pollution of pristine gas by pan and scalo ( 2007 ) . by a detailed comparison with numerical simulations of turbulent mixing in two compressible flows at mach 0.9 and 6.2
, pss showed that the convolution models provide both clear physical insights and successful fitting functions for the decay of the pristine mass fraction .
here we give a brief introduction of the convolution models , and refer the interested reader to pss for details .
there has been compelling evidence that the dominant scalar structures at small scales are 2d sheets or edges ( e.g. , pan and scannapieco 2011 and references therein ) .
the rate at which the scalar sheets are produced is determined mainly by the turbulent stretching rate at large length scales . with time , the sheets become thinner , and once the thickness of the sheets is sufficiently small for molecular diffusivity to efficiently operate , the neighboring sheets are homogenized , leading to a reduction in the pdf width . the physical picture outlined above can be approximately described by an integral equation for the concentration pdf , @xmath83 where @xmath84 and @xmath85 denote the concentrations in two nearby sheets prior to the mixing by molecular diffusivity , and the delta function in the integrand arises from the assumption that a perfect homogenization occurs instantaneously once two scalar sheets are sufficiently stretched for molecular diffusivity to take effect .
here @xmath86 is the turbulent stretching rate that controls the rate at which the pdf convolution proceeds .
the last term in eq .
( [ curl ] ) is the destruction " of the previous pdf due to the mixing event . using the properties of delta functions , eq .
( [ curl ] ) can be written as @xmath87 $ ] , which shows that turbulent mixing is essentially assumed to be a self - convolution process .
for a reason to be clarified soon , eq .
( [ curl ] ) was referred to as the discrete convolution model in pss .
it was first proposed by curl ( 1963 ) in a study of droplet interactions in a two - liquid system , and was later extended to model mixing in turbulent flows ( dopazo 1979 , janicka et al .
several variants and generalizations of eq .
( [ curl ] ) have been proposed to solve the problems of the model for turbulent mixing .
one problem of curl s model is that , for a double delta initial pdf ( eq . ( [ initialpdf ] ) )
, it produces unphysical spikes in between the initial delta functions . in order to avoid this , dopazo ( 1979 ) and janicka et al .
( 1979 ) suggested replacing the delta function in eq .
( [ curl ] ) by a general function , @xmath88 , that is smooth in between @xmath84 and @xmath85 .
pss showed that , with this modification , the model gives essentially the same prediction for the evolution of the pristine mass fraction .
another weakness of the convolution pdf models with double integral equations is that , for mixing in incompressible flows , they substantially overestimate the pdf tails at late times ( kolleman 1990 )
. however , the model offers an insightful picture for the mixing of pristine gas and provides useful fitting functions to the pristine fraction decay in certain physical regimes .
more recently , venaille and sommeria ( 2007 ) developed a continuous " version of the self - convolution model , based on an extension of the curl ( 1963 ) model in laplace space .
we first define the laplace transform , @xmath89 , of the concentration pdf as @xmath90 . using the convolution theorem , the laplace transform of eq .
( [ curl ] ) reads , @xmath91 .
\label{convolution}\ ] ] rewriting eq .
( [ convolution ] ) in a difference form , we have @xmath92 where @xmath93 with @xmath94 an infinitesimal time step . the difference equation has the following interpretation : during a timestep @xmath94 , mixing occurs in an infinitesimal fraction , @xmath95 , of the flow , and in this fraction of the flow the scalar pdf undergoes a convolution .
this suggests that in curl s model the pdf convolution occurs locally in space .
also note that , whenever a mixing event occurs , it appears as a single complete convolution in the model , and in this sense the convolution process is discrete " .
the continuous convolution model essentially assumes that the convolution occurs everywhere in the flow at any given time , but in an infinitesimal time the number of convolutions is infinitesimal and equal to @xmath95 ( duplat and villermaux 2008 ) .
the assumption can be represented by @xmath96 .
the taylor expansion of this equation gives @xmath97 $ ] .
taking the limit @xmath98 , we obtain , @xmath99 . \label{cconvolution}\ ] ] the equation was first derived by venaille and sommeria ( 2007 ) , who showed that the predicted pdf evolves toward gaussian in the long time limit . in the continuous version ,
the pdf convolution occurs globally in space .
the model prediction has been tested against experimental results by venaille and sommeria ( 2008 ) .
similar to curl s model , the continuous model can not be applied to predict the evolution of the entire pdf right at the beginning if the initial pdf is a double - delta function ( venaille and sommeria 2007 ) .
fortunately , for the problem of pristine gas pollution , the model provides a useful prediction that works immediately from the initial time ( pss ) .
a more general extension of the self - convolution model in laplace space was given in duplat and villermaux ( 2008 ) , @xmath100 .
\label{nconvolution}\ ] ] with @xmath15 and in the limit @xmath101 , the equation becomes eq .
( [ convolution ] ) for curl s original model and eq .
( [ cconvolution ] ) for the model of venaille and sommeria ( 2007 ) , respectively . in deriving eq .
( [ nconvolution ] ) , it was assumed that a fraction , @xmath102 , of the flow experiences mixing / convolution events during a time interval @xmath94 , and the number of convolutions in this fraction of the flow is @xmath103 . from the discussion above for curl s model and its continuous version
, @xmath7 characterizes the degree of spatial locality of the pdf convolution .
larger values of @xmath7 correspond to more
global " convolutions in the spatial space , and the parameter @xmath7 may be a function of time in general .
( [ nconvolution ] ) was referred to as the generalized convolution model in pss , where we found that with increasing @xmath7 the tails of the predicted pdfs become narrower .
for example , the discrete model with @xmath15 predicts exceedingly fat pdf tails , while in the continuous model ( @xmath101 ) the pdf approaches gaussian at late times . in other words , more
global " pdf convolutions produce narrower pdf tails . finally , we point out that self - convolution models were not originally intended for mixing in highly compressible flows and they do not directly account for how compressible modes and the density fluctuations in supersonic turbulence may affect the concentration pdf . the diffusivity term in the pdf equation ( see eq .
( [ simplepdfeq ] ) ) has a dependence on the density field , suggesting that the flow compressibility may have potentially important effects on the pdf evolution .
to our knowledge , the effect of compressibility has not been investigated in existing pdf models for turbulent mixing . here
we take the following approach : we compare the predictions of the convolution models for the primordial fraction against simulation results , and examine whether , by adjusting their parameters , they can be applied to study the pristine gas pollution in supersonic turbulence . indeed , we find that , by varying the parameter @xmath7 , the self - convolution models give satisfactory predictions for the the pollution of pristine gas in turbulent flows at different degrees of compressibility . nevertheless , new closure models are strongly motivated to directly and explicitly address the effects of shocks and flow compressibility on the scalar pdf in supersonic turbulence
. the pristine fraction , defined as the mass fraction of the interstellar gas with metallicity smaller than the critical value , @xmath4 , can be evaluated from the concentration pdf by @xmath104 .
the fraction can be calculated easily if the pdf evolution is known .
the threshold metallicity , @xmath4 , for the transition to pop ii star formation is small but finite , in the range from @xmath105 to @xmath106 by mass ( see bromm & yoshida 2011 , schneider et al .
2012 and references therein ) .
we also consider the fraction , @xmath107 , in the limit of an infinitesimal threshold , i.e. , @xmath108 , which corresponds to the mass fraction of exactly metal - free gas . clearly , the fraction @xmath107 is zero unless the concentration pdf , @xmath109 , has a delta function component at @xmath64 .
equations of @xmath107 can be exactly derived from the self - convolution models in 3.1 .
there is a subtle issue about the decay of the exactly metal - free fraction , @xmath107 , and the pristine fraction , @xmath110 , with a finite threshold .
pss pointed out that the nonlocal nature of the laplacian operator in the molecular diffusivity term leads to an essentially instantaneous decrease of @xmath107 .
physically , a tiny but finite fraction of the pollutant atoms can have extremely fast thermal speed , corresponding to the high tail of the maxwellian distribution , and may reach and pollute the pristine gas at large distances in a short time . even though the degree of pollution by these atoms at large distances is negligibly tiny , they do reduce the mass of gas that is _ exactly _ metal - free , and this occurs at a timescale much shorter than the sound crossing time . therefore , with the molecular diffusivity alone
, @xmath107 would decrease to zero almost instantaneously , regardless of the amplitude of the molecular diffusivity @xmath33 . on the other hand
, it takes a finite time for the molecular diffusivity to enrich the entire flow up to a finite threshold @xmath4 .
in fact , the decay of @xmath25 with @xmath4 , say , @xmath111 by molecular diffusivity alone is very slow because @xmath33 is typically tiny , @xmath112 @xmath58 s@xmath59 in galactic neutral ism ( see pan and scalo 2007 ) .
an efficient mixing rate relies on the presence of turbulent motions . an ideal model for the pollution of pristine gas should accurately capture both the rapid decay of @xmath107 and the evolution behavior of @xmath25
however , none of the models considered in pss satisfy both constraints .
for example , the mapping closure model by chen et al .
( 1989 ) does predict an instantaneous decay of @xmath107 , but a comparison with simulation results shows that its prediction for @xmath25 is poor in general , especially in highly supersonic flows . on the other hand
, the convolution pdf models introduced in 3.1 do not reduce @xmath107 to zero immediately , instead the delta function component at @xmath64 remains finite at any finite time .
this is inconsistent with the expectation of an instantaneous reduction of @xmath107 , and the reason is that the laplacian operator in the molecular diffusivity term was not directly incorporated in these models . despite this inconsistency
, pss found a very interesting result : the evolution equations of @xmath107 derived from the convolution models provide excellent fitting functions for the simulation results for the decay of the pristine fraction @xmath25 with a small but finite threshold .
here we take the same approach as pss and carry out a more systematic parameter study required to accurately span the range of astrophysical environments of interest .
we will use the @xmath107 equations from the convolution models to fit the simulation results for @xmath110 with different thresholds , @xmath4 , for scalars with different initial conditions evolving in a number of turbulent flows .
this systematic procedure gives best - fit parameters in the convolution models as functions of @xmath4 , the initial pollutant conditions and the flow mach number .
the numerically - tested @xmath107 equations with the best - fit parameters then provide a new tool to model the pollution of the primordial gas and the transition from pop iii to pop ii star formation in early galaxies .
we derive the equations of @xmath107 from the convolution models using the pdf equations in laplace space .
since the delta function at @xmath64 persists in these models , we decompose the concentration pdf into two terms , @xmath113 where @xmath107 is the fraction of exactly metal - free gas , and@xmath114 is the concentration pdf in the enriched part of the flow , which satisfies the condition @xmath115 .
the laplace transform of eq .
( [ decomp ] ) gives , @xmath116 where @xmath117 is the laplace transform of @xmath118 . from the condition @xmath119
, we have @xmath120 in the limit @xmath121 . inserting eq .
( [ lapdecomp ] ) to the pdf equation ( [ nconvolution ] ) for the generalized convolution model and taking the limit @xmath121 yields : @xmath122 for later convenience , we have replaced the turbulent stretching rate , @xmath123 , by a
convolution " timescale @xmath124 . setting @xmath15 in eq .
( [ pfnconv ] ) , we obtain the equation of @xmath107 for curl s model , @xmath125 which was first given in pan & scalo ( 2007 ) . an alternative derivation of this equation from the pdf equation in the double integral form is presented in pss . from eq .
( [ pfcurl ] ) , we see an interesting and simple physical picture for the pollution of the pristine gas by turbulent mixing : the primordial fraction is reduced when the fluid elements that are exactly metal - free and the rest of the flow that has been polluted by sources or previous mixing events , are brought close enough by turbulent stretching for the molecular diffusivity to homogenize . taking @xmath101 , eq .
( [ pfnconv ] ) becomes , @xmath126 which is the prediction of the continuous convolution model of venaille and sommeria ( 2007 ) for the pristine faction evolution .
assuming both @xmath7 and @xmath8 are constant with time , equation ( [ pfnconv ] ) has an analytic solution , @xmath127^n } , \label{pfnconvsolution}\ ] ] where @xmath68 is the initial pristine fraction .
this equation becomes @xmath128 for curl s discrete " model with @xmath15 and @xmath129 for the venaille and sommeria ( 2007 ) model with @xmath101 .
these convolution models predict that the pollution of primordial gas in turbulent flows proceeds at a timescale @xmath130 , which is essentially the timescale of turbulent stretching at large scales , anticipated by the cascade picture for turbulent mixing .
also note that the pollution timescale is essentially independent of the molecular diffusivity @xmath33 .
again , this is because the mixing rate is largely controlled by how fast the velocity field produces and feeds fine structures to the molecular diffusivity , but insensitive to the diffusion scale at which molecular diffusivity operates .
to calibrate @xmath7 and @xmath8 as function of the flow and pollutant properties , we carried out numerical simulations for mixing in compressible turbulence using the flash code ( version 3.2 ) , a multidimensional hydrodynamic code ( fryxell et al .
2000 ) that solves the riemann problem on a cartesian grid using a directionally - split piecewise - parabolic method ( colella & woodward 1984 ; colella & glaz 1985 ; fryxell , m " uller , & arnett 1989 ) .
the hydrodynamic equations were evolved in a periodic box of unit size with 512@xmath131 grid points .
simulation runs at a lower resolution ( 256@xmath131 ) were also conducted to check the potential effect of numerical diffusion .
an isothermal equation of state with unit sound speed was adopted in all our simulations .
the turbulent flows were driven and maintained at a steady state by a large - scale solenoidal external force , which was set be a gaussian stochastic vector that decorrelates exponentially with a timescale equal to a quarter of the sound crossing time .
the driving force was generated in fourier space , and it included all independent modes with wave numbers in the range from @xmath132 and @xmath133 .
each mode was given the same amount of power .
we defined a characteristic driving length scale @xmath134 , where @xmath135 is the power spectrum of the driving force . calculating @xmath136 from our forcing spectrum
, we found @xmath137 in units in which the box size is unity . by adjusting the amplitude of the driving force
, we simulated four flows with different ( density - weighted ) rms velocities , @xmath138 .
for each flow , we defined a dynamical timescale , @xmath139 , and all the simulation runs lasted for about 5 @xmath140 .
we computed the mean rms velocity by a temporal average after each flow reached a steady state , and the rms mach numbers , @xmath11 , i.e. , the ratio of the rms velocity to the sound speed , in the four flows were @xmath141 , 2.1 , 3.6 , and 6.2 , respectively .
the simulation setup for the turbulent velocity field is the same as in ps10 and pss , to which we refer the interested reader for details .
to study turbulent mixing , we evolved a number of decaying scalar fields in the four simulated flows . in each flow , we solved the advection equations of all scalar fields starting at the same time after the flow had already become fully developed and statistically stationary .
the initial condition of the scalar fields was taken to be bimodal , consisting of pure pollutants and completely unpolluted material only . such a bimodal field
was obtained by setting the pollutant concentration , @xmath142 , to unity in selected regions , representing pure pollutants , and to zero in the rest of the simulation box , corresponding to the unpolluted flow .
the rate at which the pollution of the pristine material proceeds in our simulations depends not only on the flow properties but also on the initial configuration of the pollutants .
two parameters in the initial condition are of particular interest .
the first one is the initial pollutant fraction , @xmath69 , defined as the ratio of the heavy element mass to the total flow mass in the simulation box .
the fraction is related to the initial primordial fraction , @xmath68 , by @xmath143 . obviously , with more pollutants in the flow ,
i.e. , a larger value of @xmath69 , one would expect a faster pollution of the pristine gas .
the mixing / pollution timescale also depends on how the pollutants are spatially distributed in the flow . for illustration ,
let us consider two different distribution patterns for the same amount of pollutants .
in the first pattern , the pollutants are released in the form of a single blob , while in the second the pollutants are divided into many blobs of similar sizes evenly distributed in the flow .
intuitively , the pollution process would be considerably faster in the latter case . in that case , the pollution injection scale , @xmath13 , which is essentially the average distance between the pollutant locations , is smaller , and the mixing timescale should be shorter since it is determined by the eddy turnover time at @xmath13 ( ps10 ) .
we thus expect that a smaller @xmath13 would result in a faster decay of the pristine mass fraction , and we will quantitatively examine the dependence of the pollution of the pristine flow on this parameter . in the context of the mixing of heavy elements in the interstellar media of early galaxies , the pollutant fraction , @xmath69 , is related to the number or the rate of the supernova events , and the pollutant injection scale corresponds to the average distance between the explosion locations . in order to conduct a systematic study of the parameter dependence of @xmath5
, we included in each simulated flow a total of 20 scalar fields with different initial conditions .
table 1 summarizes the fields , which are divided into 5 categories based on the geometry and the spatial distribution of the pollutants . for categories 1 and 2
, the initial pollutant configuration is a single blob located right at the center of the simulation box , and the geometrical shape of the blob was set to be a cube and a spherical ball , respectively .
clearly , for a single pollutant blob , the pollutant separation and hence the injection length scale , @xmath13 , is given by the box size .
considering that the flow driving scale @xmath144 in our simulations is 0.46 of the box size , we have @xmath145 for categories 1 and 2 . in the other three categories ,
the pollutants are divided into identical spherical blobs , equally spaced in the simulation box .
the number of blobs is 8 , 64 and 512 , respectively , for category 3 , 4 , and 5 . for scalar fields in these three categories
, @xmath13 corresponds to @xmath146 , @xmath147 and @xmath148 box size ( or equivalently @xmath149 and @xmath150 ) , respectively . for reference ,
the scale at which the energy cascade and hence the inertial range starts in our simulated flows is about @xmath147 box size .
the injection scale ( @xmath148 box size ) of category 5 scalar fields is well within the inertial range .
there are four scalar fields in each category , which differ in the initial pollutant mass fraction , @xmath69 .
the four scalar fields in category 1 , named 1a , 1b , 1c , and 1d , have @xmath151 , 0.1 , 0.01 , and 0.001 , respectively .
scalar fields in other categories are named in the same way .
these exact @xmath69 values were achieved by tuning the size of the pollutant blob(s ) . in category 1 ,
the length of the pollutant cube is set to 0.79 , 0.47 , 0.22 and 0.1 in units of the box size , or 1.7 , 1.0 , 0.48 and 0.22 in units of the driving scale @xmath144 , for scalars a , b , c and d , respectively , in the mach 0.9 flow .
for the four scalars in category 2 , the radius , @xmath152 , of the spherical ball is 0.49 , 0.29 , 0.14 , and 0.063 box size . in units of @xmath144 , @xmath153 , 0.63 , 0.30 , and 0.14 @xmath144 , respectively .
the radius of each pollutant ball for the corresponding scalar in categories 3 , 4 , and 5 is smaller by a factor of 2 , 4 , and 8 , respectively , than the @xmath152 value for category 2 .
this is because the numbers of balls in those categories are larger than that in category 2 by factors of @xmath154 , @xmath155 and @xmath156 .
the radii @xmath152 given here are the values used in the @xmath141 flow . at larger @xmath11 ,
@xmath152 for each corresponding scalar is slightly different . due to significant density fluctuations in flows at higher @xmath11
, using pollutant blobs of the same size at the same locations leads to different values of @xmath69 .
we thus tuned the pollutant size to guarantee the initial pollutant _
mass _ fraction , @xmath69 , is exactly 0.5 , 0.1 , 0.01 , and @xmath157 for the scalars in each flow .
we also made an attempt to investigate a smaller value ( @xmath158 ) of @xmath69 , which would also be interesting for mixing in the interstellar medium of early galaxies .
however , a tiny @xmath69 corresponds to a small pollutant size , and due to the limited numerical resolution , the pollutant size for @xmath159 is too close to the resolution scale of our simulations . in that case ,
numerical diffusion took effect and significantly polluted the surrounding flow from the beginning , leading to a different evolution behavior for the pristine fraction at early times than the other cases with @xmath160 . in the interstellar medium ,
the pollutant size is essentially a supernova remnant stall diameter , @xmath161 pc , with little dependence on parameters ( see thornton et al . 1998 ,
hanayama & tomisaka 2006 ) .
this is expected to lie within the inertial range of interstellar turbulence .
therefore , the real homogenization of fresh metals from supernovae by molecular diffusivity must wait for turbulent stretching to bring the concentration structures to the diffusion scale , which is tiny in comparison to the remnant size .
it is thus appropriate to consider pollutants with initial sizes significantly larger than the diffusion / resolution scale of the turbulent flow , and in the present work we do not explore scalar cases with @xmath162 .
we point out that the first three scalar fields in category 1 , i.e. , 1a , 1b , and 1c , in mach 0.9 and 6.2 flows have been studied in details in pss . in this paper , we perform a more systematic study covering a much larger parameter space . neither the viscous term in the hydrodynamic equations nor the diffusivity term in the advection - diffusion equation were explicitly included in our code .
therefore , both kinetic energy dissipation and scalar homogenization ( or dissipation ) are through numerical diffusion in our simulations .
the diffusion scale where the scalar homogenization occurs is close to the resolution scale , and so is the energy dissipation or the kolmorgorov scale . to examine whether our results for primordial gas mixing depend on the amplitude of numerical diffusion
, we also performed simulations at the resolution of @xmath163 , and the results at the two resolutions are compared in 5.4.5 .
otherwise , unless explicitly stated , the results reported below are from the @xmath164 simulations .
in fig . 1 , we show the concentration on a slice ( on the @xmath165-@xmath166 plane at @xmath167 ) of the simulation grid for three scalar fields ( rows ) at three different times ( columns ) .
the concentration is plotted on a logarithmic scale with the white color representing concentration levels below @xmath105 .
the top three panels , from left to right , correspond to case 3b in the mach 0.9 flow at @xmath168 , and @xmath169 , respectively . at @xmath170 ,
four spherical pollutant blobs lie on the @xmath171 plane . with time , turbulence stretches and
spreads out the pollutants , and structures at small scales are continuously produced .
in particular , we observe prominent cliff " structures with sharp concentration gradients .
these sheet - like structures are typical of passive scalars in incompressible turbulence ( e.g. , watanabe and gotoh 2004 ) . as the length scale of the scalar structures reaches the ( numerical ) diffusion scale , mixing occurs between the pollutants / polluted flow and the pristine regions .
the mixing process reduces the volume fraction of the pristine flow ( white regions ) , and at @xmath172 almost the entire flow is polluted .
the density - weighted concentration variance decreases from the initial value of 0.09 to 0.086 , 0.038 , and 0.015 , respectively , for the three snapshots from the left to the right .
the three panels in the second row show snapshots of the same scalar case ( 3b ) , but in the mach 6.2 flow , at @xmath173 ( left ) , 0.76 ( mid ) , and 1.41 @xmath140 ( right ) , respectively . comparing with the top panels , we see that , even at later times ( in units of @xmath140 ) , the surviving pristine volume is larger than in the @xmath141 flow , suggesting that the pollution of the pristine gas is slower in turbulent flows at higher @xmath11 .
the concentration field appears to be smoother than in the @xmath141 flow . as explained in detail in ps10 , this is because in highly supersonic turbulence the visual impression of the scalar field is dominated by expansion events , which occupy most volume of the flow domain .
since a passive scalar simply follows the flow velocity , an expanding region tends to produce coherent and smooth structures at large scales .
we note that the scalar in the supersonic flow has a smoother appearance also at small scales , and the likely reason is that the code used in our simulations applies a larger effective numerical diffusion to stabilize stronger stocks in flows with larger @xmath11 .
although compressible modes play a key role in shaping the large - scale geometry of the scalar field , the primary mixing agent is still stretching by solenoidal modes even in our simulated flows at very high @xmath11 ( ps10 ) .
the concentration variances at the three snapshots shown here are 0.078 , 0.023 , and 0.006 , respectively .
note that , even though the scalar variances in the right two panels are smaller than the corresponding snapshots in the @xmath141 flow , the remaining pristine fraction in the @xmath174 flow appears to be larger .
this is because in turbulent flows with higher @xmath11 the scalar pdf tails are broader , leading to a larger pristine fraction at the same concentration variance .
a more detailed discussion on this issue is given in 5.2 .
the bottom three panels in fig .
1 plot the evolution of the scalar field 5b in the mach 6.2 flow , which also has @xmath175 . unlike case 3b shown in the top and central panels , this field initially consists of 512 small blobs , and has a smaller injection scale , @xmath13 . at @xmath170 ,
64 blobs lie on the slice shown here . at early times
, some blobs appear to be small dots or filaments because they are being advected out of the selected slice .
the three panels correspond to @xmath176 , 0.33 , @xmath177 @xmath140 .
the mixing / pollution process proceeds much faster than case 3b in the same flow .
one reason is that , for a smaller pollutant size , turbulent stretching of the pollutant is faster , and thus mixing of each individual blob with the surrounding flow is more efficient .
also , since the separation between the pollutant blobs is small , the polluted / mixed regions by the individual pollutant blobs start to overlap quickly , resulting in a much faster erasure of the pristine flow material . as a reference ,
the scalar variances at the three snapshots from the left to the right are 0.07 , 0.038 , and 0.011 , respectively . in this subsection
, we discuss simulation results for the evolution of the concentration pdf .
( [ pdf ] ) plots the pdf as a function of time for four scalar fields .
for all scalars , the heights of the two spikes at @xmath64 and @xmath63 decrease at early times , and mixing causes a probability flux toward the central part , which gradually fills the concentration space between the two spikes .
both spikes are eventually removed , and for an initial pdf with negative skewness , or @xmath178 , the left spike lasts longer than the right one . at later times , a central peak forms around the mean concentration , and the pdf becomes unimodal . after that
, the pdf continuously narrows toward the mean value , a process described in 2.2 as anti - diffusion in concentration space . in pss
, we tested the predictions of various models for the pdf evolution against simulation data for scalar case 1b in mach 0.9 and 6.2 flows .
the initial condition of this scalar is a single cubic pollutant with @xmath179 .
it was found to be very challenging for pdf models to accurately predict the scalar pdf tails , especially for scalar fields in highly supersonic flows .
here we do not attempt to obtain successful model fits to the measured pdfs , as the main goal of this work is to understand the evolution of the pristine fraction , rather than the full details of the entire pdf .
however , including a model prediction for the pdf evolution in our figure is useful , because it provides a guideline to compare the fatness of the pdf tails for different scalar fields in different flows .
for this purpose , we consider the beta distribution function as a pdf model for passive scalar mixing , which has been shown to provide a good approximation for the pdf shape of decaying scalars with a double - delta initial condition in incompressible turbulence ( e.g. , girimaji 1991 ) .
the beta distribution function is defined as , @xmath180 where @xmath181 is the gamma function . to compare the beta distribution with the simulation results
, one can determine the two parameters , @xmath182 and @xmath183 , in eq .
( [ beta ] ) by equating the mean and variance of the beta pdf to those measured from the simulation data . for each measured pdf ( data points ) shown in fig .
( [ pdf ] ) , we plot a beta distribution ( line ) , where the beta parameters are fixed using the concentration mean and variance at the corresponding time . the left top panel in fig .
( [ pdf ] ) shows the result for case 3b in the mach 0.9 flow .
the initial condition of this case is eight equally - spaced spherical blobs with @xmath13 equal to 1/2 box size .
the total pollutant fraction , @xmath69 , of this scalar field is 0.1 .
the pdfs are measured at five different times as indicated in the legend . for this scalar ,
the fitting quality of the beta distribution functions is generally good except at far tails . in pss
, we showed that , at later evolution times , the pdf of scalar 1b in the @xmath141 flow is well fit by a gamma distribution , as predicted by the pdf model of villermaux and duplat ( 2003 ) .
the initial condition of scalar 1b is a single cubic pollutant , and it has a twice larger @xmath13 than 3b shown here .
the performance of the villermaux and duplat ( 2003 ) model is less satisfactory for scalars with smaller @xmath13 , e.g. , it significantly underestimates the pdf tails for the scalars in fig .
( [ pdf ] ) .
also that model is is invalid at the early evolution stage ( see pss ) . on the other hand ,
the beta distribution does provide acceptable fits to the measured pdfs at early times , as seen in fig .
( [ pdf ] ) .
the top right panel shows the pdf of the same scalar field in the mach 6.2 flow . using the beta distributions as a reference
, we see that at late times , the left pdf tails are broader than in the mach 0.9 case . the bottom two panels plot the results for case 5b in the same two flows .
this scalar field also has @xmath184 , but the injection scale , @xmath13 , is significantly smaller , @xmath185 box size ( see table 1 ) .
a comparison of the time series indicated by the legends in the top and bottom panels shows that the pdf variance decays much faster for scalar fields with smaller injection scale ( see 5.3 ) , but at similar values of the variance , the pdf tails are broader for scalar fields with smaller @xmath13 .
these observations are consistent with the findings of ps10 , who studied the dependence of the pdf shape on @xmath11 and @xmath13 in details , and found that the pdf tails become broader with increasing @xmath11 or decreasing @xmath13 . the physical origin of this behavior is probably related to the phenomenon of turbulent intermittency , i.e. , the existence of strong non - gaussian velocity structures at small scales .
supporting this interpretation is the fact that the degree of non - gaussianity of the velocity field increases as @xmath11 increases or as @xmath13 decreases , which coincides with the trend of the pdf tails of passive scalars .
as discussed in 3.1 , the convolution pdf models with smaller @xmath7 predict fatter tails , meaning that @xmath7 would decrease with increasing @xmath11 or decreasing @xmath13 if one attempts to fit the measured pdfs with the predictions of the convolution models . extending the intermittency argument here to mixing in supersonic flows with totally compressive driving
, we expect that , at the same @xmath11 , the passive scalar pdf would have fatter tails than in our flows with solenoidal driving .
this is because the compressively driven supersonic flows are significantly more intermittent ( federrath et al .
2010 ) .
we point out that the fatness of the pdf tails as a function of @xmath11 and @xmath13 for decaying scalars in the current study is less clear - cut than for the forced scalars examined in ps10 .
the general trend is sometimes not clearly obeyed in our simulations here , especially for scalar fields in high mach - number flows ( @xmath186 and 6.2 ) .
for example , as seen in fig .
( [ pdf ] ) , it appears that the pdf tail of scalar 5b in the mach 6.2 flow is less broad than in the mach 0.9 flow .
a possible reason is that the measurement of the pdfs of decaying scalars is less precise than in the case of forced scalars , for which the pdfs can be computed by averaging over many snapshots .
simulations with higher resolutions may help us establish a robust trend for the pdf tail of decaying scalars in turbulent flows at large @xmath11 , as they provide better statistics and better resolution of complexities , such as strong density fluctuations , in highly supersonic turbulence . finally , we stress that the pristine fraction corresponds to the probability contained in the far left tail of the pdf with @xmath187 , which is beyond the range of @xmath26 values shown in fig .
( [ pdf ] ) . nevertheless , the pdf tails shown in fig .
( [ pdf ] ) can be used to infer the trends of the pristine fraction with varying @xmath11 and @xmath188 since the pdf tail broadens with @xmath11 , we would expect that , with the same concentration variance , the pristine fraction contained in the pdf would be higher for a scalar field evolving in a flow with higher mach number or smaller pollutant injection scale .
in fact , the dependence of the pdf tails on @xmath11 and @xmath13 induces interesting effects on the pristine fraction as a function of time , which will be discussed in detail in 5.4 .
, 2.1 , 3.5 and 6.2 .
the scalar field has a total pollutant fraction , @xmath69 , of 0.1 and the injection length scale , @xmath13 , is @xmath189 box size .
mid panel : b scalar fields ( @xmath179 ) in the @xmath190 flow .
the five curves correspond to five categories in table 1 with different pollutant shape and injection length scales .
bottom panel : scalar fields from category 3 in the @xmath190 flow .
each curve is for a different value of the initial pollutant fraction , and the variance of each scalar field is normalized to its initial value . ]
( [ var ] ) plots the variance decay of a number of scalar fields .
the top panel shows the results for scalar fields 3b in the four simulated flows at different @xmath11 . for these scalar fields , @xmath191 , @xmath179 , and
the initial variance is 0.09 .
the variance decay first slows down with increasing mach number , then becomes slightly faster as @xmath11 increases from @xmath192 to 6 .
the same behavior has been found in ps10 , where a physical explanation was given ( see also 3 ) . in our simulated flows , compressible modes are inefficient in producing small - scale structures .
therefore , the mixing efficiency decreases as the fraction of kinetic energy contained in compressible modes at inertial - range scales increases . at @xmath82 ,
this fraction saturates at an equipartition value of 1/3 , and the mixing timescale becomes essentially constant .
the slightly faster mixing as @xmath11 increases to @xmath193 is because of the effect of strong compression in our simulated flow with @xmath190 . due to the limited numerical resolution
, the strongest compression events in this flow can directly squeeze the scalar structures to the diffusion scale and provide some contribution to the mixing efficiency . as discussed in 3
, the effect of compressible modes on mixing could be stronger in a highly supersonic flow with compressive driving , and the behavior of the normalized mixing timescale as a function of @xmath11 in compressively driven flows will be studied in a future work . in fig .
( [ var ] ) , we see that the variance decrease is approximately exponential . the mixing timescale @xmath78 is measured to be 0.45 , 0.48 , 0.58 , 0.57 @xmath140 for @xmath141 , 2.1 , 3.5 , and 6.2 , respectively .
the 20% increase in @xmath78 as @xmath11 goes from 0.9 to 6.2 is consistent with the results of ps10 . the middle panel of fig .
( [ var ] ) shows the variance of five b scalars in the @xmath174 flow .
each case is from one of the five categories listed in table 1 , and they all have @xmath179 .
the curves for scalar fields 1b and 2b are very close to each other .
the initial pollutant distributions of these two scalar fields are a single cube and spherical ball , respectively , and the similarity of their variance decay suggests that the mixing timescale is essentially independent of the geometrical shape of the pollutants . on the other hand , the mixing timescale decreases steadily as the average pollutant separation becomes smaller . the injection scale , @xmath13 , for scalar fields 3b , 4b , and 5b is @xmath146 , @xmath147 , and @xmath148 box size , respectively .
we attempted to measure @xmath78 by fitting the five curves with exponentials in the time interval from @xmath194 to @xmath195 , which is the time range of primary interest for the pristine gas pollution ( see 5.4 ) .
the measured values of @xmath78 are 0.72 , 0.71 , 0.57 , 0.48 , and 0.34 @xmath140 , respectively , for the five curves from top to bottom . the mixing timescale is determined by the eddy turnover time at the pollutant injection scale , and thus decreases with decreasing @xmath13 .
this physical picture also provides an explanation for the scale dependence of the mixing timescale found by de avillez and mac low ( 2002 ) in a suite of numerical simulations of mixing in supernova - driven interstellar turbulence .
in the bottom panel , we plot the variance decay of four scalar fields from category 3 in the mach 6.2 flow .
different curves correspond to different values of @xmath196 . unlike the top two panels , here we normalize the concentration variance to its initial value , @xmath197 which makes it easier to compare the variance decay timescale of different scalars . for a double - delta pdf ( eq .
( [ initialpdf ] ) ) , the initial variance @xmath198 is equal to @xmath199 . for @xmath200 ,
@xmath198 decreases with decreasing @xmath69 .
this suggests that , for scalar pdfs close to a double delta shape , the variance is not a good indicator of the pristine fraction , as the smaller variance for scalar fields with smaller @xmath69 in bottom panel of fig .
( [ var ] ) actually corresponds to a larger pristine fraction .
while a better indictor would be the variance normalized to the average concentration squared , which measures the rms of the fluctuations relative to the mean , the variance plot normalized to the initial value nevertheless provides useful information for the timescale of the mixing process . with decreasing @xmath69 ,
the radius of each individual pollutant blob becomes smaller , decreasing from about 0.5 box size ( @xmath201 ) to only 0.06 box size ( @xmath202 ) .
the top two curves for @xmath203 and 0.1 are close to each other , and the reason is that , for these two scalar fields , both the pollutant size and the pollutant separation are close to the flow driving scale , @xmath144 , and the scale ( 1/4 box size ) at which the inertial range of the flow starts .
the mixing timescales for these two scalar cases are thus given by the turnover time of large eddies of similar sizes .
the situation is different for the rest two scalar fields . for @xmath204 ,
the size of each individual blob is significantly smaller than @xmath144 .
it is also smaller than the average separation , @xmath13 ( @xmath205 ) , between the pollutant blobs . in this case
, the mixing process around each blob is not synchronized with that over the entire flow .
this divides the variance evolution into two phases .
the early phase occurs faster and is controlled by the turbulent stretching rate at smaller scales ( the pollutant size ) .
this explains the faster variance decay for smaller @xmath69 at early times .
after each blob is stretched , spread and mixed to a size close to the average pollutant separation , the mixing process starts to proceed at a single pace , and the timescale is determined by the turnover time of eddies of size @xmath13 . as seen in the bottom panel of fig .
( [ var ] ) , the variance decay is exponential for all cases at late times with essentially the same timescale ( @xmath206 ) .
the existence of two phases for scalar fields with small @xmath69 also leaves a signature in the evolution of the pristine gas fraction .
we now present results for the decay of the pristine mass fraction in our simulated flows . in pss
, we have shown results for three scalar fields , 1a , 1b and 1c from category 1 ( with @xmath203 , 0.1 and 0.01 , respectively ; see table 1 ) , evolved in two flows with @xmath141 and @xmath174 .
the pollutant injection scale @xmath13 of those fields was the box size , or about twice the flow driving scale @xmath144 .
sin this section , we consider scalar fields in category 3 in the @xmath141 and 6.2 flows as primary examples .
the injection scale of these fields is smaller , with @xmath191 . in the subsequent subsections
, we will discuss in details the dependence of the pristine fraction decay on various parameters .
( [ mach0.9four ] ) shows the mass fraction @xmath207 of the flow with concentration level below @xmath105 for scalar fields 3a , 3b , 3c , and 3d in the @xmath141 flow .
the data points are simulation results , and the lines are fitting functions based on the predictions of the self - convolution pdf models discussed in 3.2 .
the left and right panels are the same figure on linear - linear and linear - log scales , respectively .
the linear - linear scale shows the early evolution more clearly , while with a linear - log plot one can see the late - time behavior better .
the initial pollutant fraction , @xmath69 , of the four cases in this figure ranges from 0.5 ( 3a ) to @xmath24 ( 3d ) . as shown in pss , the prediction , eq .
( [ pfnconvsolution ] ) , of the self - convolution models can successfully fit the simulation results for scalar fields with @xmath208 .
the fitting lines in fig .
( [ mach0.9four ] ) for the two cases with @xmath203 and @xmath209 are the predictions of the convolution models with @xmath210 .
the initial pristine fraction @xmath68 in eq .
( [ pfnconvsolution ] ) is set to 0.5 and 0.9 , and the timescale @xmath8 is taken to be @xmath211 and @xmath212 , respectively .
both the linear - linear and the linear - log plots show that the model prediction matches the simulation data well , suggesting that the pollution process in turbulent flows may be adequately described as a self - convolution process .
if @xmath69 is smaller than @xmath213 , the evolution of the pristine fraction is more complicated , and one can not satisfactorily fit the entire evolution of @xmath25 with the convolution model , eq .
( [ pfnconvsolution ] ) , by properly choosing the parameters @xmath7 and @xmath8 . in this case
, the pollution process shows different behaviors at early and late evolution phases .
a two - phase behavior for scalar fields with small @xmath69 was actually seen earlier in the scalar variance decay ( see the bottom panel of fig .
( [ var ] ) for scalars 3c and 3d in the mach 6.2 flow ) . for these cases ,
only a small fraction of the flow material , near the pollutant blobs , experiences pdf convolution at early times , because the amount of pollutants available for mixing is limited .
this suggests that the convolution of the concentration pdf is local in space in the early phase , and , based on the physical discussion in 3.1 , the mixing process in this phase would be better described by a discrete " version of the convolution model ( with @xmath7=1 ) .
consistent with this picture , we find that the pristine fraction in the early phase is in good agreement with the prediction , eq .
( [ pfintegralsolution ] ) , of the discrete " convolution model , or equivalently eq .
( [ pfnconvsolution ] ) with @xmath7=1 . with time , more and more flow is polluted , and the mixed flow material then acts as sources for further pollution .
the pdf convolution would thus becomes more global in spatial space and hence more continuous in laplace space , leading to an increase in @xmath7 .
as described in 3.1 , @xmath7 essentially corresponds to the degree of spatial locality for the pdf convolution .
recognizing the different mixing behaviors at early and late times , we attempted to apply a two - phase fitting procedure for scalar fields with @xmath214 ( see pss ) .
for a two - phase fit , we need to determine the transition time at which the two behaviors connect .
since the generalized convolution model with a single phase provides perfect fits to scalar fields with @xmath208 , one may expect that the second phase with a more global pdf convolution starts when the pristine fraction , @xmath25 , decreases to 0.9 .
we thus first tried to obtain a fitting function that connects the two phases at the time @xmath215 when @xmath216 .
the results are shown as dashed lines in fig .
( [ mach0.9four ] ) . in these lines ,
the early phases are fit by the discrete " model , eq .
( [ pfintegralsolution ] ) , with @xmath217 for both case 3c ( @xmath218 ) and case 3d ( @xmath219 ) .
once @xmath25 decreases to 0.9 , we use the generalized model prediction , @xmath220^n$ ] ( c.f .
( [ pfnconvsolution ] ) ) with @xmath210 .
the timescale @xmath221 for the late phase is set to @xmath222 and @xmath223 for case 3c and case 3d , respectively .
the fitting values adopted for @xmath7 and @xmath8 in the late phase are close to those used for the scalar fields with @xmath208 .
this means that , once the polluted fraction becomes larger than @xmath213 , the pristine fraction decays in a similar way as the @xmath224 fields .
the fitting quality of the dashed lines appears to be acceptable . to distinguish the two convolution timescales in the early and late phases ,
we denote them as @xmath225 and @xmath226 , respectively .
we will also use @xmath226 to denote the convolution timescale for scalar fields with @xmath208 because the decay of the pristine fraction for those fields is similar to the later - phase evolution of the @xmath227 cases .
we find that one can obtain better fits for scalar fields with @xmath227 by connecting the two phases at later times .
as shown in pss , for these fields the discrete " model well matches the simulation data in an extended time range until @xmath25 drops to 0.2 - 0.3 .
this allows us to connect the early and late behaviors at a time significantly larger than @xmath215 .
it turns out that the fitting quality is actually significantly improved if we start to use the generalized model with @xmath210 at times when @xmath25 is smaller than @xmath228 .
the solid lines in fig .
( [ mach0.9four ] ) for cases 3c and 3d show the fitting functions that connect the discrete " model and the later phase at @xmath229 when @xmath25 decreases to 0.5 . in the fitting curves ,
@xmath225 for the discrete " phase is set to @xmath230 for both case 3c and case 3d . starting from @xmath231 , we use the generalized model @xmath232^n$ ] with @xmath210 .
the timescale @xmath233 is set to @xmath234 and @xmath235 for case 3c and case 3d , respectively . from fig .
( [ mach0.9four ] ) , the two - phase fitting lines connecting at @xmath229 agree with the data considerably better than the dashed lines that connect at @xmath215 .
our choice here to connect the two phases at @xmath231 is somewhat arbitrary because there is an extended time range where both the discrete " model and the @xmath210 model can match the simulation data ( pss ) .
in fact , combining the two models at any time with @xmath236 would give fitting curves of similar quality .
the parameter @xmath7 adopted in both the dashed lines and the solid lines is 10 , i.e. , the same as used for the scalar fields with @xmath208 .
this is also the case for the convolution timescale @xmath226 in the second phase .
the values of @xmath226 used in the solid lines almost coincide with those adopted in the fitting lines for scalar fields with @xmath208 , and in the dashed lines the @xmath226 values are only slightly smaller ( by @xmath237 ) .
for the early phase , the adopted values for the timescale @xmath225 in the solid and dashed lines are very close too .
our result that connecting the two phases at @xmath229 yields better fits than at @xmath215 seems to suggest that , for scalar fields with @xmath238 , the pollution process does not make an immediate transition from the discrete " to the generalized convolution model with larger @xmath7 , when the pristine fraction decreases to @xmath239 .
the transition tends to occur later . considering that the generalized model with a single phase works perfectly for scalar fields with @xmath224
, this implies that the time at which the generalized convolution phase starts is not simply controlled by the value of the pristine or polluted fraction : it appears to have some dependence on whether the initial pristine fraction is larger or smaller than @xmath240 .
this does not cause any problems in a practical application if the exact value of the initial pollutant fraction , @xmath69 , is known .
one can use the generalized convolution model with a single phase if @xmath208 , or adopt a two - phase model connecting at , say , @xmath229 if @xmath241 .
however , there is some complication when applying this procedure to the subgrid model we will construct in 7 for large - eddy simulations for the pollution of primordial gas in early galaxies . for example , if at a given time the pristine fraction in a computational cell is , say , in between 0.9 and 0.5 , then the choice of using the discrete " model or the generalized model at that moment depends on whether the pristine fraction in that cell was larger or smaller than 0.9 when it was first polluted .
this would make the implementation of our subgrid model complicated , as it requires keeping some information on the pollution history in each cell .
we advocate simply using the generalized convolution model for any cells with a pristine fraction smaller than 0.9 , as it gives acceptable , if not perfect , fits to our simulation data for @xmath227 scalars at any time after @xmath242 . in the following subsections
, we will only consider fitting functions that connect at @xmath229 for scalar fields with @xmath227 , as they are in better agreement with the simulation data .
we will tabulate the fitting parameters obtained from such fits in 5.4.6 .
if in a particular application connecting the early and later phases at @xmath215 is preferred rather than at @xmath229 , our tabulated parameters would still be applicable , as the best - fit parameters used in the fitting curves that connect at @xmath215 and @xmath229 are very close . in fig .
( [ mach6.2four ] ) , we show the simulation results and the fitting curves for scalar fields from the same category ( 3 ) , but in the mach 6.2 flow . for the fields with @xmath203 and @xmath209 ,
the data points are fit by the convolution model , eq . ( [ pfnconvsolution ] ) , with @xmath243 .
the timescale @xmath226 for the two cases is set to @xmath244 and @xmath245 , respectively .
similar to the @xmath141 case , two - phase models connecting at @xmath215 ( dashed lines ) and @xmath229 ( solid lines ) are used for the rest two cases with @xmath227 . for the dashed lines ,
the early phase is fit by the discrete " convolution model with @xmath246 for scalar field 3c and @xmath247 for case 3d , and for the late phase we used the @xmath243 convolution model with @xmath248 and @xmath249 for the two cases , respectively . for the solid lines that connect at @xmath229 , the fitting parameters for the discrete "
phase are @xmath250 and @xmath251 for case 3c for case 3d , respectively , and the late evolution stage is fit with @xmath243 and @xmath252 for both cases .
again , the fitting quality is better with a connection at @xmath253 . in all cases , the fitting parameters ,
@xmath7 and @xmath226 , adopted for the scalar fields with @xmath208 and for the late phases of the @xmath214 fields are very close , suggesting a universal decay behavior of the pristine fraction once the polluted fraction exceeds @xmath254 .
we find that , for scalar fields with @xmath227 , it is more difficult to fit the early phases as @xmath69 decreases , and the fitting quality becomes poorer with decreasing @xmath69 ( see fig .
( [ mach6.2four ] and [ mach0.9four ] ) ) . the pollutant size is smaller for smaller values of @xmath255 and this may cause some complexities for the prediction of the pristine fraction .
for example , as @xmath69 decreases to 0.001 , the blob diameter is about the size of 30 computational cells , which is close to the scale where the flow inertial range ends .
the first effect is that , with time , the size of the polluted region around each pollutant blob increases , and the turbulent stretching timescale in the polluted regions may increase with time .
this is not accounted for in the convolution models since the convolution timescale is set to be constant .
another effect arises from the fact that the turbulent stretching rate has larger spatial variations at smaller scales .
the turbulent eddy seen " by a small pollutant blob may have a stretching rate different from the average value at the pollutant size .
the increase in the amplitude of the stretching rate fluctuations with decreasing length scale indicates the turbulent intensity around smaller blobs is more random " . in the early phase , the flow mass polluted by a single blob
is expected to increase exponentially with the stretching rate .
therefore , using an average stretching rate for all the pollutant blobs may not give a precise prediction .
the overall pollution rate depends on the turbulent stretching rates seen " by all the pollutants at early times .
the blobs encountering more intense eddies provide a larger contribution to the pollution process , and vice versa .
the effect is further amplified by the phenomenon of intermittency : the pdf of the stretching intensity exhibits fatter tails toward smaller scales .
therefore , the small blobs have large chance to encounter extreme stretching events .
clearly , the effect of intermittency makes it more difficult to predict the pristine fraction for scalar fields with smaller @xmath69 .
we point out that , for a given scalar field , the parameters , @xmath7 and @xmath226 , that can fit scalar fields with @xmath208 or the late phases of @xmath256 cases are not unique .
in fact , a ( small ) range of parameter pairs ( @xmath7 , @xmath226 ) can give acceptable fits to the simulation data .
for example , if a somewhat smaller ( larger ) @xmath7 is used , one could also have a similar fit with correspondingly smaller ( larger ) value of @xmath226 .
when obtaining the best - fit parameters , we attempted to select a single value of @xmath7 that provides good fits to all scalar fields in each category . with the chosen @xmath7
, we then determine the best - fit value of @xmath226 for each scalar field in the category .
as discussed above , the timescale turns out to be similar for all cases in a given category . a comparison of figs .
( [ mach0.9four ] ) and ( [ mach6.2four ] ) shows that , when the time is normalized to the flow dynamical timescale , the pristine fraction in the mach 6.2 flow survives for significantly longer than in the mach 0.9 case ( see also pss ) .
we discuss this mach number dependence in the following subsection . in fig .
( [ machdependence ] ) , we show the evolution of the pristine fraction for scalar fields 3b ( left ) and 3c ( right ) in our four simulated flows .
as observed earlier , with @xmath40 normalized to the flow dynamical time , @xmath140 , the decrease of the pristine fraction becomes slower with increasing mach number . in the fitting lines for case 3b ( left panel ) ,
the parameter pair ( @xmath7 , @xmath226 ) is set to ( 10 , 0.26@xmath140 ) , ( 6 , 0.29 @xmath140 ) , ( 5 , 0.31@xmath140 ) , and ( 3 , 0.31@xmath140 ) for the four flows with @xmath141 , 2.1 , 3.5 , and 6.2 , respectively .
again , we see that @xmath226 first increases with @xmath11 and then saturates for @xmath82 .
this trend is similar to that of the variance decay timescale , @xmath78 , as a function of @xmath11 ( see the top panel of fig .
( [ var ] ) ) . as explained in 5.3 , @xmath78 increases as the energy fraction in compressible modes increases and then becomes roughly constant when compressible energy fraction saturates at @xmath257 .
the same reasoning also applies here for the trend of @xmath226 with @xmath11 .
similar to the discussion in 5.3 on @xmath78 , the convolution timescale may have a different behavior with @xmath11 in compressively driven flows . in 5.2
, we showed that , at a similar concentration variance , the pdf tail becomes broader as @xmath11 increases , most likely because of a the increase in turbulent intermittency .
because the pristine fraction corresponds to the far left tail of the concentration pdf , the effect also slows the pristine gas pollution at larger @xmath11 .
another effect of the broadening of the pdf tail with increasing @xmath11 is that it changes the shape of the pristine fraction vs. time curve , as seen in fig .
( [ machdependence ] ) . to fit the pristine fraction in flows at different @xmath11 , we varied the parameter , @xmath7 , in the self - convolution model , which controls the shape of the fitting function , eq .
( [ pfnconvsolution ] ) .
the best - fit value of @xmath7 decreases from 10 to 3 as @xmath11 increases from 0.9 to 6.2 .
this decrease of @xmath7 is expected from the fact that the self - convolution model with smaller @xmath7 would predict broader pdf tails ( see 3.1 ) .
the convolution model was originally proposed for mixing in incompressible turbulence , where @xmath7 was a pure parameter without a clear connection to the mixing physics .
our finding that the convolution model with properly chosen @xmath7 can well describe the pristine fraction evolution in compressible turbulent flows motivates a physical interpretation of the parameter .
a possible intuitive reason for why @xmath7 decreases with the flow mach number is that @xmath7 reflects the degree of spatial locality of the pdf convolution , with more local mixing events implying a smaller @xmath7 . in a highly supersonic turbulence ,
the majority of the flow mass resides in a small fraction of volume , i.e. , in the dense postshock regions .
therefore , mixing of the pollutants into and within local regions of high densities is crucial toward the final homogenization .
the dense postshock regions are persistent with a lifetime on the order of the flow dynamical time , and the time scale for the homogenization between different postshock regions is expected to be on the same order .
this suggests that the presence of local dense regions may suppress the possibility of a global pdf convolution . in that case , as the flow mach number increases , the convolution would become more local , leading to a decrease in @xmath7 .
based on this argument , we speculate that , in compressively driven flows at similar @xmath11 , the parameter @xmath7 would be smaller than in our simulated flows .
the convolution in a compressively driven flow is expected to be more local because the density fluctuations are stronger ( e.g. , federrath et al .
2010 ) . the right panel of fig .
( [ machdependence ] ) plots the results for case 3 c with @xmath258 in the four simulated flows .
again the decrease of the pristine fraction is slower in flows with larger @xmath11 .
as discussed earlier , a two - phase fitting scenario is needed for scalar fields with @xmath227 . using the discrete convolution model to fit the early - phase evolution , we find that the timescale @xmath225 is 0.17 , 0.19 , 0.22 and 0.23 @xmath140 for @xmath141 , 2.1 , 3.5 and 6.2 , respectively .
the two phases are connected at @xmath229 .
note that the timescale @xmath225 also increases with @xmath11 at first and then saturates for @xmath82 . for the late phase of scalar case 3 c
, we adopted the same values of @xmath7 ( i.e. , 10 , 6 , 5 , and 3 ) for the flows as used in the case of 3 b. to match the simulation data , @xmath226 in the late phase is set to 0.25 , 0.30 , 0.34 and 0.34 @xmath140 for @xmath141 , 2.1 , 3.5 and 6.2 , respectively .
again , these numbers are close to the best - fit values for case 3 b shown in the left panel . in summary
, we found that the pollution of the pristine gas is slower in flows at higher @xmath11 .
two reasons are responsible for this behavior .
first , the mixing ( or variance decay ) timescale @xmath78 becomes larger as @xmath11 increases .
second , at the same concentration variance , the left pdf tail broadens with @xmath259 and this corresponds to a larger pristine fraction .
next , we study the dependence of the pristine fraction evolution on the initial spatial configuration of the pollutants , i.e. , on how the pollutants are released into the flow .
each category in table 1 represents a different pollutant shape or distribution at the initial time . in fig .
( [ lengthscale ] ) , we compare the simulation results for scalar fields from different categories in the mach 6.2 flow .
the left panel shows five b fields with @xmath184 , and the right panel is for c cases with @xmath258 .
the initial condition for the scalar fields in categories 1 and 2 is a single pollutant cube and a single spherical blob , respectively , and the pristine fraction evolution for scalar fields in these two categories is almost the same , suggesting the geometric shape of the pollutant blob does not affect the pollution rate . on the other hand
, the pollution process has a sensitive dependence on on the injection length scale , @xmath13 . for scalar fields in first two categories ( 1 and 2 ) ,
@xmath13 is about equal to the box size , or twice the flow driving scale , @xmath144 . for categories 3 , 4 and 5 , @xmath191 , @xmath260 and @xmath261 , respectively , as expected in 4 , the decay of the pristine fraction becomes progressively faster with decreasing @xmath13 .
the four lines in each panel of fig .
( [ lengthscale ] ) are fitting functions based on the self - convolution models . since the data points almost coincide for scalar fields in categories 1 and 2 , a single fitting curve ( the line on the right ) works for both cases .
the other three fitting lines , from the right to the left , correspond to scalar fields in categories 3 , 4 , and 5 respectively .
the fitting parameters used for the b fields in the left panel are @xmath262 and 1 , and @xmath263 and @xmath264 , respectively , for the four lines from the right to the left .
the timescale @xmath226 decreases by @xmath265 as @xmath13 changes from @xmath266 to @xmath144 , and , as@xmath13 decreases further below @xmath144 , the decrease of @xmath226 is faster , dropping by @xmath267 for each factor of 2 in @xmath13 .
this trend is similar to the dependence of the variance decay timescale , @xmath78 , on @xmath268 which is controlled by the eddy turnover time at @xmath269 and decreases with decreasing @xmath13 .
this also explains the faster pollution of the pristine gas if the pollutants are injected at smaller scales . recalling that @xmath78 was measured to be @xmath270 and 0.34 @xmath140 for the same b fields in the mach 6.2 flow with @xmath271 and 0.25@xmath144 , respectively ( see the mid panel of fig .
[ var ] ) , we see that @xmath226 has a more sensitive dependence on @xmath13 than @xmath78 .
a possible reason for this is that the exposure of the pollutants to the pristine flow may be an important factor for the pollution efficiency , and , with decreasing @xmath13 , the number of pollutant blobs increases rapidly , leading to enhanced pollutant exposure .
the trend that @xmath7 becomes smaller for scalars injected at smaller scales corresponds to broadening of the pdf tails with decreasing @xmath13 found in 5.2 .
the pdf tails become broader because the flow structures seen " by the scalars with smaller @xmath13 are more intermittent , and thus the dependence of @xmath7 on @xmath13 is related to the higher degree of turbulent intermittency at smaller scales . if the turbulent flow is driven compressively , the decrease of @xmath7 with decreasing @xmath13 may be faster due to stronger intermittency of the flow .
note that broadening of the pdf tails makes the pristine fraction larger , but this effect is minor in comparison to the faster decrease of the pristine fraction caused by the smaller mixing timescale at smaller @xmath13 .
the decrease of @xmath7 with decreasing @xmath13 may also be understood from a more intuitive argument . for scalar fields with a small @xmath13 , each pollutant is stretched by a local velocity structure , and the mixing of each pollutant blob with the surrounding flow proceeds largely independently at early times .
the pollution process is almost complete when the mixed areas by the pollutant blobs start to overlap ( see bottom panels of fig . 1 ) , meaning that the pdf convolution occurs locally and independently in different regions of size @xmath13 during most of the mixing process .
as the injection scale decreases , the pdf convolution becomes more local , leading to a smaller value of @xmath7 , which corresponds to a higher degree of spatial locality in the pdf convolution ( see 3.1 ) . for c fields shown in the right panel ,
a two - phase scenario connecting at @xmath229 is used to obtain the fitting lines . in the early phase ,
the timescale , @xmath225 , in the discrete convolution model is taken to be 0.30 , 0.24 , 0.17 , and 0.1 @xmath140 , respectively , for the four fitting lines from right to left .
the dependence of @xmath225 is similar to that of @xmath226 for the b cases .
it is first reduced by 20% , as @xmath13 decreases to @xmath144 , and then decreases faster , by @xmath272 , as @xmath13 decreases further by each factor of 2 .
for the late phase , we adopted the same values ( 5 , 3 , 2 and 1 ) of @xmath7 as for the corresponding b cases shown in the left panel , and @xmath226 is set to @xmath273 , 0.22 , and 0.12 @xmath140 for scalar fields with @xmath274 , 1 , 0.5 and @xmath150 , respectively .
again , these values of @xmath226 are close to those used in the fitting lines for the corresponding b cases .
it is interesting to note that , for case 5c , @xmath15 is adopted in both the early and late phases , although the timescales @xmath275 and @xmath276 are slightly different .
we also examined the @xmath13 dependence for all the other scalar fields including those in the other three flows .
we found similar trends for the parameters , @xmath7 , @xmath225 and @xmath226 , with varying pollutant injection scale .
the results are tabulated and further discussed in 5.4.6 .
when presenting simulation results in earlier subsections , we set the threshold metallicity to @xmath277 as a representative value , but , as discussed in the introduction , the threshold value for the transition to normal star formation is uncertain .
we therefore need to study the dependence of the pristine fraction @xmath25 on the @xmath4 . in fig .
( [ threshdependence ] ) , we plot @xmath25 at different threshold values for scalar 3c in the mach 6.2 flow .
the two panels show the same figure on linear - linear and linear - log scales , respectively .
we consider the scalar case c as an example , with which we can examine the @xmath4 dependence of both convolution timescales , @xmath225 and @xmath226 , for the early and late phases , respectively . the filled circles in fig .
( [ threshdependence ] ) correspond to the fraction of exactly pristine flow material with @xmath64 .
this fraction decreases to zero almost instantaneously .
this is caused by the effect of numerical diffusion . during each timestep
, any computational cell adjacent to one that contains pollutants or has been polluted by earlier mixing events will obtain a finite , but often extremely small concentration .
this means that the exactly - pristine flow material would be completely lost in a number of steps @xmath278 with @xmath13 and @xmath279 the average pollutant separation and the computation cell size , respectively .
the timestep in our simulation is approximately given by @xmath280 , where @xmath281 is the maximum flow velocity at a given time .
therefore , the survival time of exactly - pristine gas is @xmath282 , which is much smaller than the flow dynamical time @xmath283 because @xmath284
. the almost immediate removal of exactly metal - free gas by numerical diffusion is analogous to the expectation in 3.2 that the molecular diffusivity alone tends to reduce the exactly - pristine fraction @xmath107 to zero instantaneously ( see also pss ) , although the numerical diffusion in our simulation probably has a different form and amplitude than the realistic molecular diffusivity . the open symbols in fig .
( [ threshdependence ] ) show simulation data for finite , and more realistic , threshold values in the range from @xmath285 to @xmath106 . for @xmath4 in this range
, the simulation data for @xmath25 can be fit by the self - convolution models .
the fitting lines in fig .
( [ threshdependence ] ) are obtained using a two - phase scheme which combines the early and late behaviors at @xmath231 .
fitting the early phase with the discrete convolution model , we find that the dependence of the timescale @xmath225 on @xmath4 is very weak , with @xmath286 0.226 , 0.233 , 0.244 , 0.253 and 0.271 @xmath140 for the five threshold values increasing from @xmath285 to @xmath106 . if we express the dependence as a power - law , @xmath287 , the exponent , @xmath288 , would be very small , @xmath289 .
note that the increase seems to be faster ( by about @xmath290 ) , as the threshold increases from @xmath23 to @xmath106 . for the late phase
, we fixed the parameter @xmath7 at 3 ( the same as used before for this scalar ) , and adjusted the timescale @xmath226 to match the data points for different threshold values .
the best - fit value for @xmath226 was found to be 0.33 , 0.344 , 0.355 , 0.375 and 0.39 @xmath140 for @xmath291 , and @xmath106 , respectively . on average , @xmath226 increases by 4% as @xmath4 increases by each factor of 10 , and the dependence can be roughly written as @xmath292 with @xmath293 . a similar @xmath4 dependence of the convolution timescales was also found for other cases with different @xmath13 and in flows at different @xmath11 .
there is also a general trend that the increase of the convolution timescale with the threshold becomes faster as @xmath4 increases to the highest value , @xmath106 , considered in our study . the weak power - law dependence of the convolution timescales , @xmath225 and @xmath226 , on @xmath4 may extend to a range of threshold values below @xmath285 , although as @xmath294
, the numerical diffusion would finally be able to directly act to reduce @xmath25 and the scaling of the convolution timescale with @xmath4 given earlier would fail .
the @xmath294 limit may not be of practical interest , as the critical metallicity is likely to be higher than @xmath285 by mass . in the other limit , with increasing @xmath4
, the weak power - law scaling will also break down eventually . as pointed out above
, the increase of the convolution timescales is already faster as @xmath4 increases to @xmath106 .
in fact , if @xmath4 approaches the average concentration @xmath295 ( 0.01 for the scalar field shown in fig . [ threshdependence ] ) , eq .
( [ pfnconvsolution ] ) , which was derived in the limit @xmath296 will become invalid . for illustration , let us consider the case in which @xmath4 is exactly equal to @xmath297 . in this case , the fraction @xmath298 would not decrease to zero in the long time limit ; instead it would approach 1/2 .
a more extreme example is that , if @xmath4 is larger than @xmath297 , @xmath25 would first decrease when the pollutants mix with a small amount of the flow material , and then increase and finally approach unity when the flow is completely homogenized .
this situation may occur at very early times in the history of a galaxy , before heavy elements produced by supernova explosions increased the average metallicity to above the critical threshold .
however , there is the possibility metals from the explosion of a single massive pop iii star could make @xmath299 in a small high - redshift galaxy ( frebel et al .
even if the average metallicity in the entire interstellar medium is larger than @xmath4 , there may exist local regions where the average metallicity is smaller than the threshold .
one would need to deal with this situation in a subgrid mode for the large -scale simulation for the primordial gas pollution in an early galaxy ( see 7 ) . in this work , we do not examine the evolution of @xmath25 for @xmath4 close to or even larger than @xmath297 .
we defer it to a later work . in the case of @xmath300 ,
a good approximation is perhaps to assume that @xmath25 is a constant @xmath301 considering that @xmath25 would show qualitatively different behaviors as the ratio , @xmath302 , gets close to unity , it appears appropriate to take it as a function of @xmath302 , instead of the absolute value of @xmath4 .
another motivation is that , at a given ratio @xmath302 , @xmath25 samples a concentration range at a similar distance to the central part of the pdf .
thus , in 5.4.6 , we tabulate the fitting parameters for the evolution of @xmath25 with @xmath303 as functions of the flow mach number and the pollutant injection scale . for other values of @xmath302 , the timescales , @xmath225 and @xmath226
can be inferred using the weak power - law scaling given earlier , as long as @xmath304 . finally , we examined the effect of numerical resolution .
as discussed in 3.2 , the timescale for the pollution of the pristine gas to a significant concentration level is mainly determined by the rate at which turbulence stretches the pollutants , and independent of the amplitude of the molecular or numerical diffusion if it is sufficiently small to allow a scale separation between the pollutant injection scale and the diffusion scale . to verify this expectation
, we carried out @xmath163 simulations and compared them with the results from @xmath164 runs . the scale separation mentioned above exists at both resolutions , but
the separation is limited for the @xmath163 runs .
we drive the flows in exactly the same pattern in the @xmath163 and @xmath164 runs .
however , due to the chaotic " nature of turbulence , the developed turbulent velocity field at given locations in the two runs are different .
this means that , when released to the simulated flows at different resolutions , the pollutant blobs might encounter completely different velocity structures . in this sense , the comparison of our simulation results at two resolutions is somewhat different from the usual convergence check . in fig .
( [ resolutiondependence ] ) , we plot @xmath25 with @xmath305 for scalar fields in category 3 from the @xmath163 ( open symbols ) and @xmath164 ( filled symbols ) runs , with @xmath141 and @xmath306 the filled data points and the solid fitting curves for the @xmath164 runs were already presented in figs .
( [ mach0.9four ] ) and ( [ mach6.2four ] ) . here
the early and late phases of cases ( 3c ) and ( 3d ) are connected at @xmath229 .
the fitting curves for the @xmath163 data are obtained with the same fitting scenario as in the @xmath164 case . in both the mach 0.9 and 6.2 flows ,
the pristine fraction for the two scalar fields with @xmath201 ( 3a ) and @xmath307 ( 3b ) is smaller in the @xmath163 runs .
in fact , the fraction becomes smaller than in the @xmath164 runs almost immediately after the pollutants are released into the flows .
this can be explained by considering the action of numerical diffusion on the initial concentration field . since our initial concentration field
consists of pure pollutants ( @xmath308 ) and exactly pristine gas ( @xmath309 ) , there exist sharp edges between the pollutant blobs and the pristine flow .
numerical diffusion may operate on the large concentration gradient at the edges , and a fraction of the flow material surrounding the pollutants would be polluted immediately .
this results in an instantaneous drop in the pristine fraction . in the @xmath164 runs , the effect was found to be weak , and the initial drop was slight .
the drop is significantly larger in the @xmath163 simulations due to the larger numerical diffusion , leading to smaller pristine fractions for scalars 3a and 3b in the @xmath163 runs than in the @xmath164 cases . recognizing this effect of initial drop
, we adjusted the initial pristine fraction , @xmath68 , to smaller values when fitting the 256@xmath131 data .
for scalar fields 3a and 3b , we used the same values of @xmath7 ( i.e. , @xmath210 and 3 for @xmath141 and 6.2 , respectively ) as in the corresponding @xmath164 cases . with the adjusted values of @xmath68 , the best - fit timescales
@xmath226 for these two cases in the @xmath163 runs differ slightly , only by @xmath310 , from those used to fit the @xmath164 data .
this is the case for both mach 0.9 and mach 6.2 flows .
therefore , except for the initial drop , the numerical resolution does not affect how the pristine fraction evolves for the two fields with @xmath208 , and one may claim a numerical convergence of the convolution timescale .
note that , in realistic interstellar turbulence , the effect of the initial drop would be minimal because the molecular diffusivity is much smaller than the numerical diffusion in our simulations .
also the sharp pollutant - flow edges in the simulations are artificial and may not exist in reality .
the dependence on numerical resolution is more complicated for cases 3c and 3d with @xmath311 and @xmath312 respectively . as seen in fig .
( [ resolutiondependence ] ) , the pristine fraction decay in the @xmath163 runs can be either faster or slower than in the @xmath164 simulations .
the stronger initial drop in the @xmath163 runs still exists in the early evolution phases .
however , unlike scalar fields with @xmath313 , it is not the dominant effect .
the velocity field at a given location in the @xmath163 and @xmath164 simulations is different ( see above ) , and thus the same pollutant blob may encounter very different velocity structures in the two runs . as discussed earlier , for cases with small @xmath69 ,
the pollutant size is small , and the turbulent stretching rate around the blobs would show larger variations . therefore , the stretching rate in the eddy across a small pollutant blob may deviate significantly from the mean value at that scale . since the flow mass polluted by an individual blob scales nonlinearly with the local stretching rate , the overall pollution rate for scalar fields with tiny @xmath69 can not be predicted by an average stretching rate , instead it depends on the distribution of the stretching rates over all the pollutant blobs .
this is different from the case of @xmath224 fields with large pollutant sizes , where the amplitude of the stretching rate fluctuations is smaller and the stretching rate for each blob is similar and equal to the average value .
thus , the pollution for scalar cases with small @xmath69 ( @xmath314 ) is more sensitive to the details of the stretching rates encountered by all the blobs .
if the overall stretching rates in the eddies seen " by the pollutant blobs in the @xmath163 run is relatively higher , the pollution would proceed relatively faster than the @xmath164 run and vice versa .
it appears that the origin of the observed difference at late times is stochastic and has nothing to do with the numerical diffusion / resolution .
the above picture also suggests that the difference may become larger as @xmath69 decreases further below 0.001 . when fitting the early phases of cases 3c and 3d in the @xmath163 flows , we decreased @xmath68 to account for the initial drop . in the @xmath141 flow , @xmath225 for the early phases of these two cases are close to those used to fit the @xmath164 data , and the difference is at the level of @xmath315 .
the best - fit timescales @xmath226 for the late phases of the two scalars are larger ( by @xmath316 ) than for the @xmath164 data . in the @xmath174 flow ,
the fitting parameters for scalar 3c in the @xmath163 run are the same as in the corresponding 512@xmath131 case , once the initial drop is accounted for . for case 3d in the @xmath174 flow ,
the data points from the @xmath163 and @xmath164 runs almost coincide , and we do not give a separate fit to the @xmath163 data .
it appears that the resolution dependence of the best - fit parameters is quite weak . in summary , we find that the larger numerical diffusion in the @xmath163 simulations causes a larger initial drop in the pristine fraction .
this effect is successfully accounted by adjusting the value of @xmath68 in our model fits to the simulation results .
the effect is expected to be negligibly weak in the interstellar gas where the molecular diffusivity is tiny . for scalar fields with small @xmath227 ,
the timescales to fit the simulation data differ by @xmath237 at the two resolutions , and numerical convergence may be claimed .
the origin of the random " dependence of the pristine fraction on the resolution for these fields is related to the larger fluctuations of turbulent stretching rate at smaller scales and suggests that a precise prediction of the pristine fraction in the case of tiny @xmath69 may require the detailed eddy conditions at the initial pollutant locations .
we finally point out that numerical convergence would not exist at all if the resolution did not allow a separation between the pollution injection and the diffusion length scales .
we summarize our simulation results in tables 2 , 3 , and 4 .
the parameters listed in the tables are obtained by fitting the fraction @xmath317 from simulation data for scalar fields from different categories and in different flows . here
, for each scalar , the threshold @xmath4 is set to @xmath318 .
the average concentration @xmath297 is equal to the initial pollutant fraction @xmath69 for a double - delta initial condition , eq .
( [ initialpdf ] ) . for scalar fields a , b , c and d , @xmath319 and 0.001 , and we choose @xmath4 to be @xmath320 , @xmath105 , @xmath285 and @xmath321 , respectively . the choice of a fixed ratio @xmath302 is more convenient for practical applications . the timescales in tables 2 and 4 are slightly different from those used in the figures in previous subsections , where ( except in
5.4.4 ) the threshold was fixed at @xmath322 for all values of @xmath69 .
the numbers in these two tables are in units of the flow dynamical time , @xmath140 .
the first columns of tables 2 , 3 and 4 show results for scalar fields with the injection scale @xmath13 close to the box size or @xmath323 .
these parameters are measured from scalar cases in category 1 .
measuring the parameters using category 2 fields with the same @xmath13 would give essentially the same results .
table 2 lists the timescale , @xmath324 for the early phase of scalar fields with @xmath325 . in this phase ,
the pristine fraction evolution is fit by the discrete " convolution model with @xmath15 . for scalar cases in each category ( @xmath13 ) and each flow ( @xmath11 ) , we measured @xmath225 for the early phases of fields c ( with @xmath311 ) and d ( with @xmath219 ) , and the number given in table 2 is the average of the measured values for these two fields . as found in 5.4.2 , at a given injection scale , @xmath225 first increases as @xmath11 increases from 0.9 to 2 - 3 , and then saturates for larger @xmath11 .
the overall increase in @xmath225 is about 20% for @xmath11 in the range from 0.9 to 6.2 .
this is in general agreement with the trend of the mixing timescale @xmath78 with @xmath11 found in ps10 . at a given mach number , @xmath225 decreases with decreasing injection length scale @xmath13 .
as @xmath13 decreases from @xmath326 to @xmath144 , @xmath225 is smaller by @xmath327 .
the decrease is faster for smaller @xmath13 , and a further decrease of @xmath13 by each factor of 2 reduces @xmath225 by @xmath328 .
if we express the @xmath13 dependence of @xmath225 roughly as a power law for @xmath329 , we have @xmath330 . tables 3 and 4 give the parameters @xmath7 and @xmath226 as functions of @xmath11 and @xmath13 .
these are measured for the pristine fraction evolution of scalar fields with @xmath208 or the late - time behavior of scalars with smaller @xmath69 . for a given category ( @xmath13 ) and a given flow ( @xmath11 ) ,
we choose a single value of @xmath7 , with which the self - convolution model prediction can well match the simulation data simultaneously for both the two scalar cases with @xmath208 and the late phases of the other two cases with @xmath227 . in table 3 ,
the parameter @xmath7 is taken to be @xmath331 for scalar fields with @xmath145 in the mach 0.9 flow , which corresponds to the continuous convolution model ( eq . [ pfcconvsolution ] ) .
pss showed that the continuous model can be used to obtain successful fits to category 1 scalars in the @xmath141 flow .
we find that , for a given @xmath226 , the predicted pristine fraction by the convolution model barely changes with increasing @xmath7 once @xmath7 exceeds @xmath332 .
this means that replacing @xmath331 in table 3 by any number larger than @xmath333 would also work for category 1 ( or 2 ) fields in the mach 0.9 flow . from table 3
, we see that @xmath7 decreases with increasing mach number and decreasing @xmath13 .
this is due to the higher degree of flow intermittency at larger @xmath11 ( pan & scannapieco 2011 ) or smaller @xmath13 , which causes broader concentration pdf tails .
as described previously , a smaller @xmath7 indicates a more local pdf convolution . after fixing the parameter @xmath7 for each @xmath13 and @xmath11
, we measure the timescale , @xmath226 , for scalar cases a and b and the late phases of cases c and d. the measured values for the four cases are not exactly the same , but show slight variations .
the variations are stronger at larger @xmath11 or smaller @xmath13 .
we found that the amplitude of the variations is smaller when using a fixed @xmath302 ratio rater than a fixed threshold @xmath4 .
this also justifies taking the pristine fraction as a function of @xmath302 .
the numbers given in table 4 are the averages of the best - fit values for the four scalar cases in each category and each flow .
the dependence of @xmath226 on @xmath11 and @xmath13 is very similar to that of @xmath225 shown in table 2 .
again , it increases by about 20% as @xmath11 increases from @xmath334 to @xmath335 , and then stays constant at larger @xmath11 . like @xmath225 , the decrease of @xmath226 with decreasing @xmath13 also appears to be faster at smaller @xmath13 .
it is reduced by 25% , 35% , and 40% , respectively , as @xmath13 decreases by each factor of 2 from @xmath266 to @xmath261 .
roughly , @xmath226 scales with the injection scale as @xmath336 for @xmath329 .
we point out that , when measuring the model parameters from all the scalar fields with @xmath325 , we connected the early and late phases at the time @xmath229 as the pristine fraction decreases to 0.5 .
however , as discussed in 5.4.1 , one can still use the parameters given in tables 2 , 3 , and 4 if a connection at an earlier time , @xmath215 , is preferred in a particular application .
tables 2 , 3 and 4 can be used for practical applications .
one may first fix the three parameters , @xmath225 , @xmath7 and @xmath226 , by interpolating the tabulated values according to the flow mach number , @xmath11 , and the pollutant injection scale , @xmath13 . and for interpolation purposes , one can replace @xmath101 by , say , @xmath337 for the case with @xmath141 and @xmath338 for subsonic flows with @xmath339 , we expect the parameters to be very close to those measured here for the @xmath141 flow .
as shown in ps10 and pan and scannapieco ( 2011 ) , the velocity structures at all orders in the mach 0.9 flow are essentially the same as in incompressible turbulence ( corresponding to the limit @xmath340 ) . in the other limit of large @xmath11
, the timescales would not change with @xmath11 for @xmath341 , since they already saturate at @xmath342 .
the parameter @xmath7 may keep decreasing as @xmath11 increases above 6.2 , and in that case one may obtain @xmath7 by extrapolation , with the expectation that @xmath7 has a minimum value of 1 , corresponding to the highest degree of spatial locality in the pdf convolution . for the dependence of the timescales on @xmath13
, we can use the approximate power - law scalings given above for @xmath329 .
next , depending on the initial pollutant fraction @xmath69 , one may decide whether to start with an early phase using the discrete convolution model . for different values of the ratio @xmath302 ,
@xmath7 does not change , and the timescales @xmath225 and @xmath226 may be obtained from the weak power - law scaling with @xmath4 given in 5.4.4 .
the scaling applies for @xmath304 . for convenience ,
we have computed fits to @xmath225 , @xmath7 and @xmath343 which can be used in place of interpolating along the table . because the regime in which @xmath344 is the most important one for most astrophysical systems , we have focused on this case when computing our @xmath13 dependence , and furthermore , because of the statistical noise in our measurements
, we have taken an average scaling of @xmath345 for both @xmath346 and @xmath347 imposing a strict floor of @xmath348 and the @xmath349 scaling measured above we find @xmath350 \left(\frac{l_{\rm p}}{l_{\rm f } } \right)^{0.63 } \times \notag\\ \left(\frac{z_{\rm c}}{10^{-7 } \langle z \rangle } \right)^{0.015 } , \notag \\
\tau_{\rm con2 } = \left[0.335 - 0.095 \exp(-m^2/4 ) \right ] \left(\frac{l_{\rm p}}{l_{\rm f } } \right)^{0.63}\hspace{.3 cm } \times \notag \\
\left(\frac{z_{\rm c}}{10^{-7 } \langle z \rangle } \right)^{0.02 } , \notag\\ n = 1 + 11 \ , \exp(-m/3.5 ) \left(\frac{l_{\rm p}}{l_{\rm f}}\right)^{1.3 } , \hspace{1.7 cm } \label{eq : taunfit}\end{gathered}\ ] ] which provides good fits for all mach numbers and pollution properties , as long as @xmath351 and @xmath352 we finally point out that the parameters may have a dependence on how the turbulent flow is driven . for example , in a compressively driven flow at the same mach number , @xmath7 may be smaller than measured from our simulations ( see 5.4.2 ) .
in previous sections , we have focused on understanding the fundamental physics of the pollution of pristine flow material by turbulent mixing .
we now describe how our results can be applied to investigate the pollution of primordial gas in the interstellar media of high - redshift galaxies . in this section
, we discuss using our results to obtain a qualitative estimate of the pollution timescale in early galaxies , similar to the formalism of tinsley ( 1980 ) in which the evolution within a galaxy is reduced to a few general parameters . a more accurate approach based on large - eddy simulations and subgrid
modeling will be presented in the next section . to study the mixing of heavy elements in interstellar turbulence
, we need to specify the source term in the pdf equation ( [ pdfeq ] ) , which can be evaluated by considering how the pollutants , including fresh metals from supernova explosions and low - metallicity or pristine infall gas , affect the metallicity pdf ( see 2.2 ) . if the supernova rate per unit volume in a given region of a galaxy is @xmath353 , the source term due to new metals from supernovae would be @xmath354 $ ] where it is assumed that , on average , each supernova produces an ejecta mass of @xmath355 with a mass fraction of metals @xmath356 , and that rayleigh - taylor and kelvin - helmholtz instabilities arising during the explosion mix the fresh metals with the envelope material .
in reality , the source term by supernovae may have a finite width instead of being a delta function , because the ejecta mass and the heavy element yield vary with the mass of the progenitor star .
one can refine the form of the source term by using nucleosynthesis results for the ejecta mass and metal yield as functions of the progenitor mass ( maeder 1992 ; woosley & weaver 1995 ; heger & woosley 2002 ) and accounting for the initial stellar mass function .
the @xmath357 term corresponds to the replacement of the existing pdf in a fraction of the interstellar gas by @xmath358 due to the release of new metals from supernovae , and it guarantees the source term conserves the total probability . during the formation of an early galaxy , there may exist an infall of primordial gas that continuously flows from the halo into the galaxy .
this provides another source term , @xmath359 $ ] , where @xmath360 denotes the local infall rate .
the infall rate should be take to be zero except at the boundary , where the pristine gas enters the galaxy .
again the @xmath357 term is to ensure the conservation of the total probability . clearly , new metals from supernovae and the pristine infall gas force spikes at high and low concentration levels in the pdf , respectively . we define a global pristine fraction as @xmath361 where @xmath53 is the total volume of the galaxy and @xmath362 is the total mass of the interstellar gas .
an equation for @xmath363 can be derived by performing a double integration of eq .
( [ pdfeq ] ) over space and concentration .
the advection term vanishes when integrated over space , and the double integral of the supernova source term gives @xmath364 , where @xmath365 is the total supernova rate in the galaxy .
clearly , the contribution from supernovae is always negative . on the other hand
, the infall of primordial gas contributes a positive term @xmath366 $ ] , where @xmath367 is the global infall rate . using the self - convolution model for the diffusivity term in the @xmath363 equation ,
we obtain , @xmath368 a similar equation with @xmath369 and without the infall term was first given in pan and scalo ( 2007 ) . when writing eq .
( [ eq : pfgalaxy ] ) , we have made an implicit assumption of statistical homogeneity , which may break down for several reasons .
first , the prediction of the self - convolution model for the pristine fraction evolution is tested and verified only in statistically homogeneous flows , and it may not be valid for a system with large - scale inhomogeneities .
second , eq . ( [ eq : pfgalaxy ] ) adopts single values for the parameters @xmath7 and @xmath8 , equivalent to assuming similar turbulence conditions everywhere in the interstellar gas .
finally , the first ( mixing ) term on the r.h.s . is nonlinear with @xmath363 , and this nonlinearity would affect the prediction accuracy , if , for example , the star formation and hence the metallicity have a large - scale gradient . the parameters in eq .
( [ eq : pfgalaxy ] ) should be viewed as the effective averages over the turbulence and metallicity conditions of the entire galaxy .
these suggest the solution of eq .
( [ eq : pfgalaxy ] ) only provides a rough estimate for the global pristine fraction , which can be improved by accounting for realistic complexities .
nevertheless , the equation is a useful guideline for the study of the primordial gas pollution in early galaxies .
the turbulence conditions in the interstellar media of early galaxies are essentially unknown , and thus the parameters in eq .
( [ eq : pfgalaxy ] ) can not be estimated with certainty .
here we will make various assumptions for the turbulence parameters , and discuss how the pollution of the pristine gas proceeds under different conditions .
future observations will help constrain the parameter space and give a clearer picture of the mixing process in high - redshift galaxies .
a crucial parameter for mixing in the interstellar gas is the driving length scale of the interstellar turbulence , @xmath370 if the turbulence is driven at the largest scales , e.g. , by the collapse of the baryonic matter into the potential well of the dark matter halo , then @xmath144 is close to the size of the galaxy , @xmath371 in this case , the primary energy source for turbulence is the gravitational energy , and the driving force of the interstellar turbulence is associated with the source term in eq .
( [ eq : pfgalaxy ] ) for the pristine infall .
the driving scale may also remain close to @xmath372 at late times if the infall from the halo is persistent during the galaxy evolution . on the other hand ,
if the primary energy source for interstellar turbulence is the explosion energy of supernovae , then @xmath144 is likely on the order of the typical size of supernova remnants , @xmath373 in general , we expect @xmath374 and , depending on how @xmath144 compares with @xmath372 , the pollution will proceed in qualitatively different ways .
we first consider the case where the turbulent driving scale , @xmath144 , is close to the galaxy size , @xmath372 . with @xmath375
, we may roughly think of the entire interstellar medium as corresponding to our simulation box , and the dynamical time , @xmath140 , may be calculated by dividing @xmath372 by the rms turbulent velocity . as new metals from supernovae are released to the interstellar turbulence , the supernova source term in eq .
( [ eq : pfgalaxy ] ) reduces the pristine fraction , @xmath363 .
we can start applying the self - convolution model in eq .
( [ eq : pfgalaxy ] ) to calculate @xmath363 , once the average metallicity exceeds the threshold value , @xmath4 . the parameters @xmath7 and @xmath8 can be estimated based on our simulation results tabulated in 5.4.6 . with more supernovae exploding , the pollution process would become faster due to the increased amount of pollutants .
also , the average pollutant separation and hence the injection scale , @xmath13 , will decrease with the total number of supernovae , @xmath376 assuming a random supernova distribution , @xmath13 scales like @xmath377 .
the convolution timescale @xmath8 would thus decrease with time , according to the power - law scaling of @xmath8 with @xmath13 resulting in a faster pollution rate .
a subtle and minor effect is that the increase of the average metallicity reduces the threshold to average ratio , @xmath302 , leading to a slight additional decreases in @xmath8 .
this t may be accounted for using the @xmath4-dependence of @xmath8 given in 5.4.4 . if the infall of pristine gas is persistent , the infall term in eq .
( [ eq : pfgalaxy ] ) provides a continuous source for the pristine fraction , and there may exist a quasi - steady - state for @xmath363 ( see pan and scalo 2007 ) , which is controlled by three timescales , the convolution timescale , @xmath8 , the timescale for supernova sources , @xmath378 , and the timescale for the mass accretion by infall , @xmath379 .
the estimate of @xmath363 is more complicated if the driving scale , @xmath144 , is much smaller than @xmath372 .
if @xmath380 , the correlation length scale of the turbulent velocity field is much smaller than the size of the entire interstellar medium , and one may view the interstellar medium as a collection of many independent " turbulent regions of size @xmath381 .
the pollution process in each region would be similar to that in our simulation box , with a timescale determined by the local stretching / convolution timescale , @xmath382 ( @xmath383 ) .
however , the pollution in the entire interstellar medium may not be simply described by a self - convolution model or eq .
( [ eq : pfgalaxy ] ) with a local convolution timescale .
this is because the situation in individual regions of size @xmath144 may be completely different .
for example , the regions that had supernova explosions at early times may have already been significantly polluted , while the pollution process may have not yet started in the regions that had not experienced supernovae or received any heavy elements .
thus the mixing / pollution timescale over the entire galaxy may depend on the large - scale turbulent transport of pollutants between the independent " regions .
assuming a random walk model for turbulent transport at scales @xmath384 , the transport timescale at the galactic scale may be roughly estimated as @xmath385 , which is much larger than the local stretching timescale @xmath283 and the timescale @xmath386 . if @xmath380 , another timescale of interest is @xmath387 , defined as the time needed for the average separation between the supernova remnant locations to decrease below @xmath205 .
in other words , @xmath387 represents the time for supernovae to populate the interstellar medium at a level of about one per region of size @xmath144 .
if the supernovae are randomly distributed , @xmath387 can be estimated from @xmath388 , where @xmath389 is the total number of supernovae exploded before time @xmath40 . at @xmath390 ,
only a smaller number of supernovae occur , and the supernova sources would be statistically inhomogeneous at the scale @xmath144 . in that case , eq .
( [ eq : pfgalaxy ] ) is not directly applicable as it implicitly assumes statistical homogeneity ( see above ) .
thus , the pristine fraction evolution in the @xmath380 case depends on a comparison of three timescales , @xmath140 , @xmath387 and @xmath391 . from their definitions , @xmath392 and
the amplitude of @xmath387 relative to these two timescales is crucial for how the pollution proceeds .
if the star formation or supernova rate is so high that @xmath393 , the supernovae fill the interstellar medium quickly , and its spatial distribution would appear more or less homogeneous at the scale @xmath144 before each region of size @xmath144 is significantly polluted .
this suggests that the pollution in all the
independent " regions would roughly proceed at a similar pace , and the pristine fraction evolution in each region may approximately reflect the global pristine fraction .
therefore , at @xmath394 , one may apply eq .
( [ eq : pfgalaxy ] ) to estimate the global pristine fraction using @xmath7 and @xmath8 corresponding to the physical conditions at the scale @xmath144 . in this case , the timescale for the decay of @xmath363 would be @xmath395 . if @xmath396 , the mixing of fresh metals from a supernova with the surrounding region of size @xmath144 is fast with a relatively short timescale ( @xmath382 ) , and the interstellar medium would have been completely polluted if , on average , each region of size @xmath144 in the galaxy had one supernova explosion .
this is expected to occur at time @xmath397 , and thus the pollution timescale in the entire galaxy is on the order @xmath398 . finally , if the star formation rate is very low and @xmath387 is significantly larger than @xmath391 , then the turbulent transport at large scales ( @xmath384 ) plays a crucial role in the pollution process .
the delivery of heavy elements by the large - scale transport provides the entire galaxy with pollutants before the metal deposit by supernova events covers most of the interstellar medium .
the pollution in the galaxy would be completed at @xmath391 . in this case , modeling the effect of the large - scale turbulent transport is essential .
one interesting limiting case is when the interstellar turbulence is completely driven by supernova explosions , and turbulent motions are weak outside the influence radius @xmath399 of each supernova . in that case
, we have @xmath400 in regions affected by supernovae , and the transport of metals in between supernova locations would be slow with a large timescale @xmath391 . from the discussion above
, the pollution timescale would be determined by the maximum of the two timescales , @xmath140 and @xmath387 .
we note that a quantitatively accurate prediction for this case may need to carefully account for the correlation between the metal injection and the turbulence driving force . while eq .
( [ eq : pfgalaxy ] ) provides a rough estimate for the pollution of primordial gas in early galaxies , perhaps the best tool for a quantitative prediction is a large - scale numerical simulation that can include complexities such as large - scale velocity and metallicity inhomogeneities of the interstellar medium , and the effect of large - scale transport .
so far we have ignored the advection term in the pdf equation ( [ pdfeq ] ) , which is responsible for the transport of the local pdf by the velocity field .
the transport effect on the primordial gas pollution is substantial under certain circumstances , as seen earlier in the pollution timescale estimate . in the next section
, we will establish a formulation for large - scale simulations of the pristine gas pollution in early galaxies . in the context of large - eddy simulations
, the advection term corresponds to the local pdf or pristine fraction exchange between neighboring computational cells due to both the large - scale velocity and the subgrid turbulent motions .
modeling the advection term in the pdf equation is crucial in these simulations , and we will adopt a commonly - used subgrid closure for the transport effect by subgrid turbulent motions .
the complexities present in a realistic high - redshift galaxy can only be dealt with in detail through direct numerical simulations of the pollution of pristine gas in the interstellar medium of a high - redshift galaxy .
however , it is prohibitively expensive for such simulations to resolve the scale at which homogenization by molecular diffusivity occurs in interstellar turbulence .
limited resolution implies significant numerical diffusion , which causes artificial mixing , erasing any metallicity fluctuations that would exist below the size of a computational cell .
in fact , due to the vast range of scales existing in the problem , resolving any inertial - range scales at all is extremely challenging ( scannapieco & br " uggen 2010 ) .
this results in an underestimate in the degree of metallicity fluctuations / inhomogeneity in the interstellar gas .
nevertheless , a large - scale simulations still provide useful estimates for the low - order metallicity statistics , such as the metallicity variance , if they manage to resolve a small portion of inertial range , since the majority of the scalar fluctuation power is at large scales . on the other hand ,
the problem is much more severe for the pollution of the primordial gas , which corresponds to high - order statistics of the metallicity fluctuations . since the threshold metallicity @xmath4 for the transition to pop ii star formation is tiny
, a computational cell would essentially lose all the pristine gas once it is subject to the pollution by even a small amount of heavy elements . a significant underestimate in the pristine mass fraction
is therefore expected in simulations that do not resolve a considerable inertial range . here
we propose to approach the problem using large eddy simulations ( les ) that keep track of the concentration fluctuations at subgrid scales .
in such simulations , the flow at large scales is directly computed , while the effects of turbulent motions at subgrid scales are modeled .
the existence of scale invariance in the inertial range of turbulent flows is crucial for subgrid modeling ( meneveau & katz 2000 ) , which justifies using the resolved flow structures to infer the feedback effect of small - scale fluctuations . in this section , we outline an les scenario for simulating the pollution of pristine gas in early galaxies . in 7.1 , we first derive the les equations for the interstellar turbulent flow and introduce subgrid models to close the equations , taking so - called one - equation subgrid model ( e.g. , lilly 1966 ) , which evolves the turbulent kinetic energy at subgrid scales , as an illustrative example . in 7.2 , we develop an les formulation for the pristine fraction based on an equation for the local concentration pdf filtered at the resolution scale , using the self - convolution pdf models .
the model parameters can be determined with the simulation results summarized in 5.4.6 . by retaining the subgrid concentration fluctuations
, the model provides a remedy to the over - pollution by numerical diffusion , and is expected to significantly improve the predicting power of large - scale simulations for the primordial gas pollution in high - redshift galaxies .
we start by introducing the basic filtering procedure used to derive the governing flow equations at resolved scales in les .
the procedure employs a low - pass filter function , @xmath401 , which eliminates fluctuations below the resolution scale of the simulation grid , @xmath279 .
examples of the filtering function are a window function of width @xmath402 or a gaussian function with variance @xmath403 . for any flow variable , @xmath47 , the filtered quantity @xmath404 is defined as , @xmath405 and it represents the variable at the resolved scales . from eq .
( [ filter ] ) , we have , @xmath406 where integration by parts is used to obtain the second equality . for compressible flows , it is more convenient to use the favre filtering ( e.g. , speziale et al .
1988 , moin et al .
1991 , erlebacher et al .
1992 ) , defined as , @xmath407 where a density - weighted factor is included . applying the filtering procedure to the continuity and momentum equations
gives , @xmath408 and , @xmath409 where @xmath410 , called the subgrid - scale stress tensor , is defined as @xmath411 this tensor can not be evaluated exactly because of the closure problem , and developing an adequate model for it is essential for large - eddy simulations .
the filtered pressure can be written as @xmath412 , where @xmath413 is the gas temperature , and @xmath414 is the ideal gas constant with @xmath415 , @xmath416 and @xmath417 the boltzmann constant , the molecular weight and the atomic mass unit , respectively .
the viscous stress tensor , @xmath418 , in eq .
( [ filteredmomentum ] ) is given by @xmath419 where @xmath420 is the kinematic viscosity and @xmath421 is the rate of strain tensor .
we approximate the filtered viscous stress by @xmath422 , where @xmath423 is the strain tensor at the resolution scale . for interstellar turbulence ,
various sources contribute to the driving force , @xmath424 , in the momentum equation , including , e.g. , gravity and the acceleration by supernovae . to evaluate the pressure term in eq .
( [ filteredmomentum ] ) , one needs to consider the filtered energy or temperature equation , which reads ( e.g. , garnier et al . 2009 ) ,
@xmath425 where @xmath426 is the heat capacity of the flow material , equal to @xmath427 for a monoatomic gas .
the last two terms , @xmath181 and @xmath428 on the first line are the heating rate by external sources and the cooling rate by radiation , respectively .
the two terms in the second line of equation ( [ filteredtemperature ] ) correspond to the effects of @xmath429 work and heating by viscous dissipation at subgrid scales , which we model later in this section .
the first term on the third line represents thermal conduction with @xmath430 the thermal conductivity , where we assumed @xmath431 . the last term in eq .
( [ filteredtemperature ] ) is the heat transport by subgrid turbulent motions , and the temperature flux @xmath432 is defined as , @xmath433 which will be modeled later .
a variety of subgrid models have been developed to approximate @xmath410 ( see reviews by lesieur and metais ( 1996 ) and meneveau and katz ( 2000 ) for les of incompressible flows ) . a major class of models adopt an eddy - viscosity assumption , relating the deviatoric part of @xmath410 to the resolved strain tensor , @xmath434 where the eddy viscosity , @xmath435 is usually constructed as the product of a length scale ( @xmath436 ) and a velocity scale characteristic of the subgrid turbulent motions . in an les for interstellar turbulence , @xmath437 is typically much larger than the kinematic viscosity @xmath420 , and the viscous stress term in eq .
( [ filteredmomentum ] ) may be neglected . for incompressible flows , @xmath438 in eq .
( [ eddyviscosity ] ) vanishes because @xmath439 , and the isotropic part , @xmath440 , of the subgrid stress can be absorbed in the pressure term .
therefore , one obtains a complete subgrid model for les of incompressible flows by setting @xmath441 . on the other hand , in compressible flows ,
the isotropic part must be modeled explicitly .
this part behaves like a pressure term , and is sometimes named turbulent pressure " .
note that @xmath442 , with @xmath443 the turbulent kinetic energy per unit mass at subgrid scales .
similar to the eddy - viscosity model for the subgrid stress , one may adopt an eddy - diffusivity assumption for the temperature flux , @xmath432 , caused by subgrid turbulent motions , @xmath444 where the eddy conductivity " @xmath445 is of the same order as @xmath437 , and is usually parameterized by a subgrid prandtl number , @xmath446 where @xmath447 is typically taken to be @xmath448 ( e.g. , edison 1985 , erlebacher et al . 1992 ,
jaberi et al . 1999 ) .
eddy - viscosity models differ in how @xmath437 is evaluated . in the smagorinsky ( 1963 ) model , @xmath437 is calculated by @xmath449 with @xmath450 , which essentially assumes the amplitude of the subgrid velocity fluctuations goes like @xmath451 .
the smagorinsky model has also been used in the les of compressible flows ( e.g. , moin et al .
1991 , erlebacher et al .
1992 , vreman et al .
1997 ) . for compressible flows , yoshizawa ( 1986 ) proposed to set @xmath452 for the isotropic part of the subgrid stress , which appears to underestimate the subgrid kinetic energy ( park & mahesh 2007 ) .
a variant of the eddy - viscosity model is the so - called one - equation model , where an equation for the subgrid kinetic energy , @xmath443 , is derived , modeled and solved ( e.g. , lilly 1966 , schumann 1975 , moeng 1984 , ghosal et al .
1995 , menon & kim 1996 ; for one - equation models of compressible flows , see , e.g. , schmidt et al . 2006 ,
park & mahesh 2007 , genin & menon 2010 , and chai & mahesh 2012 ) . using the solved subgrid kinetic energy , the eddy - viscosity
is then estimated by , @xmath453 in this paper , we will consider the one - equation model primarily as an example to illustrate the construction of an les for the pollution of primordial gas in interstellar turbulence .
for a compressible flow , the subgrid kinetic energy equation is given by , @xmath454 , \label{subgridkinetic}\end{gathered}\ ] ] where the first term on the r.h.s .
represents the production of subgrid kinetic energy by the cascade from resolved scales .
the two terms on the second line appeared earlier in the filtered temperature equation , corresponding to the @xmath429 work ( or pressure - dilation ) and the viscous dissipation at subgrid scales .
the pressure - dilation term is sometimes neglected for weakly compressible flows because it is difficult to model ( e.g. , moin et al .
1991 , erlebacher et al .
1992 ) , but in highly compressible flows the @xmath429 work is not negligible and needs to be accounted for . using direct numerical simulations of supersonic turbulence , ps10 showed that , despite its reversible nature , the @xmath429 work tends to convert kinetic energy to thermal energy , and thus acts as a significant kinetic energy sink in addition to the viscous dissipation .
based on their results , one can model the pressure - dilation and viscous dissipation terms together as @xmath455 , where the subgrid dynamical time , @xmath456 , is assumed to be @xmath457 . here
we have implicitly assumed that the filter size lies in the inertial range of the real flow and also assumed that the flow driving " length at subgrid scales is @xmath436 .
this may not be true if , for example , the supernova explosion is the main energy source for turbulence and the resolution scale is significantly larger than the size of supernova remnants .
we do not consider this complexity here , as the one - equation subgrid model is used largely for an illustration purpose .
the dimensionless parameter @xmath458 is expected to be a function of the subgrid mach number , @xmath459 , and the dependence of @xmath458 on @xmath460 may be determined using the simulation results of ps10 .
the last term on the first line of eq .
( [ subgridkinetic ] ) corresponds to the addition of kinetic energy at subgrid scales by the driving force , @xmath424 .
if the characteristic length scale of @xmath424 is much larger the filter size , @xmath461 , and @xmath462 would be negligible , meaning that the driving force stores kinetic energy mainly at the resolved scales .
on the other hand , a significant fraction of energy input from supernova explosions may be deposited primarily as subgrid kinetic energy , if the simulation does not resolve the typical size of supernova remnants ( scannapieco & bruggen 2010 ) . in that case , @xmath463 for isotropically expanding supernova remnants , and the subgrid kinetic energy input can be estimated as the product of the supernova explosion energy and the local supernova rate per unit volume . the transport ( or flux ) terms in the last two lines of eq .
( [ subgridkinetic ] ) are usually grouped and modeled together as a diffusion of the subgrid kinetic energy ( lilly 1966 , schumann 1975 , moeng 1984 , ghosal et al .
1995 , schmidt et al .
2006 , genin & menon 2010 , see , however , chai & mahesh 2012 for separate treatment of each individual term ) . here
we adopt an eddy - diffusion assumption for the first three flux terms and approximate them together by @xmath464 , where @xmath465 .
the parameter @xmath466 is sometimes set to be equal to @xmath467 in eq .
( [ nut ] ) for the subgrid stress tensor ( e.g. , kim & menon 1999 ) . in general
, they may be different and need to be treated separately ( e.g. , schmidt et al .
the last flux term in eq .
( [ subgridkinetic ] ) can be written as @xmath468 where the temperature flux , @xmath432 , by subgrid motions is modeled by eq .
( [ tflux ] ) . with these assumptions
, we have ( see genin & menon 2010 ) , @xmath469 \notag\\ \hspace{-1.1 cm } + \overline{\rho } \left(\widetilde{v_i f_i } - \widetilde{v}_i \hspace{0.5 mm } \widetilde{f}_i \right ) , \label{subgridkineticmodel}\end{gathered}\ ] ] which is in a closed form and can be evolved to obtain the subgrid turbulent energy .
we next consider the filtered temperature equation ( [ filteredtemperature ] ) .
we use the eddy - diffusivity model , eq .
( [ tflux ] ) , for the temperature flux , @xmath432 , in the last term of eq .
( [ filteredtemperature ] ) , and model the pressure - dilation and viscous dissipation terms as in eq .
( [ subgridkineticmodel ] ) . with these assumptions ,
we obtain , @xmath470 where the thermal conductivity term can be neglected if @xmath471 . an alternative approach to obtain @xmath472 is to model and evolve the equation of the filtered total energy , @xmath473(@xmath474 ) , per unit mass ( e.g. , vreman et al . 1997 , kosovic et al . 2002 , schmidt et al .
2006 , park & mahesh 2007 , genin & menon 2010 , scannapieco & bruggen 2010 , chai & mahesh 2012 ) . to solve the les and @xmath443 equations
, one needs to determine four parameters , @xmath467 , @xmath447 , @xmath458 , and @xmath466 .
traditionally , these are assumed to be positive constants and specified _ a priori _ and then tuned by testing against experiments or numerical simulations . on the other hand , this approach has the weaknesses of failing to fully account for the flow - dependence of these parameters , as well as not allowing the backscatter of the subgrid kinetic energy to the resolved scales , which does occur in some local regions of a turbulent flow ( piomelli et al .
these limitations motivated a dynamic procedure for subgrid modeling where the model coefficients are computed in a localized and adaptive way using the flow structures at resolved scales and the assumption of scale invariance ( e.g. germano et al .
1991 ; moin et al .
1991 , germano 1992 ; lilly 1992 ; ghosal et al .
1995 , kim & menon 1999 ; schmidt et al . 2006 ; park & mahesh 2007 ; genin & menon 2010 ; chai & mahesh 2012 ) .
here we have restricted our attention to the eddy - viscosity models and focused particularly on the one - equation model .
the interested reader is referred to , e.g. , vreman et al .
( 1997 ) , meneveau & katz ( 2000 ) , and de stefano et al .
( 2008 ) , for non - eddy - viscosity subgrid models and their dynamic versions .
two - equation subgrid models have also been developed , which , in addition to the subgrid kinetic energy , evolve another subgrid quantity , such as the dissipation rate ( e.g. , gallerano et al .
2005 ) or the characteristic length scale of subgrid turbulent motions ( e.g. , fang & menon 2006 ; dimonte & tipton 2006 ; scannapieco & bruggen 2010 ) . in this subsection ,
we construct a subgrid model for the pollution of primordial gas in early galaxies .
we first consider the equation for the filtered concentration field , which provides a general illustration for subgrid modeling of turbulent mixing . applying the filtering procedure to the advection - diffusion equation ( [ advection ] ) gives , @xmath475 where @xmath476 is the concentration flux caused by subgrid turbulent motions .
the equation is similar to the temperature equation ( [ filteredtemperature ] ) except for the pressure - dilation and viscous dissipation terms . in analogy to the subgrid temperature flux , @xmath477 , one may adopt an eddy - diffusivity assumption for the concentration flux , @xmath478 , yielding , @xmath479 where we also assumed @xmath480 .
the eddy diffusivity , @xmath481 , is of the same order as the eddy viscosity , and the subgrid schmidt number @xmath482(@xmath483 ) is sometimes set to be the same as the subgrid prandtl number @xmath484 ( e.g. , jaberi et al . 1999 ) .
somewhat smaller values , @xmath485 , have also been proposed ( e.g. , pitsch & steiner 2000 and jimenez et al . 2001 ) .
@xmath482 can also be computed from the local flow structures using the dynamic procedure discussed above ( see e.g. , moin et al . 1991 , pierce & moin 1998 ) . for the les of interstellar turbulence ,
@xmath481 is expected to be much larger than the molecular diffusivity @xmath33 .
similar to the subgrid kinetic energy , we can derive an equation for the subgrid concentration variance , defined as @xmath486 .
the equation reads ,
@xmath487 \notag\\ \hspace{-3.5 cm } + 2 \overline{\rho } ( \widetilde{sc } - \widetilde{s } \hspace{0.5 mm } \widetilde{c } ) , \label{filteredconcentrationvariance } \end{gathered}\ ] ] which is in close analogy to eq .
( [ subgridkinetic ] ) . the first term , @xmath488 , on the r.h.s
. represents the production of the concentration variance by the scalar cascade from the resolved scales .
the term on the third line corresponds to the subgrid scalar dissipation by molecular diffusivity .
we model the dissipation as @xmath489 ( see eq .
( [ ensemblevariance ] ) ) , where the subgrid mixing timescale @xmath490 is expected to scale with the subgrid dynamical time , @xmath491 .
parametrizing @xmath492 with respect to @xmath456 , we set the dissipation term to @xmath493 , where @xmath494 .
the parameter @xmath495 depends on the local subgrid mach number , @xmath496 , and also on the subgrid length scale @xmath497 at which the pollutants are injected , and it can be calibrated using the simulation results of ps10 , who tabulated the mixing timescale of passive scalars forced at different length scales in turbulent flows at a range of mach numbers . if the pollutants are forced at large scales and the subgrid fluctuations are contributed primarily by the cascade from resolved scales , and it is appropriate to set @xmath498 .
however , @xmath497 could be smaller than the resolution scale , @xmath279 , if , for example , multiple supernovae explode in a single computational cell ( scannapieco & br " uggen 2010 ) . finally , we model the transport ( flux ) terms ( the last term in the first line and all the terms in the second line ) together as a diffusion term , @xmath499 .
the eddy - diffusivity , @xmath500 , here is likely to be close to @xmath481 in the @xmath501 equation , although it is not clear if they are exactly equal ( see below ) .
these assumptions result in a closed variance equation ( c.f .
jimenez et al . 2001 ) , @xmath502 \notag \\ + 2 \overline{\rho } ( \widetilde{sc } - \widetilde{s } \widetilde{c } ) , \hspace{2.9 cm } \label{modeledconcentrationvariance } \end{gathered}\ ] ] which illustrates the basic picture for modeling the subgrid concentration fluctuations , and provides a useful guideline for formulating a subgid model for the pollution of pristine gas . a dynamic procedure for the subgrid scalar variance and dissipation
was developed in pierce and moin ( 1998 ) .
the subgrid model we construct for the pollution of pristine gas is based on the the pdf formulation in the context of les . applying a favre filter , eq .
( [ favrefilter ] ) , to the fine - grained concentration pdf @xmath503 , we define a density - weighted pdf at the resolution scale , @xmath504 an exact equation for the filtered pdf , @xmath505 , is derived in appendix a , @xmath506}_{\rho } \right ) = \frac{\partial}{\partial x_i } \left ( \overline{\rho \gamma \frac{\partial \phi}{\partial x_i } } \right ) \hspace { 3cm}\notag \\ \hspace{2.2 cm } - \frac{\partial^2}{\partial z^2 } \left(\overline{\rho } \hspace{0.5 mm } \widetilde{\phi }
\overline{\left[\gamma \left(\frac{\partial c}{\partial x_i } \right)^2 \left\vert \vphantom{\frac{1}{1 } } \right .
c = z \right]}_\rho \right)\notag \\ \hspace{.3 cm } - \frac{\partial}{\partial z } \left ( \overline{\rho } \hspace{0.5mm}\widetilde{\phi } \hspace{0.5 mm } \overline { [ s|c = z ] } _ { \rho } \right ) , \label{rhofilteredeqtext}\end{gathered}\ ] ] where @xmath507}_\rho$ ] denotes density - weighted filtering conditioned on the local concentration value .
the definition of the conditional filtering is given in appendix a. eq .
( [ rhofilteredeqtext ] ) is essentially identical to eqs .
( [ pdfeq ] ) and ( [ ensemblediffusivityterm ] ) for the ensemble - defined pdf , @xmath508 .
this implies that , first , the same closure problem exists for the advection and diffusivity terms in eq .
( [ rhofilteredeqtext ] ) , and , second , the pdf closure models in the ensemble context can be applied to the filtered pdf equation .
although our primary goal is not to solve the equation for the entire filtered pdf , we give an outline for modeling the pdf equation , which is helpful for understanding our les approach for the pristine fraction . we first consider the advection term , which is responsible for the transport of the pdf between different regions by the turbulent velocity . we write it in two terms , @xmath509}_{\rho } = \widetilde{\phi } \hspace{0.5 mm } \widetilde{v}_i + \widetilde{\phi } \left ( \overline{[v_i|c = z]}_{\rho } - \widetilde{v}_i\right)$ ] , and then modeling the second term with an eddy - diffusivity assumption gives @xmath509}_{\rho } = \widetilde{\phi } \hspace{0.5 mm } \widetilde{v_i } - \gamma_{\rm t \phi } { \partial_i \widetilde{\phi } } $ ] where @xmath510 is the eddy - diffusivity for the pdf flux by the subgrid motions ( see , e.g. , gao & obrien ( 1993 ) , colucci et al .
( 1998 ) , jaberi et al .
( 1999 ) ) .
the filtered pdf equation then becomes , @xmath511}_\rho \right ) \notag\\ \hspace{0.1 cm } - \frac{\partial}{\partial z } \left ( \overline{\rho } \hspace{0.5mm}\widetilde{\phi } \hspace{0.5 mm } \overline { [ s \vert c = z ] } _ { \rho } \right ) , \label{rhofilteredeqadvection}\end{gathered}\ ] ] where we also assumed that @xmath512 ( see , e.g. , jaberi et al .
1999 ) . taking the first - order moment of eq .
( [ rhofilteredeqadvection ] ) gives an equation for the filtered concentration @xmath501 , which is the same as eq .
( [ filteredconcentrationadvection ] ) except that @xmath481 is replaced by @xmath510 .
this suggests that @xmath513 ( gao & obrien 1993 ) . also , using the second - order moment of eq .
( [ rhofilteredeqadvection ] ) , we can derive an equation for the subgrid variance , @xmath514 , which is the same as eq .
( [ modeledconcentrationvariance ] ) except that @xmath510 replaces both @xmath481 and @xmath515 this indicates that the eddy diffusivities for the concentration mean ( @xmath481 ) and variance ( @xmath500 ) are automatically set to be equal if one models the advection term in the pdf equation with an eddy - diffusivity closure . the term on the second line of eq .
( [ rhofilteredeqadvection ] ) represents homogenization by molecular diffusivity , and can be modeled using established pdf closure approximations for turbulent mixing , such as those discussed in 3.1 . in 6.1
, we derived an expression for the source term in the ensemble pdf equation . using the same method
, we estimate the source term in the last line of eq .
( [ rhofilteredeqadvection ] ) for the filtered pdf .
the source term for new metals from supernovae is @xmath516 , where @xmath517 is the filtered number rate of supernova explosions per unit volume , and ejecta from each supernova is assumed to have the same mass @xmath518 with metallicity @xmath356 . again
the @xmath519 term ensures the conservation of the total probability . with the supernova source term in the filtered pdf equation , it is straightforward to calculate the source terms in the filtered concentration and variance equations ( [ filteredconcentrationadvection ] and [ modeledconcentrationvariance ] ) .
we find that the source terms are @xmath520 and @xmath521 $ ] in the @xmath501 and @xmath522 equations , respectively .
if a continuous infall of pristine gas from the halo exists during the formation and evolution of a galaxy , one can maintain a mass flux at the boundary of the simulation box and set @xmath523 as the boundary condition for the ( filtered ) concentration pdf .
we finally consider modeling the pollution of the primordial gas in an les .
clearly , the fine - grained pristine fraction , @xmath524 , at a given point is an integral of the fine - grained pdf , @xmath525 , from @xmath64 to the threshold , @xmath4 , and , similarly , the filtered pristine fraction , @xmath526 , at the resolution scale is given by , @xmath527 we can therefore derive an equation for @xmath526 by integrating the filtered pdf equation ( [ rhofilteredeqtext ] ) from @xmath194 to @xmath4 .
performing such an integration for the advection term in eq .
( [ rhofilteredeqtext ] ) yields @xmath528 , which corresponds to the flux of the pristine fraction into and out of a computational cell due to the transport / advection of the turbulent velocity ( see 2.1 ) .
the term can be rewritten as @xmath529 where the term in the brackets is the pristine fraction flux caused by the subgrid turbulent motions .
we model this subgrid flux with an eddy - diffusion assumption , @xmath530 where @xmath531 is the eddy diffusivity for the pristine fraction .
we use the self - convolution models ( 3.2 ) for the effect of the diffusivity term on the pristine fraction .
integrating the supernova source term from 0 to @xmath4 gives @xmath532 . with these models and assumptions , we obtain , @xmath533 where it is also assumed @xmath534 , and @xmath535 and @xmath536 correspond to the parameters , @xmath7 and @xmath8 , in the self - convolution models discussed in 3 .
as @xmath33 is likely much smaller than @xmath531 , the pristine fraction flux due to the molecular diffusivity can be neglected .
the choice for @xmath535 and @xmath536 according to the turbulence and pollutant conditions at subgrid scales will be described and discussed below .
if a pristine mass flux is enforced at the boundary of the simulation box to imitate the primordial infall gas , one should set @xmath537 as a boundary condition .
a comparison of eq .
( [ filteredprimordial ] ) with eq .
( [ modeledconcentrationvariance ] ) shows that both equations have transport or flux terms , a mixing / homogenization term by the molecular diffusivity , and a source term .
a similar analogy also exists with the equation for the subgrid kinetic energy , eq .
( [ subgridkineticmodel ] ) .
there is , however , an interesting difference .
the concentration variance equation has a term that tends to increase the subgrid variance , representing the scalar cascade from the resolved scales . on the other hand
, there is no such production term in the @xmath526 equation , because no mechanism exists in the mixing process that can produce pristine gas at subgrid scales . to derive the equation for the filtered pristine fraction
, we could also have started from eq .
( [ rhofilteredeqadvection ] ) for the filtered pdf , where the advection term is already modeled by an eddy - diffusion assumption . in that case
, we would have found that the primordial flux due to subgrid turbulent motions is given by @xmath538 .
this suggests that , when applying an eddy - diffusivity closure to the advection term in the filtered pdf equation , it is implicitly assumed that the three eddy diffusivities , @xmath481 , @xmath500 , and @xmath531 , respectively for the mean , variance and the pristine fraction , are the same and all equal to @xmath510 .
the quantitative accuracy of this assumption is not clear , although all the eddy - diffusivities are expected to be of the same order .
a simple estimate for @xmath531 is to scale it with the eddy viscosity as , @xmath539 , with the schmidt number @xmath540 for the pristine fraction in the range from @xmath541 , as in the case of @xmath481 discussed earlier ( see text below eq .
( [ filteredconcentrationadvection ] ) ) . to implement eq .
( [ filteredprimordial ] ) in an les for the pollution of primordial gas in early galaxies , we now only need to specify the two parameters , @xmath535 and @xmath536 , from the self - convolution model . when using the convolution model in eq .
( [ filteredprimordial ] ) , we have implicitly assumed that statistical homogeneity is restored at the resolution scale , @xmath279 , because the applicability of the model is tested and confirmed only in statistically homogeneous turbulent flows . with this assumption , @xmath535 and @xmath536
can be determined using our simulation results .
these parameters are functions of the flow mach number and the pollutant injection scale relative to the flow driving scale . the subgrid mach number , @xmath460 , can be easily computed by @xmath542 in the one - equation model , where @xmath443 and @xmath472 are , respectively , the subgrid kinetic energy and the filtered gas temperature . the subgrid source injection scale , @xmath497 , in a computational cell would be close to the cell size , @xmath279 , if the pollutant source was transported into the cell by advection , or if only one supernova exploded in the cell . in that case
, it is appropriate to set @xmath543 . on the other hand ,
if multiple supernova explosions occurred in a single cell , then @xmath497 would roughly go like the number of supernovae to -1/3 power , assuming a random distribution .
we assume the subgrid flow driving " scale , @xmath544 , is roughly given by the cell size , @xmath279 . with @xmath460 and the ratio @xmath545
, one can fix the parameters , @xmath535 and @xmath536 , by interpolating tables 2 , 3 and 4 in 5.4.6 , or using the fits given in eqs.([eq : taunfit ] ) ) .
the timescales @xmath225 and @xmath226 given in tables 2 and 4 are normalized to the flow dynamical time , therefore the values for @xmath536 are in units of the subgrid dynamical time , @xmath456@xmath546 .
we point out there is an uncertainty in the applicability of our tabulated parameters to computation cells with supernova explosions . in these cells ,
the effective driving is likely better described by a pure compressive force rather than solenoidal . as discussed earlier
, this may affect the parameters in the convolution model .
future simulations are needed to investigate the potential dependence of the parameters with the compressibility of the driving force .
the convolution timescale also has a dependence on the threshold metallicity , @xmath4 , relative to the mean concentration ( see 5.4.4 ) .
thus , to determine @xmath536 , we need to compute the ratio , @xmath547 , of the threshold to the mean , @xmath501 , in a cell . for that purpose , it is necessary to solve the filtered concentration equation ( [ filteredconcentrationadvection ] ) to keep track of @xmath501 in all computational cells . as mentioned earlier , the source term in this equation is given by @xmath548 .
based our results in 5.4.4 , for small values of @xmath547 ( @xmath549 ) , one can use a weak power - law scaling ( see 5.4.4 ) to rescale the convolution timescales listed in tables 2 and 4 .
however , it is possible that @xmath501 in a computational cell is close to or even smaller than @xmath4 .
as discussed in 5.4.4 , in the extreme case with @xmath550 , the evolution of @xmath526 in a cell would be qualitatively different from the prediction of the self convolution models .
a careful treatment is thus needed for cells with @xmath501 close or smaller than @xmath4 .
how the fraction , @xmath526 , evolves under this situation is not explored in the current work , and we defer it to a future paper . in the subgrid model outlined above for the pristine fraction
, we adopted a simple approach to fix the model parameters , prescribing them based on our simulation results and previous work on les .
an interesting question is whether the parameters can be determined dynamically using the resolved local flow and scalar structures .
it seems highly uncertain whether the dynamic procedure is applicable at all to the problem of how the pristine gas is polluted in a turbulent flow . as mentioned earlier , the validity of the dynamical procedure relies on the existence of scale invariance .
this can be justified , e.g. , for the cascade of kinetic energy or the concentration variance , based on kolmogorov s similarity theory of turbulence . however , unlike kinetic energy or the scalar variance , which are 2nd - order statistical measures , the pristine fraction corresponds to the extreme pdf tail , and it is unknown whether scale invariance exists for such a high - order quantity . exploring the possibility of developing a dynamic subgrid model for the pristine gas fraction would be an interesting topic for a future study .
the shift from population iii to normal star formation is a global transition of the universe that is dependent on mixing on scales smaller than a parsec ( pan & scalo 2007 ) .
this means that numerical simulations of this process will only be possible if we first develop a deep understanding of the fundamental physics of how the pristine material is polluted in turbulent flows . in an earlier paper ( pss )
, we developed a theoretical approach to modeling this process based on the pdf method for passive scalar mixing in statistically homogeneous turbulence , and we explored the evolution of the pristine fraction , @xmath25 , defined as the mass fraction of the flow with pollutant concentration below a tiny threshold @xmath4 .
then we used numerical simulations to show that a class of pdf models , called self - convolution models , provide successful fitting functions to the solution of @xmath551 which corresponds to the far left tail of the concentration pdf .
the convolution models are based on the physical picture of turbulence stretching pollutants and causing a cascade of concentration structures toward small scales .
mixing then occurs as the scale of the structures becomes sufficiently small for molecular diffusivity to operate efficiently , and the homogenization between neighboring structures corresponds to a convolution of the concentration pdf .
the picture suggests that the mixing / pollution timescale is determined by the turbulent stretching rate at the scale where the pollutants are injected , and the main result of pss was the prediction for the pristine fraction evolution , i.e. , eq .
( [ pfnconv ] ) , by the generalized self - convolution model . for convenience
, we repeat eq .
( [ pfnconv ] ) here , @xmath552 where @xmath8 is the timescale for the pdf convolution , and the parameter @xmath7 is interpreted as an indicator of the degree of spatial locality of the pdf convolution process .
a smaller @xmath7 corresponds to more local convolution and broader pdf tails . in the present work
, we briefly reviewed the formulation of pss , and conducted a systematic numerical study of the turbulent pollution process , exploring an extended parameter space .
we simulated four statistically homogeneous turbulent flows with rms mach number @xmath11 ranging from @xmath334 to @xmath193 . in each flow
, we evolved 20 decaying scalars with different initial pollutant fractions , @xmath69 , and different pollutant injection scales , @xmath553 .
the simulation data further confirmed the validity of the convolution model and allowed us to measure the model parameters , @xmath7 and @xmath8 , in eq .
( [ pfnconvconc ] ) over a wide range of turbulence and pollutant conditions .
consistent with pss , we find that , if the initial pollutant fraction @xmath224 , the simulation results for the pristine fraction can be well fit by the convolution model prediction , eq .
( [ pfnconvconc ] ) , with properly chosen parameters .
( [ pfnconvconc ] ) is solved by @xmath554^n } , \label{pfnconvsolutionconc}\ ] ] where @xmath68 is the initial pristine fraction , and we have denoted the convolution timescale as @xmath226 for these scalar fields . using eq .
( [ pfnconvsolutionconc ] ) to fit the simulation data yielded best - fit parameters @xmath7 and @xmath226 . on the other hand ,
if @xmath241 , the evolution of @xmath25 shows different behaviors at early and late times . in the early phase , the pdf convolution occurs locally in space due to the limited amount of pollutants , and the pristine fraction evolution follows the prediction of the discrete " convolution model with @xmath15 , i.e. , @xmath555 where the convolution timescale for the early phase is denoted as @xmath225 .
once a significant fraction ( 0.2 - 0.3 ) of flow is polluted , the pristine fraction evolves in the same way as the scalar fields with @xmath556 .
we therefore named the convolution timescale as @xmath226 for both scalars with @xmath224 and the late phases of @xmath557 scalars ( see 5.4 ) .
a successful two - phase fitting scenario was adopted for scalars fields with @xmath557 , which connects eqs .
( [ pfintegralsolutionconc ] ) and ( [ pfnconvsolutionconc ] ) for early and late times .
we examined the dependence of the model parameters on the flow mach number , @xmath11 .
we found that the convolution timescales , @xmath225 and @xmath226 , normalized to the flow dynamical time increase by @xmath558 as @xmath11 goes from 0.9 to 2.1 and then saturates at @xmath559 .
this is similar to the behavior of the variance decay timescale , @xmath78 , as a function of @xmath11 . for @xmath224 scalars or the late phase of scalars with @xmath241 , the parameter @xmath7 decreases with increasing @xmath11 , indicating that the pdf convolution proceeds more locally in supersonic turbulence with larger @xmath11 .
the decrease of @xmath7 is related to broader concentration pdf tails at higher @xmath11 , corresponding to a larger pristine fraction at the same concentration variance .
the pristine fraction evolution also depends on the pollutant injection scale @xmath553 .
as @xmath553 deceases , the pollution of the pristine gas is faster and the timescales , @xmath225 and @xmath226 , decrease .
this is expected as the mixing timescale scales with the eddy turnover time at @xmath553 . for scalars with @xmath224 or
the late phase of @xmath241 scalars , the parameter @xmath7 becomes smaller as @xmath553 decreases because , intuitively , the convolution is more local if the pollutants are injected at smaller scales .
the dependence of the model parameters , @xmath7 , @xmath225 and @xmath226 , on the turbulence and pollutant properties is summarized in tables 2 , 3 , and 4 , and for convenience we have fit these results with simple functions as , @xmath350 \left(\frac{l_{\rm p}}{l_{\rm f } } \right)^{0.63 } \times \notag\\ \left(\frac{z_{\rm c}}{10^{-7 } \langle z \rangle } \right)^{0.015 } , \notag \\
\tau_{\rm con2 } = \left[0.335 - 0.095 \exp(-m^2/4 ) \right ] \left(\frac{l_{\rm p}}{l_{\rm f } } \right)^{0.63}\hspace{.3 cm } \times \notag \\
\left(\frac{z_{\rm c}}{10^{-7 } \langle z \rangle } \right)^{0.02 } , \notag\\ n
= 1 + 11 \ , \exp(-m/3.5 ) \left(\frac{l_{\rm p}}{l_{\rm f}}\right)^{1.3 } , \hspace{1.7cm}\end{gathered}\ ] ] which are applicable for all mach numbers and pollution properties , as long as @xmath352 note that unlike eqs .
( [ pfnconvconc ] ) and ( [ pfnconvsolutionconc ] ) , these fits are for convenience only and not based on an underlying physical picture .
we showed that the model is valid for @xmath560 , where @xmath225 and @xmath226 only have a weak dependence on @xmath4 ( 5.4.4 ) .
if @xmath4 is close to or larger than @xmath297 , the model is no longer applicable , and we defer a study of this situation to a later work .
we also tested the convergence of the model parameters with the numerical resolution ( 5.4.5 ) .
the parameters @xmath7 and @xmath8 may have a dependence on the compressivity of the driving force , which will be systematically examined in a future work . to apply our model and simulation results to the mixing of heavy elements in the interstellar medium , we specified the source term in the concentration pdf equation , accounting for the effects of new metals from supernova explosions and the possible infall of pristine gas from the halo or the intergalactic medium .
these two sources force spikes in the pdf at high and low concentration values , respectively . with the source term , we derived an equation ( eq .
( [ eq : pfgalaxy ] ) ) for the global pristine fraction in early galaxies .
a description for how to use the equation and our simulation results to estimate the primordial gas fraction was given in 6 .
we discussed the timescales of relevant processes that control how and how fast the pollution process proceeds .
in particular , the spatial transport by turbulent motions over galactic scales may play an important role if the interstellar turbulence is driven at small scales , e.g. , by supernova explosions .
numerical simulations accounting for the interstellar environment at galactic scales are a valuable tool to study the pollution of primordial gas from a less idealized point of view .
in fact , recent efforts to track metal mixing in the context of the formation of protogalaxies have made significant improvements in tracking the spatial evolution of the metallicity , averaged over relatively large scales ( e.g. , wise & abel 2008 ) .
greif ( 2009 , 2010 ) employed a turbulent diffusion formalism to mimic mixing by smoothing over the sph kernel , and a similar approach was used in the 10 mpc sph simulations by maio ( 2010 ) and campisi ( 2011 ) , who assumed the initial metal pollution was spread over @xmath561 kpc scales by cluster winds .
another recent simulation by ritter ( 2012 ) used a finite - difference code with adaptive mesh refinement coupled to lagrangian tracer particles to keep track of the metals produced in an initially metal - free galaxy .
interestingly , they found that a cold supersonically turbulent core developed because of the fallback of metal - enhanced ejecta .
however , because the resolution scale was much larger than the scales on which turbulence - enhanced molecular diffusivity operates ( pan & scalo 2007 ) , they could not resolve a sufficiently large range of scales to track the unmixed , primordial fraction .
to overcome limitations such as these , we have developed a large - eddy simulation ( les ) approach based on our model and simulation results . in
less , the flow quantities at resolved scales are directly computed , while the feedback effect of subgrid turbulent motions is modeled . to overcome the over - pollution by numerical diffusion ,
a subgrid model was constructed to track the evolution of the concentration fluctuations below the resolution scale . using the standard filtering procedure for the les formulation , we derived an equation for the filtered concentration pdf representing metallicity fluctuations at subgrid scales , and discussed the treatment of each term in the equation .
the core of our subgrid model is equation ( [ filteredprimordial ] ) for the filtered pristine fraction ( i.e. , the pristine fraction in each computational cell ) , which was derived from the filtered pdf equation .
again , we repeat it here for convenience : @xmath562 where @xmath535 and @xmath536 are parameters for the subgrid pollution , corresponding to @xmath7 and @xmath563 in our convolution model , @xmath531 is the eddy diffusivity for the pristine fraction , @xmath517 is the filtered number rate of supernova explosions per unit volume and each supernova is assumed to have an ejecta mass @xmath355 .
this equation adopts the commonly - used eddy - diffusivity model for the transport effect of subgrid turbulent motions , and employs the convolution model for the pollution of the pristine gas within each cell .
the implementation of our subgrid model was illustrated in the context of a one - equation les model for the interstellar turbulence , which evolves the kinetic energy of subgrid turbulent motions .
together with the resolved temperature field , the subgrid kinetic energy specifies the turbulence properties in each cell , which are needed to calculate the eddy - diffusivity in the transport term and to determine the parameters in the convolution model for the subgrid pollution .
the convolution model parameters depend on the metal / supernova sources in each cell , and can be evaluated using our simulation results . the resulting physically - realistic model for the evolution of the unresolved primordial fraction serves as a prototype for future simulations aimed at interpreting many observations currently probing the nature of early galaxies . the continuing discovery of star - forming galaxies at @xmath564 in broad - band photometric searches , for example , ( as in the hubble udf12 survey , ellis 2013 ) suggests that observations of galaxies with significant primordial fractions should soon become available .
the situation for galaxies selected on the basis of strong ly@xmath565 emission ( cowie & hu 1998 ; rhoads 2000 ) is even more promising , and it may only be a matter of time before several such galaxies are clearly identified as containing primordial stars recognizable by their large ly@xmath565 equivalent width and weak he ii emission ( scannapieco 2003 ; jimenez & haiman 2006 ) .
in fact , a recent detailed analysis of deep subaru images by inoue ( 2011 ) strongly supports the interpretation that the mass fraction of stellar populations with extremely small metal abundances in @xmath566 lyman alpha emitters may be 1 - 10 % by mass , based on their very strong rest frame lyman continua .
based on similar diagnostics , kashikawa et al .
( 2012 ) recently proposed a @xmath567 ly@xmath565 emitter as a pop iii candidate although enhanced ly@xmath565 emission from a clumpy , dusty medium ( neufeld 1991 ; hansen & oh 2006 ) can not be ruled out conclusively in this case .
if any of these galaxies are convincingly demonstrated to contain primordial stars , their evolution could only be simulated using an approach such as the one outlined here . currently , a more direct constraint on the evolution of primordial gas is based on the absence of metal lines in absorption line systems in the intergalactic medium .
fumagalli ( 2011 ) used this approach to obtain upper limits of @xmath568 by mass in two lyman limit systems associated with quasars at @xmath569 and 3.4 .
simcoe ( 2012 ) used the lack of metal lines in a @xmath570 quasar spectrum that shows a large neutral hydrogen column density to obtain an upper limit of @xmath571 depending on the whether the gas is bound in a galaxy or is diffuse intergalactic gas at that redshift .
the implications of these measurements can only be fully explored through models such as ours , which capture the unresolved , unmixed fraction . finally , at least four examples of galactic stars with [ fe / h ] @xmath572 -4.5 ( @xmath573 ) are known ( christlieb et al .
2002 ; frebel et al .
2008 ; norris et al . 2007 ;
caffau et al .
2011 ) , although only one is not enhanced in carbon .
recently , yong et al . ( 2012 ) have shown convincingly that the milky way metallicity pdf is still decreasing smoothly down to at least [ fe / h ] = -4.1 ( @xmath574 ) , without the sudden cutoff claimed in earlier work .
once the rather severe selection effects are understood , this measurements could also be directly compared with our models , and even allow their two main parameters to be calibrated outside of numerical simulations .
in fact , our proposed large - eddy simulation is expected to give reliable predictions for any physical problem , astrophysical our otherwise , in which the unresolved , low concentration tail of the pdf needs to be tracked . its numerical implementation and
targeted application represents an extremely promising avenue for future studies .
we acknowledge support from nasa under theory grant no .
nnx09ad106 and astrobiology institute grant 08-nai5 - 0018 and from the national science foundation under grant ast 11 - 03608 .
all simulations were conducted at the arizona state university advanced computing center and the texas advanced computing center , using the flash code , a product of the doe asc / alliances funded center for astrophysical thermonuclear flashes at the university of chicago .
we formulate a pdf approach for large - eddy simulations of turbulent mixing .
less based on the pdf method have been applied to study reacting turbulent flows ( e.g. , gao & obrein 1993 , colucci et al .
1998 , jaberi et al . 1999 ,
pitsch 2006 ) .
we first derive an equation for the local fine - grained concentration pdf , and then apply the filtering procedure to obtain an exact equation for the filtered pdf at the resolution scale .
the derivation is similar to that in appendix a of pss for the equation of the concentration pdf defined in a statistical ensemble .
we start with the definition of the fine - grained concentration pdf as a delta function , @xmath575 because the concentration field in a given turbulent flow is single - valued at given position and time ( pss ) .
here @xmath26 is the sampling variable .
since @xmath576 depends on @xmath40 only through the variable @xmath577 , the time - derivative @xmath576 can be written as , @xmath578 similarly , the spatial gradient of @xmath579 is given by @xmath580 using eqs .
( [ pdftimederivative ] ) and ( [ pdfgradient ] ) and the advection - diffusion equation ( [ advection ] ) , we have @xmath581 , \label{finegraineq}\ ] ] where we used the fact that , except @xmath576 , all the quantities on the r.h.s .
are independent of @xmath26 . combining eq .
( [ finegraineq ] ) with the continuity equation , we obtain @xmath582 , \label{rhofinegraineq}\ ] ] which was also derived in appendix a of pss . the diffusivity term in eq .
( [ rhofinegraineq ] ) can be rewritten as , @xmath583 = \frac{\partial}{\partial x_i } \left(\rho \gamma \frac{\partial \phi}{\partial x_i } \right)- \frac{\partial^2}{\partial z^2 } \left(\rho \phi \gamma \left(\frac{\partial
c}{\partial x_i}\right)^2 \right ) , \label{diffusivity}\ ] ] where the first term on the r.h.s .
is a spatial diffusion of the local pdf .
note that eq .
( [ ensemblediffusivityterm ] ) in 2.2 is derived by taking the ensemble average of this equation and using the definition and properties of the conditional ensemble average ( see appendix a of pss ) .
we next apply a filtering procedure to eq .
( [ finegraineq ] ) .
a convolution of @xmath584 with the filter function , @xmath585 , gives a filtered pdf , @xmath586 , characterizing the concentration fluctuations within regions of the filter size ( or resolution scale ) .
for compressible turbulence , we define a filtered pdf with density weighting ( jaberi et al . 1999 ) , @xmath587 which is a specific example of eq .
( [ favrefilter ] ) . using eqs .
( [ filteredvariables ] ) and ( [ densweightedpdf ] ) in eq .
( [ rhofinegraineq ] ) , we obtain the filtered pdf equation , @xmath588 to write the equation in a more convenient form , we introduce conditional filtering based on the local concentration values . for any variable @xmath589 in the flow , we define a conditionally filtered quantity , @xmath590 since the fine - grained pdf , @xmath579 , is a delta function , the definition is straightforward to understand : the conditionally filtered variable is the average over the set of points within a filter size satisfying @xmath591 .
it is analogous to the conditional average defined in the context of a statistical ensemble ( see 2 & pss ) .
we further introduce a density - weighted conditional filtering , @xmath592}_{\rho } = \frac { \overline{\rho a|c = z } } { \overline{\rho|c = z } } = \frac { \overline { \rho \phi a } } { \overline { \rho \phi } } , \label{conditionalmass}\ ] ] where the last step follows from eq . ( [ conditionalvolume ] ) .
this definition is similar to eq .
( [ densweightconditionalave ] ) for the density - weighted conditional average over an ensemble .
combining eqs .
( [ densweightedpdf ] ) and ( [ conditionalmass ] ) , we have @xmath593}_\rho$ ] . applying this relation to the last three terms in eqs .
( [ rhofiltered ] ) , we obtain , @xmath594}_{\rho } \right ) = - \frac{\partial}{\partial z } \left ( \overline{\rho } \hspace{0.5 mm }
\widetilde{\phi } \hspace{0.5 mm } \overline { \left[\frac { 1}{\rho } \frac{\partial } { \partial x_i } \left(\rho \gamma \frac{\partial c } { \partial x_i } \right ) \left\vert \vphantom{\frac{1}{1 } } \right .
c = z \right]}_{\rho } \right ) - \frac{\partial}{\partial z } \left ( \overline{\rho } \hspace{0.5mm}\widetilde{\phi } \hspace{0.5 mm } \overline { [ s|c = z ] } _ { \rho } \right ) , \label{rhofilteredeq}\ ] ] which is equivalent to eq .
( 22 ) in jaberi et al .
( 1999 ) . using eq .
( [ diffusivity ] ) for the diffusivity term gives , @xmath595}_{\rho } \right ) = \frac{\partial}{\partial x_i } \left ( \overline{\rho \gamma \frac{\partial \phi}{\partial x_i } } \right ) - \frac{\partial^2}{\partial z^2 } \left(\overline{\rho } \hspace{0.5 mm } \widetilde{\phi } \overline{\left[\gamma \left(\frac{\partial c}{\partial x_i } \right)^2 \left\vert\vphantom{\frac{1}{1}}\right .
c = z \right]}_\rho \right ) } \\ \hspace{5.5 cm } { \displaystyle - \frac{\partial}{\partial z } \left ( \overline{\rho } \hspace{0.5mm}\widetilde{\phi } \hspace{0.5 mm } \overline { [ s|c = z ] } _ { \rho } \right)}. \end{array } \label{rhofilteredeqdiffusivity}\ ] ] note that eq .
( [ rhofilteredeq ] ) becomes identical to eq .
( [ pdfeq ] ) for the ensemble - defined pdf , if we replace @xmath596 , @xmath505 , and @xmath597}_{\rho}$ ] by @xmath44 , @xmath508 , and @xmath598 , respectively .
the equivalence between the filtered pdf and the ensemble - defined pdf has been discussed in 2 , based on the ergodic theorem and the assumption that statistical homogeneity is restored at the filter scale .
similar to the case of the ensemble pdf equation , the advection and diffusivity terms in eqs .
( [ rhofilteredeq ] ) and ( [ rhofilteredeqdiffusivity ] ) need to be modeled due to the closure problem . in 7
, we adopted an eddy - diffusivity assumption for the advection term , and the self - convolution models discussed in 3 may be applied to approximate the diffusivity term . using the convolution pdf models and the results of our simulations , a subgrid model is constructed in 7 to investigate the pollution of primordial gas in early galaxies . | we conduct a comprehensive theoretical and numerical investigation of the pollution of pristine gas in turbulent flows , designed to provide useful new tools for modeling the evolution of the first generation of stars .
the properties of such population iii ( pop iii ) stars are thought to be very different than those of later stellar generations , because cooling is dramatically different in gas with a metallicity below a critical value @xmath0 which lies between @xmath1 and @xmath2 .
the critical value is much smaller than the typical overall average metallicity , @xmath3 , and therefore the mixing efficiency of the pristine gas in the interstellar medium plays a crucial role in determining the transition from pop iii to normal star formation .
the small critical value , @xmath4 , corresponds to the far left tail of the probability distribution function ( pdf ) of the metal abundance . based on closure models for the pdf formulation of turbulent mixing , we derive evolution equations for the fraction of gas , @xmath5 , lying below @xmath0 in statistically - homogeneous compressible turbulence .
our simulation data shows that the evolution of the pristine fraction @xmath5 can be well approximated by a generalized `` self - convolution '' model , which predicts that @xmath6 where @xmath7 is a measure of the locality of the mixing or pdf convolution events and the convolution timescale @xmath8 is determined by the rate at which turbulence stretches the pollutants .
carrying out a suite of numerical simulations with turbulent mach numbers ranging from @xmath9 to @xmath10 we are able to provide accurate fits to @xmath7 and @xmath8 as a function of @xmath11 , @xmath12 and the length scale , @xmath13 , at which pollutants are added to the flow . for pristine fractions above @xmath14
mixing occurs only in the regions surrounding blobs of pollutants , such that @xmath15 . for smaller values of @xmath16 @xmath7 is larger as the mixing process becomes more global .
we show how these results can be used to construct one - zone models for the evolution of pop iii stars in a single high - redshift galaxy , as well as subgrid models for tracking the evolution of the first stars in large cosmological numerical simulations . |
accretion disks and mass - loss processes with the presence of magnetic fields govern the formation of low - mass stars ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
these mass - loss processes are non - steady , presenting pulsed events that are probably related to episodic increases in the accretion rates ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . in the case of high - mass protostars , there is increasing evidence that the associated outflows are also non - steady .
this is seen , in particular , with the detection through very long baseline interferometry ( vlbi ) h@xmath5o maser observations of short - lived ( tens of years ) , episodic outflow events , which are interpreted as due to variability in the accretion processes as in the case of low - mass protostars ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) .
cccrlrcrr & & & & & + & & & & & & & & & + source & & & & & & & & + vla 3 & 6.461 & 55.00 & 139@xmath73 & [email protected]@[email protected] & @xmath9@xmath712 & 0.25 & @xmath10 & @xmath11 + vla 2 & 6.321 & 55.95 & 12@xmath71 & @xmath120.3@xmath13 & & 0.023 & @xmath14 & @xmath15 + mm 1 & 6.292 & 57.26 & 24@xmath72 & [email protected]@[email protected] & @xmath16@xmath710 & 0.090.29 & ( 39)@xmath17 & ( 14)@xmath18 + vla 1 & 6.084 & 58.06 & 22@xmath72 & [email protected]@[email protected] & 35@xmath712 & 0.080.26 & ( 310)@xmath17 & ( 14)@xmath18 + + _ @xmath19 _ obtained by fitting a gaussian with the casa imfit task to each source in the image from fig .
[ fig - cont]b .
the major and minor axis are the full width at half maximum values of the deconvolved size . + _
@xmath20 _ mass estimated from the dust emission assuming : optically thin emission , a gas to dust ratio of 100 , a dust opacity per unit of dust mass of 1.1 @xmath21/g for vla 23 and 0.9 @xmath22 g for mm 1 and vla 1 , which are the computed values by @xcite at the observed wavelength and for dust particles with thin and thick ices , respectively .
see section 3.1 for the value of temperature used for each source . +
_ @xmath13 _ the source appears to be unresolved .
the gaussian fit yields an uncertainty for the major axis of @xmath23 arcsec .
we adopt an upper limit which is three times this uncertainty ( i.e. , at 3-@xmath24 level ) .
+ very recently , we extended these vlbi water maser studies towards the cluster of intermediate - mass young stellar objects ( ysos ) close to the optically visible herbig be star lkh@xmath0234 in the ngc 7129 star - forming region @xcite .
the ysos in this cluster ( including the star lkh@xmath0234 ) are grouped within a radius of @xmath15 arcsec ( @xmath14500 au at a distance of 0.9 kpc ; see kato et al . 2011 and references therein ) .
five of these ysos show radio continuum emission : lkh@xmath0 234 , vla 1 , vla 2 , vla 3a , and vla 3b @xcite , all of them , excepting vla 1 , with mid - infrared counterparts @xcite . from the absence of near - infrared emission in the j , h , and k - bands of vla 2 and vla 3a+3b ( hereafter vla 3 ) , but being bright in the @xmath25 , @xmath26 , and mid - infrared bands , @xcite concluded that these two objects are highly embedded protostars , having also associated h@xmath5o maser emission ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* see also fig .
1 in torrelles et al .
2014 showing the spatial distribution of the ysos in the cluster ) .
our multi - epoch vlbi h@xmath5o maser observations @xcite revealed a very compact ( @xmath10.2 arcsec , @xmath1180 au ) , short - lived ( kinematic age of @xmath140 yr ) , episodic , and highly collimated water maser bipolar outflow emerging from vla 2 . these results predicted the presence of an accretion disk associated with vla 2 as well as a relatively compact bipolar molecular outflow when observed through thermal molecular lines . in this paper , we present submillimeter array ( sma ) dust continuum , and thermal molecular line observations at @xmath11 mm wavelengths carried out towards vla 2 to detect and characterise the expected disk - yso - outflow system at scales of a few hundred au ( section 2 ) .
we show the observational results in section 3 , discussing them in section 4 .
the main conclusions of this research are presented in section 5 .
the sma observations were carried out on 2014 august 14 and september 1 in the very extended and extended configurations , respectively . the receiver was tuned to cover the 214.3 - 218.3 and 226.3 - 230.3 ghz frequency ranges in the lower ( lsb ) and upper side band ( usb ) , respectively .
the phase center of the telescope was @xmath0(j2000.0)=@xmath27 and @xmath28(j2000.0)=@xmath29 .
we used the flexible sma asic correlator , which provides 48 consecutive spectral windows of 104 mhz width .
most of these spectral windows were set to have a spectral resolution of 1.6 mhz ( i.e. , 2.13 km s@xmath4at 230 ghz ) .
however , some spectral windows , which included several lines ( e.g. , @xmath30so 6@xmath315@xmath32 , so@xmath5 22@xmath33 - 22@xmath34 , ch@xmath35oh 5@xmath364@xmath37 ,
h@xmath5s ) were set to have a spectral resolution of 0.4 mhz and 0.8 mhz ( 0.53 and 1.06 km s@xmath4 , respectively ) . unfortunately these are not the brightest lines , so they are not the best choices to trace the kinematics of the gas .
the gain calibrators were qsos 1927@xmath38739 and 2009@xmath38724 .
the bandpass calibrator was 3c454.3 .
the absolute flux scale was determined from observations of mwc349a and neptune .
the flux uncertainty was estimated to be @xmath39% .
the data were calibrated using the mir packagecqi / mircook.html ] and imaged using the miriad software @xcite .
self - calibration ( phase only ) was performed using the continuum data .
the derived gain solutions were applied to the molecular line data .
the continuum maps at 1.35 mm ( 222.3 ghz ) were obtained combining the upper and lower sidebands , the two array configurations , and using robust weighting of 0 .
the resulting synthesised beam was @xmath40@xmath8@xmath41 with a position angle p.a .
the rms noise level for the continuum image was @xmath440.6 mjy beam@xmath4 .
for the molecular line data , the dust continuum emission was subtracted in the visibility space before obtaining the channel maps .
the spectral line maps presented here were obtained by using a robust weighting of 1.0 , which yielded a synthesised beam of @xmath45 ( p.a .
@xmath46@xmath43 ) for the lsb and of @xmath47 ( p.a .
@xmath46@xmath43 ) for the usb .
the figures were created using the greg package ( from the gildas software ) .
[ fig - cont]a shows the 1.35 mm continuum map of the lkh@xmath0 234 region obtained with the sma .
the emission appears extended and elongated about 5 arcsec ( 4500 au ) in the nw - se direction , engulfing all known sources but lkh@xmath0 234 .
the overall morphology of the emission agrees with previous , lower angular resolution , 1.35 mm interferometric observations ( beam @xmath11.3 arcsec ) by @xcite .
the observed elongation is also seen , but at much larger scales ( up to @xmath10.3 pc ) , by single - dish observation of the 1.35 mm continuum and co isotopologues @xcite .
our subarcsecond angular resolution 1.35 mm images show four distinct peaks of emission coinciding with sources vla 1 , vla 2 , vla 3 @xcite , and mm 1 @xcite .
mm 1 is the only source not detected at other wavelengths .
for all the mm continuum sources detected with the sma and reported in this paper there is a systematic position offset of @xmath10.2 arcsec with respect to the positions measured in the literature at other wavelengths ( e.g. , * ? ? ?
* ; * ? ? ?
this value is comparable to the absolute astrometric uncertainty in a typical sma observation . in order to correct for the offset , we have used as a reference position the radio continuum position of vla 2 ( the most compact source both at cm and mm wavelengths ) , @xmath0(j2000 ) = 21@[email protected]@xmath49 , @xmath28(j2000 ) = 66@[email protected]@xmath52 given by @xcite , assuming that the sma mm continuum peak position of vla 2 is the same as the vla 2 position measured at cm wavelengths .
this means that we had to apply a small astrometric correction of ( @xmath53 , @xmath54 ) to our sma data , resulting also in an excellent alignment for all the other sources ( see fig .
[ fig - cont ] ) .
the total flux detected in the region by the sma at 1.35 mm is 363@xmath75 mjy . to compare this value with the one measured by @xcite with the iram plateau de bure interferometer ( pdbi ) with an angular resolution of @xmath11.3 arcsec at 1.22 mm we used the dust continuum spectral index of 2.4 derived by these authors . in this way , we find that the sma recovers about 90% of the flux measured by the pdbi .
this is @xmath165% of the total flux measured with the iram 30 m bolometer @xcite .
since there is extended 1.35 mm continuum emission around the four detected sources ( vla 1 , vla 2 , vla 3 , and mm1 ) , images have also been made by discarding the visibilities for baselines shorter than 100 k@xmath55 .
this allowed us to resolve out the extended filamentary emission and to leave only the emission arising from the four sources .
[ fig - cont]b shows the resulting map ( which yielded a synthesised beam of @xmath56@xmath8@xmath57 , p.a .
@xmath58@xmath43 ) , whereas table [ t1 ] gives the positions , fluxes , and the deconvolved sizes by gaussian fitting of the four sources . in vla 3 , the brightest source ,
the emission arises from a region with a radius of @xmath1155 au , suggesting that the dust emission may arise from a circumbinary disk around vla 3a and vla 3b .
mm 1 and vla 1 appear to have not only similar fluxes but also similar sizes .
both sources are elongated , with major and minor semiaxes of @xmath1370 and @xmath1140 au respectively , but with different orientations .
vla 2 appears to be unresolved , with a radius upper limit of @xmath1135 au ( table [ t1 ] ) .
the total flux of the four compact sources is 197@xmath74 mjy .
this leaves a flux that arises from the dust ridge of 166@xmath77 mjy . the mass , volume and column densities have been computed from the 1.35 mm dust emission of the four mm continuum sources ( table [ t1 ] ) assuming a constant temperature of 126 and 148 k for vla 2 and vla 3 , respectively , which is the value derived from fitting the methanol and so@xmath5 lines .
these two lines trace the hot and dense gas at similar scales as the dust ( see tables [ t1 ] , [ t3 ] , and section [ mol ] ) . for mm 1 and vla 1
, we assume a temperature range of 3080 k. the lack of methanol emission in these two sources suggests that it is depleted in the dust grain mantles ( the methanol starts to sublimate at a temperature of @xmath180 k , * ? ? ?
adopting a temperature of 22 k , derived from ammonia observations @xcite , the mass of the dust ridge detected by the sma is @xmath17 m@xmath2 .
.molecular lines detected[t2 ] [ cols="<,^,^,>,^ " , ] + + _ @xmath19 _ frequency from the cologne database for molecular spectroscopy @xcite . +
_ @xmath20 _ identification number shown in fig [ fig - spec ] .
@xmath13 _ the complete transition name is n=2 - 1 j=3/2 - 1/2 f=5/2 - 3/2 + _
@xmath59 _ the complete transition name is n=2 - 1 j=5/2 - 3/2 , f=7/2 - 5/2 .
this line is blended with the j=3/2 - 1/2 f=5/2 - 3/2 and j=5/2 - 3/2 f=3/2 - 1/2 lines + _ @xmath60 _ the complete transition name j=47/2 - 45/2 , @xmath61=1/2,l = f , v@xmath62=1 . + oh and so@xmath5 lines , respectively .
grey dotted lines show other molecular transitions detected .
the number is the identification label in table [ t2 ] .
the conversion from flux density to main beam brightness temperature is @xmath147 k / jy at the observed frequencies and angular resolution .
, width=359 ] the sma correlator provides 4 ghz of moderate spectral resolution data ( 0.52.0 km s@xmath4 ) .
table [ t2 ] shows the list of detected molecular transitions , including information of their frequencies and the energy of the lower rotational level .
regarding their morphology , there are two types of species .
on one hand , there are several so@xmath5 and ch@xmath35oh transitions , as well as a c@xmath63h vibrational line , that have compact emission and are detected only toward vla 2 and vla 3 .
these transitions appear to have in most cases energy levels above 100 k. on the other hand , there are several molecular transitions that show extended emission .
[ fig - spec ] shows the rich spectra over the whole observed bandwidth for vla 2 and vla 3 . for comparison
this figure also shows the vla 1 spectrum .
velocity range between @xmath64 and @xmath65 km s@xmath4 for several molecular transitions that show extended emission .
contours are 0.15 , 0.3 , 0.45 , 0.6 , 0.8 , 1.0 , 1.5 , 2.0 , 2.5 , 3.0 , 3.5 , 4.0 jy km s@xmath4 .
the crosses mark the positions of the different ysos in the region.,width=264 ] fig .
[ fig - extnd ] shows the integrated emission for the five brightest lines that show extended emission .
these images show a complex chemical behaviour , where each species traces slightly different regions of the lkh@xmath0 234 molecular gas environment . in order to simplify the description , we label in fig .
[ fig - extnd ] the different molecular structures .
hereafter , we refer to the `` dust ridge '' as the molecular structure that approximately traces the dust emission , as shown in fig .
[ fig - cont]a .
the `` s ridge '' is the curved structure that extends to the south of vla 2 .
the `` lkh@xmath0 234 se ridge '' is the structure seen only in so to the south - east of lkh@xmath0 234 .
finally , the `` lkh@xmath0 234 ne ridge '' is the structure seen to the north - east of lkh@xmath0 234 .
the length of the different ridges is @xmath15 arcsec ( @xmath14500 au ) , except for the dust ridge that in so has a length of @xmath110 arcsec ( @xmath19000 au ) .
the width of the ridges is roughly @xmath12 arcsec ( @xmath11800 au ) .
below , we briefly describe the main properties of the spatial distribution of the different molecular transitions : * the so 5@xmath314@xmath32 line traces all the aforementioned structures except the lkh@xmath0 234 ne ridge .
there is copious so emission around all the sources , with the peak emission arising toward vla 2 and vla 3 . *
the h@xmath5co line is the second most extended tracer , being detected in all structures except the lkh@xmath0 234 se ridge .
as in the case of the so line , the strongest emission arises from vla 2 and vla 3 . *
the cn emission clearly traces the lkh@xmath0 234-ne ridge .
there is also fluffy emission around , but not peaking at , vla 1 and mm 1 . *
the dcn emission traces mostly the dust ridge although there is a void around vla 3 .
there is also some emission along the lkh@xmath0 234 se ridge . *
the hc@xmath35n 25 - 24 line is only detected around vla 2 and toward the lkh@xmath0 234-ne ridge . * the sio 5 - 4 line emission ( fig .
[ fig - sio ] ) appears to be detected around vla 2 and vla 3 and along the s ridge . however , there are significant differences with respect to the other tracers .
there is blue and redshifted emission forming a compact ( total length of @xmath14 arcsec or @xmath14000 au ) , low velocity ( outflow velocities of @xmath17 km s@xmath4 with respect to the vla 2 systemic velocity , @xmath1@xmath66 km s@xmath4 ; see next subsection ) , bipolar outflow in the northeast - southwest direction ( p.a.@xmath39@xmath43 ) , that is centred on vla 2 .
the northern blueshifted lobe exhibits some weak redshifted emission .
this , and the low radial velocity of the outflow suggest that the outflow axis is close to the plane of the sky ( see also section 4.1.2 ) . at @xmath12
arcsec southwest of vla 2 the redshifted emission bends to the east following the se ridge .
the radial velocity maps of the molecular gas are shown in the ( a ) and ( b ) panels of fig .
[ fig - vel ] for the h@xmath5co and so lines , respectively .
there are complex velocity gradients , but the two more interesting ones are : ( i ) a mostly smooth gradient along the dust ridge , and ( ii ) a gradient along the s ridge ( better seen in the h@xmath5co line ) .
the second order moment map of the so 5@xmath314@xmath32 line shows that , except for vla 2 and vla 3 , the typical velocity dispersion of the molecular gas is @xmath11.01.5 km s@xmath4(panel c of fig .
[ fig - vel ] ) , which implies a line width of @xmath12.43.4 km s@xmath4 .
since the spectral resolution is @xmath12.3 km s@xmath4 , the line widths are barely resolved .
the typical velocity dispersion corrected from the channel width is @xmath121.2 km s@xmath4 .
the velocity dispersion increases toward vla 2 and vla 3 ( @xmath67 km s@xmath4 ) .
co 3@xmath682@xmath69 integrated emission ( black contours ) with the redshifted ( @xmath70 from @xmath66 to @xmath71 km s@xmath4 ) and blueshifted ( @xmath70 from @xmath72 to @xmath66 km s@xmath4 ) sio 5 - 4 averaged emission ( red and blue contours ) .
the grey ellipses in the bottom part of the panel show the synthesised beam of these two lines .
the crosses mark the different ysos of the region as in fig . 1 and 3
. h@xmath5co contours are 0.17 , 0.56 , 1.0 and 2.5 jy beam@xmath4 km s@xmath4 .
sio contours are 0.03 , 0.06 , 0.09 and 0.12 jy beam@xmath4 . _
upper panel _ : close - up of the proper motions of the h@xmath5o masers associated with vla 2 ( from * ? ? ?
] lcc source & & + & line & & + + vla 2(nw2 ) & [email protected]@[email protected] & @xmath73@xmath723 + vla 3(nw1 ) & [email protected]@[email protected] & @xmath74@xmath728 + + vla 2(nw2 ) & [email protected]@[email protected] & @xmath75@xmath770 + vla 3(nw1 ) & [email protected]@[email protected] & @xmath76@xmath744 + + _ @xmath19_deconvolved sizes obtained by fitting a gaussian to each source using the casa imfit task .
we use the integrated emission images from fig .
[ fig - vel - hc ] .
+ the so@xmath5 and ch@xmath35oh lines ( table [ t2 ] ) are only detected toward vla 2 and vla 3 .
most of these lines can only be effectively excited if the gas is very dense and hot ( because of their critical densities and energy levels ; e.g. , table [ t2 ] ) .
the so@xmath5 and ch@xmath35oh lines were used to estimate the temperature of vla 2 and vla 3 .
we performed a rotational temperature diagram analysis , where @xmath77 is the integrated intensity of a line transition , @xmath78 the so - called line strength of the transition , @xmath79 the electric dipole moment for the molecule , @xmath80 the energy level of the upper transition ( in k ) and t the gas temperature ( e.g. , * ? ? ? * ) ] using the transitions 1 , 2 , 3 , and 5 of so@xmath5 , and the transitions 7 to 15 of ch@xmath35oh ( table [ t2 ] ) .
the transition 4 of so@xmath5 was not used because of its location at the edge of the bandpass .
the energies of the transition levels covered a wide range between 30 k and 550 k. the fit of the integrated line intensity of the transitions over @xmath81 versus the energy level gave us consistent values of the slope for both molecules , for vla 2 and vla 3 .
the value of the slope is inversely proportional to the rotational temperature , assumed to be the same for all the transitions of the same species .
finally , we obtained rotational temperature values of @xmath82 k and @xmath83 k for vla 2 and vla 3 , respectively ( this is the error - weighted average of the so@xmath5 and ch@xmath35oh temperatures , see fig .
[ fig - diagrot ] ) fig .
[ fig - vel - hc ] shows the integrated emission as well as the velocity map of the so@xmath5 and ch@xmath35oh strongest lines .
the molecular emission is quite compact , although around vla 3 it is slightly elongated along the northwest southeast direction , with the molecular peaks of both species appearing closer to the continuum centimetre source vla 3a than to vla 3b . in order to measure their sizes
, we fitted two gaussians to the integrated emission shown in the aforementioned figure .
this fit toward vla 3 shows that the circumstellar molecular emission has a size and orientation similar to that of the dust emission ( tables [ t1 ] and [ t3 ] ) .
the ch@xmath35oh and so@xmath5 emissions trace gas within @xmath1300 and @xmath1160 au ( semi - major axes ) , respectively . towards vla 2 the size of the emission is similar for both lines ( a semi - major axis of @xmath1230 au ) .
the so@xmath5 and ch@xmath35oh lines show a clear velocity gradient , with a total velocity change of @xmath112 km s@xmath4 , along the major axis of vla 3 ( pa=@xmath8422@xmath43 ; table [ t1 ] and fig .
[ fig - vel - hc ] ) , which is similar to the direction of the velocity gradient seen at larger scale through other molecular lines ( see fig .
[ fig - vel ] ) .
[ fig - pv ] shows position - velocity plots along the axis that connects vla 2 and vla 3 ( which is also the dust ridge major axis ) for lines typical of warm / hot cores ( ch@xmath35oh , so@xmath5 , h@xmath5co ) overlapped with that from the sio 5 - 4 line .
there are several interesting behaviours in these position
velocity plots : ( 1 ) the center velocity of the molecular emission of the two sources is different , @xmath1@xmath66 km s@xmath4 and @xmath1@xmath85 km s@xmath4 for vla 2 and vla 3 , respectively .
we note that the @xmath8411.5 km s@xmath4 value for vla 2 is similar to the mean radial velocity of the h@xmath5o masers associated with vla 2 , @xmath8411.6 km s@xmath4 @xcite .
( 2 ) there is a clear velocity gradient along the major axis of vla 3 : northwest ( blueshifted ) to southeast ( redshifted ) . in these plots ,
the emission at the 2-@xmath24 contour level spans @xmath120 km s@xmath4 , although the most blueshifted emission is not detected in the ch@xmath35oh line . toward vla 2 ,
the molecular emission at the 2-@xmath24 contour level spans over a smaller velocity interval , @xmath112 km s@xmath4 .
however , the emission of the h@xmath5co and ch@xmath35oh lines shows only a marginal velocity gradient ( in the same direction as in vla 3 ) .
( 3 ) there is molecular gas between the two sources ( as also seen in fig .
[ fig - vel - hc ] ) , with the most blueshifted gas in vla 3 connected in velocity with a bridge of gas to vla 2 .
( 4 ) sio is detected toward vla 2 and toward the bridge of gas that connects to vla 3 .
co 3@xmath682@xmath69 and so 5@xmath314@xmath32 lines , respectively .
the colour wedge shows the velocity scale in km s@xmath4 .
_ panel ( c ) _ : second order moment ( velocity dispersion ) overlapped with the integrated emission of the so 5@xmath314@xmath32 line .
the wedge indicates the velocity dispersion range in km s@xmath4 .
the crosses indicate the positions of the ysos as in figs . 1 and 3 . , width=302 ]
in this section we discuss the properties of these two protostars and the possible interaction between their circumstellar material .
the dense and warm circumstellar material at scales of 200 au ( as traced by the 1.35 mm dust continuum emission ) is about ten times more massive in vla 3 than in vla 2 ( @xmath10.25 and @xmath10.023 m@xmath86 , respectively ; table [ t1 ] ) .
vla 3 has a bolometric luminosity of @xmath1700 l@xmath86 @xcite .
if all the luminosity would come from a single main sequence star , then the stellar mass would be @xmath16 m@xmath86 .
however , this is a binary ( vla 3a+vla 3b ) embedded yso system , so this mass value is probably an upper limit for each of the two single components ( in the case of twin components the mass of each component would be @xmath15 m@xmath86 ) , even more so considering that part of the luminosity arises from the accretion .
the stellar mass for vla 2 is even more uncertain since the bolometric luminosity is unknown . yet
, its 11.8 @xmath79 m flux is @xmath12.5 weaker than that of vla 3 @xcite , which suggests that vla 2 is also an intermediate mass star .
in fact , from the observed centimetre continuum emission of vla 2 , @xcite infer a bolometric luminosity of @xmath150 l@xmath86 for this object which can be provided by a protostar of a few solar masses .
an educated guess of the mass can be also done by comparing the velocity range of the emission for vla 2 and vla 3 : at 2-@xmath24 level the emission spreads out @xmath120 and 12 km s@xmath4 for vla 3 and vla 2 , respectively ( see section 3.2.2 ) .
if the kinematics is dominated by keplerian motions ( although with the present data we can not discard the possibility that the motions around vla 3 are just due to an overlapping of unbound motions of two independent sources , vla 3a and vla 3b ) , then the mass ratio between the two sources would be @xmath87 , where the subindices 2 and 3 refer to vla 2 and vla 3 . assuming that the radius of emission , @xmath88 , is approximately the same for the two sources ( see section [ t3 ] ) , then the vla 2 mass is roughly 40% smaller than the vla 3 , @xmath12 - 4 m@xmath89 .
oh . open and filled circles indicate the observed data for the ch@xmath35oh and so@xmath5 transitions , respectively .
the error bars include the measurement uncertainty and the calibration uncertainty ( @xmath615% ) .
the solid ( ch@xmath35oh ) and dashed ( so@xmath5 ) lines show the best fit obtained by assuming optically thin emission ( i.e. , using the standard population diagram analysis ) . for vla 3
the best fit is for a temperature of 148@xmath715 k and 144@xmath785 k for the ch@xmath35oh and so@xmath5 data , respectively . for vla 2
the best fit is 126@xmath713 k ( ch@xmath35oh ) and 127@xmath728 k ( so@xmath5 ) . ]
the sio 5 - 4 emission appears to originate from two different types of regions .
first , there is a low radial velocity ( @xmath17 km s@xmath4 ) bipolar outflow in the northeast - southwest direction arising from vla 2 .
[ fig - sio ] shows that the sio is relatively compact ( @xmath14 arcsec , @xmath14000 au ) , specially in the blueshifted lobe and , more importantly , it is contained within the dense molecular gas structure traced by the h@xmath5co @xmath90-@xmath91 line .
as mentioned in section 3.2.1 , the relatively low radial velocities , with the northern blueshifted lobe also having weak redshifted emission , suggest that the outflow is close to the plane of the sky
. there is also redhsifted sio emission @xmath14 arcsec south of vla 3 .
if this emission is associated with the sio bipolar outflow , then it would imply that the redshifted lobe has suffered a deflection of @xmath150@xmath43 .
we identify the bipolar molecular outflow seen in sio as the `` extended '' outflow counterpart of the very compact ( @xmath10.2 arcsec ,
@xmath1180 au ) , short - lived ( kinematic age @xmath140 yr ) , episodic , bipolar outflow observed through h@xmath5o maser proper motions also expanding close to the plane of the sky in the northeast - southwest direction with velocities of @xmath120 km s@xmath4 ( * ? ? ?
* see fig .
[ fig - sio ] ) .
in fact , from the proper motions and radial velocities of the h@xmath5o masers reported by torrelles et al .
( 2014 ) we estimate that the outflow axis has an inclination angle of @xmath115@xmath50 with respect to the plane of the sky . from the observed spatial extension and radial velocity of the sio bipolar outflow
, we estimate a kinematic age of @xmath1350 yr for this outflow ( after correcting for the inclination angle of @xmath115@xmath50 ) .
we therefore propose that this more extended and `` older '' outflow seen in sio has been created by the accumulation of the ejection of several episodic collimated events of material as the one observed through the h@xmath5o masers .
we note , however , that the position angle of the sio bipolar outflow is @xmath1206@xmath50 , while the position angle of the very compact h@xmath5o maser outflow is @xmath1247@xmath50 @xcite .
it is possible that this difference in position angles of @xmath140@xmath50 between the extended sio outflow and the compact h@xmath5o maser outflow is due to several short - lived episodic ejection events , which would have slightly different position angles at their origin .
co 3@xmath682@xmath69 ( top panel ) , so@xmath5 16@xmath9216@xmath93 ( central panel ) and ch@xmath35oh 8@xmath947@xmath95 ( bottom panel ) overlapped with the contour maps of their integrated emission toward vla 2 and vla 3 .
the wedge on the top indicates the velocity range ( in km s@xmath4 ) of the three panels .
, width=302 ] the other type of emission traced by sio arises at the systemic velocity of the core ( see fig .
[ fig - pv ] ) .
the sio emission is a signpost of strong shocks , since it can be generated after sputtering in shocks of the si locked in the dust grains @xcite .
however , in other star - forming regions sio is detected at velocities close to the systemic velocity of the cloud , with relatively narrow line widths and with no obvious association with the protostellar outflow activity @xcite .
it has been proposed that , in these cases , the sio can be produced in low velocity shocks ( @xmath96 km s@xmath4 ) if about 10% of the si is already in the gas phase or in the icy mantles @xcite .
although this can explain the sio emission at velocities close to the systemic velocity of the core , it is unclear what would be the origin of the initial si abundance outside of the dust grains .
sio emission at velocities close to the systemic velocity of the cloud is detected both in vla 2 and in the bridge of molecular emission connecting vla 2 and vla 3 ( hereafter `` vla 32 bridge '' ; see fig .
[ fig - pv ] ) .
thus , we think that this sio emission is probably due to the interaction ( resulting in low velocity shocks ) , either between the vla 2 and vla 3 circumstellar gas , or between the vla 2 circumstellar gas and the dust ridge gas that surrounds the two circumstellar structures .
this latter case is less probable , since the vla 32 bridge shows emission from hot core - like molecular lines that are also present in vla 2 and vla 3 , but not in the dust ridge .
therefore , we favour a kinematical scenario with a stream of gas between vla 3 and vla 2 , producing shocks and releasing sio . however , with the present data we can not discern the direction of the stream of gas ( either from vla 3 to vla 2 or from vla 2 to vla 3 ) , particularly considering that vla 3 is a binary source where we are not able to resolve the kinematics and spatial distribution of the gas contents of each component ( 3a and 3b )
nevertheless , if the stream of gas is from vla 3 to vla 2 , this scenario would increase the accretion rate onto vla 2 , and could explain why this protostar has a high outflow activity .
co ( bottom panel ) , so@xmath5 ( middle panel ) , and ch@xmath35oh ( top panel ) lines ( solid contours ) overlapped with the sio 5 - 4 line ( red contours ) . the dotted line is shown to approximately mark the velocity gradient in vla 3 .
the zero offset position corresponds to the intermediate position between vla 3 and vla 2 ( see section 4.1.3 ) .
the contour levels are 2 , 4 , 7 , 10 and 13 times the _ rms _ noise of the maps , 23 , 20 and 27 mjy beam@xmath4 for the ch@xmath35oh , so@xmath5 and h@xmath5co lines , respectively .
for the sio line the contours are 2 , 3 and 5 times the _ rms _ noise of the map , 22 mjy beam@xmath4.,width=302 ] here we discuss the apparently organised star formation process taking place in the dust ridge . 234 ( same as in fig .
[ fig - cont ] ) , superposed with the direction of the different outflow signpost detected at different wavelengths ( see sect .
[ s - outflow ] ) .
blue and red arrows indicate blue and redshifted gas , whereas grey arrows show outflow direction without velocity information .
the crosses mark the different ysos in the region as in fig . 1 . , width=283 ]
the outflow activity detected in this region is diverse , but all the signposts indicate that there is a preferential direction of the various outflow axes @xcite . besides the aforementioned h@xmath5o maser and sio outflows associated with vla 2 , vla 3a and vla 3b are radio jets with position angles of @xmath155@xmath50 and @xmath157@xmath50 , respectively @xcite .
vla 1 is also a radio jet with a position angle of @xmath145@xmath50 @xcite .
there is a large scale ( @xmath11 pc ) ne - sw co bipolar outflow , with the ne red lobe being more extended and brighter than the sw blue lobe @xcite .
interestingly , and in spite of having a similar orientation as the sio bipolar outflow ( this paper ) , the co outflow has an inverted velocity pattern , suggesting a different powering source .
there are two additional jets southwest of the dust ridge .
one detected in [ sii ] extending @xmath10.2 pc with pa@xmath44252@xmath50 @xcite .
the other one , in near - ir h@xmath5 emission , extending @xmath10.05 pc with pa@xmath44226@xmath50 @xcite .
@xcite suggest that the [ sii ] jet and the co ne red lobe are powered by vla 3 , whereas the co sw blue lobe and the molecular hydrogen jet are powered by mm 1 .
recently , @xcite have detected a very high velocity blueshifted [ feii ] knot , a few arcseconds south of vla 3 , which appears to be associated with the vla 3b radio jet .
finally , @xcite detect a chain of hh objects , hh 815 - 822 , centred around the lkh@xmath0 234 region , with an outflow axis at pa@xmath9760@xmath50/240@xmath50 and extending @xmath16 pc .
however , the driving source of this hh system is unclear .
[ fig - cont - out ] shows the dust emission traced by our sma observations with the directions of the outflows identified in the region .
it is clear that the overall trend of the activity of the outflows occurs in a direction that is approximately perpendicular to the dust ridge .
we think that a magnetic field perpendicular to the dust ridge could produce the general alignment of the different outflows in the region .
although in general there is not a clear correlation between the magnetic field and the outflow direction at core scales , @xmath98 au @xcite , there are some cases where they are aligned @xcite .
this can be tested with dust polarisation measurements along the dust ridge .
the most evolved source , lkh@xmath0 234 , is located at the south - eastern edge of the dust ridge , whereas the two apparently youngest sources , mm 1 and vla 1 , are located at the opposite side of the dust ridge .
our estimate of these sources being the youngest is based on two main facts .
first , they are not detected at mid - ir or shorter wavelengths .
second , the circumstellar gas is as dense and massive as in vla 2 and vla 3 , but it is significantly colder ( there is no indication of hot core tracers as in vla 2 and vla 3 ) .
therefore , these properties suggest an overall evolutionary sequence along the dust ridge , where star formation started in the southeastern part of the dust ridge and has proceeded with time toward the northwest .
we speculate that a compression wave has continued toward the northwest extreme , once the star formation started in lkh@xmath0 234 .
this , together with the fact that the outflow activity occurs in a direction perpendicular to the dust ridge suggests that the present star formation activity in the region around lkh@xmath0 234 is taking place in an organised fashion .
we report sma 1.35 mm continuum and spectral line observations ( angular resolution @xmath10.5 - 0.8 arcsec ) towards the intermediate - mass star - forming region around the optically visible star lkh@xmath0 234 .
we have detected a dust ridge of @xmath15 arcsec ( @xmath14500 au ) size containing the cluster of ysos vla 1 , vla 2 , vla 3a , vla 3b , and mm1 .
the mass of the dust ridge is @xmath17 m@xmath89 while the circumstellar masses around these ysos are in the range of @xmath10.020.3 m@xmath89 .
we find an evolutionary star formation sequence along the dust ridge , with the most evolved objects located at the southeastern edge of the ridge and the youngest ones located at the northeastern edge .
the outflow activity of the different ysos in the region occurs approximately perpendicular to the dust ridge .
this overall trend of orientation of the outflows could be explained by the presence of a
large - scale " magnetic field perpendicular to the dust ridge .
our sma observations reveal towards vla 2 , a compact ( @xmath14 arcsec , @xmath14000 au ) , sio bipolar outflow , moving close to the plane of the sky in the northeast - southwest direction .
the kinematic age of the sio outflow is @xmath1350 yr .
we conclude that this outflow is the large - scale " counterpart of the much more compact ( @xmath10.2 arcsec , @xmath1180 au ) , short - lived ( @xmath140 yr ) , episodic , bipolar h@xmath5o maser outflow previously reported with vlbi observations towards vla 2 . we propose that the bipolar outflow seen in sio has been created by the accumulation of material through repetitive ejections of collimated gas as the one observed through the h@xmath5o masers .
there is a bridge of molecular gas connecting vla 3 and vla 2 having sio emission at ambient velocities .
we discuss the possibility that this sio ambient emission in the vla 32 molecular bridge is due to an interaction of the gas transfer between vla 3 and vla 2 .
if the gas transfer occurs from vla 3 to vla 2 it could increase the accretion rate onto vla 2 , explaining the observed high outflow activity in this source .
ga , re , jmg , jfg , and jmt acknowledge support from mineco ( spain ) aya2014 - 57369-c3 grant ( co - funded with feder funds ) .
jmg also acknowledges support from the mecd prx15/00435 grant ( spain ) and from the si cgps award , `` magnetic fields and massive star formation '' .
re also acknowledges mdm-2014 - 0369 of iccub ( unidad de excelencia ` mara de maeztu ' ) .
sc acknowledges the support of dgapa , unam , and conacyt ( mxico ) .
aikawa y. , wakelam v. , garrod r. t. , herbst e. , 2008 , apj , 674 , 984 audard m. et al .
, 2014 , in protostars and planets vi , henrik beuther , ralf s. klessen , cornelis p. dullemond , and thomas henning ( eds . ) , university of arizona press , tucson , 914 pp .
, p. 387 cabrit s. , lagage p .- o . , mccaughrean m. , olofsson g. , 1997 , a&a , 321 , 523 carrasco - gonzlez c. , et al . , 2015 ,
sci , 348 , 114 edwards s. , snell r. l. , 1983 , apj , 270 , 605 fuente a. , neri , r. , martn - pintado j. , bachiller r. , rodrguez - franco a. , palla f. , 2001 , a&a , 366 , 873 fuente a. , rizzo j. r. , caselli p. , bachiller r. , henkel c. , 2005 , a&a , 433 , 535 frank a. et al . , 2014 , in protostars and planets vi , henrik beuther , ralf s. klessen , cornelis p. dullemond , and thomas henning ( eds . ) , university of arizona press , tucson , 914 pp .
, p. 451 girart j. m. , viti s. , williams d. a. , estalella r. , ho p. t. p. , 2002 ,
a&a , 388 , 1004 girart j. m. , rao r. , marrone d. p. , 2006
, sci , 313 , 812 hull c. l. h. , et al . , 2013 , apj , 768 , 159 hull c. l. h. , et al . , 2014 , apjs , 213 , 13 jimnez - serra i. , caselli p. , tan j. c. , hernandez a. k. , fontani f. , butler m. j. , van loo s. , 2010 , mnras , 406 , 187 kato e. , fukagawa m. , perrin m. d. , shibai h. , itoh y. , ootsubo t. , 2011 , pasj , 63 , 849 machida m. n. , inutsuka s .- i .
, matsumoto t. , 2008 , apj , 676 , 1088 marvel k. b. , 2005 , aj , 130 , 2732 mcgroarty f. , ray t. p. , bally j. , 2004 , a&a , 415 , 189 mller h. s. p. , thorwirth s. , roth d. a. , winnewisser g. , 2001 , a&a , 370 , l49 nguyen - luong q. , et al . , 2013 , apj , 775 , 88 oh h. , et al . , 2016 , apj , 817 , 148 ossenkopf v. , henning t. , 1994 , a&a , 291 , 943 pech g. , loinard l. , chandler c. j. , rodrguez l. f. , dalessio p. , brogan c. l. , wilner d. j. , ho p. t. p. , 2010 ,
apj , 712 , 1403 qiu k. , zhang q. , menten k. m. , liu h. b. , tang y .- w .
, girart j. m. , 2014 , apj , 794 , l18 rao r. , girart j. m. , lai s .- p . , marrone d. p. , 2014 ,
apj , 780 , l6 ray t. p. , poetzel r. , solf j. , mundt r. , 1990 , apj , 357 , l45 reipurth b. , bally j. , 2001 , ara&a , 39 , 403 sanna a. , reid m. j. , carrasco - gonzlez c. , menten k. m. , brunthaler a. , moscadelli l. , rygl k. l. j. , 2012 , apj , 745 , 191 sault r. j. , teuben p. j. , wright m. c. h. 1995 , astronomical data analysis software and systems iv , 77 , 433 schilke p. , walmsley c. m. , pineau des forets g. , flower d. r. , 1997 , a&a , 321 , 293 tofani g. , felli m. , taylor g. b. , hunter t. r. , 1995 , a&as , 112 , 299 torrelles j. m. et al .
, 2001b , nature , 411 , 277 torrelles j. m. , curiel , s. , estalella r. , anglada g. , gmez j. f. , cant j. , patel n. a. , trinidad m. a. , girart j. m. , carrasco - gonzlez c. , rodrguez l. f. 2014 , mnras , 442 , 148 trinidad m. a. et al .
, 2013 , mnras , 430 , 1309 trinidad m. a. , curiel , s. , torrelles j. m. , rodrguez l. f. , cant j. , gmez j. f. , patel n. , ho p. t. p. , 2004 ,
apj , 613 , 416 umemoto t. , imai h. , furuya r. s. , kitamura y. , kawabe r. , 2002 , in iau symp .
206 , cosmic masers : from proto - stars to black holes , ed .
v. migenes & m. j. reid ( san francisco : asp ) , 76 zhang q. , et al . , 2014 ,
apj , 792 , 116 | we present submillimeter array ( sma ) 1.35 mm subarcsecond angular resolution observations toward the lkh@xmath0 234 intermediate - mass star - forming region .
the dust emission arises from a filamentary structure of @xmath15 arcsec ( @xmath14500 au ) enclosing vla 1 - 3 and mm 1 , perpendicular to the different outflows detected in the region .
the most evolved objects are located at the southeastern edge of the dust filamentary structure and the youngest ones at the northeastern edge .
the circumstellar structures around vla 1 , vla 3 , and mm 1 have radii between @xmath1200 and @xmath1375 au and masses in the @xmath10.080.3 m@xmath2 range .
the 1.35 mm emission of vla 2 arises from an unresolved ( r@xmath3 au ) circumstellar disk with a mass of @xmath10.02 m@xmath2 .
this source is powering a compact ( @xmath14000 au ) , low radial velocity ( @xmath17 km s@xmath4 ) sio bipolar outflow , close to the plane of the sky .
we conclude that this outflow is the large - scale " counterpart of the short - lived , episodic , bipolar outflow observed through h@xmath5o masers at much smaller scales ( @xmath6180 au ) , and that has been created by the accumulation of the ejection of several episodic collimated events of material .
the circumstellar gas around vla 2 and vla 3 is hot ( @xmath1130 k ) and exhibits velocity gradients that could trace rotation .
there is a bridge of warm and dense molecular gas connecting vla 2 and vla 3 .
we discuss the possibility that this bridge could trace a stream of gas between vla 3 and vla 2 , increasing the accretion rate onto vla 2 to explain why this source has an important outflow activity .
[ firstpage ] masers stars : formation ism : individual objects : lkh@xmath0 234-vla 2 ism : jets and outflows
ism : molecules |
the observed similarities in the timing properties of different black hole ( bh ) systems suggest that the same physical mechanism is at work in sources spanning a wide range of masses ( e.g. uttley et al 2005 , mchardy et al 2006 ) .
this observational fact represents an important achievement in the context of the theory of unification of bh accretion . a breakthrough in this respect
was the discovery of the mass - scaling law that regulates the characteristic time - scales of variability , identified with the high frequency break in the power spectral density function ( psd ) .
this relation holds for objects of widely different size ( i.e. over about eight order of magnitudes in mass , from bh x - ray binaries , bhxb , up to active galactic nuclei , agn , mchardy et al 2006 , koerding et al 2007 ) and is in agreement with expectations from standard accretion disk models , whereby all the characteristic time - scales depend linearly on the bh mass , @xmath3 ( e.g. shakura & sunyaev 1973 , treves et al 1988 ) . +
another fundamental analogy emerges from the comparison of time lags between x - ray energy bands .
hard / positive lags ( i.e. hard x - ray variations lagging soft x - ray variations ) are generally observed in both bhxbs and agn , and can be interpreted in terms of propagation of mass accretion rate fluctuations in the disc ( kotov et al 2001 and arvalo & uttley 2006 ) .
those lags are usually detected at relatively low fequencies ( i.e. below the psd high frequency break ) .
+ on the other hand , a relatively new and interesting perspective comes from the study of high frequency , soft / negative lags ( i.e. soft x - ray band variations lagging hard x - ray band variations ) .
the first significant detection of a soft / negative lag in agn light curves was reported in the narrow line seyfert 1 ( nlsy1 ) galaxy 1h0707 - 495 ( fabian et al 2009 , zoghbi et al 2010 ) , and interpreted as the signature of relativistic reflection which `` reverberates '' in response to continuum changes after a time equal to the light - crossing time from the source to the reflecting region .
a different interpretation has been proposed in terms of a complex system of scatterers / absorbers located close to the line of sight , but at hundreds of gravitational radii , @xmath4 , from the central source ( miller et al 2010 ) , thus requiring us to have a special line of sight to the source .
+ subsequently , soft lags have been observed in several other sources ( e.g. tripathi et al 2011 , emmanoulopoulos et al 2011 , zoghbi & fabian 2011 ) .
one of the largest - mass sources ( @xmath5 ) showing a soft / negative lag is pg 1211 + 143 ( de marco et al 2011 ) , whose lag spectrum appears `` shifted '' by about one order of magnitude ( with the lag frequency , @xmath6 , and amplitude , @xmath7 , found at @xmath8hz and @xmath9 s , respectively ) with respect to those measured in other low mass sources ( i.e. @xmath10hz and @xmath11 s , emmanoulopoulos et al 2011 ) . in de marco et al ( 2011 ) we speculated that this difference could be simply explained if the time - scales in pg 1211 + 143 are scaled - up by the bh mass , estimated to be about 10 - 100 times the mass of the other agn for which x - ray lags have been recorded so far . in this paper
we explore this hypothesis , by studying the properties of the lag frequency spectrum in a large sample of sources , spanning a wide range of masses ( about 2.5 decades ) .
the time - scales of detected soft / negative lags are then used to study correlations with relevant parameters , such as luminosity and bh mass .
the sources analysed are extracted from the _ caixavar _ sample a subsample of the caixa sample by bianchi et al ( 2009a ) presented in ponti et al ( 2012 ) , which includes all the well - exposed , x - ray unobscured ( @xmath12@xmath13 ) , radio quiet agn observed by xmm-_newton _ in targeted observations as of june 2010 .
we considered all the sources having at least one observation with a longer than 40 ks exposure ( 41 sources ) , and selected those with published black hole mass , @xmath3 , estimates ( 39 sources ) .
finally , we selected all the sources showing significant variability in their light curves . to this aim we made use of the excess variance estimates provided by ponti et al ( 2012 ) , computed by sampling the 2 - 10 kev light curve of each source in 40 ks segments (
note that also single - interval measurements were considered ) , and assuming a time resolution of 250 s. all the sources having excess variance different from zero at @xmath142@xmath15 confidence level were included in our sample , i.e. excluding all the sources consistent with having constant flux on the time - scales of interest for this work ( @xmath16ks ) . with the latter selection
our final sample was reduced to a total of 32 sources . for each source all the available observations in the xmm-_newton _ archive have been used in our analysis , apart from those highly corrupted by background flares . in the computation of the lag - frequency spectra ,
multiple observations have been combined to obtain better statistics .
+ only data from the epic - pn camera were used , because of its high effective area and s / n over the 0.3 - 10 kev energy band .
data reduction was performed using xmm science analysis system ( sas v. 10.0 ) , starting from the observation data files ( odf ) and following standard procedures .
filtered events are characterised by pattern @xmath174 , and are free from background flares .
typical source extraction regions are 45 arcsec radius circles .
spectra were extracted and used to select the energy bands for the computation of time lags .
they were grouped to a minimum of 20 counts per bin , while response matrices were generated through the rmfgen and arfgen sas tasks .
the analysis of the time series was carried out using routines implemented through idl v. 6.1 .
correction of sporadic count rate drops ( usually involving single time bins ) in the time series , which occur as a consequence of event losses due to telemetry dropouts , was performed by rescaling the count rate within the bin for the effective fractional time bin length .
we computed time lag - frequency spectra between light curves in the soft and hard x - ray energy bands .
the soft and hard energy bands were selected so as to single out energies dominated by the soft excess and the primary power law . to this aim
we adopted a phenomenological approach , although we note that several tests have been done by slightly varying the selected energy bounds , and the obtained results are all consistent with each other .
specifically , we first fitted the data in the 1 - 4 kev energy band with a simple power law absorbed by a cold column of gas ( which accounts for any warm absorber - induced extra curvature at high energies ) , with all the parameters left free to vary . for the soft band , we fixed at 0.3 kev the low energy bound , and selected , as high energy bound , the energy at which the soft excess significantly deviates ( at more than @xmath18 ) from the extrapolation of the best fit absorbed - power law . as the hard energy band we used the 1 - 5 kev range , excluding energies where significant deviations from the best fit power law were observed , e.g. due to the presence of troughs produced by complex absorption ( at @xmath19 kev ) or to the presence of a broad fe k line red tail ( at @xmath20 kev ) .
data above 5 kev were , in general , excluded to avoid contamination from components of the fe k line complex .
a detailed compilation of all the adopted energy bands can be found in tables [ tab : obs ] and [ tab : obs2 ] .
+ the time lag - frequency spectra are computed following the techniques described in detail by nowak et al ( 1999 ) and applied to xmm-_newton _ data in de marco et al ( 2011 ) .
the time lag is derived from the formula @xmath21 , where @xmath22 is the frequency - dependent phase of the average cross power spectrum between the soft and hard time series .
the resulting lag - frequency spectra are rebinned multiplicatively , i.e. the bin size is set equal to @xmath23times the frequency value , with step size , @xmath24 , chosen between @xmath25 depending on the quality of the data .
sources for which a negative lag was reported in the literature have been re - analyzed with our procedures . | we carried out a systematic analysis of time lags between x - ray energy bands in a large sample ( 32 sources ) of unabsorbed , radio quiet active galactic nuclei ( agn ) , observed by xmm-_newton_. the analysis of x - ray lags ( up to the highest / shortest frequencies / time - scales ) , is performed in the fourier - frequency domain , between energy bands where the soft excess ( soft band ) and the primary power law ( hard band ) dominate the emission .
we report a total of 15 out of 32 sources displaying a high frequency soft lag in their light curves .
all 15 are at a significance level exceeding 97 per cent and 11 are at a level exceeding 99 per cent . of these soft lags ,
7 have not been previously reported in the literature , thus this work significantly increases the number of known sources with a soft / negative lag .
the characteristic time - scales of the soft / negative lag are relatively short ( with typical frequencies and amplitudes of @xmath0 hz and @xmath1 s , respectively ) , and show a highly significant ( @xmath2 ) correlation with the black hole mass .
the measured correlations indicate that soft lags are systematically shifted to lower frequencies and higher absolute amplitudes as the mass of the source increases .
to first approximation , all the sources in the sample are consistent with having similar mass - scaled lag properties .
these results strongly suggest the existence of a mass - scaling law for the soft / negative lag , that holds for agn spanning a large range of masses ( about 2.5 orders of magnitude ) , thus supporting the idea that soft lags originate in the innermost regions of agn and are powerful tools for testing their physics and geometry .
[ firstpage ] galaxies : active , galaxies : nuclei , x - rays : galaxies |
the modelling of the spectral energy distribution ( sed ) of galaxies and its interpretation in terms of stellar population parameters ( e.g. age , star formation history , metallicity ) via stellar population synthesis ( sps ) is a fundamental tool to understand the properties and evolution of galaxies .
yet , we are far from a complete comprehension and a reliable modelling of some stellar evolutionary phases which strongly affect the energy output of stellar populations . among them , the so - called thermally pulsing - asymptotic giant branch ( tp - agb ) phase has been the focus of debate among modellers for several years ( e.g. @xcite ) . this late stage of the agb phase for low and intermediate mass stars ( @xmath4 57 m@xmath5 ) culminates in stellar populations of ages between 0.5 and 1.5 gyr .
tp - agb stars emit mainly in the near infrared ( nir ) spectral range given the low temperatures of these stars .
quantitative predictions of the lifetimes , luminosities and spectral shapes and features of these stars as well as of their impact on the integrated spectra of even `` simple '' stellar populations , are still very uncertain and strongly model - dependent ( e.g. * ? ? ?
one further complication derives from the fact that tp - agb stars are likely to be embedded in envelopes of dust produced by themselves ( e.g. * ? ? ?
* ; * ? ? ?
* ) , which attenuates the emerging optical and nir radiation by a substantial fraction @xcite . the seminal work of ( * ?
* ma05 hereafter ) , who adopted the `` fuel consumption theorem '' approach and calibrated the flux contribution of this phase against optical and nir photometry of magellanic clouds globular clusters , dramatically pointed out how large the impact of tp - agb stars on galaxy seds may possibly be .
the ma05 models make two clear and testable predictions in contrast to `` classical '' models ( such as * ? ? ?
* bc03 hereafter ) , which rely on standard , less extreme prescriptions for the tp - agb phase : in particular , _
i ) _ the nir flux of stellar populations of ages between @xmath60.5 and @xmath7 gyr is enhanced by a factor up to 3 ( as a function of age and metallicity ) ; in addition , _ ii ) _ strong , sharp absorption features appear in the spectrum , especially at nir wavelengths ( and most notably at 1.1 , 1.41 and 1.77 @xmath2 m ) .
these features correspond to the band - heads of carbon composite molecules and should depend on metallicity .
they represent the undebatable fingerprints of tp - agb stars as predicted by ma05 models .
this can be clearly seen in fig .
[ fig : ma05vsbc03 ] , where we confront the optical - nir spectra of simple stellar populations ( ssp ) of three ages ( 0.5 , 1 and 1.5 gyr ) at three different metallicities ( 0.5 , 1 and 22.5 @xmath8 ) as derived from the ma05 models ( black lines ) and the bc03 models ( red lines ) respectively .
observationally , detections of near - infrared cn absorption bands have been reported in the integrated spectra of only a few seyfert galaxies @xcite and globular clusters dominated by the stochastic presence of bright carbon stars @xcite .
in fact , from a theoretical point of view , carbon absorption features are not a required signature of agb - star populations .
sophisticated models of agb stellar evolution predict that convective dredge - up of carbon produced by nuclear fusion during the envelope thermal pulses will increase the abundance of carbon relative to oxygen near the surface , which can eventually lead to the production of a carbon star ( e.g. * ? ? ?
* ) . while the efficiency of this process is expected to increase as the metallicity decreases , the integrated spectrum of most intermediate - age stellar populations could well be dominated oxygen - rich agb stars without strong carbon absorption features .
the largely different nir flux predictions for tp - agb stars by different models have strong implications in terms of sed interpretation , leading to substantially lower stellar mass and age estimates ( by up to a factor 2 or more ) when rest - frame nir bands are included in the analysis of galaxies with ages around 1 gyr ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the popularity of nir passbands as tracers of stellar mass , both at low and high redshift , thanks to their ( generally ) low sensitivity to age , metallicity and dust in terms of mass - to - light ratio ( m / l ) ( e.g. * ? ? ?
* ; * ? ? ?
* ) , clearly calls for a proper assessment of the models in this respect ( see also * ? ? ?
verifying the reliability of sps models in the regime of ages around 1 gyr is therefore of paramount relevance for studies of galaxy properties and their evolution .
since 2005 , a number of contrasting results and claims have been produced about the appropriateness of the ma05 predictions .
@xcite showed that the ma05 models fit the seds of high - z ( @xmath9 ) galaxies better than bc03 models .
however the same work shows that the performance of the two models differ only marginally in terms of goodness of fit if one allows for dust reddening , although the inferred stellar mass is substantially different . a better agreement of ma05 models with the observed optical - nir colours is also reported by @xcite , who used combined optical spectroscopy and optical - nir photometry to constrain the star formation history of two galaxies .
@xcite also report detections of the 1.77 @xmath2 m feature in two magellanic clouds globular clusters , in agreement with ma05 predictions .
on the other hand , studies of magellanic clouds globular clusters found significant discrepancies between observations and ma05 predictions .
@xcite found that the co line strength ( in the nir k band ) is much weaker than predicted by ma05 in clusters of age @xmath6 1 gyr .
@xcite studied the colours of magellanic clouds globular clusters as a function of their age and showed that ma05 models predict optical and nir colours that are too red . moreover
the predicted age dependence of these colours does not agree with the data .
as opposed , a substantial agreement with models based on much less extreme tp - agb prescriptions is found . as already pointed out by @xcite ,
post - starburst galaxies ( psb hereafter ) are the best galaxies to look at in order to find the fingerprints of ( tp-)agb stars on their sed .
psbs have a star formation history dominated by a burst in the exact age range when tp - agb stars are expected to have the strongest impact , i.e. between 0.5 and 1.5 gyr prior to the epoch of observation .
their optical spectrum is characterised by a well developed balmer / d4000 break , the lack of strong emission lines ( which testify a substantial cessation of star formation activity ) and by the very strong balmer absorption lines typical of a stars ( hence the alternative denomination of e+a or k+a galaxies ) .
@xcite compared the optical - nir colours of the sample of psbs spectroscopically selected by @xcite to the colours of different sps models .
they showed ( see their fig .
16 ) that @xmath10 , @xmath11 colours are not reproduced by ma05 models : not only would the blue portion of this colour plane be only explained by unrealistically low metallicities , but also the age sequences in ma05 models produce trends that are orthogonal to the observed age trends . @xcite
performed another photometric test , in which they fitted the stacked ( rest - frame ) uv - optical - nir sed of 62 psbs at @xmath12 , colour selected from the _ newfirm _ medium band survey ( nmbs ) to have ages around 1 gyr , according to both bc03 and ma05 models .
also in this case it is found that bc03 models can reproduce the observed seds much better , in particular the observed blue optical - nir colours , which are not reproduced by ma05 . in this paper
we present the results of the experiment that we designed to test for the two main predictions of ma05 models , namely the presence of sharp nir spectral features and the boosted nir flux , on a sample of psbs carefully selected from optical spectroscopy and newly observed in nir spectroscopy ( h and k band ) with isaac at the eso - vlt .
the selection criteria and properties of the sample are fully detailed in section [ sec : sample_obs ] . here
we would like to stress the uniqueness of this work . _
i ) _ for the first time we cover both the optical and the nir range of psbs with flux calibrated spectra to look directly for the nir spectral features at 1.41 and 1.77 @xmath2 m predicted by the ma05 models and resulting from the large impact of tp - agb c - rich stars .
these features are unaccessible or extremely difficult to measure from the ground as they roughly coincide with the atmospheric gaps between j and h , and between h and k bands , respectively , for nearby objects .
therefore we observed psbs in the redshift range 0.150.25 so that the c - features move well into the observable windows and are minimally affected by uncertain telluric corrections , as shown in fig .
[ fig : ma05vsbc03 ] . _
ii ) _ contrary to previous works based either on colour selection @xcite or on a purely phenomenological spectroscopic selection @xcite , the selection of our sample is based on both phenomenological criteria ( high balmer line strength and low emission lines , * ? ? ?
* ) and bayesian estimates of light weighted ages from optical spectral indices as in @xcite .
this allows us to focus our observations and analysis on the _ precise _ range of stellar ages where tp - agb stars are expected to contribute the most , for a variety of stellar metallicities .
the paper is organised as follows . in section [ sec : sample_obs ] we present the sample , the new isaac nir spectroscopic observations and the data reduction , including spectrophotometric calibrations and integration with optical data .
in section [ sec : analysis ] we present the results and discuss possible spurious effects and contaminations that might affect our conclusions . a comparison with the analysis of @xcite is also presented .
finally , in section [ sec : summary ] we summarise our findings and propose our conclusions and future developments .
we draw our psb targets from the spectroscopic sample of the sloan digital sky survey ( sdss , * ? ? ?
* ; * ? ? ?
* ) data release 7 ( dr7 , * ? ? ?
the initial pre - selection was based on the line equivalent width ( ew ) criteria defined in @xcite , namely : ew(h@xmath13)@xmath14 , ew(h@xmath15)@xmath16 , ew([oii])@xmath17 .
emission , the sum of the doublet is considered . ] in order to ensure that the two strong nir features corresponding to the band - heads of cn ( 1.41 @xmath2 m ) and c@xmath18 ( 1.77 @xmath2 m ) due to carbon rich agb stars ( cf .
* ) are observable from the ground , we require the galaxies to be at @xmath19 , so that 1.41 @xmath2 m and 1.77 @xmath2 m ( at rest ) move into the h- and k - band windows , respectively .
this selection is based on the sdss - dr7 casjobs catalog .
we further restrict the sample by applying the visibility cuts in order for targets to be observable from paranal during the winter semester : dec@xmath20 deg and ra between 0 and 230 deg or ra@xmath21 deg .
finally we require that the stellar light - weighted age of the galaxies is in the range 0.5 to 1.5 gyr , i.e. at the maximum of the predicted tp - agb luminosity contribution , based on the estimates computed as in @xcite and available in the mpa - jhu catalog .
given the relative rarity of the post starburst phase and the restrictive age , coordinate and redshift cuts ( the latter particularly relevant as it is almost at the extreme boundary of the sdss main galaxy sample ) , the final sample is composed of sixteen galaxies only .
the properties of the sample are summarised in table [ tab : sample ] and in fig .
[ fig : sample_diagn ] .
this figure illustrates the distribution of our galaxies in the classical h@xmath22 vs @xmath23 plane . in this plane the main sequence of galaxies with smooth continuum star formation history is displayed by the grey intensity levels , which represent the distribution of sdss - dr4 galaxies of increasing ages from top - left to bottom right .
galaxies lying above this sequence , such as our galaxies , represented by the red dots with error bars , are characterised by relatively recent bursts of star formation with ages between a few hundred myr to a couple of gyr , which dominate the light of the galaxy . for comparison ,
tracks of simple stellar populations ( ssp ) with different metallicities from bc03 are shown ; the typical bell - shape of these ssp tracks peaks in h@xmath22 at ages of 0.30.5 gyr ( depending on metallicity ) and decreases to h@xmath24 35 for ages of a few gyr . by design
, our sample populates the intermediate range of burst ages in this diagram .
the statistical estimates of light - weighted stellar ages , metallicity and mass computed from optical spectral absorption indices ( @xmath23 , @xmath25 , @xmath26 , @xmath27^\prime}$ ] , @xmath28}$ ] ) with the bayesian method of @xcite and used in the selection are reported in tab . [
tab : sample ] ( columns 6 , 7 and 8) , along with the i d ( 1 ) , coordinates ( 2 , 3 ) , redshift ( 4 ) and apparent @xmath29-band petrosian magnitudes ( 5 ) of the galaxies from the sdss .
the sample comprises intermediate mass galaxies , between @xmath30 and @xmath31 , with ages in the range @xmath32 gyr and metallicities between 0.2 and 2.3 solar .
most galaxies have indeed @xmath33 , as expected given their stellar mass . despite the paucity of galaxies at low age ( @xmath34 gyr ) _ and _ low metallicity ( @xmath35 ) , the sample provides a good coverage of the relevant ages for the tp - agb phase and of the metallicities most typical for galaxies around the characteristic luminosity @xmath36 .
[ cols= " < , > , > , > , > , > , > , > " , ] -d4000@xmath37 diagram , a proxy for burstiness vs. age of the stellar population .
red points with error bars are the sixteen galaxies selected for the present study ; the underlying grey - scale image represents the density distribution of a complete sample of galaxies from sdss - dr4 .
as one can see , our sample selects galaxies dominated by a burst in a range of young to intermediate ages .
note that the estimates of h@xmath38 used in this plot are not exactly the same as the e.w .
measurements used in the sample selection ( therefore values @xmath39 are not inconsistent with the selection criteria reported in the text ) . ]
we note that the estimates of the stellar population parameters of @xcite rely on the comparison with synthetic spectral libraries based on bc03 models , therefore they could be biased with respect to estimates based on other synthesis models , most notably the ma05 ones
. however , optical spectral indices short - wards of 5500 are generally well agreed upon among different modellers and , most importantly , are only weakly affected by tp - agb stars , which mainly influence the nir spectral region .
thus age and metallicity estimates from optical absorption features are robust against different tp - agb prescriptions and the use of different models .
this is even more the case for spectra dominated by a stars , where balmer absorptions are very strong .
furthermore , in sec .
[ sec : analysis ] ( [ sec : sspcomp ] and [ sec : cspcomp ] in particular ) we thoroughly explore if and how different ( combinations of ) ages and metallicities can possibly explain the observed properties of our galaxies , irrespective of their original selection .
based on the sdss optical images and , to minor extent , on our isaac nir images , we note that the majority of galaxies display smooth and centrally concentrated light distributions , typical of early type galaxies , with the only exception of psbj0227@xmath400015 , which has an extended and irregular `` halo '' .
however , other four galaxies ( psbj1039@xmath410604 , psbj1046@xmath410714 , psbj1119@xmath411313 and psbj1206@xmath410918 ) have close and possibly interacting companions ; in many cases , faint structures ( like shells or streamers ) can be tentatively seen .
this suggests a relatively recent merger or interaction as a possible origin of the ceased starburst in these galaxies .
the sixteen galaxies of the sample were observed in the programme i d 086.b-0733 ( p.i .
: s. zibetti ) at the ut3 melipal of the eso - vlt at the paranal observatory .
the isaac instrument @xcite has been used in two different spectroscopic setups to obtain low resolution spectra in the h and k bands .
1.5 arcsec slit and sws1-lr grism were used in both setups , with central wavelengths of 1.65 and 2.2 @xmath2 m respectively .
this gives a spectral resolution of @xmath42 ( 50 ) and @xmath43 ( 70 ) in h and k respectively .
observations were carried out in service mode with clear ( not necessarily photometric ) sky and maximum seeing of 1.4 arcsec to minimise slit losses and match the fibre size and seeing of the sdss optical spectra as well as possible .
standard abba on - slit nodding with 70 arcsec offsets and 20 arcsec jittering were adopted to allow accurate sky subtraction and bad pixels and cosmic ray rejection .
eight exposures of 180 sec each were taken in h band and six exposures of 150 sec each were taken in k , for total 24 minutes of integration in h and 15 minutes in k per target .
this eventually yields typical signal - to - noise ratio ( snr ) of @xmath44 and 25 per resolution element in h and k respectively .
spectro - photometric standard stars were observed with identical setup immediately after or before the science observations to provide a reliable estimate of the spectral sensitivity function and compute the corrections for telluric absorption .
standard data reduction procedures have been implemented in iraf .
wavelength calibration has been performed based on the sky oh lines as measured on a median combined image of the scientific frames .
this approach turned out to be much more stable and reliable than using the arc lamp calibrations taken in day time .
we have tried different approaches to optimally correct the spectra for telluric absorption . as our best solution we adopted the following strategy : the empirical high - resolution transmissivity function of the average atmosphere above paranal ( provided by eso ) is finely tweaked by scaling and changing the resolution until we can exactly compensate for the observed absorption at various wavelengths on the spectrum of the spectrophotometric standard ; the coadded spectrum of each galaxy
is then corrected for the same absorption pattern as for the closest ( in time and airmass ) standard star .
we have applied the telluric correction to both standard stars and galaxies _ before _ computing the sensitivity function and applying the flux calibration . in order to assemble a sed including the visible range and the h and k nir bands , absolute flux calibration is required . to this goal
we have obtained high snr images in @xmath45 and @xmath46 bands in photometric conditions with isaac as part of the same program .
`` total '' integrated magnitude derived from these images are used to rescale the spectra . by `` total '' magnitude
we adopt the magnitude integrated within the petrosian aperture , as defined by the sdss photometric pipeline on the sdss @xmath29-band images in data release 8 ( dr8 , * ? ? ?
dr8 is chosen as it fixes a number of issues especially with sky subtraction . @xmath45 and
@xmath46 magnitudes are used to rescale the respective isaac spectra in the nir , while the optical spectrum from the sdss is normalised according to the @xmath29-band petrosian magnitude0205 and psbj0151@xmath400056 ) we find fainter magnitude by more than 0.03 mag ( + 0.040 and + 0.033 respectively ) .
these small discrepancies are well within the errors adopted in sec .
[ sec : sspcomp ] and [ sec : cspcomp ] . ] .
conversions between nir magnitudes natively calibrated in the vega system for the isaac filters into the corresponding flux densities for our spectra are computed based on the spectrum of vega of @xcite , distributed by bc03 .
typical total photometric uncertainties for isaac petrosian magnitudes are estimated @xmath47 mag and include photometric zero point fluctuations ( @xmath48 mag , derived from repeated observations of standard stars during the nights ) and background subtraction uncertainties . for the sdss
we assume errors of 0.02 mag , which is the maximum reported for our objects by sdss - dr8 .
foreground galactic extinction correction is finally applied using the optical depth derived from @xcite and assuming the standard extinction laws of @xcite and @xcite .
the spectra of our galaxies , both sdss and isaac , are taken in relatively small apertures or slits centred on the brightness peak : 3-arcsec
fibres are used in the sdss with a typical seeing of @xmath49 arcsec while 1.5-arcsec long - slits with seeing of 1.4 arcsec or better are used with isaac . relative aperture size and seeing
are designed to conspire and collect a similar fraction of the galaxy light . by rescaling the spectra according to the `` total '' petrosian magnitudes we implicitly assume that stellar population gradients are of limited relevance and the central region of the galaxies are representative of the whole . @xcite
have analysed a small sample of post - starburst galaxies with integral field spectroscopy and showed that , in reality , a large fraction of them have cores that are younger than the rest of the galaxy , hence our assumption does not strictly hold . unfortunately , because of different seeing conditions and geometry of the apertures , a precise and homogeneous rescaling of the spectra is very hard to achieve . however ,
by comparing fluxes and colours that are obtained in the petrosian and in the fibre apertures we can show that the aperture effects for our sample are small enough not to affect our conclusions .
first of all , we note that at the relatively high redshift of our sample the apparent size of the galaxies is typically 2.5 arcsec ( median petrosian radius , maximum 3.1 arcsec ) , so that the 3-arcsec sdss fibre typically collects already half of the total flux .
the isaac spectra roughly collect the same fraction of flux .
the long slit geometry in this case integrates into the spectra not only the central part , but also the outskirts of the galaxy , so that the isaac spectra are indeed even more representative of the whole galaxy . in second place ,
we have checked by how much `` total '' petrosian colours differ from fibre colours in the optical .
fibre colours are computed in such a way that they correspond as closely as possible to the sed portion effectively captured in the sdss spectra .
the strongest effects are expected and observed in @xmath50 , since this colour roughly corresponds to @xmath51 at @xmath1 and therefore is the most sensitive to stellar population properties by bracketing the d4000 break .
we find that , on average , fibre @xmath50 is bluer by 0.027 mag with respect to the corresponding total petrosian colours , with a rms of 0.037 .
only four galaxies ( out of 16 ) have a fibre colour which is bluer than total by more than 0.05 mag : psbj1046@xmath410714 , psbj1015@xmath410103 , psbj0227@xmath400015 and psbj0328@xmath410045 .
we dub these galaxies `` blue core '' in the following . for the remainder
, differences in colour can be considered as not significant , although the data hint at starbursts preferentially occurring in the central regions of galaxies rather than in the outskirts , consistently with @xcite .
looking at @xmath52 , which roughly corresponds to the rest - frame optical colour @xmath53 that will be discussed in the following , we find even smaller differences of 0.012 mag on average ( 0.021 mag rms ) .
these are well within the spectrophotometric calibration errors . finally , we estimate the relative aperture corrections between the optical and nir bands . specifically , we consider the aperture corrections in @xmath29 and @xmath45-band , defined by @xmath54 . for the @xmath29-band
we use the quantities provided by the sdss database . for the @xmath45-band we use the total petrosian magnitude as derived above ,
while to obtain fibre magnitudes we first convolve the isaac images to a common psf of fwhm@xmath552 arcsec and then integrate the flux within a circular aperture of 3 arcsecond in diameter , as done in the sdss pipeline . since for most of the isaac images we lack a sufficient number of high snr stars to compute an accurate psf ,
the final fwhm is approximated within 0.1 arcsec .
we find an average difference of aperture corrections in @xmath29 and @xmath45-band @xmath56 mag with rms of 0.1 mag , indicating that on average our galaxies are slightly more concentrated in @xmath29 than in @xmath45 , although with very marginal significance .
this is consistent with the existence of few blue cores and , in general , of very small systematic sed variations between the spectral and the total apertures .
this analysis of the aperture effects justifies our choice to use the total petrosian magnitude to compute the relative normalisation of the optical and nir spectra .
total petrosian magnitudes are much more stable than magnitudes integrated in small apertures comparable with the seeing , which therefore rely on uncertain psf convolution . on the other hand , we have shown that they do not introduce significant biases between different spectral regions .
the systematic aperture bias of 0.026 mag between @xmath29 and @xmath45 is in fact well within the photometric uncertainties estimated in the previous section . in the following analysis
we will adopt 0.1 mag as standard error on the optical - nir colours and 0.05 mag on the optical colours , to take into account both photometric errors and aperture effects .
the composite optical - nir spectra obtained as explained in the previous section are presented in fig .
[ fig : specs_full ] , in the rest frame and normalised to the flux density at 5500 , in order to ease a prompt comparison between all galaxies .
the actual normalisation refers to the median flux in the range 54805520 , in order to minimise the effect of noise on the individual spectral pixels .
the sdss portion of the spectra covers approximately the rest - frame range 30007500 , isaac h 1.251.50 @xmath2 m , and isaac k 1.652.00 @xmath2 m . for each galaxy ,
three panels display the full optical - nir spectral range ( top left plot ) , a zoom into the optical region that includes the most significant diagnostic lines ( top right plot ) , and the nir h k range ( bottom plot ) .
green lines are used to plot the spectra at high resolution : the original @xmath57 and @xmath58 resolution at @xmath59 and @xmath60 per pixel in the rest frame is used for the h and k spectra respectively , while for the sdss spectra ( original resolution @xmath61 at @xmath62 per pixel in the rest frame ) a smoothing over 3 pixels is applied to make the plot more readable .
black thick lines are used to ( over)plot heavily smoothed versions of the spectra , in order to ease the identification of possible spectral features , especially in the nir bands : smoothing over 15 pixels is adopted both in the optical and the nir spectra , resulting in rest frame resolution of @xmath63 , 60 and 90 in optical , h and k ranges , respectively .
vertical dotted lines mark the position of the main ( expected ) spectral features : the balmer series ( up to h@xmath64 9 - 2 ) and the two features of cn ( 1.41 @xmath2 m ) and c@xmath18 ( 1.77 @xmath2 m ) due to carbon rich agb stars @xcite . for each galaxy in the bottom panel
we report also the light - weighted stellar age and metallicity and corresponding uncertainties derived from optical absorption features as in @xcite .
all galaxies display typical e+a or post - starburst spectra with very strong balmer absorptions , well developed d4000 break and blue continuum .
the actual slope of the optical continuum generally shows the well known dependence on age and metallicity , although dust attenuation may affect the slope as well . we do not attempt to put any constraint on this parameter , but discuss its possible impact on our conclusions below .
three galaxies , psbj0227@xmath400015 , psbj0328@xmath400045 , and psbj1150@xmath400107 show significant h@xmath15 emission ( and some higher ionisation lines for psbj0328@xmath400045 ) , indicating some residual ongoing star formation , which is also confirmed by ( marginal ) detections at 22 @xmath2 m ( w4 ) in the wise all sky survey @xcite , measured with snr of 4.4 and 2.8 , and 2.0 , respectively .
however , the equivalent width in all cases matches the @xcite criteria , being @xmath65 , @xmath66 , and @xmath67 , respectively , all larger than @xmath68 . given the observed strength of all balmer absorption lines , the contribution of newly formed stars ( or agn ) can be largely neglected in comparison to the radiation by the intermediate - age population we aim at studying here .
it is worth noting that two of these galaxies ( psbj0227@xmath400015 and psbj0328@xmath400045 ) also have a blue core , supporting the idea that blue cores might be related to some residual star formation extending close to the present time .
the absence of emission lines and blue cores from the majority of the galaxies in our sample further confirms their nature of genuine post - starburst galaxies .
what is most important for the scope of this work is the lack of any evidence for sharp features in the nir in any of the 16 galaxies .
most of them display completely featureless continua with approximately uniform slope within the two nir windows .
some tentative evidence for a wavelength - dependent slope is seen in a few objects : psbj1039@xmath410604 , psbj1046@xmath410714 and especially psbj1314@xmath410430 appear to change their slope around 1.41 @xmath2 m , while in psbj1119@xmath411313 and psbj1133@xmath410205 a change of slope can be tentatively seen around 1.77 @xmath2 m .
similar mild changes of slope , most likely due to molecular bands in the atmospheres of cool stars , are compatible with bc03 but are in fact substantially dissimilar from the sharp drops by a few 10% in flux predicted by ma05 models ( see fig .
[ fig : ma05vsbc03 ] ) . to further support these conclusions , we stacked all 16 spectra in h and k separately , after normalising them in a 1000 - wide region around 1.41 and 1.77 @xmath2 m , respectively .
the results are displayed in fig .
[ fig : stack_hk ] , with the black lines showing the average spectrum and the shaded green area indicating the 16 - 84% inter - percentile range ( approximately @xmath69 range ) .
this figure shows that _ on average _ carbon molecules in the atmospheres of evolved stars produce band - head drops that are definitely less than 5% of the continuum .
stellar absorption indices in the optical range are extremely powerful diagnostics of stellar population properties .
decades of development and tests have led to substantial agreement on their interpretation in terms of ages and global metallicity . in this section
we exploit part of the wealth of information that we can extract from the optical sdss spectra of our galaxies for two main purposes : _
i ) _ probe the ability of `` tp - agb light '' models ( represented by bc03 ) and `` tp - agb heavy '' models ( ma05 ) to reproduce the observed trends and distributions in the parameter space of optical absorption indices and optical - nir colours ; _ ii ) _ investigate whether the lack of nir spectral features can be due to selection biases such that we misinterpreted optical diagnostics ( e.g. due to the use of bc03-based model libraries ) and ended up missing the relevant phase for tp - agb stars .
we start considering only age - sensitive absorption indices and optical - nir colours and build the diagnostic plots presented in fig .
[ fig : diag_optnir_ssp ] for bc03 and ma05 in the _ top row _ and _ bottom row _ , respectively : coloured symbols and lines show the predicted optical - nir colour @xmath70 as a function of @xmath25 , @xmath23 , and the optical colour @xmath53 ( from _ left _ to the _ right _ ) . , @xmath71 and @xmath72 denote the ab magnitudes corresponding to the flux density around 0.55 , 0.70 and 1.40 @xmath2 m ( rest ) , respectively . ] for bc03 , ages range from 50 myr ( lower ages are not relevant for the present analysis as they would yield much bluer colours than observed ) to 13.6 gyr .
metallicities range from 1/50 solar to 2.5 times solar , and are represented with different symbols and colours from blue to red ( solar in orange , see figure caption ) .
symbols of increasing size are used to mark discrete increasing ages , with the crucial age of 1 gyr identified by a star .
the spectral indices @xmath25 and @xmath23 are measured after convolving the model spectra to match the resolution and average velocity dispersion of the galaxies ( @xmath73 ) .
the ma05 ssp spectra are distributed over the full optical - nir range only at low resolution ( 20 sampling ) , therefore can not be used to compute reliable indices , @xmath25 in particular . by up to @xmath74 and to overestimate @xmath23 by up to @xmath75 . ]
@xcite delivered high resolution spectra for a subsample of the ma05 age - metallicity grid , over the optical range : only three metallicities are available ( 0.5 , 1.0 and 2.0 solar ) , with the lowest ages being 200 , 30 , 400 myr , respectively .
we use these models ( in their stelib version ) , convolved to the effective resolution of our data , to compute the @xmath25 and @xmath23 plotted in the first two panels of the _ bottom row _ of fig .
[ fig : diag_optnir_ssp ] ; optical - nir colours are computed from the original ma05 seds of corresponding age and metallicity . for the colour - colour plot the full grid of ma05
is used instead , which allows extending the metallicity down to 1/20 solar and the age down to 50 myr as for bc03 .
observational data points derived from our composite optical - nir spectra are plotted as error bars . and the optical - nir colour @xmath70 respectively , following the discussion in sec .
[ subsec : aper_eff ] .
errors on indices are derived from the mpa - jhu catalog and , in the case of @xmath23 , incremented _ a posteriori _ to take into account flux calibration errors , as statistically derived by comparing duplicate observations of a large sample of sdss galaxies . ] as shown by the _ top _ row of fig .
[ fig : diag_optnir_ssp ] , bc03 ssps provide a good match to the observed spectral diagnostics , with the only exception of a couple of galaxies , psbj1150@xmath400107 and psbj1206@xmath410918 , which appear to have a too blue @xmath70 nir - optical colour for the given @xmath53 optical colour ( or , reversely , a too red @xmath53 for the given @xmath70 ) and will be discussed in sec .
[ subsec : deviant_objs ] . in particular
, galaxies cover the range in age and metallicity corresponding to light - weighted ages and metallicities listed in table [ tab : sample ] , which are mostly relevant for the tp - agb phase . as far as the comparison with the ma05 models
is concerned , we note that the observed age - sensitive indices approximately match the range of ages given by our light - weighted estimates . most of the observed @xmath25 and @xmath23 values correspond to ssp ages between 0.5 and 1 ( 1.5 ) gyr for the solar ( sub - solar ) metallicity track , while ages smaller than 0.5 gyr would be inferred from the 2 times solar metallicity track . for the given indices , however , the predicted @xmath70 nir - optical colours are on average 0.5 mag redder than the observed ones .
this is consistent with the ma05 `` tp - agb heavy '' models having the nir very much boosted with respect to bc03 in the age range around 1 gyr implied by the balmer indices . in the rightmost panel of fig .
[ fig : diag_optnir_ssp ] , _ bottom _ row , we see that , unless a metallicity as low as 1/20 solar is representative of most galaxies and ages are significantly larger than 1 gyr , the ma05 models predict @xmath70 nir - optical colours that are on average 0.5 mag redder than observed , for the observed optical @xmath53 colours .
we note that the inclusion of dust with a standard effective attenuation curve in the models does not alleviate the tension between the ma05 models and observations .
we illustrate this in fig .
[ fig : diag_optnir_ssp ] by drawing arrows which represent the offset to apply to each model when an optical extinction of @xmath76 mag and an effective attenuation law @xmath77 ( following * ? ? ?
* for an ssp ) are applied . as expected , the effect of dust is to move model tracks to redder colours : this would worsen the disagreement in the index - colour planes .
the reddening vector is approximately parallel to the model tracks in the colour - colour plane : accounting for dust attenuation in the models would not substantially change the ( dis)agreement with the observations in terms of colours , although it would require even lower metallicities and ages .
vs [ mgfe]@xmath78 diagnostic plot , showing that our galaxies need to have metallicity not lower than @xmath6 40% solar in order to be consistent with the observed strength of metal sensitive absorption lines .
the four evolutionary tracks in colour are from bc03 models , colour coded and marked as in fig .
[ fig : diag_optnir_ssp ] . ]
the hypothesis of a metallicity as low as 1/20 solar is not only unlikely to apply to a full sample of relatively massive galaxies with @xmath79 , but is also inconsistent with direct metallicity constraints which can be derived from absorption indices . in table
[ tab : sample ] we already showed metallicity estimates derived from the vast library ( based on bc03 ) and bayesian method of @xcite : these estimates range from 0.2 @xmath8 minimum to 2.3 @xmath8 maximum , roughly uniformly distributed . in fig .
[ fig : hbeta_mgfep ] we show that metallicities as low as 1/20 solar are indeed ruled out already from a simple comparison of standard absorption line indices on which different models are in substantial agreement , namely [ mgfe]@xmath78 and h@xmath80 .
[ mgfe]@xmath78 in particular is chosen as optimal metallicity indicator following @xcite who demonstrated its insensitivity to @xmath81fe abundance ratio ; h@xmath80 serves to constrain the age of the stellar population and hence alleviate the well known age - metallicity degeneracy .
[ fig : hbeta_mgfep ] shows the indices measured in our galaxies as crosses with error bars , while tracks of different colours are derived from bc03 ssps in the age range 50 myr 13.6 gyr ( symbols are the same as in fig . [
fig : diag_optnir_ssp ] ) .
four different metallicities , 1/50 , 0.4 , 1 and 2.5 solar are plotted in purple , blue , green orange and red , respectively .
it is apparent from this plot that metallicities significantly lower than 0.4 solar are inconsistent with the data - points .
we have checked that the same conclusions are obtained if indices based on bc03 models are replaced with those computed by thomas and collaborators @xcite : the only difference between the two sets of models is that thomas models shift the lower limit allowed by the data to slightly higher metallicities .
a possible explanation why the strong tp - agb features predicted by ma05 models are not seen in our spectra is that in the nir there is a substantial contribution by old stars , which are almost completely outshined by the younger population in the optical bands .
in fact , it is quite unlikely that the entire stellar population of galaxies as massive as several @xmath82 is formed in a burst @xmath62 gyr old and a more realistic scenario is that the burst shines atop an old component . to explore this hypothesis
we have repeated the test presented in fig .
[ fig : diag_optnir_ssp ] using composite stellar populations ( csp ) instead of ssps . in a first experiment ,
each csp is composed of two components : a fixed 10-gyr old , solar metallicity ssp ( which we will call `` old component '' in the following ) and an ssp of variable age , from 50 myr to 13.6 gyr , and metallicity , from 1/20 to 22.5 solar ( which we will call `` young '' component or `` burst '' , and corresponds to the ssps of figure [ fig : diag_optnir_ssp ] ) . in the following we refer to these models as csp - s , where `` s '' indicates that the metallicity of the old component is solar .
this choice appears reasonable considering that @xmath83 is typical for the old component of massive galaxies with @xmath84 ( e.g. * ? ? ?
* ; * ? ? ?
we have varied the relative fraction of ( initial ) mass in the old component and in the burst from 0:1 to 0.9:0.1 .
[ fig : diag_optnir_csp ] shows the same diagnostic plots we already used to analyse pure ssps ( fig .
[ fig : diag_optnir_ssp ] ) , but now plotting csp - s models for the extreme `` contamination '' , i.e. for 90% of the formed mass in the old component and 10% in the burst ( as was also assumed by * ? ? ?
symbols and colours are the same as in fig .
[ fig : diag_optnir_ssp ] and refer to age and metallicity of the variable burst component alone .
csp - s models , either bc03 or ma05 , can easily reproduce the distribution of data - points in @xmath25 and @xmath23 : the contamination by the old component is compensated by a younger burst component ( @xmath85 myr ) with respect to those that provided the best match in the pure burst ( ssp ) models .
such a combination would easily explain why we do not see the sharp nir features due to tp - agb stars predicted by ma05 : not only part of the nir emission is provided by the old and tp - agb free stellar population , but also the young component is younger that the peak of the tp - agb phase .
however , problems arise when optical - nir colours are considered . the addition of an old component at solar metallicity to a young burst yields significantly redder optical - nir colours with respect to a pure burst . as a consequence , bc03 models , which provided a decent match in the case of a pure ssp ,
are now shifted to inconsistently red colours with respect to the observations . in the case of ma05 models , which already had troubles being too red in optical - nir colours at almost any age ,
they become inconsistent with the observed colours of almost all galaxies .
mixtures with an intermediate mass ratio between the old population and the burst produce also intermediate results in terms of indices and colours between the pure burst models ( fig .
[ fig : diag_optnir_ssp ] ) and the csp - s ones presented in fig .
[ fig : diag_optnir_csp ] .
in fact , as shown in fig . [
fig : diag_optnir_csp_20 ] , a more moderate ratio between `` old '' and `` burst '' stars of 2:1 yields a decent match between the bc03 models and the observations .
as for the ma05 models , however , the fact that already ssps are too red in optical - nir colours results in no combination of `` old '' vs `` burst '' stars being able to reproduce the observed colours .
we repeated the experiment with the 9:1 ratio , but changing the metallicity of the old component from @xmath86 fixed to a variable metallicity that matches the one of the `` young '' component .
we tag these models as csp - y , where `` y '' indicates that the metallicity of the old component is the same as in the young component .
the results are shown in fig .
[ fig : diag_optnir_csp_samez ] .
bc03 models ( fig .
[ fig : diag_optnir_csp_samez ] _ top _ ) are able to cover most of the parameter space occupied by the data - points ; however a significant number of galaxies are too blue in @xmath70 with respect to models at given @xmath23
. moreover , the metallicities implied by colours are systematically lower ( and to large extent inconsistent ) with those implied by the absorption indices alone . as one can argue in the light of the results of pure ssp and 2:1 ratio models , less extreme fractions of the old component can provide a much better match to the data . as for the ma05 models , fig .
[ fig : diag_optnir_csp_samez ] _ bottom _ shows that not even with this kind of mix it is possible to match the distribution of @xmath70 vs @xmath23 , the optical - nir colours of the models being too red with respect to the data .
we reiterate that any dust with the most commonly adopted attenuation curves would only worsen this disagreement . in summary , adding a substantial fraction of old stellar populations to reduce the impact of tp - agb stars predicted by ma05 in the nir may indeed explain the lack of observed nir features , while remaining consistent with the observed optical spectral indices
. however , such a combination fails to reproduce the observed optical - nir colours at given @xmath25 , @xmath23 and optical colours , no matter if the metallicity of the `` old '' component is the same as the one of the `` young '' component or is fixed at the solar value ( as it is probably more realistic for an old stellar population in a system of a few times @xmath87 ) .
we can not exclude that more complicated and _ ad hoc _ star formation histories and metallicity mixtures might be able to produce a closer match between observations and ma05 , but it appears unlikely that such a fine tuning actually applies to all 16 galaxies in the sample . on the other hand ,
bc03 models appear to reproduce the observations much more easily by assuming ssps or csps with even massive `` old '' components of variable sub - solar metallicity ; in order for models with an `` old '' component at solar metallicity to match the observations , `` old '' fractions less than 0.9 are required in most cases : it is sufficient to lower such fraction to 0.6 - 0.7 in order to obtain a generally good agreement .
we further illustrate these points by directly comparing ssp and csp model spectra with an actual spectrum of our psb sample in fig .
[ fig : example_csp_j1006 ] , where psbj1006@xmath411308 is used as a mere example which optimally serves to this goal , as it is broadly representative of the full sample . for both bc03 ( shown in the left - hand panel ) and ma05 ( in the right - hand panel ) , we pick the best fitting models from each of the three grids of ssp , csp - s and csp - y ( assuming old : young=9:1 ) .
they are overplotted to the observed spectrum ( solid black line ) , in blue for the ssp , in solid red for the csp - s and dashed green for the csp - y .
the `` best fitting '' models are those yielding the minimum @xmath88 over the optimal set of optical absorption indices defined in @xcite ( @xmath23 , @xmath25 , @xmath26 , @xmath27^\prime}$ ] , @xmath28}$ ] ) . due to the coarse grid in metallicity ,
the use of fixed old fractions and the neglect of dust attenuation , we warn the reader not to over - interpret these fits in terms of derived physical parameters .
nonetheless , the following key observations can be derived from fig .
[ fig : example_csp_j1006 ] .
_ i ) _ all best fit models can reproduce the observed optical features extremely well ( see insets ) .
the slope of the optical continuum is also generally well reproduced , although for bc03 the ssp fit is too blue in the optical range , which might be due either to a small mismatch in metallicity because of the coarse grid or to small amounts of dust , which are not included in our simple models . _
ii ) _ the nir spectrum of the best fitting ma05 ssp , with an age of 0.6 gyr and metallicity 2 times solar , has the typical nir boosting due to the ma05 implementation of the tp - agb phase , clearly recognisable from the sharp features around 1.41 and 1.77 @xmath2 m
. however , no evidence of such features appears in the observed spectrum ; moreover , the observed spectrum is almost a factor of two dimmer than the ma05 ssp in the nir .
in contrast , the bc03 ssp provides a very good fit in the nir both in terms of ( lack of ) features and of overall flux . _
iii ) _ the use of csps with a substantial fraction ( 90% ) of 10-gyr old stars yields a relative enhancement in the nir over optical for bc03 models , for both csp - s and csp - y .
the plot suggests that a smaller old fraction and lower metallicity than the best fitting pure ssp might provide a very close match to the nir - to - optical ratio .
csps for ma05 ( which by coincidence converge to the same solution for both csp - s and csp - y , i.e. the solid red line and the dashed green one overlap ) do actually yield much weaker nir features and also lower nir flux relative to optical .
however , this stays significantly higher ( @xmath44% ) than observed .
psbj1150@xmath400107 and psbj1206@xmath410918 are characterised by extremely blue optical - nir colours ( @xmath89 ) and intermediate / red optical colours ( @xmath90 ) . when plotted on the colour - colour diagrams of fig .
[ fig : diag_optnir_ssp ] , [ fig : diag_optnir_csp ] , [ fig : diag_optnir_csp_20 ] and [ fig : diag_optnir_csp_samez ] , they significantly depart from the sequences both of observed data - points and model tracks .
none of them is able to reproduce such a combination of intermediate / red optical colours and very blue optical - nir ones : for the given @xmath70 , an offset of @xmath91 to @xmath92 mag in @xmath53 would be required to bring them back in agreement with the models , or , for the given @xmath53 , an offset of @xmath93 mag in @xmath70 . from the arrows plotted in the colour - colour diagrams ,
it is immediate to see that a standard effective attenuation la @xcite is unable to move the models closer to these two objects : in fact , one would need an effective attenuation curve which is extremely flat , thus causing similar attenuation at 0.70 and 1.40 @xmath2 m . with standard dust grains ,
for which the absorption cross section decreases with wavelength , this can be only obtained by invoking a peculiar geometry in which back scattering plays a major role .
significant residual star formation is witnessed in psbj1150@xmath400107 by ew(h@xmath15)@xmath94 and a marginal detection at 22 @xmath2 m in the wise all - sky survey .
psbj1206@xmath410918 , however , does not display any sign of star formation , with ew(h@xmath15)@xmath95 and no detection at 22 @xmath2 m .
this suggests that residual star formation or processes related to nuclear activity are not the cause of the anomaly common to these two galaxies .
we note that the shape of the spectra of these two galaxies is peculiar in two main aspects : _
i ) _ the continuum appears to bend down , i.e. the slope becomes steeper , going from @xmath96 to 0.8 @xmath2 m , whereas all other galaxies display an approximate constant slope or even an up - bending shape in this region ; _ ii ) _ a significant and featureless drop long - ward of @xmath97 @xmath2 m is observed instead of the fe and mg absorption complexes , which is probably due to the low metallicity of these two galaxies , among the lowest of the entire sample .
these two features appear to be responsible for the anomalous colours , by making the spectral slope between 0.7 and 1.4 @xmath2 m too steep relative to the one in the optical range 0.550.70 @xmath2 m .
the fact that models fail to reproduce these colours may point to problems in the low metallicity regime , below @xmath98 .
we note that , according to the ma05 models , the largest enhancement and the most dramatic features in the nir are expected for @xmath99 . instead these two low - metallicity galaxies exhibit bluer optical - nir colours than even `` tp - agb light '' models ( e.g. , bc03 ) . as already mentioned in sec .
[ sec : intro ] , @xcite have obtained a detailed ( rest - frame ) uv - optical - nir sed for a sample of 62 psb galaxies observed in the _ newfirm _ medium band survey ( nmbs ) in the redshift range 0.72 .
their composite sed is essentially a stack of the sed of their sample .
they applied maximum likelihood fitting using both bc03 and ma05 on different wavelength ranges ( optical only or optical plus nir ) and , similarly to the results we have presented in this work , they found that ma05 can not reproduce the blue optical - nir colours observed , whereas bc03 provide an almost perfect match . in fig .
[ fig : comp_kriek ] we compare the sed published by @xcite , represented by the red points with error bars , with the stack of our 16 spectra .
both seds are normalised at 5500 .
the black line is the median of our 16 spectra at each wavelength , while the shaded area displays the 1684 inter - percentile range .
the similarity of the stacked seds from the two works is striking , typically consistent within 11.5 @xmath100 .
our spectra are slightly bluer and steeper in the range 5000 to @xmath101 m , which might derive from some relative bias in the respective sample selections .
, with our objects being somehow more extreme ( e.g. in terms of burst fraction , see also sec .
[ sec : summary ] ) .
in fact , @xcite define their sample based on k - corrected rest - frame colour cuts which are hardly comparable to our spectroscopic criteria .
alternatively , this comparison might indicate that nir - bright evolved ( tp - agb ) stars have a slightly larger influence on the sed of psb galaxies at @xmath102 than at @xmath103 , consistently with expectations , e.g. , of @xcite .
apart from this minor colour difference , our detailed spectroscopic study of individual psb galaxies confirms and reinforces the previous results by @xcite , based on stacking and medium band photometry and at a very different age of the universe .
we have presented the first spectroscopic study of post - starburst galaxies covering both the optical ( sdss ) and nir ( our new vlt - isaac observations ) wavelength ranges , which was specifically designed to detect the two main signatures of the presence of tp - agb stars predicted by the @xcite models in contrast to `` tp - agb light '' models ( such as bc03 ) : _ i ) _ the strong nir absorption features caused by carbon composite molecular band - heads ; and _ ii ) _ the boosting of the nir flux . to this goal , we have carefully selected a sample of 16 galaxies with post - burst ages corresponding to the peak contribution by tp - agb stars to the nir flux ( 0.51.5 gyr ) , in a broad range of stellar metallicities ( @xmath1042.5 @xmath8 ) .
these galaxies are at high enough redshift ( 0.150.25 ) to allow clean measurements of the nir features at rest wavelengths 1.41 and 1.77 @xmath2 m through the atmospheric h and k windows . despite these ideal conditions for the detection of the characteristic signatures of tp - agb stars predicted by the ma05 models , the 16 galaxies in our sample all display smooth , featureless nir spectra and blue optical - nir colours .
interestingly , these observed spectral properties are generally well reproduced using the bc03 models assuming simple ( i.e. single - age ) stellar populations .
we have investigated whether the observations could be reproduced with ma05 models as well , by combining stellar populations of different ages .
we find that diluting the light of a young ( i.e. a few 100 myr old ) starburst with that of a massive old component allows the ma05 models to reproduce the typical `` e+a '' optical spectra of post - starburst galaxies characterised by strong balmer absorption lines and a well developed d4000 break . at the same time , such models exhibit very weak nir features , in agreement with the observations , because tp - agb stars from the young component have not developed yet , while the nir emission from old stellar populations is smoother .
however , these models still can not account for the data , because the predicted ratio of nir to optical flux is significantly larger than observed at the metallicity of our galaxies . hence
, the observed lack of deep nir absorption features and the modest ratio of nir to optical flux in our data can not be simultaneously accounted for by a simple `` light dilution '' effect .
we note that the observation of metal absorption lines at optical wavelength does not support the idea that the faint nir continuum ( and the consequent dilution and/or absence of strong nir absorption features ) could arise from extremely low metallicities .
the ma05 models predict too red optical - nir colours for any choice of ratio between the old and the young component . on the contrary ,
models based on bc03 fail for `` old '' mass fractions as high as 90% , but produce colours that are consistent with ( most of ) the observations for `` old '' mass fractions up to 6070% , which leaves enough space to accommodate more reasonable sfhs than a ( possibly unrealistic ) single burst .
we note that the work of @xcite indicates old fractions close to 90% for their seven psb galaxies , which would be inconsistent with our seds .
however , this estimate might be biased by their assumption of fixed solar metallicity .
moreover , and probably more importantly , it is possible that our objects are somehow more extreme in terms of burst fraction , because , due to our redshift constraints , we select objects close to the faintest limit of the spectroscopic sample of the sdss . in this way we would select the intrinsically most luminous psb galaxies at @xmath103 , hence those with the largest burst fraction ( i.e. with the lowest `` old '' fraction )
: in fact , for a given total mass , a higher burst fraction yields higher luminosity . the lack of strong colour gradients in most of our psb galaxies lends support to this hypothesis , suggesting that the starburst substantially affected the full extent of the galaxy , rather than the centre only .
finally , @xcite point out that their sample is less extreme than typical samples of psb galaxies in the literature due to the small physical size of the spectroscopic fiber ( few hundred parsecs ) , which bias their selection towards central rather than extended starbursts . in conclusion
, the consistency of our seds with bc03-based csps and `` old '' mass fractions up to 6070% does not raise any problem for the physical interpretation of our psb galaxies using the bc03 models .
we also find that attenuation by interstellar dust , using standard prescriptions , can only worsen the disagreement ( when present ) between spectral evolution models and the observed optical - nir colours of post - starburst galaxies in our sample .
this is particularly true for the ma05 models , which predict systematically redder colours than observed .
we note that heavier - than - implemented attenuation by _
circumstellar _ dust of the most luminous c stars about to eject their envelopes could potentially reduce the contribution by these stars to the integrated spectrum and perhaps improve the agreement of the ma05 models with observations .
a more accurate theoretical and empirical assessment of the effects of circumstellar dust absorption is definitely key to make progress in the sed modelling of the stellar population affected by tp - agb stars ( e.g. * ? ? ?
* ; * ? ? ?
the results presented in this work confirm the previous claim by @xcite that bc03 , and more generally `` tp - agb light '' models , provide a better agreement with the spectral properties of galaxies hosting a stellar population at the peak of the tp - agb phase than ma05 models .
the important novelty of our work is to base these conclusions on a higher - resolution , high - quality spectrophotometric dataset and an optimal spectroscopic sample selection .
somehow in contrast with these studies and our work in particular , @xcite have detected the 1.77 @xmath2 m feature in a couple of magellanic cloud s globular clusters and found them in agreement with ma05 predictions .
this result calls for more observations in order to obtain a representative sample of globular clusters over a broader range of metallicities ( the two in which the detection was obtained are both at metallicity between 0.2 and 0.6 solar ) and beat down stochasticity effects : in fact , a single ( tp-)agb star can produce a substantial fraction of the total cluster nir luminosity in the mass range explored .
one further possibility to reduce the apparent disagreement between the @xcite results and ours might reside in a technical aspect of their challenging observations : the addition of targeted agb stars outside of the central region to integrate the central spectrum and obtain full spatial coverage of the globular clusters might lead to agb stars being over - represented with respect to the stars that contribute the bulk of the diffuse emission . in the light of this complex and partly controversial observational framework , the apparent agreement of the bc03 models ( of either simple or composite stellar populations ) with the observations presented in this paper should not be necessarily interpreted as an indication that the description of the tp - agb phase in these models is adequate .
new models incorporating more recent developments in the theory and observation of tp - agb stars ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) are being assembled ( charlot & bruzual , in preparation ) . by providing a better account of observations , new models of this type
will allow deeper insight into the physical properties of galaxies from spectral interpretations , especially at high redshift , where a large number of galaxies is going through the evolutionary phase affected by tp - agb stars . in this perspective , the present work provides key and very stringent constraints for the new generation of stellar population synthesis models .
we thank the anonymous referee for useful comments and suggestions that significantly improved the clarity of the presentation . + the dark cosmology centre
is funded by the danish national research foundation .
+ we also thank paolo cassata , harald kuntschner and mariya lyubenova for useful comments . s.z .
acknowledges the generous hospitality of the dark cosmology centre and stimulating discussions with jens hjorth and sune toft .
d.p . acknowledges the kind hospitality at the max - planck - institut fr extraterrestrische physik .
g. , 2007 , in vallenari a. , tantalo r. , portinari l. , moretti a. , eds , from stars to galaxies : building the pieces to build up the universe vol .
374 of astronomical society of the pacific conference series , stellar populations : high spectral resolution libraries . improved tp - agb treatment . | we present vlt - isaac near infrared ( nir ) spectro - photometric observations of 16 post - starburst galaxies aimed at constraining the debated influence of tp - agb stars on the spectral energy distribution ( sed ) of galaxies with stellar ages between 0.5 and 2 gyr , hence critical for high - redshift studies .
post - starburst galaxies are characterised by negligible on - going star formation and a sed dominated by the stellar population formed in a recent ( @xmath0 gyr ) burst . by selecting post - starburst galaxies with mean luminosity - weighted ages between 0.5 and 1.5 gyr and a broad range of metallicities ( based on sdss optical spectroscopy )
, we explore the parameter space over which the relative energy output of tp - agb stars peaks .
a key feature of the present study is that we target galaxies at @xmath1 , so that two main spectral features of tp - agb stars ( c - molecule band - head drops at 1.41 and 1.77 @xmath2 m , blended with strong telluric absorption features , hence hardly observable from the ground , for targets at @xmath3 ) move inside the h and k atmospheric windows and can be constrained for the first time to high accuracy .
our observations provide key constraints to stellar population synthesis models .
our main results are : _
i ) _ the nir regions around 1.41 and 1.77 @xmath2 m ( rest - frame ) are featureless for all galaxies in our sample over the whole range of relevant ages and metallicities at variance with the maraston ( 2005 ) `` tp - agb heavy '' models , which exhibit marked drops there ; _ ii ) _ no flux boosting is observed in the nir .
the optical - nir seds of most of our post - starburst galaxies can be consistently reproduced with bruzual & charlot ( 2003 ) models , using either simple stellar populations ( ssp ) of corresponding light - weighted ages and metallicities , or a more realistic burst plus an underlying old population containing up to approximately 60% of the total stellar mass .
in contrast , all combinations of this kind based on the maraston ( 2005 ) models are unable to simultaneously reproduce the smoothness of the nir spectra and the relatively blue optical - nir colours in the observations .
the data collected in this study appear to disfavour `` tp - agb heavy '' models with respect to `` tp - agb light '' ones .
[ firstpage ] galaxies : stellar content stars : agb and post - agb infrared : stars infrared : galaxies galaxies : general galaxies : photometry |
the extragalactic background light ( ebl ) is a measurement of the sum of the light produced by all extragalactic sources over cosmic time .
when it is integrated over the full spectral range , the so - called cosmic background is a fossil record of the overall activity of all galaxies from their birth until now .
it can be considered as the global energetic budget available for any model aiming at simulating the birth and fate of galaxies during the hubble time .
however the physical origin of this light will remain unknown until we have pinpointed the individual sources responsible for it .
the goal of the present paper is to demonstrate that an important new result has come from the combination of a series of deep extragalactic surveys performed in the mid - infrared ( mir ) at 15@xmath0 m with the isocam camera ( cesarsky et al .
1996a ) onboard the infrared space observatory ( iso , kessler et al .
1996 ) : we suggest here that the galaxies detected in these surveys , which median redshift of @xmath13 was measured from a sub - sample of isocam galaxies , contribute dominantly to the cosmic infrared background ( cirb ) , i.e. the ebl integrated over all wavelengths within @xmath14= 5 to 1000@xmath0 m .
the cirb was recently detected and measured thanks to the cosmic background explorer ( cobe ) instruments firas ( far infrared absolute spectrometer ) and dirbe ( diffuse infrared background experiment ) ( puget et al .
1996 , fixsen et al .
1998 , lagache et al .
1999 , 2000 , hauser et al .
1998 , dwek et al .
1998 , finkbeiner et al .
2000 ) from 100@xmath0 m to 1 mm .
it peaks around @xmath15 140@xmath0 m and was found to represent at least half and maybe two thirds of the overall cosmic background ( see gispert , lagache & puget 2000 ) .
hence the cirb reflects the bulk of the star formation that took place over the history of the universe . by resolving it into individual galaxies
, we would therefore pinpoint the times and places where most stars seen in the local universe were formed .
two physical processes were considered for its origin : nucleosynthesis , i.e. stellar radiation in star forming galaxies , and accretion around a black hole , i.e. active galactic nuclei . in both cases ,
the light is not directly coming from its physical source but is reprocessed by dust , i.e. absorbed and re - radiated thermally by the `` warm '' dust .
both processes are probably related ( see genzel et al .
1998 ) , but energetic considerations , based on the presence of massive black holes and on the amount of heavy elements in local galaxies , suggest that star formation should by far dominate in the cirb over agn activity ( madau & pozzetti 2000 , franceschini et al .
however , until the individual galaxies responsible for the cirb are found and studied in detail , this result will remain theoretical . the spectral energy distribution ( sed ) in the ir of local galaxies peaks above @xmath1060@xmath0 m and typically around 80 @xmath1 20@xmath0 m ( see sanders & mirabel 1996 ) .
as a result , the distant galaxies responsible for the peak of the cirb detected by cobe around @xmath16 @xmath10 140@xmath0 m should be located below @xmath4 1.3 and present a redshift distribution peaked around @xmath4 0.8 , if their seds do not strongly differ from those of local galaxies .
as we will see this is also the redshift range of the galaxies detected at 15@xmath0 m with isocam .
the isocam extragalactic surveys were performed with two filters , lw2 ( 5 - 8.5@xmath0 m ) and lw3 ( 12 - 18@xmath0 m ) , centered at 6.75 and 15@xmath0 m respectively . the 6.75@xmath0 m sample of sources is strongly contaminated by galactic stars , whereas stars are rather easily distinguished from galaxies at 15@xmath0 m using optical - mir colour - colour plots . as a consequence , we are only concerned here by the 15@xmath0 m galaxies . moreover ,
the observed 6.75@xmath0 m light is no more produced by dust emission for galaxies more distant than @xmath4 0.4 because of k - correction ( redshifted stellar light dominates the 6.75@xmath0 m band above this redshift ) , whereas the observed 15@xmath0 m light is mostly due to dust emission for galaxies up to @xmath4 2 .
about 1000 galaxies detected in the 15@xmath0 m surveys were used to produce number counts ( i.e. surface density of galaxies as a function of flux density ; see elbaz et al .
the steep slope of the 15@xmath0 m counts below @xmath17 mjy indicates the presence of an excess of faint sources by one order of magnitude in comparison with predictions assuming no evolution of the 15@xmath0 m luminosity function with redshift .
the presence of broad emission features in the mir spectrum of galaxies alone can not explain the shape of the number counts and a strong evolution of either the whole luminosity function ( xu 2000 , chary & elbaz 2001 ) or preferentially of a sub - population of starburst galaxies evolving both in luminosity and density ( franceschini et al . 2001 , chary & elbaz 2001 , xu et al . 2001 ) is required in order to fit the isocam 15@xmath0 m counts . in the present paper , we suggest that these isocam galaxies are in fact dusty starbursts responsible for the bulk of the cirb . in sect .
[ resolve ] , we compare the sensitivity of different extragalactic surveys in several wavelength ranges to detect the galaxies responsible for the cirb . it is suggested that mir is presently the most efficient technique to detect dusty starbursts up to @xmath4 1.3 . in sect .
[ ebl15 ] , we calculate the 15@xmath0 m integrated galaxy light ( igl ) due to isocam galaxies . the 15@xmath0 m igl is the sum of the 15@xmath0 m fluxes from individual galaxies , down to a given sensitivity limit , per unit area .
it represents a lower limit to the 15@xmath0 m ebl , which remains unknown .
once the redshift distribution and sed of these galaxies is determined , it becomes possible to estimate their contribution to the cirb . in sect .
[ mirtracer ] , we demonstrate that mir luminosities at 6.75 , 12 and 15@xmath0 m are strongly correlated with the bolometric ir luminosity ( from 8 to 1000@xmath0 m ) for local galaxies .
the correlations presented in chary & elbaz ( 2001 ) are confirmed here with a larger sample of galaxies . before spectroscopic redshifts
are obtained for the full sample of isocam galaxies used to produce these number counts , the redshift distribution of these galaxies can be inferred from a few sub - samples : hdfn ( aussel et al . 1999 , 2001 ) , cfrs-14 ( flores et al .
1999 , 2002 ) , cfrs-03 ( flores et al . 2002 ) .
the ultra - deep isocam survey of the hdfn samples a flux density range where most of the evolution observed in the number counts takes place and where the bulk of the 15@xmath0 m igl is produced .
this field is complete in spectroscopic redshifts , so it is used to estimate the bolometric ir luminosities and star formation rates of the isocam galaxies in sect .
[ bolir ] .
this result relies on two assumptions : + - that the main source for the mir light in isocam galaxies is star formation and not accretion around a black hole .
+ - that the correlations found in the local universe between the mir and bolometric ir luminosity of galaxies remain valid up to @xmath4 1 .
the first assumption is discussed and justified in sect .
[ agnfrac ] , where soft and hard x - ray data from the chandra and xmm - newton x - ray observatories are combined with isocam data on galaxies in the hdfn and lockman hole regions respectively .
the issue of the robustness of the mir - fir correlations in the distant universe is addressed in sect .
[ radio ] , where ir luminosities are also computed from radio ( 1.4 ghz ) flux densities for a sub - sample of 24 distant and 109 local isocam galaxies . in sect .
[ ldens ] , we compute the cosmic density of ir light due to luminous ir galaxies ( @xmath18 ) at @xmath19 1 . in sect .
[ cirb ] , we evaluate the contribution of the isocam galaxies to the cirb , more precisely to its peak emission around @xmath20 140@xmath0 m . finally , the nature of isocam galaxies is discussed in the conclusions ( sect .
[ conclusions ] ) . in the following
, we will use the terms ulig for galaxies with an ir luminosity @xmath21\geq 10^{12}~l_{\sun}$ ] , lig , when @xmath22 and luminous ir galaxies for both ( @xmath23 ) . throughout this paper
, we will assume h@xmath24= 75 km s@xmath3 mpc@xmath3 , @xmath25= 0.3 and @xmath26 .
one of the major results from the iras satellite was to demonstrate that the bolometric luminosity of galaxies more luminous than @xmath27 was underestimated by more than one order of magnitude before the ir luminosity was accounted for ( soifer et al .
these luminous ir galaxies are not typical of the local galaxy population , since local galaxies radiate only about 30@xmath5 of their bolometric luminosity in the dust regime from 8 to 1000@xmath0 m and only @xmath10 2@xmath5 of the local bolometric luminosity density is due to luminous ir galaxies . if we convert the ir luminosity into a star formation rate ( sfr ) using the formula of kennicutt ( 1998 ) , @xmath28 ( l_{\sun } ) \label{eqsfr}\ ] ] we find that luminous ir galaxies ( @xmath29 ) form stars at a rate larger than @xmath1020@xmath11 yr@xmath3 .
( [ eqsfr ] ) assumes continuous bursts lasting 10 - 100 myr , solar abundance and a salpeter imf ( kennicutt 1998 ) .
iras surveyed about 95@xmath5 of the sky down to a completeness limit of @xmath10 0.5 mjy at 60@xmath0 m ( @xmath10 1.5 jy at 100@xmath0 m ) and the differential counts are well fitted by an euclidean slope , @xmath30 ( soifer et al .
the fluxes of all galaxies detected down to this sensitivity limit by iras add up to a 60@xmath0 m igl of @xmath10 0.15 nw m@xmath2 sr@xmath3 .
this is less than 1@xmath5 of the value of the cirb measured by cobe at @xmath20 140@xmath0 m , @xmath31= ( 25@xmath17 ) nw m@xmath2 sr@xmath3 .
deeper fir extragalactic surveys were performed with isophot ( lemke et al . 1996 ) onboard iso at 170@xmath0 m . at this depth ,
the differential counts are no more fitted by an euclidean slope , expected in the case of no evolution of galaxies with redshift , but instead a strong excess of faint sources was found . at the confusion limit of isophot of about 120 mjy , less than 10@xmath5 of the value of the cirb at 170@xmath0 m is resolved into individual galaxies ( dole et al .
a preliminary follow - up of the isophot galaxies suggests that a large fraction of these galaxies are located at low redshift . in the sub - millimeter range ,
the scuba bolometer array on the james clerk maxwell telescope ( jcmt ) has produced deep images at 450 and 850@xmath0 m .
number counts at both wavelengths present a steep slope compatible with a strong evolution as compared to the local universe ( see smail et al .
2001 and references therein ) . at 450@xmath0 m ,
a depth of 10 mjy is reached and the combined fluxes of all scuba galaxies produce about 15@xmath5 of the cirb measured by cobe - firas at this wavelength ( smail et al .
2001 ) . at 850@xmath0 m
, scuba is confusion limited at @xmath10 2 mjy ( hughes et al .
1998 , barger , cowie & sanders 1999 , eales et al .
2000 , smail et al . 2001 ) , because of its large point spread function ( psf ) of 15@xmath32 full width half maximum ( fwhm ) .
about 20@xmath5 of the value of the cirb measured by cobe - firas at 850@xmath0 m ( the 850@xmath0 m ebl ) is resolved into galaxies at this depth . however , using gravitational lensing this limit can be lowered to @xmath10 1 mjy , where @xmath10 60@xmath5 of the cirb is resolved ( smail et al .
2001 , blain et al .
1999a ) .
the 850@xmath0 m ebl measured by cobe - firas ( @xmath10 [email protected] nw m@xmath2 sr@xmath3 ) is 50 times lower than the peak value of the cirb at @xmath20 140@xmath0 m . in order to compute the contribution of scuba galaxies to the peak of the cirb
it is therefore necessary to determine their redshift distribution and sed . until now
, very few redshifts have been obtained due to the large psf of scuba and optical faintness of these galaxies .
hence the fraction of the cirb resolved into individual galaxies by scuba remains highly uncertain .
the deepest isocam surveys reach a completeness limit of @xmath10 100@xmath0jy at 15@xmath0 m ( without lensing , but this limit goes down to @xmath10 30 - 50@xmath0jy including lensing , see sect . [ ebl15 ] ) . for
a given galaxy sed and redshift , this flux density can be converted into a bolometric ir luminosity ( @xmath33[8 - 1000@xmath0 m ] ) .
any galaxy more luminous than this @xmath6 will be detected in the survey . in order to compare the sensitivity to distant luminous ir galaxies of isocam , isophot and scuba , we have calculated @xmath6 corresponding to their sensitivity limits for a given redshift between @xmath34=0 and 3 . on one hand , we used the sed of m 82 ( fig .
[ fig : sedsm82a220]a ) for galaxies of all luminosities . on the other hand
, we used a library of template seds from chary & elbaz ( 2001 ) . in the following
, we will call it the `` multi - template '' technique .
this library of 105 seds was constructed under the constraint that it reproduces the correlations observed between the 6.75 , 12 , 15 , 60 , 100 and 850@xmath0 m luminosities of local galaxies . the seds were interpolated from a sample of observed seds , including mir spectra obtained with the isocam cvf .
they cover the luminosity range @xmath35= 8.5 to 13.5 ( see fig .
[ fig : sedsm82a220]b ) .
we did not use arp 220 s sed in fig .
[ fig : sedsm82a220]a , because this galaxy presents a large fir over mir luminosity ratio , which is not representative of galaxies within the same luminosity range ( see figs.[fig : correl]c , d ) .
the radio continuum ( @xmath14= 21 cm , @xmath36= 1.4 ghz ) is also a tracer of @xmath6 because of the tight correlation between both luminosities in the local universe ( see yun , reddy & condon 2001 and references therein ) .
if this correlation remains valid in the distant universe then we can translate the sensitivity limit of the deepest radio surveys into a minimum @xmath6 accessible at a given redshift assuming a radio sed .
the correlation between the radio and fir luminosities is usually described by the `` @xmath37 '' parameter ( condon et al . 1991 ) : @xmath38 where @xmath39 is the monochromatic luminosity at 1.4 ghz and @xmath40 is the fir luminosity between 40 and 120@xmath0 m , as defined by helou et al .
( 1988 ) , @xmath41 where @xmath42 and @xmath43 are the iras flux densities at 60 and 100@xmath0 m in jy .
the relationship between @xmath44[40 - 120@xmath0 m ] and @xmath6[8 - 1000@xmath0 m ] was computed with the sample of 300 galaxies from the bright galaxy sample ( bgs , soifer et al .
1987 ) detected in the four iras bands ( 12 , 25 , 60 and 100@xmath0 m ) : @xmath45 @xmath6 is defined as ( sanders & mirabel 1996 ) : @xmath46$\\ & & $ \times\,1.8\times10^{-14}~\times4\pi d$(m)$^2 $ \label{eq : ir } \end{tabular}\ ] ] where @xmath47 and @xmath48 are the iras flux densities at 12 and 25@xmath0 m in jy .
yun , reddy & condon ( 2001 ) measured a value of @xmath37= [email protected] from a flux limited ( @xmath49 2 jy ) sample of 1809 iras galaxies .
the deepest radio surveys presently available were performed in the hdfn by richards ( 2000 ) and garrett et al .
( 2000 ) with the vla and wsrt respectively .
both surveys reach the same sensitivity limit of 40@xmath0jy at 1.4 ghz ( 5-@xmath50 ) . in order to convert this flux density into a @xmath51 for a given redshift , we used a power index of @xmath52= [email protected] ( @xmath53 ) , as suggested by yun , reddy & condon ( 2001 ) for starburst galaxies .
[ fig : sensm82 ] presents the sensitivity of the deepest mir , fir , sub - millimeter and radio surveys in the form of l@xmath54 ( or sfr ) as a function of redshift .
the sensitivity limits used for isocam , isophot and scuba are taken from the deepest existing surveys at these wavelengths , as previously described .
[ fig : sensm82]a ( with m 82 ) and fig .
[ fig : sensm82]b ( `` multi - template '' technique ) both clearly show that the faintest ir galaxies are best detected at 15@xmath0 m up to @xmath4 1.3 .
the right axis of the plots shows the corresponding minimum sfr that a galaxy must harbor to be detected at a given redshift in the surveys . at @xmath4 1 , isocam ( plain line ) detects all ligs while scuba ( dot - dashed line ) detects only uligs .
the difference in sensitivity between isocam and the deepest radio surveys is about a factor 2 .
isophot ( dotted line ) is only sensitive to uligs above @xmath4 0.5 . in fig . [
fig : sensm82]b , where we used the library of template seds from chary & elbaz ( 2001 ) , the plain line corresponding to isocam rises faster above @xmath4 1 than in fig .
[ fig : sensm82]a , where we used m 82 s sed .
this different behavior comes from the fact that at these redshifts isocam measures fluxes at about 7@xmath0 m in the rest - frame of the galaxy and we will show that @xmath6/@xmath55 increases with @xmath6 ( sect .
[ mirtracer ] ) . above z@xmath102 ,
the sub - millimeter becomes the most efficient technique , although only galaxies more luminous than a few 10@xmath56 l@xmath57 can be detected above a sensitivity limit of 2 mjy at 850@xmath0 m .
hence the unlensed isocam and scuba surveys are not sampling the same redshift and luminosity ranges .
this statement is confirmed observationally in the hdfn itself where none of the isocam sources are detected by scuba ( hughes et al . 1998 ) and over a larger scale in the canada france redshift survey 14 ( cfrs-14 , eales et al .
2000 ) . in this latter field of @xmath1050@xmath58 ,
only the two brightest 15@xmath0 m isocam sources are detected at 850@xmath0 m , among a sample of 50 isocam ( flores et al .
1999 ) and 19 scuba sources .
this confirms that both instruments detect different sets of objects .
more common objects between isocam and scuba are expected if they are gravitationally lensed , since the sensitivity limits of both instruments are then decreased by a factor of about two . indeed in the clusters a 370 and a 2390 , three of the four lensed scuba galaxies ( smail et al .
2001 ) are also detected with isocam ( altieri et al .
1999 , metcalfe 2000 ) .
above the earth s atmosphere , the 15@xmath0 m light is strongly dominated by the zodiacal emission from interplanetary dust and it has not yet been possible to make a direct measurement of the 15@xmath0 m background , or ebl .
individual galaxies contribute to this background and a lower limit to the 15@xmath0 m ebl can be obtained by adding up the fluxes of all isocam galaxies detected per unit area down to a given flux limit .
the resulting value is called the 15@xmath0 m integrated galaxy light ( igl ) . the statistical reliability with which the surface density of mir sources is computed depends on their flux : large and shallow fields are required for bright sources which are less numerous , whereas faint sources require smaller and deeper fields .
the optimization is obtained by observing similar volumes of the sky but in pencil beams for faint sources and in wide surveys for bright sources , instead of one large and deep survey which would be unnecessarily time consuming .
together with the need to account for cosmic variance , this explains why we have combined a set of 10 surveys located in 7 regions of the northern and southern hemispheres , in order to estimate the 15@xmath0 m igl .
these surveys cover five orders of magnitude in flux density from @xmath59 @xmath0jy to @xmath60 jy ( including iras counts at 12@xmath0 m )
. the number counts presented in elbaz et al .
( 1999 ) are converted here into a differential contribution to the 15@xmath0 m igl as a function of flux density , estimated from the following formula : @xmath61 where @xmath62(sr@xmath3 ) is the surface density of sources with a flux density @xmath63[15@xmath0m]@xmath64 ( mjy ) over a flux density bin @xmath65(mjy ) ( 1 mjy= 10@xmath66 nw m@xmath2 hz@xmath3 ) and @xmath67(hz ) is the frequency of the 15@xmath0 m photons .
below @xmath68 3 mjy , about 600 galaxies were used to produce the points with errors bars in fig .
[ fig : debl]a . fig .
[ fig : debl]b shows the 15@xmath0 m igl as a function of depth .
it corresponds to the integral of fig .
[ fig : debl]a , where the data below 3 mjy were fitted with a polynomial of degree 3 and the 1-@xmath50 error bars on @xmath69 were obtained from the polynomial fit to the upper and lower limits of the data points .
the 15@xmath0 m igl does not converge above a sensitivity limit of @xmath68 50@xmath0jy , but the flattening of the curve below @xmath70 mjy suggests that most of the 15@xmath0 m ebl should arise from the galaxies already unveiled by isocam . above the completeness limit of @xmath71jy ,
we computed an igl of : @xmath72 where the error bar corresponds to the 68@xmath5 confidence level ( i.e. 1-@xmath50 ) .
the error bar combines the uncertainty on the measurements for each individual survey with cosmic variance .
each flux bin is covered by two to three independent surveys .
isocam number counts are only complete above @xmath68 100@xmath0jy , but aussel et al .
( 1999 ) computed number counts down to @xmath71jy after correcting for incompleteness and a consistent result is obtained by altieri et al .
( 1999 ) after correcting for lensing magnification a sample of galaxies in the field of the galaxy cluster abell 2390 .
only 9 sources are detected below @xmath71jy by altieri et al .
( 1999 ) , where the flux determination and correction for completeness are less robust and are not confirmed by another survey . a tentative value of @xmath73 3.3 @xmath1 1.3 nw m@xmath2 sr@xmath3 was quoted in altieri et al .
( 1999 ) , above @xmath74= 30@xmath0jy . this value may be slightly overestimated if we consider that the models of franceschini et al .
( 2001 ) and chary & elbaz ( 2001 ) , which reproduce the number counts from isocam at 15@xmath0 m , from isophot at 90 and 170@xmath0 m and from scuba at 850@xmath0 m , as well as the shape of the cirb from 100 to 1000@xmath0 m , consistently predict a 15@xmath0 m ebl of : @xmath75 if this prediction from the models is correct then about @xmath76 of the 15@xmath0 m ebl is resolved into individual galaxies by the isocam surveys .
this result is consistent with the upper limit on the 15@xmath0 m ebl estimated by stanev & franceschini ( 1998 ) of : @xmath77 this upper limit was computed from the 1997 @xmath78-ray outburst of the blazar mkn 501 ( @xmath79 0.034 ) as a result of the opacity of mir photons to @xmath78-ray photons , which annihilate with them through @xmath80 pair production .
it was recently confirmed by renault et al .
( 2001 ) , who found an upper limit of 4.7 nw m@xmath2 sr@xmath3 from 5 to 15@xmath0 m .
finally , the 15@xmath0 m background ( ebl ) must be contained between the following limits : @xmath81 .relative contributions to the 15@xmath0 m igl as a function of flux density . ref . :
( 1 ) altieri et al .
( 1999 ) , metcalfe ( 2000 ) , ( 2 ) aussel et al .
( 1999 , 2001 ) , ( 3 ) elbaz et al .
( 1999 ) , ( 4 ) serjeant et al .
( 2000 ) , renormalized by a factor 0.5 as in genzel & cesarsky ( 2000 ) , following suggestion from s.serjeant for photometric calibration , ( 5 ) interpolation in a flux density range where we do not have enough statistics .
we used the same slope as in the iras counts , ( 6 ) iras counts from rush , malkan & spinoglio ( 1993 ) at 12@xmath0 m . [ cols="<,<,<,<",options="header " , ] [ tab : cirb ]
the cosmic ir background is a fossil record of the light radiated by galaxies since their formation .
it peaks around @xmath82 140@xmath0 m whereas the spectral energy distribution of galaxies peaks above 60@xmath0 m .
this suggests that the bulk of the cosmic ir background is due to galaxies located below @xmath4 1.3 .
we have shown that the best technique currently available to unveil dusty galaxies up to @xmath4 1.3 is provided by the isocam mir extragalactic surveys .
we have computed the contribution of isocam galaxies to the 15@xmath0 m background , the 15@xmath0 m integrated galaxy light , and found a value of @xmath83 ( [email protected] ) nw m@xmath2 sr@xmath3 .
this is about ten times below the cosmic background measured by cobe at @xmath84 140@xmath0 m : @xmath85 ( 25@xmath17 ) nw m@xmath2 sr@xmath3 .
we have demonstrated that the mir luminosities at 6.75 , 12 and 15@xmath0 m were correlated with each other and with the bolometric ir luminosity for local galaxies .
this suggests that the contribution of isocam galaxies to the cirb can be computed from @xmath86 , unless distant galaxies seds strongly differ from local ones .
the redshift distribution of isocam galaxies was measured from the spectroscopically complete sample of galaxies in the region of the hdfn .
this redshift distribution is consistent with the twice larger sample of isocam galaxies detected in the cfrs fields cfrs-14 and cfrs-03 ( flores et al .
1999 , 2002 ) .
at the median redshift of @xmath87 0.8 , the observed 15 and 140@xmath0 m wavelengths correspond to about 7@xmath0 m ( isocam - lw2 filter ) and 80@xmath0 m ( iras bands ) in the rest - frame of the galaxies .
luminosities at both wavelengths are correlated ( see fig . [
fig : correl]d ) .
if the correlations between mir and fir luminosities remain valid up to @xmath4 1 , then they can be used to compute @xmath31 .
we have checked with a sample of galaxies detected both in the mir with isocam and in the radio with the vla and wsrt , that the mir - fir and radio - fir correlations are consistent up to @xmath4 1 .
this comparison independently validates our estimate of the bolometric ir luminosity of the isocam galaxies , although it is not clear whether the radio - fir correlation works also up to @xmath4 1 .
the fraction of active nuclei responsible for the 15@xmath0 m luminosity of isocam galaxies was estimated from the deepest soft and hard x - ray surveys available at present by the xmm - newton and chandra x - ray observatories in the lockman hole and hdfn respectively ( fadda et al . 2001 ) .
it was found that about ( 12@xmath15)@xmath5 of the isocam galaxies are powered by an agn and that the agn contribution to @xmath86 was about ( 17@xmath16)@xmath5 . the agn contribution to @xmath31
was found to be as low as @xmath10 4@xmath5 assuming that they all share the sed of the local seyfert 2 , ngc 1068 .
this is a conservative choice since ngc 1068 presents the flattest ir sed that we know .
however , we note that the cosmic x - ray background ( cxb ) peaks around 30 kev ( see fig.1 of wilman , fabian & nulsen 2000 ) , while both xmm - newton and chandra were limited to energies below 10 kev .
it is therefore possible that a population of hard x - ray agns was missed by these surveys .
but as claimed by the authors of these deep x - ray surveys , the bulk of the cxb had been resolved into individual galaxies in the lockman hole and hdfn images . moreover , using estimates of the present comoving density of black holes , madau & pozzetti ( 2000 ) calculated that less than 20@xmath5 of the cirb could be due to dusty agns . for the remaining star forming galaxies
, we used a library of template seds , reproducing the mir - fir correlations , to compute their ir luminosity and contribution to @xmath31 .
we find that ligs and uligs produce about 60@xmath5 of @xmath86 .
the comoving density of ir luminosity produced by these luminous ir galaxies was about ( 70@xmath135 ) times larger at @xmath4 1 than today , while in the same redshift interval the @xmath88-band or uv luminosity densities only decreased by a factor @xmath10 3 .
since the ir luminosity measures the dusty star formation rate of a galaxy , this also implies that the comoving density of star formation , due to luminous ir galaxies , decreased by a similar factor in this redshift range , i.e. much more than expected by studies at other wavelengths affected by dust extinction .
this indicates that a large fraction of present day stars were formed during a dusty starburst event .
we estimate a contribution of isocam galaxies to the peak of the cirb at @xmath15 140@xmath0 m of ( 16@xmath15 ) nw m@xmath2 sr@xmath3 as compared to the measured value of ( 25@xmath17 ) nw m@xmath2 sr@xmath3 from cobe .
this study therefore suggests that the isocam galaxies are responsible for the bulk of the cirb .
we have started a systematic spectroscopic follow - up of these galaxies with the aim of studying their physical properties and the origin of their large sfr .
franceschini et al .
( 2001 ) estimated their baryonic masses to be of the order of @xmath89 by fitting their optical and near - ir luminosities with template seds ( from silva et al .
1998 ) and assuming a salpeter initial mass function ( from 0.15 to 100@xmath11 ) .
their colors are similar to field galaxies of similar magnitudes ( cohen 2001 ) , hence they could not have been selected on the basis of their optical colors .
the technique which consists in using the spectral slope in the uv domain to correct luminosities from extinction ( meurer , heckman & calzetti 1999 ) fails to differentiate the luminous dusty galaxies detected with isocam from other field galaxies in the hdfn+ff ( cohen 2001 ) .
this was to be expected since this technique only works for galaxies with @xmath90 ( meurer et al .
2000 ) while most isocam galaxies are more luminous than this threshold . a property of the isocam galaxies that may give a hint on their origin is their association with small groups of galaxies . a preliminary study of 22 isocam - hdfn galaxies by cohen et al .
( 2000 ) found that nearly all isocam galaxies belonged to small groups , while the fraction of field galaxies with similar optical magnitudes belonging to such groups was 68@xmath5 .
the study of the full sample of isocam - hdfn galaxies by aussel et al .
( 2001 ) shows that all of them belong to physical groups , hence suggests that dynamical effects such as merging and tidal interactions are responsible for their large sfr . the space ir telescope facility ( sirtf ) will soon provide a powerful insight on the fir emission of distant dusty starbursts and its 24@xmath0 m band , less affected by confusion ,
should prolongate their detection up to @xmath91 .
extending the redshift range surveyed by isocam to such redshifts is particularly important to measure the direct ir emission of the distant population of lyman break galaxies , whose dust extinction remains highly uncertain ( steidel et al .
however , the direct measurement of the mir and fir emission of distant galaxies will only become possible with the launch of the next generation space telescope ( ngst ) and herschel ( first ) satellite scheduled for 2009 and 2007 .
de wishes to thank the american astronomical society for its support through the chretien international research grant and joel primack and david koo for supporting his research through nasa grants nag5 - 8218 and nag5 - 3507 .
we wish to thank h.flores and a.blain for fruitful comments and a.boselli for sharing his isocam catalog with us .
this research has made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
allamandola , l.j . ,
tielens , a.g.g.m . , barker , j.r .
1989 , apjs 71 , 733 almaini , o. , lawrence , a. , boyle , b.j .
1999 , mnras 305 , l59 altieri , b. , metcalfe , l. , kneib , j.p . ,
1999 , a&a 343l , 65 anantharamaiah , k.r . ,
viallefond , f. , mohan , n.r . , goss , w.m . ,
zhao , j.h .
2000 , apj 537 , 613 andriesse , c.d . , 1978 , a&a 66
, 169 aussel , h. , cesarsky , c.j . ,
elbaz , d. , starck , j.l .
1999 , a&a 342 , 313 aussel , h. , coia , d. , mazzei , p. , de zotti , g. , franceschini , a. 2000 , a&as 141 , 257 aussel , h. , et al .
2001 , in preparation barger , a.j .
, cowie , l.l . , sanders , d.b . 1999 , apj 518 , l5 bernstein , r.a . ,
freedman , w. l. , madore , b.f .
2001 , apj , accepted ( astro - ph/0112193 ) blain , a.w . ,
smail , i. , ivison , r.j . ,
kneib , j .- p .
1999a , mnras 302 , 632 blain , a.w . , kneib , j .- p . , ivison , r.j . , smail , i. 1999b , apj 512 , l87 boselli , a. , lequeux , j. , sauvage , m. , et al .
1998 , a&a 335 , 53 brandt , w.n .
, hornschemeier , a.e . , alexander , d.m .
2001 , aj 122 , 1 brandt , w.n . ,
alexander , d.m . ,
hornschemeier , a.e . , et al .
2002 , aj , in press ( astro - ph/0108404 ) cambrsy , l. , reach , w.t . , beichman , c.a . ,
jarrett , t.h .
2001 , apj 555 , 563 cesarsky , d. , lequeux , j. , abergel , a. , et al .
1996a , a&a 315 , l305 cesarsky , c.j . , abergel , a. , agnse , p. , et al .
1996b , a&a 315 , l32 charmandaris , v. , laurent , o. , mirabel , i.f .
, et al . 1999 ,
ap&ss 266 , 99 chary , r.r . , elbaz , d. 2001 , apj 556 , 562 cohen , j.g . ,
hogg , d.w .
, blandford , r. , et al .
2000 , apj 538 , 29 cohen , j.g .
2001 , aj 121 , 2895 coleman , g. d. , wu , c .- c . , weedman , d. w. 1980 , apjs 43 , 393 condon , j.j . , anderson , m.l .
, helou , g. 1991 , apj 376 , 95 condon , j.j . , 1992 ,
ara&a 30 , 575 condon , j.j . ,
cotton , w.d .
, greisen , e.w . , et al .
1998 , aj 115 , 1693 connolly , a.j . , szalay , a.s . , dickinson , m.e . , et al .
1997 , apj 486 , l11 cowie , l.l . ,
songaila , a. , barger , a.j .
1999 , apj 118 , 603 dale , d.a . ,
silbermann , n.a . ,
helou , g. , et al .
2001 , aj 120 , 583 dsert , f .- x .
, boulanger , f. , puget , j .- l .
1990 , a&a 237 , 215 de vaucouleurs , g. , de vaucouleurs , a. , corwin , h.g.jr . , et al .
1991 , third reference catalogue of bright galaxies , ( new york : springer - verlag ) ( rc3 ) dole , h. , gispert , r. , lagache , g. , et al .
2001 , a&a 372 , 364 dwek , e. , arendt , r.g . , hauser , m. , et al .
1998 , apj 508 , 106 dwek , e. , arendt , r.g . 1998 , apj 508 , l9 eales , s. , lilly , s. , webb , t. , et al . 2000 ,
aj 120 , 2244 elbaz , d. , cesarsky , c.j . ,
fadda , d. , et al .
1999 , a&a 351 , l37 elbaz , d. , flores , h. , chanial , p. , et al .
2002 , a&a 381 , l1 fadda , d. , elbaz , d. , duc , p .- a .
2000 , a&a 361 , 827 fadda , d. , flores , h. , hasinger , g. , et al .
2001 , a&a in press ( astro - ph/0111412 ) finkbeiner , d.p . ,
davis , m. , schlegel , d.j .
2000 , apj 524 , 867 fischer , j. , luhman , m.l . , satyapal , s. 1999 , ap&ss 266 , 91 fixsen , d. j. , dwek , e. , mather , j.c . ,
bennett , c.l . , shafer , r.a .
1998 , apj 508 , 123 flores , h. , hammer , f. , thuan , t.x . ,
et al . 1999 , apj 517 , 148 flores , h. , et al .
2002 , in preparation frster schreiber , n.m . , laurent , o. , sauvage , m. , et al . 2000 , a&a , in preparation frster schreiber , n.m . , genzel , r. , lutz , d. , kunze , d. , sternberg , a. 2001 , apj 552 , 544 franceschini , a. , aussel , h. , cesarsky , c.j . ,
elbaz , d. , fadda , d. 2001 , a&a 378 , 1 franceschini , a. , fadda , d. , cesarsky , c.j .
, elbaz , d. , flores , h. , granato , g.l .
2002 , apj accepted ( astro - ph/0111413 ) garrett , m.a . , de bruyn , a.g . ,
giroletti , m. , et al .
2000 , a&a 361 , l41 gorjian , v. , wright , e. l. , chary , r.r .
2000 , apj 536 , 550 genzel , r. , lutz , d. , sturm , e. , et al .
1998 , apj 498 , 579 genzel , r. , cesarsky , c.j .
2000 , ara&a 38 , 761 gispert , r. , lagache , g. , puget , j.l .
2000 , a&a 360 , 1 haas , m. , klaas , u. , mller , s.a.h . ,
chini , r. , coulson , i. 2001 , a&a 367 , l9 hasinger , g. , altieri , b. , arnaud , m. , et al .
2001 , a&a 365 , l45 hauser , m.g . ,
arendt , r.g . ,
kelsall , t. , et al .
1998 , apj 508 , 25 helou , g. , khan , i.r . ,
malek , r. , boehmer , l. 1988 , apjs 68 , 151 hornschemeier , a.e . ,
brandt , w.n .
, garmire , g.p . ,
et al . 2001 , apj 554 , 742 hughes , d. h. , gear , w.k . , robson , e.i .
1994 , mnras 270 , 641 hughes , d. h. , serjeant , s. , dunlop , j. , et al .
1998 , nature 394 , 241 ikebe , y. , leighly , k. , tanaka , y. , et al .
2000 , mnras 316 , 433 jones , a.p . , dhendecourt , l. 2000 , a&a 355 , 1191 kennicutt , r.c.jr 1998 , ara&a 36 , 189 kessler , m. , steinz , j. , anderegg , m. , et al .
1996 , a&a 315 , l27 klaas , u. , haas m. , heinrichsen , i. , schulz , b. 1997 , a&a 325 , l21 lagache , g. , abergel , a. , boulanger , f. , desert , f .- x . , puget , j .- l .
1999 , a&a 344 , 322 lagache , g. , haffner , l.m .
, reynolds , r.j . , tufte , s.l .
2000 , a&a 354 , l247 laurent , o. , 1999 , phd thesis , universit de paris xi .
laurent , o. , mirabel , i.f . , charmandaris , v. , et al . 2000 ,
a&a 359 , 887 le floch , e. , mirabel , i.f . , laurent , o. , et al .
2001 , a&a 367 , l487 lger , a. , puget , j.l . , 1984 ,
a&a 137 , l5 lemke , d. , klaas , u. , abolins , j. , et al .
1996 , a&a 315 , l64 lilly , s.j . , lefvre , o. , hammer , f. , crampton , d. 1996 , apj 460 , l1 madau , p. , pozzetti , l. 2000 ,
mnras 312 , l9 matsuhara , h. , kawara , k. , sato , y. , et al . 2000 ,
a&a 361 , 407 metcalfe , l. 2000 , in `` the extragalactic infrared background and its cosmological implications '' , iau symposium 204 , p.18 meurer , g.r . ,
heckman , t.m .
, calzetti , d. , 1999 , apj 521 , 64 meurer , g.r . ,
heckman , t.m .
, seibert , m. , et al .
2000 , to appear in proc .
jd9 , iau manchester 2000 , ` cold gas and dust at high redshift ' ed .
d.j.wilner , highlights of astronomy vol . 12 .
pasp ( astro - ph/0011201 ) mirabel , i.f . ,
vigroux , l. , charmandaris , v. , et al .
1998 , a&a 333 , l1 moutou , c. , verstraete , l. , lger , a. , sellgren , k. , schmidt , w. 2000 , a&a 354 , l17 okumura , k. , 1998 , esa isocam psf report , www.iso.vilspa.esa.es/users/expl@xmath92lib/cam/[email protected] pozzetti , l. , madau , p. , zamorani , g. , ferguson , h.c . , bruzual , g.a .
1998 , mnras 298 , 1133 pozzetti , l. , madau , p. 2001 , in `` the extragalactic infrared background and its cosmological implications '' , iau symposium 204 , eds .
m.harwit and m.g.hauser ( astro - ph/0011359 ) puget , j .- l . , lger , a. , 1989 , ara&a 27 , 37 puget , j.l . , abergel , a. , bernard , j.p . , et al .
1996 , a&a 308 , l5 puget , j.l . , lagache , g. , clements , d. , et al .
1999 , a&a 345 , 29 puget , j.l . , lagache , g. 2001 , in `` the extragalactic infrared background and its cosmological implications '' , iau symposium 204 , eds .
m.harwit and m.g.hauser , p.19 ( astro - ph/0101105 ) renault , c. , barrau , a. , lagache , g. , puget , j .- l .
2001 , a&a 371 , 771 richards , e.a .
2000 , apj 533 , 611 rigopoulou , d. , spoon , h.w.w . ,
genzel , r. , et al .
1999 , aj 118 , 2625 roussel , h. , vigroux , l. , bosma , a. , et al .
2001 , a&a 369 , 473 rush , b. , malkan , m.a . , spinoglio , l. 1993 , apjs 89 , 1 sanders , d.b . , mirabel , i.f .
1996 , ara&a 34 , 749 serjeant , s. , oliver , s. , rowan - robinson , m. , et al .
2000 , mnras 316 , 768 silva , l. , granato , g.l . ,
bressan , a. , danese , l. 1998 , apj 509 , 103 smail , i. , ivison , r.j . ,
blain , a.w .
, kneib , j .- p .
2001 , mnras in press ( astro - ph/0112100 ) smith , c.h . ,
aitken , d.k . , & roche , p.f .
1989 , mnras 241 , 425 soifer , b.t . , sanders , d.b . ,
madore , b.f . , et al . 1987 , apj 320 , 238 stanev , t. , franceschini , a. 1998 , apj 494 , l159 steidel , c.c . ,
adelberger , k.l . ,
giavalisco , m. , dickinson , m. , pettini , m. 1999 , apj 519 , 1 telesco , c.m . ,
becklin , e.e .
, wynn - williams , c.g . , harper , d.a .
1984 , apj 282 , 427 tran , q.d . ,
lutz , d. , genzel , r. , et al . 2001 , apj 552 , 527 xu , c. , hacking , p.b . ,
fang , f. , et al .
1998 , apj 508 , 576 xu , c. 2000 , apj 541 , 134 xu , c. , lonsdale , c.j . ,
shupe , d.l .
, olinger , j. , masci , f. 2001 , apj 562 , 179 vigroux , l. , mirabel , f. , altieri , b. , et al .
1996 , a&a 315 , l93 wilman , r.j . ,
fabian , a.c .
, nulsen , p.e.j .
2000 , mnras 319 , 583 wright , e.l .
2000 , apj 553 , 538 wright , e.l . , reese , e.d .
2000 , apj 545 , 43 yun , m.s . , reddy , n.a . , condon , j.j .
2001 , apj 554 , 803 | deep extragalactic surveys with isocam revealed the presence of a large density of faint mid - infrared ( mir ) sources .
we have computed the 15@xmath0 m integrated galaxy light produced by these galaxies above a sensitivity limit of 50 @xmath0jy .
it sets a lower limit to the 15@xmath0 m extragalactic background light of ( 2.4 @xmath1 0.5 ) nw m@xmath2 hz@xmath3 .
the redshift distribution of the isocam galaxies is inferred from the spectroscopically complete sample of galaxies in the hubble deep field north ( hdfn ) .
it peaks around @xmath4 0.8 in agreement with studies in other fields .
the rest - frame 15@xmath0 m and bolometric infrared ( 8 - 1000@xmath0 m ) luminosities of isocam galaxies are computed using the correlations that we establish between the 6.75 , 12 , 15@xmath0 m and infrared ( ir ) luminosities of local galaxies .
the resulting ir luminosities were double - checked using radio ( 1.4 ghz ) flux densities from the ultra - deep vla and wsrt surveys of the hdfn on a sample of 24 galaxies as well as on a sample of 109 local galaxies in common between isocam and the nrao vla sky survey ( nvss ) .
this comparison shows for the first time that mir and radio luminosities correlate up to @xmath4 1 .
this result validates the bolometric ir luminosities derived from mir luminosities unless both the radio - far infrared ( fir ) and the mir - fir correlations become invalid around @xmath4 1 .
the fraction of ir light produced by active nuclei was computed from the cross - correlation with the deepest x - ray surveys from the chandra and xmm - newton observatories in the hdfn and lockman hole respectively .
we find that at most 20@xmath5 of the 15@xmath0 m integrated galaxy light is due to active galactic nuclei ( agns ) unless a large population of agns was missed by chandra and xmm - newton . about 75@xmath5 of the isocam galaxies
are found to belong to the class of luminous infrared galaxies ( @xmath6 @xmath7 @xmath8 @xmath9 ) .
they exhibit star formation rates of the order of @xmath10 100 @xmath11 @xmath12 .
the comoving density of infrared light due to these luminous ir galaxies was more than 40 times larger at @xmath4 1 than today .
the contribution of isocam galaxies to the peak of the cosmic infrared background ( cirb ) at 140@xmath0 m was computed from the mir - fir correlations for star forming galaxies and from the spectral energy distribution of the seyfert 2 , ngc 1068 , for agns .
we find that the galaxies unveiled by isocam surveys are responsible for the bulk of the cirb , i.e ( 16 @xmath1 5 ) nw m@xmath2 hz@xmath3 as compared to the ( 25 @xmath1 7 ) nw m@xmath2 hz@xmath3 measured with the cobe satellite , with less than 10@xmath5 due to agns . since the cirb contains most of the light radiated over the history of star formation in the universe , this means that a large fraction of present - day stars must have formed during a dusty starburst event similar to those revealed by isocam . |
authors thank gleb kozlov for fruitful discussions .
we acknowledge the financial support by the deutsche forschungsgemeinschaft and the russian foundation of basic research in the frame of the icrc trr 160 ( rfbr project no
. 15 - 52 - 12016 nnio_a ) .
the project ` spangl4q ' acknowledges financial support from the future and emerging technologies ( fet ) programme within the seventh framework programme for research of the european commission , under fet - open grant no . fp7 - 284743 .
thanks the russian foundation of basic research for partial financial support ( contract no .
14 - 02 - 31735 mol_a ) .
s.v.p . and i.a.yu .
acknowledge partial financial support from the russian ministry of science and education ( contract no .
11.g34.31.0067 ) and spbu ( grant no .
11.38.213.2014 ) .
m.b . acknowledges support from the russian ministry of science and education ( contract number 14.z50.31.0021 ) .
we consider optical excitation of an inhomogeneous ensemble of two - level systems by two short laser pulses with central frequency @xmath11 close to the average frequency of the ensemble @xmath71 .
for each individual two - level system the incident electromagnetic field induces optical transitions creating a coherent superposition of states .
let @xmath72 denote the ground state and @xmath73 denote the dipole active , excited state .
the temporal evolution of the density matrix is described by the lindblad equation : @xmath74+\gamma.\ ] ] here @xmath46 , where @xmath47 is the hamiltonian of the unperturbed system .
@xmath42 describes relaxation processes phenomenologically .
@xmath75 is the perturbation operator describing the interaction with the electromagnetic wave of a pulse : @xmath76 , @xmath6 is the dipole moment of the optical transition to state @xmath73 , @xmath77 is the envelope of the electric field of the pulse . here
the rotating wave approximation is used . for simplicity
, we consider pulses with rectangular shape of amplitude @xmath5 and assume that the pulse duration , @xmath49 , is significantly shorter than all relaxation processes . at the end of a short pulse ,
the density matrix elements are @xcite : _ 00(_p)&=&_00(0)- _
p + & + & ( 1- _ p ) , + _ 11(_p)&=&_11(0)+ _
p + & - & ( 1- _ p ) , + _ 01(_p)&= & ( i_p ) .
[ eq : eq5a ] here , @xmath12 is the detuning between the optical frequency @xmath11 and the resonance frequency of the individual two - level system @xmath13 .
@xmath78 is proportional to the smooth envelope of the pulse : @xmath79 in the rectangular pulse approximation , @xmath78 is constant during the pulse action , thus , @xmath80 , where @xmath16 is the rabi frequency . let us consider now the interaction of the two - level system with two pulses separated by the time , @xmath2 .
we assume that before excitation with the first pulse all two - level systems are in the ground state @xmath72 and , therefore , only @xmath81 .
the density matrix elements after this first pulse are @xcite : _
00(_p)&=&_00(0 ) -(1-_1 _ p ) , + _ 11(_p)&= & ( 1-_1 _ p ) , + _ 01(_p)&= & ( i_p ) f_1^*_00(0)2_1 ^ 2 .
[ eq : eq5a ] here the index @xmath82 indicates the first pulse . after the pulse ,
the elements of the density matrix relax with different characteristic times .
we assume that @xmath83 describing the optical coherence relaxes with the dephasing time @xmath84 : @xmath85\exp\big[-(t-\tau_p)/t_{2}\big]$ ] .
after the action of the second pulse we need only the following term in @xmath83 for calculation of the photon echo : _ 01(_12 + 2_p ) & & ( i_p ) . finally , after two pulses : _
01(t ) & & + & & ^2(_2_p/2 ) .
the measured photon echo signal from the ensemble of two - level systems is proportional to the sum over all @xmath13 : @xmath86 , where @xmath87 is the distribution function of the ensemble .
if the distribution function is gaussian with a central frequency @xmath88 and a dispersion @xmath10 , this gives the following form of the echo signal : & & .
+ [ p_pe_abs ]
in order to calculate the two - dimensional fourier spectrum ( 2dfs ) of a photon echo ( pe ) , we integrate the pe field over the two temporal varaibles that determine the delay between the laser pulses , @xmath89 , and the laboratory time , @xmath90 . to that end we rewrite eq.([p_pe_abs ] ) : figure [ som_fig ] demonstrates the results of corresponding 2dfs calculation after eq.([2dfts ] ) for tls ensemble excitation with different areas , @xmath64 of the first pulse .
the 2dfs spectra are normalized to unity . the first case shown in fig .
[ som_fig](a ) accords with a traditional hahn echo , for which no temporal shift of the pe is expected . here
, the 2dfs has only a real part .
note that the weak fringes in the plot are an artifact due to taking the fourier transform over finite intervals . in case of weak excitation ( linear regime ) of a strongly inhomogeneous system similar to the one
studied experimentally , the temporal shift of the pe gives rise to a significant imaginary part of the 2dfs with a dispersive shape and a sign reversal along the diagonal , as shown in fig .
[ som_fig](b ) .
when increasing the power of the first excitation pulse further to a @xmath23-pulse , see fig .
[ som_fig](c ) , also the real part of the 2dfs undergoes strong changes : its sign becomes reversed compared to the weak power regime and it shows a double peak structure , while the imaginary part of the spectrum is now the dominant contribution . ) for three different cases : ( a ) hahn - echo ; ( b ) linear excitation regime with weak intensity of the first pulse , @xmath91 ; ( c ) excitation in the non - linear regime with the first pulse being a @xmath23-pulse .
the 2dfs spectra are normalized to unity
. left and right panels denote the real and imaginary parts of the @xmath92 spectra , respectively , plotted in the same scaling .
other parameters in the calculations are : @xmath93 mev , @xmath94 ps . ] | an ensemble of quantum dot excitons may be used for coherent information manipulation . due to the ensemble inhomogeneity
any optical information retrieval occurs in form of a photon echo .
we show that the inhomogeneity can lead to a significant deviation from the conventional echo timing sequence .
variation of the area of the initial rotation pulse , which generates excitons in a dot sub - ensemble only , reveals this complex picture of photon echo formation .
we observe a retarded echo for @xmath0 pulses , while for @xmath1 the echo is advanced in time as evidenced through monitoring the rabi oscillations in the time - resolved photon echo amplitude from ( in , ga)as / gaas self - assembled quantum dot structures and confirmed by detailed calculations .
manipulation of electronic states by laser pulses requires precise timing of the pulse sequence .
a particular example is ultrafast coherent control of exciton complexes confined in semiconductor nanostructures @xcite which may be used for manipulating quantum states @xcite or even storing quantum information @xcite . for an inhomogeneous ensemble of two - level systems ( tls ) such as quantum dots ( qds )
the information retrieval induces echoes for the recovered signal @xcite . to be of practical use
, the echo formation time needs to be determined with high accuracy . the scenario is well established when a weakly inhomogeneous ensemble of tls is resonantly addressed by ultrashort laser pulses whose spectrum covers the distribution of optical transitions in the ensemble . in this conventional picture of a photon echo ( pe )
, a first excitation pulse creates a macroscopic polarization which subsequently dephases due to inhomogeneity .
a second subsequent optical rephasing pulse retrieves the macroscopic polarization @xcite with subsequent emission of a pe pulse delayed by a time exactly equal to the interval between the rephasing and excitation pulse , @xmath2 .
typically the temporal profile of the pe pulse has a gaussian shape reflecting the inhomogeneous frequency distribution of the excited ensemble . in semiconductors
this picture may be complicated by several features : ( 1 ) due to the many - body interactions a description by a distribution of independent tls is as a rule not appropriate , but complex coherent transients are recorded @xcite .
however , localization of excitons strongly reduces exciton - exciton interactions @xcite .
therefore , qds with particularly strong three - dimensional localization can be often reasonably well considered as tls with long coherence times .
indeed , rabi oscillations of exciton complexes were demonstrated in qds using various techniques @xcite .
( 2 ) in addition , fluctuations in qd size , shape and composition lead to a considerable inhomogeneous broadening of the optical transitions in qds . simultaneously excitation with ultrashort pulses having a broader spectrum than the inhomogeneity in many cases is prohibited because of the demand of addressing a specific optical transition .
the relatively narrow energy spacing of different electronic excitations limits the spectral width that the laser pulses may have , to avoid excitation of multiple transitions .
therefore a careful consideration of the optical frequency detuning with respect to the tls optical resonance is mandatory when considering the pe transient formation , even in the limit of small excitation levels .
generally , it is expected that for strongly inhomogeneous tls ensembles the pe temporal profile may undergo significant distortion compared to the standard @xmath3 sequence , showing either retarded or advanced echo pulses @xcite .
this is due to the non - negligible inhomogeneity - induced dephasing of oscillators during the optical excitation pulse , so that the finite pulse duration should be taken into account .
surprisingly , these variations of the pe transients in tls were only reported in ruby @xcite , where the characteristic excitation and depolarization times are in the range of tens of nanoseconds , i.e. orders of magnitude longer than in semiconductor nanostructures .
so far , the pe transients in qds were considered in the conventional @xmath3 picture only .
the most interesting from the standpoint of applications is the non - linear regime , when the inhomogeneous system under study may be considered as an ensemble of two - level systems ( tls ) coherently driven by the optical field , so that even rabi oscillations can be observed @xcite .
this regime is characterized by the following relation for the saturation factor : t_2 1 , [ satfactor ] where @xmath4 is the rabi frequency expressed via laser pulse amplitude @xmath5 and transition dipole moment @xmath6 , @xmath7 is the decoherence time of an individual oscillator , and @xmath8 is planck constant . during the process of optical excitation , coherent evolution of an individual oscillator from an ensemble with a gaussian distribution of frequencies @xmath9 with standard deviation @xmath10 directly depends on its detuning from the light frequency @xmath11 : @xmath12 , where @xmath13 is the oscillator frequency .
hence , the generalized rabi frequency is defined by @xmath14 . while generalized rabi frequency increases with the detuning , the maximum possible population of the tls has the following dependence : , maximum possible population corresponds to the envelope function of population inversion @xmath15 . ] w()=-1 .
[ population ] . here , two limiting cases are feasible , as outlined in fig .
[ experimental_data ] .
when @xmath16 is significantly larger than the inhomogeneity parameter @xmath10 , detuning plays insufficient role , and emission of the pe pulse occurs precisely at @xmath17 by the analogy with the hahn echo @xcite , as shown in fig . [ experimental_data](a ) . on the bloch sphere , in the reference frame rotating at the light frequency @xmath11 ,
the horizontal drift due to detuning during the excitation is negligible . in the case ,
when the inhomogeneous ensemble is relatively broad , which is often the case for semiconductor qd ensembles , as compared with atoms and rare - earth crystals , the magnitude of the rabi frequency limits the fraction of the ensemble interacting with the pulse , see fig .
[ experimental_data](b ) . .
( b ) theoretical modeling of the data from ( a ) using eq .
( [ pdet ] ) with the following parameters : pump duration @xmath18 ps , inhomogeneity @xmath19 mev , and rephasing pulse area @xmath20 . in the colormaps ,
dark - blue indicates the zero level , dark - red corresponds to maximum signal .
( c ) pl spectrum of the sample with the energy position of excitation , @xmath21 , indicated by the arrow .
( d ) time - integrated ( ti ) modulus of the pe amplitude measured as function of the first pulse amplitude .
( e ) pe transients for @xmath22 , @xmath23 and @xmath1 , respectively . ] in this letter , we demonstrate that the pe transients are strongly influenced by the inhomogeneity of a qd ensemble .
we study the dynamics of the pe from an ensemble of ( in , ga)as / gaas semiconductor qds excited by picosecond laser pulses . by varying the amplitude of the first excitation pulse we find a complex evolution of rabi oscillations in the amplitude of the pe , deviating strongly from the @xmath24 expectation .
namely , we observe temporal shifts of the pe pulse relative to the expected pe arrival time .
we analyze the experimental results measured on excitons using a transparent analytical model .
the pe experiment was carried out on a sample with a single layer of ( in , ga)as qds which is inserted into a gaas @xmath25-microcavity .
the qd density is about @xmath26 @xmath27 and the barrier contains a si @xmath10-layer with donor density @xmath28 @xmath27 , separated from the qd - layer by 10 nm .
the bragg mirrors consist of alternating gaas and alas layers with thicknesses of 68 nm and 82 nm , respectively .
they have 5 and 18 such pairs in the top and bottom mirror , respectively . with a microcavity @xmath29-factor of @xmath30 ,
the photon lifetime is much shorter than the laser pulse duration of several picoseconds .
therefore , the main influence of the microcavity is to enhance the light - matter interaction for those qds with optical frequency in resonance with the cavity mode .
further , the broad exciton photoluminescence ( pl ) spectrum from the qds with a width of 30 mev is convoluted with the relatively narrow photon cavity mode to result in a relatively narrow emission band with full - width of 6.5 mev .
the energy of the cavity mode depends on the sample spot due to the wedged cavity design .
a pl spectrum measured at the studied sample spot discussed subsequently is shown in fig .
[ experimental_data](c ) . to study the pe dynamics ,
a degenerate transient four - wave - mixing ( fwm ) experiment was performed in reflection geometry .
the sample was placed into a helium bath cryostat and cooled down to a temperature of @xmath31 k. we used a tunable mode - locked ti : sa laser as source of the optical pulses with duration of about 2.5 ps and repetition rate of about 76 mhz .
we implemented an interferometric heterodyne detection using a reference beam to detect the absolute value of the fwm amplitude with temporal resolution .
the excitation pulses were linearly cross - polarized resulting a linearly polarized pe signal.4-density matrix formalism in order to account for the linear polarizations as shown in ref .
@xcite .
it was checked , however , that the effect of the pe shifts demonstrated here is valid for any polarization configuration . ] a detailed description of the experimental technique can be found elsewhere @xcite . in the experiment ,
the temporal profiles of the pe were measured as a function of the laser pulse amplitude .
figure [ experimental_data ] summarizes the experimental data obtained when the amplitude of the first excitation pulse is varied .
this pulse amplitude is proportional to @xmath32 , where @xmath33 is the time - integrated intensity of the laser pulses .
the delay between the centers of rephasing and excitation pulses is set to 66.7 ps .
clearly , footprints of rabi oscillations are observed in fig .
[ experimental_data](a ) .
most strikingly , sizable temporal shifts of the pe maxima for @xmath0 and @xmath1-excitation pulses ( around 0.8 mw@xmath34 and 1.8 mw@xmath34 , respectively ) as well as a splitting of the pe profile around @xmath23-pulse amplitude ( around 1.3 mw@xmath34 ) become evident in fig . [
experimental_data](e ) , where the corresponding pe transients are plotted .
the time - integrated pe signal also manifests rabi oscillations , as seen from fig .
[ experimental_data](d ) , but the oscillations are much less pronounced than in the time - resolved data .
note , that variation of the second pulse amplitude does not lead to any additional modification of the pe temporal profile . in order to describe the pe from a qd ensemble we calculate the temporal evolution of a set of 2@xmath352 density matrices interacting with two sequent laser pulses and analyze the resulting macroscopic polarization of the medium , @xmath36 , given by the mean value of the dipole moment operator @xmath37 : p = tr ( ) . here , @xmath38 is the density matrix of the tls and @xmath39 is the trace of the matrix .
the temporal evolution of the density matrix is described by the lindblad equation : @xmath40+\gamma$ ] , where @xmath41 is the hamiltonian of the system and @xmath42 describes relaxation processes phenomenologically .
we consider the case , when the individual oscillator coherence time @xmath43 , which is justified by @xmath7 being ultimately limited by the exciton lifetime of about 0.5 ns .
thus , the only consequence of relaxation in the system is the pe exponential decay @xmath44 .
therefore , we will neglect relaxation considering @xmath45 , for simplicity . in our model ,
the hamiltonian consists of two parts : @xmath46 , where @xmath47 describes the unperturbed system and @xmath48 gives the interaction with light .
we assume rectangular shaped pulses , each with with duration @xmath49 . during the optical excitation
, we calculate the ensemble dephasing due to the detuning of every individual oscillator with frequency @xmath13 from the light frequency @xmath11 : @xmath12 . in the rotating - wave approximation , this defines the optical nutation of the tls on the bloch sphere with a generalized rabi frequency @xmath14 , where @xmath50 is proportional to the optical field amplitude @xmath5 .
accordingly , the analysis of the coherent optical response of the system includes the following steps : first , we calculate the interaction of the unperturbed system with the first excitation pulse ( @xmath51 ) acting during time @xmath49 , then the system evolves freely in absence of excitation ( @xmath52 ) during the time interval @xmath2 , followed by the excitation with the second rephasing pulse ( @xmath53 ) during time @xmath49 .
finally , we analyze the macroscopic polarization of the qds that is available for pe pulse formation by integrating the individual responses over the inhomogeneous ensemble of tls , defined by a gaussian distribution with standard deviation @xmath10 . within these approximations , it is possible to perform the analysis analytically and calculate the following expression for the modulus of the detected excitation - dependent part of the pe amplitude ( see appendix a containing details of the theoretical approach . ) : @xmath54 d\delta\bigg| . \label{pdet}\end{aligned}\ ] ] here @xmath55 , and @xmath56 are the generalized rabi frequencies of the first and the second pulse , respectively . in this notation ,
delay between the exciting pulse centers used in experiment is @xmath57 and expected pe arrival time according to the standard treatment is @xmath58 .
this equation is similar to the one obtained for spin echoes measured in highly inhomogeneous fields @xcite .
it can be easily seen , that variation of the second , rephasing pulse amplitude and , hence , @xmath59 does not affect the symmetry of the pe profile . therefore , we will consider only the effect coming from variation of the first , excitation pulse amplitude
. using eq .
( [ pdet ] ) for two different ensemble with different inhomogeneity @xmath10 : ( a ) @xmath60 ; and ( b ) @xmath61 .
the time scale is normalized by the pulse duration .
dark - blue corresponds to zero signal level , dark - red gives the maximum value . ]
figure [ simulations ] shows results of theoretical calculations performed using eq .
( [ pdet ] ) for two ensembles with different widths of the inhomogeneity : @xmath60 and 1.2 . for an ensemble broadening much narrower than the pulse spectrum , @xmath60 ,
clean rabi oscillations are seen in the pe temporal profile as function of the excitation pulse area @xmath62 , shown in fig .
[ simulations](a ) .
in particular the photon echo arrives as expected at twice the separation between rephasing and excitation pulse .
this case is equivalent to an ideal hahn echo , where the echo signal is generated at @xmath63 .
when , however , the inhomogeneous ensemble width becomes comparable with the pulse spectral width ( @xmath61 ) , strong modifications of the temporal pe profile appear , as can be seen in fig .
[ simulations](b ) .
the key feature of these modifications is the effective temporal shift of the pe amplitude that is particularly prominent for the first two maxima of the rabi oscillations when increasing the excitation power : for a @xmath0-pulse the pe maximum is retarded relative to the expected pe arrival time , while for a @xmath1-pulse the pe is advanced .
these shifts do not depend on the pulses delay time @xmath2 in the limit of negligible damping and decrease with further increasing @xmath64 .
-pulse : case of hahn echo .
( b ) @xmath0-pulse : retarded echo .
( c ) @xmath1-pulse : advanced echo .
the colored lines on the bloch spheres ( left panels ) denote different trajectories of the bloch vectors of individual oscillators with different detuning from the laser frequency in directions indicated by the arrows .
the right panels show the temporal evolution of the modulus of the macroscopic polarization , @xmath65 , after excitation pulse action .
the dashed lines denote extrapolation of @xmath65 to @xmath66 . ] the temporal shifts of the pe can be qualitatively explained using the bloch sphere representation , as outlined in fig .
[ drift_explanation ] . in fig .
[ drift_explanation](a ) , the ideal case of a hahn echo is given as a reference , where excitation with a negligibly short excitation @xmath0-pulse results in an unshifted echo signal .
when the excitation pulse has a finite duration , as shown in fig .
[ drift_explanation](b ) , oscillators experience dephasing during pulse action , so that the excitation distributes the oscillators along the equator of the bloch sphere followed by their precession in the equatorial plane .
this precession extrapolated to negative times results in focusing all the bloch vectors on the same meridian , which corresponds to a situation where the maximum macroscopic polarization is shifted to @xmath66 . in case of excitation by a @xmath1-pulse as shown in fig .
[ drift_explanation](c ) , after pulse action all oscillators proceed along the equator towards each other , which leads to a shift of the macroscopic polarization maximum to @xmath67 . since the rephasing pulse , which is not necessarily a @xmath23-pulse , acts essentially as a refocusing pulse , which in effect reverses the temporally coherent evolution of the ensemble oscillators , the dynamics of the macroscopic dipole moment around the expected echo arrival time , @xmath68 , is similar to a mirrored copy of its dynamics around the time @xmath69 , as shown by the right panels of fig .
[ drift_explanation ] .
thus , the pe pulse is retarded for the a @xmath0-excitation pulse and advanced for a @xmath1-pulse , when the ensemble dephasing during pulse action is sufficiently strong .
this can be achieved either by high inhomogeneity in the excited ensemble and/or long pulse duration . the analysis of the experimental data in fig .
[ experimental_data](a ) was performed using the theoretical modeling of the pe effect at varying excitation pulse amplitudes after eq .
( [ pdet ] ) .
the adjustable parameters are @xmath10 and @xmath49 .
the theoretical fit depicted in fig .
[ experimental_data](b ) gives a pulse duration @xmath70 ps and an inhomogeneous ensemble full - width at half - maximum @xmath19 mev .
this broadening is much smaller than the pl linewidth of the qd structure ( 6.5 mev ) indicating that a much narrower sub - ensemble contributes to the pe compared to the ensemble contributing to the pl emission .
the origin of this discrepancy will be subject of further studies
. a possible reason might be the interaction with the microcavity .
the evident rabi oscillations pattern in the pe sequence tells that qd structures under resonant excitation can be well described by tls . in conclusion ,
the importance of the oscillator detuning from the optical frequency in forming the macroscopic coherent response in semiconductor qds was addressed in literature already before @xcite .
however , no consequence of the timing of the associated pe had been reported .
the essential consequence of our study is that even significantly inhomogeneous semiconductor systems still demonstrate a comprehensible coherent behavior , but great care needs to be exercised to consider the consequences of the inhomogeneity in detail .
when doing so , such systems are very well suited candidates for the coherent manipulations in coherent optoelectronics . in particular , the observed temporal shifts of the pe amplitude need to be taken into account when a precise timing of the optical signals matters .
our findings have strong impact on interpretation of two - dimensional fourier transform images , which are currently widely used because of their intuitive visualization of underlying physics @xcite .
the complex - valued envelope of the pe - transients from strongly inhomogeneous systems results in a changeover of the corresponding 2d image from a resonance profile in case of a hahn echo to a dispersive profile ( see appendix b ) . |
the cyclotron motion of charge carriers with a well defined effective mass @xmath1 and the related resonant absorption of light at the cyclotron frequency @xmath2 are basic physical phenomena representative of the magneto - optical response of conventional condensed - matter systems . well - established and often applied , cyclotron resonance ( cr ) became through the years a common technique routinely used in solid - state physics , inherently connected with the effective mass of elementary electronic excitations.@xcite recently fabricated graphene@xcite immediately became a challenge for cr measurements and implied necessity to comprehend the phenomenon of cyclotron resonance in a system with massless charge carriers.@xcite special character of cr in graphene can be exemplified on the basis of a simple quasi - classical approach.@xcite if we solve the classical equation of motion for a charged particle with an energy @xmath3 which depends linearly on the momentum @xmath4 , i.e. @xmath5 , we find out that the precession - like ( cyclotron ) motion remains characteristic also for these dirac - like particles , nevertheless , with a cyclotron frequency modified to @xmath6 .
hence , the classical cyclotron motion and consequently also the cyclotron resonance of massless dirac fermions scale linearly with the applied magnetic field @xmath7 , similar to conventional massive particles , but they depend on the particle energy , which appears in a form of an energy - dependent cyclotron mass defined by the einstein relation @xmath8 .
surprisingly , this classical , i.e. linear with the magnetic field , cyclotron resonance has not been clearly observed in graphene so far .
instead , the optical response of graphene has been experimentally explored@xcite only in the quantum regime , when all detected transitions correspond to excitations between well - separated landau levels ( lls ) .
these measurements provided us with a unique insight into the ll spectrum of graphene and directly visualized its intriguing @xmath9-scaling , but no information about the quasi - classical limit could be extracted .
consequently , it remains a challenge to demonstrate experimentally the quasi - classical cyclotron resonance in graphene , or interestingly , to study the crossover between this classical ( linear in @xmath7 ) and fully quantum - mechanical ( linear in @xmath9 ) magneto - optical response of graphene the crossover , which is not present in a system of massive particles , where the response always remains linear in the applied magnetic field .
here we report on magneto - optical measurements on highly - doped epitaxial graphene grown on the silicon - terminated surface of sic .
we identify the cyclotron - resonance - like response related to the massless dirac fermions in the vicinity of the fermi level , which evolves nearly linearly with the applied magnetic field , as expected from ( quasi-)classical considerations .
these results are compared to the currently well - understood response of multilayer epitaxial graphene prepared on the carbon - terminated surface of sic , where majority of ( rotationally stacked ) graphene sheets is nearly undoped and where the fully quantum - mechanical treatment is appropriate for its description .
in our study , we present data taken on two particular samples of graphene prepared by thermal decomposition on the surface of semi - insulating silicon carbide.@xcite both specimen have been grown at 1600@xmath10c in ar atmosphere using a commercially available horizontal chemical vapor deposition hot - wall reactor ( epigress v508 ) , inductively heated by a rf generator .
the sample denoted as a in our study , grown on the c - terminated surface of silicon carbide ( 4h - sic[0001 ] ) , represents a standard multilayer epitaxial graphene ( meg ) specimen .
the majority of sheets in this structure is practically undoped ( residual @xmath4-like doping expected at the level of @xmath11 @xmath12 , see refs . and ) with the electronic band structure similar if not identical to an isolated graphene monolayer , provided by the characteristic rotational stacking of layers.@xcite this fact is reflected in the micro - raman spectra taken on this sample , which revealed dominantly single - component ( lorentzian - like ) 2d band , @xcite nevertheless , some bernal - stacked residuals have been observed on selected locations as well .
the sample b has been prepared on the si - terminated surface of sic ( 4h - sic[0001 ] ) and also characterized in the micro - raman experiment . similarly to the previous sample , the dominantly observed single - component 2d band serves as an indication for linearity of electronic bands .
it points towards the presence of a single graphene sheet ( in addition to the non - graphene - like buffer layer@xcite ) , as in contrast to meg on the c - face , the few - layer graphene stacks on the si - face exhibit bernal and not rotational ordering.@xcite in addition , such monolayers are typically rather highly doped by the charge transfer from the substrate.@xcite to measure the ( non - polarized ) infrared transmittance of our samples , we used the radiation of a globar , which was analyzed by a fourier transform spectrometer and delivered to the sample via light - pipe optics .
the light was then detected by a composite bolometer kept at @xmath13 k placed directly below the sample .
measurements were done in the faraday configuration in a superconducting solenoid .
both investigated samples were fully opaque in the energy range 85 - 220 mev due to phonon - related absorption in sic .
let us start our discussion with the data obtained on the sample a , i.e. on the meg structure grown on the c - terminated surface of sic , see fig .
[ spkt - c - face ] , whose magneto - optical response is relatively well understood.@xcite transmission spectra are in this kind of specimens dominated by relatively sharp absorption lines with a typical @xmath9-like scaling ( see fig . [ fanchart - c - face ] ) and taking the ll spectrum of graphene : @xmath14 , we easily identify them as inter - ll transitions l@xmath15@xmath16l@xmath17 and l@xmath18@xmath16l@xmath19 with @xmath20 . the corresponding fermi velocity can be evaluated as @xmath21 m.s@xmath22 . as a matter of fact , this response is equivalent to that of an exfoliated graphene monolayer@xcite and in turn it serves as a direct indication that layers in meg indeed behave as independent , electrically isolated , graphene sheets .
apparently , here we deal with graphene in its fully quantum regime , i.e. we study optical excitations between well - separated landau levels , whose @xmath9-scaling and multi - mode character are directly related to the linearity of the electronic bands.@xcite to characterize our sample further , we can set the upper limit for the carrier density that can not significantly exceed @xmath23 @xmath12 , as the transition l@xmath24l@xmath25 is observed down to the magnetic field of 100 mt ( the filling factor @xmath26 ) .
we assume that the graphene sheets are slightly @xmath27doped as follows from the recent arpes study done on equivalent specimens .
@xcite we should also note that the dominant graphene - like lines are often accompanied by additional spectral features , which are usually significantly weaker , nevertheless still well - defined in the spectra , located especially at low energies , see e.g. the transition denoted by stars in fig .
[ spkt - c - face ] .
we will come back to these features later on after we discuss the response of the sample b. the sample b exhibits a very much different behavior in our magneto - transmission experiment and shows no @xmath9-scaled spectral features , which are , as we have learned above , characteristic of graphene in its quantum regime .
instead , it clearly shows magnetic - field - induced transmission at low energies accompanied by a relatively wide absorption minimum , which moves towards higher energies with increasing @xmath7 .
we found this behavior to be characteristic for all graphene specimens prepared on si - terminated surface of sic ( not shown in this paper ) , nevertheless , the total intensity of the observed spectral features as well as the position of the absorption minimum slightly varied from sample to sample . to explain this behavior , we should recall that epitaxial graphene on the si - terminated surface is usually highly doped by the charge transfer from the substrate , reaching carrier densities up to @xmath28 @xmath12 ( @xmath29 ev ) , see e.g. ref . , with quality ( expressed in terms of scattering time or mobility ) comparable to meg on the c - face of sic.@xcite therefore , at magnetic fields of a few tesla , lls should not be resolved around the fermi level and the system should remain in the quasi - classical regime .
interestingly , owing to the non - equidistant spacing of lls in graphene , the tuning from a fully quantum regime to the classical one is possible just by changing the fermi level .
if we adopt this , the qualitative explanation of the observed behavior becomes straightforward .
the relative magneto - transmission spectrum @xmath30 reflects two specific effects : the low - energy free carrier ( drude - like ) absorption , which is gradually suppressed with the increasing field , and the cyclotron resonance itself .
the former one gives rise to the field - induced transmission at low energies that decreases monotonically with the photon energy , and the latter one is manifested by an absorption minimum close to the energy of the cyclotron resonance @xmath31 .
let us now proceed with the quantitative analysis of this data , which will allow us to determine the characteristic field - dependence of @xmath31 . in the framework of the classical theory for cyclotron resonance , the optical conductivity
is given by @xmath32 which describes both the cr active ( @xmath33 ) and inactive ( @xmath34 ) polarization modes of light , as well as the drude - like low - energy absorption ( at @xmath35 ) .
@xmath36 here denotes zero - field dc conductivity and @xmath37 is a phenomenologically introduced damping .
we should also point out that in graphene , contrary to conventional systems of massive particles , the cyclotron frequency @xmath31 is not independent of @xmath36 and @xmath37 as the `` effective '' effective mass of dirac fermions ( at the fermi level ) can be determined from these two parameters .
the transmission of circularly - polarized light through the studied specimen ( graphene sheet on an isolating substrate ) can be expressed by formula:@xcite @xmath38 where @xmath39 which is derived for a dielectric slab with the index of refraction @xmath40 ( @xmath41 ) with an infinitely thin conducting layer on one of its surfaces ( described by @xmath42 ) . @xmath43 and @xmath44 stand for the real speed of light and the vacuum permittivity , respectively .
the dependence of @xmath45 on the thickness of the slab ( i.e. interference effects ) has been already averaged out . within the framework of our simple model
, we can easily calculate the relative change of the sample transmission with the magnetic field @xmath46/2t(0)$ ] , and compare it with our experimental data . as the conductivity as well as the damping parameter can in principle vary with @xmath7 , we have first found their appropriate values at zero field , @xmath47 @xmath48 and @xmath49 mev , which we afterwards kept constant and varied only @xmath50 and @xmath51 to reproduce our experimental traces . as shown in figs .
[ spkt - si - face ] and [ fanchart - si - face ] , our simple theoretical model is able to reproduce experimental transmission spectra very well and the cr frequency indeed follows a clear linear in @xmath7 dependence .
the conductivity @xmath52 and the cr half - width @xmath51 exhibited in our fitting procedure a weak decrease and increase , respectively , but practically remain within the interval of uncertainty estimated for @xmath36 and @xmath53 and they can be therefore taken as nearly constant in the investigated range of fields . the width ( fwhm ) of the cyclotron resonance absorption peak @xmath54 has been found comparable with the cyclotron energy @xmath31 what justifies the quasi - classical regime in the investigated sample .
we should also clarify that a relatively broad cr resonance implies that the minimum in the transmission spectra does not exactly match to the position of the cr mode .
this is explained in the inset of fig .
[ spkt - si - face ] , where the transmission curves @xmath55 at @xmath56 t and @xmath57 are depicted as calculated from parameters obtained in our data analysis .
we also plot curves @xmath58 and @xmath59 , which show expected transmission in opposite circular polarizations of light , i.e. in the cr active and inactive modes , respectively .
let us note that all these transmission curves share a common baseline at @xmath60 , which corresponds to the light transmission through a dielectric slab without any conducting sheet ( or away from resonances )
. this value can be easily obtained from eq . for @xmath61 .
let us now return to the main finding of our analysis , i.e. to the clear linear in @xmath7 dependence of the cyclotron resonance as shown in fig .
[ fanchart - si - face ] .
if we evaluate the effective mass corresponding to this cr mode , we obtain @xmath62 , which according to the einstein relation @xmath63 implies the fermi level of @xmath64 mev in a reasonably good agreement with conclusions of arpes measurements on similar samples.@xcite in this estimation , we used the fermi velocity of @xmath65 m.s@xmath22 determined from the measurement on the sample a , but very similar velocities are reported by arpes for graphene on the si - terminated surface of sic as well .
@xcite the observation of the cr mode that is linear in the applied field @xmath7 justifies our simple approach in which ( otherwise massless ) dirac fermions are considered as massive particles , the mass of which is determined by the fermi energy . on the other hand
, this confirmation implies another constrain on parameters used in our fitting .
namely , the conductivity @xmath36 can be directly estimated within the boltzmann transport theory from the scattering rate @xmath66 and from the fermi energy : @xmath67 .
this relation was respected in our fitting procedure and required an additional self - consistent loop to keep the correct relation between @xmath36 and @xmath53 on the basis of @xmath68 extracted from the fitting .
obviously , this approach works only in case of specimens with a full coverage by the graphene monolayer , otherwise the conductivity derived from @xmath53 and @xmath68 exceeds the value obtained directly from the fitting .
the ratio of these two values thus could serve as a simple parameter that describes fraction of coverage by graphene .
we should note that the evaluated mass of @xmath69 is surprisingly close the effective mass of @xmath70 point electrons in bulk graphite @xmath710.058@xmath72,@xcite but this similarity represents just a coincidence . to prove this , we can for instance argue that bulk graphite reaches the quantum limit at fields around 7 t,@xcite and therefore it can not exhibit cr line that corresponds to the effective mass of @xmath70 point electrons .
instead , due to the formal similarity of the band structure at the @xmath70 point with the graphene bilayer , it should provide the cr - like absorption at energy by a factor of @xmath73 higher , as indeed observed experimentally.@xcite to summarize , the highly - doped graphene sheet on the si - terminated surface allows us to study the cr resonance of masssless dirac fermions in its fully classical limit , when it becomes strictly linear with the applied magnetic field .
nevertheless , let us look at the same problem using the quantum - mechanical description , i.e. interpret our data as intra - band excitations between adjacent ( but in reality significantly broadened ) landau levels , as schematically shown in the inset of fig .
[ fanchart - si - face ] . to clarify the linear in @xmath7 response , found in a system with a @xmath9-scaled ll spectrum , we only need to consider the fermi level @xmath68 located high in the conduction ( or equivalently low in the valence ) band and independent of the magnetic field .
this can be exactly ensured by the pinning the fermi level by the surface states in the substrate or fulfilled approximatively , when lls significantly overlap . at a given magnetic field , the @xmath40-th
ll approaches the fermi level , @xmath74 , and the active cr - like transition ( between adjacent lls ) has the energy of @xmath75 , i.e. it indeed becomes linear in @xmath7 and moves with the slope related to the cyclotron mass @xmath0 .
the analysis of data obtained on highly - doped graphene in the sample b allows us to come back to the low - energy features in spectra taken on the sample a. also in this kind of samples , we usually find relatively weak spectral features whose energy moves nearly linearly in @xmath7 .
even though the sample a dominantly consists of nearly undoped graphene sheets , the increase of the magneto - transmission at lower energies serves as an indication of some free carriers in the sample , similarly to the sample b. in fact , also in meg on the c - face of sic , first few layers nearby the interface are usually doped by charge transfer from the substrate .
the transition denoted by stars ( see figs .
[ spkt - c - face ] and [ fanchart - c - face ] ) thus could be assigned to the cr response of these doped layers , and its characteristic slope @xmath764.5
mev.t@xmath22 ( effective mass of @xmath77 ) would imply the fermi level of @xmath78 mev . on the other hand
, the carrier density in these doped layers gradually decreases with the distance from the substrate , which should prevent observation of a well - defined cr mode ( whose energy depends on the density of massless particles ) , while the transition denoted by the star remains relatively narrow in our data .
more likely , this line originates in bernal - stacked residuals , especially in the graphene bilayer which has been identified as a minor but not negligible component of meg.@xcite the masssive dirac fermions in the bilayer are characterized by an effective mass of @xmath79 ( @xmath80 mev@xcite ) which does not match to the effective mass derived for the `` star '' transition , but the characteristic ll spectrum in the bilayer , @xmath81,@xcite implies in the quantum limit appearance of the cr mode ( l@xmath82l@xmath83 transitions ) whose slope is by a factor of @xmath73 higher .
the transition denoted by the star , with a slope corresponding to the effective mass of @xmath77 , thus could be assigned an undoped graphene bilayer in which the cr absorption should correspond to the mass of @xmath84 .
nevertheless , the definitive conclusion can not be drawn on the basic of our current data .
we have explored magneto - optical response of graphene in two distinctively different regimes .
whereas the weakly doped sheets in multilayer epitaxial graphene on the c - terminated surface allowed us to study the optical response of graphene in the quantum limit , massless dirac fermions in the quasi - classical regime can be explored in highly doped epitaxial graphene prepared on the si - terminated surface of sic .
the response of the former system clearly shows @xmath9-like scaling , typical of landau levels in systems with linear electronic bands , the latter case proves that cyclotron resonance of massless particles remains in the quasi - classical limit linear with the magnetic field . _ additional note : _ during our work on the manuscript we became aware of faraday rotation measurements on epitaxial graphene,@xcite in which a similar conclusion about cyclotron resonance of massless dirac fermions in the quasi - classical limit has been made .
this work has been supported by the projects 395/n - pics - fr/2009/0 and projects mtkd - ct-2005 - 029671 , furthermore by grants 670/n - esf - epi/2010/0 , 671/n - esf - epi/2010/0 and gra/10/e006 within the esf eurographene programme ( epigrat ) .
we acknowledge also funding received from euromagnetii under the eu contract no .
228043 and via projects msm0021620834 and gacr no . p204/10/1020 . | cyclotron resonance in highly doped graphene has been explored using infrared magneto - transmission .
contrary to previous work , which only focused on the magneto - optical properties of graphene in the quantum regime , here we study the quasi - classical response of this system .
we show that it has a character of classical cyclotron resonance , with an energy which is linear in the applied magnetic field and with an effective cyclotron mass defined by the position of the fermi level @xmath0 . |
the study of the behaviour of hadronic matter in the density - temperature , @xmath2 , diagram allows to have a deeper understanding of matter under extreme conditions . in this context , the high density , low temperature limit can be addressed for a fermion system using the landau fermi liquid theory ( flt ) @xcite . from a theoretical point of view the properties of this type of normal quantum systems
can be studied calculating the interaction matrix element of quasiparticle ( qp ) excitations close to the fermi surface .
the inclusion of an additional component in the problem , a magnetic field , @xmath3 , allows further testing the properties of magnetized fermi liquids .
the role of magnetic fields in bulk properties and equation of state has been partially analyzed in the past for nuclear matter @xcite @xcite and quark matter @xcite @xcite . due to the tiny value of the neutron magnetic moment @xmath4 ( @xmath5 @xmath6 mev @xmath7 ) @xcite and in order to provide a sizable magnetization ,
huge magnetic fields are needed .
the only scenarios where we have indication of such intense fields are , first , from estimates of the background magnetic fields created in heavy - ion collisions like those at rhic @xcite and , second , in a subgroup of pulsars called magnetars . for these astrophysical objects surface magnetic field strengths are of the order @xmath8 g @xcite .
recent numerical simulations @xcite of formation of proto - neutron stars show that the field configuration plays a significant role in the dynamics of the core if the initial magnetic field is large enough .
in particular , in the rapid cooling of the newly formed neutron - rich object neutrino transport is an important ingredient @xcite .
however , some of these simulations lack from accurate and consistent neutrino transport , missing the impact of magnetic fields in the microphysics input that affects the dynamics of the collapsing dense objects . in most of the existing calculations of nuclear matter ( either symmetric , pure neutron or beta - equilibrated ) the effect due to the presence of strong magnetic fields and the consistently induced spin polarization are discarded in a first approximation .
either relativistic @xcite or effective approaches @xcite have been used to obtain some insight into the equation of state ( eos ) or some structure properties @xcite in presence of magnetic fields .
these include a possible transition to a ferromagnetic state , although simulations using realistic potentials seem to prevent it @xcite .
in general , a non - vanishing magnetization in a low temperature nuclear plasma @xcite produces a resolution of some degenerated observables as obtained in the context of the flt @xcite .
in this work we are interested in the response of a spin - polarized pure neutron plasma to a weak neutrino probe . it can be seen @xcite that for the density range @xmath9 , where the quark deconfinement is not expected to take place , and for magnetic field strengths of maximum strength @xmath10 g , allowed in principle by the scalar virial theorem , the neutral system is mostly neutrons .
the maximum magnetic field strength we will consider is @xmath11 ( as measured in units of the electron critical field @xmath12 with @xmath13 g ) and the neutron fraction is @xmath14 @xcite .
so the neutral plasma is mostly neutrons but leptons and additional baryons are also present in a tiny fraction that should be considered for full application in an astrophysical scenario where @xmath15-equilibrium holds .
we are interested in exploring the effect of a strong magnetic field and the spin polarization of a pure neutron plasma through the structure functions , which provide information on density and spin density in - medium correlations .
the homogeneous system under study is under the presence of an internal magnetic field , @xmath16 populated by species with paricle density @xmath17 , where @xmath18 is the spin @xmath19-projection .
@xmath20 is the spin excess and @xmath21 is the total particle density . for given thermodynamical conditions @xmath22
is obtained by minimizing the helmholtz free energy per particle , @xmath23 , where @xmath24 is the energy per particle .
note that parallel ( antiparallel ) aligned magnetic moments ( spins ) are energetically favoured .
we have considered an effective approach to describe the nuclear interaction using zero - range skyrme forces @xcite with two of the most widely used parametrizations given by the lyon group sly4 and sly7 @xcite and finite range gogny with d1p @xcite and d1s @xcite forces .
all of them provide good values for binding of nuclei and also for neutron matter eos . in the context of the flt
the properties of non - magnetized systems at low temperature have been evaluated @xcite by calculating the qp matrix element around the fermi surface where the only dependence is on fermionic densities and the qp scattering angle , @xmath25 , involved . in the usual formalism , for the non - magnetized case the qp matrix element is written as a multipolar expansion in legendre polinomials , @xmath26 p_l ( cos\theta ) , \label{elemento}\ ] ] @xmath27 and @xmath28 are the so - called landau parameters of multipolarity @xmath29 . in the more general case where any two possible spin orientations @xmath30 are taken into account ,
the polarized qp matrix elements @xcite @xcite are a crucial ingredient to compute the response functions to a a weakly interacting neutrino probe that excites a collective mode @xmath31 under the presence of a magnetic field @xmath3 .
the lindhard function in the system , @xmath32 , satisfies the bethe - salpeter equation and can be written in the dipolar ( @xmath33 ) case in the random phase approximation ( rpa ) as a coupled system , @xmath34 @xmath35 with the auxiliar definitions , @xmath36 and @xmath37 .
notice that the qp propagators @xmath38 have been given in @xcite and the expressions for the coefficients @xmath39 can be written @xcite in the landau limit as @xmath40 and @xmath41 where @xmath42 .
the qp effective mass in a magnetized system depends on the polarized dipolar coefficients @xcite , @xmath43 \label{mef}\ ] ] where @xmath44 is the quasiparticle level density at each polarized fermi surface with momentum @xmath45 .
the generalized parameters @xmath46 are obtained by derivating the helmhotz free energy with respect to the polarized density component , @xmath47 @xcite , setting momenta on the polarized fermi surfaces and expanding the resulting expression as a series in legendre polinomials of multipolarity @xmath29 .
these generalized parameters fullfill the following relations recovering the usual ones in flt in the limit @xmath48 @xcite , @xmath49 @xmath50 with the generalized paramters and using the expressions in eq .
( [ chil1 ] ) the corresponding lindhard function for the isovector ( @xmath51 ) response of the plasma can be written as , @xmath52 and for the vector - axial ( @xmath53 ) response as , @xmath54 then the previous expression of the lindhard function in rpa approximation @xcite include in - medium correlations at zero temperature . from them , one can obtain the structure functions given by , @xmath55 the structure function allows to calculate the non - relativistic differential cross section of neutrinos scattering off matter via neutral currents from @xcite @xmath56 \label{cs}\ ] ] where @xmath57 is the incoming ( outgoing ) neutrino energy and @xmath58 @xmath59 is the neutrino incoming ( outgoing ) three - momentum .
the transferred energy is @xmath60 and the transferred three - momentum is @xmath61 .
the neutral current vector and axial vector charges are @xmath62 and @xmath63 where @xmath64 @xcite .
@xmath65 is the fermi coupling constant .
once the response has been evaluated it is straightforward to evaluate the neutrino mean free paths in the medium , @xmath66 .
in this section we include the effect of in - medium correlations in the neutron magnetized system as obtained in the hartree - fock approximation in the presence of a strong magnetic field . in fig .
[ fig1 ] the ratio of effective neutron mass as compared to the free value at saturation density , @xmath67 , is shown as a function of the logarithm of the magnetic field strength for the skyrme sly7 ( a ) and gogny d1p ( b ) parametrizations .
for each model upper ( lower ) curves correspond to spin up ( down ) polarized particles .
with skyrme description the intense field affects more both effective nucleon mass absolute and relative ( up - down polarized components ) values than with gogny .
the impact of density effects on the mean free path can be seen in fig .
we plot the ratio of change of neutrino mean free paths in the rpa dipolar approximation for a fixed value of magnetic field strength , @xmath68 g , with respect to the field - free case , @xmath69 , as a function of the density ( in units of nuclear saturation density , @xmath67 ) .
we consider skyrme sly7 ( solid line ) , sly4 ( long dashed line ) gogny d1p ( short dashed line ) and d1s ( dotted line ) parametrizations and set as a typical value of neutrino incoming energy @xmath70 mev .
while gogny forces show almost unchanged or very mild reduction of mean free paths at densities in the range @xmath1\rho_0 $ ] , the skyrme forces show a high density dramatic increase with respect to the field - free case .
note that all standard skyrme forces predict the onset of a ferromagnetic transition in the range @xmath71\rho_0 $ ] , and in our selection of interactions for the study cases it is near @xmath72 .
however this feature is not present in the gogny forces that prevent ferromagnetic transitions . at densities
@xmath1\rho_0 $ ] , effects due to the energetic contribution of the magnetic perturbation , introduced by the neutron magnetic moment , are small at the selected field ( @xmath68 g ) with respect to changes in other single particle properties like effective masses . for even lower densities
( i.e. @xmath73 ) it can be seen ( see fig . 7 in @xcite ) that the @xmath74 term produces a relevant contribution to the magnetization . for application in astrophysical scenarios and at low densities
one should consider that the effect of a non - zero proton fraction determines the appearance of pasta phases @xcite where electromagnetic and nuclear interactions are frustrated and clustering of matter arises .
as density grows , at fixed values of @xmath3 , the spin polarization decreases forming a _ plateau _ at intermediate densities before the possible appearance of a phase transition in the system . in fig .
[ fig3 ] we plot the ratio of change of neutrino mean free paths in the rpa dipolar approximation computed for a generic value of magnetic field strength with respect to the field - free case as a function of the logarithm ( base 10 ) of the magnetic field strength , @xmath75 .
we set a value of density @xmath76 and use sly7 ( solid line ) , sly4 ( long dashed line ) , d1p ( short dashed line ) and d1s ( dotted line ) parametrizations . for fields below @xmath77 g
there is almost no change in the ratio but for larger strengths there is a decrease ( increase ) as computed with gogny ( skyrme ) forces . for this high density case
the change can be @xmath78 as computed with the sly7 parametrization while the gogny d1p predicts a relative change @xmath79 .
note that the main contribution to the mean free paths comes from the fact that , as shown in fig .
[ fig1 ] , the skyrme parametrization predicts a larger change in the absolute and relative values of the two effective masses of the spin polarized components .
the landau parameters and the energetic contribution of the magnetic perturbation @xcite show a minor contribution to the structure functions , that in turn determine the mean free paths .
it is worth mentioning that the lindhard function,@xmath80 , has a rich structure in ( @xmath81 ) that has been studied in @xcite , however , the smallnes of the magnetic perturbation is washed out in the response of the system by the influence of the magnetization @xcite and density effects in the neutron effective mass . as we can see from fig .
[ fig3 ] , this result shows not only quantitative but also qualitative differences in the neutrino transparency of magnetized neutron matter .
we have investigated for the first time in the context of the landau theory of normal fermi liquids , the effect of a strong magnetic field on the variation of the neutrino mean free path in a partially magnetized pure neutron system within the framework of the non - relativistic hartree - fock approximation comparing skyrme and gogny forces .
we find that for fields up to the maximum strength studied in this work , @xmath82 g , skyrme forces show at high density an enhancement of neutrino transparency of the system , while gogny forces predict a small decrease .
these results can be explanined due to the fact that for the density , and @xmath3 field range considered in this work the variation of landau parameters is a minor contribution compared to the effective mass and magnetization .
* acknowledgments * + we acknowledge discussions with j. navarro and a. polls .
this work has been partially funded by the spanish ministry of science and education under projects fis2006 - 05319 , fis2009 - 07238 and junta de castilla y leon excellence program gr234 .
we also acknowledge support by compstar , a research networking programme of the european science foundation .
99 g. baym and c. pethick , landau fermi liquid theory , ed . wiley - vch verlag gmbh and co. kgaa , weinheim ( 2004 ) c. y. cardall , m. prakash and j. lattimer , astrophys .
j. * 554 * 322 ( 2001 ) .
d. bandyopadhyay , s. chakrabarty p. dey and s. pal , phys .
rev . d * 58 * 121301 ( 1998 ) .
s. chakrabarty , phys .
d * 54 * 1306 ( 1996 ) .
d. p. menezes et al , phys .
c * 79 * 035807 ( 2009 ) .
w. m. yao et al ( particle data group ) , j. phys .
g * 33 * 1 ( 2006 ) .
d. e. kharzeev , l. d. mclerran and h. j. wiringa , nuc .
phys . a * 803 * 227 ( 2008 ) .
d. lazzati , nature * 434 * 1075 ( 2005 ) . c. thomson and r. duncan , l. woltjer , mon . not .
r. astron .
soc . * 275 * 255 ( 1995 ) .
h. sawai et al , astrophys . j.*631 * 446 ( 2005 ) .
d. g. yakovlev and c. j. pethick , ann .
rev . of astron . and astrophys .
* 42 * 169 ( 2004 ) .
a. broderick , m. prakash and j. m. lattimer , apj , * 537 * 351 ( 2000 ) .
s. chakrabarty , d. bandyopadhyay and s. pal , phys .
* 78 * 1898 ( 1997 ) .
i. vidaa and i. bombaci , phys .
c * 66 * 045801 ( 2002 ) .
s. fantoni , a. sarsa and k. e. schmidt , phys .
* 87 * ( 2001 ) 181101 .
m. a. prez - garca , phys .
c * 77 * ( 2008 ) 065806 m. a. prez - garca , a. polls and j. navarro , phys .
c * 80 * 025802 ( 2009 ) m. a. prez - garca , phys .
c * 80 * 045804 ( 2009 ) d. vautherin and d. m. brink , phys .
c * 5 * 626 ( 1972 ) .
e. chabanat et al . , nuc .
phys a * 627 * 710 ( 1997 ) e. chabanat et al . , nuc .
phys a * 635 * 231 ( 1998 ) .
m. farine , d. von - eiff , p. schuck , j. f. berger , j. decharg and m. girod , j phys .
g * 25 * 863 ( 1999 ) .
j. f. berger , m. girod and d. gogny , comput .
phys commun .
* 63 * 365 ( 1991 ) .
s. o. bkcman , c. g. kllman and s. sjberg , phys . lett .
* 43b * ( 1973 ) j. navarro , unpublished notes . c. garca - recio , j. navarro , n. van giai and l. l. salcedo , ann . of phys .
* 214 * ( 1992 ) 293 j. margueron , j. navarro and nguyen van giai , phys .
c * 74 * , 015805 ( 2006 ) n. iwamoto and c. j. pethick , phys .
d * 25 * 313 ( 1982 ) .
horowitz , m.a .
prez - garca , and j. piekarewicz , phys .
c * 69 * 045804 ( 2004 ) . | neutrino mean free paths in magnetized neutron matter are calculated using the hartree - fock approximation with effective skyrme and gogny forces in the framework of the landau fermi liquid theory .
it is shown that describing nuclear interaction with skyrme forces and for magnetic field strengths @xmath0 , the neutrino mean free paths stay almost unchanged at intermediate densities but they largely increase at high densities when they are compared to the field - free case results .
however the description with gogny forces differs from the previous and mean free paths stay almonst unchanged or decrease at densities @xmath1\rho_0 $ ] .
this different behaviour can be explained due to the combination of common mild variation of the landau parameters with both types of forces and the values of the nucleon effective mass and induced magnetization of matter under presence of a strong magnetic field as described with the two parametrizations of the nuclear interaction . |
the steadily enlarging collection of unusual heavy hadrons represents a challenge for theory . will it be possible to find a principle ( or a few ) that describe the new phenomena ? or will hadronic phenomenology devolve into a sort of astrophysics , where every heavenly object has its own unique interpretation ?
this report briefly reviews current ideas in hadronic phenomenology and then concludes with an argument that simple extensions to current models can indeed provide the missing guiding principles that are required to understand " the new crop of hadrons . that this is not an easy process is illustrated by the long history of di - positronium . in 1946 ,
wheeler suggested that positronium - positronium bound states are possible@xcite .
that same year ore proved that the system was unbound , and the next year proved that it was bound@xcite .
the _ observation _ of di - positronium had to wait another sixty years@xcite . in this case
the process of reducing quantum electrodynamics to an effective nonrelativistic theory is relatively clean something that can not be said of quantum chromodynamics !
the notion that qcd may manifest at low energy with more complicated structure than @xmath0 or @xmath1 dates from the beginnings of qcd , with fritzsch and collaborators pointing out that gluonic degrees of freedom must appear somewhere in the spectrum@xcite .
the possibility of mixed quark and gluonic matter was also realized early@xcite , as was the notion of multiquark states@xcite .
qcd is notorious for its stubbornness in revealing its secrets . that , and its rather advanced age of 40 years , have led to a proliferation of ideas concerning the emergent low energy behavior of the theory .
some of these are listed in this section .
the simplest idea concerning quark interactions relies on asymptotic freedom to claim that one gluon exchange should dominate at short interquark distances .
this idea is verified by explicit lattice calculations@xcite ; however , this indicates perturbative behavior at _ very _ small distances of order 0.1 fm or less . at somewhat larger interquark separations
the static interaction remains coulombic but behaves as @xmath2 , as is expected from string dynamics@xcite .
this is our first indication that things are not as simple as might naively appear .
although the notion appears trivial , it can cause trouble elsewhere .
for example , the multipole expansion makes use of the quark interaction in a color octet@xcite , which is assumed to be perturbative .
however , it is possible ( and likely ) that this form of the octet interaction is only valid at very small separations because stringy dynamics is quite robust@xcite .
one can , of course , attempt to extend the notion of one gluon exchange interactions by computing higher order corrections in perturbation theory .
efforts along these lines date from the early 80s@xcite and expand on contemporary efforts to classify possible quark interactions in terms of gluonic matrix elements@xcite .
such corrections are possibly relevant to phenomenology because they induce a nontrivial flavor dependence to the interaction : @xmath3 some of the phenomenology of this relevant to the @xmath4 system is worked out in ref.@xcite .
spin - dependence of the interquark interaction is also of interest .
this has been pursued for several decades , with phenomenological ideas settling on the notion of scalar " confinement coupled to
vector " gluon exchange@xcite .
we note , however , that how spin - dependence manifests itself is far from obvious in qcd , and that it certainly is _ not _ of the form of simple relativistic model interactions .
in particular , a scalar confining interaction for baryons implies an _ anticonfining _ interaction for mesons ( along with many other problems such as an unstable vacuum ) .
a better approach is to develop effective field theory capable of addressing this issue .
recent effort in this regard is contained in ref .
alternatively , it is possible to extract spin - dependent interactions in lattice gauge theory@xcite .
this leads to strong constraints in the heavy quark sector .
it is possible that other physical mechanisms modify the short range interaction between quarks .
for example , pion exchange of the type @xmath5\ ] ] has been suggested to be important in the baryon spectrum@xcite .
the important phenomenological feature of this interaction is the flavor dependence evidenced in the factor @xmath6 . in a similar fashion
, instanton - induced interactions have been incorporated into models of the baryon spectrum@xcite .
unfortunately , our knowledge of the baryon spectrum is too rough to be able to choose between options based on the data alone .
there is a long history of lattice gauge computations of static quark interactions .
the usefulness of modern calculations and their relevance to the new exotic hadrons is discussed in section 1.3 of ref .
@xcite and ref .
@xcite . of particular interest is the possibility of determining adiabatic potentials in novel configurations .
for example , quark interactions can be measured in the presence of excited gluonic degrees of freedom@xcite , which is evidently useful for computing the spectrum of heavy quark hybrid mesons .
measurements such as this can be used to validate a series of pre - lattice gauge theory ideas concerning gluonic excitations such as the flux tube model@xcite , the constituent vector particle ( see the last of ref .
@xcite ) , string excitations ( see giles and tye in ref .
@xcite ) , and bag models@xcite . for example
, recent results strongly indicate that an efficient representations of low lying gluonic degrees of freedom is as a quasiparticle with @xmath7@xcite .
this single observation essentially eliminates all of the older gluon models .
a minor industry computing adiabatic potentials for a variety of multiquark configurations exists .
an early attempt at the @xmath8 interaction is contained in ref .
@xcite and more modern results for four - quark and five - quark configurations exist@xcite .
the relatively simple job of determining whether the static three - quark interaction is dominated by two - body interactions or three - body ( the mercedes star " interaction ) is still not settled .
currently it appears that the interaction is predominantly two - body at short distances and slowly transitions to a three - body interaction as the interquark separation grows@xcite .
nascent applications of these ideas to @xmath9 states are presented in ref .
@xcite .
the possibility that one pion exchange is important at short range has been mentioned already .
here we focus on its evident utility for describing long range interactions .
this is , of course , the famous observation of yukawa , and is underpinned by spontaneous chiral symmetry breaking .
pion ( and other meson ) exchange forms a cornerstone for longstanding descriptions of nuclear interactions associated with western european cities@xcite .
twenty years ago tornqvist promoted the idea that pion exchange should be taken seriously for describing the interactions between mesons and suggested that a series of hadronic molecules could exist@xcite .
this has given rise to a ( very ) large collection of theoretical speculation concerning possible binding in many meson - meson systems .
the most famous of these is the @xmath10@xcite .
pion mediated interactions are typically dominated by @xmath11-wave couplings ( for example @xmath12 and @xmath13 ) .
however the possibility of strong @xmath14-wave couplings also exists .
this has been explored by close and collaborators@xcite .
as mentioned , pionic interactions are predicated on chiral symmetry breaking and , as pointed out long ago by weinberg , this means that an effective field theoretic description of low energy pionic interactions is possible .
more recently this idea was extended to nuclear interactions ( for a review of this large field see ref .
an alternative approach has been in the area of unitarized chiral models " in which model assumptions are used to extend the theory to infinite order in the chiral expansion@xcite .
while this certainly removes rigor from the effective field theory it opens many phenomena to modelling .
for example , a recent application to the lhcb pentaquark is contained in ref .
@xcite .
chiral symmetry also comes into play in a novel suggestion due to nowak , rho , and zahed@xcite .
these authors constructed an effective field theory that combined chiral with heavy quark symmetries .
although not terribly predictive , the model does relate scalar - vector doublets of opposite parity and hence predicts @xmath15 this prediction works remarkably well for the @xmath16 and @xmath4 mesons and hence is possibly important in describing the enigmatic @xmath17@xcite .
of course other possibilities exist , including novel short flavor dependent interactions mentioned above ( ref .
@xcite ) and @xmath18 molecular interpretations@xcite . in general
it is possible that meson exchange contributes to hadronic properties , and that it does so in an important way in some cases .
for example , the jlich group has argued that triangle diagrams populated by @xmath16 and @xmath19 mesons dominate hidden flavor decays of charmonia ( such as @xmath20)@xcite .
these diagrams also contribute to the cusp effect , to be discussed below .
the idea that hadrons can be described by singularities in pionic ( or sigma ) fields date to work by skyrme on nuclear interactions@xcite .
several decades later it was suggested that classical pion fields could be created in heavy ion collisions a concept that now goes by the name of disordered chiral condensate"@xcite .
more relevant to this report is the notion that su(3 ) chiral solitons can be
collectively quantized " and that this is justified in the large @xmath21 limit .
indeed , the large @xmath21 limit permits the use of classical and static hedgehog field configurations and suppresses the effects of quantum fluctuations about these configurations by order @xmath22 .
these observations were used by diakonov , petrov , and polyakov to predict an exotic decuplet of five - quark baryons@xcite , the most famous of which was the @xmath23 .
this initiated a storm of activity , initially propelled by low statistics experiments that confirmed the novel prediction . in the end
the pentaquark faded away as high statistics experiments were carried out@xcite .
contemporaneously , it was pointed out that the pentaquark excitations are finite in the large @xmath21 limit , which implies that the pentaquark field configuration is not collective , in contrast with the model assumptions@xcite .
dubynskiy and voloshin have recently postulated the concept of hadrocharmonium " , wherein a compact charmonium state is imbedded in a larger hadron consisting of light degrees of freedom@xcite .
the interaction is provided by a color van der waals force , which is sufficiently strong to provide binding but sufficiently weak to maintain the character of the charmonium component .
the mechanism is analogous to older ideas concerning the possible binding of charmonia and nuclei@xcite and is presumably related to the newly discovered lhcb pentaquark , which was observed in the @xmath24 system@xcite .
a rich spectrum of hadrocharmonium states is expected , in particular resonances decaying into @xmath25 and one or two pions , and resonances with charmonium bound to higher excitations of the light hadronic matter .
the idea of diquarks " predates qcd@xcite .
the original notion was that a pair of quarks could combine to form a degree of freedom with bosonic quantum numbers in a color ( and flavor ) @xmath26 representation .
this idea is often softened these days to refer to a quark - quark correlation in a hadron . in this way
a baryon can be thought of as a diquark ( @xmath27 ) bound to a quark ( @xmath28 ) as follows : @xmath29_{\bar 3 } q \to \bar q q.\ ] ] similarly @xmath30 [ \bar
q \bar q ] \to \bar q q$ ] and @xmath31 [ qq ] \bar q \to \bar q \bar q \bar q$ ] .
the most famous application of these ideas was to the light scalar nonet , which is well - known for its anomalous properties@xcite .
more recently the idea has been expanded to all hadrons@xcite , the old " pentaquarks@xcite , and the @xmath9 state@xcite .
the latter effort assumes scalar and vector @xmath32 $ ] diquarks of mass 1933 mev and combines them to make an infinite spectrum of exotic four - quark states .
for example , @xmath33 decay in which a kaon is produced in conjunction with @xmath34 quarks .
these are postulated to coalesce into diquarks @xmath35 $ ] and @xmath36 $ ] that separate rapidly .
the diquarks separate to large distance , suppressing the fall apart " mode of the putative exotic state .
hadronization occurs at large inter - diquark distance , which enhances to the coupling to excited charmonium , such as @xmath37 , in agreement with experiment for the @xmath38 .
the model has also been applied to the @xmath39 via the hypothesis @xmath40_3 [ cu]_{\bar 3}$]@xcite .
notice that the existence of a state " relies on the dynamics of the production process .
thus this is a new kind of hadron that hinges crucially on the production and decay of the state .
this stands in contrast to the usual interpretation of a hadronic state as a stationary eigenstate of the qcd hamiltonian .
many of the @xmath9 states lie near threshold and therefore are often associated with weakly bound molecular states , in analogue with the @xmath41 interpretation of the @xmath10 .
intriguingly , several of the new states lie just _ above _ threshold : @xmath42 [ @xmath41 ] , @xmath44 [ @xmath45 , @xmath46 [ @xmath47 , and @xmath48 [ @xmath49 . this strongly suggests that these experimental enhancements may be due to threshold cusp effects rather than quark binding .
that something nontrivial can happen at a threshold can be seen in the following two - channel nonrelativistic example .
consider @xmath50 and @xmath51 scattering described by the s - matrix @xmath52 near an @xmath14-wave threshold at @xmath53 , @xmath54 where @xmath55 is a constant and @xmath56 under these conditions @xmath57 as the threshold is approached from above the cross section is well - behaved , however @xmath58 , indicating a discontinuity .
continuing below threshold shows that this can appear as a cusp in the cross section .
the importance of this simple phenomenon to interpreting hadronic reactions has been stressed for many years by bugg@xcite .
more recently it has been applied to the @xmath59 and @xmath60 states with results that show very good agreement with the experimental data with no need for a resonance assumption@xcite .
the model employed in ref .
@xcite was criticized in ref . @xcite as being acausal ; however , it was subsequently shown that nearly identical results can be obtained with a simple unitary and causal model@xcite .
models similar to those of ref .
@xcite have been applied to the @xmath59 states but with different modelling assumptions@xcite . perhaps not surprisingly , these lead to different conclusions about the nature of the @xmath59 states .
nevertheless , the conclusion appears clear : threshold effects should be accounted for in analyses when their presence is deemed to be important .
this brief survey illustrates the breadth of ideas concerning low energy qcd .
it is inevitable that many of them will wither away as experimental data are accumulated .
of course , the capabilities of the lattice gauge theory community continue to grow , both in terms of hardware and in algorithmic sophistication .
thus there is hope that definitive answers will eventually be obtained .
this is good news in view of the difficulty in deciphering something as simple as the short range interaction between quarks !
new @xmath9 states number in the twenties at present , and undoubtedly more are on the way .
it is likely that many of these are statistical fluctuations and will disappear . the remaining demand interpretation .
some physicists think that the @xmath9 states point to the inadequacy of our current , rather weak , understanding of low energy qcd , and that a radical overhaul of phenomenology must be made .
i am inclined to be both more conservative and more hopeful
. it might be that the new states simply highlight the reality of doing experiment high in the spectrum .
it is precisely here that gluonic degrees of freedom must make their presence felt . and
it is here that the effects of light degrees of freedom must become apparent , especially near thresholds .
perhaps what we are experiencing is merely the opening of quark and gluonic degrees of freedom that qcd demands must happen .
the earliest documented work on hybrid mesons is contained in t. barnes , caltech thesis ( 1977 ) .
the earliest published mention of these states is in r.l .
jaffe and k. johnson , phys .
60b * , 201 ( 1976 ) .
the first published explicit models are s .- h .
h. tye , phys . rev . * 13 * , 3416 ( 1976 ) , r. giles and s .-
h h. tye , phys .
lett . * 37 * , 1175 ( 1976 ) and d. horn and j. mandula , phys .
d17 , 898 ( 1978 ) .
o. lakhina and e. s. swanson , phys .
b * 582 * , 172 ( 2004 ) .
s.n . gupta and s.f .
radford , phys .
d * 24 * , 2309 ( 1981 ) ; s.n .
gupta , s.f .
radford , w.w .
repko , phys .
d * 28 * , 1716 ( 1983 ) ; j.t .
pantaleone , s.h.h .
tye , and y.j .
ng , phys . rev .
d * 33 * , 777 ( 1986 ) .
l. y. glozman and d. o. riska , phys .
* 268 * ( 1996 ) 263 .
b. metsch , aip conf .
* 619 * , 589 ( 2002 ) ; u. loring , b. c. metsch and h. r. petry , eur .
j. a * 10 * , 447 ( 2001 ) ; u. loring , b. c. metsch and h. r. petry , eur .
j. a * 10 * , 395 ( 2001 ) .
r. a. briceno _ et al .
_ , `` issues and opportunities in exotic hadrons '' , arxiv:1511.06779 [ hep - ph ] .
e. braaten , c. langmack , and d. h. smith , phys .
d * 90 * , 014044 ( 2014 ) .
n. isgur and j. paton , phys .
d * 31 * , 2910 ( 1985 ) .
t. barnes , nucl .
b158 * , 171 ( 1979 ) ; p. hasenfratz , r.r .
horgan , j. kuti and j.m .
richard , phys . lett . *
95b * , 299 ( 1980 ) ; f. de viron and j. weyers , nucl . phys .
* b185 * , 391 ( 1981 ) ; m.s .
chanowitz and s.r .
sharpe , phys .
b132 * , 413 ( 1983 ) .
h. ichie , v. bornyakov , t. streuer and g. schierholz , the flux distribution of the three quark system in su(3 ) " , arxiv : hep - lat/0212024 ; c. alexandrou , p. de forcrand and a. tsapalis , phys .
d * 65 * , 054503 ( 2002 ) ; t.t .
takahashi , h. suganuma , y. nemoto , and h. matsufuru , phys .
d * 65 * , 114509 ( 2002 ) .
paris : w.n .
cottingham , m. lacombe , b. loiseau , j.m .
richard and r. vinh mau , phys .
d * 8 * , 800 ( 1973 ) ; m. lacombe , b. loiseau , j.m .
richard , r. vinh mau , j. cote , p. pires and r. de tourreil , phys .
c * 21 * , 861 ( 1980 ) .
nijmegen : m.m .
nagels , t.a .
rijken , and j.j .
de swart , phys .
d * 15 * , 2547 ( 1977 ) ; m.m .
nagels , t.a .
rijken , and j.j .
de swart , phys .
d * 20 * , 1633 ( 1979 ) ; v.g.j .
stoks , r.a.m .
kompl , m.c.m .
rentmeester , and j.j .
de swart , phys . rev .
c * 48 * , 792 ( 1993 ) .
f. e. close and p. r. page , phys .
b * 578 * , 119 ( 2004 ) ; e. s. swanson , phys .
b * 588 * , 189 ( 2004 ) ; n. a. tornqvist , phys .
b * 590 * , 209 ( 2004 ) .
f. close and c. downum , phys .
lett . * 102 * , 242003 ( 2009 ) ; f. close , c. downum , and c.e .
thomas , phys .
d * 81 * , 074033 ( 2010 ) .
see , for example , e.e .
kolomeitsev and m.f.m .
lutz , phys .
b * 582 * , 39 ( 2004 ) ; j. nieves and e. ruiz arriola , nucl .
a * 679 * , 57 ( 2000 ) ; j. nieves and e. ruiz arriola , phys .
d * 80 * , 045023 ( 2009 ) .
anselm , phys .
b * 217 * , 169 ( 1989 ) ; a.a .
anselm and m.g .
ryskin , phys .
b * 266 * , 482 ( 1991 ) ; j. d. bjorken , k. l. kowalski and c. c. taylor , observing disoriented chiral condensates " , hep - ph/9309235 ; k. rajagopal and f. wilczek , nucl . phys .
b * 404 * , 577 ( 1993 ) .
brodsky , i. schmidt , and g. f. de terramond , phys .
* 64 * , 1011 ( 1990 ) ; m.e .
luke , a.v .
manohar , and m.j .
savage , phys .
b * 288 * , 355 ( 1992 ) ; s.j .
brodsky , g.a .
miller , phys .
b * 412 * , 125 ( 1997 ) .
r. aaij _ et al . _
[ lhcb collaboration ] , phys .
* 115 * , 072001 ( 2015 ) .
m. ida and r. kobayashi , prog .
phys . * 36 * , 846 ( 1966 ) ; d.b . lichtenberg and l.j .
tassie , phys . rev .
* 155 * , 1601 ( 1967 ) .
a. selem and f. wilczek , hadron systematics and emergent diquarks " , hep - ph/0602128 .
r. l. jaffe , phys .
rept . * 409 * , 1 ( 2005 ) .
l. maiani , f. piccinini , a. d. polosa and v. riquer , phys .
d * 71 * , 014028 ( 2005 ) ; l. maiani , a. d. polosa and v. riquer , phys .
lett . * 99 * , 182003 ( 2007 ) ; l. maiani , a. d. polosa and v. riquer , the charged z(4433 ) : towards a new spectroscopy " , arxiv:0708.3997 [ hep - ph ] .
e. s. swanson , cusps and exotic charmonia " ,
arxiv:1504.07952 [ hep - ph ] . f. k. guo , c. hanhart , q. wang and q. zhao , phys . rev .
d * 91 * , 051504 ( 2015 ) ; z. y. zhou and z. xiao , phys .
d * 92 * , 094024 ( 2015 ) . | various ideas associated with exotic hadrons , and hadronic structure in general , are briefly reviewed . |
the raman scattering technique @xcite has enjoyed widespread applications in molecular spectroscopy and light amplification .
the stimulated raman scattering ( srs ) process is well understood with a full quantum theory @xcite . for raman scattering in an atomic ensemble , due to the limited number of atoms and limited interaction length , the conversion efficiency of the pump field to stokes laser is low and the srs regime is seldom reached . in order to obtain a high efficiency frequency conversion , a seed stokes field is usually injected into the atomic ensemble and is used to generate the srs in a short time .
another method to improve conversion efficiency is coherent anti - stokes raman spectroscopy ( cars ) @xcite , which utilizes the atomic coherence built in stimulated raman process to greatly enhance the anti - stokes component .
this technique was developed into two regimes : one is femtosecond adaptive spectroscopic techniques applied to cars ( fast cars ) , which makes it easy to detect minute biological and chemical agents scully02,beadie , sautenkov , pestov , another is enhanced four - wave mixing ( fwm ) using electromagnetically induced transparency ( eit ) @xcite for generation of non - classical correlations , non - classical photon - pairs lukin98,lv , zibrov99,wal , kuzmich , matsukevich104,kolchin , balic , thompson , du , ooi , yuan and single photons @xcite where the weak generated signals can avoid resonant absorption due to eit .
these works can be understood as a combined write - read process . in the write process , a spin wave ( or an atomic coherence ) is written in the atomic ensemble . in the read process
, the stored spin wave is retrieved by coherent conversion from the atomic states into the anti - stokes field .
there are many ways to prepare atomic coherence , such as eit and stimulated raman adiabatic passage ( stirap ) @xcite .
@xcite achieved a high - frequency conversion efficiency from 425 to 293 nm with the help of an atomic coherence prepared via eit on a raman transition .
sautenkov _
_ @xcite demonstrated the enhancement of coherent anti - stokes laser in rb atomic vapor by a maximal atomic coherence prepared by fractional stirap .
recently , our group observed an enhanced raman scattering ( ers ) effect by the prebuilt atomic spin wave @xcite .
this effect can be understood as a write - write process shown in fig .
[ fig1 ] . in the first step , the stokes field @xmath0
is produced and a spin wave is written into the atomic ensemble by the first write field @xmath1 , which is the same as the first step of the write - read process . in the second step after some delay , differing from the cars process , a second write field @xmath2 is applied for another raman scattering , thus another stokes field @xmath3 is generated .
: initial state preparation laser ; @xmath4 , @xmath5 : the rabi frequencies of write lasers ; @xmath0 , @xmath3 : the generated raman stokes signals ; @xmath6 , @xmath7 : the detunings .
( b ) the time sequence for different fields .
@xmath8 : the delay time . ] in this article , we present a theoretical treatment for the observed ers effect . here , we will show that the intensity of the second stokes field @xmath9 can be enhanced compared to the case with no spin wave prepared initially and depends on the intensity and the spatial distribution of initially prepared atomic spin wave .
our article is organized as follows . in sec .
[ raman ] , we describe our ers based on the spin wave initially prepared by a weak write laser , and give the intensities of the stokes laser ( @xmath10 in the counterpropagation case and in the copropagation case . in sec .
analysis , we numerically calculate the intensities of the usual raman scattering ( urs ) and ers , which shows that the intensities of ers are larger than that of urs and the intensity of counterpropagation case is larger that of copropagation case . in sec
. [ discussion ] , we discuss the ers and urs . finally , we conclude with a summary of our results .
in order to explain why the raman scattering effect shown in fig . 1 can be enhanced , we consider a three - level raman system composed of states @xmath11 .
due to the first write field @xmath12 is absent in the four - level system , the process is the familiar urs . when two write fields @xmath13 and @xmath14 with respective rabi frequencies @xmath15 and
@xmath2 are driven on the atomic system according to the time sequence shown in fig .
[ fig1](b ) , the intensity of stokes field @xmath16 will be enhanced compared to the urs case due to the coherence @xmath17 between states @xmath18 and @xmath19 is built by the first write field @xmath12 @xcite .
the enhancement of the intensity of stokes field @xmath3 is also dependent on the spatial distribution of the flipped atoms generated by the first write field @xmath12 . in the undepleted pump approximation ,
the interaction hamiltonian in a rotating wave frame is given by @xcite @xmath20\},\end{aligned}\]]where @xmath21 , @xmath22 is the atom - field coupling constant , the stokes electric field @xmath23h.c.@xmath24 $ ] , @xmath25 is the number of atoms , and @xmath26 is the quantization volume .
the collective atomic operators are @xmath27 , where @xmath28 is the transition operator of the @xmath29th atom between states @xmath30 and @xmath31 , and there is a small and macroscopic volume containing @xmath32 @xmath33 atoms around position @xmath34 . the stokes field operator @xmath35 obeys the wave equation @xmath36 the equations of motion for the atomic operators in the heisenberg picture are @xmath37where @xmath38 is the coherence ( @xmath39 ) decay rate and @xmath40 is the decay rate of the excited state @xmath41 .
@xmath42 are the langevin noise operators for the atomic operator . due to large detuning @xmath43 , @xmath44
, we can adiabatically eliminate the optical coherence @xmath45 and @xmath46 . in the following ,
we consider the number of atoms scattered to @xmath19 at all times much smaller than @xmath25 , and obtain @xmath47,~~ \label{eq6 } \\
\tilde{\sigma}_{41 } & = & \frac{\omega _ { w_{2}}^{\ast } } { \delta }
e^{-i\mathbf{k}% _ { w_{2}}\cdot \mathbf{r}}\tilde{\sigma}_{11}+i\frac{\hat{f}_{41}}{\delta } . \label{eq7}\end{aligned}\]]using eq . ( [ eq6 ] ) , we have the stokes propagating equation @xmath48where @xmath49 is the creation operator of the atomic spin wave where the wave vector @xmath50 , and @xmath51 is the coupling rate between the collective spin excitation @xmath52 and the stokes field @xmath3 . using eqs .
( [ eq6 ] ) and ( [ eq7 ] ) we have @xmath53where @xmath54 , @xmath55 , @xmath56 , and @xmath57 is the ac stark shift .
@xmath58 is the langevin noise operator and @xmath59 .
the population difference @xmath60 is slowly varying in a time of the order of an optical pulse due to large detuning @xmath61 . from eq .
( [ eq10 ] ) , the atomic population difference is @xmath62 , where @xmath63 .
therefore , @xmath64 in eq .
( [ eq9 ] ) can be replaced by @xmath60 which describes the small changes in population difference .
the equations of motion ( @xmath65)-(@xmath66 ) are the same as that of urs case except the initial conditions are different . in the case of urs @xcite ,
the atomic medium is initially optically pumped to the ground state @xmath18 , then the parameter @xmath67 and the other atomic operators are zero , but in the case of ers , the initial state is a superposition state of states @xmath68 , @xmath69 , and @xmath70 . assume the write field @xmath71
corresponding to a focused beam and the fresnel number @xmath72 ( @xmath73 is the cross - sectional area , and @xmath74 is the cell length ) is of the order of unity , then only a single transverse spatial mode contributes strongly to emission along the direction of the write field @xmath71 .
therefore the above model can be simplified as a one - dimensional model , the propagating quantized stokes field @xmath3 obeys the equation of motion @xcite @xmath75where @xmath76 .
using the moving frame @xmath77 , @xmath78 , the solution is raymer81,andre @xmath79}h_{2}(z^{\prime } , 0,t^{\prime } , t^{\prime \prime } ) \hat{\mathcal{e% } } _ { s_{2}}(0,t^{\prime \prime } ) dt^{\prime \prime } \notag \\ & & + e^{-\gamma _ { 2}(t^{\prime } ) } \int_{0}^{z^{\prime } } g_{s_{2}}(z^{\prime } , z^{\prime \prime } , t^{\prime } , 0)\hat{s}_{a_{2}}^{\dagger } ( z^{\prime \prime } , 0)dz^{\prime \prime } + \int_{0}^{t^{\prime } } e^{-[\gamma _ { 2}(t^{\prime } ) -\gamma _ { 2}(t^{\prime \prime } ) ] } \hat{f}_{s_{2}}^{\dagger } ( z^{\prime } , t^{\prime \prime } ) dt^{\prime \prime } \notag \\ & & + \int_{0}^{t^{\prime } } e^{-[\gamma _ { 2}(t^{\prime } ) -\gamma _ { 2}(t^{\prime \prime } ) ] } \int_{0}^{z^{\prime } } g_{s_{2}}(z^{\prime } , z^{\prime \prime } , t^{\prime } , t^{\prime \prime } ) \hat{f}_{s_{2}}^{\dagger } ( z^{\prime \prime } , t^{\prime \prime } ) dz^{\prime \prime } dt^{\prime \prime } , \label{spinwave } \\
\hat{\mathcal{e}}_{s_{2}}(z^{\prime } , t^{\prime } ) & = & \hat{\mathcal{e}}% _ { s_{2}}(0,t^{\prime } ) + \frac{\chi _ { 2}(t^{\prime } ) } { c}\int_{0}^{t^{\prime } } \mathcal{w}(t^{\prime \prime } )
\chi _ { 2}(t^{\prime \prime } ) e^{-[\gamma _ { 2}(t^{\prime } ) -\gamma _ { 2}(t^{\prime \prime } ) ] } \hat{\mathcal{e}}% _ { s_{2}}(0,t^{\prime \prime } ) g_{e_{2}}(z^{\prime } , 0,t^{\prime } , t^{\prime \prime } ) dt^{\prime \prime } \notag \\ & & + \frac{i\chi _ { 2}(t^{\prime } ) } { c}e^{-\gamma _ { 2}(t^{\prime } ) } \int_{0}^{z^{\prime } } h_{2}(z^{\prime } , z^{\prime \prime } , t^{\prime } , 0)% \hat{s}_{a_{2}}^{\dagger } ( z^{\prime \prime } , 0)dz^{\prime \prime } \notag \\ & & + \frac{i\chi _ { 2}(t^{\prime } ) } { c}\int_{0}^{t^{\prime } } e^{-[\gamma _ { 2}(t^{\prime } ) -\gamma _ { 2}(t^{\prime \prime } ) ] } \int_{0}^{z^{\prime } } h_{2}(z^{\prime } , z^{\prime \prime } , t^{\prime } , t^{\prime \prime } ) \hat{f}% _ { s_{2}}^{\dagger } ( z^{\prime \prime } , t^{\prime \prime } ) dz^{\prime \prime } dt^{\prime \prime } , \label{stokes}\end{aligned}\ ] ] in order to explain the ers , we compare the intensity of the ers with that of the urs .
using eq .
( [ stokes ] ) , the intensity at the end of the atomic cell is given by @xmath85}\int_{0}^{l}h_{2}(l , z^{\prime } , t^{\prime } , 0)\int_{0}^{l}h_{2}(l , z^{\prime \prime } , t^{\prime } , 0)\langle \hat{s}% _ { a_{2}}(z^{\prime } , 0)\hat{s}_{a_{2}}^{\dag } ( z^{\prime \prime } , 0)\rangle dz^{\prime } dz^{\prime \prime } \notag \\ & + & \frac{2\hbar \omega _ { s_{2}}\chi _ { 2}^{2}(t^{\prime } ) l}{c}\gamma _ { s_{2}}\int_{0}^{t^{\prime } } e^{-2\mathrm{re}[\gamma _ { 2}(t^{\prime } ) -\gamma _ { 2}(t^{\prime \prime } ) ] } [ i_{0}^{2}(2\sqrt{\frac{[p_{2}(t^{\prime } ) -p_{2}(t^{\prime \prime } ) ] l}{c}})-i_{1}^{2}(2\sqrt{\frac{[p_{2}(t^{\prime } ) -p_{2}(t^{\prime \prime } ) ] l}{c}})]dt^{\prime \prime } , ~~ \label{intensity1}\end{aligned}\ ] ] where the first and second terms are from the third and fourth terms in eq .
( [ stokes ] ) . according to the commutation relation of spin wave , the term @xmath86 in eq .
( [ intensity1 ] ) is @xmath87for urs , @xmath88 equals zero because no initial spin wave is prepared , but @xmath89 for ers .
the term @xmath90 due to the initially written spin wave , is the key for the appearance of the ers for the write field @xmath71 . for convenience , we assume the write fields intensity being constant @xmath91 $ ] , after being switched on at @xmath92 , so @xmath93 , and @xmath94 with @xmath95 . in the case of urs ( considering a raman system composed of states @xmath96 , @xmath19 , and @xmath41 ) , no initial stokes field is externally incident on the ensemble and no initial spin wave is written into the ensemble , so the initial conditions for the stokes field and the spin wave are@xmath97 @xmath98then from eqs .
( [ intensity1])-([initial1 ] ) the intensity of the stokes field in urs case is @xmath99+e^{-2\gamma _ { s_{2}}t^{\prime } } \notag \\ & \times & \lbrack i_{0}(2\sqrt{\frac{\eta ( t^{\prime } ) \chi _ { 2}^{2}l}{c}% t^{\prime } } ) ^{2}-i_{1}(2\sqrt{\frac{\eta ( t^{\prime } ) \chi _ { 2}^{2}l}{c}% t^{\prime } } ) ^{2}]\ } , \notag \\ & & \label{intensity2}\end{aligned}\ ] ] where @xmath100 $ ] , and @xmath67 in the case of the urs . in the case of ers , like urs , no stokes field is externally incident on the ensemble , but the atomic ensemble contains a spin wave @xmath101 written by the first write field @xmath15 , which the form is @xmath102 , where @xmath103 is the duration of the first write pulse @xcite . when the second write field @xmath104 is driven on the ensemble ,
the initial conditions are given by@xmath97 @xmath105then from eqs .
( [ intensity1 ] ) , ( [ commutation ] ) and ( [ initial2 ] ) , the intensity of the stokes field in the ers case is @xmath106where @xmath107 is the usual raman intensity as eq .
( [ intensity2 ] ) except here @xmath108 in ers case , and @xmath109 is the additional intensity generated by the initially prepared spin wave .
next we investigate the additional intensity @xmath110 .
we first analyze the term @xmath111 .
we consider two different propagation geometries : copropagating or counterpropagating of this two write fields @xmath15 and @xmath71 .
for the copropagating case , we have @xmath112 , so @xmath113and for two write fields counter - propagating case , we have @xmath114 , so @xmath115with @xmath116/\chi _ { 1}^{2}l}i_{0}(2\sqrt{\zeta \frac{z^{\prime } } { l}}% ) ^{2}d\zeta , \label{density}\]]where @xmath117 , @xmath118 ( @xmath119 , @xmath120 ) are the dimensionless strengths annotate , @xmath121 $ ] denotes the real part of a complex number @xmath122 , and the coefficient @xmath123 $ ] is the order of optical depth . figure [ fig2 ] shows the spatial distribution of the flipped - atom number prepared by the first write field @xmath15 from eq .
( [ density ] ) , where the flipped density becomes larger toward the end part of the atomic ensemble , and the flipped density increases with increment of the dimensionless strength @xmath124 .
according to the co - propagating case and the counter - propagating case , the additional intensities @xmath110 can be expressed as @xmath125 and @xmath126 .
thus the intensities of stokes field @xmath3 are written as @xmath127the spatial distribution will result in different ers intensities @xmath128 and @xmath129 for the copropagating case and the counterpropagating case , respectively .
the ratio @xmath130 of the additional intensities @xmath126 and @xmath125 is shown in fig .
[ fig3 ] , from which we know that the additional intensity of the counter - propagating case is much larger that of the co - propagating case . the ratio increases with increment of the dimensionless strength @xmath131 , and also increases with increment of the dimensionless strength @xmath124 . for @xmath132 ( solid line ) and @xmath133 ( dashed line ) with @xmath134=10 $ ] . ] versus the dimensionless strength @xmath135 for @xmath132 ( solid line ) and @xmath136 ( dashed line ) with @xmath137=10 $ ] . ]
@xmath153}+2le^{-2\mathrm{re}[\gamma _ { 1}(t_{1})]}\int_{0}^{z^{\prime \prime } } g_{s_{1}}(z^{\prime \prime } , z^{\prime \prime \prime } , t_{1},0)dz^{\prime \prime \prime } \notag \\ & & + l^{2}e^{-2\mathrm{re}[\gamma _ { 1}(t_{1})]}\int_{0}^{z^{\prime \prime } } g_{s_{1}}^{2}(z^{\prime \prime } , z^{\prime \prime \prime } , t_{1},0)dz^{\prime \prime \prime } + 4\gamma _ { s_{1}}l\int_{0}^{t_{1}}e^{-2% \mathrm{re}[\gamma _ { 1}(t_{1})-\gamma _ { 1}(t^{\prime \prime } ) ] } \int_{0}^{z^{\prime \prime } } g_{s_{1}}(z^{\prime \prime } , z^{\prime \prime \prime } , t_{1},0)dz^{\prime \prime \prime } dt^{\prime \prime } \notag \\ & & + 2\gamma _ { s_{1}}l\int_{0}^{t_{1}}e^{-2\mathrm{re}[\gamma _ { 1}(t_{1})-\gamma _ { 1}(t^{\prime \prime } ) ] } dt^{\prime \prime } + 2\gamma _ { s_{1}}l^{2}\int_{0}^{t_{1}}e^{-2\mathrm{re}[\gamma _ { 1}(t_{1})-\gamma _ { 1}(t^{\prime \prime } ) ] } \int_{0}^{z^{\prime \prime } } g_{s_{1}}^{2}(z^{\prime \prime } , z^{\prime \prime \prime } , t_{1},t^{\prime \prime } ) dz^{\prime \prime \prime } dt^{\prime \prime } , \label{eq32}\end{aligned}\ ] ] and@xmath154\frac{z^{\prime } -z^{\prime \prime } } { c } } ) , \notag \\ & & \gamma _ { 1}(t^{\prime } ) = \gamma _ { s}t^{\prime } + \bar{\gamma}% _ { 1}\int_{0}^{t^{\prime } } [ e^{-30(t^{\prime \prime } /t_{1}-0.5)^{2}}-e^{-7.5}]^{2}dt^{\prime \prime } , \notag \\ & & p_{1}(t^{\prime } ) = \bar{p}_{1}\int_{0}^{t^{\prime } } [ e^{-30(t^{\prime \prime } /t_{1}-0.5)^{2}}-e^{-7.5}]^{2}dt^{\prime \prime }
.\end{aligned}\]]where @xmath155 , and @xmath156 .
for two write fields counter - propagating case , @xmath157 .
figure [ fig4 ] numerically shows the intensities of three cases @xmath158 , @xmath159 , and @xmath160 using eqs .
( [ intensity1 ] ) and ( [ eq32 ] ) . in fig .
[ fig4 ] , we choose the initial population difference @xmath161 , and the other experimental parameters are the same as that were used in ref .
when @xmath162 , the extent of enhancements agrees with our recent experimental results @xcite , where @xmath163 is the order of the optical depth .
when only the write light @xmath71 is turned on , a usual raman scattering occurs , as shown in the line marked with circles in fig
. fig4 . when the write fields @xmath15 and @xmath71 are turned on according to the timing diagram in fig .
[ fig1](b ) and are in the copropagation configuration , an enhancement in @xmath164 occurs , where the experimental result corresponds to the line marked with square and the theoretical numerical calculation corresponds to the solid line in fig . [ fig4](a ) . when the write fields @xmath15 and @xmath71 are in the counterpropagation configuration , a much bigger enhancement effect in @xmath3 is shown in fig .
[ fig4](b ) , where the line marked with pentagram ( the solid line ) is from the experimental ( theoretical ) data . from fig .
[ fig4 ] it is easily seen that the intensity of two write fields counterpropagating case is larger than that of two write fields copropagating case .
the reason why the counter - propagating case has a larger enhancement effect than the copropagating case is as follows . in the copropagating case , when the second write field @xmath71 enters the atomic medium , it first encounters a very small number of the flipped atoms , which is equivalent to a small seed for amplification . on the other hand , in the case of counterpropagating ,
once the second write field @xmath71 enters the medium , it immediately encounters a maximum number of the flipped atoms and starts to amplify it all the way through the atomic medium , which is equivalent to a large seed for amplification . ) , the copropagation case of ers , and the counterpropagation case of ers , respectively .
the parameters are as follows : @xmath165 , @xmath166 , @xmath167 ghz , @xmath168 khz , @xmath169 mhz , and @xmath170 mhz . ] ) , the copropagation case of ers , and the counterpropagation case of ers , respectively .
the parameters are as follows : @xmath165 , @xmath166 , @xmath167 ghz , @xmath168 khz , @xmath169 mhz , and @xmath170 mhz . ]
in this section , let us compare the enhanced raman process with the usual raman process . assume the classical write field does not undergo depletion , and then the usual raman process is generated from spontaneous raman scattering at very small times to transient srs at moderately small times , and finally to steady - state srs .
that is to say , the spontaneous stokes scattering which is from the vacuum acts as the source term to generate srs , so the usual raman process need a period of time to srs and its phase is random . but the enhanced raman process is generated based on the prepared spin wave and it directly and quickly to the srs from the initiation .
the phase of the enhanced stokes laser is from the spin wave and is nonrandom .
the atomic spin wave will act like an input seed to the raman amplification process in the same way as the input stokes field would , so the intensity of stokes field @xmath3 is enhanced .
our scheme can also be explained by the language of nonlinear optics .
after the first write laser is driven on the atomic ensemble , the atomic ensemble is turned into a new medium within a certain coherence time , where the nonlinear coefficient is larger compared to no initial coherence case .
then when another write laser is driven on the atomic ensemble , an enhanced stokes field will occur .
in conclusion , we theoretically demonstrated an enhanced raman effect , which is another mechanism to realize srs instantaneous and to increase the conversion efficiency .
the conversion efficiency of the second write laser is high due to the initially prepared spin wave by the first write field , and the enhancement of the stokes field intensity in two - write - fields counterpropagating case is much larger than that in the two - write - fields copropagating case .
the ers is useful in quantum information , and in nonlinear optics and to detect minute biological and chemical agents . | we present a scheme to obtain the coherently enhanced raman scattering in atomic vapor which is induced by a spin wave initially written by a weak write laser .
the enhancement of raman scattering is dependent on the number and the spatial distribution of the flipped atoms generated by the weak write laser .
such an enhanced raman scattering may have practical applications in quantum information , nonlinear optics and laser spectroscopy because of its simplicity . |
the meson light - cone distribution amplitude ( lcda ) relates to the matrix elements of the nonlocal light - ray operators sandwiched between the hadronic state and the vacuum .
it provides underlying links between the hadronic phenomena at the small and large distances . the leading - twist ( twist-2 )
lcda arouses people s great interests due to its role in hard exclusive processes @xcite , i.e. it always provides dominant contributions to those processes .
the investigation on the lcdas has been the subject of numerous studies , especially for the simpler lcdas of the pseudoscalar mesons pion , kaon and etc . as for the vector @xmath0 meson , its lcdas
are much more involved due to its complex structures : there are chiral - even or chiral - odd lcdas for the @xmath0 meson , the @xmath0 meson has longitudinal ( @xmath10 ) and transverse ( @xmath11 ) polarization states .
several approaches have been adopted to study the @xmath0-meson lcda properties , such as the qcd sum rules @xcite , the light - cone quark model @xcite , the ads / qcd model @xcite and etc .
the properties of the @xmath0-meson lcda are also helpful for us to understand other vector mesons lcdas .
that is , by taking the @xmath12-breaking effect into consideration , one can further study the @xmath13 meson lcdas @xcite .
.the @xmath0-meson lcdas up to twist-4 and @xmath14-order .
[ cols="^,^,^,^,^",options="header " , ] we take two types of semi - leptonic decays as a try to determine the ckm matrix element @xmath8 . one is the @xmath15-type via the process @xmath16 with branching ratio and lifetime @xmath17 and @xmath18ps @xcite .
another is the @xmath19-type via the process @xmath20 with branching ratio and lifetime @xmath21 and @xmath22ps @xcite .
our predicted weighted averages of @xmath8 are given in table [ gammatota3 ] , the babar predictions are also included as a comparison .
the errors come from the choices of the @xmath23-quark mass , the borel window and the threshold parameter @xmath24 , the experimental uncertainties are from the measured lifetimes and decay ratios .
table [ gammatota3 ] show that @xmath8 increases with the increment of @xmath9 .
we observe that by choosing @xmath25 $ ] , its central value is consistent with the babar prediction @xcite .
the babar predictions are based on the lcsr @xcite or isgw @xcite predictions on @xmath3 tffs , respectively .
we have constructed a convenient model ( [ da : wh ] ) for the longitudinal leading - twist lcda @xmath26 , in which a single parameter @xmath27 controls its longitudinal behavior . by varying @xmath28 $ ]
, the @xmath26 shall evolve from a single - peaked behavior to a doubly - humped behavior .
we have discussed its properties via the @xmath3 tffs by using the lcsr approach .
those chiral lcsrs for the tffs shall be dominated ( over @xmath29 ) by @xmath1 either directly or indirectly , then those tffs do provide us a platform to test the properties of the leading - twist @xmath1 . after extrapolating the tffs to their allowable region ,
we compare them to those of lattice qcd predictions .
those tffs become smaller with the increment of @xmath2 , and either a too smaller or a too larger @xmath2 is not allowed by the lattice qcd estimations .
if we have a precise lattice qcd estimation , we can get a more strong constraint on the @xmath0-meson da behavior . in comparison to the babar prediction on the differential decay width @xmath30 @xcite and also the tffs from lattice
qcd , we find that smaller @xmath31 ( or @xmath32 ) is not suitable . as a further comparison with the babar prediction on the @xmath8
, we observe that by choosing @xmath25 $ ] , its central values show a better agreement with the babar prediction @xcite .
thus , we can predict that @xmath26 prefers a doubly - humped behavior other than the sing - peaked behavior of the transverse leading - twist lcda @xmath33 as suggested by ref.@xcite . as a final remark
, the present approach can also be applied to study other vector meson lcdas .
for example , by taking the @xmath12-breaking effect into consideration , one can further study the @xmath13 meson lcdas .
a detailed discussion on the @xmath13 meson lcdas under the same approach of the present paper is in preparation . +
* acknowledgments * : this work was supported in part by natural science foundation of china under grant no.11075225 and no.11275280 , and by the fundamental research funds for the central universities under grant no.cqdxwl-2012-z002 . | in the present paper , we suggest a convenient model for the vector @xmath0-meson longitudinal leading - twist distribution amplitude @xmath1 , whose distribution is controlled by a single parameter @xmath2 . by choosing proper chiral current in the correlator , we obtain new light - cone sum rules ( lcsr ) for the @xmath3 tffs @xmath4 , @xmath5 and @xmath6 , in which the @xmath7-order @xmath1 provides dominant contributions .
then we make a detailed discussion on the @xmath1 properties via those @xmath3 tffs .
a proper choice of @xmath2 can make all the tffs agree with the lattice qcd predictions .
a prediction of @xmath8 has also been presented by using the extrapolated tffs , which indicates that a larger @xmath9 leads to a larger @xmath8 . to compare with the babar data on @xmath8 ,
the longitudinal leading - twist da @xmath1 prefers a doubly - humped behavior . |
in this paper , we consider the following optimization problem : @xmath6 where @xmath2 is a linear map from @xmath7 to @xmath8 , @xmath1 is a proper closed function on @xmath8 and @xmath0 is twice continuously differentiable on @xmath7 with a bounded hessian .
we also assume that the proximal ( set - valued ) mappings @xmath9 are well - defined and are simple to compute for all @xmath10 and for any @xmath4 . here
, @xmath11 denotes the set of minimizers , and the simplicity is understood in the sense that _ at least one _ element of the set of minimizers can be obtained efficiently .
concrete examples of such @xmath1 that arise in applications include functions listed in ( * ? ? ?
* table 1 ) , the @xmath5 regularization @xcite , the @xmath12 regularization , and the indicator functions of the set of vectors with cardinality at most @xmath13 @xcite , matrices with rank at most @xmath14 and @xmath13-sparse vectors in simplex @xcite , etc .
moreover , for a large class of nonconvex functions , a general algorithm has been proposed recently in @xcite for computing the proximal mapping .
the model problem with @xmath0 and @xmath1 satisfying the above assumptions encompasses many important applications in engineering and machine learning ; see , for example , @xcite .
in particular , many sparse learning problems are in the form of with @xmath0 being a loss function , @xmath2 being the identity map and @xmath1 being a regularizer ; see , for example , @xcite for the use of the @xmath12 norm as a regularizer , @xcite for the use of the @xmath15 norm , @xcite for the use of the nuclear norm , and @xcite and the references therein for the use of various continuous difference - of - convex functions with simple proximal mappings . for the case when @xmath2 is not the identity map , an application in stochastic realization where @xmath0 is a least squares loss function , @xmath1 is the rank function and @xmath2 is the linear map that takes the variable @xmath16 into a block hankel matrix was discussed in ( * ? ? ?
* section ii ) .
when @xmath2 is the identity map , the proximal gradient algorithm @xcite ( also known as forward - backward splitting algorithm ) can be applied whose subproblem involves a computation of the proximal mapping of @xmath3 for some @xmath4 .
it is known that when @xmath0 and @xmath1 are convex , the sequence generated from this algorithm is convergent to a globally optimal solution if the step - size is chosen from @xmath17 , where @xmath18 is any number larger than the lipschitz continuity modulus of @xmath19 . for nonconvex @xmath0 and @xmath1
, the step - size can be chosen from @xmath20 so that any cluster point of the sequence generated is stationary ( * ? ? ?
* proposition 2.3 ) ( see section [ sec2 ] for the definition of stationary points ) , and convergence of the whole sequence is guaranteed if the sequence generated is bounded and @xmath21 satisfies the kurdyka - ojasiewicz ( kl ) property ( * ? ? ? * theorem 5.1 , remark 5.2(a ) ) . on the other hand , when @xmath2 is a general linear map so that the computation of the proximal mapping of @xmath22 , @xmath4 , is not necessarily simple , the proximal gradient algorithm can not be applied efficiently . in the case when @xmath0 and @xmath1 are both convex
, one feasible approach is to apply the alternating direction method of multipliers ( admm ) @xcite .
this has been widely used recently ; see , for example @xcite .
while it is tempting to directly apply the admm to the nonconvex problem , convergence has only been shown under specific assumptions .
in particular , in @xcite , the authors studied an application that can be modeled as with @xmath23 , @xmath1 being some risk measures and @xmath2 typically being an injective linear map coming from data .
they showed that any cluster point gives a stationary point , assuming square summability of the successive changes in the dual iterates .
more recently , in @xcite , the authors considered the case when @xmath0 is a nonconvex quadratic and @xmath1 is the sum of the @xmath24 norm and the indicator function of the euclidean norm ball .
they showed that if the penalty parameter is chosen sufficiently large ( with an explicit lower bound ) and the dual iterates satisfy a particular assumption , then any cluster point gives a stationary point .
in particular , their assumption is satisfied if @xmath2 is surjective . motivated by the findings in @xcite , in this paper , we focus on the case when @xmath2 is surjective and consider both the admm ( for a general surjective @xmath2 ) and the proximal gradient algorithm ( for @xmath2 being the identity ) .
the contributions of this paper are as follows : * first , we characterize cluster points of the sequence generated from the admm .
in particular , we show that if the ( fixed ) penalty parameter in the admm is chosen sufficiently large ( with a computable lower bound ) , and a cluster point of the sequence generated exists , then it gives a stationary point of problem .
moreover , our analysis allows replacing @xmath0 in the admm subproblems by its local quadratic approximations so that in each iteration of this variant , the subproblems only involve computing the proximal mapping of @xmath3 for some @xmath4 and solving an unconstrained convex quadratic minimization problem .
furthermore , we also give simple sufficient conditions to guarantee the boundedness of the sequence generated .
these conditions are satisfied in a wide range of applications ; see examples [ examplenew:3 ] , [ examplenew:1 ] and [ examplenew:2 ] .
* second , under the additional assumption that @xmath0 and @xmath1 are semi - algebraic functions , we show that if a cluster point of the sequence generated from the admm exists , it is actually convergent .
our assumption on semi - algebraicity not only can be easily verified or recognized , but also covers a broad class of optimization problems such as problems involving quadratic functions , polyhedral norms and the cardinality function . *
third , we give a concrete 2-dimensional counterexample in example [ ex7:nonconverge ] showing that the admm can be divergent when @xmath2 is assumed to be injective ( instead of surjective ) .
* finally , for the particular case when @xmath2 equals the identity map , we show that the proximal gradient algorithm can be applied with a slightly more flexible step - size rule when @xmath0 is nonconvex ( see theorem [ prop : prox ] for the precise statement ) .
the rest of the paper is organized as follows .
we discuss notation and preliminary materials in the next section .
convergence of the admm is analyzed in section [ sec : msur ] , and section [ sec : mi ] is devoted to the analysis of the proximal gradient algorithm .
some numerical results are presented in section [ sec : num ] to illustrate the algorithms .
we give concluding remarks and discuss future research directions in section [ sec : con ] .
we denote the @xmath25-dimensional euclidean space as @xmath7 , and use @xmath26 to denote the inner product and @xmath27 to denote the norm induced from the inner product .
linear maps are denoted by scripted letters .
the identity map is denoted by @xmath28 . for a linear map @xmath2
, @xmath29 denotes the adjoint linear map with respect to the inner product and @xmath30 is the induced operator norm of @xmath2 .
a linear self - map @xmath31 is called symmetric if @xmath32 . for a symmetric linear self - map @xmath31
, we use @xmath33 to denote its induced quadratic form given by @xmath34 for all @xmath16 , and use @xmath35 ( resp . , @xmath36 ) to denote the maximum ( resp .
, minimum ) eigenvalue of @xmath31 .
a symmetric linear self - map @xmath31 is called positive semidefinite , denoted by @xmath37 ( resp . ,
positive definite , @xmath38 ) if @xmath39 ( resp .
, @xmath40 ) for all nonzero @xmath16 . for two symmetric linear self - maps @xmath41 and @xmath42
, we use @xmath43 ( resp . , @xmath44 ) to denote @xmath45 ( resp . , @xmath46 ) .
an extended - real - valued function @xmath47 is called proper if it is finite somewhere and never equals @xmath48 .
such a function is called closed if it is lower semicontinuous .
given a proper function @xmath49 $ ] , we use the symbol @xmath50 to indicate @xmath51 and @xmath52 .
the domain of @xmath47 is denoted by @xmath53 and is defined as @xmath54 .
our basic _ subdifferential _ of @xmath47 at @xmath55 ( known also as the limiting subdifferential ) is defined by ( see , for example , ( * ? ? ?
* definition 8.3 ) ) @xmath56 it follows immediately from the above definition that this subdifferential has the following robustness property : @xmath57 for a convex function @xmath47 the subdifferential reduces to the classical subdifferential in convex analysis ( see , for example , ( * ? ? ?
* theorem 1.93 ) ) @xmath58 moreover , for a continuously differentiable function @xmath47 , the subdifferential reduces to the derivative of @xmath47 denoted by @xmath59 . for a function @xmath47 with more than one group of variables
, we use @xmath60 ( resp . , @xmath61 ) to denote the subdifferential ( resp . ,
derivative ) of @xmath47 with respect to the variable @xmath16 .
furthermore , we write @xmath62 . in general , the subdifferential set can be nonconvex ( e.g. , for @xmath63 at @xmath64 ) while @xmath65 enjoys comprehensive calculus rules based on _ variational / extremal principles _ of variational analysis @xcite .
in particular , when @xmath2 is a surjective linear map , using ( * ? ? ? * exercise 8.8(c ) ) and ( * ? ? ? * exercise 10.7 ) , we see that @xmath66 for any @xmath67 .
hence , at an optimal solution @xmath68 , the following necessary optimality condition always holds : @xmath69 throughout this paper , we say that @xmath70 is a stationary point of if @xmath70 satisfies in place of @xmath68 . for a continuously differentiable function @xmath71 on @xmath7 ,
the bregman distance @xmath72 is defined as @xmath73 for any @xmath74 , @xmath75 . if @xmath71 is twice continuously differentiable and there exists @xmath76 so that the hessian @xmath77 satisfies @xmath78 ^ 2\preceq \q$ ] for all @xmath16 , then for any @xmath74 and @xmath79 in @xmath7 , we have @xmath80 dt\right\|^2\\ & \le \left(\int_0 ^ 1\left\|\nabla^2\phi(x_2 + t(x_1 - x_2))\cdot[x_1 - x_2]\right\| dt\right)^2\\ & = \left(\int_0 ^ 1\sqrt{\langle x_1-x_2,[\nabla^2\phi(x_2 + t(x_1 - x_2))]^2\cdot[x_1 - x_2]\rangle } dt\right)^2 \le \|x_1 - x_2\|^2_{\q}. \end{split}\ ] ] on the other hand , if there exists @xmath76 so that @xmath81 for all @xmath16 , then @xmath82\rangle ds \
\ge \frac12 \|x_1 - x_2\|^2_{\q } \end{split}\ ] ] for any @xmath74 and @xmath79 in @xmath7 .
a semi - algebraic set @xmath83 is a finite union of sets of the form @xmath84 where @xmath85 and @xmath86 are polynomials with real coefficients in @xmath25 variables .
in other words , @xmath87 is a union of finitely many sets , each defined by finitely many polynomial equalities and strict inequalities .
a map @xmath88 is semi - algebraic if @xmath89 is a semi - algebraic set .
semi - algebraic sets and semi - algebraic mappings enjoy many nice structural properties .
one important property which we will use later on is the kurdyka - ojasiewicz ( kl ) property .
[ def : kl ] * ( kl property & kl function ) * a proper function @xmath47 is said to have the kurdyka - ojasiewicz ( kl ) property at @xmath90 if there exist @xmath91 $ ] , a neighborhood @xmath92 of @xmath93 and a continuous concave function @xmath94 such that : 1 .
@xmath95 and @xmath96 is continuously differentiable on @xmath97 with positive derivatives ; 2 . for all @xmath98 satisfying @xmath99
, it holds that @xmath100 a proper closed function @xmath47 satisfying the kl property at all points in @xmath101 is called a kl function .
it is known that a proper closed semi - algebraic function is a kl function as such a function satisfies the kl property for all points in @xmath101 with @xmath102 for some @xmath103 and some @xmath104 ( for example , see ( * ? ? ?
* section 4.3 ) ; further discussion can be found in ( * ? ? ?
* corollary 16 ) and ( * ? ? ?
* section 2 ) ) .
in this section , we study the alternating direction method of multipliers for finding a stationary point of . to describe the algorithm , we first reformulate as @xmath105 to decouple the linear map and the nonsmooth part .
recall that the augmented lagrangian function for the above problem is defined , for each @xmath106 , as : @xmath107 our algorithm is then presented as follows : + + notice that the first subproblem is essentially computing the proximal mapping of @xmath3 for some @xmath4 .
the above algorithm is called the proximal admm since , in the second subproblem , we allow a proximal term @xmath72 and hence a choice of @xmath71 to simplify this subproblem . if @xmath108 , then this algorithm reduces to the usual admm described in , for example , @xcite . for other popular non - trivial choices of @xmath71 ,
see remark [ rem0 ] below .
we next study global convergence of the above algorithm under suitable assumptions .
specifically , we consider the following assumption .
[ assumption ] 1 .
@xmath109 for some @xmath110 ; and there exist @xmath111 , @xmath112 such that for all @xmath16 , @xmath113 .
2 . @xmath106 and @xmath71 are chosen so that * there exist @xmath114 so that @xmath115 ^ 2\succeq \t^2_2 $ ] for all @xmath16 ; * @xmath116 for some @xmath117 ; * with @xmath118 ^ 2 $ ] for all @xmath16 , there exists @xmath119 so that @xmath120 * ( comments on assumption 1)*[rem0 ] point ( i ) says @xmath2 is surjective . the first and second points in ( ii ) would be satisfied if @xmath121 is chosen to be @xmath122 , where @xmath18 is at least as large as the lipschitz continuity modulus of @xmath123 . in this case , one can pick @xmath124 and @xmath125 .
this choice is of particular interest since it simplifies the @xmath16-update in to a convex quadratic programming problem ; see ( * ? ? ?
* section 2.1 ) . indeed , under this choice , we have @xmath126 and hence the second subproblem becomes @xmath127 finally , point 3 in ( ii )
can always be enforced by picking @xmath128 sufficiently large if @xmath71 , @xmath41 and @xmath42 , are chosen independently of @xmath128 .
in addition , in the case where @xmath129 and hence @xmath125 , it is not hard to show that the requirement that @xmath130 for some @xmath119 is indeed equivalent to imposing @xmath131 . before stating our convergence results , we note first that from the optimality conditions , the iterates generated satisfy @xmath132 hence , if @xmath133 and if for a cluster point @xmath134 of the sequence @xmath135 , we have @xmath136 along a convergent subsequence @xmath137 that converges to @xmath134 , then @xmath138 is a stationary point of . to see this , notice from and the definition of @xmath139 that @xmath140 passing to the limit in along the subsequence @xmath137 and invoking , and , it follows that @xmath141 in particular , @xmath138 is a stationary point of the model problem .
we now state our global convergence result .
our first conclusion establishes under assumption [ assumption ] , and so , any cluster point of the sequence generated from the proximal admm produces a stationary point of our model problem such that holds . in the case where @xmath0 is a nonconvex quadratic function with a negative semi - definite hessian matrix and @xmath1 is the sum of the @xmath15 norm and the indicator function of the euclidean norm ball , the convergence of the admm ( i.e. , proximal admm with @xmath142 ) was established in @xcite .
our convergence analysis below follows the recent work in ( * ? ? ?
* section 3.3 ) and @xcite .
specifically , we follow the idea in @xcite to study the behavior of the augmented lagrangian function along the sequence generated from the proximal admm ; we note that this was subsequently also used in ( * ? ? ?
* section 3.3 ) .
we then bound the changes in @xmath143 by those of @xmath144 , following the brilliant observation in ( * ? ? ?
* section 3.3 ) that the changes in the dual iterates can be controlled by the changes in the primal iterates that correspond to the quadratic in their objective .
however , we would like to point out two major modifications : * ( i ) * the proof in ( * ? ? ? * section 3.3 ) can not be directly applied because our subproblem corresponding to the @xmath145-update is not convex due to the possible nonconvexity of @xmath1 .
our analysis is also complicated by the introduction of the proximal term . *
( ii ) * using the special structure of their problem , the authors in ( * ? ? ?
* section 3.3 ) established that the augmented lagrangian for their problem is uniformly bounded below along the sequence generated from their admm .
in contrast , we _ assume _ existence of cluster points in our convergence analysis below and will discuss sufficient conditions for such an assumption in theorem [ thm : boundedness2 ] .
on the other hand , we have to point out that although our sufficient conditions for boundedness of sequence are general enough to cover a wide range of applications , they do _ not _ cover the particular problem studied in @xcite .
our second conclusion , which is new in the literature studying convergence of admm in the nonconvex scenarios , states that if the algorithm is suitably initialized , we can get a strict improvement in the objective values .
in particular , if suitably initialized , one will not end up with a stationary point with a larger objective value .
[ thm : main ] suppose that assumption [ assumption ] holds .
then we have the following results . 1 . *
( global subsequential convergence ) * if the sequence @xmath135 generated from the proximal admm has a cluster point @xmath134 , then holds .
moreover , @xmath138 is a stationary point of such that holds .
( strict improvement in objective values ) * suppose that the algorithm is initialized at a non - stationary @xmath146 with @xmath147 , and @xmath148 satisfying @xmath149 .
then for any cluster point @xmath134 of the sequence @xmath135 , if exists , we have @xmath150 [ rem1 ] the proximal admm does not necessarily guarantee that the objective value of is decreasing along the sequence @xmath144 generated
. however , under the assumptions in theorem [ thm : main ] , any cluster point of the sequence generated from the proximal admm improves the starting ( non - stationary ) objective value .
we now describe one way of choosing the initialization
as suggested in ( ii ) when @xmath1 is nonconvex . in this case , it is common to approximate @xmath1 by a proper closed convex function @xmath151 and obtain a relaxation to , i.e. , @xmath152 then any stationary point @xmath70 of this relaxed problem , if exists , satisfies @xmath153 .
thus , if @xmath154 , then one can initialize the proximal admm by taking @xmath155 and @xmath156 with @xmath157 , so that the conditions in ( ii ) are satisfied . we start by showing that holds .
first , observe from the second relation in that @xmath158 consequently , we have @xmath159 taking norm on both sides , squaring and making use of ( i ) in assumption [ assumption ] , we obtain further that @xmath160 where @xmath161 is defined in point 3 in ( ii ) of assumption [ assumption ] , and we made use of the relation @xmath162 for the first inequality , while the last inequality follows from points 1 and 3 in ( ii ) of assumption [ assumption ] , and .
on the other hand , from the definition of @xmath139 , we have @xmath163 which implies @xmath164 in view of and , to establish , it suffices to show that @xmath165 we now prove .
we start by noting that @xmath166 next , recall from ( * ? ? ?
* page 553 , ex.17 ) that the operation of taking positive square root preserves the positive semidefinite ordering .
thus , point 1 in ( ii ) of assumption [ assumption ] implies that @xmath167 for all @xmath16 . from this and point 2 in ( ii ) of assumption [ assumption ] ,
we see further that the function @xmath168 is strongly convex with modulus at least @xmath169 . using this , the definition of @xmath170 ( as a minimizer ) and , we have @xmath171 moreover , using the definition of @xmath172 as a minimizer , we have @xmath173 summing , and
, we obtain that @xmath174 summing the above relation from @xmath175 with @xmath176 , we see that @xmath177 where @xmath178 due to point 3 in ( ii ) of assumption [ assumption ] ; and the last inequality follows from @xmath179 . now , suppose that @xmath134 is a cluster point of the sequence @xmath135 and consider a convergent subsequence , i.e. , @xmath180 from lower semicontinuity of @xmath18 , we see that @xmath181 where the last inequality follows from the properness assumption on @xmath1 . on the other hand , putting @xmath182 and @xmath183 in , we see that @xmath184 passing to the limit in and making use of and ( ii ) in assumption [ assumption ] , we conclude that @xmath185 the desired relation now follows from this and the fact that @xmath186 .
consequently , holds .
we next show that holds along the convergent subsequence in . indeed , from the definition of @xmath187 ( as a minimizer )
, we have @xmath188 taking limit and using , we see that @xmath189 on the other hand , from lower semicontinuity , and , we have @xmath190 the above two relations show that @xmath191 .
this together with and the discussions preceding this theorem shows that @xmath138 is a stationary point of and that holds .
this proves ( i ) .
next , we suppose that the algorithm is initialized at a non - stationary @xmath146 with @xmath147 and @xmath148 chosen with @xmath149 ; we also write @xmath192 .
we first show that @xmath193 . to this end
, we notice that @xmath194 proceeding as in , we have @xmath195 on the other hand , combining the relations @xmath196 and @xmath192 , we see that @xmath197 consequently , if @xmath198 , then it follows from and that @xmath199 and @xmath200 .
this together with implies that @xmath201 i.e. , @xmath146 is a stationary point .
since @xmath146 is non - stationary by assumption , we must have @xmath202 .
we now derive an upper bound on @xmath203 for any @xmath204 . to this end , using the definition of augmented lagrangian function , the @xmath205-update and , we have @xmath206 combining this relation with and , we obtain the following estimate @xmath207 on the other hand , by specializing to @xmath208 and recalling that @xmath209 , we see that @xmath210 combining , and the definition of @xmath211 , we obtain @xmath212 where the strictly inequality follows from the fact that @xmath193 , and the fact that @xmath209 .
the conclusion of the theorem now follows by taking limit in the above inequality along any convergent subsequence , and noting that @xmath192 by assumption , and that @xmath213 .
we illustrate in the following examples how the parameters can be chosen in special cases .
[ example:3 ] suppose that @xmath214 and that @xmath19 is lipschitz continuous with modulus bounded by @xmath18 .
then one can take @xmath215 and @xmath216 .
moreover , assumption [ assumption](i ) holds with @xmath217 .
furthermore , one can take @xmath218 so that @xmath219 , @xmath125 and @xmath220 .
for the second and third points of assumption [ assumption](ii ) to hold , one can choose @xmath221 and then @xmath128 can be chosen so that @xmath222 and that @xmath223 these can be achieved by picking @xmath224
. [ example:4 ] suppose again that @xmath214 and @xmath225 for some linear map @xmath226 and vector @xmath227
. then one can take @xmath142 so that @xmath228 , and @xmath229 , @xmath230 , @xmath220 , where @xmath231 .
observe that assumption [ assumption](i ) holds with @xmath217 .
for the second and third points of assumption [ assumption](ii ) to hold , we only need to pick @xmath128 so that @xmath232 , i.e. , @xmath233 , while @xmath234 can be any number chosen from @xmath235 .
[ example:5 ] suppose that @xmath2 is a general surjective linear map and @xmath0 is strongly convex .
specifically , assume that @xmath236 for some @xmath93 so that @xmath237
. then we can take @xmath108 and hence @xmath238 , @xmath239 .
assumption [ assumption](i ) holds with @xmath240 .
the second point of assumption [ assumption](ii ) holds with @xmath241 .
for the third point to hold , it suffices to pick @xmath242 , while @xmath234 can be any number chosen from @xmath243 .
we next give some sufficient conditions under which the sequence @xmath135 generated from the proximal admm under assumption [ assumption ] is bounded .
this would guarantee the existence of cluster point , which is the assumption required in theorem [ thm : main ] . *
( boundedness of sequence generated from the proximal admm)*[thm : boundedness2 ] suppose that assumption [ assumption ] holds , and @xmath128 is further chosen so that there exists @xmath244 with @xmath245 suppose that either 1 . @xmath2 is invertible and @xmath246 ; or 2 .
@xmath247 and @xmath248 .
then the sequence @xmath135 generated from the proximal admm is bounded .
first , observe from that @xmath249 where the last inequality follows from point 3 in ( ii ) of assumption [ assumption ] .
in particular , the sequence @xmath250 is decreasing and consequently , we have , for all @xmath251 , that @xmath252 next , recall from that @xmath253 plugging this into , we see further that @xmath254 where @xmath255 , and @xmath256 is chosen so that @xmath257 , i.e. , @xmath258 . now
, suppose that the conditions in ( i ) hold .
note that @xmath246 implies @xmath248 .
this together with and @xmath257 implies that @xmath259 , @xmath260 , and @xmath261 are bounded .
boundedness of @xmath143 follows from these and .
moreover , the boundedness of @xmath144 follows from the boundedness of @xmath259 , @xmath143 , the invertibility of @xmath2 and the third relation in .
next , consider the conditions in ( ii ) . since @xmath1 is bounded below , and the coerciveness of @xmath262 give the boundedness of @xmath144 .
the boundedness of @xmath143 follows from this and .
finally , the boundedness of @xmath259 follows from these and the third relation in .
this completes the proof .
notice that in order to guarantee boundedness of the sequence generated from the proximal admm , we have to choose @xmath128 to satisfy both assumption [ assumption ] and .
we illustrate the conditions in theorem [ thm : boundedness2 ] in the next few examples .
in particular , we shall see that such a choice of @xmath128 does exist in the following examples .
[ examplenew:3 ] consider the problem in example [ example:3 ] , and suppose in addition that @xmath263 for some linear map @xmath226 and vector @xmath227 , and that @xmath1 is coercive , i.e. , @xmath246 .
this includes the model of @xmath264 regularization considered in @xcite . since @xmath265
, we have @xmath266 where @xmath231 .
thus , holds with @xmath217 and @xmath267 , where @xmath221 .
hence , the sequence generated from the proximal admm is bounded , according to theorem [ thm : boundedness2 ] ( i ) .
[ examplenew:1 ] consider the problem in example [ example:4 ] , and suppose in addition that @xmath1 is coercive , i.e. , @xmath246 .
this covers the model of @xmath264 regularization considered in @xcite .
we show that @xmath135 is bounded by verifying the conditions in theorem [ thm : boundedness2 ] .
indeed , we have from that holds with @xmath217 and @xmath268 ; recall that @xmath231 and @xmath234 can be chosen from @xmath235 in this example .
the conclusion now follows from theorem [ thm : boundedness2 ] ( i ) .
[ examplenew:2 ] consider the problem in example [ example:5 ] , and assume in addition that @xmath248 .
we show that @xmath135 is bounded by showing that holds for our choice of @xmath128 .
the conclusion will then follow from theorem [ thm : boundedness2 ] ( ii ) . to this end , note that @xmath269 and thus @xmath270 thus , holds with @xmath271 ; recall that @xmath234 can be chosen from @xmath243 in this example .
we further comment on the condition .
in particular , we shall argue that for a fairly large class of twice continuously differentiable function @xmath0 with a bounded hessian , there exists @xmath272 so that @xmath273 actually , let @xmath0 be a twice continuously differentiable function with a bounded hessian and @xmath274 .
then it is well known that @xmath275 where @xmath18 is a lipschitz continuity modulus of @xmath123 .
we include a simple proof for the convenience of the readers .
indeed , @xmath276 where the first inequality follows from the fact that @xmath0 is bounded from below by @xmath277 , and the second inequality follows from the fact that the gradient is lipschitz continuous with modulus @xmath18 .
consequently , for a twice continuously differentiable function @xmath0 with a bounded hessian , the condition holds for some @xmath278 if and only if @xmath0 is bounded below .
we now study convergence of the whole sequence generated by the admm ( i.e. , proximal admm with @xmath142 ) when the objective function is semi - algebraic .
the proof of this theorem relies heavily on the kl property . for recent applications of kl property to convergence analysis of a broad class of optimization methods ,
see @xcite .
we would like to point out that our analysis is adapted from @xcite , and we can not directly apply the results there since some of their assumptions are not satisfied in our settings .
we will further comment on this in remark [ rem4 ] . *
( global convergence for the whole sequence)*[th:2 ] suppose that assumption [ assumption ] holds with @xmath129 ( and hence @xmath142 ) , and that @xmath0 and @xmath1 are semi - algebraic functions .
suppose further that the sequence @xmath135 generated from the admm has a cluster point @xmath134 .
then the sequence @xmath135 converges to @xmath134 and @xmath138 is a stationary point of .
moreover , @xmath279 the conclusion that @xmath138 is a stationary point of follows from theorem [ thm : main ] . moreover , holds .
we now establish convergence .
first , consider the subdifferential of @xmath280 at @xmath281 .
specifically , we have @xmath282 where the last two equalities follow from the second and third relations in .
similarly , @xmath283 since @xmath284 from . the above relations together with the assumption that @xmath285 and imply the existence of a constant @xmath286 so that @xmath287 moreover , from and @xmath129 ( and hence @xmath125 ) , we see that @xmath288 for some @xmath289 . in particular , @xmath290 is decreasing .
since @xmath280 is also bounded below along the subsequence in , we conclude that @xmath291 exists .
we now show that @xmath292 ; here , we write @xmath293 for notational simplicity . to this end ,
notice from the definition of @xmath172 as a minimizer that @xmath294 using this relation , and the continuity of @xmath280 with respect to the @xmath16 and @xmath205 variables , we have @xmath295 where @xmath296 is a subsequence that converges to @xmath134 .
on the other hand , from , we see that @xmath297 also converges to @xmath134 .
this together with the lower semicontinuity of @xmath280 imply @xmath298 combining , and the existence of @xmath299 , we conclude that @xmath300 as claimed .
furthermore , if @xmath301 for some @xmath251 , since the sequence is decreasing , we must have @xmath302 for all @xmath303 . from , we see that @xmath304 and hence @xmath305 from the fact that @xmath129 and , for all @xmath303 .
consequently , we conclude from that @xmath306 for all @xmath307 , meaning that the algorithm terminates finitely .
since the conclusion of this theorem holds trivially if the algorithm terminates finitely , from now on , we only consider the case where @xmath308 for all @xmath251 .
next , notice that the function @xmath309 is semi - algebraic due to the semi - algebraicity of @xmath0 and @xmath1 .
thus , it is a kl function from ( * ? ? ?
* section 4.3 ) . from the property of kl functions , there exist @xmath310 , a neighborhood @xmath92 of @xmath134 and a continuous concave function @xmath94 as described in definition [ def : kl ] so that for all @xmath311 satisfying @xmath312 , we have @xmath313 pick @xmath314 so that @xmath315 and set @xmath316 . from the second relation in and , we obtain for any @xmath251 that @xmath317
hence @xmath318 whenever @xmath319 and @xmath251 .
moreover , from the definition of @xmath139 and , we see that whenever @xmath251 , @xmath320 since there exists @xmath321 so that for all @xmath322 , we have @xmath323 ( such an @xmath324 exists due to ) , it follows that @xmath325 whenever @xmath319 and @xmath322 .
thus , if @xmath319 and @xmath322 , we have @xmath326 .
moreover , it is not hard to see that there exists @xmath327 with @xmath328 such that 1 .
@xmath329 ; 2 .
@xmath330 ; 3 . @xmath331 .
indeed , these properties follow from the fact that @xmath134 is a cluster point , and that @xmath332 for all @xmath251 .
we next show that , if @xmath319 and @xmath333 for some fixed @xmath322 , then @xmath334 .
\end{split}\ ] ] to see this , notice that @xmath319 and @xmath322 implies @xmath326 .
hence , holds for @xmath335 . combining , , and the concavity of @xmath71
, we conclude that for all such @xmath336 @xmath337\\ & \ge { \rm dist}(0,\partial l_\beta(x^t , y^t , z^t))\cdot[\varphi(l_\beta(x^t , y^t , z^t ) - l^ * ) - \varphi(l_\beta(x^{t+1},y^{t+1},z^{t+1 } ) - l^*)]\\ & \ge { \rm dist}(0,\partial l_\beta(x^t , y^t , z^t))\cdot\varphi'(l_\beta(x^t , y^t , z^t ) - l^*)\cdot[l_\beta(x^t , y^t , z^t ) - l_\beta(x^{t+1},y^{t+1},z^{t+1})]\\ & \ge d \|x^{t+1}-x^t\|^2 .
\end{split}\ ] ] dividing both sides by @xmath338 , taking square root , using the inequality @xmath339 as in the proof of ( * ? ? ?
* lemma 2.6 ) , and rearranging terms , we conclude that holds .
we now show that @xmath319 whenever @xmath340 .
we establish this claim by induction , and our proof is similar to the proof of ( * ? ? ?
* lemma 2.6 ) .
the claim is true for @xmath341 by construction . for @xmath342
, we have @xmath343 where the first inequality follows from . now
, suppose the claim is true for @xmath344 for some @xmath345 ; i.e. , @xmath346 .
we now consider the case when @xmath347 : @xmath348\\ & \le \|x^n - x^*\| + 2\|x^n - x^{n+1}\|\\ & \ \
+ \frac{c}{d}\sum_{j=1}^{k-1}[\varphi(l_\beta(x^{n+j},y^{n+j},z^{n+j } ) - l^ * ) - \varphi(l_\beta(x^{n+j+1},y^{n+j+1},z^{n+j+1 } ) - l^*)]\\ & \le \|x^n - x^*\| + 2\|x^n - x^{n+1}\| + \frac{c}{d}\varphi(l_\beta(x^{n+1},y^{n+1},z^{n+1 } ) - l^ * ) , \end{split}\ ] ] where the first inequality follows from , the monotonicity of @xmath290 from , and the induction assumption that @xmath346 .
moreover , in view of and the definition of @xmath349 , we see that the last expression above is less than @xmath349 . hence , @xmath350 as claimed , and we have shown that @xmath319 for @xmath340 by induction . since @xmath319 for @xmath340 , we can sum from @xmath341 to @xmath351 .
invoking , we arrive at @xmath352 which implies that holds . convergence of @xmath144 follows immediately from this . convergence of @xmath259 follows from the convergence of @xmath144 , the relation @xmath353 from , and .
finally , the convergence of @xmath143 follows from the surjectivity of @xmath2 , and the relation @xmath354 from .
this completes the proof . *
( comments on theorem [ th:2])*[rem4 ] * a close inspection of the above proof shows that the conclusion of theorem [ th:2 ] continues to hold as long as the augmented lagrangian @xmath280 is a kl - function .
here , we only state the case where @xmath0 and @xmath1 are semi - algebraic because this simple sufficient condition can be easily verified . *
although a general convergence analysis framework was established in @xcite for a broad class of optimization problems , it is not clear to us whether their results can be applied directly here . indeed , to ensure convergence , three basic properties * h1 * , * h2 * and * h3 * were imposed in ( * ? ? ?
* page 99 ) .
in particular , their property * h1 * ( sufficient descent property ) in our case reads : @xmath355 for some @xmath356 . on the other hand , ( [ eq : smallrelation2 ] ) in our proof only gives us that @xmath357 , which is not sufficient for property * h1 * to hold . * in theorem [ th:2 ] ,
we only discussed the case where @xmath108 . this condition is used to ensure that @xmath290 is a decreasing sequence that is at least as large as @xmath358 .
it would be interesting to see whether the analysis here can be further extended to the case where @xmath359 . before ending this section , we comment on the behavior of admm in the case where @xmath2 is assumed to be injective ( instead of surjective ) . as suggested by the numerical experiments in @xcite and our preliminary numerical tests ,
it is conceivable that the admm does _ not _ cluster at a stationary point in general when applied to solving problem with an injective @xmath2 .
we hereby give a concrete 2-dimensional example for non - convergence , motivated by the recent counterexample in ( * ? ? ?
* remark 6 ) for the convergence of douglas - rachford splitting method in a nonconvex setting . ;
see @xcite .
moreover , it has been brought to our attention during the revision process of this paper that the known equivalence between the admm and the dr splitting method in the convex case ( see , for example , ( * ? ? ? * remark 3.14 ) ) can be passed through to the nonconvex cases .
thus , the global convergence results in this paper concerning the admm can be specialized to obtain global convergence of the dr splitting method in some nonconvex settings .
we note that the global convergence of the dr splitting method in the nonconvex settings has been studied in @xcite based on a new specially constructed merit function . ]
[ ex7:nonconverge]*(divergence of admm when @xmath2 is injective ) * fix @xmath360 $ ] and set @xmath361 and @xmath362 . then @xmath363 .
consider the optimization problem @xmath364 this problem corresponds to with @xmath365 , @xmath366 where @xmath367 , and @xmath2 is the linear map so that @xmath368 ; the problem can be equivalently reformulated as @xmath369 and the admm can be applied .
let @xmath370 and @xmath371 denote the multipliers corresponding to the first and second equality constraints , respectively .
the iterates in ( with @xmath142 ) now take the form @xmath372 for concreteness , whenever ambiguity arises in updating @xmath373 via the projection onto the nonconvex ( discrete ) set @xmath338 , we choose the element in @xmath338 that is closest to the previous iterate @xmath374 . for each @xmath375 , consider the initializations @xmath376 , @xmath377 and @xmath378 .
then it is routine to show that the admm described in will exhibit a discrete limit cycle of length @xmath379 .
specifically , @xmath380 for any @xmath381 and @xmath382 .
moreover , @xmath383 in particular , the sequence @xmath144 is not convergent and the successive change of the @xmath205-update does not converge to zero .
in this section , we look at the model problem in the case where @xmath384 . since the objective is the sum of a smooth and a possibly nonsmooth part with a simple proximal mapping , it is natural to consider the proximal gradient algorithm ( also known as the forward - backward splitting algorithm ) . in this approach ,
one considers the update @xmath385 from our assumption on @xmath1 , the update can be performed efficiently via a computation of the proximal mapping of @xmath386 .
when @xmath387 , where @xmath388 , it is not hard to show that any cluster point @xmath138 of the sequence generated above is a stationary point of ; see , for example , @xcite . in what follows
, we analyze the convergence under a slightly more flexible step - size rule .
[ prop : prox ] suppose that there exists a twice continuously differentiable convex function @xmath389 and @xmath390 such that for all @xmath16 , @xmath391 let @xmath144 be generated from with @xmath392 . then the algorithm is a descent algorithm
. moreover , any cluster point @xmath138 of @xmath144 , if exists , is a stationary point .
for the algorithm to converge faster , intuitively , a larger step - size @xmath128 should be chosen ; see also table [ table3 ] .
condition indicates that the concave " part of the smooth objective @xmath0 does not impose any restrictions on the choice of step - size .
this could result in an @xmath393 smaller than the lipschitz continuity modulus of @xmath123 , and hence allow a choice of a larger @xmath128 . on the other hand , since the algorithm is a descent algorithm by theorem [ prop : prox ] , the sequence generated from would be bounded under standard coerciveness assumptions on the objective function .
notice from assumption that @xmath394 is lipschitz continuous with lipschitz continuity modulus at most @xmath393 .
hence @xmath395 from this we see further that @xmath396 where the first inequality follows from , the last inequality follows from the definition of @xmath170 and the subdifferential inequality applied to the function @xmath389 .
since @xmath397 implies @xmath398 , shows that the algorithm is a descent algorithm .
rearranging terms in and summing from @xmath399 to any @xmath400 , we see further that @xmath401 now , let @xmath138 be a cluster point and take any convergent subsequence @xmath402 that converges to @xmath138 . taking limit on both sides of the above inequality along the convergent subsequence , one can see that @xmath403 .
finally , we wish to show that @xmath404 . to this end ,
note first that since @xmath403 , we also have @xmath405 .
then it follows from lower semicontinuity of @xmath1 that @xmath406 . on the other hand , from
, we have @xmath407 which gives @xmath408 . hence , @xmath409 .
now , using this , @xmath410 , and taking limit along the convergent subsequence in the following relation obtained from @xmath411 we see that the conclusion concerning stationary point holds .
we illustrate the above theorem in the following examples .
suppose that @xmath0 admits an explicit representation as a difference of two convex twice continuously differentiable functions @xmath412 , and that @xmath413 has a lipschitz continuous gradient with modulus at most @xmath414
. then holds with @xmath415 and @xmath416 .
hence , the step - size can be chosen from @xmath417 .
a concrete example of this kind is given by @xmath418 , where @xmath419 is a symmetric indefinite matrix . then holds with @xmath420 , where @xmath421 is the projection of @xmath419 onto the cone of nonpositive semidefinite matrices , and @xmath422 .
the step - size @xmath128 can be chosen within the open interval @xmath423 . in the case when @xmath262 is a concave quadratic , say , for example , @xmath424 for some linear map @xmath226 , it is easy to see that holds with @xmath425 for _ any _ positive number @xmath393 .
thus , step - size can be chosen to be any positive number .
suppose that @xmath0 has a lipschitz continuous gradient and it is known that all the eigenvalues of @xmath426 , for any @xmath16 , lie in the interval @xmath427 $ ] with @xmath428 . if @xmath429 , it is clear that @xmath19 is lipschitz continuous with modulus bounded by @xmath430 , and hence the step - size for the proximal gradient algorithm can be chosen from @xmath431 . on the other hand , if @xmath432 , then it is easy to see that holds with @xmath433 and @xmath434 .
hence , the step - size can be chosen from @xmath435 .
we next comment on the convergence of the whole sequence .
we consider the conditions * h1 * through * h3 * on ( * ? ? ?
* page 99 ) .
first , it is easy to see from that * h1 * is satisfied with @xmath436 .
next , notice from that if @xmath437 , then @xmath438 .
moreover , from the definition of @xmath439 , we have @xmath440 for any @xmath441 .
this shows that the condition * h2 * is satisfied with @xmath442 .
finally , ( * ? ? ?
* remark 5.2 ) shows that * h3 * is satisfied . thus , we conclude from ( * ? ? ?
* theorem 2.9 ) that if @xmath443 is a kl - function and a cluster point @xmath138 of the sequence @xmath144 exists , then the whole sequence converges to @xmath138 .
a line - search strategy can also be incorporated to possibly speed up the above algorithm ; see @xcite for the case when @xmath1 is a continuous difference - of - convex function .
the convergence analysis there can be directly adapted .
the result of theorem [ prop : prox ] concerning the interval of viable step - sizes can be used in designing the initial step - size for backtracking in the line - search procedure .
in this section , we perform numerical experiments to illustrate our algorithms .
all codes are written in matlab .
all experiments are performed on a 32-bit desktop machine with an intel@xmath444 i7 - 3770 cpu ( 3.40 ghz ) and a 4.00 gb ram , equipped with matlab 7.13 ( 2011b ) . [
[ minimizing - constraints - violation . ] ] minimizing constraints violation .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we consider the problem of finding the closest point to a given @xmath445 that violates at most @xmath14 out of @xmath446 equations .
the problem is presented as follows : @xmath447 where @xmath448 has full row rank , @xmath449 , @xmath450 .
this can be seen as a special case of by taking @xmath236 and @xmath451 to be the indicator function of the set @xmath452 , which is a proper closed function ; here , @xmath453 is the @xmath12 norm that counts the number of nonzero entries in the vector @xmath145 .
we apply the admm ( i.e. , proximal admm with @xmath142 ) with parameters specified as in example [ example:5 ] , and pick @xmath454 so that @xmath242 . from example
[ examplenew:2 ] , the sequence generated from the admm is always bounded and hence convergence of the sequence is guaranteed by theorem [ th:2 ] .
we compare our model against the standard convex model with the @xmath12 norm replaced by the @xmath15 norm .
this latter model is solved by sdpt3 ( version 4.0 ) , called via cvx ( version 1.22 ) , using default settings . for the admm
, we consider two initializations : setting all variables at the origin ( @xmath455 init . ) , or setting @xmath146 to be the approximate solution @xmath70 obtained from solving the convex model , @xmath192 and @xmath456 ( @xmath15 init . ) . as discussed in remark [ rem1 ] , when @xmath70 is feasible for , this latter initialization satisfies the conditions in theorem [ thm : main](ii ) .
we terminate the admm when the sum of successive changes is small , i.e. , when @xmath457 in our experiments , we consider random instances . in particular , to guarantee that the problem is feasible for a fixed @xmath14 , we generate the matrix @xmath2 and the right hand side @xmath227 using the following matlab codes : .... m = randn(m , n ) ; x_orig = randn(n,1 ) ; j = randperm(m ) ; b = randn(m,1 ) ; b(j(1:m - r ) ) = m(j(1:m - r),:)*x_orig ; % subsystem has a solution .... we then generate @xmath93 with i.i.d .
standard gaussian entries .
we consider @xmath458 , @xmath459 , @xmath460 , @xmath461 and @xmath462 , @xmath463 , @xmath464 , @xmath465 and @xmath466 .
we generate one random instance for each @xmath467 and solve and the corresponding @xmath15 relaxation .
the computational results are shown in table [ table2 ] , where we report the number of violated constraints ( vio ) by the approximate solution @xmath16 obtained , defined as @xmath468 , and the distance from @xmath93 ( dist ) defined as @xmath469 .
we also report the number of iterations the admm takes , as well as the cpu time of both the admm initialized at the origin and sdpt3 called using cvx .
we see that the model allows an explicit control on the number of violated constraints .
in addition , comparing with the @xmath15 model , the @xmath12 model solved using the admm always gives a solution closer to @xmath93 .
finally , the solution obtained from the admm initialized from an approximate solution of the @xmath15 model can be slightly closer to @xmath93 than the solution obtained from the zero initialization , depending on the particular problem instance .
.computational results for perturbation with bounded number of violated equalities . [ cols="^,^,^,^,^,^,^,^,^,^,^,^,^ " , ]
in this paper , we study the proximal admm and the proximal gradient algorithm for solving problem with a general surjective @xmath2 and @xmath214 , respectively . we prove that any cluster point of the sequence generated from the algorithms gives a stationary point by assuming merely a specific choice of parameters and the existence of a cluster point .
we also show that if the functions @xmath0 and @xmath1 are in addition semi - algebraic and the sequence generated by the admm ( i.e. , proximal admm with @xmath108 ) clusters , then the sequence is actually convergent .
furthermore , we give simple sufficient conditions for the boundedness of the sequence generated from the proximal admm .
one interesting future research direction would be to adapt other splitting methods for convex problems to solve , especially in the case when @xmath2 is injective , and study their convergence properties .
the second author would like to thank ernie esser and gabriel goh for enlightening discussions .
the authors would also like to thank the anonymous referees for suggestions that help improve the manuscript .
b. p. w. ames and m. hong .
alternating direction method of multipliers for sparse zero - variance discriminant analysis and principal component analysis .
preprint , january 2014 .
available at ` http://arxiv.org/abs/1401.5492 ` .
h. attouch , j. bolte , p. redont and a. soubeyran .
proximal alternating minimization and projection methods for nonconvex problems .
an approach based on the kurdyka - lojasiewicz inequality .
35 , pp . 438457 ( 2010 ) .
h. attouch , j. bolte and b. f. svaiter .
convergence of descent methods for semi - algebraic and tame problems : proximal algorithms , forward - backward splitting , and regularized gauss - seidel methods .
137 , ser .
a , pp . 91129 ( 2013 ) .
m. fortin and r. glowinski .
on decomposition - coordination methods using an augmented lagrangian . in m. fortin and r. glowinski , eds . , _ augmented lagrangion methods : applications to the solution of boundary problems .
_ north - holland , amsterdam , 1983 .
applications of the method of multipliers to variational inequalities . in m. fortin and r.
glowinski , eds .
, _ augmented lagrangion methods : applications to the solution of boundary problems . _ north - holland , amsterdam , 1983 . p. gong , c. zhang , z. lu , j. huang and j. ye . a general iterative shrinkage and thresholding algorithm for non - convex regularized optimization problems . the 30th international conference on machine learning ( icml 2013 ) . | we consider the problem of minimizing the sum of a smooth function @xmath0 with a bounded hessian , and a nonsmooth function .
we assume that the latter function is a composition of a proper closed function @xmath1 and a surjective linear map @xmath2 , with the proximal mappings of @xmath3 , @xmath4 , simple to compute .
this problem is nonconvex in general and encompasses many important applications in engineering and machine learning . in this paper , we examined two types of splitting methods for solving this nonconvex optimization problem : alternating direction method of multipliers and proximal gradient algorithm . for the direct adaptation of the alternating direction method of multipliers ,
we show that , if the penalty parameter is chosen sufficiently large and the sequence generated has a cluster point , then it gives a stationary point of the nonconvex problem .
we also establish convergence of the whole sequence under an additional assumption that the functions @xmath0 and @xmath1 are semi - algebraic .
furthermore , we give simple sufficient conditions to guarantee boundedness of the sequence generated .
these conditions can be satisfied for a wide range of applications including the least squares problem with the @xmath5 regularization . finally ,
when @xmath2 is the identity so that the proximal gradient algorithm can be efficiently applied , we show that any cluster point is stationary under a slightly more flexible constant step - size rule than what is known in the literature for a nonconvex @xmath0 . |
the relativistic heavy ion collisions at sps and rhic may be the only way to create the extreme conditions necessary to produce a new state of matter quark - gluon plasma ( qgp ) in the laboratory@xcite .
one can attempt to understand the energy density achieved in the collisions by studying the multiplicity and transverse energy distributions through hydrodynamic models@xcite . at sps energies , the global quantities like average multiplicity
, multiplicity distribution and rapidity distribution can be well described by soft processes only , namely by the number of participant nucleons only@xcite .
however , at rhic energies , the measured pseudorapidity density normalized per participant pair for central au - au collisions shows that @xmath1 more particles are produced than at sps@xcite .
this indicates that the yield of particles created by hard scattering processes becomes important at rhic@xcite .
one can decompose the multiplicity at fixed impact parameter into a soft component and a hard component as@xcite @xmath2 where @xmath3 and @xmath0 are the participant number and binary collision number , respectively .
however , the two - component expression ( [ softhard ] ) can be effectively described by a simple power - law form , @xmath4 which is then similar to that measured at sps@xcite . a natural question
then arises : can one find other global observables which are more sensitive to the hard processes than the multiplicity itself , and which can not be effectively described in the models with only soft processes ? as is well known , the multiplicity moments are important characteristics in multiparticle production
. the properties of the multiplicity distribution can be completely described by the normalized moments @xmath5 in ref.@xcite @xmath6 were investigated at sps energies with a general wounded nucleon model .
it was found that the normalized multiplicity moments are independent of the concrete behavior of elementary nucleon - nucleon collisions , but dominated by the normalized participant moments @xmath7 provided that the colliding nuclei are not too light . in this paper
we investigate how the hard processes change the normalized multiplicity moments .
we extend the study in ref.@xcite to including the hard component .
we will focus on the sensitivity of @xmath6 to the colliding energy , nuclear geometry and especially to the geometry fluctuations .
at fixed impact parameter @xmath8 , the nuclear geometry of soft and hard processes is expressed in terms of @xmath3 and @xmath0@xcite , respectively , @xmath9\ , \nonumber\\ n_c(b ) = \int d^2{\bf s } & & \sigma_n t_a({\bf s})t_b({\bf b - s})\ , \end{aligned}\ ] ] where @xmath10 is the nucleon - nucleon inelastic cross section , and @xmath11 and @xmath12 are the local participant densities in the plane orthogonal to the collision axis defined as @xmath13 if the average multiplicity distribution of each soft source is @xmath14 , and the average multiplicity distribution of each hard source is @xmath15 , the multiplicity distribution of an @xmath16 collision at impact parameter @xmath8 is the supposition of the contributions of @xmath3 soft sources and @xmath0 hard sources : @xmath17 it will be seen later that the results in this paper are not concerned with the concrete form of @xmath14 and @xmath15 . taking into account all the processes with different @xmath8 , the final - state multiplicity distribution of the nucleus - nucleus collisions is @xmath18 where @xmath19 is the distribution function of @xmath3 and @xmath0 . since @xmath0 calculated with ( [ npnc ] ) is a monotonous function of @xmath3 , @xmath20 is just the distribution function @xmath21 which can be obtained from ( [ npnc ] ) as @xmath22 we introduce now generating functions@xcite @xmath23 and @xmath24 for the whole system and the elementary soft and hard sources , @xmath25^{n_p } \left[f_c(\theta)\right]^{n_c}\ , \nonumber\\ f_p(\theta ) & \equiv & \sum_{n_p } \theta^{n_p } g_p(n_p)\ , \nonumber\\ f_c(\theta ) & \equiv & \sum_{n_c } \theta^{n_c } g_c(n_c)\ , \ \
-1\le \theta \le 1\ .\end{aligned}\ ] ] differentiating ( [ ffpfc ] ) with respect to @xmath26 and making use of the relations , @xmath27 we derive the multiplicity moments @xmath28 in terms of the elementary soft and hard moments @xmath29 and @xmath30 and the nuclear geometry moments
@xmath31 and @xmath32 , @xmath33 with the definition of the moments , @xmath34 where we have used the minimum participant number @xmath35 to select events .
@xmath36 means minimum - bias events and very large @xmath35 corresponds to central events . with the known multiplicity moments ,
the normalized moments @xmath37 can be expressed as an expansion in the inverse number of average participants @xmath38 , @xmath39 where the average ratio of hard to soft component @xmath40 depends on the elementary nucleon - nucleon dynamics and the nuclear geometry .
if we do not consider peripheral interactions alone , @xmath41 , we can then consider only the zeroth order in the expansion ( [ cis ] ) . in this case , only the average ratio of hard to soft component remains , the other dynamics of elementary soft and hard processes hidden in @xmath29 and @xmath30 with @xmath42 is washed away by the nuclear geometry .
when the hard contribution can be neglected , namely @xmath43 , the normalized multiplicity moments are just the normalized participant moments , @xmath44 this is the case discussed in ref.@xcite at sps energies .
let s first determine the soft and hard components @xmath45 and @xmath46 in elementary nucleon - nucleon collisions . to this end
, we compare the average multiplicity with the experimental data for central @xmath47 collisions .
since we did not introduce rapidity dependence in our discussion , we consider only the central rapidity region where the data show a plateau structure for different centrality bins . by comparing the average participant number @xmath48 , the average multiplicity per participant pair @xmath49 and the multiplicity for @xmath50 @xmath51 with the experimental data@xcite in the central rapidity region @xmath52 at rhic and the parametrization of the @xmath50 data@xcite @xmath53 we can determine at different energies the average ratio @xmath54 and the minimum participant number @xmath35 which is used to select centrality in calculating geometry moments . using a wood - saxon distribution , @xmath55 .
[ tab1 ] with the parameters @xmath56 for @xmath57 and taking @xmath58 at @xmath59 a gev ( @xmath60 for @xmath61 a gev)@xcite , the two parameters are shown in tab.([tab1 ] ) .
we see that at rhic energies , @xmath62 , the soft component is still more important than the hard component .
the influence of nuclear geometry is twofold : the average numbers @xmath48 and @xmath63 and the fluctuations of @xmath3 and @xmath0 around their average values . for central collisions the average numbers @xmath48 and @xmath63 are huge , but the fluctuations are small .
this can be seen clearly in tab.([tab1 ] ) where @xmath64 and @xmath65 . for minimum - bias events the average numbers are relatively small , but
the fluctuations are the maximum .
the multiplicity @xmath66 is only related to the average values @xmath48 and @xmath63 . when the hard contribution vanishes , the average multiplicity is proportional to @xmath48 .
the hard contribution reflected in the ratio @xmath54 leads to an extra @xmath63 dependence .
the centrality dependence of the average multiplicity per participant pair ( [ data1 ] ) can be calculated by changing the minimum participant number @xmath35 from @xmath67 to @xmath68 . in fig.([fig1 ] ) it is compared with the data in the central rapidity region @xmath52 for the central @xmath47 collisions at @xmath69 a gev@xcite . the extra geometry dependence induced by the hard component is weak . since @xmath70 and @xmath71 , the multiplicity moments @xmath28 for @xmath72 are associated with both the average numbers @xmath48 and @xmath63 and the fluctuations in @xmath3 and @xmath0 . from eq.([cis ] ) the normalized moments @xmath6 depend on the fluctuations and the average ratio @xmath54 of hard to soft component .
fig.([fig2 ] ) shows the centrality and energy dependence of @xmath54 . at any energy
the centrality dependence is very weak .
therefore , the behavior of the normalized moments @xmath6 is mainly controlled by the fluctuations in @xmath3 and @xmath0 .
let s first consider the limit of no fluctuations , @xmath73 . in this limit ,
@xmath74 we have @xmath75 in this case there is no difference between the two - component and one - component model . although fluctuations around the average numbers always exist , and it is difficult to choose events with the same impact parameter @xmath8 , namely with the same @xmath3 and @xmath0 , in experiments , for very central collisions with large @xmath48 and @xmath63 , @xmath3 and @xmath0 fluctuate in a narrow region , the case is then similar to the above limit
the fluctuations grow up when the minimum participant number @xmath35 decreases from its maximum value @xmath76 .
fig.([fig3 ] ) shows the centrality dependence of the fluctuations @xmath77 . as the orders @xmath78 and @xmath79 are not too small , the fluctuations are very strong for minimum - bias events . in order to see the contribution from the hard processes , we define the ratio of the normalized moments with and without consideration of the hard component , @xmath80 the centrality and energy dependence of @xmath81 is shown in fig.([fig4 ] ) .
while there is no remarkable difference between @xmath82 and @xmath6 in central collisions , the big fluctuations in @xmath3 and @xmath0 in minimum - bias events enhance the hard contribution , and this enhancement become more and more important when colliding energy increases . at @xmath83a gev ,
the hard contribution to @xmath84 is almost @xmath85 .
the effect of the hard scattering processes on the average multiplicity can be effectively described in the one - component model by modifying the participant number@xcite , @xmath86 by comparing the average multiplicity @xmath87 with the rhic data listed in tab.([tab1 ] ) , we can determine the power @xmath88 and the average contribution of each effective soft source @xmath89 . corresponding to the colliding energy @xmath90 a gev , we have @xmath91 , respectively . in the effective one - component model ,
the normalized moments are just the effective participant moments , @xmath92 when the peripheral interactions are not considered alone .
while the contribution of the hard processes to the average multiplicity through the average binary collision number @xmath63 can be equivalently expressed by increasing the average participant number from @xmath48 to @xmath93 , the fluctuations in @xmath0 can not be effectively included in the fluctuations in @xmath93 .
this can be seen clearly in fig.([fig5 ] ) which shows the ratio @xmath94 as a function of the centrality for @xmath47 collisions . from the comparison with fig.([fig4 ] ) , @xmath95 , the fluctuations in @xmath0 are partly included in the fluctuations in the effective participant number @xmath96
. however , the difference between the two - component model and the effective one - component model is still remarkable in minimum - bias events , especially for the higher order moments and at high energies .
the huge average participant number @xmath48 and binary collision number @xmath63 in relativistic heavy ion collisions make it difficult to extract dynamic information on hard processes from the geometry background .
different from the multiplicity moments @xmath28 which depend on both the average numbers @xmath97 and @xmath63 and the fluctuations in @xmath3 and @xmath0 strongly , the normalized moments @xmath98 have only weak @xmath48 and @xmath63 dependence , and are mainly associated with the fluctuations in @xmath3 and @xmath0 . therefore , the geometry background for @xmath6 is not so complicated as that for @xmath28 .
we have investigated the normalized moments @xmath6 in the frame of a general two - component model . when the hard component can be neglected at sps energies , @xmath6
are completely determined by the geometry fluctuations , the dynamics is totally washed away . when the hard processes become important at rhic energies , the average ratio of hard to soft component depends on the centrality weakly , and @xmath6 are dominated by the fluctuations . for central collisions where the fluctuations are weak , @xmath6 approach to @xmath99 , the dynamic information can not be seen in @xmath6 .
however , the big fluctuations in minimum - bias events make us to see clearly the difference between the models with and without hard component .
while the average effect of the hard processes can be effectively described in the one - component model by modifying the participant number , we have found that the fluctuations in the binary collision number can not be fully included in the fluctuations in the effective participant number . | the normalized multiplicity moments and their relation with soft and hard processes in relativistic heavy ion collisions are analyzed in a general two - component model .
it is found that , the strong fluctuations in binary collision number @xmath0 in minimum - bias events can enhance the hard component , especially for the higher order moments .
this enhancement can not be effectively described by modifying the participant number in the one - component model . |
the study of the magnetic ordering in geometrically frustrated antiferromagnets ( gfaf ) is not a new problem in condensed matter physics .
it was initiated in the fifties by wannier , houtappel , and anderson@xcite .
a further milestone was set by villain@xcite , who first argued that the pyrochlore heisenberg antiferromagnet remains paramagnetic down to 0 k. however , there has been a renewed interest during recent years on these lattices , mainly because , experimentally , a rich phenomenology has been found at low temperatures and novel phases have been identified , even though these systems are relatively simple crystals@xcite . in these materials ,
the elementary unit of the magnetic structure is the triangle , which makes it impossible to satisfy all the antiferromagnetic bonds at the same time , with the result of a macroscopically degenerate ground state . among the lattices with the highest degree of frustration
are the pyrochlore and kagome ones . in the former ,
the magnetic ions occupy the corners of a 3d arrangement of corner sharing tetrahedra ; in the latter , the magnetic ions occupy the corners of a 2d arrangement of corner sharing triangles ( see fig .
[ fig.lattices ] ) . to summarize a very long story ,
it is found that the susceptibility in materials belonging to this class exhibits a high temperature phase in which it follows the curie
weiss ( cw ) law .
below the curie
weiss temperature , geometrical frustration inhibits the formation of a long range ordered ( lro ) state , and the system remains paramagnetic , even though there are strong correlations between units .
this phase is universally present in these systems , and it is called the cooperative paramagnetic phase .
finally , at a certain temperature , @xmath1 , which depends on the particular material and , usually , is well below the cw temperature , there appear non universal phases : some of the systems remain paramagnetic down to the lowest temperature reached experimentally@xcite , some of them exhibit non collinear ordered states@xcite , or even some of them form a spin glass state@xcite , even though the amount of disorder in the structure is very small .
it is thus easy to understand the amount of theoretical attention these systems have generated@xcite . in the simplest theoretical description of gfaf ,
the spins on the lattice are regarded as heisenberg spins with only nearest neighbor ( nn ) interactions . in this picture
, it is predicted that the non trivial degeneracy of the ground state inhibits the formation of a lro state , and the system remains paramagnetic down to zero temperature@xcite .
however , due to the presence of frustration , the nn exchange does not fix an energy scale on the problem , and any small perturbation can break the non trivial degeneracy of the ground state and lead to some kind of ordered state .
therefore , it is especially important to incorporate these possible perturbations in any model that tries to explain the low temperature phases of these systems .
examples of perturbations present in real systems are next nearest neighbor ( nnn ) interactions , small anisotropies , long range dipole dipole interaction , or dilution by non magnetic impurities , to cite some . in a series of recent papers@xcite
, the present authors have generalized the well known constant coupling method of magnetism@xcite , to deal with geometrically frustrated lattices , in both the classical and quantum cases , in the cooperative paramagnetic region .
also , the connection between this technique and the so called phenomenological renormalization group ( rg ) methods for ising spins@xcite has been furnished in ref . , where it was shown that the generalized constant coupling ( gcc ) method can be regarded as the zero - th order approximation of one of these rg techniques , for vectorial or scalar spin systems , the so called , mean field renormalization group method , when correlations outside a cluster are neglected , and the anomalous critical exponent of the system is taken to be zero .
therefore , it is not so surprising that magnetic properties calculated within the framework of the classical gcc method were found to be in excellent agreement with monte carlo calculations for those same quantities@xcite , for both the kagome and pyrochlore lattices . the main idea behind this technique is very simple , and it is based in the experimental and numerical observation that correlations in gfaf are always short ranged@xcite .
therefore , in order to calculate any thermodynamic quantity , we can start by calculating the partition function of an isolated cluster with @xmath2 spins and , later , add the interactions with the surrounding @xmath3 spins outside the cluster ( see fig .
[ fig.lattices ] ) by means of an effective field which is fixed by a self - consistency condition .
this , almost phenomenological , formulation of the method , allows one to easily compute thermodynamic quantities , mainly the susceptibility , in the paramagnetic region , by using a single effective field to characterize all the interactions with spins outside the cluster .
however , it is not very useful to study the critical properties of these systems , except for ferromagnetic interactions , which is not as interesting as the antiferromagnetic case , due to the absence of geometrical frustration ( for heisenberg spins ) .
the reason is very simple .
it is now generally accepted that the ordered state of these gfaf , if any , is not a simple antiferromagnetic configuration , as in bipartite lattices , but a non - collinear ( nc ) one , in which the total spin momentum of the unit is identically zero@xcite .
obviously , we can not characterize such an ordered state by using a single order parameter fixed by a single effective field ( ef ) .
what is even worse , even we know that we have to introduce different order parameters for each of the spins in the cluster , and different ( ef ) for the neighboring spins outside the cluster , it is not easy to intuitively see how these ef enter into the expressions of the order parameters .
therefore , we need to go beyond the phenomenological formulation of the gcc method introduced in the aforementioned works .
this is precisely the intention of this work .
we construct the gcc method from first principles , by using rigorous identities in terms of averages over finite size clusters . in this way
, the ef naturally emerge as averages over the microscopic spin variables .
this microscopic formulation of the gcc method not only allows us to recover all the results obtained refs .
and in the paramagnetic regime but , also , gives us the possible ordered configurations of the system , and the conditions for a transition to one of these states to occur .
moreover , as the gcc method is constructed in the real space , it is especially easy to include further perturbations in the system .
particularly , we study the effect of next nearest neighbor ( nnn ) interactions and site dilution by non - magnetic impurities . apart from the microscopic formulation of the gcc method itself , we present various interesting results in this work .
we find the possible ordered configurations for systems formed by corner sharing units .
these systems have the interesting property that the corresponding lattices are formed by units with @xmath4 spins ( @xmath5 the dimensionality ) , and each of the spins in the unit has @xmath3 nn outside the unit .
thus , for @xmath6 , we have the pyrochlore lattice ; for @xmath7 , the kagome lattice , and for @xmath8 , the linear chain .
we have studied this last case for the sake of completeness , even though this system is not frustrated .
it is found that , apart from the trivial ferromagnetic configuration for ferromagnetic interactions , the ordered state would correspond to a non - collinear configuration given by the condition that the total magnetization of the unit is zero ( for the especial case of the linear chain this corresponds to standard antiferromagnetic order ) . for nn interactions only , there is no transition to such an ordered state at any finite temperature , for none of the systems considered , as it is now generally accepted@xcite .
however , when nnn interactions are taken into account , there is a transition to such a nc ordered state for some ranges of the coupling parameters .
we have only studied the pyrochlore lattice in this case , as we do not think the results of the present method for the critical behavior of these systems are reliable in dimension lower than 3 ( actually , the results of any model based on the idea of finite size clusters ) , as the long wavelength fluctuations around the nc ordered state are not properly taken into account , and they are very likely to be specially important in 2 dimensions ( kagome lattice ) .
also , the effect of site dilution is considered in the present work .
it is found that , in this framework , dilution by itself is not enough to stabilize a nc ordered state .
finally , we also consider both dilution and nnn together , and the phase diagrams for different amounts of dilution are calculated .
the remaining of the paper is organized as follows : in the following section , we introduce the systems studied in this work , the hamiltonian of these systems , the notation , and the self - consistency condition that fixes the ef . in section [ sec.order.parameters ]
we present the calculation of the order parameters for finite cluster of size 1 and @xmath2 ( @xmath9 , and @xmath10 for the pyrochlore , kagome , and linear chain , respectively ) .
the connection with the original formulation of the gcc method is furnished in section [ sec.connection ] . in section
[ sec.critical ] we study the conditions for a transition to a long range ordered state . in section [ sec.nnn ]
we include nnn interactions and study how this additional interaction modifies the conclusions of section [ sec.critical ] . in section [ sec.dilution ]
we develop the formalism to deal with site dilution by non - magnetic impurities , and apply it to the present problem . in section [ sec.nnn.dilution ]
we consider simultaneously the effect of both nnn interactions and site dilution .
section [ sec.conclusions ] is devoted to the discussion of the results and the conclusions .
finally , we have included two appendixes to clarify some of the most involved details of the evaluation of the partition function of the different clusters subject to inhomogeneous ef .
as stated above , the original formulation of the gcc method@xcite is not well suited to study the critical behavior of gfaf .
indeed , it is now generally accepted that the ground state of the pyrochlore and kagome lattices is fixed by the condition that the total spin momentum of the tetrahedral or triangular unit , respectively , is zero , though frustration inhibits the formation of such a state , and the system remains paramagnetic in the whole temperature range , if only nn interactions are present .
obviously , we can not characterize such an ordered state with a single order parameter , as we did in refs . and , which restricts the applicability of the gcc method to the study of the paramagnetic region or , at most , the ferromagnetic case , which is of less interest , due to the absence of geometrical frustration .
a relatively simple way of circumventing this shortcoming of the method has been pointed out in refs . and , and
essentially consists on introducing different sublattices characterized by different order parameters .
the minimum number of sublattices we need to introduce is given by the number of spins in the cluster , that is , 3 and 4 for the kagome and pyrochlore lattices , respectively ( see fig . [ fig.lattices ] ) .
with this subdivision , spins belonging to one sublattice only interact with spins belonging to different sublattices .
therefore , the dimensionless heisenberg hamiltonian of a lattice formed by corner sharing clusters with @xmath2 spins , and only nn interactions , in the presence of a uniform magnetic field can be put in the form@xmath11 where @xmath12 are the classical heisenberg spins of unit length .
the @xmath13 index labels the sublattice to which the considered spin belongs ( and takes the values @xmath14 ) , whereas the @xmath15 index labels the spins belonging to a given sublattice .
@xmath16 ( @xmath17 for ferromagnetic interactions and @xmath18 for antiferromagnetic ones ) and @xmath19 with @xmath20 the applied magnetic field .
the central idea of the gcc method is to calculate the order parameter ( for each sublattice ) for two clusters of different sizes , @xmath2 and @xmath21 , replacing the effect of the spins not included in the cluster by fixed effective fields , which act as a symmetry breaking field ( sbf ) and , by using the self - consistency condition@xmath22 obtain an equation to determine these effective fields and , thus , the magnetization .
the order parameter is calculated by making use of the callen - suzuki identity@xcite@xmath23 where the partial trace is taken over the set of @xmath2 variables specified by the finite size cluster hamiltonian @xmath24 ; @xmath25 is the corresponding order parameter , and @xmath26 indicates the usual canonical thermal average over the ensemble defined by the complete hamiltonian @xmath27 .
let us consider the cluster defined by the spin 1 in sublattice @xmath13 ( see fig . [ fig.1spin.clusters ] ) .
the hamiltonian of this cluster is given by@xmath28 where the subindex @xmath29 makes reference to the fact that the hamiltonian corresponds to a cluster of 1 spin belonging to sublattice @xmath13 , and @xmath30 stands for the symmetry breaking field acting on the spin of the 1-spin cluster , which belongs to sublattice @xmath13 .
the sum over the @xmath15 index is performed over the nn of the @xmath29 spin , with @xmath31 the number of nn in each of the sublattices ( @xmath32 for both the kagome and pyrochlore lattices ) . by using the callen suzuki identity
we can calculate the order parameter in sublattice @xmath13 for a cluster formed by 1 spin@xmath33 as commented above , @xmath34 represents the partial trace with respect to the variables of the spin @xmath29 , and the symbol @xmath26 stands for the usual canonical thermal average taken over the ensemble defined by the complete hamiltonian @xmath27 .
this equation can be put in the more convenient form@xmath35 for classical vectorial spins of fixed length , the trace operation is given by@xmath36 where the integration is performed over the angular degrees of freedom of the spin .
this integral is trivially evaluated to give@xmath37 and , taking into account the well known relation @xmath38 , where @xmath39 is a unitary vector in the direction of @xmath40 , we can easily calculate the order parameter for the 1-spin cluster as@xmath41 where @xmath42 is the langevin function@xcite and @xmath43 is an unitary vector in the direction of the corresponding sbf . near a critical point , as we expect the sbf to be very small , we can expand the previous expression and keep only linear terms in the sbf@xmath44 by defining the effective fields @xmath45 ( @xmath46 ) , we arrive to the final expression for the order parameter@xmath47 where we have taken into account that each spin has 2 nn on each sublattice different from the considered one .
for example , for the pyrochlore lattice with no applied magnetic field we have @xmath48 where @xmath13 , @xmath49 , @xmath50 , and @xmath51 can take the values @xmath52 , @xmath53 , @xmath54 , or @xmath55 .
for the kagome lattice@xmath56 where @xmath13 , @xmath49 , and @xmath50 can take the values @xmath52 , @xmath53 , or @xmath54 . for a cluster with 2 spins
, we have a linear chain , for which @xmath57 where @xmath13 and @xmath49 can take the values @xmath52 or @xmath53 .
another interesting point is that if we keep all the terms in the expansion of the order parameter , and make use of the decoupling approximation @xmath58 and further assume @xmath59 , we recover the usual self consistent curie
weiss equation for the order parameter@xmath60 however , this expression would predict a transition to an ordered state ( for antiferromagnetic interactions ) at a finite temperature given by @xmath61 , which it is well known to be incorrect for gfaf . that is one of the reasons why constructing the effective fields by the self - consistency condition
provides a much better theory for gfaf , as we will see below .
the calculation of the order parameter for a cluster with @xmath2 spins in the presence of the sbf created by the neighboring spins is a highly non - trivial problem , as we will see in the following .
the size of the clusters is chosen so that we consider the smallest number of spins that include all the different sublattices .
that is , @xmath63 for a kagome lattice ( a triangular cluster ) , @xmath64 for a pyrochlore ( a tetrahedral cluster ) , etc ( see fig .
[ fig.pspin.clusters ] ) .
the hamiltonian for such a cluster can be put as@xmath65 in this hamiltonian , the first term represents the interaction between the spins inside the cluster of size @xmath2 , whereas the second term represents the interaction of those same spins with the symmetry breaking fields @xmath66 created by the @xmath3 spins outside the cluster .
the sum over @xmath15 goes over all the nn spins except the ones already included in the cluster .
the order parameter for each of the sublattices included in the cluster , obtained from the @xmath2-spin cluster can be now put as@xmath67 or , more conveniently@xmath68 in these expressions , @xmath69 represents the trace over the spin degrees of freedom in the @xmath2-spin cluster . the reduced trace operation for heisenberg spins
is given by@xmath70 it is impossible to find a closed expression for this partition function of the cluster in the presence of inhomogeneous sbf ( the case with no sbf has been studied by moessner and berlinsky in ref . ) .
however , as we will be mainly interested in the region near the critical point ( if any ) in the absence of applied field , where the sbf are expected to be small , and in the paramagnetic region , where the sbf will be proportional to the applied field and , again , small , it is enough for our purposes to calculate reduced partition function up to second order in the sbf .
the main steps of the calculation are presented in appendix [ appendix1 ] , and the result is given by @xmath71 for @xmath72 , where @xmath73 and@xmath74 expressions for these functions are given in appendix [ appendix2 ] . incidentally
, the first term in corresponds to the partition function for non - interacting clusters with no applied field , which were first calculated by moessner and berlinsky@xcite .
the additional terms in allow us to study not only the paramagnetic regime , but also the behavior of the system near a critical point , as we will see below .
once we have this reduced partition function we can calculate the order parameter up to first order in the sbf by taking into account @xmath75 taking into account the definition of the sbf , and the fact that @xmath76 , after a little algebra we arrive at@xmath77 \vec{h}_{0}+\left ( 1-(p-2)a_{p}(k)\right ) \sum _ { \beta \neq \alpha } \vec{h}'_{\beta } -(p-1)a_{p}(k)\vec{h}'_{\alpha } \right\ } , \ ] ] where we have introduced the function@xmath78
in the following , we will find more useful to express all the thermodynamic quantities in terms of the dimensionless temperature @xmath79 in terms of this parameter , we have@xmath80 and@xmath81 \vec{h}_{0}+\left ( 1-(p-2)a_{p}(\widetilde{t})\right ) \sum _ { \beta \neq \alpha } \vec{h}'_{\beta } -(p-1)a_{p}(\widetilde{t})\vec{h}'_{\alpha } \right\ } , \ ] ] where@xmath82 for ferromagnetic interactions and@xmath83 for antiferromagnetic ones . the self consistency condition ( see eq . ) @xmath84 leads to the following system of linear equations@xmath85 \sum _
{ \beta \neq \alpha } \vec{h}'_{\beta } = -(p-1)\ , a_{p}(\widetilde{t})\ , \vec{h}_{0},\ ] ] that has the solution@xmath86 } \ , \vec{h}_{0}.\ ] ] substituting back in we find@xmath87 that is , the susceptibility is given by@xmath88 by direct substitution , it is simple to verify the relation@xmath89 where @xmath90 is defined as@xcite @xmath91 so we can finally put@xmath92 in complete agreement with the result in ref . .
in the absence of applied field , the system of equations takes the simpler form@xmath93 \sum _ { \beta \neq \alpha } \vec{h}'_{\beta } = \vec{0},\ ] ] and it has two different non - trivial solutions .
the first one occurs if @xmath94 or , equivalently,@xmath95 and corresponds to an ordered state in which@xmath96 we will call this kind of state a non - collinear ( nc ) ordered state .
actually , this condition includes , as particular cases , states in which spins in different sublattices are antiferromagnetically aligned with each other , that is , collinear states . however , the most general state described by this condition is a non - collinear one .
the second solution of occurs if @xmath97 or , equivalently , @xmath98 and corresponds to an ordered state in which @xmath99 that is , ferromagnetic order .
the previous condition for the existence at any finite temperature of a ferromagnetic ordered state is equivalent to studying if the function @xmath100 changes its sign for any finite value of @xmath101 .
this function has been depicted in fig .
[ fig.ferro ] for the pyrochlore lattice , kagome lattice , and linear chain .
we can see that the present method correctly predicts the existence of a transition for the pyrochlore lattice for a value finite value of @xmath101 ( @xmath102 ) , and no finite @xmath103 transition for the linear chain , but incorrectly predicts a transition for the kagome lattice ( @xmath104 ) .
this is due to the fact that the long wavelength fluctuations around the ferromagnetic state are not taken into account in the present method , and these fluctuations destroy the ordered state in 2 dimensions . on the other hand , in the linear chain case ,
even short range correlations , properly described in the model , are enough to destroy any form of long range order .
the condition for the existence of a non - collinear ordered state is equivalent to study if the function @xmath105 changes its sign for any finite value of @xmath101 .
this function has been depicted in fig .
[ fig.noncollinear ] .
we can see that the model correctly predicts that there is no transition to an ordered state for any of the systems considered . for comparison
, we have also applied the present gcc method to the heisenberg model with nn interactions on a simple cubic ( s.c . ) lattice . in the framework of the cw theory ,
it is predicted that both the s.c . and
the pyrochlore lattices should have the same transition temperature to a lro state , as they have the same number of nn .
for the s.c . , it is enough , for our purposes , to consider 2 different sublattices .
each spin in one sublattice has 6 nn in the other sublattice .
once we take into account this difference with respect to the pyrochlore lattice ( which has @xmath3 nn , as mentioned above ) , we obtain that , for ferromagnetic nn interactions , there is a transition to a ferromagnetic state for @xmath106 , to be compared with more refined estimates of this quantity from high - precision mc calculations@xcite , @xmath107 .
for antiferromagnetic interactions , a transition to an antiferromagnetic ( af ) state occurs at @xmath108 .
therefore , we can see that the present method is able to distinguish between the different topologies of the lattices , even when they have the same number of nn .
furthermore , the estimation of the critical point for the s.c . is very good ( up to ~10% ) , even though we have considered the smallest possible cluster that can describe the af ordered state of the s.c .
lattice .
as mentioned in the introduction , one consequence of the presence of geometrical frustration in the system is that the nn exchange does not define an energy scale in the problem .
therefore , any perturbation is a strong perturbation , and can break the non - trivial degeneracy of the ground state , selecting one ordered state . in the following sections
, we will study if nnn interactions and site dilution by non - magnetic impurities , which are always present in real systems , really break this non - trivial degeneracy , that is , if these effects are able to select an ordered state at finite temperature .
furthermore , we will focus the discussion on the pyrochlore lattice , for the reasons explained in the introduction .
in order to include the effect of nnn interactions , we only need to redefine the sbf in the following way@xmath109 , \ ] ] where @xmath110 stands for the number of nnn on each sublattice ( @xmath111 for the kagome lattice and @xmath112 for the pyrochlore lattice ) and @xmath113 , with @xmath114 the nn exchange coupling and @xmath115 the nnn exchange coupling . for the cluster with @xmath2 spins@xmath116 , \ ] ] and@xmath117 \right\ } .\ ] ] therefore , the self - consistency condition that fixes the effective fields is given by@xmath118 \sum _ { \beta \neq \alpha } \vec{h}'_{\beta } = -(p-1)a_{p}(\widetilde{t})\vec{h}_{0},\ ] ] and its solution is given by@xmath119 } \ , \vec{h}_{0}.\ ] ] substituting back into the expression of the 1-spin cluster order parameter we find the expression for the susceptibility@xmath120 or , in terms of the @xmath121 function defined above@xmath122 this susceptibility has been depicted in fig .
[ fig.suscep.nnn ] for different values of @xmath123 and antiferromagnetic nn interactions .
effect of nnn interactions on the susceptibility of the pyrochlore lattice . ] in zero applied field , the system has two non - trivial solutions , similar to the ones found for nn interactions only .
the ferromagnetic solution occurs if @xmath124 and the non - collinear solution occurs if @xmath125 these conditions have been studied and the main results are summarized in fig .
[ fig.phase.diagram.nnn ] . as we can see from those figures
, nnn interactions can stabilize different ordered states depending on the value of @xmath123 and the sign of the nn interaction . for antiferromagnetic nn interactions and @xmath126 ,
the system remains paramagnetic down to 0 k. for antiferromagnetic nn and nnn interactions , a nc ordered estate is selected at finite temperature for @xmath127 .
however , for antiferromagnetic nn interactions , and ferromagnetic nnn interactions , a ferromagnetic ordered state is selected for @xmath128 . on the other hand , the case of ferromagnetic nn interactions is slightly more complex . for ferromagnetic nnn interactions , a ferromagnetic ordered state is selected for @xmath129 . in the range @xmath130 , the system remains paramagnetic down to 0 k and , in the range @xmath128 , a nc ordered state is selected . in any case , from the analysis of this section , it seems very unlikely that the non collinear ordered states experimentally found in some pyrochlore systems are due to nnn interactions@xcite , as in real pyrochlore systems @xmath131 , a value well inside the region of the phase diagram in which the system is found to remain paramagnetic down to 0 k. it is also important to stress that the present treatment of nnn interactions is only approximate .
in fact , in order to treat this problem consistently , we should include new sublattices such as spins in one sublattice do not interact with spins on the same sublattice .
in doing so , it can happen that new kinds of ordered states arise@xcite .
however , our main intention here is to check if the nc state given by is selected by nnn interactions .
the general problem with more sublattices will be studied elsewhere .
in order to study the effect of site dilution on the susceptibility and on the conditions for the appearance of an ordered state , we need to average both the sbf and the order parameters over the distribution of non - magnetic impurities . for simplicity
, we will assume that the distribution of non - magnetic impurities is purely random . under this condition ,
the number of units with @xmath132 spins , for a lattice which , in the absence of dilution , is formed by units with @xmath2 spins , for a concentration of non - magnetic impurities @xmath133 , is given by@xcite@xmath134 however , this distribution is normalized with respect to units , whereas we need a distribution function normalized with respect to a single spin .
the adequate distribution function , for the sbf , is then@xmath135 and verifies the important relations we will use below@xmath136 @xmath137 the corresponding distribution function for averaging over the order parameter , however , has a different normalization . indeed ,
if we average the sbf for the @xmath138spin cluster with ferromagnetic interactions with respect to we have@xmath139 in agreement with the intuitive idea that , upon dilution , the applied field is unaffected , and the ef changes according with the reduction of the average coordination number by a factor @xmath140 , indicating that the distribution function for the sbf is properly normalized .
if we now average the order parameter with respect to the same distribution function , we find@xmath141 and this result is incorrect , as it does not take into account the reduction in the total number of magnetic ions in the lattice upon dilution .
therefore , the correctly normalized distribution function we have to use in order to average over the order parameter is given by@xmath142 in the following , we will use the following notation to denote averages with respect to dilution@xmath143 _ { p}=\sum _ { q=1}^{p}\mathcal{p}_{q}^{p}(x)\ , f_{q},\ ] ] where @xmath144 is any quantity , and it will be understood that we are using the distribution when averaging over sbf and the distribution when averaging over the order parameter . furthermore , for the sake of clarity , we will consider the antiferromagnetic case with no applied field separately from the paramagnetic and ferromagnetic with no applied field cases . in the paramagnetic case and ferromagnetic case in the absence of an applied field , we have seen that the system can be described by a single effective field , @xmath145 . the average with respect to dilution of the sbf for the 1-spin cluster , as we have seen above ,
is given by@xmath146 _
j\right| \widetilde{t}}\vec{h}'+\frac{\vec{h}_{0}}{\left| j\right| \widetilde{t}},\ ] ] where we have only considered the contribution of nn to the sbf .
now , the averaged 1-spin order parameter is given by@xmath147 _
{ p}=\frac{1-x}{3\left| j\right| \widetilde{t}}\left [ \vec{h}_{0}+2(p-1)(1-x)\ , \vec{h}'\right ] .\ ] ] analogously , for the @xmath2-spin cluster we have@xmath148 _
{ p}=\frac{(p-1)(1-x)}{\left| j\right|
\widetilde{t}}\vec{h}'+\frac{\vec{h}_{0}}{\left| j\right| \widetilde{t}},\ ] ] @xmath149 _ { p}=\frac{q-1}{\left| j\right| \widetilde{t}}\left ( ( p-1)(1-x)\vec{h}'+\vec{h}_{0}\right ) , \ ] ] and the corresponding averaged order parameter is given by@xmath150 _
{ p}=\frac{1-x}{3\left| j\right| \widetilde{t}}\left ( 1-\bar{a}_{p}(\widetilde{t})\right ) \left ( \vec{h}_{0}+(p-1)(1-x)\vec{h}'\right ) , \ ] ] where @xmath151 therefore the averaged self - consistency condition @xmath147 _ { p}=\left [ \vec{m}_{q\alpha } \right ] _ { p},\ ] ] leads us to the following equation for the effective field @xmath145@xmath152 that has the solution@xmath153 and the corresponding susceptibility is found to be@xmath154 _ { p}=\frac{1-x}{3\left| j\right| \widetilde{t}}\frac{1-\bar{a}_{p}}{1+\bar{a}_{p}},\ ] ] which is of the same form as , except for that @xmath155 is replaced by the corresponding average over dilution .
interestingly , it is easily shown that this average can be expressed in terms of an average over the @xmath121 function introduced before .
indeed , if we define@xmath156 where @xmath157 is the average of @xmath158 with respect to dilution , given by@xmath159 we have the identity@xmath160 and we can put the averaged susceptibility in the form@xmath154 _
{ p}=\frac{1-x}{3\left| j\right| \widetilde{t}}\frac{1+\bar{\varepsilon } _ { p}}{1-\bar{\varepsilon }
_ { p}}.\ ] ] incidentally , this is precisely the susceptibility calculated by the present authors in ref . ,
so we recover all the results presented there for the paramagnetic case . in the absence of applied field , it can be easily shown from eq . that the condition for the formation of a ferromagnetic ordered state is given by@xmath161 to check the consistency of the results
, we have studied this condition for different amounts of dilution for the ferromagnetic pyrochlore , and the dependence of the transition temperature with @xmath133 is depicted in fig .
[ fig.dilution ] .
we can see that there is a critical value @xmath162 above which there is no transition .
this value is in reasonable agreement with the value of the percolation threshold for the pyrochlore lattice , @xmath163 , whereas our model predicts a value @xmath164 .
this result is reasonable , as the transition temperatures predicted by our model are slightly larger than the real ones , due to neglecting long wavelength fluctuations .
let us now consider the antiferromagnetic case with no applied field , and let us calculate the condition for the existence of non - collinear order upon dilution . in this case , we know from the analysis in the previous sections that the effective fields @xmath165 verify the relation@xmath166 at the transition point , if any . therefore , we have the following expressions for the dilution averaged sbf in this case@xmath146 _ { p}=2\ ,
\left [ \left\langle \vec{\xi } _ { q\alpha } \right\rangle \right ] _ { p}=-\frac{2\vec{h}'_{\alpha } } { \left| j\right| \widetilde{t}},\ ] ] and@xmath149 _ { p}=\frac{\vec{h}'_{\alpha } } { \left| j\right| \widetilde{t}}.\ ] ] the order parameter for a unit with @xmath167 spins is then given by@xmath168 \vec{h}'_{\alpha } .\ ] ] if we now naively average the self consistency condition as above , we would not be able to obtain the condition for the transition to a non - collinear state in terms of the @xmath169 or @xmath170 functions introduced before , as a @xmath171 prefactor is lacking in the expression for @xmath172 .
thus , we construct the following averages instead@xmath173 _ { p}=-\frac{2(p-1)(1-x)}{3\left| j\right| \widetilde{t}}\vec{h}'_{\alpha } , \ ] ] and@xmath174 _ { p}=-\frac{1-x}{3\left| j\right| \widetilde{t}}\left ( ( p-1)(1-x)+\bar{a}_{p}\ , \right ) \ , \vec{h}'_{\alpha } .\ ] ] equating both quantities , we easily obtain the condition for the existence of non - collinear order upon site dilution@xmath175 where we have made use of relation . as can be seen from fig .
[ fig.cond.nc.dilution ] , the function @xmath176 does not change its sign for any value of @xmath133 , indicating that a non - collinear ordered state is not selected by site dilution , in contrast with the qualitative picture suggested in other works .
of course , in real systems , the effects analyzed above are always present simultaneously .
therefore , it is important to study how the conclusions reached in the previous sections are modified when we simultaneously consider both nnn interactions and site dilution .
we will only present the main results of the analysis , as the derivation of the final expressions follows very closely the previous derivations .
the averaged susceptibility is obtained to be@xmath154 _ { p}=\frac{1-x}{3\left| j_{1}\right| \widetilde{t}}\ , \frac{1+\bar{\varepsilon } _ { p}}{1-(1+z'\ , \lambda ) \bar{\varepsilon } _ { p}}.\ ] ] the condition for the existence of a ferromagnetic ordered state
is given by @xmath177 whereas the corresponding one for a non - collinear ordered state is given by @xmath178 the phase diagrams , for different concentrations of non - magnetic impurities are depicted in fig .
[ fig.phase.diagram.nnn.dilution ] . also , in fig .
[ fig.lambdac ] we have depicted the values of @xmath179 , that is , the value of @xmath180 above which a non - collinear ordered state is formed at a finite temperature , for a given value of @xmath133 .
we have only depicted this quantity for antiferromagnetic nn interactions , as the corresponding @xmath181 for ferromagnetic nn interactions does not change upon dilution .
minimum value of @xmath182 for which a nc ordered state is formed for different concentrations of non - magnetic impurities . ]
in this work we have presented a microscopic formulation of the generalized constant coupling method for geometrically frustrated magnets . by introducing different bravais sublattices characterized by different order parameters we are able to describe more general kinds of long range ordered states , including non - collinear ones .
these order parameters are evaluated by making use of the suzuki - callen identity for finite size clusters , in terms of symmetry breaking fields arising from interactions with units surrounding the considered one .
the symmetry breaking fields are easily related with the phenomenological effective fields introduced in the original formulation of the generalized constant coupling method , in terms of averages over the neighboring spin variables outside the unit , and they are fixed by a self - consistency condition . though we have not stressed this point , this self - consistency condition is , in fact , a particular case of the more general scaling hypothesis between the order parameters for finite clusters of different sizes@xcite . when a magnetic field is applied to the system
, this method allows us to easily compute the susceptibility , and we recover previous results obtained in the phenomenological formulation of the gcc method , which were shown to be in excellent agreement with mc results for both the kagome and pyrochlore lattice .
however , the present formulation , allows us to also study the critical behavior of the system when no applied magnetic field is present .
the possible ordered configurations for these systems naturally arise from the equations that fix the order parameters . in completely agreement with the accepted idea , for ferromagnetic interactions a ferromagnetic configuration in which all the spins point along the same direction is found . for antiferromagnetic interactions , a non - collinear state in which the total magnetization of each cluster is identically zero
also , the conditions for the formation of such ordered states arise from the calculation .
it is found that , for nearest neighbor antiferromagnetic interactions , frustration inhibits the formation of a long range ordered state , and the system remains paramagnetic down to 0 k. we have briefly compared the results for these frustrated geometries with the results for a simple cubic lattice obtained with this same method , and we found that the generalized constant coupling method correctly distinguishes between different topologies of the lattice , even in case where the number of nearest neighbors is the same , as it happens with the pyrochlore and cubic lattices .
another advantage of the present formulation is that it is constructed in the real space , which makes it very easy to include additional perturbations to the system , by modifying the form of the symmetry breaking fields .
particularly , we have studied the effect of next nearest neighbor interactions and site dilution by non - magnetic impurities for the pyrochlore lattice . in the former case , the phase diagram has been calculated .
it is found that , for certain values of the ratio @xmath113 , a non - collinear or ferromagnetic ordered state can be stabilized . however , the minimum value of @xmath123 above which a non - collinear configuration is estable seems to be too large to explain the formation of long range order found in some systems , notably in gd@xmath183ti@xmath183o@xmath184@xcite .
moreover , it seems clear from neutron scattering that the ordered configuration in this case is not compatible with the condition that the total magnetization of the tetrahedral cluster is zero@xcite .
however , the main perturbation in this system seems to come from long range dipolar interactions , which are highly anysotropic . further theoretical work in this direction , in the present framework , is still necessary . regarding the effects of site dilution
, it has been argued that this kind of effect , always present in real materials , could break the non - trivial degeneracy of the ground state and cause a transition to some kind of long range ordered state at finite temperature .
however , in our calculation , we find that this is not case : site dilution by itself does not induce a transition , for any amount of dilution .
it is necessary to include simultaneously both dilution and next nearest neighbor interactions .
a word of caution must be said here , however . in the last part of this work ,
we have focused our attention on studying if a non - collinear ordered state given by the rule that the total magnetization of the tetrahedral unit is zero is estabilized by any of these perturbations . however , these are not the only possible ordered configurations when we include these effects . for example ,
reimers and coworkers@xcite showed that , in the framework of mean field theory , next nearest neighbor interactions can induce both non - collinear ordered states as the ones studied in this work , or ordered states characterized by an inconmensurate wave vector .
it is not easy to describe these inconmesurate states with a real space method , and it could be the case that site dilution would lead to such an inconmesurate state . therefore , more theoretical work in that direction should be necesary before ruling out the possibility of a long range ordered state induced by small amounts of site dilution or , what would be even more interesting , the possibility of a transition to some kind of spin glass state induced by dilution@xcite . in conclusion ,
the formulation of the generalized constant coupling method presented in this work provides us a conceptual framework in which both thermodynamic quantities and the critical behavior of geometrically frustrated magnets can be properly described
. the method can be easily generalized to deal with further perturbations always present in real systems .
we have presented the classical limit of the method . however , in order to compare the calculated quantities with experimental data for these systems , the corresponding quantum generalized constant coupling method should be used .
this issue , the occurrence of inconmensurate states due to dilution , and the effect of the inclusion of long range dipolar interactions in the hamiltonian are open issues which deserve further theoretical work .
in this appendix we present the main step for the evaluation of the integral @xmath185 which can be put in the more convenient form@xmath186 by using the fourier transform representation of the dirac-@xmath51@xmath187 we can further put@xmath188 where we have introduced the vector @xmath189 , and made use of the fact@xmath190 with @xmath191 the total spin of the unit , up to an unimportant global constant in the hamiltonian .
the integral over @xmath192 can be done as follows .
taking into account@xmath193 where we have assumed that the length of @xmath192 is 1 ( this can be always done by redefining @xmath194 with @xmath195 the length of the spin ) , the integral over @xmath192 can be put as + @xmath196 where @xmath197 ( we have factored out a ( -1 ) in the definition of @xmath198 ) .
next , we can do the integral over @xmath191 @xmath199 which only converges for @xmath18 , and we are left with an integral over @xmath200@xmath201 which can not be calculated in a closed form . an interesting point here
is that , if we take @xmath202 , we recover the partition function calculated in the original gcc model .
the only thing we can do at this point is to expand the partition function in powers of the sbf , up to second order , as the rest of terms will be assumed to be small near the critical point once we take the thermal average . with a little algebra
it is easy to show that@xmath203 where @xmath204 is the angle defined by @xmath205 and @xmath200 , and@xmath206
@xmath207 @xmath208 @xmath209 by making use of this result , we can put @xmath210 + the expressions of the form @xmath211 are the projection of @xmath205 over @xmath200 , and can be put in the more convenient form@xmath212 if we consider the imaginary part of the previous series , and integrate over the angular part of @xmath200 in , we have integrals of the form @xmath213 which are easily shown to be zero
. we also have integrals of the form@xmath214 = \frac{4\pi } { 3}\xi _ { p\alpha } ^{2}\ ] ] and@xmath215 = \frac{4\pi } { 3}\vec{\xi } _ { p\alpha } \cdot \vec{\xi } _ { p\beta } .\ ] ] substituting these results back in we are left with the integral over @xmath132@xmath216 which , taking into account @xmath217 can be simplified to the form@xmath218 each of the integrals can be put as a closed expressions , and are given in the next section . for the moment , we will use the notation @xmath219
and@xmath220 finally , up to second order in the sbf , we can put@xmath221
the @xmath224 function for different values of @xmath2 and antiferromagnetic interactions ( @xmath18 ) is given as follows@xmath225 @xmath226 @xmath227 @xmath228 in these expressions , @xmath229 represents the error function .
it is important to notice that , if we make the replacement @xmath230 , we immediately recover the expressions of ref . .
for ferromagnetic interactions ( @xmath17 ) , the expressions are completely similar to the ones above , with the substitutions @xmath231 and @xmath232 , with @xmath233 the imaginary error function . to quote an example@xmath234 the @xmath235 functions are obtained to be , for antiferromagnetic interactions@xmath236 @xmath237 , \ ] ] @xmath238 -\frac{(1+k)\left ( 3+e^{8k}-4e^{2k}\right ) } { k^{3}}.\ ] ] the corresponding expressions for the ferromagnetic case can be obtained by using the prescription indicated above .
it is important to notice that @xmath235 is not defined for @xmath239 but it easy to see that this function does not appear in any calculation , as it corresponds to non - interacting spins . | a microscopic derivation of the classical generalized constant coupling ( gcc ) model for geometrically frustrated magnets is presented . starting from the classical heisenberg hamiltonian ,
the partition function for clusters with @xmath0 spins in the presence of the inhomogeneous symmetry breaking fields ( sbf ) created by spins outside the unit is calculated .
the effective fields characterizing the interaction between units naturally arise as averages over the sbf .
these effective fields are fixed by a self - consistency condition . in the paramagnetic regime
, we recover all the results previously obtained in a more phenomenological way , which were shown to be in excellent agreement with monte carlo calculations for these lattices .
however , this microscopic formulation allows us to also study the behavior of the system in the absence of applied magnetic field .
it is found that , for antiferromagnetic interactions , the equilibrium configuration is a non - collinear configuration in which the total magnetization of the unit is zero , and the condition under which such an ordered state occurs is also obtained from the calculation .
however , frustration inhibits the formation of such an state , and the system remains paramagnetic down to 0 k , if only nearest neighbor interactions are taken into account , for all the systems considered , in agreement with the now generally accepted idea . for completeness , the ferromagnetic case is also studied . furthermore ,
as the present method is formulated in the real space , it is very easy to study the effect of further perturbations in the hamiltonian by modifying the form of the sbf .
we do this to include the effect of next nearest neighbor interactions ( nnn ) and site dilution by non - magnetic impurities for the pyrochlore lattice .
it is found that nnn interactions can stabilize a non - collinear ordered state , or ferromagnetic one , depending on the relation between nn and nnn interactions , in agreement with mean field calculations , and the phase diagram is calculated .
however , site dilution is not enough by itself to form such an ordered state . |
it has been apparent for quite some time that there exists a strong connection between subfactors , large random matrices and free probability theory .
perhaps the most clear instance of this connection is that all three theories have an underlying _ planar _ structure .
for example , the standard invariant of a subfactor ( i.e. , the system of higher relative commutants ) is in a natural way a planar algebra @xcite .
traces of polynomials in random matrices naturally count certain planar objects ( @xcite ) .
finally , the combinatorics of free probability theory is intimately tied with that of non - crossing ( i.e. , planar ) partitions @xcite .
furthermore , techniques from some of these subjects proved useful for applications to others .
for example , there are many connections between work in free probability theory and certain computations in the paper @xcite . random matrices and free probability theory were used to construct subfactors @xcite .
more recently , mingo and speicher and guionnet and maurel - segala @xcite have found combinatorial expressions , involving planar diagrams , for the that the large-@xmath2 asymptotics of moments of polynomials in certain random matrices . in this paper
we exploit for the first time the _ graded algebra _ coming from a planar algebra @xmath3 to obtain a subfactor @xmath4 whose standard invariant is @xmath3 .
the essential ingredient is a _ trace _ on the graded algebra coming from free probability / random matrices , whose use in this context was inspired by @xcite , which promises to be a source of further developments in this direction .
we take the point of view that all of the three subjects mentioned above are intimately related to the notion of a planar algebra .
specifically , the underlying idea is that a planar algebra , endowed with its graded multiplication @xmath5 and trace @xmath6 is a natural replacement for the ring of polynomials occurring in both free probability theory @xcite and the theory of random matrices with a potential @xcite . to be more precise , a _ subfactor planar algebra _
( spa ) @xmath3 will be a graded vector space @xmath7 which is an algebra over the planar operad of @xcite and satisfies certain dimension and positivity conditions outlined in @xmath8[sec : planaralgebra ] .
every extremal finite index subfactor has an spa as its _
standard invariant_. given an spa @xmath3 , we define the sequence @xmath9 , @xmath10 of complex @xmath11-algebras with @xmath12 ( @xmath13 if @xmath14 ) and multiplications @xmath15 given by tangles as in @xmath8[sec : planaralgebra ] .
on each @xmath9 we define a trace @xmath16 using the sum of all temperley - lieb tangles .
the trace @xmath6 comes directly from wick s theorem applied to large @xmath2 limit of a certain gaussian matrix model using wishart matrices , defined in @xmath8[sec : randommatrixmodel ] .
but once calculated , this trace can be defined entirely in terms of planar algebras .
a rather important special case is when the matrix models may be taken as @xmath17 independent hermitian matrices . then the algebra @xmath18 is the even degree subalgebra of @xmath19 , the non - commutative polynomials in @xmath17 self - adjoint variables @xmath20 ( with @xmath21 ) .
the trace @xmath6 is then the one discovered by voiculescu in the context of his free probability theory @xcite .
it can be realized as the vacuum expectation value on the full fock space on a real @xmath17-dimensional vector space with basis @xmath20 , by the representation of @xmath19 which sends @xmath22 to @xmath23 , @xmath24 being the left creation operator of @xmath22 ( see @xcite ) .
the higher multiplications @xmath25 are then given ( on monomials of even degree @xmath26 ) by@xmath27 our main result is , with notation as above and an spa @xmath3 , of index parameter @xmath28 , \(i ) for each @xmath29 , @xmath30 is a faithful tracial state on @xmath9 and the gns completion of @xmath9 is a ii@xmath0 factor @xmath31 as long as @xmath32 ; \(ii ) there are unital inclusions @xmath33 which extend to @xmath34and projections @xmath35 , such that @xmath36 is the tower of basic constructions for the subfactor @xmath37 ; + ( iii ) the relative commutants @xmath38 are canonically identified with the vector spaces @xmath39 and this identification is a homomorphism of planar @xmath11-algebras .
this theorem gives a new proof of the breakthrough result of popa @xcite , showing that any subfactor planar algebra @xmath3 can indeed be realized by the system of higher relative commutants of a ii@xmath0 subfactor .
the key ingredient in the proofs will be representations of the algebras @xmath9 on fock spaces . in order to define these
we will suppose that @xmath3 is given as a planar subalgebra of the full planar algebra @xmath40 of some bipartite graph @xmath41 as in @xcite .
this is always possible one may for instance take @xmath42 to be the principal graph of @xmath3 .
a basis of @xmath40 is formed by loops on @xmath42 starting and ending in @xmath43 but we will define a slightly different planar algebra structure from that of @xcite , better adapted to graded multiplication . the fock space will then be spanned ( orthogonally ) by paths of varying lengths on @xmath42 , ending in @xmath43 .
it is naturally @xmath44-graded .
note that @xmath42 may be infinite so we need to make a choice of perron - frobenius eigenvector and eigenvalue for the adjacency matrix of @xmath42 .
there will not necessarily be a markov trace on @xmath40 so we work instead with the center - valued trace .
this will restrict to a markov trace on @xmath3 .
as in the theory of graph @xmath45-algebras @xcite , each edge @xmath46 of @xmath42 defines an operator @xmath47 ( of grading @xmath48 ) on the fock space , creating an edge on a path .
a loop of edges @xmath49 in @xmath40 is then represented by the product @xmath50 where @xmath51 is a version of @xmath52 according to the parity of @xmath53 , where the factors @xmath54 are determined by the perron - frobenius eigenvector . we also make use of the fact that the fock space of loops and the resulting ii@xmath0 factor can be embedded into a type iii factor canonically associated to the graph and its perron - frobenius eigenvector using a free version of the second quantization procedure .
the material is organized as follows : to aid the reader , we list here some notation used in the paper . *
bi - partite graph ( [ sub : graphplanar ] ) : @xmath42 ; vertices : @xmath42 ; even / odd vertices : @xmath55 ; edges : @xmath56 ; positively / negatively oriented edges : @xmath57 ; edges starting at @xmath58 : @xmath59 ; edges ending at @xmath58 : @xmath60 .
all loops of length @xmath29 starting at even / odd vertex : @xmath61 ; all loops starting at a positive / negative vertex : @xmath62 .
* planar algebra ( def .
[ def : planaralg ] ) : @xmath3 ; @xmath29-box space : @xmath63 ; positive / negative part : @xmath64 ; @xmath65 .
planar algebra of a graph ( [ sub : graphplanar ] ) : @xmath40 , @xmath66 etc .
subfactor planar algebra : def .
[ subfactorplanar ] .
* graded multiplications : @xmath25 ( def .
[ def : grk ] ) , the algebra @xmath9 .
trace on @xmath9 : @xmath67 ( def . [ def : trk ] ) .
we begin with a definition of planar algebra which will be recognizably equivalent to other definitions @xcite and suited to the purposes of this paper . [ tangle ] ( planar @xmath29-tangles . )
a planar @xmath29-tangle will consist of a smoothly embedded disc @xmath68 @xmath69 in @xmath70 minus the interiors of a finite ( possibly empty ) set of disjoint smoothly embedded discs @xmath71 in the interior of @xmath68 .
each disc @xmath72 , @xmath73 , will have an even number @xmath74 of marked points on its boundary ( with @xmath75 ) .
inside @xmath68 and outside @xmath71 there is also a finite set of disjoint smoothly embedded curves called strings which are either closed curves or whose boundaries are marked points of the @xmath72 s .
each marked point is a boundary point of some string , and the strings meet the boundaries of the discs transversally , only in the marked points .
the connected components of the complement of the strings in @xmath76 are called regions .
those parts of the boundaries of the discs between adjacent marked points ( and the whole boundary if there are no marked points ) will be called intervals .
the regions of the tangle will be shaded black and white so that two regions whose boundaries intersect are shaded differently .
( such a shading is always possible , since there is an even number of marked points . )
the shading will be considered to extend to the intervals which are part of the boundary of a region .
finally , to each disc in a tangle there is a distinguished interval on its boundary ( which may be shaded black or white ) .
the set of internal discs of a tangle @xmath77 will be denoted @xmath78 .
observe that diffeomorphisms of @xmath70 act on planar tangles in the obvious way .
in particular if @xmath79 is a diffeomorphism it induces a map @xmath80 we will often have to draw pictures of tangles . to indicate the distinguished interval on the boundary of a disc
we will place a * , near to that disc , in the region whose boundary contains the distinguished interval .
an example of a @xmath81-tangle illustrating all the above ingredients is given below;@xmath82 we will often use pictures with a given number of strings to illustrate a situation where the number of strings is arbitrary . we hope this will not lead to misinterpretation .
similarly if the shading is implicit or both possible shadings are intended we will suppress the shading . given planar @xmath29 and @xmath83-tangles @xmath77 and @xmath84 respectively , we say they are composable if 1 . the outside boundary of @xmath84 is equal to the boundary of one of the inside discs of @xmath77 where equality means that the marked points are the same , the shadings of the intervals are the same and the distinguished intervals are the same . and , 2 . the union of the strings of @xmath84 and those of @xmath77 are smooth curves .
if @xmath77 and @xmath84 are composable we define the composition @xmath85 to be the union @xmath86 .
the strings of @xmath85 are the unions of the strings of @xmath77 and @xmath84 .
since the shadings of @xmath77 and @xmath84 agree on their common boundary curve , it is easy to see that @xmath86 is a planar @xmath29-tangle . this composition operation is often called " gluing " as one may think of @xmath84 as being glued inside @xmath77 .
we will now define a notion of planar algebra .
axioms can be subtracted to obtain more general objects but for convenience in this paper the term " planar algebra
" will imply all the properties . before giving the formal definition we recall the notion of the cartesian product of vector spaces over an index set @xmath87,@xmath88 .
this is the set of functions @xmath89 from @xmath87 to the union of the @xmath90 with @xmath91 .
vector space operations are point - wise .
multilinearity is defined in the obvious way , and one converts multinearity into linearity in the usual way to obtain @xmath92 , the tensor product indexed by @xmath87 . a cartesian product over the empty set will mean the scalars .
[ def : planaralg]a ( unital ) planar algebra @xmath3 will be a family of @xmath44-graded vector spaces indexed by the set @xmath93 , where @xmath63 will denote the @xmath94 graded space indexed by @xmath29 .
to each planar tangle @xmath77 there will be a multilinear map @xmath95 where @xmath96 is the vector space indexed by half the number of marked boundary points of @xmath68 and graded by @xmath97 if the distinguished interval of @xmath68 is shaded white and @xmath98 if it is shaded black .
the maps @xmath99 are subject to the following two requirements : 1 .
( isotopy invariance ) if @xmath100 is an diffeomorphism of @xmath70 then @xmath101 where the sets of internal discs of @xmath77 and @xmath102 are identified using @xmath100 .
2 . ( naturality )
@xmath103 where the right hand side of the equation is defined as follows : first observe that @xmath104 is naturally identified with @xmath105 , where @xmath106 is the disc of @xmath77 containing @xmath84 .
thus given a function @xmath89 on @xmath104 to the appropriate vector spaces , we may define a function @xmath107 on @xmath78 by @xmath108 finally the formula @xmath109 defines the right hand side .
+ the natural notation for @xmath110 is to write in the @xmath111 into @xmath112 .
this is just like the notation
@xmath113 for a function of several variables , where the @xmath114 are the @xmath115 , and the internal discs correspond to the spaces between the commas .
( we also call the internal disks `` input discs '' ) .
thus if @xmath116 and @xmath117 are in @xmath118 , @xmath119 is in @xmath120 and @xmath121 is in @xmath122 then the following picture is an element of @xmath123@xmath124 + the vector spaces @xmath125 will possess a conjugate linear involution @xmath11 , @xmath126 with the compatibility requirement : @xmath127 whenever @xmath79 is an orientation diffeomorphism . +
observe that @xmath128 become _ unital commutative @xmath11-algebras _ under the multiplication operation ( with either shading):@xmath129 \circlenode{a}{a}\\ \circlenode{b}{b } \endpsmatrix } \end{array}\ ] ] + [ subfactorplanar ] a planar algebra @xmath3 will be a planar algebra satisfying the following four conditions : 1 .
@xmath130 for all @xmath131 2 .
@xmath132 + condition ( 2 ) allows us to canonically identify @xmath128 with @xmath133 as * -algebras , @xmath48 being @xmath134 ( a 0-tangle with nothing in it@xmath135 . +
this further allows us to define a sesquilinear form on each @xmath125 by : @xmath136 \circlenode{a}{b^ * } \\ \circlenode{b}{a } \nccurve[anglea=-80,angleb=80]{a}{b } \nput[labelsep=0]{-160}{b } { * } \nput[labelsep=0]{190}{a } { * } \nccurve[anglea=-60,angleb=60]{a}{b } \nccurve[anglea=-40,angleb=40]{a}{b } \nccurve[anglea=260,angleb=100]{a}{b } \nccurve[anglea=240,angleb=120]{a}{b } \nccurve[anglea=220,angleb=140]{a}{b } \endpsmatrix}}\end{array}}\ ] ] where the outside region is shaded according to @xmath94 .
the form @xmath137 is positive definite .
4 . @xmath138 where @xmath139 and @xmath140 are the following two 0-tangles:@xmath141{inner}{inner } \nput[labelsep=0]{200}{inner } { * } \end{array } } \qquad t_2= \circlenode{outer1 } { \begin{array}{c } \circlenode{inner1}{t } \nput[labelsep=0]{200}{inner1 } { * } \ncloop[anglea=180,angleb=0,linearc=0.1,arm=0.17,loopsize=0.46]{inner1}{inner1 } \end{array } } \ ] ] the last condition is topologically natural and corresponds to extremality of the subfactor ( @xcite,@xcite ) .
this condition means that the partition function of a fully - labeled zero - tangle ( when considered without its boundary disc ) is actually well - defined for that zero - tangle on the sphere @xmath142 obtained by adding a point at @xmath143 to @xmath70 .
it is natural then to suppress the outer disc of a @xmath144-tangle in pictures .
[ delta ] once @xmath128 have been identified with the scalars there is a canonical scalar @xmath28 associated with a subfactor planar algebra with the property that the multilinear map associated to any tangle containing a closed string is equal to @xmath28 times the multilinear map of the same tangle with the closed string removed . by positivity @xmath145 and it is well known that in fact the possible values of @xmath28 form the set @xmath146 @xcite . since @xmath147 it is clear that all the spaces @xmath148 are redundant and subfactor planar algebra could be axiomatized in terms of @xmath149 .
for this reason we will use in what follows @xmath150 to denote @xmath149 ( even in the non - subfactor case ) .
[ rem : usualops]in the development of planar algebras the following structures played a major role : 1 .
multiplication : each @xmath125 is a * -algebra with the involution defined above and the multiplications : @xmath151 \circlenode{a}{b } \\
\circlenode{b}{a } \ncdiag[arm=0,anglea=60,angleb=81]{a}{outer } \ncdiag[arm=0,anglea=80,angleb=87]{a}{outer } \ncdiag[arm=0,anglea=100,angleb=93]{a}{outer } \ncdiag[arm=0,anglea=120,angleb=99]{a}{outer } \nccurve[anglea=-80,angleb=80]{a}{b } \nput[labelsep=0]{130}{b } { * } \nput[labelsep=0]{130}{a } { * } \nccurve[anglea=-60,angleb=60]{a}{b } \nccurve[anglea=260,angleb=100]{a}{b } \nccurve[anglea=240,angleb=120]{a}{b } \ncdiag[arm=0,anglea=-60,angleb=-81]{b}{outer } \ncdiag[arm=0,anglea=-80,angleb=-87]{b}{outer } \ncdiag[arm=0,anglea=-100,angleb=-93]{b}{outer } \ncdiag[arm=0,anglea=-120,angleb=-99]{b}{outer } \nput[labelsep=-0.2]{110}{outer } { * } \endpsmatrix\end{array } } } \end{array}}\ ] ] there are two choices of shadings which give in general non - isomorphic algebra structures .
( we shall refer to this multiplication sometimes as the `` usual '' multiplication on @xmath125 ) .
trace : each @xmath125 is equipped with a linear map @xmath152 which is given by @xmath153{x}{x } \ncarc[arcangle=30,ncurv=4]{x}{x } \nput[labelsep=0]{170}{x } { * } \nput[labelsep=0.05]{90}{x}{\cdots } \end{array } } \end{array}\ ] ] 3 .
the temperley - lieb tangles : each tangle consisting of an outside box all of whose @xmath154 boundary points are connected by ( non - crossing ) strings inside of the box determines an element of @xmath63 , pairing depending on the shading of the region containing @xmath11 .
the set of such tangles is denoted by @xmath155 .
the jones projections : @xmath156 is given by the temperley - lieb tangle ( having @xmath154 boundary points ) : @xmath157{\vbox to2em{\vfill\hbox to10em{\hfill } } } \ncbar[ncurv=-1,arm=-0.4,angle=90,offseta=0.2,offsetb=0.2,linearc=0.2]{box}{box } \ncbar[ncurv=-1,arm=-0.8,angle=90,offseta=0.95,offsetb=0.95,linearc=0.2]{box}{box } \ncbar[ncurv=-1,arm=-0.4,angle=90,offseta=1.5,offsetb=-1.1,linearc=0.2]{box}{box } \ncbar[ncurv=-1,arm=-0.4,angle=90,offseta=-1.5,offsetb=1.1,linearc=0.2]{box}{box } \nput[labelsep=-0.75em , offset=0.55]{90}{box}{\cdots } \nput[labelsep=-0.75em , offset=-0.55]{90}{box}{\cdots } \nput[labelsep=-0.75em , offset=0]{90}{box } { * } } \end{array}.\ ] ] [ rectangles ] ( rectangles ) .
it is sometimes very convenient to use rectangles rather than circles for the input and output discs . strictly speaking this
is not allowed since the boundaries are supposed to be smooth . but nothing will happen at the corners of the rectangles so one may simply interpret a picture of a rectangle as one with smoothed corners .
use of horizontal rectangles also makes it possible to avoid specifying the first interval which we will always suppose to be the one containing the left hand vertical part of the rectangle .
( outer disks and shading ) .
we occasionally omit the outer disk when describing a planar algebra element , especially in the case that there are no boundary points on the outer disk .
also , unless the shading is explicitly indicated in a picture , we follow the convention that the region adjacent to the boundary region marked with a @xmath11 is unshaded ( white ) . if @xmath3 is a planar algebra we define a graded algebra @xmath158 as follows . as a graded vector space @xmath159 and the graded product @xmath160
is given by the tangle below which puts the element of @xmath150 entirely to the left of the element of @xmath150:@xmath161 \circlenode{a}{\vbox to1.1em{\vfill\hbox to1.1em{\hfill$a$\hfill}\vfill } } & \circlenode{b}{\vbox to1.1em{\vfill\hbox to1.1em{\hfill$b$\hfill}\vfill } } \endpsmatrix \ncdiag[arm=0,anglea=110,angleb=140]{a}{outer } \ncdiag[arm=0,anglea=100,angleb=130]{a}{outer } \ncdiag[arm=0,anglea=90,angleb=120]{a}{outer } \ncdiag[arm=0,anglea=80,angleb=110]{a}{outer } \nput[labelsep=0]{140}{a } { * } \nput[labelsep=0]{140}{b } { * } \nput[labelsep=-0.22]{147}{outer } { * } \ncdiag[arm=0,anglea=60,angleb=30]{b}{outer } \ncdiag[arm=0,anglea=70,angleb=40]{b}{outer } \ncdiag[arm=0,anglea=80,angleb=50]{b}{outer } \ncdiag[arm=0,anglea=90,angleb=60]{b}{outer } \ncdiag[arm=0,anglea=100,angleb=70]{b}{outer } \ncdiag[arm=0,anglea=110,angleb=80]{b}{outer } \end{array } } \end{array}\ ] ] ( the shading in the picture above is determined by saying that the region adjacent to the marked interval on the outer box is unshaded ( white ) ; as before , @xmath11 s denote the marked intervals on the disks ) .
note that one could also define a dual structure changing @xmath97 to @xmath98 and changing the shading in the above figure . as a graded algebra
a subfactor planar algebra is just the free graded algebra on a certain graded vector space as we shall see .
if @xmath162 is a subfactor planar algebra let @xmath163 be the 2-sided ideal of @xmath164 spanned by all elements of degree @xmath48 or more .
@xmath163 . each graded piece of @xmath163
has an innner product as defined above . for each @xmath165 let @xmath166 be the orthogonal complement of @xmath167 in @xmath168 . with notation as above
, @xmath169 is the free graded algebra generated freely by @xmath170 .
let @xmath171 be an ordered @xmath29-tuple of integers with @xmath172 and @xmath173 .
then multiplication defines linear maps @xmath174 by induction the images of @xmath175 span @xmath176 as @xmath177 varies .
so , together with @xmath166 they span @xmath178 .
thus the theorem follows from the two assertions : + 1 ) each @xmath175 is injective .
+ 2 ) the images of the @xmath175 are orthogonal for different @xmath177 .
+ to see 1 ) , note that each @xmath175 is an isometry if we give @xmath179 the hilbert space tensor product structure . to see 2 ) , let @xmath177 and @xmath180 be two distinct partitions of @xmath181 as above .
suppose @xmath182 .
consider the following picture : + this is the inner product of an element @xmath183 in @xmath184 with an element @xmath185 in @xmath186 .
( here @xmath187 and @xmath188 .
one may evaluate the tangle inside the dashed curve to obtain an element of @xmath189 .
thus the figure is actually the inner product of @xmath190 with an element of @xmath191 , thus it is zero .
so the images of @xmath175 and @xmath192 are orthogonal unless @xmath193 .
continuing in this way we see that the images of @xmath175 and @xmath192 are orthogonal unless @xmath194 .
writing elements of @xmath164 as sums of products of elements orthogonal to @xmath191 times arbitrary elements gives , by an easy argument with generating functions , @xmath195 where @xmath196 is the generating function for @xmath197 , and @xmath198 is the generating function for @xmath199 . in general
if @xmath200 is the generating function for the dimensions of the graded vector space @xmath201 we have @xmath202 so that @xmath203 .
although the graded algebra structure is not commutative even up to a sign , the presence of the cyclic group action gives a kind of `` cyclic commutativity '' as follows where @xmath180 denotes the action of the counterclockwise rotation tangle on @xmath204 : if @xmath162 is a planar algebra then @xmath205 just draw the picture .
the multiplication in an exterior algebra can be made to satisfy exactly the same commutativity formula by making the cyclic group act by the appropriate sign in each degree .
besides @xmath206 we will need other " shifted " graded * -algebra structures on @xmath207 in order to define a subfactor and analyze its tower .
[ def : grk]given a planar algebra @xmath208 and an integer @xmath209
we make @xmath210 into an associative ( unital ) * -algebra with multiplication @xmath211 given by the following formula : @xmath212 \circlenode{a}{\vbox to2.2em{\vfill \hbox{$a$}\vfill } } & \circlenode{b}{\vbox to2.2em{\vfill\hbox{$b$}\vfill } } \ncarc[arcanglea=30,arcangleb=30,ncurv=1]{a}{b } \ncarc[arcanglea=40,arcangleb=40,ncurv=1]{a}{b } \ncarc[arcanglea=50,arcangleb=50,ncurv=1]{a}{b } \nput[labelsep=-0.02]{115}{b } { * } \ncdiag[anglea=155,angleb=170,arm=-0,linearc=0.2]{a}{outer } \ncdiag[anglea=145,angleb=160,arm=-0,linearc=0.2]{a}{outer } \ncdiag[anglea=135,angleb=150,arm=-0,linearc=0.2]{a}{outer } \nput[labelsep=-0.02]{115}{a } { * } \nput[labelsep=-0.25]{140}{outer } { * } \ncdiag[anglea=100,angleb=130,arm=-0,linearc=0.2]{a}{outer } \ncdiag[anglea=65,angleb=92,arm=0]{a}{outer } \nput[labelsep=-0.25]{112}{outer}{\cdots } \nput[labelsep=-0.05,rot=43]{155}{outer}{\overbrace{\quad\qquad}^k } \nput[labelsep=-0.45,rot=50]{136}{b}{\text{\tiny $ \scriptscriptstyle{\underbrace{\quad\!\!}_k}$ } } \ncdiag[anglea=95,angleb=65,arm=0]{b}{outer } \ncdiag[anglea=25,angleb=10,arm=0]{b}{outer } \nput[labelsep=-0.25,rot=-28]{37}{outer}{\cdots } \endpsmatrix \end{array } } \end{array}\ ] ] the involution ( denoted by @xmath213 to distinguish it from the usual involution @xmath11 on @xmath39 ) is given by@xmath214{a}{outer } \ncdiag[anglea=145,angleb=160,arm=-0,linearc=0.2]{a}{outer } \ncdiag[anglea=135,angleb=150,arm=-0,linearc=0.2]{a}{outer } \nput[labelsep=-0.25]{137}{outer } { * } \ncdiag[anglea=120,angleb=120,arm=-0,linearc=0.2]{a}{outer } \ncdiag[anglea=60,angleb=60,arm=0]{a}{outer } \nput[labelsep=-0.25]{90}{outer}{\cdots } \nput[labelsep=-0.0]{52.5}{a } { * } \nput[labelsep=-0.05,rot=63]{158}{outer}{\hbox{\tiny$\overbrace{\quad}^k$ } } \nput[labelsep=-0.05,rot=-63]{22}{outer}{\hbox{\tiny$\overbrace{\quad}^k$ } } \ncdiag[anglea=25,angleb=10,arm=-0,linearc=0.2]{a}{outer } \ncdiag[anglea=35,angleb=20,arm=-0,linearc=0.2]{a}{outer } \ncdiag[anglea=45,angleb=30,arm=-0,linearc=0.2]{a}{outer } } \ ] ] the shading in both figures above is determined by the condition that the marked boundary region @xmath11 is adjacent to an unshaded ( white ) region . here
@xmath215 means @xmath216 where @xmath217 is an orientation - reversing diffeomorphism ( cf .
[ def : planaralg](2 ) ) .
we denote this * -algebra by @xmath9 .
any planar algebra contains in a canonical way the temperley - lieb planar algebra @xmath218 .
indeed , @xmath218 is spanned by tl diagrams : a tl diagram is a diagram that has no inner disks , and all of whose strings connect points on the outer disk .
any such diagram is naturally an element of @xmath3 .
[ def : trk]let @xmath219 be the sum of all tl diagrams having @xmath220 points on the outer disk represented pictorially below ( for @xmath221):@xmath222{t}{outert } \ncdiag[anglea=-60,angleb=-60,arm=-0,linearc=0.2]{t}{outert } \ncdiag[anglea=-80,angleb=-80,arm=-0,linearc=0.2]{t}{outert } \ncdiag[anglea=-100,angleb=-100,arm=-0,linearc=0.2]{t}{outert } \ncdiag[anglea=-120,angleb=-120,arm=-0,linearc=0.2]{t}{outert } \ncdiag[anglea=-140,angleb=-140,arm=-0,linearc=0.2]{t}{outert } } \ ] ] ( the position of the @xmath11 is irrelevant , by since the set of tl diagrams is invariant under a rotation by @xmath223 ) .
the trace @xmath224 is defined for @xmath225 , @xmath226 , and is valued in the zero box space of @xmath3 : @xmath227{\vbox to 6em{\vfill } \psmatrix[rowsep=0.2,colsep=0.5 ] & \ovalnode{t}{t_n } & \\ & \rnode{x}{\psframebox[framearc=0.4]{\vbox to2em{\vfill\hbox to 7em{\hfill$x$\hfill}\vfill } } } & \ncdiag[anglea=-60,angleb=60,arm=-0,linearc=0.2]{t}{x } \ncdiag[anglea=-120,angleb=120,arm=-0,linearc=0.2]{t}{x } \nput[labelsep=0]{123}{x } { * } \ncbar[angle=90,arm=1.3,offseta=1.1,offsetb=1.1,linearc=0.2]{x}{x } \ncbar[angle=90,arm=1.1,offseta=0.6,offsetb=0.6,linearc=0.2]{x}{x } \nput[labelsep=0.1]{90}{x}{\hbox{\tiny$\cdots$ } } \nput[offset=0.84]{90}{x}{\hbox{\tiny$\cdots$ } } \nput[offset=-0.86]{90}{x}{\hbox{\tiny$\cdots$ } } \nput[labelsep=-0.55,offset=0.84]{90}{x}{\underbrace{\quad}_k } \endpsmatrix } } \ ] ] where @xmath228 ( in other words , there are @xmath29 strings surrounding @xmath219 ) .
@xmath67 is a trace on @xmath9 if endowed with the multiplication @xmath25 .
this follows from the fact that the set of all tl diagrams on @xmath229 points is invariant under rotations by @xmath230 . before proceeding further
, we consider an example .
let us assume that @xmath3 is a subfactor planar algebra , so that in particular @xmath128 are one - dimensional and @xmath67 is scalar - valued .
let @xmath231 be the following element of tl : @xmath232{\vbox to0.8em{\vfill\hbox to2em{\hfill } } } } \ncbar[arm=-0.6em , linearc=0.3em , offseta=0.5em , offsetb=0.5em , angle=90]{c}{c } \nput[labelsep=-0.2,offset=0.8em]{90}{c } { * } $ ] .
let us denote by @xmath79 the moment generating function of @xmath231 .
thus we let @xmath233 be the unique scalar defined by @xmath234 we shall presently compute @xmath233 by using planar algebra methods .
let @xmath219 be the element of the planar algebra defined as the sum of all the temperley lieb diagrams connecting the @xmath220 boundary points , [ lem : mgfcup]@xmath235 the trace of @xmath236 is given by the picture ( corresponding to @xmath221 ) : @xmath237{\quad t_n \quad } } \ncbar[offseta=-0.6,offsetb=0.3,angle=90,ncurv=1,linearc=0.1]{t}{t } \ncbar[offseta=0.15,offsetb=0.15,angle=90,ncurv=1,linearc=0.1]{t}{t } \ncbar[offseta=0.3,offsetb=-0.6,angle=90,ncurv=1,linearc=0.1]{t}{t } } \ ] ] group the tl diagrams in @xmath219 according to where the first boundary point of @xmath236 is connected .
adding all those diagrams where it is connected to its nearest neighbor we get @xmath238 .
proceeding similarly we get , for @xmath239 , contributions of the form:@xmath240 { \vbox to5em { \vfill\hbox{$ \psmatrix[colsep=0.6 ] & \rnode{f1}{}\ \
\rnode{t}{\psframebox[framearc=0.4]{\qquad t_k \qquad } } \ncbar[offseta=-0.6,offsetb=0.3,angle=90,ncurv=1,linearc=0.1]{t}{t } \ncbar[offseta=0.15,offsetb=0.15,angle=90,ncurv=1,linearc=0.1]{t}{t } \ncbar[offseta=0.3,offsetb=-0.6,angle=90,ncurv=1,linearc=0.1]{t}{t } \ \ \rnode{f2 } { } & \rnode{tn}{\psframebox[framearc=0.4]{t_{n - k-1 } } } \ncbar[offseta=-0.6,offsetb=0.3,angle=90,ncurv=1,linearc=0.1]{tn}{tn } \ncbar[offseta=0.15,offsetb=0.15,angle=90,ncurv=1,linearc=0.1]{tn}{tn } \ncbar[offseta=0.3,offsetb=-0.6,angle=90,ncurv=1,linearc=0.1]{tn}{tn } & \ncbar[angle=-90,linearc=0.2]{f1}{f2 } \ncbar[offseta=0.9,angle=90,linearc=0.2]{t}{f1 } \ncbar[offseta=-0.9,angle=90,linearc=0.2]{t}{f2 } \endpsmatrix$ } \vfill } } \ ] ] if the first term in the picture is rotated by one we may use the rotational invariance of @xmath241 to see that it is just @xmath242 .
thus we have , for each @xmath243 , @xmath244 multiplying both sides by @xmath245 and summing from @xmath246 to @xmath143 we see that @xmath247 solving the quadratic equation and checking the first term to get the right solution we obtain our answer .
the function @xmath79 in lemma [ lem : mgfcup ] is that of a free poisson random variable having @xmath248-transform @xmath249 ( see @xcite ) .
we shall give an alternative computation using free convolution later in the paper ( see lemma [ lem : noatoms ] ) .
the following lemma can be easily proved by drawing the appropriate pictures : the obvious linear embedding of @xmath39 into @xmath250 is an algebra @xmath11-homomorphism from @xmath39 endowed with its usual @xmath11-algebra structure of remark [ rem : usualops ] , to @xmath9 taken with multiplication @xmath25 and conjugation @xmath213 as in definition [ def : grk ] .
moreover , this embedding carries the trace @xmath67 to the usual trace @xmath251 on @xmath39 .
let @xmath252 be a ( locally finite ) bipartite graph with adjacency matrix @xmath253 possessing an eigenvector @xmath254 ( @xmath58 being a vertex of @xmath42 ) with @xmath255 for all @xmath58 and @xmath256 .
note that although @xmath257 may be unbounded as a function of @xmath42 , the ratios @xmath258 where @xmath58 and @xmath259 are adjacent , are bounded by the eigenvector condition .
we shall denote by @xmath56 the set of oriented edges of @xmath42 , taken with all possible orientations .
thus @xmath260 where @xmath261 consists of all edges of @xmath42 oriented so as to start at a vertex in @xmath43 and end at a vertex in @xmath262 , and @xmath263 will consists of all edges of @xmath42 oriented so as to start in @xmath262 and end in @xmath43 .
for @xmath264 we ll denote by @xmath265 the edge with the opposite orientation . in @xcite
a planar algebra was associated with the above data with the property that closed strings may be removed multiplicatively as in remark [ delta ] .
we quickly redo this planar algebra with a slightly different ( but isomorphic ) structure , emphasizing those elements that arise when @xmath42 is infinite .
with @xmath266 as above we will define the planar algebra @xmath267 where @xmath268 is the vector space of bounded functions on loops on @xmath42 of length @xmath220 starting and ending in @xmath43 for the plus sign and @xmath262 for the minus sign .
( spin state ) given a planar tangle @xmath77 , and a bipartite graph @xmath42 as above a _ spin state _
@xmath269 will be a function from the regions of @xmath77 to the vertices of @xmath42 , shaded regions being mapped to @xmath43 and unshaded ones to @xmath262 , together with a function from the strings of @xmath77 to the edges of @xmath42 such that if a string @xmath84 is part of the boundary of the regions @xmath116 and @xmath117 then @xmath270 is an edge connecting @xmath271 and @xmath272 .
note that a state @xmath269 determines a function @xmath273 in the obvious way - if we follow a disc of @xmath77 around clockwise , the intervals , beginning at the distinguished one , touch regions of @xmath77 to which @xmath269 has assigned vertices of @xmath42 and the strings connected to the marked boundary points of a disc @xmath68 have been assigned edges of @xmath42 connecting the vertices on either side .
we will call @xmath274 the loop _ induced _ on @xmath68 by @xmath269 .
( the curvature factor of a spin state . ) given a tangle and a spin state @xmath269 as above , define the curvature factor @xmath275 as follows .
first isotope the tangle so that all discs are horizontal rectangles ( with the first boundary interval on the left as in remark [ rectangles ] ) and all marked points are on the top edges of the rectangles .
arrange also for all singularities of the y coordinate on the strings to be generic ( maxima or minima ) .
near such a maximum ( resp .
minimum ) we see regions above and below , one of which is convex , labeled by adjacent ( on @xmath42 ) vertices @xmath276 and @xmath277 according to @xmath269 .
assign the number @xmath278 to this singularity . then the curvature factor is@xmath279 ( the planar algebra of a bipartite graph . )
we now define the action of a planar tangle @xmath77 on @xmath280 . we are given a function @xmath281 functions on \{loops on @xmath42 } and we have to define a function on loops appropriate to the boundary of @xmath77 , in a multilinear way .
so given a loop @xmath282 appropriate to the boundary , define @xmath283 where the sum runs over all @xmath269 which induce @xmath282 on the boundary of @xmath77 .
the main thing to note in this definition is that the sum is finite since there are only a finite number of states inducing @xmath282 on the boundary , and it defines a bounded function since all the @xmath248 are bounded and so is the factor @xmath275 .
we leave it as an exercise to show that this definition of @xmath99 is compatible with the gluing of tangles and the * -structure where the * of a loop is that loop read backwards . also that the eigenvector property of @xmath257 guarantees that contractible closed strings in tangles can be removed with a multiplicative factor of @xmath28 . also that this planar algebra structure is isomorphic to that of @xcite , the only change being in how tangles are isotoped in order to define the factor @xmath275 .
the reason for the change is that we are mostly dealing with the _ graded _
algebra for which the isotopy we use is the most natural .
each of the vector spaces @xmath268 is infinite dimensional if @xmath42 is infinite .
moreover @xmath284 are the abelian von neumann algebras @xmath285 which act on the @xmath268 .
( note that the graded product and the usual product are the same on these subalgebras ) .
the trace tangle when applied to any element of the planar algebra @xmath284 produces an element of @xmath285 .
we thus get a bilinear conditional expectation @xmath286 from @xmath284 ( taken with its usual product ) onto @xmath285 .
the inner product tangles of definition [ subfactorplanar ] thus become @xmath285-valued inner products , satisfying @xmath287 .
it will follow from a representation of the graph planar algebra on a hilbert space that the conditional expectation @xmath286 ( and thus the inner product ) is non - negative definite . in the next few sections
, we shall work out several examples , which make explicit the operations of graded multiplication on @xmath40 , and which will be useful in the rest of the paper .
all of the facts mentioned below are straightforward consequences of the definition of the graph planar algebra .
we will sometimes use the word `` loop '' to also mean the planar algebra element given by the delta function on the set of all loops supported on the given loop . as a matter of convenience ,
when inserting a loop into an internal disc of a tangle we will line up the edges of the loop with the boundary points of the disc , starting with the one first in clockwise order after @xmath11 .
this convention is useful , since given a string meeting the disc in question at a certain boundary point , any state @xmath269 which has a nonzero contribution to the sum @xmath99 of will have to assign the edge of this boundary point to that string . for an edge @xmath46 we ll write @xmath288 for its starting vertex and @xmath289 for its ending vertex . for a vertex @xmath58 we ll write @xmath59 for the set of all edges starting at @xmath58 ( i.e. , @xmath290 ) , and we ll denote by @xmath60 the set of all edges that end at @xmath58 . we ll also use the notation@xmath291^{1/2}.\ ] ]
let @xmath292 be the set of all loops of length @xmath154 starting at an even vertex , and @xmath293 be the set of all loops of length @xmath154 starting at an odd vertex .
from now on , fix an integer @xmath294 and consider the algebra @xmath295 with its graded multiplication @xmath296 .
let @xmath297 be a loop , @xmath298 where @xmath299 and @xmath300 are edges of @xmath42 .
let @xmath301 then the graded product @xmath302 is given by:@xmath303 { \vbox to4em{\vfill } \rnode{a}{\psframebox[framearc=0.4 ] { \vbox to2em{\hbox to 20em{\hfill}\vfill}}}\ \nput[offset=9em , labelsep=-0.5]{90}{a}{e_1 } \nput[offset=7.5em , labelsep=-0.4]{90}{a}{\cdots } \nput[offset=7.5em , labelsep=0.4]{90}{a}{\cdots } \nput[offset=6em , labelsep=-0.5]{90}{a}{e_t } \nput[offset=4.8em , labelsep=-0.25]{90}{a } { * } \nput[offset=4.5em , labelsep=0.5]{90}{a } { * } \nput[offset=3em , labelsep=-0.5]{90}{a}{e_{t+1 } } \nput[offset=1.5em , labelsep=-0.4]{90}{a}{\cdots } \nput[offset=1.5em , labelsep=0.4]{90}{a}{\cdots } \nput[offset=0em , labelsep=-0.5]{90}{a}{e_{k } } \nput[offset=-1.5em , labelsep=-0.5]{90}{a}{f_{k}^{o } } \nput[offset=-3em , labelsep=-0.4]{90}{a}{\cdots } \nput[offset=-3em , labelsep=0.4]{90}{a}{\cdots } \nput[offset=-4.5em , labelsep=-0.5]{90}{a}{f_{t+1}^{o } } \nput[offset=-6em , labelsep=-0.5]{90}{a}{f_{t}^{o } } \nput[offset=-7.5em , labelsep=-0.4]{90}{a}{\cdots } \nput[offset=-7.5em , labelsep=0]{90}{a}{\cdots } \nput[offset=-9em , labelsep=-0.5]{90}{a}{f_{1}^{o } } \ncdiag[arm=0,offseta=9em , offsetb=-19.5em , angle=90]{a}{o
} \ncdiag[arm=0,offseta=6em , offsetb=-16.5em , angle=90]{a}{o } \ncdiag[arm=0,offseta=3em , offsetb=-13.5em , angle=90]{a}{o } \ncdiag[arm=0,offseta=0em , offsetb=-10.5em , angle=90]{a}{o } \ncdiag[arm=0,offseta=-1.5em , offsetb=-9em , angle=90]{a}{o } \ncdiag[arm=0,offseta=-4.5em , offsetb=-6em , angle=90]{a}{o } \rnode{b}{\psframebox[framearc=0.4 ] { \vbox to2em{\hbox to 20em{\hfill}\vfill } } \nput[offset=9em , labelsep=-0.5]{90}{b}{e'_1 } \nput[offset=7.5em , labelsep=-0.4]{90}{b}{\cdots } \nput[offset=7.5em , labelsep=0]{90}{b}{\cdots } \nput[offset=6em , labelsep=-0.5]{90}{b}{e'_t } \nput[offset=4.8em , labelsep=-0.25]{90}{b } { * } \nput[offset=3em , labelsep=-0.5]{90}{b}{e'_{t+1 } } \nput[offset=1.5em , labelsep=-0.4]{90}{b}{\cdots } \nput[offset=1.5em , labelsep=0.4]{90}{b}{\cdots } \nput[offset=0em , labelsep=-0.5]{90}{b}{e'_{k } } \nput[offset=-1.5em , labelsep=-0.5]{90}{b}{{f'_{k}}^{o } } \nput[offset=-3em , labelsep=-0.4]{90}{b}{\cdots } \nput[offset=-3em , labelsep=0.4]{90}{b}{\cdots } \nput[offset=-4.5em , labelsep=-0.5]{90}{b}{{f'_{t+1}}^{o } } \nput[offset=-6em , labelsep=-0.5]{90}{b}{{f'_{t}}^{o } } \nput[offset=-7.5em , labelsep=-0.4]{90}{b}{\cdots } \nput[offset=-7.5em , labelsep=0.4]{90}{b}{\cdots } \nput[offset=-9em , labelsep=-0.5]{90}{b}{{f'_{1}}^{o } } \ncdiag[arm=0,offseta=-9em , offsetb=19.5em , angle=90]{b}{o } \ncdiag[arm=0,offseta=-6em , offsetb=16.5em , angle=90]{b}{o } \ncdiag[arm=0,offseta=-4.5em , offsetb=15em , angle=90]{b}{o } \ncdiag[arm=0,offseta=-0em , offsetb=10.5em , angle=90]{b}{o } \ncdiag[arm=0,offseta=-1.5em , offsetb=12em , angle=90]{b}{o } \ncdiag[arm=0,offseta=3.0em , offsetb=7.5em , angle=90]{b}{o } \ncbar[linearc=0.2,arm=0.5,offseta=-6em , offsetb=-6em , angle=90]{a}{b } \ncbar[linearc=0.2,arm=0.2,offseta=-9em , offsetb=-9em , angle=90]{a}{b } } } } \end{array}\ ] ] which translates into the following formula:@xmath304^{1/2}\cdot\\ & & \cdot e_{t+1}\cdots e_{k}f_{k}^{o}\cdots f_{t+1}^{o}\ e_{t+1}'\cdots e_{k'}'f_{k'}'^{o}\cdots f_{1}'^{o}\cdots e_1 \cdots e_t .\end{aligned}\ ] ] apart from the perron - frobenious factors , @xmath305 corresponds to a kind of amalgamated concatenation of paths , although the edges of the path are should be cyclically permuted .
if for a path @xmath306 ( parity according to @xmath294 ) we denote by @xmath307 the path that starts at the @xmath308-th edge of @xmath309 , then we have:@xmath310 where @xmath311 is zero if the last @xmath294 edges of @xmath309 do not form the inverse of the path formed by the first @xmath294 segments of @xmath312 , and is the concatenation of @xmath309 ( with last @xmath294 segments removed ) and @xmath312 ( with first @xmath294 segments removed ) otherwise . in particular , if @xmath313 , given two paths @xmath309 , @xmath312 in @xmath314 the graded multiplication @xmath5 is just concatenation of paths ( note that in this case @xmath315 is the identity map ) .
the ( usual ) trace @xmath251 is given by@xmath316(where again the infinite sums are locally finite ) .
let us now set @xmath313 and identify in terms of paths the element of @xmath318 corresponding to any tl picture .
suppose that we are given a box @xmath319 with @xmath154 boundary points ( arranged so that all boundary points are at the top and @xmath11 is at position @xmath144 from the top - left ) .
assume also that there are @xmath29 non - crossing curves inside @xmath319 which connect pairs of boundary points together .
let @xmath177 be the associated non - crossing pairing .
the associated element of the planar algebra is the function @xmath320 on loops , defined on a loop @xmath309 by:@xmath321 if @xmath322 where @xmath323 and @xmath324 , then one can think of @xmath320 as the following locally finite sum of delta functions:@xmath325 an example of @xmath326 associated to the pairing @xmath327 ( thus @xmath328 and @xmath313 ) is presented below:@xmath329{\vbox to3.5em{\vfill\hbox to 19em{\hfill } } } } \nput[offset=8.25em , labelsep=-0.35]{90}{x}{e_1 } \nput[offset=6.75em , labelsep=-0.35]{90}{x}{e_2 } \nput[offset=5.25em , labelsep=-0.35]{90}{x}{e_2^o } \nput[offset=3.75em , labelsep=-0.35]{90}{x}{e_1^o } \nput[offset=2.25em , labelsep=-0.35]{90}{x}{e_3 } \nput[offset=9em , labelsep=-0.25]{90}{x } { * } \nput[offset=0.75em , labelsep=-0.35]{90}{x}{e_4 } \nput[offset=-0.75em , labelsep=-0.35]{90}{x}{e_5 } \nput[offset=-2.25em , labelsep=-0.35]{90}{x}{e_5^o } \nput[offset=-3.75em , labelsep=-0.35]{90}{x}{e_4^o } \nput[offset=-5.25em , labelsep=-0.35]{90}{x}{e_6 } \nput[offset=-6.75em , labelsep=-0.35]{90}{x}{e_6^o } \nput[offset=-8.25em , labelsep=-0.35]{90}{x}{e_3^o } \ncbar[linestyle = dotted , nodesep=-0.4,arm=-0.4,linearc=0.1,angle=90,offseta=8.25em , offsetb=-3.75em]{x}{x } \ncbar[linestyle = dotted , nodesep=-0.4,arm=-0.2,linearc=0.1,angle=90,offseta=6.75em , offsetb=-5.25em]{x}{x } \ncbar[linestyle = dotted , nodesep=-0.4,arm=-0.6,linearc=0.1,angle=90,offseta=2.25em , offsetb=8.25em]{x}{x } \ncbar[linestyle = dotted , nodesep=-0.4,arm=-0.4,linearc=0.1,angle=90,offseta=0.75em , offsetb=3.75em]{x}{x } \ncbar[linestyle = dotted , nodesep=-0.4,arm=-0.2,linearc=0.1,angle=90,offseta=-0.75em , offsetb=2.25em]{x}{x } \ncbar[linestyle = dotted , nodesep=-0.4,arm=-0.2,linearc=0.1,angle=90,offseta=-5.25em , offsetb=6.75em]{x}{x } \end{array}\ ] ] ( the dotted lines are for illustration purposes only and are not part of the planar diagram ) . in this way ,
given a @xmath155 element @xmath319 we get an associated element @xmath330 .
this embedding is the canonical inclusion of the temperley - lieb planar algebra into @xmath40 . as before ,
we denote by @xmath241 the element@xmath332 obtained by summing over all @xmath155 diagrams .
let @xmath128 be the zero - box space , i.e. , as a linear space it is @xmath285 .
the algebras @xmath284 , when considered with the graded multiplication @xmath5 , are abelian , and are in the center of @xmath331 .
recall that @xmath333 is a @xmath334-bilinear map determined by @xmath335 ; one can check that @xmath336 , where @xmath58 denotes the delta function at @xmath55 . the center - valued trace @xmath337 is given by the equation @xmath338 here as before @xmath241 is the sum of all @xmath218 diagrams .
[ lemma : tr0formula]let @xmath339 and let @xmath340 be defined by @xmath341 ( i.e. , the value of @xmath342 , viewed as a function on @xmath42 ) .
let @xmath343 be a loop .
then if @xmath344 starts at @xmath58,@xmath345 where the sum is over all non - crossing pairings of @xmath154 integers and the product is taken over all tuples @xmath346 , @xmath347 which are paired by @xmath177 .
if @xmath344 does not start at @xmath58 , @xmath348 .
furthermore , @xmath349 is uniquely determined by the recursive formula@xmath350 and the formula @xmath351 we note that although the support of an element @xmath352 , viewed as a function on paths , may not be finite , the support of @xmath353 is always finite , since this element is supported on paths of a fixed length starting and ending at @xmath58 .
thus the value of @xmath354 is well - defined . moreover , to know the value of @xmath354 , it is sufficient to know its value on elements of @xmath355 that have finite support .
clearly , the recursive formula gives rise to a uniquely defined linear functional on all finitely - supported elements of @xmath355 ( these elements are , of course , viewed as functions on paths ) . by the comments above
, we shall therefore prove the lemma if we prove that both the functional @xmath349 and the functional@xmath356 satisfy this recursive relation .
let @xmath357 .
then @xmath48 is paired with some integer @xmath358 .
thus @xmath359 .
thus@xmath360 furthermore , @xmath361 is given by the claimed formula .
thus @xmath362 satisfies the recursive relation .
we now turn to showing that @xmath349 satisfies the same recursive relation .
note that @xmath348 unless @xmath344 starts at @xmath58 . note that if @xmath363 and @xmath364 then @xmath365 unless @xmath366 ( an opposite of a path is a path with the order of edges and also all edges reversed ) .
furthermore , if @xmath367 , then@xmath368 the set @xmath218 of all temperley - lieb diagrams can be written as a union@xmath369 in a manner similar to decomposing the partitions ( @xmath358 denotes the other endpoint of the string ending at @xmath48 ) .
let us assume that @xmath344 starts at @xmath58 .
let us denote by @xmath370{a}{b}$ ] the diagram in which @xmath48 is connected to @xmath358 and @xmath371 , @xmath372
. then @xmath373{a}{b}}\rangle.\end{aligned}\ ] ] now , recall that@xmath374 so that@xmath375{a}{b } } & = & \sum_{f_{1}\ldots f_{q-1}e\ldots f_{2k-1}e^{o}\in l_{k}^{+}}\sigma(e)\prod _ { \parbox{2cm}{\tiny\begin{center}$\{i_{p},j_{p}\}\subset b_{1}$ \\ { or } $ \{i_{p,}j_{p}\}\subset b_{2}$\end{center } } } \delta_{f_{i_{p}}=f_{j_{p}}^{o}}\sigma(f_{i_{p } } ) ef_{1}\cdots f_{q-1}e^{o}f_{q+1}\cdots f_{2k}\\ & = & \sum_{e}\sigma(e)ew_{b_{1}}e^{o}w_{b_{2}}.\end{aligned}\ ] ] moreover,@xmath376{a}{b}}\rangle = 0\ ] ] unless @xmath344 has the form @xmath377 with @xmath378 a loop having length @xmath379 and @xmath46 an edge . in this case,@xmath380{a}{b}}\rangle=\langle x_{1 } , w_{b_{1}}\rangle \langle x_{2 } , w_{b_{2}}\rangle \sigma(e)\sigma(e^{o})\sigma(e ) = \langle x_{1 } , w_{b_{1}}\rangle \langle x_{2 } , w_{b_{2}}\rangle \sigma(e).\ ] ]
lastly , if @xmath381 then@xmath382 it follows that @xmath349 satisfies the same recursive formula as @xmath383 and , in particular , @xmath384 .
let us denote by @xmath231 the element @xmath385 .
then@xmath386=\sum_{e}\left[\frac{\mu(t(e))}{\mu(s(e))}\right]s(e)\\ & = & \sum_{v\in\gamma_{+}}v\frac{1}{\mu(v)}\sum_{s(e)=v}\mu(t(e))=\sum_{v\in\gamma_{+}}\delta v,\end{aligned}\ ] ] since @xmath387 .
it is not the planar algebras @xmath40 that are of real interest , but some of their planar subalgebras . in particular those with finite dimensional @xmath150 and 1-dimensional @xmath128 for which the inner product is thus scalar valued and inherits positive definiteness from @xmath40 .
the following theorem , which follows from popa s work on the theory of @xmath388-lattices ( see e.g. theorem 2.9 ( 4 ) in @xcite ) , shows that any subfactor planar algebra is a sub - planar algebra of a planar algebra of a discrete bipartite graph .
let @xmath3 be an ( extremal ) subfactor planar algebra , realized as the @xmath388-lattice @xmath389 with principal graph @xmath42 and associated perron - frobenius eigenvector @xmath257 .
let @xmath390 be as in theorem 2.9(4 ) in @xcite .
then : + ( a ) the graph planar algebra @xmath40 is the planar algebra of the inclusion @xmath391 ; in other words , @xmath392 ; + ( b ) the isomorphism @xmath393 gives rise to a planar algebra inclusion of @xmath3 into @xmath40 .
the algebras @xmath394 and @xmath395 were constructed in @xcite as certain non - unital inductive limits of the algebra @xmath389 .
pictorially , this construction corresponds to e.g. taking @xmath395 to be the inductive limit of the algebras @xmath396 using the non - unital inclusion given by the following picture ( the region containing @xmath11 is unshaded):@xmath397 { \vbox
to3em { \vfill \hbox{$ \rnode{x } { \psframebox[framearc=0.4 ] { \vbox to1em { \vfill \hbox to6em { \hfill $ x$ \hfill } \vfill } } } $ } \vfill } } } \ncdiag[angle=90,arm=0,offseta=2.5em , offsetb=-2.5em]{x}{outer } \ncdiag[angle=90,arm=0,offseta=2.0em , offsetb=-2.0em]{x}{outer } \ncdiag[angle=90,arm=0,offseta=1.5em , offsetb=-1.5em]{x}{outer } \nput[offset=-0em]{90}{x}{\cdots } \ncdiag[angle=90,arm=0,offseta=-2.0em , offsetb=2.0em]{x}{outer } \ncdiag[angle=90,arm=0,offseta=-2.5em , offsetb=2.5em]{x}{outer } \nput[offset=3.1em , labelsep=0.1]{90}{x } { * } \ncdiag[angle=-90,arm=0,offseta=2.5em , offsetb=-2.5em]{x}{outer } \ncdiag[angle=-90,arm=0,offseta=-2.0em , offsetb=2.0em]{x}{outer } \ncdiag[angle=-90,arm=0,offseta=2.0em , offsetb=-2.0em]{x}{outer } \nput[offset=-0em]{-90}{x}{\cdots } \ncdiag[angle=-90,arm=0,offseta=-1.5em , offsetb=1.5em]{x}{outer } \ncdiag[angle=-90,arm=0,offseta=-2.5em , offsetb=2.5em]{x}{outer } \end{array } \mapsto \begin{array}{c } \rnode{outer } { \psframebox[framearc=0.4 ] { \vbox to3em { \vfill \hbox{$ \rnode{x } { \psframebox[framearc=0.4 ] { \vbox to1em { \vfill \hbox to6em { \hfill $ x$ \hfill } \vfill } } } $ \hskip 3em } \vfill } } } \ncdiag[angle=90,arm=0,offseta=2.5em , offsetb=-4.0em]{x}{outer } \ncdiag[angle=90,arm=0,offseta=2.0em , offsetb=-3.5em]{x}{outer } \ncdiag[angle=90,arm=0,offseta=1.5em , offsetb=-3.0em]{x}{outer } \nput[offset=-0.25em]{90}{x}{\cdots } \ncdiag[angle=90,arm=0,offseta=-2.0em , offsetb=0.5em]{x}{outer } \ncdiag[angle=90,arm=0,offseta=-2.5em , offsetb=1.0em]{x}{outer } \nput[offset=3.1em , labelsep=0.1]{90}{x } { * } \ncdiag[angle=-90,arm=0,offseta=2.5em , offsetb=-1.0em]{x}{outer } \ncdiag[angle=-90,arm=0,offseta=2.0em , offsetb=-0.5em]{x}{outer } \ncdiag[angle=-90,arm=0,offseta=-1.5em , offsetb=3.0em]{x}{outer } \nput[offset=-0.25em]{-90}{x}{\cdots } \ncdiag[angle=-90,arm=0,offseta=-2.0em , offsetb=3.5em]{x}{outer } \ncdiag[angle=-90,arm=0,offseta=-2.5em , offsetb=4.0em]{x}{outer } \ncbar[angle=90,arm=-0.3,linearc=0.1,offsetb=2.0em , offseta=-4.0em]{outer}{outer } \ncbar[angle=-90,arm=-0.3,linearc=0.1,offseta=2.0em , offsetb=-4.0em]{outer}{outer } \end{array } \in p_{k+2}\ ] ] the algebra @xmath394 then consists of all diagrams having a vertical through - string on the left ( again , region containing @xmath11 is unshaded ) : @xmath398 { \psmatrix[colsep=1em , rowsep=0.3 ] \\ & \rnode{f } { } & \rnode{x } { \psframebox[framearc=0.4]{\hbox { \vbox to1em { \vfill \hbox to6em { \hfill $ x$ \hfill } \vfill } } } } \\ \vbox{\vfill } \endpsmatrix } } \end{array } \ncdiag[arm=0,angle=90,linearc=0.2,offseta=3em , offsetb=-2em]{x}{outer } \ncdiag[arm=0,angle=90,linearc=0.2,offseta=2em , offsetb=-1em]{x}{outer } \ncdiag[arm=0,angle=90,linearc=0.2,offseta=1em , offsetb=-0em]{x}{outer } \ncdiag[arm=0,angle=90,linearc=0.2,offseta=-2em , offsetb=3em]{x}{outer } \ncdiag[arm=0,angle=90,linearc=0.2,offseta=-3em , offsetb=4em]{x}{outer } \ncdiag[arm=0,angle=-90,linearc=0.2,offseta=3em , offsetb=-4em]{x}{outer } \ncdiag[arm=0,angle=-90,linearc=0.2,offseta=2em , offsetb=-3em]{x}{outer } \ncdiag[arm=0,angle=-90,linearc=0.2,offseta=-1em , offsetb=0em]{x}{outer } \ncdiag[arm=0,angle=-90,linearc=0.2,offseta=-2em , offsetb=1em]{x}{outer } \ncdiag[arm=0,angle=-90,linearc=0.2,offseta=-3em , offsetb=2em]{x}{outer } \nput[offset=-0.5em]{90}{x}{\cdots } \nput[offset=0.5em]{-90}{x}{\cdots } \ncdiag[arm=0,angle=90,offseta=3.5em , offsetb=0.25em]{outer}{f } \ncdiag[arm=0,angle=-90,offseta=-3.5em , offsetb=-0.25em]{outer}{f } \nput[offset=4.1em , labelsep=-0.25]{90}{outer } { * } \right\}\ ] ]
given @xmath42 as above , let @xmath55 .
let @xmath399 where @xmath400 .
we shall denote by @xmath57 the set of edges of @xmath42 which are positively or negatively oriented ( according to the sign @xmath94 ) .
we shall make the convention that together with any edge @xmath401 there is also its opposite edge @xmath402 .
we endow @xmath403 with a ( semi - finite ) trace @xmath404 given on the minimal projections of @xmath403 by the formula@xmath405 let @xmath406 be integers . for each choice of @xmath407 choose integers @xmath408 with the property that for each fixed vertex @xmath58 , @xmath409 as @xmath410 . in the foregoing
, we will consider ( infinite if the graph @xmath42 is infinite ) matrices whose entries are indexed by the set @xmath411 .
such an entry will be denoted @xmath412 , where @xmath413 , @xmath414 , @xmath415 and @xmath416 .
given such a matrix @xmath417 , we compute its trace as follows:@xmath418 our matrices will be such that @xmath419 unless @xmath420 belong to a finite set , so that the sum above is finite . for @xmath339 consider the diagonal matrix @xmath421 given by@xmath422 note that the joint law of @xmath423 converges as @xmath410 to the joint law of @xmath424 .
consider then for a positively oriented edge @xmath425 from @xmath58 to @xmath426 the @xmath427 matrix @xmath428 defined as follows .
the entry @xmath429 is zero unless @xmath430 and @xmath431 . otherwise ,
@xmath432 is ( up to scaling ) a random gaussian matrix ; in other words , the entries form a family of independent complex gaussian random variables , each of variance @xmath433 .
we shall moreover choose the matrices @xmath428 in such a way that the entries of matrices corresponding to different positively oriented edges are independent .
thus the variables @xmath434 are assumed to be independent . for a negatively oriented edge @xmath89 , set @xmath435 . for a loop @xmath436@xmath437 , set @xmath438 .
note that @xmath439 is a homomorphism from the algebra @xmath440 to the algebra of random matrices .
let @xmath56 denote the expected value of a random variable .
then the matrices @xmath428 satisfy : ( a ) @xmath441 ; ( b ) @xmath442 is independent of @xmath2 and converges to @xmath443 as @xmath410 ; ( c ) for any @xmath444 , @xmath436 @xmath445 ( here @xmath446 means the value of the function @xmath447 at @xmath339 ) .
\(a ) and ( b ) are both straightforward ; note that@xmath448 to see ( c ) , we first note that if @xmath449 then @xmath450 as @xmath451 and then @xmath410 . on the other hand , @xmath452 .
denote by @xmath286 the map the conditional expectation onto the algebra @xmath403
. then we have that if @xmath381 , @xmath453 is a multiple of @xmath421 .
since @xmath286 is @xmath404-preserving , we have@xmath454 in particular , we see that@xmath455 it is known ( see e.g. @xcite ) that the variables @xmath456 converge in distribution ( jointly also with elements of @xmath403 ) to a family of @xmath403-valued semicircular variables with variance@xmath457 hence if @xmath437 , then for any @xmath458 , @xmath459 by lemma [ lemma : tr0formula ] , we see that@xmath460 as claimed . since the trace @xmath404 is positive and faithful on @xmath403 we conclude that the center - valued trace @xmath6 is non - negative : @xmath461 if @xmath462 .
recall that a bi - partite graph can be used as a bratteli diagram to describe an inclusion of two algebras .
let @xmath463 be an inclusion of multi - matrix algebras corresponding to the graph @xmath42 .
this means @xmath464 and @xmath465 .
in particular , each @xmath466 corresponds to a central projection @xmath467 in @xmath319 ( the unit of the @xmath58-th direct summand ) , and each @xmath468 corresponds to a central projection @xmath469 .
the inclusion @xmath470 is such that @xmath471 if there is no edge between @xmath58 and @xmath426 .
if there are @xmath472 edges between @xmath58 and @xmath426 , then @xmath473 is included into @xmath474 with index @xmath472 . in particular
, this means that @xmath475 and also that we can choose @xmath472 orthogonal projections @xmath476 in @xmath477 with the property that @xmath478 and the inclusion of @xmath479 into @xmath477 is given by @xmath480 .
choose also isometries @xmath481 so that @xmath482 . let @xmath251 be the semi - finite trace on @xmath483 determined by the requirement that @xmath484 , @xmath485 .
let @xmath486 be a semicircular element , free from @xmath483 ( this only makes sense in the case that @xmath487 ; more precisely , we shall consider a large projection @xmath488 in the center of @xmath483 and consider an element @xmath486 free from @xmath489 with respect to @xmath490 ; our computations will not depend on @xmath488 once it is large enough ) . to a positive edge @xmath46 ,
we associate : ( i ) a central projection @xmath491 ; ( ii ) a projection @xmath492@xmath493 .
let @xmath494 if @xmath425 and @xmath495 if @xmath496 .
note that @xmath497 unless @xmath498 .
we can think of @xmath499 as a limit of a `` @xmath500 '' random block matrix , since its left and right support projections , @xmath467 and @xmath501 , have traces @xmath502 and @xmath503 .
in fact , one can model @xmath486 by a suitable gue random matrix in the limit when its size goes to infinity , in which case the variables @xmath499 are indeed approximated in law by random blocks as their sizes go to infinity .
furthermore , if @xmath425,@xmath504 thus also @xmath505 .
it follows that if we denote by @xmath286 the conditional expectation onto the center of @xmath483 , then , keeping in mind that @xmath506,@xmath507 thus the variables @xmath508 have the same joint law as the variables @xmath509 that we constructed in the previous section .
let @xmath42 be a bi - partite graph , as before . consider the real vector space @xmath510 with basis given by the ( oriented ) edges @xmath56 of the graph
; we denote , as before , by @xmath261 the set of positively oriented edges .
then @xmath510 is equipped with a natural conjugation which takes an edge to its opposite , @xmath511 and can thus be endowed with a complex structure : @xmath512 for any positively - oriented @xmath46 .
the inner product on @xmath510 is determined by requiring that @xmath513 unless @xmath514 and @xmath515^{1/2}.\ ] ] ( note that @xmath511 is _ not _ isometric ) .
as before , we shall use the notation@xmath291^{1/2}=\vert e\vert^{-2}.\ ] ] let @xmath403 denote the abelian algebra @xmath516 , where as before @xmath42 denotes the set of vertices of @xmath42 .
then @xmath510 is naturally an @xmath517-bimodule : given @xmath46 an edge in @xmath56 , define@xmath518 moreover , @xmath510 has a natural @xmath403-valued inner product:@xmath519 we now consider the fock space @xcite@xmath521(here @xmath522 denotes the relative bimodule tensor product ) . for @xmath264
we consider the operator@xmath523 its adjoint is given by@xmath524 note that the norm of this operator is given by@xmath525 let also@xmath526 note that @xmath527 .
let @xmath528 be the algebra of bounded adjointable operators on @xmath529 and let @xmath530 be the natural conditional expectation given by@xmath531 each vertex @xmath339 determines a state on @xmath528 given by@xmath532 where @xmath533 is the point evaluation at @xmath58 .
then@xmath534 is a weight on @xmath528 .
note that @xmath217 is finite on all finite words in @xmath535 , and therefore defines a semifinite weight on the von neumann algebra @xmath536 ( in the gns representation associated to @xmath217 ) .
[ lem : noatoms](i ) the weight @xmath217 and the conditional expectation @xmath56 are faithful on this algebra .
\(ii ) the modular group of @xmath217 is determined by @xmath537^{it}c(e)=\sigma(e)^{2it}c(e)$ ] .
\(iii ) consider @xmath492@xmath538 , where @xmath59 denotes the set of all edges that start at @xmath58 .
then for each @xmath58 , the law of @xmath231 with respect to @xmath349 has no atoms and is the free poisson law with @xmath248-transform @xmath249 .
in particular , @xmath539 is bounded for all @xmath58 and thus the ( possibly infinite ) direct sum defining @xmath231 yields a bounded operator . the gns vector space @xmath540 associated to @xmath349 can be identified with the subspace of the fock space @xmath541 spanned by tensors of the form @xmath542 , @xmath543 for which @xmath544 form a path ( i.e.
, are `` composable '' : @xmath545 ) .
if we denote by @xmath546 the operator @xmath547 and by @xmath548 the operator @xmath549 , then we have@xmath550 let @xmath3 be the set of all paths in @xmath42 and @xmath551 be the set of all paths starting at @xmath58 .
for a path @xmath552 , let @xmath553 and similarly for @xmath554 we then see that the joint laws associated to the vacuum expectation state of the variables@xmath555 have the same law .
indeed , @xmath556 if @xmath557 .
it follows that the von neumann algebra generated by @xmath558 in the gns representation @xmath559 associated to @xmath349 can be embedded into the von neumann algebra @xmath560 in such a way that the restriction of the state @xmath561 to the former algebra is exactly @xmath349 .
but it is known @xcite that @xmath562 is faithful , and so @xmath349 is faithful ( on the image in the gns construction @xmath559 ) .
furthermore , the modular group of @xmath562 is given by@xmath563^{it}\hat{c}(e).\ ] ] it follows that@xmath564^{it}\pi_{v}(c(e)).\ ] ] it is clear that the gns vector space for @xmath217 is just the direct sum of the gns vector spaces for @xmath349 taken over all vertices @xmath58 .
thus @xmath217 is faithful and so @xmath56 is faithful ( on the possibly larger algebra @xmath565 .
thus ( i ) holds .
let now@xmath566 then @xmath486 has the same law for @xmath562 as does @xmath231 for @xmath349 .
note that @xmath567 , with @xmath568 .
thus @xmath569 are free and so the law of @xmath486 satisfies@xmath570 now , for each @xmath46 , @xmath571 has free poisson distribution with @xmath248-transform @xmath572 ( see ( * ? ? ?
* remark 4.4 on p. 347 ) ) .
thus the law of @xmath569 has only one atom of mass @xmath573 at zero ( this expression is to be interpreted as zero if it is negative ) .
it follows from additivity of @xmath248-transform @xcite that the law of @xmath486 is free poisson with @xmath248-transform@xmath574 which will have an atom iff @xmath575 . since @xmath576
, the law of @xmath486 has no atoms . thus ( iii ) holds . finally , it is also clear that ( ii ) holds since a similar formula holds in the gns representation of each @xmath349 and @xmath577 .
let @xmath578 be the set of all loops in @xmath42 .
then the algebra @xmath579 belongs to the fixed point of the modular group acting on the algebra @xmath580 .
since @xmath426 is a loop , the factors @xmath581 associated to each factor in @xmath582 cancel .
let @xmath583 , @xmath584 ( the set of all loops starting at a positive vertex and of length @xmath154 ) .
let @xmath585 be a loop given by @xmath586
. then @xmath587 . in particular ,
@xmath588 . @xmath589 .
moreover , @xmath590 . the rest follows from lemma [ lemma : tr0formula ] .
[ lemma : embedfreegroup]let @xmath591 be the set of loops starting at an even vertex . consider @xmath592 with its semi - finite weight @xmath217 .
then each even @xmath339 , defines a central projection in @xmath593 and@xmath594 for each @xmath58 , the algebra @xmath595 can be canonically embedded into a free group factor .
if @xmath585 is a loop starting at @xmath58 , then @xmath596 .
we have seen before that @xmath597 with its state @xmath349 can be embedded into a free araki - woods factor associated to @xmath510 and taken with its free quasi - free state , in a state - preserving way .
the image of @xmath593 under @xmath559 is precisely @xmath595 , and this image clearly lies in the centralizer of the free quasi - free state .
the free quasi - free state is periodic ( the modular group , restricted to @xmath598 has as its eigenvectors the edges of @xmath42 ) and therefore the centralizer is a free group factor .
note that @xmath599 is faithful .
let @xmath600 be a loop on @xmath42 , starting at an even vertex .
then the map @xmath601 extends to a trace - preserving embedding with dense range of @xmath602 into @xmath603 .
thus @xmath6 is a faithful center - valued trace .
clearly the theorem is true on elements of @xmath604 that have finite support , i.e. , are finite linear combinations of loops .
we have to check that this embedding makes sense for elements of @xmath604 which , as functions on loops , have infinite support .
let @xmath605 .
then for any @xmath466 , @xmath606 has finite support .
moreover , by assumption @xmath607 has finite @xmath608 norm .
but the value of @xmath609 at @xmath58 is exactly @xmath610 and is therefore uniformly bounded as a function of @xmath58 .
moreover , note that each @xmath611 belongs to the span of words of length @xmath154 in operators @xmath612 .
the eigenvector condition implies that the ratios @xmath613 for @xmath420 adjacent are bounded , and also that the valence of the graph is bounded .
it follows that the linear dimension of the space of all loops of length @xmath29 starting at a vertex @xmath58 is uniformly bounded , by a constant independent of @xmath58 . moreover , the norms of the orthogonal basis for this space ( consisting of the various loops ) are bounded both above and below uniformly in @xmath58
. thus the restrictions of the operator norm and the @xmath614-norm to the finite - dimensional linear span of loops of length @xmath29 starting at @xmath58 are equivalent , and the constants in the equivalence can be chosen to be uniform in @xmath58 .
it follows that @xmath615 is uniformly bounded in norm ( independent of @xmath58 ) . since the projections @xmath616
are orthogonal , it follows that @xmath582 , defined as the ultraweakly - convergent sum @xmath617 , is a bounded operator in @xmath593 . since the map @xmath601 is bilinear over @xmath334 , and is an algebra homomorphism when restricted to finite linear combinations of loops , it is easy to see that it is an algebra homomorphism on all of @xmath618
. let @xmath59 denote the set of all edges starting at an even vertex @xmath58 and let @xmath261 denote the set of all positively oriented edges ( i.e. , ones that start at an even vertex ) . recall that @xmath619 if we let @xmath28 be the perron - frobenius eigenvalue , then@xmath620v+\sum_{e\in e_{+}}\sigma(e)\ell(e)\ell(e)^{*}\\ & = & 2\sum_{e\in\gamma_{+}(v)}\sigma(e)\re(\ell(e)\ell(e^{o}))+\delta+\sum_{e\in e_{+}}\sigma(e)\ell(e)\ell(e)^{*}.\end{aligned}\ ] ] here we used@xmath621 so that@xmath622 let @xmath623 be the set of all vectors in _
@xmath529 _ starting and ending in a positive vertex and let @xmath624 . since if @xmath625@xmath626 ( because of the normalizations of the lengths of @xmath627 we have that the sum @xmath628 is the same as the sum @xmath629 where the summation is over an orthonormal basis ) .
thus@xmath630 where @xmath631 is the projection onto @xmath632 and @xmath28 is the perron - frobenius eigenvalue . as consequence
, we note that we can now identify the position of @xmath633 inside of _ @xmath634 _ identified with a subspace of @xmath635 : [ lem : xiandl2a]let @xmath58 be an even vertex .
then @xmath636 is the closed linear span of the orthogonal system of vectors@xmath637 moreover,@xmath638 where @xmath28 is the perron - frobenius eigenvalue .
we note that @xmath639 . moreover , the linear span of @xmath640 is clearly stable under the action of @xmath486 .
thus it is sufficient to prove that if @xmath641 for @xmath642 are in @xmath643 then also @xmath644 .
but this follows from noting that @xmath645 , where @xmath646 is a tensor of smaller degree in @xmath643 .
furthermore,@xmath647\vert e\otimes e^{o}\vert_{2}^{2}=\sum_{e\in\gamma_{+}(v)}\frac{\mu(t(e))}{\mu(v)}=\frac{1}{\mu(v)}\sum_{j}\gamma_{vj}\mu_{j}=\delta.\ ] ] recall that @xmath648 .
let as before @xmath649 , @xmath650 and @xmath651 .
[ lem : relcommcup](i ) @xmath652 is a singular masa in @xmath595 .
+ ( ii ) @xmath653 .
+ ( iii ) consider the algebra @xmath654 .
then @xmath655 .
we first note that any @xmath444 commutes with @xmath649 . in particular , @xmath656 .
hence @xmath657=0 $ ] implies that @xmath658w=[y , vxw]=0 $ ] for all @xmath659 .
hence @xmath660 is the closure of @xmath661 .
consider the full fock space @xmath662 as in the proof of lemma [ lem : noatoms ] , where @xmath510 is as before a hilbert space having as basis edges of @xmath42 .
thus @xmath662 is spanned by all tensors of the form @xmath663 , where @xmath664 .
let @xmath665 , and consider @xmath666 given by @xmath667 .
let @xmath668 acting on @xmath662 , and consider the subalgebra @xmath669 given by @xmath670[\hat{\ell}(f)+\hat{\ell}(f^{o})^{*}]:e , f\in e)$ ] .
then clearly @xmath671 can be identified with @xmath672 .
furthermore , @xmath488 is invariant under the modular group associated to @xmath349 ( the vector state associated to the vacuum vector in @xmath666 ) .
thus the modular group of @xmath488 is the restriction to @xmath488 of the modular group of @xmath3 .
fix @xmath659 . denote by @xmath673 the element @xmath674 .
denote by @xmath675 the element @xmath676 ( here @xmath677 denotes the right creation operator and @xmath678 is the modular conjugation ) .
note that @xmath679 . we now make the identification @xmath680 obtained by sending a tensor @xmath681 associated to a _ path _ @xmath682 starting at @xmath58 and ending at @xmath426 to the tensor @xmath683 .
it is not hard to see that@xmath684 it follows that the laws of @xmath675 and @xmath673 ( with respect to the vacuum state on @xmath685 ) are the same as that of @xmath486 and have no atoms ; thus @xmath686 and @xmath687 are diffuse .
in particular , if @xmath688 satisfies @xmath689 , then @xmath690 satisfies @xmath691 .
consequently , we would prove ( iii ) if we could show : \(a ) if @xmath692 , @xmath693 for @xmath694 only occurs if @xmath695 and \(b ) if @xmath696 and @xmath697 for some @xmath698 then @xmath699 .
let @xmath700 .
assume first that @xmath701 .
let @xmath702 . put @xmath703 .
then @xmath704 in such a way that the left and right multiplication by @xmath486 on @xmath705 correspond to the actions of @xmath673 and @xmath706 . in particular
, @xmath707 is invariant under both @xmath673 and @xmath706 .
the image of @xmath708 under @xmath709 lies in the closure of the direct sum@xmath710 ( this direct sum is identified with a subspace @xmath711 by identifying @xmath712 and @xmath713 with @xmath646 if @xmath714 ) . each direct summand in this sum is invariant under both @xmath706 and @xmath673 since their actions respect the tensor product decompositions @xmath715 : @xmath706 acts as @xmath716 and @xmath673 acts as @xmath717 .
now , for any choice of orthonormal basis for @xmath718 , @xmath719 , we have for all @xmath720:@xmath721 and consequently @xmath722 is isomorphic to an ( infinite ) multiple of @xmath723 as a bimodule over @xmath686 acting on the left copy of @xmath707 and @xmath724 acting on the right copy of @xmath707 .
since the spectral measure of @xmath673 is non - atomic , it follows that there can be no vector @xmath725 contained in @xmath722 satisfying @xmath726 , since such a vector would give rise ( via an isomorphism of @xmath723 with hilbert - schmidt operators on this space ) to a hilbert - schmidt operator on @xmath707 , commuting with @xmath673 .
thus the only possible @xmath725 satisfying @xmath726 and lying in the image of @xmath727 must be contained in @xmath704 .
thus we have proved ( b ) . to prove ( a ) , we note that if @xmath692 , and we let @xmath728 , the image of @xmath729 lies in@xmath730 ( once again identified with a subspace of @xmath711 as before ) , which is isomorphic to an infinite multiple of @xmath731 as a bimodule over @xmath686 acting on the left copy of @xmath707 and @xmath732 acting on the right copy of @xmath733 .
once again , we see that there can be no vector @xmath725 satisfying @xmath734 in this space .
thus ( a ) is also proved .
thus we have proved ( iii ) .
note that we have actually proved that @xmath735 when viewed as a bimodule over @xmath736 is the direct sum of @xmath636 and an ( infinite ) multiple of the coarse @xmath737-bimodule @xmath738 .
recall ( see e.g. @xcite or @xcite ) that the normalizer of @xmath652 is contained in its quasi - normalizer @xmath739 , which consists of those elements @xmath646 in @xmath595 for which the associated bimodule @xmath740 is `` discrete '' .
this bimodule can not be discrete if it contains a sub - bimodule isomorphic to a compression of the coarse @xmath737-bimodule .
thus the only @xmath741 must lie in @xmath636 and thus in @xmath652 .
it follows that the normalizer of @xmath652 in @xmath595 is contained in @xmath652 .
thus @xmath652 is a singular masa and so ( i ) follows .
now ( ii ) easily follows from ( i ) .
we now consider the following sum @xmath744^{1/2}c(e)c(f)c(f)^{*}c(e)^{*}\ ] ] taken over all loops that start at an even vertex .
the pictorial representation of this planar algebra element is : @xmath745{\vbox to0.8em{\vfill\hbox to3.2em{\hfill } } } } \ncbar[arm=-0.6em , linearc=0.3em , offseta=0.5em , offsetb=0.5em , angle=90]{c}{c } \ncbar[arm=-0.9em , linearc=0.3em , offseta=0.9em , offsetb=0.9em , angle=90]{c}{c } \nput[labelsep=-0.20]{167}{c } { * } .\ ] ] [ lem : factorwhennotstar]let @xmath58 be a fixed even vertex .
assume that there is a path of length @xmath746 from @xmath58 to @xmath58 not of the form @xmath747 .
then algebra @xmath748 has a trivial relative commutant inside of the algebra @xmath749 , where @xmath654 . because of lemma [ lem : relcommcup ] , we know that the relative commutant of @xmath748 inside of @xmath749 is contained in @xmath750 .
let @xmath751 where @xmath752 and @xmath753 is a path from @xmath58 to @xmath58 .
set@xmath754 where the sum is over all paths @xmath755 from @xmath58 to @xmath58 .
then if @xmath756 , and @xmath757 is as in lemma [ lem : xiandl2a ] , we have : @xmath758\rangle & = & \langle\eta\otimes\xi^{\otimes k},\sum_{e , f}\sigma(e)\sigma(f)c(e)c(f)c(f^{o})c(e^{o})\xi^{\otimes l}-\sigma(e)\sigma(f)\xi^{\otimes l}c(e)c(f)c(f^{o})c(e^{o})\rangle\\ & = & \langle\eta\otimes\xi^{\otimes k},\sum_{e , f}\sigma(e)\sigma(f)e\otimes f\otimes f^{o}\otimes e^{o}\otimes\xi^{\otimes l}\\ & & + \sum_{e , f}\sigma(e)\sigma(f)\frac{\mu(t(e))^{1/2}}{\mu(v)^{1/2}}e\otimes f\otimes f^{o}\otimes\xi^{\otimes(l-1)}\rangle.\end{aligned}\ ] ] thus@xmath759\rangle=\begin{cases } \frac{\mu(t(f_{1}))^{1/2}}{\mu(v)^{1/2 } } , & l = k\\ \frac{\mu(t(f_{1}))^{1/2}}{\mu(v)^{1/2}}\cdot\frac{\mu(t(e_{1}))^{1/2}}{\mu(v)^{1/2 } } , & l = k+1\\ 0 , & \textrm{otherwise}.\end{cases}\ ] ] thus if we consider@xmath760 and assume that @xmath761=0 $ ] and @xmath762 ( so that @xmath763 ) , we obtain:@xmath764\rangle\\ & = & \frac{\mu(t(f_{1}))^{1/2}}{\mu(v)^{1/2}}\left(\alpha_{k}+\frac{\mu(t(e_{1}))^{1/2}}{\mu(v)^{1/2}}\alpha_{k+1}\right),\qquad k\geq1.\end{aligned}\ ] ] since the choice of @xmath765 was arbitrary , we find that @xmath766 if @xmath767 , not all @xmath768 are zero ; from this recursive relation we deduce that @xmath503 are all equal to the same value , @xmath769 , independent of @xmath770 and that ( after rescaling @xmath309 by a non - zero constant ) we may assume that @xmath771 where @xmath772 .
on the other hand,@xmath773 so that@xmath774 thus if @xmath775 is the valence of @xmath42 at @xmath58 , we find that @xmath776 , so that @xmath777 . using the fact that @xmath778 , we compute:@xmath779 which is impossible .
thus @xmath780=0 $ ] forces @xmath781 .
[ lem : whenstar]let @xmath42 be a connected bi - partite graph with @xmath782 vertices , @xmath339 even and assume that the hypothesis of lemma [ lem : factorwhennotstar ] is not satisfied .
then the remaining vertices @xmath783 of @xmath42 are all connected to @xmath58 by a single edge , and @xmath42 has no other edges .
we can now prove that the relative commutant of @xmath743 can be controlled , if the graph @xmath42 is not too small .
the cases we exclude are @xmath784 ( a graph with a single vertex and no edges ) and @xmath785 ( a graph with exactly two vertices connected by a single edge ) . in these cases ,
@xmath231 and @xmath742 commute ( in fact , they are equal ) . in either of these cases ,
the perron - frobenius eigenvalue is @xmath48 , which is of little interest to us .
[ thm : relcommutanttrivial]assume that @xmath786 and @xmath787 and let @xmath339 even .
then ( i ) the relative commutant @xmath788 is trivial . in particular , the center of @xmath593 is the algebra @xmath789 .
( ii ) @xmath790 ( where @xmath791 , and @xmath792 ) .
because of lemma [ lem : factorwhennotstar ] and lemma [ lem : whenstar ] , it remains to consider the case in which @xmath42 is a graph with @xmath793 vertices @xmath794 with a single edge between @xmath58 and each @xmath795 and no other edges . since @xmath796 $ ] , we find that @xmath797 and therefore @xmath798 .
moreover , one can normalize the perron - frobenius eigenvector to be @xmath799 for all @xmath800 .
thus@xmath801 let @xmath802 . then @xmath803 = \sum_{i}c(e_{i})c(e_{i}^{o})c(e_{i})c(e_{i}^{o})\xi^{\otimes k}-\xi^{\otimes k}c(e_{i})c(e_{i}^{o})c(e_{i})c(e_{i}^{o})\\ = \xi^{\otimes k-2}+4\xi^{\otimes k-1}+6\xi^{\otimes k}+3\xi^{\otimes k+1}+\sum_{i}e_{i}\otimes e_{i}^{o}\otimes e_{i}\otimes e_{i}^{o}\otimes\xi^{\otimes k-1}+\sum_{i}e_{i}\otimes e_{i}^{o}\otimes e_{i}\otimes e_{i}^{o}\otimes\xi^{\otimes k}\\ -\xi^{\otimes k-2}-4\xi^{\otimes k-1}-6\xi^{\otimes k}-3\xi^{\otimes k+1}-\sum_{i}\xi^{\otimes k-1}\otimes e_{i}\otimes e_{i}^{o}\otimes e_{i}\otimes e_{i}^{o}+\sum_{i}\xi^{\otimes k}e_{i}\otimes e_{i}^{o}\otimes e_{i}\otimes e_{i}^{o}\\ = \sum_{i}e_{i}\otimes e_{i}^{o}\otimes e_{i}\otimes e_{i}^{o}\otimes(\xi^{\otimes k}+\xi^{\otimes k-1})-(\xi^{\otimes k}+\xi^{\otimes k-1})\otimes\sum_{i}e_{i}\otimes e_{i}^{o}\otimes e_{i}\otimes e_{i}^{o}\\ = \zeta\otimes(\xi^{\otimes k}+\xi^{\otimes k-1})-(\xi^{\otimes k}+\xi^{\otimes k-1})\otimes\zeta,\end{gathered}\ ] ] where we have set @xmath804 .
let @xmath805 .
then @xmath806 for all @xmath807 . on the other hand , @xmath808 .
it follows that for any @xmath802 , @xmath809\rangle & = \langle\zeta\otimes(\xi^{\otimes k}+\xi^{\otimes k-1})-(\xi^{\otimes k}+\xi^{\otimes k-1})\otimes\zeta,\eta\otimes\xi^{\otimes l}\rangle\\ = & -\langle\xi^{\otimes k}\otimes\zeta,\eta\otimes\xi^{\otimes l}\rangle-\langle\xi^{\otimes k-1}\otimes\zeta,\eta\otimes\xi^{\otimes l}\rangle\\ = & -\delta_{l = k}\vert\xi^{\otimes(l-1)}\vert-\delta_{l = k-1}\vert\xi^{\otimes(l-1)}\vert.\end{aligned}\ ] ] it follows that if @xmath810 , and we assume that @xmath780=0 $ ] , then we get for all @xmath811:@xmath812,\eta\otimes\xi^{\otimes l}\rangle & = & \sum_{k}\alpha_{k}\langle[z,\xi^{\otimes k}],\eta\otimes\xi^{\otimes l}\rangle\\ & = & -\alpha_{l}\vert\xi^{\otimes(l-1)}\vert-\alpha_{l+1}\vert\xi^{\otimes(l-1)}\vert.\end{aligned}\ ] ] it follows that @xmath768 , @xmath813 , is a constant sequence .
but the sequence @xmath814 is in @xmath815 and thus must be zero .
it follows that @xmath816 commutes with @xmath134 , then @xmath817 .
but in this case , @xmath761\omega=\alpha_{1}[\cup , z]\omega$ ] .
the only tensors of degree @xmath818 in @xmath819 are @xmath820 , and the only terms of this degree in @xmath821 are @xmath822 , which are not equal .
thus @xmath761=0 $ ] implies that also @xmath823 and so @xmath309 must be a scalar . to see ( ii ) , note first that any @xmath466 is a projection in the relative commutant of @xmath824 . since the projections corresponding to different vertices are orthogonal , it follows that any element @xmath344 in the relative commutant is weakly - convergent infinite sum @xmath825 , where @xmath826 .
let @xmath3 be an ( extremal ) subfactor planar algebra embedded into the planar algebra of some graph @xmath42 .
thus @xmath828 can be viewed as a subalgebra of @xmath829 .
moreover , @xmath830 , and so @xmath231 and @xmath742 both belong to @xmath18 .
therefore , the center of @xmath831 is contained in the relative commutant of @xmath743 inside of _ @xmath593_. by theorem [ thm : relcommutanttrivial ] , this relative commutant is the intersection of the algebra @xmath832 identified with the zero box space in @xmath40 .
assume that the zero - box space of @xmath3 is one - dimensional
. then @xmath833 .
note that @xmath834 is the restriction to @xmath835 of the conditional expectation @xmath56 from @xmath593 onto @xmath836 ( which is the center of @xmath593 ) . since this conditional expectation is normal , if @xmath837 , then @xmath838 . on the other hand
, @xmath839 is the limit ( in the weak - operator topology ) of some sequence @xmath840 . for each @xmath53
, @xmath841 belongs to the zero - box space of @xmath842 .
since @xmath843 and the zero - box space is one - dimensional @xmath844 must be a multiple of @xmath845 and hence @xmath846 .
thus if the zero box space of @xmath3 is one - dimensional , and since @xmath847 is diffuse , we automatically get : let @xmath3 be a planar algebra with one - dimensional zero box space and of index @xmath32 .
then @xmath848 is a type ii@xmath0 factor . since @xmath849 , we see by lemma [ lemma : embedfreegroup ] that @xmath827 can be actually embedded into a direct sum of free group factors
. in particular , @xmath827 has the haagerup property and is @xmath850-embeddable .
we first recall from section [ sec : planaralgebra ] that if we identify elements of @xmath40 with paths , then the multiplication @xmath855 on @xmath856 can be expressed as follows .
let @xmath586 and @xmath857 be two paths .
denote by @xmath858 the path obtained from @xmath426 by following the path @xmath426 , but starting at the first point of @xmath426 ( rather than its starting point ) .
then@xmath859 ( note that the factor @xmath860 is exactly the norm @xmath861 ) .
the vector space @xmath851 is an algebra with multiplication @xmath855 .
thus @xmath869 is a @xmath11-homomorphism from @xmath870 onto @xmath851 .
the unit of @xmath851 is the element@xmath871 ( here @xmath263 is the set of all odd edges , i.e. , ones _ ending _ at an even vertex ) .
we now define a tracial weight @xmath852 on @xmath851:@xmath874 ( the first @xmath28 is the perron - frobenius eigenvalue ; note that @xmath514 forces @xmath875 ) .
in other words,@xmath876 the last observation shows that @xmath877 is a non - negative functional .
let us consider @xmath890 , @xmath891 with @xmath892 in the domain of @xmath852
. then @xmath893 .
moreover,@xmath894 and @xmath895 .
thus if @xmath896 , each of the positive terms in the sum above must be zero and so @xmath897 for all @xmath898 .
it follows that @xmath899 .
define now the map @xmath900 by@xmath901 note that@xmath902 and moreover@xmath903 so that @xmath904 is an @xmath905-linear projection .
moreover , we see that @xmath906 so that @xmath907 is the trace - preserving conditional expectation from @xmath908 .
it follows that @xmath909 extends also to the von neumann algebra generated by @xmath851 .
the algebras @xmath910 with semi - finite traces @xmath912 are defined in a similar way .
the algebra @xmath910 is the linear span@xmath913 where the parity of the loops is chosen to match the parity of @xmath181 . for a loop @xmath914 ,
we set@xmath915 then map @xmath916 defines a @xmath11-homomorphism from @xmath917 to @xmath918 .
( recall that @xmath919 denotes the loop obtained by replacing the starting point in @xmath426 by the @xmath29-th point on the path @xmath426 ) .
let@xmath920 the inclusion @xmath921 is given by@xmath922 one can check that @xmath53 is again a trace - preserving inclusion .
the conditional expectation @xmath923 is given by@xmath924 as before , set@xmath925 it is not hard to check that this is the unique trace - preserving @xmath926 linear conditional expectation from @xmath910 to @xmath927 and that the trace @xmath911 is faithful ( the argument is exactly the same as in the case @xmath246 ) .
moreover , one can easily check that @xmath928 if we identify @xmath910 with @xmath929 .
let @xmath946 be the set of all paths in @xmath42 of length @xmath181 ending at an even vertex ( and starting at an even or odd vertex , according to the parity of @xmath181 ) .
for @xmath947 , let @xmath948 .
then for any @xmath949,@xmath950 moreover,@xmath951 where @xmath952 are some constants .
since the sum above is finite , it follows that equations and continue to hold whenever @xmath953 , i.e. after passing to weak limits .
let us assume now that @xmath959 .
then by choosing @xmath960 as above , we see from lemma [ lem : relcomcontainspn ] that @xmath961 .
using equations and , we conclude that @xmath486 is a finite linear combination of terms of the form@xmath962 and that each such term must belong to the relative commutant @xmath963 .
we can thus assume that @xmath964 with @xmath965
. then @xmath966=\left(\frac{\mu(t(e_{n}))}{\mu(s(e_{1}))}\right)^{1/2}\ell(e_{1})\cdots\ell(e_{n})[x,\cup]\ell(f_{n})^{*}\cdots\ell(f_{1})^{*}\ ] ] and similarly@xmath967=\left(\frac{\mu(t(e_{n}))}{\mu(s(e_{1}))}\right)^{1/2}\ell(e_{1})\cdots\ell(e_{n})[x,\cup]\ell(f_{n})^{*}\cdots\ell(f_{1})^{*}.\ ] ] thus if @xmath486 is in the relative commutant of @xmath968 , then @xmath22 must be in the relative commutant of @xmath969 in @xmath970 , which we know to be @xmath836 ( theorem [ thm : relcommutanttrivial ] ) .
it follows that@xmath971 since the reverse inclusion holds by lemma [ lem : relcomcontainspn ] , equality holds . in particular , @xmath972 .
we now see from the definitions in section [ sec : planaralgebra ] that the latter algebra is exactly the planar algebra @xmath973 taken with its usual multiplication .
thus it follows that@xmath974 we claim that the latter intersection is exactly @xmath149 . to see this ,
write any @xmath953 as@xmath975 with @xmath976 , as before .
let @xmath977 , where @xmath978 is the ( normal ) conditional expectation given by .
then @xmath979 is a weakly - continuous map , and moreover @xmath980 if @xmath981 . thus if @xmath982 , then @xmath486 is the limit ( in the weak operator topology ) of a sequence @xmath983 .
but then @xmath984 . since the zero - box space of @xmath3 is one - dimensional , it follows that @xmath985 ( since then @xmath986 and so @xmath987 .
thus @xmath988 and the theorem is proved .
[ thm : hrcommutsasalgarepks]let @xmath934 be a subfactor planar subalgebra of index @xmath989 .
let @xmath990 .
then @xmath991 as algebras ( here @xmath992 is taken with ordinary multiplication ) in a way that preserves jones projections . since @xmath993 , lemma [ lem : relcommcupcup ] shows that @xmath994 .
thus it is enough to prove that @xmath995 .
but this is immediate , since @xmath39 commutes with @xmath996 and thus with @xmath827 .
the correspondence takes jones projections to jones projections ( as is immediate from the pictures ) .
we first check that @xmath999 . indeed , @xmath1000{\vbox
to2em{\vfill\hbox to10em{\hfill } } } \ncbar[ncurv=-1,arm=-0.4,angle=90,offseta=0.2,offsetb=0.2,linearc=0.2]{box}{box } \ncbar[ncurv=-1,arm=-0.8,angle=90,offseta=0.95,offsetb=0.95,linearc=0.2]{box}{box } \ncbar[ncurv=-1,arm=-0.4,angle=90,offseta=1.5,offsetb=-1.1,linearc=0.2]{box}{box } \ncbar[ncurv=-1,arm=-0.4,angle=90,offseta=-1.5,offsetb=1.1,linearc=0.2]{box}{box } \nput[labelsep=-0.75em , offset=0.55]{90}{box}{\cdots } \nput[labelsep=-0.75em , offset=-0.55]{90}{box}{\cdots } \nput[labelsep=-0.75em , offset=0]{90}{box } { * } } \end{array}}^{k\textrm { strings total}}\end{aligned}\ ] ] and since @xmath1001 , it has the form @xmath1002{\vbox to3em{\vskip 0.3em \hbox to10em{\hfill$\circlenode{a}{a}$\hfill}\vfill } } } \ncbar[linearc=0.2 , arm=-1.2 , angle=90 , offseta=1.5 , offsetb=1.5]{x}{x } \ncbar[linearc=0.2 , arm=-1.0 , angle=90 , offseta=1.1 , offsetb=1.1]{x}{x } \ncdiag[arma=0,armb=-0.1,linearc=0.1,anglea=140,angleb=90,offsetb=-0.9]{a}{x } \ncdiag[arma=0,armb=-0.1,linearc=0.1,anglea=40,angleb=90,offsetb=0.9]{a}{x } \nput[labelsep=0.05]{88}{a } { * } \nput[labelsep=0.17]{55}{a}{\cdots } \nput[labelsep=0.17]{125}{a}{\cdots } \nput[labelsep=0.37]{115}{a}{\overbrace{\ \qquad\ } ^{k-2 } } \ncdiag[arma=0,armb=-0.1,anglea=80,angleb=90,offsetb=0.2]{a}{x } \ncdiag[arma=0,armb=-0.1,anglea=100,angleb=90,offsetb=-0.1]{a}{x } \end{array}.\ ] ] we now compute : @xmath1003{\vbox to6em{\vfill } \rnode{box}{\psframebox[framearc=0.4,linestyle = dashed , linewidth=0.02]{\vbox to3em{\vfill\hbox to10em{\hfill } } } \ncbar[ncurv=-1,arm=-0.4,angle=90,offseta=0.2,offsetb=0.2,linearc=0.2]{box}{box } \ncbar[ncurv=-1,arm=-0.8,angle=90,offseta=0.95,offsetb=0.95,linearc=0.2]{box}{box } \ncbar[ncurv=-1,arm=-0.4,angle=90,offseta=1.5,offsetb=-1.1,linearc=0.2]{box}{box } \ncbar[ncurv=-1,arm=-0.4,angle=90,offseta=-1.5,offsetb=1.1,linearc=0.2]{box}{box } \nput[labelsep=-0.75em , offset=0.55]{90}{box}{\cdots } \nput[labelsep=-0.75em , offset=-0.55]{90}{box}{\cdots } } \qquad \rnode{x}{\psframebox[framearc=0.4,linestyle = dashed , linewidth=0.02]{\vbox to3em{\vskip 0.3em \hbox to10em{\hfill$\circlenode{a}{a}$\hfill}\vfill } } } \ncbar[linearc=0.2 , arm=-1.2 , angle=90 , offseta=1.5 , offsetb=1.5]{x}{x } \ncbar[linearc=0.2 , arm=-1.0 , angle=90 , offseta=1.1 , offsetb=1.1]{x}{x } \ncdiag[arma=0,armb=-0.1,linearc=0.1,anglea=140,angleb=90,offsetb=-0.9]{a}{x } \ncdiag[arma=0,armb=-0.1,linearc=0.1,anglea=40,angleb=90,offsetb=0.9]{a}{x } \nput[labelsep=0.17]{55}{a}{\cdots } \nput[labelsep=0.17]{125}{a}{\cdots } \ncdiag[arma=0,armb=-0.1,linearc=0.1,anglea=80,angleb=90,offsetb=0.2]{a}{x } \ncdiag[arma=0,armb=-0.1,linearc=0.1,anglea=100,angleb=90,offsetb=-0.1]{a}{x } \ncbar[linearc=0.2,arm=0.2,angle=90,offseta=-1.5,offsetb=-1.5]{box}{x } \ncbar[linearc=0.2,arm=0.4,angle=90,offseta=-1.1,offsetb=-1.1]{box}{x } \ncbar[linearc=0.2,arm=0.6,angle=90,offseta=-0.95,offsetb=-0.9]{box}{x } \ncbar[linearc=0.2,arm=0.8,angle=90,offseta=-0.2,offsetb=-0.1]{box}{x } } } \ncdiag[arm=0,angle=90,offseta=1.5,offsetb=-3.8]{box}{outer } \ncdiag[arm=0,angle=90,offseta=1.1,offsetb=-3.4]{box}{outer } \ncdiag[arm=0,angle=90,offseta=0.95,offsetb=-3.25]{box}{outer } \ncdiag[arm=0,angle=90,offseta=0.2,offsetb=-2.5]{box}{outer } \ncdiag[arm=0,angle=90,offseta=-1.1,offsetb=3.4]{x}{outer } \ncdiag[arm=0,angle=90,offseta=-0.2,offsetb=2.45]{x}{outer } \ncdiag[arm=0,angle=90,offseta=-0.9,offsetb=3.15]{x}{outer } \ncdiag[arm=0,angle=90,offseta=-1.5,offsetb=3.75]{x}{outer } \nput[labelsep=-0.25,offset=2.25]{90}{outer } { * } \end{array } \\ & = & \begin{array}{c } \rnode{x}{\psframebox[framearc=0.4]{\vbox to3em{\vskip 0.3em \hbox to10em{\hfill$\circlenode{a}{a}$\hfill}\vfill } } } \ncbar[linearc=0.2 , arm=-0.5 , angle=90 , offseta=1.5 , offsetb=-1.1]{x}{x } \ncbar[linearc=0.2 , arm=-0.5 , angle=90 , offseta=-1.1 , offsetb=1.5]{x}{x } \ncdiag[arma=0,armb=-0.1,linearc=0.1,anglea=140,angleb=90,offsetb=-0.9]{a}{x } \ncdiag[arma=0,armb=-0.1,linearc=0.1,anglea=40,angleb=90,offsetb=0.9]{a}{x } \nput[labelsep=0.05]{88}{a } { * } \nput[labelsep=0.17]{55}{a}{\cdots } \nput[labelsep=0.17]{125}{a}{\cdots } \ncdiag[arma=0,armb=-0.1,linearc=0.1,anglea=80,angleb=90,offsetb=0.2]{a}{x } \ncdiag[arma=0,armb=-0.1,linearc=0.1,anglea=100,angleb=90,offsetb=-0.1]{a}{x } \end{array } \\ & = & \delta x\wedge_k \mathbf{e}_k \qquad \textrm{(by symmetry)}.\end{aligned}\ ] ] ( here dashed lines indicate removed boxes ) .
next , we check that @xmath1004 for @xmath1005{\vbox to3em{\vskip 0.3em \hbox to10em{\hfill$\circlenode{a}{a}$\hfill}\vfill } } } \ncbar[linearc=0.2 , arm=-1.2 , angle=90 , offseta=1.5 , offsetb=1.5]{x}{x } \ncdiag[arma=0,armb=-0.1,linearc=0.1,anglea=140,angleb=90,offsetb=-0.9]{a}{x } \ncdiag[arma=0,armb=-0.1,linearc=0.1,anglea=40,angleb=90,offsetb=0.9]{a}{x } \nput[labelsep=0.05]{88}{a } { * } \nput[labelsep=0.17]{55}{a}{\cdots } \nput[labelsep=0.17]{125}{a}{\cdots } \ncdiag[arma=0,armb=-0.1,linearc=0.1,anglea=80,angleb=90,offsetb=0.2]{a}{x } \ncdiag[arma=0,armb=-0.1,linearc=0.1,anglea=100,angleb=90,offsetb=-0.1]{a}{x } \nput[labelsep=0.37]{115}{a}{\overbrace{\ \qquad\ } ^{k-1 } } \end{array } \in m_{k-1}.$ ] note that it follows from the formula for @xmath1006 ( or from an explicit computation using the trace ) that @xmath1007{\vbox to3em{\vskip 0.3em \hbox to8em{\hfill$\circlenode{a}{a}$\hfill}\vfill } } } \ncdiag[arma=0,armb=-0.1,linearc=0.1,anglea=140,angleb=90,offsetb=-0.9]{a}{x } \ncdiag[arma=0,armb=-0.1,linearc=0.1,anglea=40,angleb=90,offsetb=0.9]{a}{x } \nput[labelsep=0.05]{88}{a } { * } \nput[labelsep=0.17]{55}{a}{\cdots } \nput[labelsep=0.17]{125}{a}{\cdots } \ncdiag[arma=0,armb=-0.1,linearc=0.1,anglea=80,angleb=90,offsetb=0.2]{a}{x } \ncdiag[arma=0,armb=-0.1,linearc=0.1,anglea=100,angleb=90,offsetb=-0.1]{a}{x } \nput[labelsep=0.37]{115}{a}{\overbrace{\ \qquad\ } ^{k-2 } } \ncloop[linearc=0.1,anglea=180,angleb=0,loopsize=0.6]{a}{a } \end{array}\ ] ] now , @xmath1008{\vbox to6em{\vfill } \rnode{box}{\psframebox[framearc=0.4,linestyle = dashed , linewidth=0.02]{\vbox to3em{\vfill\hbox to10em{\hfill } } } } \ncbar[ncurv=-1,arm=-0.4,angle=90,offseta=0.2,offsetb=0.2,linearc=0.2]{box}{box } \ncbar[ncurv=-1,arm=-0.8,angle=90,offseta=0.95,offsetb=0.95,linearc=0.2]{box}{box } \ncbar[ncurv=-1,arm=-0.4,angle=90,offseta=1.5,offsetb=-1.1,linearc=0.2]{box}{box } \ncbar[ncurv=-1,arm=-0.4,angle=90,offseta=-1.5,offsetb=1.1,linearc=0.2]{box}{box } \nput[labelsep=-0.75em , offset=0.55]{90}{box}{\cdots } \nput[labelsep=-0.75em , offset=-0.55]{90}{box}{\cdots } \quad \rnode{x}{\psframebox[framearc=0.4,linestyle = dashed , linewidth=0.02]{\vbox
to3em{\vskip 0.3em \hbox to10em{\hfill$\circlenode{a}{a}$\hfill}\vfill } } } \ncbar[linearc=0.2 , arm=-1.2 , angle=90 , offseta=1.5 , offsetb=1.5]{x}{x } \ncdiag[arm=0,anglea=140,angleb=90,offsetb=-0.9]{a}{x } \ncdiag[arm=0,anglea=40,angleb=90,offsetb=0.9]{a}{x } \nput[labelsep=0.17]{55}{a}{\cdots } \nput[labelsep=0.17]{125}{a}{\cdots } \ncdiag[arm=0,anglea=60,angleb=90,offsetb=0.2]{a}{x } \ncdiag[arm=0,anglea=120,angleb=90,offsetb=-0.2]{a}{x } \ncdiag[arm=0,anglea=160,angleb=90,offsetb=-1.3]{a}{x } \ncdiag[arm=0,anglea=20,angleb=90,offsetb=1.3]{a}{x } \quad \rnode{box1}{\psframebox[framearc=0.4,linestyle = dashed , linewidth=0.02]{\vbox to3em{\vfill\hbox to10em{\hfill } } } } \ncbar[ncurv=-1,arm=-0.4,angle=90,offseta=0.2,offsetb=0.2,linearc=0.2]{box1}{box1 } \ncbar[ncurv=-1,arm=-0.8,angle=90,offseta=0.95,offsetb=0.95,linearc=0.2]{box1}{box1 } \ncbar[ncurv=-1,arm=-0.4,angle=90,offseta=1.5,offsetb=-1.1,linearc=0.2]{box1}{box1 } \ncbar[ncurv=-1,arm=-0.4,angle=90,offseta=-1.5,offsetb=1.1,linearc=0.2]{box1}{box1 } \nput[labelsep=-0.75em , offset=0.55]{90}{box1}{\cdots } \nput[labelsep=-0.75em , offset=-0.55]{90}{box1}{\cdots } } } \ncbar[linearc=0.2,arm=0.2,angle=90,offseta=-1.5,offsetb=-1.5]{box}{x } \ncbar[linearc=0.2,arm=0.4,angle=90,offseta=-1.1,offsetb=-1.3]{box}{x } \ncbar[linearc=0.2,arm=0.6,angle=90,offseta=-0.95,offsetb=-0.9]{box}{x } \ncbar[linearc=0.2,arm=0.8,angle=90,offseta=-0.2,offsetb=-0.2]{box}{x } \ncbar[linearc=0.2,arm=0.2,angle=90,offseta=-1.5,offsetb=-1.5]{x}{box1 } \ncbar[linearc=0.2,arm=0.4,angle=90,offseta=-1.3,offsetb=-1.1]{x}{box1 } \ncbar[linearc=0.2,arm=0.6,angle=90,offseta=-0.9,offsetb=-0.95]{x}{box1 } \ncbar[linearc=0.2,arm=0.8,angle=90,offseta=-0.2,offsetb=-0.2]{x}{box1 } \ncdiag[arm=0,angle=90,offseta=1.5,offsetb=-5.67]{box}{outer } \ncdiag[arm=0,angle=90,offseta=1.1,offsetb=-5.27]{box}{outer } \ncdiag[arm=0,angle=90,offseta=0.95,offsetb=-5.12]{box}{outer } \ncdiag[arm=0,angle=90,offseta=0.2,offsetb=-4.37]{box}{outer } \nput[labelsep=-0.25,offset=4.1]{90}{outer } { * } \ncdiag[arm=0,angle=90,offseta=-1.5,offsetb=5.67]{box1}{outer } \ncdiag[arm=0,angle=90,offseta=-1.1,offsetb=5.27]{box1}{outer } \ncdiag[arm=0,angle=90,offseta=-0.95,offsetb=5.12]{box1}{outer } \ncdiag[arm=0,angle=90,offseta=-0.2,offsetb=4.37]{box1}{outer } \ncdiag[arma=1.13,nodesepa=-0.02,linearc=0.1,armb=-0.15,anglea=90,offseta=0.075,angleb=90,offsetb=-0.5]{a}{outer } \ncdiag[arma=1.13,nodesepa=-0.02,linearc=0.1,armb=-0.15,anglea=90,offseta=-0.075,angleb=90,offsetb=0.5]{a}{outer } \nput[labelsep=-0.2]{90}{outer}{\cdots } \end{array}\\ & = & \delta^{-2 } \begin{array}{c } \rnode{x}{\psframebox[framearc=0.4]{\vbox to3em{\vskip 0.3em \hbox to10em{\hfill$\circlenode{a}{a}$\hfill}\vfill } } } \ncbar[linearc=0.2 , arm=-0.5 , angle=90 , offseta=1.5 , offsetb=-1.1]{x}{x } \ncbar[linearc=0.2 , arm=-0.5 , angle=90 , offseta=-1.1 , offsetb=1.5]{x}{x } \ncdiag[arma=0,armb=-0.1,linearc=0.1,anglea=140,angleb=90,offsetb=-0.9]{a}{x } \ncdiag[arma=0,armb=-0.1,linearc=0.1,anglea=40,angleb=90,offsetb=0.9]{a}{x } \ncloop[linearc=0.1,anglea=180,angleb=0,loopsize=0.6]{a}{a } \nput[labelsep=0.05]{88}{a } { * } \nput[labelsep=0.17]{55}{a}{\cdots } \nput[labelsep=0.17]{125}{a}{\cdots } \ncdiag[arma=0,armb=-0.1,linearc=0.1,anglea=80,angleb=90,offsetb=0.2]{a}{x } \ncdiag[arma=0,armb=-0.1,linearc=0.1,anglea=100,angleb=90,offsetb=-0.1]{a}{x } \end{array } \quad=\quad \delta^{-1 } \begin{array}{c } \rnode{x}{\psframebox[framearc=0.4]{\vbox to3em{\vskip 0.3em \hbox to10em{\hfill$\ovalnode{a}{e_{m_{k-2 } } ( a)}$\hfill}\vfill } } } \ncbar[linearc=0.2 , arm=-0.4 , angle=90 , offseta=1.5 , offsetb=-1.1]{x}{x } \ncbar[linearc=0.2 , arm=-0.4 , angle=90 , offseta=-1.1 , offsetb=1.5]{x}{x } \ncdiag[arma=0,armb=-0.1,linearc=0.1,anglea=150,angleb=90,offsetb=-0.9]{a}{x } \ncdiag[arma=0,armb=-0.1,linearc=0.1,anglea=30,angleb=90,offsetb=0.9]{a}{x } \nput[labelsep=0.05]{88}{a } { * } \nput[labelsep=0.1]{55}{a}{\cdots } \nput[labelsep=0.1]{125}{a}{\cdots } \ncdiag[arma=0,armb=-0.1,linearc=0.1,anglea=80,angleb=90,offsetb=0.2]{a}{x } \ncdiag[arma=0,armb=-0.1,linearc=0.1,anglea=100,angleb=90,offsetb=-0.1]{a}{x } \end{array } \\ & = & \mathcal{e}_{k-2 } ( x)\wedge_{k}\mathbf{e}_{k}.\end{aligned}\ ] ] finally , we check that the trace @xmath67 is @xmath388-markov
. let @xmath1009 .
then : @xmath1010 { \vbox to4.3em{\vfill\hbox{$ \begin{array}{cc}\rnode{box}{\psframebox[framearc=0.4]{\vbox to2em{\vfill\hbox to10em{\hfill}\vfill } \ncbar[ncurv=-1,arm=-0.4,angle=90,offseta=0.2,offsetb=0.2,linearc=0.2]{box}{box } \ncbar[ncurv=-1,arm=-0.8,angle=90,offseta=0.95,offsetb=0.95,linearc=0.2]{box}{box } \ncbar[ncurv=-1,arm=-0.4,angle=90,offseta=1.5,offsetb=-1.1,linearc=0.2]{box}{box } \ncbar[ncurv=-1,arm=-0.4,angle=90,offseta=-1.5,offsetb=1.1,linearc=0.2]{box}{box } \nput[labelsep=-0.75em , offset=0.55]{90}{box}{\cdots } \nput[labelsep=-0.75em , offset=-0.55]{90}{box}{\cdots } } } & \rnode{f1}{}\quad\vbox to2em{\hbox{$ \rnode{x}{\psframebox[framearc=0.4]{\vbox to 1em{\vfill \hbox to6em{\hfill $ x$\hfill } \vfill } } } \quad\rnode{f2}{}\quad$}\vfill } \ncbar[angle=90,offseta=-1.5,offsetb=0,arm=0.2,linearc=0.1]{box}{f1 } \ncbar[angle=90,offseta=-0.95,offsetb=-0.8,arm=0.6,linearc=0.1]{box}{x } \ncbar[angle=90,offseta=-1.1,offsetb=-1.0,arm=0.4,linearc=0.1]{box}{x } \ncbar[angle=90,offseta=-0.2,offsetb=-0.2,arm=0.8,linearc=0.1]{box}{x } \nput[offset=0.5]{90}{x}{\cdots } \ncbar[angle=-90,arm=0.13,linearc=0.2]{f1}{f2 } \ncdiag[arm=0,angle=90,offseta=-0.2,offsetb=2.125]{x}{outer } \ncdiag[arm=0,angle=90,offseta=-1,offsetb=2.925]{x}{outer } \ncdiag[arm=0,angle=90,offseta=0,offsetb=3.45]{f2}{outer } \nput[offset=-0.6]{90}{x}{\cdots } \ncdiag[arm=0,angle=90,offseta=1.5,offsetb=-3.4]{box}{outer } \ncdiag[arm=0,angle=90,offseta=1.1,offsetb=-3.0]{box}{outer } \ncdiag[arm=0,angle=90,offseta=0.95,offsetb=-2.85]{box}{outer } \ncdiag[arm=0,angle=90,offseta=0.2,offsetb=-2.1]{box}{outer } \nput[labelsep=-0.25,offset=1.9]{90}{outer } { * } \end{array}$}}}}\end{array}\right\ } \\[1em ] & = & tr_k \left\ { \begin{array}{c } \rnode{outer}{\psframebox[framearc=0.4 ] { \psmatrix[colsep=0.4,rowsep=0.3 ] \\ & \qquad \rnode{f } { } \quad \rnode{x}{\psframebox[framearc=0.4 ] { \vbox to 1em{\vfill \hbox to6em{\hfill $ x$\hfill } \vfill } } } \quad \rnode{f2 } { } \quad \\ \vbox{\vfill } \endpsmatrix } } \end{array } \ncbar[angle=90,linearc=0.1,arm=0.4,offsetb=-0.8]{f}{x } \ncbar[angle=-90,linearc=0.1,arm=0.4]{f}{f2 } \ncdiag[angle=90,arm=0,offsetb=1.92]{f2}{outer } \ncdiag[angle=90,arm=0,offseta=0.385,offsetb=0]{x}{outer } \ncdiag[angle=90,arm=0,offseta=-0.2,offsetb=0.585]{x}{outer } \nput[offset=-0.42,labelsep=-0.25]{90}{outer } { * } \nput[offset=0.1,labelsep=0.1]{90}{x}{\cdots } \ncdiag[angle=90,arm=0,offseta=-0.8,offsetb=1.185]{x}{outer } \nput[offset=-0.5,labelsep=0.1]{90}{x}{\cdots } \ncbar[angle=90,linearc=0.1,arm=-0.4,offseta=2.02,offsetb=-1.52]{outer}{outer } \right\ } \\[1em ] & = & \begin{array}{c } \psmatrix[rowsep=0.4,colsep=0.2 ] \\ \rnode{f1 } { } \quad & \rnode{f2 } { } & \rnode{f3 } { } & \rnode{t } { \psframebox[framearc=0.4 ] { \vbox to1em { \vfill \hbox to4em{\hfill$t_n$\hfill } \vfill } } } \\ & & & \rnode{x } { \psframebox[framearc=0.4 ] { \vbox to1em { \vfill \hbox to10em{\hfill$x$\hfill } \vfill } } } & \rnode{f4 } { } \ncdiag[anglea=90,angleb=-90,arm=0,offseta=0.6,offsetb=0.6]{x}{t } \ncdiag[anglea=90,angleb=-90,arm=0,offseta=0.3,offsetb=0.3]{x}{t } \nput{90}{x}{\cdots } \nput[offset=0.7,labelsep=0.1]{90}{x } { * } \ncdiag[anglea=90,angleb=-90,arm=0,offseta=-0.3,offsetb=-0.3]{x}{t } \ncdiag[anglea=90,angleb=-90,arm=0,offseta=-0.6,offsetb=-0.6]{x}{t } \ncbar[angle=-90,linearc=0.1,arm=0.3]{f1}{f2 } \ncbar[angle=90,arm=1.2,linearc=0.2,offseta=1.0,offsetb=1.0]{x}{x } \ncbar[angle=90,arm=1.4,linearc=0.2,offseta=1.4,offsetb=1.4]{x}{x } \ncbar[angle=90,arm=0.4,linearc=0.2,offseta=0,offsetb=-1.6]{f3}{x } \ncbar[angle=-90,linearc=0.4]{f3}{f4 } \ncbar[angle=90,linearc=0.4,arm=1.2]{f4}{f1 } \ncbar[angle=90,linearc=0.4,arm=1.0,offsetb=1.6]{f2}{x } \nput[offset=1.2]{90}{x}{\hbox{\tiny$\cdots$ } } \nput[offset=-1.2]{90}{x}{\hbox{\tiny$\cdots$ } } \endpsmatrix \end{array } \\[1em ] & = & \begin{array}{c } \psmatrix[rowsep=0.4,colsep=0.2 ] \\ & \rnode{t } { \psframebox[framearc=0.4 ] { \vbox to1em { \vfill \hbox to4em{\hfill$t_n$\hfill } \vfill } } } \\ \rnode{f3 } { } & \rnode{x } { \psframebox[framearc=0.4 ] { \vbox to1em { \vfill \hbox to10em{\hfill$x$\hfill } \vfill } } } & \rnode{f4 } { } \ncdiag[anglea=90,angleb=-90,arm=0,offseta=0.6,offsetb=0.6]{x}{t } \ncdiag[anglea=90,angleb=-90,arm=0,offseta=0.3,offsetb=0.3]{x}{t } \nput{90}{x}{\cdots } \nput[offset=0.7,labelsep=0.1]{90}{x } { * } \ncdiag[anglea=90,angleb=-90,arm=0,offseta=-0.3,offsetb=-0.3]{x}{t } \ncdiag[anglea=90,angleb=-90,arm=0,offseta=-0.6,offsetb=-0.6]{x}{t } \ncbar[angle=90,arm=1.2,linearc=0.2,offseta=1.0,offsetb=1.0]{x}{x } \ncbar[angle=90,arm=1.4,linearc=0.2,offseta=1.4,offsetb=1.4]{x}{x } \ncbar[angle=-90,linearc=0.2]{f3}{f4 } \ncbar[angle=90,linearc=0.2,arm=0.4,offsetb=1.6]{f4}{x } \ncbar[angle=90,arm=0.4,linearc=0.2,offseta=0,offsetb=-1.6]{f3}{x } \nput[offset=1.2]{90}{x}{\hbox{\tiny$\cdots$ } } \nput[offset=-1.2]{90}{x}{\hbox{\tiny$\cdots$ } } \endpsmatrix \end{array } = \begin{array}{c } \psmatrix[rowsep=0.4,colsep=0.2 ] \\ \rnode{t } { \psframebox[framearc=0.4 ] { \vbox to1em { \vfill \hbox to4em{\hfill$t_n$\hfill } \vfill } } } \\
\rnode{x } { \psframebox[framearc=0.4 ] { \vbox to1em { \vfill \hbox to10em{\hfill$x$\hfill } \vfill } } } \ncdiag[anglea=90,angleb=-90,arm=0,offseta=0.6,offsetb=0.6]{x}{t } \ncdiag[anglea=90,angleb=-90,arm=0,offseta=0.3,offsetb=0.3]{x}{t } \nput{90}{x}{\cdots } \nput[offset=0.7,labelsep=0.1]{90}{x } { * } \ncdiag[anglea=90,angleb=-90,arm=0,offseta=-0.3,offsetb=-0.3]{x}{t } \ncdiag[anglea=90,angleb=-90,arm=0,offseta=-0.6,offsetb=-0.6]{x}{t } \ncbar[angle=90,arm=1.2,linearc=0.2,offseta=1.0,offsetb=1.0]{x}{x } \ncbar[angle=90,arm=1.4,linearc=0.2,offseta=1.4,offsetb=1.4]{x}{x } \ncbar[angle=90,arm=1.6,linearc=0.2,offseta=1.6,offsetb=1.6]{x}{x } \nput[offset=1.2]{90}{x}{\hbox{\tiny$\cdots$ } } \nput[offset=-1.2]{90}{x}{\hbox{\tiny$\cdots$ } } \endpsmatrix \end{array } \\[1em ] & = & tr_{k-1}(x).\end{aligned}\ ] ] we first note that because of the markov property and the jones relations between the projections @xmath1012 , the algebras @xmath1013 , @xmath1014 are exactly the algebras appearing in the basic construction for @xmath37 .
hence clearly @xmath1015 .
now suppose that for some @xmath181 this inclusion were strict ; choose smallest such @xmath181 ( necessarily @xmath1016 since @xmath1017 and @xmath1018 ) .
then the projection @xmath1019 is the jones projection for @xmath1020 and also for @xmath1021 .
thus the index of @xmath1020 is the same as that of @xmath1022 .
but since @xmath1015 , multiplicativity of index entails @xmath1023=1 $ ] and thus @xmath1024 , a contradiction .
we have seen in theorem [ thm : hrcommutsasalgarepks ] that the multiplication induced by @xmath1029 ( hence @xmath31 ) on @xmath39 is precisely that of the multiplication tangle . by @xcite , to conclude that the planar algebra structure defined on @xmath3 by this identification with the higher relative commutants for @xmath37 we have to check the following . for these
, 1 ) follows from the definition of extremality in @xcite and a simple diagrammatic manipulation involving spherical invariance of the partition function .
2 ) was proved as part of theorem [ thm : hrcommutsasalgarepks ] .
properties 3 ) and 4 ) are just obvious pictures .
the only one that requires any thought is 5 ) which we now prove .
any element in @xmath1033 is in the image of the map from @xmath1034 to @xmath38 defined by the following annular tangle:@xmath1035{\vbox to3em{\vfill } \qquad\qquad\circlenode{x}{\ a\ } \qquad\qquad \ncdiag[arm=0,anglea=30,angleb=30]{x}{outer } \ncdiag[arm=0,anglea=60,angleb=60]{x}{outer } \ncdiag[arm=0,anglea=120,angleb=120]{x}{outer } \ncdiag[arm=0,anglea=150,angleb=150]{x}{outer } \nput[labelsep=-0.25]{135}{outer}{\cdots } \nput[labelsep=-0.25]{45}{outer}{\cdots } \ncbar[arm=-0.3,linearc=0.2,angle=90,offseta=0.2,offsetb=0.2]{outer}{outer } \nput[labelsep=-0.22]{90}{outer } { * } } } \end{array}.\ ] ] ( the shading is determined by the stars being in unshaded regions , the position of @xmath11 on the inside box being irrelevant . )
[ proof of claim . ]
it is a simple diagrammatic calculation to show that the image of this tangle does indeed commute with @xmath1036 . on the other hand
the tangle defines an injective map ( the inverse tangle is obvious ) and from general subfactor theory the dimensions of @xmath1034 and @xmath1033 are the same .
if @xmath403 is in @xmath38 , identified with @xmath39 , then@xmath1037{\vbox to3em{\vfill } \qquad\qquad\ovalnode{x}{\quad a\quad } \qquad\qquad \ncdiag[arm=0,anglea=20,angleb=30]{x}{outer } \ncdiag[arm=0,anglea=45,angleb=60]{x}{outer } \ncdiag[arm=0,anglea=135,angleb=120]{x}{outer } \ncdiag[arm=0,anglea=160,angleb=150]{x}{outer } \nput[labelsep=-0.25]{135}{outer}{\cdots } \nput[labelsep=-0.25]{45}{outer}{\cdots } \ncbar[arm=-0.3,linearc=0.2,angle=90,offseta=0.2,offsetb=0.2]{outer}{outer } \nput[labelsep=-0.22]{90}{outer } { * } \ncbar[arm=0.3,linearc=0.2,angle=90,offseta=0.2,offsetb=0.2]{x}{x } \nput[labelsep=0]{90}{x } { * } } } \end{array}.\ ] ] v.f.r .
annular structure of subfactors . in e.
ghys , p. de la harpe , v.f.r .
jones , v. sergiescu , and t. tsuboi , editors , _ essays on geometry and related topics _ , volume 32 of _ lenseignment mathematique _ , 2001 .
m. pimsner . a class of @xmath1042-algebras generalizing both cuntz - krieger algebras and crossed products by @xmath1043 . in d .-
voiculescu , editor , _ free probability _ , volume 12 of _ fields institute communications _ , pages 189212 .
american mathematical society , 1997 .
symmetries of some reduced free product @xmath1042-algebras , in `` operator algebras and their connections with topology and ergodic theory '' , lecture notes in math .
1132 , springer verlag , 1985 , 556588 . dan voiculescu .
lectures on free probability theory . in _ lectures on probability theory and statistics ( saint - flour , 1998 ) _ , volume 1738 of _ lecture notes in math .
_ , pages 279349 .
springer , berlin , 2000 . | using a family of graded algebra structures on a planar algebra and a family of traces coming from random matrix theory , we obtain a tower of non - commutative probability spaces , naturally associated to a given planar algebra .
the associated von neumann algebras are ii@xmath0 factors whose inclusions realize the given planar algebra as a system of higher relative commutants .
we thus give an alternative proof to a result of popa that every planar algebra can be realized by a subfactor .
department of mathematics , uc berkeley , berkeley , ca 94720 , [email protected] .
research supported by nsf grant dms-0401734 + @xmath1 department of mathematics , ucla , los angeles , ca 94720 , [email protected] . research supported by nsf grant dms-0555680 ] |
a family @xmath5 of holomorphic functions defined on a domain @xmath6 is said to be normal in @xmath7 if every sequence in @xmath5 has a subsequence converging uniformly on compact subsets of @xmath7 to a function @xmath8 .
the limit function @xmath8 is holomorphic on @xmath7 ( by weierstrass theorem ) or the constant infinity .
various authors while studying the normality of a family of holomorphic functions take the limit function @xmath9 but for studying the normal families from complex dynamics point of view , one needs to include the case where the limit function @xmath10 . the former approach we shall call as restrictive approach while the later will be called the general approach towards normal families .
the concept of normal families was introduced by p. montel in 1907 @xcite . for
comprehensive account of normal families of meromorphic functions on domains in @xmath11 one can refer to joel schiff s text @xcite , c.t .
chuang s text @xcite and zalcman s survey article @xcite . with the renewed interest in normal families of meromorphic functions , arising largely from the important role they play in complex dynamics
, it seems sensible to talk about normal families of holomorphic functions on different domains of different spaces thereby enabling one to study the dynamics of such functions . in this article
we have considered the families of bicomplex holomorphic functions on bicomplex domains . since
this article lays the foundations of the subject * normal families of bicomplex holomorphic functions * for future investigations in various possible directions , it is necessary to adopt a dual approach towards the study of normality of families of bicomplex holomorphic functions on bicomplex domains .
the first approach is restrictive approach which gives rise to more interesting results when the normal families are studied in their own right .
for example the converse of montel theorem holds under this approach .
the second approach is the general approach in which though the converse of montel theorem fails to hold but is essentially required when the normal families are studied from the bicomplex dynamics point of view . during our discussions , we shall come across the situations where the differences lead to interesting conclusions .
besides complete discussion on montel theorem in various situations , we have defined fatou , julia and filled - in julia sets of bicomplex polynomials and their representations in terms of their complex counterparts are obtained for the specific case of non - degenerate bicomplex polynomials of degree @xmath12 .
also , some 3d visual examples of bicomplex julia sets are given for the specific slice @xmath0 .
bicomplex numbers are defined as : = \{z_1+z_2@xmath13 | z_1 , z_2 ( @xmath14 ) } [ enstetra ] where the imaginary units @xmath15 and @xmath16 are governed by the rules : @xmath17 , @xmath18 and @xmath14@xmath13&= & @xmath13@xmath14&= & @xmath19 , + @xmath14@xmath19&= & @xmath19@xmath14&= & -@xmath13 , + @xmath13@xmath19&= & @xmath19@xmath13&= & -@xmath14 .
+ note that we define @xmath20 and @xmath21 for @xmath22 .
hence , it is easy to see that the multiplication of two bicomplex numbers is commutative .
in fact , the bicomplex numbers @xmath23 are * unique * among the _ complex clifford algebras _ ( see @xcite ) in that they are commutative but not division algebra .
it is also convenient to write the set of bicomplex numbers as : = \{w_0+w_1@xmath14+w_2@xmath13+w_3@xmath19 | w_0 , w_1,w_2,w_3 } . in particular , in equation ( [ enstetra ] ) ,
if we put @xmath24 and @xmath25 with @xmath26 , then we obtain the following subalgebra of hyperbolic numbers , also called duplex numbers ( see , e.g. @xcite , @xcite ) : @xmath27 complex conjugation plays an important role both for algebraic and geometric properties of @xmath28 . for bicomplex numbers ,
there are three possible conjugations .
let @xmath29 and @xmath30 such that @xmath31 .
then we define the three conjugations as : [ eq : dag ] @xmath32 where @xmath33 is the standard complex conjugate of complex number @xmath34 .
if we say that the bicomplex number @xmath35 has the `` signature '' @xmath36 , then the conjugations of type 1,2 or 3 of @xmath37 have , respectively , the signatures @xmath38 , @xmath39 and @xmath40 .
we can verify easily that the composition of the conjugates gives the four - dimensional klein group : [ cols="^,^,^,^,^",options="header " , ] [ eq : groupedag ] where @xmath41 .
the three kinds of conjugation all have some of the standard properties of conjugations , such as : @xmath42 for @xmath43 and @xmath44 . + we know that the product of a standard complex number with its conjugate gives the square of the euclidean metric in @xmath45 .
the analogs of this , for bicomplex numbers , are the following .
let @xmath46 and @xmath47 , then we have that : @xmath48 |w|^{2}_{{\ensuremath{{\bf i_2}}}}&:=w\cdot w^{\dag_{1}}=\left(|z_1|^2-|z_2|^2\right)+2{\mathrm{re}}(z_1\overline z_2){\ensuremath{{\bf i_2}}}\in \mathbb{c}({\ensuremath{{\bf i_2 } } } ) , \\*[2ex ] |w|^{2}_{{\ensuremath{{\bf j}}}}&:=w\cdot w^{\dag_{3}}=\left(|z_1|^2+|z_2|^2\right)-2{\mathrm{im}}(z_1\overline z_2){\ensuremath{{\bf j}}}\in \mathbb{d},\end{aligned}\ ] ] where the subscript of the square modulus refers to the subalgebra @xmath49 or @xmath50 of @xmath51 in which @xmath37 is projected
. note that for @xmath46 and @xmath52 , we can define the usual ( euclidean in @xmath53 ) norm of @xmath37 as @xmath54 .
it is easy to verify that @xmath55 .
hence , the inverse of @xmath37 is given by w^-1= . from this
, we find that the set @xmath56 of zero divisors of @xmath51 , called the _ null - cone _ , is given by @xmath57 , which can be rewritten as = \{z(@xmath14@xmath13)| z(@xmath14)}. it is also possible to define differentiability of a function at a point of @xmath51 : let @xmath58 be an open set of @xmath51 and @xmath59 .
then , @xmath60 is said to be @xmath51-differentiable at @xmath61 with derivative equal to @xmath62 if @xmath63 * note : * the subscript of the limit is there to recall that the division is possible only if @xmath64 is invertible .
we also say that the function @xmath8 is bicomplex holomorphic ( @xmath51-holomorphic ) on an open set u if and only if @xmath8 is @xmath51-differentiable at each point of u. using @xmath65 , a bicomplex number @xmath37 can be seen as an element @xmath66 of @xmath67 , so a function @xmath68 of @xmath51 can be seen as a mapping @xmath69 of @xmath67 . here
we have a characterization of such mappings : let @xmath58 be an open set and @xmath60 such that @xmath70 , and let @xmath68 .
then @xmath8 is @xmath51-holomorphic on @xmath58 if and only if @xmath71 and @xmath72 + moreover , @xmath73 and @xmath74 is invertible if and only if @xmath75 .
[ theobasic ] this theorem can be obtained from results in @xcite and @xcite .
moreover , by the hartogs theorem @xcite , it is possible to show that @xmath76 " can be dropped from the hypotheses .
hence , it is natural to define the corresponding class of mappings for @xmath67 : the class of @xmath51-holomorphic mappings on a open set @xmath77 is defined as follows : @xmath78 it is the subclass of holomorphic mappings of @xmath67 satisfying the complexified cauchy - riemann equations .
we remark that @xmath79 in terms of @xmath67 if and only if @xmath8 is @xmath51-differentiable on @xmath58 .
it is also important to know that every bicomplex number @xmath80 has the following unique idempotent representation : @xmath81 where @xmath82 and @xmath83 .
this representation is very useful because addition , multiplication and division can be done term - by - term .
it is also easy to verify the following characterization of the non - invertible elements .
an element @xmath65 is in the null - cone if and only if @xmath84 or @xmath85 .
[ null ] the notion of holomorphicity can also be seen with this kind of notation .
for this we need to define the projections @xmath86 as @xmath87 and @xmath88 .
we say that @xmath89 is a @xmath51-cartesian set determined by @xmath90 and @xmath91 if @xmath92 . in @xcite
it is shown that if @xmath90 and @xmath91 are domains ( open and connected ) of @xmath93 then @xmath94 is also a domain of @xmath51 .
then , a way to construct some discus " ( of center 0 ) in @xmath51 is to take the @xmath51-cartesian product of two discs ( of center 0 ) in @xmath93 .
hence , we define the discus " with center @xmath95 of radius @xmath96 and @xmath97 of @xmath51 as follows @xcite : @xmath98 where @xmath99 is the open ball with center @xmath100 and radius @xmath101 . in the particular case
where @xmath102 , @xmath103 will be called the @xmath51-disc with center @xmath104 and radius @xmath105 . in particular , we define @xmath106 .
we remark that @xmath107 is , in fact , the lie ball ( see @xcite ) of radius @xmath105 in @xmath51 .
now , it is possible to state the following striking theorems ( see @xcite ) : let @xmath90 and @xmath91 be open sets in @xmath93 . if @xmath108 and @xmath109 are holomorphic functions of @xmath93 on @xmath90 and @xmath91 respectively , then the function @xmath110 defined as @xmath111 is @xmath51-holomorphic on the open set @xmath94 and @xmath112 @xmath113 [ theo5 ] let @xmath114 be an open set in @xmath51 , and let @xmath115 be a @xmath51-holomorphic function on x. then there exist holomorphic functions @xmath116 and @xmath117 with @xmath118 and @xmath119 , such that : @xmath120 @xmath121 [ theo4 ]
since the concepts like * uniform boundedness * , * local uniform boundedness * , and * uniform convergence on compact sets * are defined for functions on any metric space and do not depend on bicomplex holomorphic functions , we assume these concepts in our discussion and refer the reader to any standard text on analysis ( e.g. see @xcite and @xcite ) . we start our discussion with the following definition of normality .
a family @xmath5 of bicomplex holomorphic functions defined on a domain @xmath122 is said to be * normal * in @xmath7 if every sequence in @xmath5 contains a subsequence which converges locally uniformly on @xmath7 .
@xmath5 is said to be * normal at a point * @xmath123 if it is normal in some neighborhood of @xmath124 in @xmath7 .
[ basicnormal ] let us consider @xmath125 be a @xmath51-holomorphic function on @xmath7 .
then by theorem [ theo4 ] , there exist holomorphic functions @xmath126 and @xmath127 such that @xmath128 we define the * norm * of @xmath8 on @xmath7 as @xmath129 one can easily see that * @xmath130 ; * @xmath131 ; * @xmath132 ; * @xmath133 . thus , the linear space of bicomplex holomorphic functions on a domain @xmath122 is a normed space under the above norm .
we start with a * uniformly bounded * family @xmath5 of bicomplex holomorphic functions . in this case
, we can verify directly the following result .
a family @xmath5 of bicomplex holomorphic functions defined on a bicomplex cartesian domain @xmath7 is uniformly bounded on @xmath7 if and only if @xmath134 is uniformly bounded on @xmath135 [ unibounded ] if we consider now a * locally uniformly bounded * family @xmath5 of bicomplex holomorphic functions , we can prove a similar result since a set @xmath136 is compact if and only if @xmath137 is compact for @xmath138 .
a family @xmath5 of bicomplex holomorphic functions defined on a bicomplex cartesian domain @xmath7 is locally uniformly bounded on @xmath7 if and only if @xmath134 is locally uniformly bounded on @xmath139 [ thmp ] _ proof _ let @xmath5 be locally uniformly bounded on @xmath7 .
then for every compact set @xmath140 there is a constant @xmath141 such that @xmath142 thus , @xmath143 @xmath144 therefore , @xmath145 and @xmath146 now , let @xmath147 be a compact subset of @xmath148 . then there is ( always ) a compact subset @xmath149 of @xmath150 ( even singleton will do ) such that @xmath151 say , is a compact subset of @xmath7 with @xmath152 thus ( [ eq04 ] ) holds for any compact subset of @xmath148 , and similarly for ( [ eq05 ] ) .
conversely , suppose @xmath153 is locally uniformly bounded on @xmath154 let @xmath155 be any compact subset of @xmath7 . then by continuity of @xmath156 ,
@xmath157 is compact subset of @xmath158 and hence there are constants @xmath159 and @xmath160 such that @xmath161 and @xmath162 therefore , @xmath163 since @xmath164 , ( [ eq06 ] ) holds for @xmath155 also and this completes the proof .
@xmath165 what happens if @xmath7 is not a bicomplex cartesian product ? in the case of uniformly bounded family of bicomplex holomorphic functions ( theorem [ unibounded ] ) , it is easy to verify that the result is true for any domain . in the case of
locally uniformly bounded family of bicomplex holomorphic functions , we need to recall the following results from the bicomplex function theory .
a domain @xmath166 is a domain of holomorphism for functions of a bicomplex variable if and only if @xmath7 is a @xmath51-cartesian set ( @xcite , theorem 15.11 ) , and if @xmath7 is not a domain of holomorphism then @xmath167 , and there exists a holomorphic function which is a bicomplex holomorphic continuation of the given function from @xmath7 to @xmath168 ( @xcite corollary 15.4 ) .
[ rek ] a family @xmath5 of bicomplex holomorphic functions defined on a arbitrary bicomplex domain @xmath7 is locally uniformly bounded on @xmath7 if and only if @xmath134 is locally uniformly bounded on @xmath139 [ lem01 ] _ proof _ if @xmath134 is locally uniformly bounded on @xmath169 for @xmath138 , from remark [ rek ] , we can extend @xmath7 to @xmath168 and apply theorem [ thmp ] to obtain that @xmath5 is locally uniformly bounded on @xmath168 . for the other side ,
we need to recall that a family @xmath5 is locally uniformly bounded on @xmath7 if and only if the family @xmath5 is * locally bounded * on @xmath7 i.e. for each @xmath170 there is a positive number @xmath171 and a neighbourhood @xmath172 such that @xmath173 for all @xmath174 and all @xmath175 ( see @xcite ) . since @xmath172 is a bicomplex cartesian product of two discs in the plane , it is easy to verify that the family @xmath153 is bounded by @xmath176 on @xmath177 for @xmath138 . as @xmath178 was arbitrary , @xmath134 is locally bounded on @xmath179 .
@xmath165 we are now ready to prove the bicomplex version of the montel theorem .
let @xmath5 be a family of bicomplex holomorphic functions defined on a bicomplex domain @xmath7 .
if @xmath180 is normal on @xmath169 for @xmath138 then @xmath5 is normal on @xmath7 .
[ normal ] _ proof _ suppose that @xmath134 is normal on @xmath181 we want to show that @xmath5 is normal in @xmath7 .
let @xmath182 be any sequence in @xmath5 and @xmath155 be any compact subset of @xmath7 .
then @xmath183 is a sequence in @xmath184 since @xmath185 is normal in @xmath186 then @xmath187 has a subsequence @xmath188 which converges uniformly on @xmath189 to a @xmath190-function .
now , consider @xmath191 in @xmath5 .
then @xmath192 is a sequence in @xmath193 .
since @xmath193 is normal in @xmath186 then @xmath194 has a subsequence @xmath195 which converges uniformly on @xmath196 to a @xmath190-function .
this implies that @xmath197 is a subsequence of @xmath198 which converges uniformly on @xmath199 to a bicomplex function showing that @xmath5 is normal in @xmath7 .
@xmath165 * ( montel ) * every locally uniformly bounded family of bicomplex holomorphic functions defined on a bicomplex domain is a normal family .
[ montel ] _ proof _ let @xmath5 be a locally uniformly bounded family of bicomplex holomorphic functions defined on a domain @xmath166 . using theorem [ lem01 ] , we have that @xmath134 is locally uniformly bounded on @xmath139 hence , from the classical montel theorem , @xmath134 is normal on @xmath169 for @xmath138 and by lemma [ normal ] we obtain that @xmath5 is normal on @xmath7.@xmath165 * note : * the converse of bicomplex montel theorem is also true . indeed , suppose that @xmath5 is normal and not locally uniformly bounded in @xmath7 .
then in some closed discus @xmath200 in the domain @xmath7 , for each @xmath201 there is a function @xmath202 and a point @xmath203 such that @xmath204 since @xmath5 is normal , there is a subsequence @xmath205 of @xmath206 converging uniformly on @xmath200 to a bicomplex ( holomorphic ) function @xmath8 .
that is , for some positive integer @xmath207 , we have @xmath208 thus , if @xmath209 , then @xmath210 and this is a contradiction . the above discussion permits to establish the following results .
the family @xmath5 of bicomplex holomorphic functions is normal on the arbitrary domain @xmath7 if and only if @xmath153= @xmath211 is normal on @xmath169 for @xmath212 [ maintheo ] if the family @xmath5 of bicomplex holomorphic functions is normal on a arbitrary domain @xmath213 , then @xmath5 is normal on the larger domain @xmath168 .
a family of bicomplex holomorphic functions @xmath5 is normal on a arbitrary domain @xmath7 if and only if @xmath5 is normal at each point of @xmath7 .
in this section , we want to show that it is possible to see the bicomplex montel theorem ( theorem [ montel ] ) as a particular case of the following montel theorem of several complex variables ( see @xcite ) .
let @xmath214 be an open set and @xmath215 be a family of holomorphic mappings .
then the following are equivalent : * the family @xmath216 is locally uniformly bounded . *
the family @xmath216 is relatively compact in @xmath217 .
[ bmc2 ] since , @xmath218 , we obtain directly the desired result using the fact that a family @xmath216 is relatively compact in @xmath217 , i.e. @xmath219 is compact in @xmath217 , if and only if @xmath216 is a normal family ( see @xcite ) .
moreover , theorem [ bmc2 ] will be proven for the specific class @xmath220 instead of @xmath221 if we can show that @xmath220 is closed in @xmath221 with the compact convergence topology .
this is a direct consequence of the following * bicomplex weierstrass theorem*. let @xmath206 be a sequence of bicomplex holomorphic functions which converges locally uniformly to a function @xmath8 on a @xmath51-disc @xmath222 . then @xmath8 is bicomplex holomorphic in @xmath222 .
[ weierslem ] _ proof _ since @xmath223 is @xmath51-holomorphic on @xmath222 @xmath224 , we have from theorem [ theo4 ] that @xmath225 is holomorphic for @xmath138 , @xmath224 .
since @xmath222 is a bicomplex cartesian product , by the weierstrass theorem of one complex variable , the sequence @xmath226 must converges locally uniformly to the holomorphic function @xmath227 on @xmath228 for @xmath138 .
therefore , from theorem [ theo5 ] , the function @xmath229 is @xmath51-holomorphic on @xmath222.@xmath165 * ( weierstrass ) * let @xmath206 be a sequence of bicomplex holomorphic functions on a domain @xmath7 which converges uniformly on compact subsets of @xmath7 to a function @xmath8 .
then @xmath8 is bicomplex holomorphic in @xmath7 .
_ proof _ for an arbitrary @xmath170 , choose a @xmath51-disc @xmath172 . since @xmath230 locally uniformly on @xmath7 , by lemma [ weierslem ]
, @xmath8 is @xmath51-holomorphic on @xmath7 . as @xmath178 was arbitrary , @xmath231 is @xmath51-holomorphic on @xmath7.@xmath165
to carry further the study of normal families of bicomplex holomorphic functions particularly to consider the dynamics of bicomplex holomorphic functions , we propose the following more general definition of normality . a family @xmath5 of bicomplex holomorphic functions defined on a domain @xmath122 is said to be * normal * in @xmath7 if every sequence in @xmath5 contains a subsequence which on compact subsets of @xmath7 either converges uniformly to a limit function or converges uniformly to @xmath232 .
@xmath5 is said to be * normal at a point * @xmath123 if it is normal in some neighborhood of @xmath124 in @xmath7 .
[ compnormal ] we say that a sequence @xmath233 of bicomplex numbers converges to @xmath232 if and only if the norm @xmath234 congerges to @xmath232 . we note that our proofs of the bicomplex montel theorem work also in this situation .
however , as for one complex variable , the converse of theorem [ montel ] will not remain valid with this more complete definition of normality ( see @xcite ) .
both the situations in the last definition may occur simultaneously .
for example , consider the family @xmath235 of bicomplex holomorphic functions on @xmath236 .
then , by using the idempotent representation and results from one complex variable theory of normal families , we find that this family is normal on @xmath237 , where @xmath238 and @xmath239 on the set @xmath240 , normality is under the first situation whereas on the set @xmath241 the normality is under the second situation .
consider the family @xmath242 then @xmath243 , but @xmath244 for @xmath245 .
it follows that @xmath5 can not be normal in any domain containing the origin .
now , let us prove that theorem [ maintheo ] is only true in one direction with this more general definition of normality .
let @xmath5 be a family of bicomplex holomorphic functions defined on a domain @xmath7 .
if @xmath134 is normal on @xmath169 for @xmath138 then @xmath5 is normal on @xmath7 .
[ normal2 ] _ proof _ suppose that @xmath134 is normal on @xmath181 we want to show that @xmath5 is normal in @xmath7 .
let @xmath182 be any sequence in @xmath5 and @xmath155 be any compact subset of @xmath7 .
then @xmath183 is a sequence in @xmath184 since @xmath185 is normal in @xmath186 then @xmath187 has a subsequence @xmath188 which converges uniformly on @xmath189 to either a @xmath190-function or to @xmath232 .
now , consider @xmath191 in @xmath5 .
then @xmath192 is a sequence in @xmath193 .
since @xmath193 is normal in @xmath186 then @xmath194 has a subsequence @xmath195 which converges uniformly on @xmath196 to either a @xmath190-function or to @xmath232 .
this implies that @xmath197 is a subsequence of @xmath198 which converges uniformly on @xmath199 to a bicomplex function or to @xmath232 showing that @xmath5 is normal in @xmath7 .
@xmath165 here is the counterexample for the other side .
let @xmath90 and @xmath91 be domains in @xmath93 containing the origin .
let @xmath246 .
then @xmath7 is not a bicomplex cartesian domain because @xmath247 .
now the family @xmath248 is normal in the domain @xmath7 ( by the proposed definition of normality as above ) but @xmath134 is not normal in @xmath249 as it contains the origin .
moreover , the next examples show that the converse of theorem [ normal2 ] is not true even if the domain @xmath7 is a bicomplex cartesian product .
consider the family @xmath250 on @xmath190 .
then @xmath251 is normal on the punctured disc @xmath252 but not normal on the disc @xmath253 .
however , the bicomplex family @xmath254 where @xmath255 is normal in the following bicomplex cartesian product : @xmath256 since the limit function is identically infinite . consider the family @xmath257 on @xmath190 .
then @xmath251 is normal on @xmath258 where here the limit function is identically infinite , but not normal on @xmath190 since @xmath259 .
however , the bicomplex family @xmath260 where @xmath261 , is normal in the following bicomplex cartesian product : @xmath262 since the limit function is identically infinite .
[ maincounter ]
we conclude this article with the following general definition of * fatou * and * julia * sets for bicomplex polynomials
. let @xmath263 be a bicomplex polynomials .
we define the bicomplex julia set for @xmath264 as @xmath265 and the bicomplex fatou ( or stable ) set as @xmath266 hence , about each point @xmath267 , there is a neighborhood @xmath268 in which @xmath269 is a normal family .
therefore , @xmath270 is an open set , the connected components of which are the maximal domains of normality of @xmath269 , and @xmath271 is a closed set . from theorem [ normal2 ] , we obtain the following inclusion : @xmath272^{\circ n}(z_1-z_2{\ensuremath{{\bf i_1}}})\}\}\mbox { or } \label{inc}\\ & & \{[\mathcal{p}_2(p)]^{\circ n}(z_1+z_2{\ensuremath{{\bf i_1}}})\}\mbox { is not normal \large \}}\\ & = & [ \mathcal{j}_1(\mathcal{p}_1(p))\times_e \mathbb{c}({\ensuremath{{\bf i_1}}})]\cup [ \mathbb{c}({\ensuremath{{\bf i_1}}})\times_e \mathcal{j}_1(\mathcal{p}_2(p))].\end{aligned}\ ] ] however , from example [ maincounter ] , we know that ( [ inc ] ) can not be transformed into equality .
in fact , to obtain a characterization of bicomplex julia sets in terms of one variable dynamics we need to use the concept of filled - in julia set . as for the complex case ,
the bicomplex * filled - in julia * set @xmath273 of a polynomial @xmath264 is defined as the set of all points @xmath274 of dynamical space that have bounded orbit with respect to @xmath264 , that is to say : @xmath275 we remark that @xmath276 is a closed set . as for the classical case ( see @xcite , p.65 ) , we need to consider polynomials of degree @xmath12 to be able to see a bicomplex julia set as the boundary of a bicomplex filled - in julia set .
in fact , to decompose @xmath277 in terms of two complex polynomials of degree @xmath12 , we must also consider non - degenerate bicomplex polynomials of the form @xmath278 where @xmath279 . under these specifications , we have the following result .
let @xmath263 be a non - degenerate bicomplex polynomials of degree @xmath12 .
then , @xmath280 [ filled ] now , using the concept of normality in terms of * definition [ basicnormal ] * , we obtain the following characterization of @xmath281 : @xmath282 moreover , using the idempotent representation , it is easy to see that the bicomplex filled - in julia set @xmath276 can be expressed in terms of two filled - in julia sets in the plane .
more specifically , @xmath283 hence , since @xmath284=[\partial \mathcal{k}_1(\mathcal{p}_1(p))\times_e \mathcal{k}_1(\mathcal{p}_2(p))]\cup [ \mathcal{k}_1(\mathcal{p}_1(p))\times_e \partial \mathcal{k}_1(\mathcal{p}_2(p))]$ ] , we have the following characterization of the bicomplex julia set @xmath271 in terms of one complex variable dynamics .
let @xmath263 be a non - degenerate bicomplex polynomials of degree @xmath12 .
then , @xmath285\cup [ \mathcal{k}_1(\mathcal{p}_1(p))\times_e \mathcal{j}_1(\mathcal{p}_2(p ) ) ] \label{put}.\ ] ] [ grabache ] consider the bicomplex polynomial : @xmath286 we can verify that @xmath287 for @xmath22 . in the complex plane ( in @xmath288 ) , it is well known that @xmath289 and @xmath290 ( see @xcite or @xcite ) . hence , using theorem [ grabache ] , we obtain that @xmath291\cup [ \{z:|z|<1\}\times_e \{z:|z|=1\}].\ ] ] in the particular case of the bicomplex quadratic polynomial @xmath292 the * fundamental * definition of bicomplex julia set of this article ( see [ fund ] ) coincide with the definition , using boundary of bicomplex filled - in julia set , introduced by d. rochon in @xcite ( see theorem [ filled ] ) . moreover , using some distance estimation formulas that can be used to ray traced slices of bicomplex filled - in julia sets in dimension three ( see @xcite ) , we obtain some visual examples ( see fig . 1 , 2 , 3 and 4 ) of bicomplex julia sets @xmath293 for the specific slice @xmath0 .
consider the bicomplex polynomial : @xmath294 we can verify that @xmath295 for @xmath22 . in the complex plane ( in @xmath288 ) , it is well known that @xmath296 is a cantor set .
we shall denote such cantor set by @xmath297 .
hence , using theorem [ grabache ] , we obtain that @xmath298 , title="fig:",width=147 ] , title="fig:",width=147 ] , title="fig:",width=147 ] consider the bicomplex polynomial : @xmath299 we can verify that @xmath300 for @xmath22 . in the complex plane ( in @xmath288 )
, it is well known that @xmath301 is the so - called * airplane*. we shall denote this set by @xmath302 . hence , using theorem [ grabache ] , we obtain that @xmath303\cup [ \mathcal{a}\times_e \partial \mathcal{a } ] \ \ \mbox{(fig .
2)}.\ ] ] , title="fig:",width=147 ] , title="fig:",width=147 ] , title="fig:",width=147 ] consider the bicomplex polynomial : @xmath304.\ ] ] we can verify that @xmath305)=z^2 + 0.26 $ ] and @xmath305)=z^2 - 1.754878 $ ] .
hence , using theorem [ grabache ] , we obtain that @xmath306 , title="fig:",width=147 ] , title="fig:",width=147 ] , title="fig:",width=147 ] consider the bicomplex polynomial : @xmath307.\ ] ] we can verify that @xmath308)=z^2+(-0.123 + 0.745{\ensuremath{{\bf i_1}}})$ ] and @xmath308)=z^2+(-0.391 - 0.587{\ensuremath{{\bf i_1}}})$ ] . in the complex plane ( in @xmath288 ) , it is well known that @xmath309 is the so - called * douady s rabbit * and @xmath310 is a * siegel disk*. hence , using theorem [ grabache ] , we obtain that @xmath311\cup [ \mathcal{d}\times_e \partial \mathcal{s } ] \ \ \mbox{(fig .
4)}.\ ] ] , title="fig:",width=226 ] , title="fig:",width=226 ] by using the definition of bicomplex fatou set as the complement of bicomplex julia set ( [ put ] ) leads us to characterize the bicomplex fatou set of non - degenerate bicomplex polynomials of degree @xmath12 as @xmath312\cup [ \mathcal{f}_1(\mathcal{p}_1(p))_{\infty}\times_e \mathcal{j}_1(\mathcal{p}_2(p))]\ ] ] @xmath313\label{do}\ ] ] where @xmath314 denotes the unbounded component of the fatou set of projections of @xmath315 consider the bicomplex polynomial : @xmath286 in the complex plane ( in @xmath288 ) , it is well known that @xmath316 and @xmath317 .
hence , using ( [ do ] ) , we obtain that @xmath318\ ] ] @xmath319\ ] ] @xmath320.\ ] ] as a direct consequence of theorem [ grabache ] , the bicomplex julia set @xmath271 is completely invariant under the substitution @xmath321 when @xmath264 is a non - degenerate bicomplex polynomials of degree @xmath12 .
the next theorem will prove this result in general .
let @xmath322 be any bicomplex function .
the function @xmath8 is said to be strongly non - constant on @xmath7 if @xmath227 is non - constant on @xmath169 for @xmath138 .
let @xmath8 be an entire strongly non - constant @xmath51-holomorphic function and @xmath323 then , + 1 . if a point @xmath324 , then @xmath325 ; + 2 . if @xmath324 and @xmath326 is a point such that @xmath327 then @xmath328 .
_ proof _ since our definition of normality ( def .
[ compnormal ] ) is analogous to the related notion in the complex plane , the proof of the theorem is same as the proof of the corresponding result in the plane ( see @xcite , theorem 2.17 and 2.18 ) . *
note : * the condition for @xmath8 to be strongly non - constant is needed in the proof of ( 2 . ) to be able to use the open mapping theorem in each components.@xmath165
the research of d.r . is partly supported by grants from crsng of canada and fqrnt of qubec . | in this article , we introduce the concept of normal families of bicomplex holomorphic functions to obtain a bicomplex montel theorem .
moreover , we give a general definition of fatou and julia sets for bicomplex polynomials and we obtain a characterization of bicomplex fatou and julia sets in terms of fatou set , julia set and filled - in julia set of one complex variable . some 3d visual examples of bicomplex julia sets are also given for the specific slice @xmath0 .
* k.s .
charak@xmath1,d .
rochon@xmath2,n . sharma@xmath3 * @xmath4 department of mathematics , university of jammu , jammu-180 006 , india . + e - mail : [email protected] @xmath2 dpartement de mathmatiques et dinformatique , universit du qubec trois - rivires , c.p
. 500 trois - rivires , qubec , canada g9a 5h7 .
+ e - mail : [email protected] , web : www.3dfractals.com @xmath3 department of mathematics , university of jammu , jammu-180 006 , india .
+ e - mail : [email protected] * keywords : * bicomplex numbers , complex clifford algebras , normal families , montel theorem , bicomplex dynamics , julia set , fatou set . |
recently , two rotation - powered x - ray pulsars were discovered in the young supernova remnants [email protected] and n157b .
the pulsar ax j1811.5@xmath01926 in the galactic remnant [email protected] , discovered in archival _ asca _ data , has a 65 ms period ( torii et al .
psr j0537@xmath06910 , discovered in n157b in the large magellanic cloud ( lmc ) , was found in _
data and has a period of 16 ms , making it the fastest known pulsar associated with a remnant ( marshall et al .
the discovery of these two pulsars brings the total number of known young crab - like pulsars in plerionic nebulae to five and doubles the number of such systems known in the lmc .
the discovery of pulsed radio emission from these systems would be important for several reasons . a determination of the dispersion measure ( dm ) for these pulsars would constrain the electron distribution toward the lmc as well as the taylor & cordes ( 1993 ) galactic dm - distance model .
in addition , determining the radio luminosity distribution for young pulsars is important for population synthesis models
. furthermore , ground - based radio telescope facilities are often more practical for long - term timing of pulsars than are x - ray instruments . for young pulsars with high spin - down rates , long - term timing can determine accurate braking indexes . in the several cases where young pulsars show infrequent but sudden spin - ups known as glitches
, radio timing can also give information on the interior physics of the neutron star ( e.g. shemar & lyne 1996 ) .
in addition , measuring a time - of - arrival phase offset between the radio and x - ray pulses helps constrain the pulsar magnetic field geometry and radiation emission mechanism ( e.g. romani & yadigaroglu 1995 ) .
we have undertaken a search for pulsed radio emission from these two pulsars using the 64-m radio telescope in parkes , nsw , australia .
we conducted the search at two center frequencies , 660 mhz and 1374 mhz .
for the 1374 mhz observations , we used the center beam of the newly installed 20-cm multibeam receiver with a filter - bank back end designed and built at jodrell bank .
the search setup used was similar to that of the current parkes multibeam pulsar survey , described elsewhere ( camilo et al .
ax j1811.5@xmath01926 was observed for a single 4.5 hr integration at 1374 mhz while psr j0537@xmath06910 was observed twice at 660 mhz in separate 4 hr integrations and once at 1374 mhz for 6 hr .
data were recorded on magnetic tape for processing offline . as part of this same project
, we searched for and found the pulsar psr j1617@xmath05055 at 1374 mhz near the supernova remnant rcw 103 ( kaspi et al .
1998 ) which was also first discovered as an x - ray pulsar from archival _ asca _ data ( torii et al .
1998 ) .
since the predicted topocentric period for ax j1811.5@xmath01926 was uncertain due to the unknown period derivative of the pulsar , we performed a standard spectral analysis , after first dedispersing the data at a variety of dms from 0 to 1477 pc @xmath2 .
after each dedispersion , the frequency channels were summed and a power spectrum computed from the resulting time series .
this power spectrum was then searched for significant spikes .
each candidate frequency was then further investigated by dedispersing and folding the original data at the candidate dm and period .
a search was then performed for a small range in dm and period around the nominal dm and period and the highest resulting signal - to - noise profile was computed . since the ephemeris for psr j0537@xmath06910 has a well - constrained period derivative ( marshall et al .
1998 ) , we folded the data at the predicted topocentric period at a variety of dms spanning 0 to 300 pc @xmath2 , which includes the expected dm range of 50 to 200 pc @xmath2 for lmc pulsars , based on previous pulsar discoveries ( mcconnell et al .
1991 , manchester et al .
1993 ) . in each dm trial , the data were searched in a period range spanning the nominal folding period .
for the 660 mhz data , this period range was @xmath3150 ns , and for the 1374 mhz data the range was @xmath395 ns , corresponding to @xmath336@xmath4 and @xmath323@xmath4 from the expected period , respectively . we have also searched the dedispersed time series for crab - like giant radio pulses at the expected range of dms . in this search , each dedispersed time series was rebinned at a variety of widths and searched for samples greatly exceeding the noise .
potential giant pulse candidates were investigated by examining the non - dispersed time series for a strong signal in the same part of the time series in order to rule out signals from terrestrial interference . for pulsar ax j1811.5@xmath01926 , we searched a dm range from 40 to 670 pc @xmath2 and found no convincing giant pulse candidates . for psr j0537@xmath06910 , we searched the two 660 mhz data sets and the 1374 mhz data set over the expected lmc dm range of 50 to 200 pc @xmath2 .
we found no convincing giant pulse candidates in any of the three n157b data sets .
we did not detect radio pulsations from either of the two x - ray pulsars . by assuming distances to the remnants and using the taylor & cordes ( 1993 ) dm - distance model to estimate temporal smearing of pulses from dispersion
, we set upper limits for pulsed emission of 0.07 mjy and 0.06 mjy at 1374 mhz for ax j1811.5@xmath01926 and psr j0537@xmath06910 , respectively . we have also set an upper limit of 0.18 mjy for the 660 mhz flux for psr j0537@xmath06910 . by assuming a radio power law index of @xmath5 = 2 and distances of @xmath6 kpc for [email protected]
( green et al . 1988 ) and @xmath7 kpc for n157b ( gould 1995 ) , we obtain 400 mhz luminosity ( @xmath8 ) upper limits for the two pulsars , for comparison with other sources . for ax j1811.5@xmath01926 , @xmath9 mjy kpc@xmath1 . for psr j0537@xmath06910 ,
the luminosity upper limit is @xmath10 mjy kpc@xmath1 using the more constraining 660 mhz flux limit .
these results are outlined in table 1 .
* table 1 - summary of radio observations * [ cols="<,^,^ " , ] of the five known radio pulsars having characteristic age @xmath11 kyr ( the crab pulsar , psr b1509@xmath058 , psr b0540@xmath069 , psr b1610@xmath050 , and psr j1617@xmath05055 ) , psr b1509@xmath058 has the lowest pulsed radio luminosity , @xmath12 mjy kpc@xmath1 .
the largest luminosities belong to the crab pulsar and psr b0540@xmath069 , which have @xmath13 mjy kpc@xmath1 and @xmath14 mjy kpc@xmath1 respectively .
of course , all luminosity estimates are distance - dependent , hence subject to some uncertainty .
our upper limit of @xmath15 mjy kpc@xmath1 for ax j1811.5@xmath01926 is significantly lower than the luminosities of the crab pulsar and psr b0540@xmath069 and is below the value for psr b1509@xmath058 .
this suggests that ax j1811.5@xmath01926 should have detectable radio emission . from the taylor & cordes ( 1993 ) dm - distance model
, we estimate that the expected dm for ax j1811.5@xmath01926 is @xmath16350 pc @xmath2 , corresponding to about 3 ms dispersion smearing in each frequency channel , and that the multipath scattering from plasma inhomogeneities should be less than 1 ms .
thus , neither of these effects should significantly affect the detectability of the 65 ms periodicity .
therefore , the non - detection means that either the radio beam does not intersect our line of sight or that the radio luminosities of very young pulsars can be lower than what has been generally thought ( e.g. narayan & ostriker 1990 ) .
if the latter is true , then deeper radio searches with higher sensitivity may reveal a previously undiscovered population of young faint pulsars .
our luminosity limit is not as stringent for psr j0537@xmath06910 due to its large distance .
also , the poorly known interstellar / intergalactic medium in this direction could significantly scatter radio pulses beyond detectability , particularly at 660 mhz .
we conclude from our search that the x - ray pulsar ax j1811.5@xmath01926 in [email protected] has pulsed radio emission that is certainly of significantly lower flux than all other known pulsars of comparable age , and is probably either underluminous relative to these sources or is not beaming toward earth . our much less stringent limits on the radio luminosity for psr j0537@xmath06910 in n157b are unconstraining .
next generation telescopes such as the square kilometer array would be ideal instruments to carry out deeper searches toward these sources . | using the parkes radio telescope , we have searched for pulsed radio emission from the recently discovered x - ray pulsars ax j1811.5@xmath01926 and psr j0537@xmath06910 in the supernova remnants [email protected] and n157b , respectively .
we detected no significant pulsed radio emission from these pulsars and have set an upper limit of 0.07 mjy on the 1374 mhz flux from ax j1811.5@xmath01926 and upper limits of 0.18 mjy and 0.06 mjy for the flux from psr j0537@xmath06910 at 660 mhz and 1374 mhz , respectively . assuming a power law radio spectral index of 2
, these flux limits correspond to luminosity limits at 400 mhz of 20 mjy kpc@xmath1 for ax j1811.5@xmath01926 and 1100 mjy kpc@xmath1 for psr j0537@xmath06910 .
our luminosity limit for ax j1811.5@xmath01926 is lower than the observed luminosities of other young radio pulsars .
the upper limit on the luminosity for psr j0537@xmath06910 in n157b is not significantly constraining due to the large distance to the pulsar .
we have also searched for giant radio pulses from both pulsars and have found no convincing candidates .
# 1 # 2 _ mem .
soc .
astron .
it .
_ * # 1 * , # 2 # 1 # 2 _ the messenger _ * # 1 * , # 2 # 1 # 2 _ astron .
nach . _ * # 1 * , # 2 # 1 # 2 _ astron .
astrophys . _ * # 1 * , # 2 # 1 # 2 _ astron .
astrophys .
lett . _ * # 1 * , l#2 # 1 # 2 _ astron
. astrophys .
rev . _ * # 1 * , # 2 # 1 # 2 _ astron .
astrophys .
suppl .
ser . _ * # 1 * , # 2 # 1 # 2 _ astron .
j. _ * # 1 * , # 2 # 1 # 2 _ ann .
rev .
astron .
astrophys . _ * # 1 * , # 2 # 1 # 2 _ astrophys .
j. _ * # 1 * , # 2 # 1 # 2 _ astrophys .
j. lett .
_ * # 1 * , l#2 # 1 # 2 _ astrophys .
j. suppl .
_ * # 1 * , # 2 # 1 # 2 _ astrophys .
space sci .
_ * # 1 * , # 2 # 1 # 2 _ adv .
space res .
_ * # 1 * , # 2 # 1 # 2 _ bull .
astron .
inst .
czechosl . _ * # 1 * , # 2 # 1 # 2 _ j. quant .
spectrosc .
radiat .
transfer _ * # 1 * , # 2 # 1 # 2 _ mon . not .
r. astr .
soc . _ * # 1 * , # 2 # 1 # 2 _ mem .
r. astr .
soc . _ * # 1 * , # 2 # 1 # 2 _ phys .
lett .
rev . _ * # 1 * , # 2 # 1 # 2 _ publ .
astron .
. japan _ * # 1 * , # 2 # 1 # 2 _ publ .
astr .
soc .
pacific _ * # 1 * , # 2 # 1 # 2 _ nature _ * # 1 * , # 2 epsf.sty |
the interstellar media ( ism ) of galaxies are both the archaeological record of metal production during the star - formation history of the galaxy and the reservoir from which current star - formation draws its material . despite the great interest in tracing the star formation history and evolution of galaxies
surprisingly little is known about the largest sink for metals in the ism , namely dust grains ( see e.g. whittet 1992 ; edmunds 2001 ) . after the _ iras _ all - sky survey many studies were published of the dust properties in both ` active ' and ` normal ' galaxies .
an influential study was conducted by devereux & young ( 1990 ; henceforth dy90 ) in which they derived dust masses from _ iras _ observations of a sample of 58 spiral galaxies and by comparison with co ( as a tracer of h@xmath5 ) and h i measurements of
the gas concluded that the average derived total gas - to - dust mass ratio was @xmath61100 , compared to the value @xmath6100 generally found in our own galaxy . dy90 and others recognized that this almost certainly indicated that _
iras _ measurements at 60 to 100 @xmath0 m were simply not sensitive to the cool dust that may dominate the total dust mass of a galaxy . following the launch of _ iso _ and
the commissioning of scuba on the jcmt and mambo on the iram 30-m telescope , longer wavelength measurements of dust in spiral galaxies have become possible and unsurprisingly several detailed studies have confirmed that indeed there is more dust in spiral galaxies than _ iras _ data alone would suggest ( gu@xmath7lin et al .
1993 ; braine et al .
1995 , 1997 ; neininger & dumke 1999 ; neininger et al . 1996 ; dumke et al . 1997
; alton et al . 1998 ; haas et al .
1998 ; davies et al . 1999 ; contursi et al 2001 ; popescu et al . 2002 ; hippelein et al . 2003 ; meijerink et al .
the scuba local universe galaxy survey ( slugs ; dunne et al .
2000 ; dunne & eales 2001 ) made brief observations of a large sample of _ iras_-selected galaxies .
they concluded that two dust temperature components were required to fit the spectra , both with an emissivity index @xmath8 . in an overlapping study , james et al .
( 2002 ) used scuba observations of galaxies with known metallicity and the assumption that a constant fraction of metals in a galaxy are bound up in dust grains to derive an absolute dust emissivity consistent with the an extrapolation of the far - infrared results of hildebrand ( 1983 ) and casey ( 1991 ) with @xmath92 .
we have also previously published a study of the dust in ngc 3079 ( stevens & gear 2000 ) .
this paper is a follow - up with a much larger sample . for reasons of brevity
we do not in this paper attempt a very detailed study of the morphology of each individual galaxy , which the quality of the images certainly merit , but rather concentrate on the more global properties of the sample and defer more detailed comparison with e.g. co and h i images to a future paper . [ cols="<,^,^,^,^,^,^,^ " , ] \(1 ) test samples . ( 2 ) k - s statistic .
( 3 ) probability that datasets are drawn from different distributions .
because we have a mixture of field and virgo cluster spirals in our sample we can try to make a comparison of the gas to dust ratios in the different environments .
it is well known that some virgo spirals are deficient in h i , presumably due to stripping , however h@xmath5 is not depleted ( kenney & young 1986 , 1988a , b , 1989 ; stark et al .
leggett , brand & mountain ( 1987 ) claimed that there was no difference in the _ iras _ emission from field and virgo spirals , consistent with the idea that _ iras _ traced warm dust associated with well - bound molecular clouds .
however , doyon & joseph ( 1989 ) found evidence that h i deficient galaxies in the virgo cluster have lower 60- and 100-@xmath0 m flux densities and cooler far - infrared colour temperatures than those with normal h i content . they conclude that for a typical spiral in the core of the cluster , at least half the diffuse dust has been stripped .
if true then then we might expect to see a difference at longer wavelengths . specifically ,
if the cold dust that is mixed with the h i is also stripped from the virgo galaxies then this difference would be reflected in the @xmath10 ratios determined in the model fits since these give the mass of dust in the cold component relative to that in the warm component .
the values found for the virgo galaxies are indeed all relatively small ( 9 , 17 , 6 and 5 ) with a mean value of just @xmath11 . for comparison ,
the non - virgo members have a mean value of @xmath12 or @xmath13 excluding ngc 3310 .
however , the samples are not formally distinct ; a two - sample k - s test returns a ks statistic of 0.64 with a probability of 0.89 ( 1.6 @xmath14 ) that the two datasets are drawn from different distributions .
observations of much larger samples would be required to investigate the effect of environment any further .
we have presented high - quality submillimetre images of 14 spiral galaxies .
simple two - component model fits to the global seds show that : \(1 ) the dust is typically radiating at temperatures of @xmath15 k and @xmath16 k and in most cases the warm component dominates the total luminosity .
\(2 ) the dust mass in the warm component correlates with the mass in molecular hydrogen while the dust mass in the cold component correlates with the mass in atomic hydrogen .
the opposite correlations are not significant .
this result suggests that the simple two - component models provide a good approximation to the real physical nature of these galaxies .
\(3 ) the mean gas - to - dust mass ratio of @xmath4 is entirely consistent with that reported for the milky way .
there is evidence that the ratio is higher in the warm , molecular component .
\(4 ) comparing our model results with those found for the slugs survey , we find that our galaxies have , on average , similar dust masses but lower far - infrared luminosities and lower dust temperatures for both the warm and cold components .
we suggest that the lower dust temperatures found for our galaxies can be linked to reduced star - formation activity and thus less intense isrfs than those present in the slugs galaxies .
+ the next step will be to analyse the emission from the individual galaxies in more detail .
in particular to investigate both the correspondence of dust and gas emission and to search for differences in submillimetre spectral index as a function of galactocentric radius .
ma acknowledges the support of a pparc studentship . we thank loretta dunne for insightful comments on the pre - submitted manuscript , ashley james for helpful discussions on dust properties and the referee for constructive criticism .
guest user , canadian astronomy data centre , which is operated by the dominion astrophysical observatory for the national research council of canada s herzberg institute of astrophysics .
this research has made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
the dss images presented in this paper were obtained from the multimission archive at the space telescope science institute ( mast ) .
stsci is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas5 - 26555 .
support for mast for non - hst data is provided by the nasa office of space science via grant nag5 - 7584 and by other grants and contracts .
alton p.b .
, bianchi s. , rand r.j . , xilouris e. , davies j.i .
trewhella m. , 1998 , apj , 507 , l125 alton p.b .
, xilouris e.m .
, misiriotos a. , dasyra k.m .
, dumke m. , 2004 , a&a , 425 , 109 amure m. , 2003 , ph.d .
thesis , cardiff university bendo g.j . , et al .
2003 , aj , 125 , 2361 braine j. , combes f. , van driel w. , 1993 , a&a , 280 , 451 braine j. , krugel e. , sievers , a. , wielebinski r. , 1995 , a&a , 344 , l1 braine j. , gulin m. , dumke m. , brouillet n. , herpin f. , wielebinski r. , 1997 , a&a , 326 , 963 casey s.c . , 1991 , apj , 371 , 183 cayatte v. , van gorkom j.h . , balkowski c. , kotanyi c. , 1990 , aj , 100 , 604 chini r. , kreysa e. , krgel e. , mezger p.g . , 1986 ,
a&a , 166 , l8 chini r. , krgel e. , lemke r. , ward - thompson d. , 1995 , a&a , 295 , 317 contursi a. , boselli a. , gavazzi g. , bertagna e. , tuffs r. , lequeux j. , 2001 , a&a , 365 , 11 dahlem m. , aalto s. , klein u. , booth r. , mebold u. , wielebinski r. , lesch h. , 1990 , a&a , 240 , 237 dahlem m. , ehle m. , ryder s.d . , 2001 , a&a , 373 , 485 davies j.i . , alton p. , trewhella m. , evans r. , bianchi s. , 1999 , mnras , 304 , 495 devereux n.a . , young j.s . , 1990 , apj , 359 , 42 ( dy90 ) de vaucouleurs g. de vaucouleurs a. corwin h. , 1976 , second reference catalogue of bright galaxies , university of texas press , austin doyon r. , joseph r.d . , 1989 , mnras , 239 , 347 draine b.t . , lee h. m. , 1984 , apj , 285 , 89 dumke m. , krause m. , wielebinski r. , 2004 , a&a , 414 , 475 dumke m. , braine j. , krause m. , zylka r. , wielebinski r. , guelin m. , 1997 , a&a , 325 , 124 dunne l. , eales s. , edmunds m. , ivison r. , alexander p. , clements d.l . , 2000 , mnras , 315 , 115 dunne l. , eales s. a. , 2001 , mnras , 327 , 697 edmunds m.g . , 2001 , mnras , 328 , 223 giovanelli r. , haynes m.p . , 1983 , aj 88 , 881 . gu@xmath7lin m. , zylka r. , mezger p.g . , haslam c.g.t .
, kreysa e. , lemke r. , sievers a.w . , 1993 ,
a&a , 279 , l37 .
haas m. , lemke d. , stickel m. , hippelein h. , kunkel m. , herbstmeier u. mattila k .. 1998 , a&a , 338 , l33 haynes m.p .
, giovanelli r. , roberts m.s .
1979 , apj 229,83 .
hildebrand r.h . , 1983 , qjras , 24 , 267 hippelein h. , haas m. , tuffs r.j . , lemke d. , stickel m. , klaas u. , vlk h.j . , 2003 , a&a , 407 , 137 holland w.s . , et al . , 1999 ,
mnras , 303 , 659 huchtmeier w.k . , richter o .-
g , 1989 , a general catalog of h i observations of galaxies ; the reference catalog , xix .
springer - verlag , new york .
james a. , dunne l. , eales s.a .
, edmunds m.g . , 2002 , mnras , 335 , 753 jenness t. , stevens j.a . ,
archibald e. , economou f. , jessop n.e . ,
robson e.i . , 2002 , mnras , 336 , 14 kenney j.d . , young j. , 1986 , apj , 301 , l13 kenney j.d . , young j. , 1988a , apj , 326 , 588 kenney j.d . , young j. , 1988b , apjs , 66 , 261 kenney j.d . , young j. , 1989 , apj , 344 , 761 klass u. , walker h. j. , 2002 , a&a , 391 , 911 leggett s.k . , brand p. , mountain c.m .
1987 , mnras , 228 , l11 moshir m. , et al . , 1990 , iras faint source catalogue , version 2.0 meijerink r. , tilanus r.p.j . , dullemond c.p . , israel f.p .
, van der werf p.p .
, 2004 , a&a , in press ( astro - ph/0409511 ) mulder p.s . , van driel w. , braine j. , 1995 , a&a , 300 , 687 neininger n. , gu@xmath7lin m. , garcia - burillo s. , zylka r. wielebinski r. , 1996 , a&a , 310 , 725 neininger n. , dumke m. 1999 , proc.nat.acad.sci , 96 , 5360 popescu c.c .
, tuffs r.j . ,
, pierini d. , madore b.f . , 2002 , apj , 567 , 221 reach w.t .
, et al . , 1995 , 451 , 188 rand r.j . , 1994 , a&a , 285 , 833 rice w. , lonsdale c.j .
, soifer b.t . , neugebauer g. , koplan e.l . , lloyd l.a . ,
de jong t. , habing h.j . , 1988 , apjs , 68 , 91 ryder s.d .
, zasov a.v . ,
mcintyre v.j . , walsh w. , silchenko o. k. , 1998 , mnras , 293 , 411 sakomoto k. , okumura s.k . , ishizuki s. , scoville n.z . 1999 ,
apjs , 124 , 403 sodroski t.j . , odegard n. , arendt r.g .
, dwek e. , weiland j.l . , hauser m.g .
, kelsall t. , 1997 , apj , 480 , 173 soifer b.t . , boehmer l. , neugebauer g. , sanders d.b . , 1989 ,
aj , 98 , 766 stark a.a . , knapp g.r . ,
bally j. , wilson r.w . , penzias a.a . ,
roe h.e . , 1986 ,
apj , 310 , 660 stevens j.a . ,
gear w.k . , 2000 ,
mnras , 312 , l5 tuffs r.j . , et al . , 2002 , apjs , 139 , 37 van driel w. et al . , 1995 , aj , 109 , 942 warmels r.h . , 1986 , phd thesis , university of groningen wevers b.m.h.r . ,
van der kruit p.c . ,
allen r.j . , 1986 ,
a&as 66 , 505 whittet d.c.b . , 1992 ,
dust in the galactic environment , institute of physics ( iop ) publishing , bristol young j.s . , xie s. , kenney j.d.p .
, rice w.l . , 1989 ,
apjs , 70 , 669 young j.s .
et al . , 1995 ,
apjs , 98 , 219 | we present and analyse high - quality scuba 850- and 450-@xmath0 m images of 14 local spiral galaxies , including the detection of dust well out into the extended disk in many cases .
we use these data in conjunction with published far - infrared flux densities from _ iras _ and _ iso _ , and millimetre - wave measurements from ground - based facilities to deduce the global properties of the dust in these galaxies ; in particular temperature and mass .
we find that simple two - temperature greybody models of fixed dust emissivity index @xmath1 and with typical temperatures of @xmath2 k and @xmath3 k provide good fits to the overall spectral energy distributions .
the dust mass in the cold component correlates with the mass in atomic hydrogen and the mass in the warm component correlates with the mass in molecular hydrogen .
these results thus fit the simple picture in which the cold dust is heated predominantly by the interstellar radiation field while the hot dust is heated predominantly by ob stars in more active regions , although we argue that there is some mixing .
the mean gas - to - dust mass ratio is @xmath4 , very similar to that found within our own galaxy and roughly a factor of 10 lower than that derived from _
iras _ data alone .
the gas - to - dust mass ratios in the warm , molecular component are on average higher than those in the cold , atomic component .
we compare our modelling results with similar results for more luminous spiral galaxies selected at far - infrared wavelengths by the scuba local universe galaxy survey and find that whilst the total dust mass distributions of the two samples are indistinguishable , they have significantly different dust temperature distributions in both the warm and cold components .
we suggest that this difference might be related to the level of star - formation activity in these systems , with the more active galaxies having more intense interstellar radiation fields and higher dust temperatures .
[ firstpage ] galaxies : spiral - galaxies : ism - submillimetre . |
1 . schematic of homodyne locking experiment . + fig .
2 . error signal ( top trace ) and seed transmission through the opo cavity ( bottom trace ) of the opo cavity .
3 . squeezing spectrum at the first fsr of the opo .
schematic of universal squeezer @xcite .
( b ) alternative implementation . | we report on the successful implementation of a new approach to locking the frequencies of an opo - based squeezed - vacuum source and its driving laser .
the technique allows the simultaneous measurement of the phase - shifts induced by a cavity , which may be used for the purposes of frequency - locking , as well as the simultaneous measurement of the sub - quantum - noise - limited ( sub - qnl ) phase quadrature output of the opo .
the homodyne locking technique is cheap , easy to implement and has the distinct advantage that subsequent homodyne measurements are automatically phase - locked .
the homodyne locking technique is also unique in that it is a sub - qnl frequency discriminator .
nonclassical states are considered key resources in quantum information processing applications @xcite , quantum key distribution systems @xcite and high - precision measurements such as gravitational wave interferometry @xcite .
one particular nonclassical state for which a number of applications have been proposed is the vacuum - squeezed state , which is commonly produced via a sub - threshold optical parametric oscillator ( opo ) in continuous - wave systems @xcite . in such systems ,
a squeezed vacuum is created via the non - linear interaction between a classical pump field and the modes of a cavity @xcite .
mechanical disturbances to the cavity make it impossible to create , measure or use the squeezed vacuum if the opo cavity is left uncontrolled .
consequently , all such systems are frequency - locked to the driving laser . in order to frequency - lock the opo and driving laser together
, an error signal must be generated that is bipolar and proportional to their frequency detuning .
common approaches to extracting an appropriate error signal include the pound - drever - hall ( pdh ) modulation - based technique @xcite or modulation - free techniques such as hnsch - couillaud @xcite and tilt - locking @xcite . these are all quite different techniques , but they share in common the idea that they are implemented separately from the homodyne measurements of the squeezed cavity mode .
such an approach leaves the squeezed output mode undisturbed and ready to direct towards subsequent experiments or measurement apparatus .
here we report on the successful implementation of an alternative scheme , which we call homodyne locking , that allows simultaneous measurement of the phase - shift induced by the cavity ( for the purposes of frequency locking ) as well as phase - quadrature homodyne detection of the squeezed output of the opo .
the homodyne locking technique makes use of the principle that a non - resonant mode , which provides the phase reference for frequency - locking , can also be used as the local oscillator in a homodyne detector .
the homodyne locking technique has the distinct advantage that any subsequent homodyne detector is automatically locked to a known phase angle .
this locking technique is easy and cost - effective to implement as well as , uniquely , being a sub - quantum - noise - limited frequency discriminator .
the homodyne locking technique is based on the observation that measurements of the phase quadrature of the output of a cavity can be used to estimate the frequency detuning between the input to the cavity and the cavity mode of interest @xcite . that is , assuming a singly - resonant cavity , no pump depletion , and parametric de - amplification , the sub - threshold opo is described by the following set of equations
@xcite : @xmath0 here , @xmath1 denotes the annihilation operator for the squeezed cavity mode defined in an appropriate rotating frame .
the detuning @xmath2 represents the frequency deviation of the cavity s resonant frequency from the nominal laser frequency @xmath3 .
the cavity decay rates are determined by mirror reflectivities @xmath4 and are given by @xmath5 for a cavity round trip time of @xmath6 .
the total decay rate for the fundamental is @xmath7 .
the seed field incident on the front face of the cavity is given by @xmath8 and the squeezed output field is @xmath9 .
the amplitude and phase quadratures of any field @xmath10 are denoted @xmath11 and @xmath12 respectively .
the intra - cavity annihilation operator can be decomposed into @xmath13 as can the seed @xmath14 .
. [ cavity-1 ] , can be solved in steady - state to find that @xmath15 - x_{s}^{\pm}. \label{alpha}\ ] ] from here on , we will assume without loss of generality that the input field is real .
linearising eq .
[ alpha ] near @xmath16 , we find that steady - state measurements of the phase quadrature of the output field are linearly proportional to the detuning and hence can be used as an error signal .
. [ cavity-1 ] can also be solved in frequency space for the fluctuating terms to yield an expression for the squeezing spectrum : @xmath17 + \delta x_l^{\pm } 2\sqrt{\kappa_{s}\kappa_l } } { ( \kappa+i\omega ) \mp \chi } \label{squeezingsimple}\ ] ] here we have set @xmath16 and @xmath18 is a frequency measured relative to one of the longitudinal resonances of the opo cavity .
we assume that @xmath18 is small compared to the free - spectral - range ( fsr ) of the opo cavity .
therefore , a single measurement of the phase quadrature of the output field affords both a direct measurement of the squeezed output of the sub - threshold opo , eq .
[ squeezingsimple ] , and the detuning between the cavity resonance and the input field as required for frequency locking , eq .
[ alpha ] . fig .
[ schem ] illustrates schematically the experiment used to demonstrate homodyne locking .
the 532 nm , frequency - doubled output of a diode - pumped , miniature monolithic nd : yag laser is used to pump a sub - threshold opo .
the nonlinear crystal is periodically - poled ktp , with a phase - matching temperature of @xmath19c .
the opo has a free - spectral range of 199 mhz and is operated with a parametric amplification of 3.9 db and de - amplification of 2.6db .
a small fraction of the source laser power is tapped off prior to frequency - doubling and is used as the seed to the opo and the local oscillator for the homodyne measurements quadrature homodyne measurement requires a bright , coherently related local oscillator ( lo ) with which to interfere the field of interest @xcite .
the phase of the lo relative to the field of interest determines the measurement quadrature . in the experiments described here the lo co - propagates with the seed to ensure that phase fluctuations are common - mode and hence that the homodyne detector is passively locked to measure the phase quadrature as required by eqs .
[ alpha ] and [ squeezingsimple ] . in the experiment
described here 1% of the power of the input field is set to the polarisation appropriate to the nonlinear crystal in the opo cavity .
the remaining 99% of the power is in the orthogonal , non - resonant polarisation and acts as the lo .
this is not the only means of implementing passive locking of the homodyne detector . in a previous experiment involving an empty cavity ,
the same effect was achieved by using a spatially offset , but co - propagating lo @xcite .
the upper trace of fig .
[ error ] shows a plot of @xmath20 measured using this technique . in this polarization implementation ,
the seed and the local oscillator each produce a dispersion shaped error signal when resonant with the cavity , but with flipped sign relative to each other . the lower trace in fig .
[ error ] is a plot of the measured transmission of the seed field through the opo cavity .
the error signal is used to lock the opo cavity to a resonance for the seed mode , which corresponds to one of the peaks in the lower trace in fig .
[ error ] .
the fluctuating component of the same homodyne measurement that lead to fig . [ error ] can be used to measure the squeezed output of the opo . fig .
[ sqz ] is a plot of the variance of the homodyne measurement , @xmath21 , at frequencies near the first resonance of the opo , made when the feedback controller is active .
maximum squeezing of approximately 2db has been observed at the resonant frequency of 199mhz .
coupled with a measured de - amplification of 2.6db , this measurement suggests that the overall escape and detection efficiency of the system is greater than 87% @xcite . with thoughtful design , the homodyne locking technique can work even with an experiment interposed between the opo and the homodyne detector .
furthermore , a co - propagating local oscillator turns out to be a most useful addition to subsequent experiments .
for example , consider fig .
[ unisqz](a ) in which the schematic diagram for a universal squeezer is reproduced from ref .
@xcite . here
, an input field is interferred with a phase - quadrature - squeezed vacuum on a beamsplitter of transmittivity @xmath22 .
phase quadrature detection of one port of the beamsplitter is performed in the usual fashion @xcite .
the homodyne photocurrent is amplified by @xmath23 and fed forward to the other port of the beamsplitter .
after feed - forward , the output is a phase - quadrature squeezed version of the input , with excellent purity @xcite . not shown in fig .
[ unisqz](a ) are the requirements to lock the frequency of the opo cavity and the phase of the lo .
fig .
[ unisqz](b ) shows a schematic diagram for an alternative implementation . in that implementation
, the homodyne locking technique can be used to simultaneously lock the frequency of the opo cavity and the lo phase .
the diagram is shown with polarisation modes , but the idea could be equally well implemented using displaced beams , higher order spatial modes and so forth . treating the input and output of the opo cavity as vector quantities , @xmath24 and @xmath25 , it is straightforward to trace the change in notation through eqs .
[ cavity-1 ] , [ alpha ] and [ squeezingsimple ] .
we treat @xmath26 as being the desired cavity mode , and @xmath27 as an orthogonal mode in one of the degrees of freedom .
the latter is not resonant with the opo cavity , but it co - propagates with the former and interacts with all of the same linear optical elements .
the steady - state value of the homodyne photocurrent is @xmath28 as required to frequency - lock the opo cavity and the linearised fluctuating component of the homodyne photocurrent is as required for the universal squeezing operation @xcite when @xmath16 . in summary
, we have proposed a new approach to controlling the resonant frequency of a sub - threshold opo .
this approach is based on direct measurement of the squeezed phase quadrature output of the opo , rather than on the traditional approach where the squeezed output is not used for frequency - locking .
the homodyne locking technique is cheap , easy to implement and , uniquely , is a sub - quantum - noise - limited frequency discriminator .
we have demonstrated the homodyne locking technique on a sub - threshold opo producing approximately 2db of squeezing .
we have also shown , via an explicit example , that homodyne locking remains useful when an experiment is interposed between the opo and the homodyne detector .
indeed this approach greatly simplifies the experiment as it relieves the necessity of locking subsequent homodyne detectors . _
acknowledgments - _ this work was supported by the australian research council .
99 m. a. nielsen and i. l. chuang , _ quantum computation and quantum information _ , cambridge university press , uk , 2000 . n. j. cerf and p. grangier , j. opt . soc . am .
b , * 24 * , 324 ( 2007 ) .
v. chickarmane and b. bhawal , phys .
lett . a , * 190 * 22 ( 1994 ) . h .- a .
bachor and t. c. ralph , _ a guide to experiments in quantum optics _ ( 2 ed . ) , wiley - vch , berlin , 2004 .
r. w. p. drever , j. l. hall , f. v. kowalski , j. hough , g. m. ford , a. j. munley and h. ward , appl .
phys .
b. , * 31 * , 97 , ( 1983 ) .
t. hnsch , b. couillaud , opt . commun . * 35 * , 441 ( 1980 ) .
d. a. shaddock , m. b. gray , and d. e. mcclelland , opt .
lett .
, * 24 * 1499 ( 1999 )
. e. h. huntington , m. r. james , i. r. petersen , proc .
ieee cdc , new orleans , usa , 2007 .
m. heurs , e. h. huntington , s. z. s. hassen , i. r.petersen and m. r. james , proc .
qcmc , canada , paper number p2 - 41 , 2008 ; and s. z. s. hassen , m. heurs , e. h. huntington , i. r. petersen and m. r. james , under review j. phys . b. ( 2009 ) .
y. takeno , m. yukawa , h. yonezawa , and a. furusawa , opt .
exp .
, * 15 * , 4321 ( 2007 ) .
jun - ichi yoshikawa , t. hayashi , t. akiyama , n. takei , u. l. andersen and a. furusawa , proc .
qels , 2006 paper qfd2 .
r. filip , p. marek , u. l. andersen , phys .
rev .
a 71 , 042308 ( 2005 ) . |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.